diff options
Diffstat (limited to 'lib/Lex')
| -rw-r--r-- | lib/Lex/CMakeLists.txt | 26 | ||||
| -rw-r--r-- | lib/Lex/HeaderMap.cpp | 245 | ||||
| -rw-r--r-- | lib/Lex/HeaderSearch.cpp | 446 | ||||
| -rw-r--r-- | lib/Lex/Lexer.cpp | 1809 | ||||
| -rw-r--r-- | lib/Lex/LiteralSupport.cpp | 929 | ||||
| -rw-r--r-- | lib/Lex/MacroArgs.cpp | 240 | ||||
| -rw-r--r-- | lib/Lex/MacroArgs.h | 109 | ||||
| -rw-r--r-- | lib/Lex/MacroInfo.cpp | 75 | ||||
| -rw-r--r-- | lib/Lex/Makefile | 28 | ||||
| -rw-r--r-- | lib/Lex/PPCaching.cpp | 113 | ||||
| -rw-r--r-- | lib/Lex/PPDirectives.cpp | 1665 | ||||
| -rw-r--r-- | lib/Lex/PPExpressions.cpp | 717 | ||||
| -rw-r--r-- | lib/Lex/PPLexerChange.cpp | 345 | ||||
| -rw-r--r-- | lib/Lex/PPMacroExpansion.cpp | 605 | ||||
| -rw-r--r-- | lib/Lex/PTHLexer.cpp | 701 | ||||
| -rw-r--r-- | lib/Lex/Pragma.cpp | 699 | ||||
| -rw-r--r-- | lib/Lex/Preprocessor.cpp | 478 | ||||
| -rw-r--r-- | lib/Lex/PreprocessorLexer.cpp | 45 | ||||
| -rw-r--r-- | lib/Lex/ScratchBuffer.cpp | 73 | ||||
| -rw-r--r-- | lib/Lex/TokenConcatenation.cpp | 219 | ||||
| -rw-r--r-- | lib/Lex/TokenLexer.cpp | 542 |
21 files changed, 10109 insertions, 0 deletions
diff --git a/lib/Lex/CMakeLists.txt b/lib/Lex/CMakeLists.txt new file mode 100644 index 0000000..a7237a7 --- /dev/null +++ b/lib/Lex/CMakeLists.txt @@ -0,0 +1,26 @@ +set(LLVM_NO_RTTI 1) + +# TODO: Add -maltivec when ARCH is PowerPC. + +add_clang_library(clangLex + HeaderMap.cpp + HeaderSearch.cpp + Lexer.cpp + LiteralSupport.cpp + MacroArgs.cpp + MacroInfo.cpp + PPCaching.cpp + PPDirectives.cpp + PPExpressions.cpp + PPLexerChange.cpp + PPMacroExpansion.cpp + Pragma.cpp + Preprocessor.cpp + PreprocessorLexer.cpp + PTHLexer.cpp + ScratchBuffer.cpp + TokenLexer.cpp + TokenConcatenation.cpp + ) + +add_dependencies(clangLex ClangDiagnosticLex) diff --git a/lib/Lex/HeaderMap.cpp b/lib/Lex/HeaderMap.cpp new file mode 100644 index 0000000..4c8b70e --- /dev/null +++ b/lib/Lex/HeaderMap.cpp @@ -0,0 +1,245 @@ +//===--- HeaderMap.cpp - A file that acts like dir of symlinks ------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements the HeaderMap interface. +// +//===----------------------------------------------------------------------===// + +#include "clang/Lex/HeaderMap.h" +#include "clang/Basic/FileManager.h" +#include "llvm/ADT/OwningPtr.h" +#include "llvm/ADT/SmallString.h" +#include "llvm/Support/DataTypes.h" +#include "llvm/Support/MathExtras.h" +#include "llvm/Support/MemoryBuffer.h" +#include <cstdio> +using namespace clang; + +//===----------------------------------------------------------------------===// +// Data Structures and Manifest Constants +//===----------------------------------------------------------------------===// + +enum { + HMAP_HeaderMagicNumber = ('h' << 24) | ('m' << 16) | ('a' << 8) | 'p', + HMAP_HeaderVersion = 1, + + HMAP_EmptyBucketKey = 0 +}; + +namespace clang { +struct HMapBucket { + uint32_t Key; // Offset (into strings) of key. + + uint32_t Prefix; // Offset (into strings) of value prefix. + uint32_t Suffix; // Offset (into strings) of value suffix. +}; + +struct HMapHeader { + uint32_t Magic; // Magic word, also indicates byte order. + uint16_t Version; // Version number -- currently 1. + uint16_t Reserved; // Reserved for future use - zero for now. + uint32_t StringsOffset; // Offset to start of string pool. + uint32_t NumEntries; // Number of entries in the string table. + uint32_t NumBuckets; // Number of buckets (always a power of 2). + uint32_t MaxValueLength; // Length of longest result path (excluding nul). + // An array of 'NumBuckets' HMapBucket objects follows this header. + // Strings follow the buckets, at StringsOffset. +}; +} // end namespace clang. + +/// HashHMapKey - This is the 'well known' hash function required by the file +/// format, used to look up keys in the hash table. The hash table uses simple +/// linear probing based on this function. +static inline unsigned HashHMapKey(const char *S, const char *End) { + unsigned Result = 0; + + for (; S != End; S++) + Result += tolower(*S) * 13; + return Result; +} + + + +//===----------------------------------------------------------------------===// +// Verification and Construction +//===----------------------------------------------------------------------===// + +/// HeaderMap::Create - This attempts to load the specified file as a header +/// map. If it doesn't look like a HeaderMap, it gives up and returns null. +/// If it looks like a HeaderMap but is obviously corrupted, it puts a reason +/// into the string error argument and returns null. +const HeaderMap *HeaderMap::Create(const FileEntry *FE) { + // If the file is too small to be a header map, ignore it. + unsigned FileSize = FE->getSize(); + if (FileSize <= sizeof(HMapHeader)) return 0; + + llvm::OwningPtr<const llvm::MemoryBuffer> FileBuffer( + llvm::MemoryBuffer::getFile(FE->getName(), 0, FE->getSize())); + if (FileBuffer == 0) return 0; // Unreadable file? + const char *FileStart = FileBuffer->getBufferStart(); + + // We know the file is at least as big as the header, check it now. + const HMapHeader *Header = reinterpret_cast<const HMapHeader*>(FileStart); + + // Sniff it to see if it's a headermap by checking the magic number and + // version. + bool NeedsByteSwap; + if (Header->Magic == HMAP_HeaderMagicNumber && + Header->Version == HMAP_HeaderVersion) + NeedsByteSwap = false; + else if (Header->Magic == llvm::ByteSwap_32(HMAP_HeaderMagicNumber) && + Header->Version == llvm::ByteSwap_16(HMAP_HeaderVersion)) + NeedsByteSwap = true; // Mixed endianness headermap. + else + return 0; // Not a header map. + + if (Header->Reserved != 0) return 0; + + // Okay, everything looks good, create the header map. + return new HeaderMap(FileBuffer.take(), NeedsByteSwap); +} + +HeaderMap::~HeaderMap() { + delete FileBuffer; +} + +//===----------------------------------------------------------------------===// +// Utility Methods +//===----------------------------------------------------------------------===// + + +/// getFileName - Return the filename of the headermap. +const char *HeaderMap::getFileName() const { + return FileBuffer->getBufferIdentifier(); +} + +unsigned HeaderMap::getEndianAdjustedWord(unsigned X) const { + if (!NeedsBSwap) return X; + return llvm::ByteSwap_32(X); +} + +/// getHeader - Return a reference to the file header, in unbyte-swapped form. +/// This method cannot fail. +const HMapHeader &HeaderMap::getHeader() const { + // We know the file is at least as big as the header. Return it. + return *reinterpret_cast<const HMapHeader*>(FileBuffer->getBufferStart()); +} + +/// getBucket - Return the specified hash table bucket from the header map, +/// bswap'ing its fields as appropriate. If the bucket number is not valid, +/// this return a bucket with an empty key (0). +HMapBucket HeaderMap::getBucket(unsigned BucketNo) const { + HMapBucket Result; + Result.Key = HMAP_EmptyBucketKey; + + const HMapBucket *BucketArray = + reinterpret_cast<const HMapBucket*>(FileBuffer->getBufferStart() + + sizeof(HMapHeader)); + + const HMapBucket *BucketPtr = BucketArray+BucketNo; + if ((char*)(BucketPtr+1) > FileBuffer->getBufferEnd()) { + Result.Prefix = 0; + Result.Suffix = 0; + return Result; // Invalid buffer, corrupt hmap. + } + + // Otherwise, the bucket is valid. Load the values, bswapping as needed. + Result.Key = getEndianAdjustedWord(BucketPtr->Key); + Result.Prefix = getEndianAdjustedWord(BucketPtr->Prefix); + Result.Suffix = getEndianAdjustedWord(BucketPtr->Suffix); + return Result; +} + +/// getString - Look up the specified string in the string table. If the string +/// index is not valid, it returns an empty string. +const char *HeaderMap::getString(unsigned StrTabIdx) const { + // Add the start of the string table to the idx. + StrTabIdx += getEndianAdjustedWord(getHeader().StringsOffset); + + // Check for invalid index. + if (StrTabIdx >= FileBuffer->getBufferSize()) + return 0; + + // Otherwise, we have a valid pointer into the file. Just return it. We know + // that the "string" can not overrun the end of the file, because the buffer + // is nul terminated by virtue of being a MemoryBuffer. + return FileBuffer->getBufferStart()+StrTabIdx; +} + +/// StringsEqualWithoutCase - Compare the specified two strings for case- +/// insensitive equality, returning true if they are equal. Both strings are +/// known to have the same length. +static bool StringsEqualWithoutCase(const char *S1, const char *S2, + unsigned Len) { + for (; Len; ++S1, ++S2, --Len) + if (tolower(*S1) != tolower(*S2)) + return false; + return true; +} + +//===----------------------------------------------------------------------===// +// The Main Drivers +//===----------------------------------------------------------------------===// + +/// dump - Print the contents of this headermap to stderr. +void HeaderMap::dump() const { + const HMapHeader &Hdr = getHeader(); + unsigned NumBuckets = getEndianAdjustedWord(Hdr.NumBuckets); + + fprintf(stderr, "Header Map %s:\n %d buckets, %d entries\n", + getFileName(), NumBuckets, + getEndianAdjustedWord(Hdr.NumEntries)); + + for (unsigned i = 0; i != NumBuckets; ++i) { + HMapBucket B = getBucket(i); + if (B.Key == HMAP_EmptyBucketKey) continue; + + const char *Key = getString(B.Key); + const char *Prefix = getString(B.Prefix); + const char *Suffix = getString(B.Suffix); + fprintf(stderr, " %d. %s -> '%s' '%s'\n", i, Key, Prefix, Suffix); + } +} + +/// LookupFile - Check to see if the specified relative filename is located in +/// this HeaderMap. If so, open it and return its FileEntry. +const FileEntry *HeaderMap::LookupFile(const char *FilenameStart, + const char *FilenameEnd, + FileManager &FM) const { + const HMapHeader &Hdr = getHeader(); + unsigned NumBuckets = getEndianAdjustedWord(Hdr.NumBuckets); + + // If the number of buckets is not a power of two, the headermap is corrupt. + // Don't probe infinitely. + if (NumBuckets & (NumBuckets-1)) + return 0; + + // Linearly probe the hash table. + for (unsigned Bucket = HashHMapKey(FilenameStart, FilenameEnd);; ++Bucket) { + HMapBucket B = getBucket(Bucket & (NumBuckets-1)); + if (B.Key == HMAP_EmptyBucketKey) return 0; // Hash miss. + + // See if the key matches. If not, probe on. + const char *Key = getString(B.Key); + unsigned BucketKeyLen = strlen(Key); + if (BucketKeyLen != unsigned(FilenameEnd-FilenameStart)) + continue; + + // See if the actual strings equal. + if (!StringsEqualWithoutCase(FilenameStart, Key, BucketKeyLen)) + continue; + + // If so, we have a match in the hash table. Construct the destination + // path. + llvm::SmallString<1024> DestPath; + DestPath += getString(B.Prefix); + DestPath += getString(B.Suffix); + return FM.getFile(DestPath.begin(), DestPath.end()); + } +} diff --git a/lib/Lex/HeaderSearch.cpp b/lib/Lex/HeaderSearch.cpp new file mode 100644 index 0000000..129fa1a --- /dev/null +++ b/lib/Lex/HeaderSearch.cpp @@ -0,0 +1,446 @@ +//===--- HeaderSearch.cpp - Resolve Header File Locations ---===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements the DirectoryLookup and HeaderSearch interfaces. +// +//===----------------------------------------------------------------------===// + +#include "clang/Lex/HeaderSearch.h" +#include "clang/Lex/HeaderMap.h" +#include "clang/Basic/FileManager.h" +#include "clang/Basic/IdentifierTable.h" +#include "llvm/System/Path.h" +#include "llvm/ADT/SmallString.h" +#include <cstdio> +using namespace clang; + +const IdentifierInfo * +HeaderFileInfo::getControllingMacro(ExternalIdentifierLookup *External) { + if (ControllingMacro) + return ControllingMacro; + + if (!ControllingMacroID || !External) + return 0; + + ControllingMacro = External->GetIdentifier(ControllingMacroID); + return ControllingMacro; +} + +HeaderSearch::HeaderSearch(FileManager &FM) : FileMgr(FM), FrameworkMap(64) { + SystemDirIdx = 0; + NoCurDirSearch = false; + + ExternalLookup = 0; + NumIncluded = 0; + NumMultiIncludeFileOptzn = 0; + NumFrameworkLookups = NumSubFrameworkLookups = 0; +} + +HeaderSearch::~HeaderSearch() { + // Delete headermaps. + for (unsigned i = 0, e = HeaderMaps.size(); i != e; ++i) + delete HeaderMaps[i].second; +} + +void HeaderSearch::PrintStats() { + fprintf(stderr, "\n*** HeaderSearch Stats:\n"); + fprintf(stderr, "%d files tracked.\n", (int)FileInfo.size()); + unsigned NumOnceOnlyFiles = 0, MaxNumIncludes = 0, NumSingleIncludedFiles = 0; + for (unsigned i = 0, e = FileInfo.size(); i != e; ++i) { + NumOnceOnlyFiles += FileInfo[i].isImport; + if (MaxNumIncludes < FileInfo[i].NumIncludes) + MaxNumIncludes = FileInfo[i].NumIncludes; + NumSingleIncludedFiles += FileInfo[i].NumIncludes == 1; + } + fprintf(stderr, " %d #import/#pragma once files.\n", NumOnceOnlyFiles); + fprintf(stderr, " %d included exactly once.\n", NumSingleIncludedFiles); + fprintf(stderr, " %d max times a file is included.\n", MaxNumIncludes); + + fprintf(stderr, " %d #include/#include_next/#import.\n", NumIncluded); + fprintf(stderr, " %d #includes skipped due to" + " the multi-include optimization.\n", NumMultiIncludeFileOptzn); + + fprintf(stderr, "%d framework lookups.\n", NumFrameworkLookups); + fprintf(stderr, "%d subframework lookups.\n", NumSubFrameworkLookups); +} + +/// CreateHeaderMap - This method returns a HeaderMap for the specified +/// FileEntry, uniquing them through the the 'HeaderMaps' datastructure. +const HeaderMap *HeaderSearch::CreateHeaderMap(const FileEntry *FE) { + // We expect the number of headermaps to be small, and almost always empty. + // If it ever grows, use of a linear search should be re-evaluated. + if (!HeaderMaps.empty()) { + for (unsigned i = 0, e = HeaderMaps.size(); i != e; ++i) + // Pointer equality comparison of FileEntries works because they are + // already uniqued by inode. + if (HeaderMaps[i].first == FE) + return HeaderMaps[i].second; + } + + if (const HeaderMap *HM = HeaderMap::Create(FE)) { + HeaderMaps.push_back(std::make_pair(FE, HM)); + return HM; + } + + return 0; +} + +//===----------------------------------------------------------------------===// +// File lookup within a DirectoryLookup scope +//===----------------------------------------------------------------------===// + +/// getName - Return the directory or filename corresponding to this lookup +/// object. +const char *DirectoryLookup::getName() const { + if (isNormalDir()) + return getDir()->getName(); + if (isFramework()) + return getFrameworkDir()->getName(); + assert(isHeaderMap() && "Unknown DirectoryLookup"); + return getHeaderMap()->getFileName(); +} + + +/// LookupFile - Lookup the specified file in this search path, returning it +/// if it exists or returning null if not. +const FileEntry *DirectoryLookup::LookupFile(const char *FilenameStart, + const char *FilenameEnd, + HeaderSearch &HS) const { + llvm::SmallString<1024> TmpDir; + if (isNormalDir()) { + // Concatenate the requested file onto the directory. + // FIXME: Portability. Filename concatenation should be in sys::Path. + TmpDir += getDir()->getName(); + TmpDir.push_back('/'); + TmpDir.append(FilenameStart, FilenameEnd); + return HS.getFileMgr().getFile(TmpDir.begin(), TmpDir.end()); + } + + if (isFramework()) + return DoFrameworkLookup(FilenameStart, FilenameEnd, HS); + + assert(isHeaderMap() && "Unknown directory lookup"); + return getHeaderMap()->LookupFile(FilenameStart, FilenameEnd,HS.getFileMgr()); +} + + +/// DoFrameworkLookup - Do a lookup of the specified file in the current +/// DirectoryLookup, which is a framework directory. +const FileEntry *DirectoryLookup::DoFrameworkLookup(const char *FilenameStart, + const char *FilenameEnd, + HeaderSearch &HS) const { + FileManager &FileMgr = HS.getFileMgr(); + + // Framework names must have a '/' in the filename. + const char *SlashPos = std::find(FilenameStart, FilenameEnd, '/'); + if (SlashPos == FilenameEnd) return 0; + + // Find out if this is the home for the specified framework, by checking + // HeaderSearch. Possible answer are yes/no and unknown. + const DirectoryEntry *&FrameworkDirCache = + HS.LookupFrameworkCache(FilenameStart, SlashPos); + + // If it is known and in some other directory, fail. + if (FrameworkDirCache && FrameworkDirCache != getFrameworkDir()) + return 0; + + // Otherwise, construct the path to this framework dir. + + // FrameworkName = "/System/Library/Frameworks/" + llvm::SmallString<1024> FrameworkName; + FrameworkName += getFrameworkDir()->getName(); + if (FrameworkName.empty() || FrameworkName.back() != '/') + FrameworkName.push_back('/'); + + // FrameworkName = "/System/Library/Frameworks/Cocoa" + FrameworkName.append(FilenameStart, SlashPos); + + // FrameworkName = "/System/Library/Frameworks/Cocoa.framework/" + FrameworkName += ".framework/"; + + // If the cache entry is still unresolved, query to see if the cache entry is + // still unresolved. If so, check its existence now. + if (FrameworkDirCache == 0) { + HS.IncrementFrameworkLookupCount(); + + // If the framework dir doesn't exist, we fail. + // FIXME: It's probably more efficient to query this with FileMgr.getDir. + if (!llvm::sys::Path(std::string(FrameworkName.begin(), + FrameworkName.end())).exists()) + return 0; + + // Otherwise, if it does, remember that this is the right direntry for this + // framework. + FrameworkDirCache = getFrameworkDir(); + } + + // Check "/System/Library/Frameworks/Cocoa.framework/Headers/file.h" + unsigned OrigSize = FrameworkName.size(); + + FrameworkName += "Headers/"; + FrameworkName.append(SlashPos+1, FilenameEnd); + if (const FileEntry *FE = FileMgr.getFile(FrameworkName.begin(), + FrameworkName.end())) { + return FE; + } + + // Check "/System/Library/Frameworks/Cocoa.framework/PrivateHeaders/file.h" + const char *Private = "Private"; + FrameworkName.insert(FrameworkName.begin()+OrigSize, Private, + Private+strlen(Private)); + return FileMgr.getFile(FrameworkName.begin(), FrameworkName.end()); +} + + +//===----------------------------------------------------------------------===// +// Header File Location. +//===----------------------------------------------------------------------===// + + +/// LookupFile - Given a "foo" or <foo> reference, look up the indicated file, +/// return null on failure. isAngled indicates whether the file reference is +/// for system #include's or not (i.e. using <> instead of ""). CurFileEnt, if +/// non-null, indicates where the #including file is, in case a relative search +/// is needed. +const FileEntry *HeaderSearch::LookupFile(const char *FilenameStart, + const char *FilenameEnd, + bool isAngled, + const DirectoryLookup *FromDir, + const DirectoryLookup *&CurDir, + const FileEntry *CurFileEnt) { + // If 'Filename' is absolute, check to see if it exists and no searching. + // FIXME: Portability. This should be a sys::Path interface, this doesn't + // handle things like C:\foo.txt right, nor win32 \\network\device\blah. + if (FilenameStart[0] == '/') { + CurDir = 0; + + // If this was an #include_next "/absolute/file", fail. + if (FromDir) return 0; + + // Otherwise, just return the file. + return FileMgr.getFile(FilenameStart, FilenameEnd); + } + + // Step #0, unless disabled, check to see if the file is in the #includer's + // directory. This has to be based on CurFileEnt, not CurDir, because + // CurFileEnt could be a #include of a subdirectory (#include "foo/bar.h") and + // a subsequent include of "baz.h" should resolve to "whatever/foo/baz.h". + // This search is not done for <> headers. + if (CurFileEnt && !isAngled && !NoCurDirSearch) { + llvm::SmallString<1024> TmpDir; + // Concatenate the requested file onto the directory. + // FIXME: Portability. Filename concatenation should be in sys::Path. + TmpDir += CurFileEnt->getDir()->getName(); + TmpDir.push_back('/'); + TmpDir.append(FilenameStart, FilenameEnd); + if (const FileEntry *FE = FileMgr.getFile(TmpDir.begin(), TmpDir.end())) { + // Leave CurDir unset. + // This file is a system header or C++ unfriendly if the old file is. + // + // Note that the temporary 'DirInfo' is required here, as either call to + // getFileInfo could resize the vector and we don't want to rely on order + // of evaluation. + unsigned DirInfo = getFileInfo(CurFileEnt).DirInfo; + getFileInfo(FE).DirInfo = DirInfo; + return FE; + } + } + + CurDir = 0; + + // If this is a system #include, ignore the user #include locs. + unsigned i = isAngled ? SystemDirIdx : 0; + + // If this is a #include_next request, start searching after the directory the + // file was found in. + if (FromDir) + i = FromDir-&SearchDirs[0]; + + // Cache all of the lookups performed by this method. Many headers are + // multiply included, and the "pragma once" optimization prevents them from + // being relex/pp'd, but they would still have to search through a + // (potentially huge) series of SearchDirs to find it. + std::pair<unsigned, unsigned> &CacheLookup = + LookupFileCache.GetOrCreateValue(FilenameStart, FilenameEnd).getValue(); + + // If the entry has been previously looked up, the first value will be + // non-zero. If the value is equal to i (the start point of our search), then + // this is a matching hit. + if (CacheLookup.first == i+1) { + // Skip querying potentially lots of directories for this lookup. + i = CacheLookup.second; + } else { + // Otherwise, this is the first query, or the previous query didn't match + // our search start. We will fill in our found location below, so prime the + // start point value. + CacheLookup.first = i+1; + } + + // Check each directory in sequence to see if it contains this file. + for (; i != SearchDirs.size(); ++i) { + const FileEntry *FE = + SearchDirs[i].LookupFile(FilenameStart, FilenameEnd, *this); + if (!FE) continue; + + CurDir = &SearchDirs[i]; + + // This file is a system header or C++ unfriendly if the dir is. + getFileInfo(FE).DirInfo = CurDir->getDirCharacteristic(); + + // Remember this location for the next lookup we do. + CacheLookup.second = i; + return FE; + } + + // Otherwise, didn't find it. Remember we didn't find this. + CacheLookup.second = SearchDirs.size(); + return 0; +} + +/// LookupSubframeworkHeader - Look up a subframework for the specified +/// #include file. For example, if #include'ing <HIToolbox/HIToolbox.h> from +/// within ".../Carbon.framework/Headers/Carbon.h", check to see if HIToolbox +/// is a subframework within Carbon.framework. If so, return the FileEntry +/// for the designated file, otherwise return null. +const FileEntry *HeaderSearch:: +LookupSubframeworkHeader(const char *FilenameStart, + const char *FilenameEnd, + const FileEntry *ContextFileEnt) { + assert(ContextFileEnt && "No context file?"); + + // Framework names must have a '/' in the filename. Find it. + const char *SlashPos = std::find(FilenameStart, FilenameEnd, '/'); + if (SlashPos == FilenameEnd) return 0; + + // Look up the base framework name of the ContextFileEnt. + const char *ContextName = ContextFileEnt->getName(); + + // If the context info wasn't a framework, couldn't be a subframework. + const char *FrameworkPos = strstr(ContextName, ".framework/"); + if (FrameworkPos == 0) + return 0; + + llvm::SmallString<1024> FrameworkName(ContextName, + FrameworkPos+strlen(".framework/")); + + // Append Frameworks/HIToolbox.framework/ + FrameworkName += "Frameworks/"; + FrameworkName.append(FilenameStart, SlashPos); + FrameworkName += ".framework/"; + + llvm::StringMapEntry<const DirectoryEntry *> &CacheLookup = + FrameworkMap.GetOrCreateValue(FilenameStart, SlashPos); + + // Some other location? + if (CacheLookup.getValue() && + CacheLookup.getKeyLength() == FrameworkName.size() && + memcmp(CacheLookup.getKeyData(), &FrameworkName[0], + CacheLookup.getKeyLength()) != 0) + return 0; + + // Cache subframework. + if (CacheLookup.getValue() == 0) { + ++NumSubFrameworkLookups; + + // If the framework dir doesn't exist, we fail. + const DirectoryEntry *Dir = FileMgr.getDirectory(FrameworkName.begin(), + FrameworkName.end()); + if (Dir == 0) return 0; + + // Otherwise, if it does, remember that this is the right direntry for this + // framework. + CacheLookup.setValue(Dir); + } + + const FileEntry *FE = 0; + + // Check ".../Frameworks/HIToolbox.framework/Headers/HIToolbox.h" + llvm::SmallString<1024> HeadersFilename(FrameworkName); + HeadersFilename += "Headers/"; + HeadersFilename.append(SlashPos+1, FilenameEnd); + if (!(FE = FileMgr.getFile(HeadersFilename.begin(), + HeadersFilename.end()))) { + + // Check ".../Frameworks/HIToolbox.framework/PrivateHeaders/HIToolbox.h" + HeadersFilename = FrameworkName; + HeadersFilename += "PrivateHeaders/"; + HeadersFilename.append(SlashPos+1, FilenameEnd); + if (!(FE = FileMgr.getFile(HeadersFilename.begin(), HeadersFilename.end()))) + return 0; + } + + // This file is a system header or C++ unfriendly if the old file is. + // + // Note that the temporary 'DirInfo' is required here, as either call to + // getFileInfo could resize the vector and we don't want to rely on order + // of evaluation. + unsigned DirInfo = getFileInfo(ContextFileEnt).DirInfo; + getFileInfo(FE).DirInfo = DirInfo; + return FE; +} + +//===----------------------------------------------------------------------===// +// File Info Management. +//===----------------------------------------------------------------------===// + + +/// getFileInfo - Return the HeaderFileInfo structure for the specified +/// FileEntry. +HeaderFileInfo &HeaderSearch::getFileInfo(const FileEntry *FE) { + if (FE->getUID() >= FileInfo.size()) + FileInfo.resize(FE->getUID()+1); + return FileInfo[FE->getUID()]; +} + +void HeaderSearch::setHeaderFileInfoForUID(HeaderFileInfo HFI, unsigned UID) { + if (UID >= FileInfo.size()) + FileInfo.resize(UID+1); + FileInfo[UID] = HFI; +} + +/// ShouldEnterIncludeFile - Mark the specified file as a target of of a +/// #include, #include_next, or #import directive. Return false if #including +/// the file will have no effect or true if we should include it. +bool HeaderSearch::ShouldEnterIncludeFile(const FileEntry *File, bool isImport){ + ++NumIncluded; // Count # of attempted #includes. + + // Get information about this file. + HeaderFileInfo &FileInfo = getFileInfo(File); + + // If this is a #import directive, check that we have not already imported + // this header. + if (isImport) { + // If this has already been imported, don't import it again. + FileInfo.isImport = true; + + // Has this already been #import'ed or #include'd? + if (FileInfo.NumIncludes) return false; + } else { + // Otherwise, if this is a #include of a file that was previously #import'd + // or if this is the second #include of a #pragma once file, ignore it. + if (FileInfo.isImport) + return false; + } + + // Next, check to see if the file is wrapped with #ifndef guards. If so, and + // if the macro that guards it is defined, we know the #include has no effect. + if (const IdentifierInfo *ControllingMacro + = FileInfo.getControllingMacro(ExternalLookup)) + if (ControllingMacro->hasMacroDefinition()) { + ++NumMultiIncludeFileOptzn; + return false; + } + + // Increment the number of times this file has been included. + ++FileInfo.NumIncludes; + + return true; +} + + diff --git a/lib/Lex/Lexer.cpp b/lib/Lex/Lexer.cpp new file mode 100644 index 0000000..c2ffd6d --- /dev/null +++ b/lib/Lex/Lexer.cpp @@ -0,0 +1,1809 @@ +//===--- Lexer.cpp - C Language Family Lexer ------------------------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements the Lexer and Token interfaces. +// +//===----------------------------------------------------------------------===// +// +// TODO: GCC Diagnostics emitted by the lexer: +// PEDWARN: (form feed|vertical tab) in preprocessing directive +// +// Universal characters, unicode, char mapping: +// WARNING: `%.*s' is not in NFKC +// WARNING: `%.*s' is not in NFC +// +// Other: +// TODO: Options to support: +// -fexec-charset,-fwide-exec-charset +// +//===----------------------------------------------------------------------===// + +#include "clang/Lex/Lexer.h" +#include "clang/Lex/Preprocessor.h" +#include "clang/Lex/LexDiagnostic.h" +#include "clang/Basic/SourceManager.h" +#include "llvm/Support/Compiler.h" +#include "llvm/Support/MemoryBuffer.h" +#include <cctype> +using namespace clang; + +static void InitCharacterInfo(); + +//===----------------------------------------------------------------------===// +// Token Class Implementation +//===----------------------------------------------------------------------===// + +/// isObjCAtKeyword - Return true if we have an ObjC keyword identifier. +bool Token::isObjCAtKeyword(tok::ObjCKeywordKind objcKey) const { + if (IdentifierInfo *II = getIdentifierInfo()) + return II->getObjCKeywordID() == objcKey; + return false; +} + +/// getObjCKeywordID - Return the ObjC keyword kind. +tok::ObjCKeywordKind Token::getObjCKeywordID() const { + IdentifierInfo *specId = getIdentifierInfo(); + return specId ? specId->getObjCKeywordID() : tok::objc_not_keyword; +} + + +//===----------------------------------------------------------------------===// +// Lexer Class Implementation +//===----------------------------------------------------------------------===// + +void Lexer::InitLexer(const char *BufStart, const char *BufPtr, + const char *BufEnd) { + InitCharacterInfo(); + + BufferStart = BufStart; + BufferPtr = BufPtr; + BufferEnd = BufEnd; + + assert(BufEnd[0] == 0 && + "We assume that the input buffer has a null character at the end" + " to simplify lexing!"); + + Is_PragmaLexer = false; + + // Start of the file is a start of line. + IsAtStartOfLine = true; + + // We are not after parsing a #. + ParsingPreprocessorDirective = false; + + // We are not after parsing #include. + ParsingFilename = false; + + // We are not in raw mode. Raw mode disables diagnostics and interpretation + // of tokens (e.g. identifiers, thus disabling macro expansion). It is used + // to quickly lex the tokens of the buffer, e.g. when handling a "#if 0" block + // or otherwise skipping over tokens. + LexingRawMode = false; + + // Default to not keeping comments. + ExtendedTokenMode = 0; +} + +/// Lexer constructor - Create a new lexer object for the specified buffer +/// with the specified preprocessor managing the lexing process. This lexer +/// assumes that the associated file buffer and Preprocessor objects will +/// outlive it, so it doesn't take ownership of either of them. +Lexer::Lexer(FileID FID, Preprocessor &PP) + : PreprocessorLexer(&PP, FID), + FileLoc(PP.getSourceManager().getLocForStartOfFile(FID)), + Features(PP.getLangOptions()) { + + const llvm::MemoryBuffer *InputFile = PP.getSourceManager().getBuffer(FID); + + InitLexer(InputFile->getBufferStart(), InputFile->getBufferStart(), + InputFile->getBufferEnd()); + + // Default to keeping comments if the preprocessor wants them. + SetCommentRetentionState(PP.getCommentRetentionState()); +} + +/// Lexer constructor - Create a new raw lexer object. This object is only +/// suitable for calls to 'LexRawToken'. This lexer assumes that the text +/// range will outlive it, so it doesn't take ownership of it. +Lexer::Lexer(SourceLocation fileloc, const LangOptions &features, + const char *BufStart, const char *BufPtr, const char *BufEnd) + : FileLoc(fileloc), Features(features) { + + InitLexer(BufStart, BufPtr, BufEnd); + + // We *are* in raw mode. + LexingRawMode = true; +} + +/// Lexer constructor - Create a new raw lexer object. This object is only +/// suitable for calls to 'LexRawToken'. This lexer assumes that the text +/// range will outlive it, so it doesn't take ownership of it. +Lexer::Lexer(FileID FID, const SourceManager &SM, const LangOptions &features) + : FileLoc(SM.getLocForStartOfFile(FID)), Features(features) { + const llvm::MemoryBuffer *FromFile = SM.getBuffer(FID); + + InitLexer(FromFile->getBufferStart(), FromFile->getBufferStart(), + FromFile->getBufferEnd()); + + // We *are* in raw mode. + LexingRawMode = true; +} + +/// Create_PragmaLexer: Lexer constructor - Create a new lexer object for +/// _Pragma expansion. This has a variety of magic semantics that this method +/// sets up. It returns a new'd Lexer that must be delete'd when done. +/// +/// On entrance to this routine, TokStartLoc is a macro location which has a +/// spelling loc that indicates the bytes to be lexed for the token and an +/// instantiation location that indicates where all lexed tokens should be +/// "expanded from". +/// +/// FIXME: It would really be nice to make _Pragma just be a wrapper around a +/// normal lexer that remaps tokens as they fly by. This would require making +/// Preprocessor::Lex virtual. Given that, we could just dump in a magic lexer +/// interface that could handle this stuff. This would pull GetMappedTokenLoc +/// out of the critical path of the lexer! +/// +Lexer *Lexer::Create_PragmaLexer(SourceLocation SpellingLoc, + SourceLocation InstantiationLocStart, + SourceLocation InstantiationLocEnd, + unsigned TokLen, Preprocessor &PP) { + SourceManager &SM = PP.getSourceManager(); + + // Create the lexer as if we were going to lex the file normally. + FileID SpellingFID = SM.getFileID(SpellingLoc); + Lexer *L = new Lexer(SpellingFID, PP); + + // Now that the lexer is created, change the start/end locations so that we + // just lex the subsection of the file that we want. This is lexing from a + // scratch buffer. + const char *StrData = SM.getCharacterData(SpellingLoc); + + L->BufferPtr = StrData; + L->BufferEnd = StrData+TokLen; + assert(L->BufferEnd[0] == 0 && "Buffer is not nul terminated!"); + + // Set the SourceLocation with the remapping information. This ensures that + // GetMappedTokenLoc will remap the tokens as they are lexed. + L->FileLoc = SM.createInstantiationLoc(SM.getLocForStartOfFile(SpellingFID), + InstantiationLocStart, + InstantiationLocEnd, TokLen); + + // Ensure that the lexer thinks it is inside a directive, so that end \n will + // return an EOM token. + L->ParsingPreprocessorDirective = true; + + // This lexer really is for _Pragma. + L->Is_PragmaLexer = true; + return L; +} + + +/// Stringify - Convert the specified string into a C string, with surrounding +/// ""'s, and with escaped \ and " characters. +std::string Lexer::Stringify(const std::string &Str, bool Charify) { + std::string Result = Str; + char Quote = Charify ? '\'' : '"'; + for (unsigned i = 0, e = Result.size(); i != e; ++i) { + if (Result[i] == '\\' || Result[i] == Quote) { + Result.insert(Result.begin()+i, '\\'); + ++i; ++e; + } + } + return Result; +} + +/// Stringify - Convert the specified string into a C string by escaping '\' +/// and " characters. This does not add surrounding ""'s to the string. +void Lexer::Stringify(llvm::SmallVectorImpl<char> &Str) { + for (unsigned i = 0, e = Str.size(); i != e; ++i) { + if (Str[i] == '\\' || Str[i] == '"') { + Str.insert(Str.begin()+i, '\\'); + ++i; ++e; + } + } +} + + +/// MeasureTokenLength - Relex the token at the specified location and return +/// its length in bytes in the input file. If the token needs cleaning (e.g. +/// includes a trigraph or an escaped newline) then this count includes bytes +/// that are part of that. +unsigned Lexer::MeasureTokenLength(SourceLocation Loc, + const SourceManager &SM, + const LangOptions &LangOpts) { + // TODO: this could be special cased for common tokens like identifiers, ')', + // etc to make this faster, if it mattered. Just look at StrData[0] to handle + // all obviously single-char tokens. This could use + // Lexer::isObviouslySimpleCharacter for example to handle identifiers or + // something. + + // If this comes from a macro expansion, we really do want the macro name, not + // the token this macro expanded to. + Loc = SM.getInstantiationLoc(Loc); + std::pair<FileID, unsigned> LocInfo = SM.getDecomposedLoc(Loc); + std::pair<const char *,const char *> Buffer = SM.getBufferData(LocInfo.first); + const char *StrData = Buffer.first+LocInfo.second; + + // Create a lexer starting at the beginning of this token. + Lexer TheLexer(Loc, LangOpts, Buffer.first, StrData, Buffer.second); + Token TheTok; + TheLexer.LexFromRawLexer(TheTok); + return TheTok.getLength(); +} + +//===----------------------------------------------------------------------===// +// Character information. +//===----------------------------------------------------------------------===// + +static unsigned char CharInfo[256]; + +enum { + CHAR_HORZ_WS = 0x01, // ' ', '\t', '\f', '\v'. Note, no '\0' + CHAR_VERT_WS = 0x02, // '\r', '\n' + CHAR_LETTER = 0x04, // a-z,A-Z + CHAR_NUMBER = 0x08, // 0-9 + CHAR_UNDER = 0x10, // _ + CHAR_PERIOD = 0x20 // . +}; + +static void InitCharacterInfo() { + static bool isInited = false; + if (isInited) return; + isInited = true; + + // Intiialize the CharInfo table. + // TODO: statically initialize this. + CharInfo[(int)' '] = CharInfo[(int)'\t'] = + CharInfo[(int)'\f'] = CharInfo[(int)'\v'] = CHAR_HORZ_WS; + CharInfo[(int)'\n'] = CharInfo[(int)'\r'] = CHAR_VERT_WS; + + CharInfo[(int)'_'] = CHAR_UNDER; + CharInfo[(int)'.'] = CHAR_PERIOD; + for (unsigned i = 'a'; i <= 'z'; ++i) + CharInfo[i] = CharInfo[i+'A'-'a'] = CHAR_LETTER; + for (unsigned i = '0'; i <= '9'; ++i) + CharInfo[i] = CHAR_NUMBER; +} + +/// isIdentifierBody - Return true if this is the body character of an +/// identifier, which is [a-zA-Z0-9_]. +static inline bool isIdentifierBody(unsigned char c) { + return (CharInfo[c] & (CHAR_LETTER|CHAR_NUMBER|CHAR_UNDER)) ? true : false; +} + +/// isHorizontalWhitespace - Return true if this character is horizontal +/// whitespace: ' ', '\t', '\f', '\v'. Note that this returns false for '\0'. +static inline bool isHorizontalWhitespace(unsigned char c) { + return (CharInfo[c] & CHAR_HORZ_WS) ? true : false; +} + +/// isWhitespace - Return true if this character is horizontal or vertical +/// whitespace: ' ', '\t', '\f', '\v', '\n', '\r'. Note that this returns false +/// for '\0'. +static inline bool isWhitespace(unsigned char c) { + return (CharInfo[c] & (CHAR_HORZ_WS|CHAR_VERT_WS)) ? true : false; +} + +/// isNumberBody - Return true if this is the body character of an +/// preprocessing number, which is [a-zA-Z0-9_.]. +static inline bool isNumberBody(unsigned char c) { + return (CharInfo[c] & (CHAR_LETTER|CHAR_NUMBER|CHAR_UNDER|CHAR_PERIOD)) ? + true : false; +} + + +//===----------------------------------------------------------------------===// +// Diagnostics forwarding code. +//===----------------------------------------------------------------------===// + +/// GetMappedTokenLoc - If lexing out of a 'mapped buffer', where we pretend the +/// lexer buffer was all instantiated at a single point, perform the mapping. +/// This is currently only used for _Pragma implementation, so it is the slow +/// path of the hot getSourceLocation method. Do not allow it to be inlined. +static SourceLocation GetMappedTokenLoc(Preprocessor &PP, + SourceLocation FileLoc, + unsigned CharNo, + unsigned TokLen) DISABLE_INLINE; +static SourceLocation GetMappedTokenLoc(Preprocessor &PP, + SourceLocation FileLoc, + unsigned CharNo, unsigned TokLen) { + assert(FileLoc.isMacroID() && "Must be an instantiation"); + + // Otherwise, we're lexing "mapped tokens". This is used for things like + // _Pragma handling. Combine the instantiation location of FileLoc with the + // spelling location. + SourceManager &SM = PP.getSourceManager(); + + // Create a new SLoc which is expanded from Instantiation(FileLoc) but whose + // characters come from spelling(FileLoc)+Offset. + SourceLocation SpellingLoc = SM.getSpellingLoc(FileLoc); + SpellingLoc = SpellingLoc.getFileLocWithOffset(CharNo); + + // Figure out the expansion loc range, which is the range covered by the + // original _Pragma(...) sequence. + std::pair<SourceLocation,SourceLocation> II = + SM.getImmediateInstantiationRange(FileLoc); + + return SM.createInstantiationLoc(SpellingLoc, II.first, II.second, TokLen); +} + +/// getSourceLocation - Return a source location identifier for the specified +/// offset in the current file. +SourceLocation Lexer::getSourceLocation(const char *Loc, + unsigned TokLen) const { + assert(Loc >= BufferStart && Loc <= BufferEnd && + "Location out of range for this buffer!"); + + // In the normal case, we're just lexing from a simple file buffer, return + // the file id from FileLoc with the offset specified. + unsigned CharNo = Loc-BufferStart; + if (FileLoc.isFileID()) + return FileLoc.getFileLocWithOffset(CharNo); + + // Otherwise, this is the _Pragma lexer case, which pretends that all of the + // tokens are lexed from where the _Pragma was defined. + assert(PP && "This doesn't work on raw lexers"); + return GetMappedTokenLoc(*PP, FileLoc, CharNo, TokLen); +} + +/// Diag - Forwarding function for diagnostics. This translate a source +/// position in the current buffer into a SourceLocation object for rendering. +DiagnosticBuilder Lexer::Diag(const char *Loc, unsigned DiagID) const { + return PP->Diag(getSourceLocation(Loc), DiagID); +} + +//===----------------------------------------------------------------------===// +// Trigraph and Escaped Newline Handling Code. +//===----------------------------------------------------------------------===// + +/// GetTrigraphCharForLetter - Given a character that occurs after a ?? pair, +/// return the decoded trigraph letter it corresponds to, or '\0' if nothing. +static char GetTrigraphCharForLetter(char Letter) { + switch (Letter) { + default: return 0; + case '=': return '#'; + case ')': return ']'; + case '(': return '['; + case '!': return '|'; + case '\'': return '^'; + case '>': return '}'; + case '/': return '\\'; + case '<': return '{'; + case '-': return '~'; + } +} + +/// DecodeTrigraphChar - If the specified character is a legal trigraph when +/// prefixed with ??, emit a trigraph warning. If trigraphs are enabled, +/// return the result character. Finally, emit a warning about trigraph use +/// whether trigraphs are enabled or not. +static char DecodeTrigraphChar(const char *CP, Lexer *L) { + char Res = GetTrigraphCharForLetter(*CP); + if (!Res || !L) return Res; + + if (!L->getFeatures().Trigraphs) { + if (!L->isLexingRawMode()) + L->Diag(CP-2, diag::trigraph_ignored); + return 0; + } + + if (!L->isLexingRawMode()) + L->Diag(CP-2, diag::trigraph_converted) << std::string()+Res; + return Res; +} + +/// getEscapedNewLineSize - Return the size of the specified escaped newline, +/// or 0 if it is not an escaped newline. P[-1] is known to be a "\" or a +/// trigraph equivalent on entry to this function. +unsigned Lexer::getEscapedNewLineSize(const char *Ptr) { + unsigned Size = 0; + while (isWhitespace(Ptr[Size])) { + ++Size; + + if (Ptr[Size-1] != '\n' && Ptr[Size-1] != '\r') + continue; + + // If this is a \r\n or \n\r, skip the other half. + if ((Ptr[Size] == '\r' || Ptr[Size] == '\n') && + Ptr[Size-1] != Ptr[Size]) + ++Size; + + return Size; + } + + // Not an escaped newline, must be a \t or something else. + return 0; +} + +/// SkipEscapedNewLines - If P points to an escaped newline (or a series of +/// them), skip over them and return the first non-escaped-newline found, +/// otherwise return P. +const char *Lexer::SkipEscapedNewLines(const char *P) { + while (1) { + const char *AfterEscape; + if (*P == '\\') { + AfterEscape = P+1; + } else if (*P == '?') { + // If not a trigraph for escape, bail out. + if (P[1] != '?' || P[2] != '/') + return P; + AfterEscape = P+3; + } else { + return P; + } + + unsigned NewLineSize = Lexer::getEscapedNewLineSize(AfterEscape); + if (NewLineSize == 0) return P; + P = AfterEscape+NewLineSize; + } +} + + +/// getCharAndSizeSlow - Peek a single 'character' from the specified buffer, +/// get its size, and return it. This is tricky in several cases: +/// 1. If currently at the start of a trigraph, we warn about the trigraph, +/// then either return the trigraph (skipping 3 chars) or the '?', +/// depending on whether trigraphs are enabled or not. +/// 2. If this is an escaped newline (potentially with whitespace between +/// the backslash and newline), implicitly skip the newline and return +/// the char after it. +/// 3. If this is a UCN, return it. FIXME: C++ UCN's? +/// +/// This handles the slow/uncommon case of the getCharAndSize method. Here we +/// know that we can accumulate into Size, and that we have already incremented +/// Ptr by Size bytes. +/// +/// NOTE: When this method is updated, getCharAndSizeSlowNoWarn (below) should +/// be updated to match. +/// +char Lexer::getCharAndSizeSlow(const char *Ptr, unsigned &Size, + Token *Tok) { + // If we have a slash, look for an escaped newline. + if (Ptr[0] == '\\') { + ++Size; + ++Ptr; +Slash: + // Common case, backslash-char where the char is not whitespace. + if (!isWhitespace(Ptr[0])) return '\\'; + + // See if we have optional whitespace characters followed by a newline. + if (unsigned EscapedNewLineSize = getEscapedNewLineSize(Ptr)) { + // Remember that this token needs to be cleaned. + if (Tok) Tok->setFlag(Token::NeedsCleaning); + + // Warn if there was whitespace between the backslash and newline. + if (EscapedNewLineSize != 1 && Tok && !isLexingRawMode()) + Diag(Ptr, diag::backslash_newline_space); + + // Found backslash<whitespace><newline>. Parse the char after it. + Size += EscapedNewLineSize; + Ptr += EscapedNewLineSize; + // Use slow version to accumulate a correct size field. + return getCharAndSizeSlow(Ptr, Size, Tok); + } + + // Otherwise, this is not an escaped newline, just return the slash. + return '\\'; + } + + // If this is a trigraph, process it. + if (Ptr[0] == '?' && Ptr[1] == '?') { + // If this is actually a legal trigraph (not something like "??x"), emit + // a trigraph warning. If so, and if trigraphs are enabled, return it. + if (char C = DecodeTrigraphChar(Ptr+2, Tok ? this : 0)) { + // Remember that this token needs to be cleaned. + if (Tok) Tok->setFlag(Token::NeedsCleaning); + + Ptr += 3; + Size += 3; + if (C == '\\') goto Slash; + return C; + } + } + + // If this is neither, return a single character. + ++Size; + return *Ptr; +} + + +/// getCharAndSizeSlowNoWarn - Handle the slow/uncommon case of the +/// getCharAndSizeNoWarn method. Here we know that we can accumulate into Size, +/// and that we have already incremented Ptr by Size bytes. +/// +/// NOTE: When this method is updated, getCharAndSizeSlow (above) should +/// be updated to match. +char Lexer::getCharAndSizeSlowNoWarn(const char *Ptr, unsigned &Size, + const LangOptions &Features) { + // If we have a slash, look for an escaped newline. + if (Ptr[0] == '\\') { + ++Size; + ++Ptr; +Slash: + // Common case, backslash-char where the char is not whitespace. + if (!isWhitespace(Ptr[0])) return '\\'; + + // See if we have optional whitespace characters followed by a newline. + if (unsigned EscapedNewLineSize = getEscapedNewLineSize(Ptr)) { + // Found backslash<whitespace><newline>. Parse the char after it. + Size += EscapedNewLineSize; + Ptr += EscapedNewLineSize; + + // Use slow version to accumulate a correct size field. + return getCharAndSizeSlowNoWarn(Ptr, Size, Features); + } + + // Otherwise, this is not an escaped newline, just return the slash. + return '\\'; + } + + // If this is a trigraph, process it. + if (Features.Trigraphs && Ptr[0] == '?' && Ptr[1] == '?') { + // If this is actually a legal trigraph (not something like "??x"), return + // it. + if (char C = GetTrigraphCharForLetter(Ptr[2])) { + Ptr += 3; + Size += 3; + if (C == '\\') goto Slash; + return C; + } + } + + // If this is neither, return a single character. + ++Size; + return *Ptr; +} + +//===----------------------------------------------------------------------===// +// Helper methods for lexing. +//===----------------------------------------------------------------------===// + +void Lexer::LexIdentifier(Token &Result, const char *CurPtr) { + // Match [_A-Za-z0-9]*, we have already matched [_A-Za-z$] + unsigned Size; + unsigned char C = *CurPtr++; + while (isIdentifierBody(C)) { + C = *CurPtr++; + } + --CurPtr; // Back up over the skipped character. + + // Fast path, no $,\,? in identifier found. '\' might be an escaped newline + // or UCN, and ? might be a trigraph for '\', an escaped newline or UCN. + // FIXME: UCNs. + if (C != '\\' && C != '?' && (C != '$' || !Features.DollarIdents)) { +FinishIdentifier: + const char *IdStart = BufferPtr; + FormTokenWithChars(Result, CurPtr, tok::identifier); + + // If we are in raw mode, return this identifier raw. There is no need to + // look up identifier information or attempt to macro expand it. + if (LexingRawMode) return; + + // Fill in Result.IdentifierInfo, looking up the identifier in the + // identifier table. + IdentifierInfo *II = PP->LookUpIdentifierInfo(Result, IdStart); + + // Change the kind of this identifier to the appropriate token kind, e.g. + // turning "for" into a keyword. + Result.setKind(II->getTokenID()); + + // Finally, now that we know we have an identifier, pass this off to the + // preprocessor, which may macro expand it or something. + if (II->isHandleIdentifierCase()) + PP->HandleIdentifier(Result); + return; + } + + // Otherwise, $,\,? in identifier found. Enter slower path. + + C = getCharAndSize(CurPtr, Size); + while (1) { + if (C == '$') { + // If we hit a $ and they are not supported in identifiers, we are done. + if (!Features.DollarIdents) goto FinishIdentifier; + + // Otherwise, emit a diagnostic and continue. + if (!isLexingRawMode()) + Diag(CurPtr, diag::ext_dollar_in_identifier); + CurPtr = ConsumeChar(CurPtr, Size, Result); + C = getCharAndSize(CurPtr, Size); + continue; + } else if (!isIdentifierBody(C)) { // FIXME: UCNs. + // Found end of identifier. + goto FinishIdentifier; + } + + // Otherwise, this character is good, consume it. + CurPtr = ConsumeChar(CurPtr, Size, Result); + + C = getCharAndSize(CurPtr, Size); + while (isIdentifierBody(C)) { // FIXME: UCNs. + CurPtr = ConsumeChar(CurPtr, Size, Result); + C = getCharAndSize(CurPtr, Size); + } + } +} + + +/// LexNumericConstant - Lex the remainder of a integer or floating point +/// constant. From[-1] is the first character lexed. Return the end of the +/// constant. +void Lexer::LexNumericConstant(Token &Result, const char *CurPtr) { + unsigned Size; + char C = getCharAndSize(CurPtr, Size); + char PrevCh = 0; + while (isNumberBody(C)) { // FIXME: UCNs? + CurPtr = ConsumeChar(CurPtr, Size, Result); + PrevCh = C; + C = getCharAndSize(CurPtr, Size); + } + + // If we fell out, check for a sign, due to 1e+12. If we have one, continue. + if ((C == '-' || C == '+') && (PrevCh == 'E' || PrevCh == 'e')) + return LexNumericConstant(Result, ConsumeChar(CurPtr, Size, Result)); + + // If we have a hex FP constant, continue. + if ((C == '-' || C == '+') && (PrevCh == 'P' || PrevCh == 'p')) + return LexNumericConstant(Result, ConsumeChar(CurPtr, Size, Result)); + + // Update the location of token as well as BufferPtr. + const char *TokStart = BufferPtr; + FormTokenWithChars(Result, CurPtr, tok::numeric_constant); + Result.setLiteralData(TokStart); +} + +/// LexStringLiteral - Lex the remainder of a string literal, after having lexed +/// either " or L". +void Lexer::LexStringLiteral(Token &Result, const char *CurPtr, bool Wide) { + const char *NulCharacter = 0; // Does this string contain the \0 character? + + char C = getAndAdvanceChar(CurPtr, Result); + while (C != '"') { + // Skip escaped characters. + if (C == '\\') { + // Skip the escaped character. + C = getAndAdvanceChar(CurPtr, Result); + } else if (C == '\n' || C == '\r' || // Newline. + (C == 0 && CurPtr-1 == BufferEnd)) { // End of file. + if (!isLexingRawMode() && !Features.AsmPreprocessor) + Diag(BufferPtr, diag::err_unterminated_string); + FormTokenWithChars(Result, CurPtr-1, tok::unknown); + return; + } else if (C == 0) { + NulCharacter = CurPtr-1; + } + C = getAndAdvanceChar(CurPtr, Result); + } + + // If a nul character existed in the string, warn about it. + if (NulCharacter && !isLexingRawMode()) + Diag(NulCharacter, diag::null_in_string); + + // Update the location of the token as well as the BufferPtr instance var. + const char *TokStart = BufferPtr; + FormTokenWithChars(Result, CurPtr, + Wide ? tok::wide_string_literal : tok::string_literal); + Result.setLiteralData(TokStart); +} + +/// LexAngledStringLiteral - Lex the remainder of an angled string literal, +/// after having lexed the '<' character. This is used for #include filenames. +void Lexer::LexAngledStringLiteral(Token &Result, const char *CurPtr) { + const char *NulCharacter = 0; // Does this string contain the \0 character? + const char *AfterLessPos = CurPtr; + char C = getAndAdvanceChar(CurPtr, Result); + while (C != '>') { + // Skip escaped characters. + if (C == '\\') { + // Skip the escaped character. + C = getAndAdvanceChar(CurPtr, Result); + } else if (C == '\n' || C == '\r' || // Newline. + (C == 0 && CurPtr-1 == BufferEnd)) { // End of file. + // If the filename is unterminated, then it must just be a lone < + // character. Return this as such. + FormTokenWithChars(Result, AfterLessPos, tok::less); + return; + } else if (C == 0) { + NulCharacter = CurPtr-1; + } + C = getAndAdvanceChar(CurPtr, Result); + } + + // If a nul character existed in the string, warn about it. + if (NulCharacter && !isLexingRawMode()) + Diag(NulCharacter, diag::null_in_string); + + // Update the location of token as well as BufferPtr. + const char *TokStart = BufferPtr; + FormTokenWithChars(Result, CurPtr, tok::angle_string_literal); + Result.setLiteralData(TokStart); +} + + +/// LexCharConstant - Lex the remainder of a character constant, after having +/// lexed either ' or L'. +void Lexer::LexCharConstant(Token &Result, const char *CurPtr) { + const char *NulCharacter = 0; // Does this character contain the \0 character? + + // Handle the common case of 'x' and '\y' efficiently. + char C = getAndAdvanceChar(CurPtr, Result); + if (C == '\'') { + if (!isLexingRawMode() && !Features.AsmPreprocessor) + Diag(BufferPtr, diag::err_empty_character); + FormTokenWithChars(Result, CurPtr, tok::unknown); + return; + } else if (C == '\\') { + // Skip the escaped character. + // FIXME: UCN's. + C = getAndAdvanceChar(CurPtr, Result); + } + + if (C && C != '\n' && C != '\r' && CurPtr[0] == '\'') { + ++CurPtr; + } else { + // Fall back on generic code for embedded nulls, newlines, wide chars. + do { + // Skip escaped characters. + if (C == '\\') { + // Skip the escaped character. + C = getAndAdvanceChar(CurPtr, Result); + } else if (C == '\n' || C == '\r' || // Newline. + (C == 0 && CurPtr-1 == BufferEnd)) { // End of file. + if (!isLexingRawMode() && !Features.AsmPreprocessor) + Diag(BufferPtr, diag::err_unterminated_char); + FormTokenWithChars(Result, CurPtr-1, tok::unknown); + return; + } else if (C == 0) { + NulCharacter = CurPtr-1; + } + C = getAndAdvanceChar(CurPtr, Result); + } while (C != '\''); + } + + if (NulCharacter && !isLexingRawMode()) + Diag(NulCharacter, diag::null_in_char); + + // Update the location of token as well as BufferPtr. + const char *TokStart = BufferPtr; + FormTokenWithChars(Result, CurPtr, tok::char_constant); + Result.setLiteralData(TokStart); +} + +/// SkipWhitespace - Efficiently skip over a series of whitespace characters. +/// Update BufferPtr to point to the next non-whitespace character and return. +/// +/// This method forms a token and returns true if KeepWhitespaceMode is enabled. +/// +bool Lexer::SkipWhitespace(Token &Result, const char *CurPtr) { + // Whitespace - Skip it, then return the token after the whitespace. + unsigned char Char = *CurPtr; // Skip consequtive spaces efficiently. + while (1) { + // Skip horizontal whitespace very aggressively. + while (isHorizontalWhitespace(Char)) + Char = *++CurPtr; + + // Otherwise if we have something other than whitespace, we're done. + if (Char != '\n' && Char != '\r') + break; + + if (ParsingPreprocessorDirective) { + // End of preprocessor directive line, let LexTokenInternal handle this. + BufferPtr = CurPtr; + return false; + } + + // ok, but handle newline. + // The returned token is at the start of the line. + Result.setFlag(Token::StartOfLine); + // No leading whitespace seen so far. + Result.clearFlag(Token::LeadingSpace); + Char = *++CurPtr; + } + + // If this isn't immediately after a newline, there is leading space. + char PrevChar = CurPtr[-1]; + if (PrevChar != '\n' && PrevChar != '\r') + Result.setFlag(Token::LeadingSpace); + + // If the client wants us to return whitespace, return it now. + if (isKeepWhitespaceMode()) { + FormTokenWithChars(Result, CurPtr, tok::unknown); + return true; + } + + BufferPtr = CurPtr; + return false; +} + +// SkipBCPLComment - We have just read the // characters from input. Skip until +// we find the newline character thats terminate the comment. Then update +/// BufferPtr and return. If we're in KeepCommentMode, this will form the token +/// and return true. +bool Lexer::SkipBCPLComment(Token &Result, const char *CurPtr) { + // If BCPL comments aren't explicitly enabled for this language, emit an + // extension warning. + if (!Features.BCPLComment && !isLexingRawMode()) { + Diag(BufferPtr, diag::ext_bcpl_comment); + + // Mark them enabled so we only emit one warning for this translation + // unit. + Features.BCPLComment = true; + } + + // Scan over the body of the comment. The common case, when scanning, is that + // the comment contains normal ascii characters with nothing interesting in + // them. As such, optimize for this case with the inner loop. + char C; + do { + C = *CurPtr; + // FIXME: Speedup BCPL comment lexing. Just scan for a \n or \r character. + // If we find a \n character, scan backwards, checking to see if it's an + // escaped newline, like we do for block comments. + + // Skip over characters in the fast loop. + while (C != 0 && // Potentially EOF. + C != '\\' && // Potentially escaped newline. + C != '?' && // Potentially trigraph. + C != '\n' && C != '\r') // Newline or DOS-style newline. + C = *++CurPtr; + + // If this is a newline, we're done. + if (C == '\n' || C == '\r') + break; // Found the newline? Break out! + + // Otherwise, this is a hard case. Fall back on getAndAdvanceChar to + // properly decode the character. Read it in raw mode to avoid emitting + // diagnostics about things like trigraphs. If we see an escaped newline, + // we'll handle it below. + const char *OldPtr = CurPtr; + bool OldRawMode = isLexingRawMode(); + LexingRawMode = true; + C = getAndAdvanceChar(CurPtr, Result); + LexingRawMode = OldRawMode; + + // If the char that we finally got was a \n, then we must have had something + // like \<newline><newline>. We don't want to have consumed the second + // newline, we want CurPtr, to end up pointing to it down below. + if (C == '\n' || C == '\r') { + --CurPtr; + C = 'x'; // doesn't matter what this is. + } + + // If we read multiple characters, and one of those characters was a \r or + // \n, then we had an escaped newline within the comment. Emit diagnostic + // unless the next line is also a // comment. + if (CurPtr != OldPtr+1 && C != '/' && CurPtr[0] != '/') { + for (; OldPtr != CurPtr; ++OldPtr) + if (OldPtr[0] == '\n' || OldPtr[0] == '\r') { + // Okay, we found a // comment that ends in a newline, if the next + // line is also a // comment, but has spaces, don't emit a diagnostic. + if (isspace(C)) { + const char *ForwardPtr = CurPtr; + while (isspace(*ForwardPtr)) // Skip whitespace. + ++ForwardPtr; + if (ForwardPtr[0] == '/' && ForwardPtr[1] == '/') + break; + } + + if (!isLexingRawMode()) + Diag(OldPtr-1, diag::ext_multi_line_bcpl_comment); + break; + } + } + + if (CurPtr == BufferEnd+1) { --CurPtr; break; } + } while (C != '\n' && C != '\r'); + + // Found but did not consume the newline. + + // If we are returning comments as tokens, return this comment as a token. + if (inKeepCommentMode()) + return SaveBCPLComment(Result, CurPtr); + + // If we are inside a preprocessor directive and we see the end of line, + // return immediately, so that the lexer can return this as an EOM token. + if (ParsingPreprocessorDirective || CurPtr == BufferEnd) { + BufferPtr = CurPtr; + return false; + } + + // Otherwise, eat the \n character. We don't care if this is a \n\r or + // \r\n sequence. This is an efficiency hack (because we know the \n can't + // contribute to another token), it isn't needed for correctness. Note that + // this is ok even in KeepWhitespaceMode, because we would have returned the + /// comment above in that mode. + ++CurPtr; + + // The next returned token is at the start of the line. + Result.setFlag(Token::StartOfLine); + // No leading whitespace seen so far. + Result.clearFlag(Token::LeadingSpace); + BufferPtr = CurPtr; + return false; +} + +/// SaveBCPLComment - If in save-comment mode, package up this BCPL comment in +/// an appropriate way and return it. +bool Lexer::SaveBCPLComment(Token &Result, const char *CurPtr) { + // If we're not in a preprocessor directive, just return the // comment + // directly. + FormTokenWithChars(Result, CurPtr, tok::comment); + + if (!ParsingPreprocessorDirective) + return true; + + // If this BCPL-style comment is in a macro definition, transmogrify it into + // a C-style block comment. + std::string Spelling = PP->getSpelling(Result); + assert(Spelling[0] == '/' && Spelling[1] == '/' && "Not bcpl comment?"); + Spelling[1] = '*'; // Change prefix to "/*". + Spelling += "*/"; // add suffix. + + Result.setKind(tok::comment); + PP->CreateString(&Spelling[0], Spelling.size(), Result, + Result.getLocation()); + return true; +} + +/// isBlockCommentEndOfEscapedNewLine - Return true if the specified newline +/// character (either \n or \r) is part of an escaped newline sequence. Issue a +/// diagnostic if so. We know that the newline is inside of a block comment. +static bool isEndOfBlockCommentWithEscapedNewLine(const char *CurPtr, + Lexer *L) { + assert(CurPtr[0] == '\n' || CurPtr[0] == '\r'); + + // Back up off the newline. + --CurPtr; + + // If this is a two-character newline sequence, skip the other character. + if (CurPtr[0] == '\n' || CurPtr[0] == '\r') { + // \n\n or \r\r -> not escaped newline. + if (CurPtr[0] == CurPtr[1]) + return false; + // \n\r or \r\n -> skip the newline. + --CurPtr; + } + + // If we have horizontal whitespace, skip over it. We allow whitespace + // between the slash and newline. + bool HasSpace = false; + while (isHorizontalWhitespace(*CurPtr) || *CurPtr == 0) { + --CurPtr; + HasSpace = true; + } + + // If we have a slash, we know this is an escaped newline. + if (*CurPtr == '\\') { + if (CurPtr[-1] != '*') return false; + } else { + // It isn't a slash, is it the ?? / trigraph? + if (CurPtr[0] != '/' || CurPtr[-1] != '?' || CurPtr[-2] != '?' || + CurPtr[-3] != '*') + return false; + + // This is the trigraph ending the comment. Emit a stern warning! + CurPtr -= 2; + + // If no trigraphs are enabled, warn that we ignored this trigraph and + // ignore this * character. + if (!L->getFeatures().Trigraphs) { + if (!L->isLexingRawMode()) + L->Diag(CurPtr, diag::trigraph_ignored_block_comment); + return false; + } + if (!L->isLexingRawMode()) + L->Diag(CurPtr, diag::trigraph_ends_block_comment); + } + + // Warn about having an escaped newline between the */ characters. + if (!L->isLexingRawMode()) + L->Diag(CurPtr, diag::escaped_newline_block_comment_end); + + // If there was space between the backslash and newline, warn about it. + if (HasSpace && !L->isLexingRawMode()) + L->Diag(CurPtr, diag::backslash_newline_space); + + return true; +} + +#ifdef __SSE2__ +#include <emmintrin.h> +#elif __ALTIVEC__ +#include <altivec.h> +#undef bool +#endif + +/// SkipBlockComment - We have just read the /* characters from input. Read +/// until we find the */ characters that terminate the comment. Note that we +/// don't bother decoding trigraphs or escaped newlines in block comments, +/// because they cannot cause the comment to end. The only thing that can +/// happen is the comment could end with an escaped newline between the */ end +/// of comment. +/// +/// If KeepCommentMode is enabled, this forms a token from the comment and +/// returns true. +bool Lexer::SkipBlockComment(Token &Result, const char *CurPtr) { + // Scan one character past where we should, looking for a '/' character. Once + // we find it, check to see if it was preceeded by a *. This common + // optimization helps people who like to put a lot of * characters in their + // comments. + + // The first character we get with newlines and trigraphs skipped to handle + // the degenerate /*/ case below correctly if the * has an escaped newline + // after it. + unsigned CharSize; + unsigned char C = getCharAndSize(CurPtr, CharSize); + CurPtr += CharSize; + if (C == 0 && CurPtr == BufferEnd+1) { + if (!isLexingRawMode()) + Diag(BufferPtr, diag::err_unterminated_block_comment); + --CurPtr; + + // KeepWhitespaceMode should return this broken comment as a token. Since + // it isn't a well formed comment, just return it as an 'unknown' token. + if (isKeepWhitespaceMode()) { + FormTokenWithChars(Result, CurPtr, tok::unknown); + return true; + } + + BufferPtr = CurPtr; + return false; + } + + // Check to see if the first character after the '/*' is another /. If so, + // then this slash does not end the block comment, it is part of it. + if (C == '/') + C = *CurPtr++; + + while (1) { + // Skip over all non-interesting characters until we find end of buffer or a + // (probably ending) '/' character. + if (CurPtr + 24 < BufferEnd) { + // While not aligned to a 16-byte boundary. + while (C != '/' && ((intptr_t)CurPtr & 0x0F) != 0) + C = *CurPtr++; + + if (C == '/') goto FoundSlash; + +#ifdef __SSE2__ + __m128i Slashes = _mm_set_epi8('/', '/', '/', '/', '/', '/', '/', '/', + '/', '/', '/', '/', '/', '/', '/', '/'); + while (CurPtr+16 <= BufferEnd && + _mm_movemask_epi8(_mm_cmpeq_epi8(*(__m128i*)CurPtr, Slashes)) == 0) + CurPtr += 16; +#elif __ALTIVEC__ + __vector unsigned char Slashes = { + '/', '/', '/', '/', '/', '/', '/', '/', + '/', '/', '/', '/', '/', '/', '/', '/' + }; + while (CurPtr+16 <= BufferEnd && + !vec_any_eq(*(vector unsigned char*)CurPtr, Slashes)) + CurPtr += 16; +#else + // Scan for '/' quickly. Many block comments are very large. + while (CurPtr[0] != '/' && + CurPtr[1] != '/' && + CurPtr[2] != '/' && + CurPtr[3] != '/' && + CurPtr+4 < BufferEnd) { + CurPtr += 4; + } +#endif + + // It has to be one of the bytes scanned, increment to it and read one. + C = *CurPtr++; + } + + // Loop to scan the remainder. + while (C != '/' && C != '\0') + C = *CurPtr++; + + FoundSlash: + if (C == '/') { + if (CurPtr[-2] == '*') // We found the final */. We're done! + break; + + if ((CurPtr[-2] == '\n' || CurPtr[-2] == '\r')) { + if (isEndOfBlockCommentWithEscapedNewLine(CurPtr-2, this)) { + // We found the final */, though it had an escaped newline between the + // * and /. We're done! + break; + } + } + if (CurPtr[0] == '*' && CurPtr[1] != '/') { + // If this is a /* inside of the comment, emit a warning. Don't do this + // if this is a /*/, which will end the comment. This misses cases with + // embedded escaped newlines, but oh well. + if (!isLexingRawMode()) + Diag(CurPtr-1, diag::warn_nested_block_comment); + } + } else if (C == 0 && CurPtr == BufferEnd+1) { + if (!isLexingRawMode()) + Diag(BufferPtr, diag::err_unterminated_block_comment); + // Note: the user probably forgot a */. We could continue immediately + // after the /*, but this would involve lexing a lot of what really is the + // comment, which surely would confuse the parser. + --CurPtr; + + // KeepWhitespaceMode should return this broken comment as a token. Since + // it isn't a well formed comment, just return it as an 'unknown' token. + if (isKeepWhitespaceMode()) { + FormTokenWithChars(Result, CurPtr, tok::unknown); + return true; + } + + BufferPtr = CurPtr; + return false; + } + C = *CurPtr++; + } + + // If we are returning comments as tokens, return this comment as a token. + if (inKeepCommentMode()) { + FormTokenWithChars(Result, CurPtr, tok::comment); + return true; + } + + // It is common for the tokens immediately after a /**/ comment to be + // whitespace. Instead of going through the big switch, handle it + // efficiently now. This is safe even in KeepWhitespaceMode because we would + // have already returned above with the comment as a token. + if (isHorizontalWhitespace(*CurPtr)) { + Result.setFlag(Token::LeadingSpace); + SkipWhitespace(Result, CurPtr+1); + return false; + } + + // Otherwise, just return so that the next character will be lexed as a token. + BufferPtr = CurPtr; + Result.setFlag(Token::LeadingSpace); + return false; +} + +//===----------------------------------------------------------------------===// +// Primary Lexing Entry Points +//===----------------------------------------------------------------------===// + +/// ReadToEndOfLine - Read the rest of the current preprocessor line as an +/// uninterpreted string. This switches the lexer out of directive mode. +std::string Lexer::ReadToEndOfLine() { + assert(ParsingPreprocessorDirective && ParsingFilename == false && + "Must be in a preprocessing directive!"); + std::string Result; + Token Tmp; + + // CurPtr - Cache BufferPtr in an automatic variable. + const char *CurPtr = BufferPtr; + while (1) { + char Char = getAndAdvanceChar(CurPtr, Tmp); + switch (Char) { + default: + Result += Char; + break; + case 0: // Null. + // Found end of file? + if (CurPtr-1 != BufferEnd) { + // Nope, normal character, continue. + Result += Char; + break; + } + // FALL THROUGH. + case '\r': + case '\n': + // Okay, we found the end of the line. First, back up past the \0, \r, \n. + assert(CurPtr[-1] == Char && "Trigraphs for newline?"); + BufferPtr = CurPtr-1; + + // Next, lex the character, which should handle the EOM transition. + Lex(Tmp); + assert(Tmp.is(tok::eom) && "Unexpected token!"); + + // Finally, we're done, return the string we found. + return Result; + } + } +} + +/// LexEndOfFile - CurPtr points to the end of this file. Handle this +/// condition, reporting diagnostics and handling other edge cases as required. +/// This returns true if Result contains a token, false if PP.Lex should be +/// called again. +bool Lexer::LexEndOfFile(Token &Result, const char *CurPtr) { + // If we hit the end of the file while parsing a preprocessor directive, + // end the preprocessor directive first. The next token returned will + // then be the end of file. + if (ParsingPreprocessorDirective) { + // Done parsing the "line". + ParsingPreprocessorDirective = false; + // Update the location of token as well as BufferPtr. + FormTokenWithChars(Result, CurPtr, tok::eom); + + // Restore comment saving mode, in case it was disabled for directive. + SetCommentRetentionState(PP->getCommentRetentionState()); + return true; // Have a token. + } + + // If we are in raw mode, return this event as an EOF token. Let the caller + // that put us in raw mode handle the event. + if (isLexingRawMode()) { + Result.startToken(); + BufferPtr = BufferEnd; + FormTokenWithChars(Result, BufferEnd, tok::eof); + return true; + } + + // Otherwise, issue diagnostics for unterminated #if and missing newline. + + // If we are in a #if directive, emit an error. + while (!ConditionalStack.empty()) { + PP->Diag(ConditionalStack.back().IfLoc, + diag::err_pp_unterminated_conditional); + ConditionalStack.pop_back(); + } + + // C99 5.1.1.2p2: If the file is non-empty and didn't end in a newline, issue + // a pedwarn. + if (CurPtr != BufferStart && (CurPtr[-1] != '\n' && CurPtr[-1] != '\r')) + Diag(BufferEnd, diag::ext_no_newline_eof) + << CodeModificationHint::CreateInsertion(getSourceLocation(BufferEnd), + "\n"); + + BufferPtr = CurPtr; + + // Finally, let the preprocessor handle this. + return PP->HandleEndOfFile(Result); +} + +/// isNextPPTokenLParen - Return 1 if the next unexpanded token lexed from +/// the specified lexer will return a tok::l_paren token, 0 if it is something +/// else and 2 if there are no more tokens in the buffer controlled by the +/// lexer. +unsigned Lexer::isNextPPTokenLParen() { + assert(!LexingRawMode && "How can we expand a macro from a skipping buffer?"); + + // Switch to 'skipping' mode. This will ensure that we can lex a token + // without emitting diagnostics, disables macro expansion, and will cause EOF + // to return an EOF token instead of popping the include stack. + LexingRawMode = true; + + // Save state that can be changed while lexing so that we can restore it. + const char *TmpBufferPtr = BufferPtr; + bool inPPDirectiveMode = ParsingPreprocessorDirective; + + Token Tok; + Tok.startToken(); + LexTokenInternal(Tok); + + // Restore state that may have changed. + BufferPtr = TmpBufferPtr; + ParsingPreprocessorDirective = inPPDirectiveMode; + + // Restore the lexer back to non-skipping mode. + LexingRawMode = false; + + if (Tok.is(tok::eof)) + return 2; + return Tok.is(tok::l_paren); +} + + +/// LexTokenInternal - This implements a simple C family lexer. It is an +/// extremely performance critical piece of code. This assumes that the buffer +/// has a null character at the end of the file. Return true if an error +/// occurred and compilation should terminate, false if normal. This returns a +/// preprocessing token, not a normal token, as such, it is an internal +/// interface. It assumes that the Flags of result have been cleared before +/// calling this. +void Lexer::LexTokenInternal(Token &Result) { +LexNextToken: + // New token, can't need cleaning yet. + Result.clearFlag(Token::NeedsCleaning); + Result.setIdentifierInfo(0); + + // CurPtr - Cache BufferPtr in an automatic variable. + const char *CurPtr = BufferPtr; + + // Small amounts of horizontal whitespace is very common between tokens. + if ((*CurPtr == ' ') || (*CurPtr == '\t')) { + ++CurPtr; + while ((*CurPtr == ' ') || (*CurPtr == '\t')) + ++CurPtr; + + // If we are keeping whitespace and other tokens, just return what we just + // skipped. The next lexer invocation will return the token after the + // whitespace. + if (isKeepWhitespaceMode()) { + FormTokenWithChars(Result, CurPtr, tok::unknown); + return; + } + + BufferPtr = CurPtr; + Result.setFlag(Token::LeadingSpace); + } + + unsigned SizeTmp, SizeTmp2; // Temporaries for use in cases below. + + // Read a character, advancing over it. + char Char = getAndAdvanceChar(CurPtr, Result); + tok::TokenKind Kind; + + switch (Char) { + case 0: // Null. + // Found end of file? + if (CurPtr-1 == BufferEnd) { + // Read the PP instance variable into an automatic variable, because + // LexEndOfFile will often delete 'this'. + Preprocessor *PPCache = PP; + if (LexEndOfFile(Result, CurPtr-1)) // Retreat back into the file. + return; // Got a token to return. + assert(PPCache && "Raw buffer::LexEndOfFile should return a token"); + return PPCache->Lex(Result); + } + + if (!isLexingRawMode()) + Diag(CurPtr-1, diag::null_in_file); + Result.setFlag(Token::LeadingSpace); + if (SkipWhitespace(Result, CurPtr)) + return; // KeepWhitespaceMode + + goto LexNextToken; // GCC isn't tail call eliminating. + case '\n': + case '\r': + // If we are inside a preprocessor directive and we see the end of line, + // we know we are done with the directive, so return an EOM token. + if (ParsingPreprocessorDirective) { + // Done parsing the "line". + ParsingPreprocessorDirective = false; + + // Restore comment saving mode, in case it was disabled for directive. + SetCommentRetentionState(PP->getCommentRetentionState()); + + // Since we consumed a newline, we are back at the start of a line. + IsAtStartOfLine = true; + + Kind = tok::eom; + break; + } + // The returned token is at the start of the line. + Result.setFlag(Token::StartOfLine); + // No leading whitespace seen so far. + Result.clearFlag(Token::LeadingSpace); + + if (SkipWhitespace(Result, CurPtr)) + return; // KeepWhitespaceMode + goto LexNextToken; // GCC isn't tail call eliminating. + case ' ': + case '\t': + case '\f': + case '\v': + SkipHorizontalWhitespace: + Result.setFlag(Token::LeadingSpace); + if (SkipWhitespace(Result, CurPtr)) + return; // KeepWhitespaceMode + + SkipIgnoredUnits: + CurPtr = BufferPtr; + + // If the next token is obviously a // or /* */ comment, skip it efficiently + // too (without going through the big switch stmt). + if (CurPtr[0] == '/' && CurPtr[1] == '/' && !inKeepCommentMode() && + Features.BCPLComment) { + SkipBCPLComment(Result, CurPtr+2); + goto SkipIgnoredUnits; + } else if (CurPtr[0] == '/' && CurPtr[1] == '*' && !inKeepCommentMode()) { + SkipBlockComment(Result, CurPtr+2); + goto SkipIgnoredUnits; + } else if (isHorizontalWhitespace(*CurPtr)) { + goto SkipHorizontalWhitespace; + } + goto LexNextToken; // GCC isn't tail call eliminating. + + // C99 6.4.4.1: Integer Constants. + // C99 6.4.4.2: Floating Constants. + case '0': case '1': case '2': case '3': case '4': + case '5': case '6': case '7': case '8': case '9': + // Notify MIOpt that we read a non-whitespace/non-comment token. + MIOpt.ReadToken(); + return LexNumericConstant(Result, CurPtr); + + case 'L': // Identifier (Loony) or wide literal (L'x' or L"xyz"). + // Notify MIOpt that we read a non-whitespace/non-comment token. + MIOpt.ReadToken(); + Char = getCharAndSize(CurPtr, SizeTmp); + + // Wide string literal. + if (Char == '"') + return LexStringLiteral(Result, ConsumeChar(CurPtr, SizeTmp, Result), + true); + + // Wide character constant. + if (Char == '\'') + return LexCharConstant(Result, ConsumeChar(CurPtr, SizeTmp, Result)); + // FALL THROUGH, treating L like the start of an identifier. + + // C99 6.4.2: Identifiers. + case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': case 'G': + case 'H': case 'I': case 'J': case 'K': /*'L'*/case 'M': case 'N': + case 'O': case 'P': case 'Q': case 'R': case 'S': case 'T': case 'U': + case 'V': case 'W': case 'X': case 'Y': case 'Z': + case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': case 'g': + case 'h': case 'i': case 'j': case 'k': case 'l': case 'm': case 'n': + case 'o': case 'p': case 'q': case 'r': case 's': case 't': case 'u': + case 'v': case 'w': case 'x': case 'y': case 'z': + case '_': + // Notify MIOpt that we read a non-whitespace/non-comment token. + MIOpt.ReadToken(); + return LexIdentifier(Result, CurPtr); + + case '$': // $ in identifiers. + if (Features.DollarIdents) { + if (!isLexingRawMode()) + Diag(CurPtr-1, diag::ext_dollar_in_identifier); + // Notify MIOpt that we read a non-whitespace/non-comment token. + MIOpt.ReadToken(); + return LexIdentifier(Result, CurPtr); + } + + Kind = tok::unknown; + break; + + // C99 6.4.4: Character Constants. + case '\'': + // Notify MIOpt that we read a non-whitespace/non-comment token. + MIOpt.ReadToken(); + return LexCharConstant(Result, CurPtr); + + // C99 6.4.5: String Literals. + case '"': + // Notify MIOpt that we read a non-whitespace/non-comment token. + MIOpt.ReadToken(); + return LexStringLiteral(Result, CurPtr, false); + + // C99 6.4.6: Punctuators. + case '?': + Kind = tok::question; + break; + case '[': + Kind = tok::l_square; + break; + case ']': + Kind = tok::r_square; + break; + case '(': + Kind = tok::l_paren; + break; + case ')': + Kind = tok::r_paren; + break; + case '{': + Kind = tok::l_brace; + break; + case '}': + Kind = tok::r_brace; + break; + case '.': + Char = getCharAndSize(CurPtr, SizeTmp); + if (Char >= '0' && Char <= '9') { + // Notify MIOpt that we read a non-whitespace/non-comment token. + MIOpt.ReadToken(); + + return LexNumericConstant(Result, ConsumeChar(CurPtr, SizeTmp, Result)); + } else if (Features.CPlusPlus && Char == '*') { + Kind = tok::periodstar; + CurPtr += SizeTmp; + } else if (Char == '.' && + getCharAndSize(CurPtr+SizeTmp, SizeTmp2) == '.') { + Kind = tok::ellipsis; + CurPtr = ConsumeChar(ConsumeChar(CurPtr, SizeTmp, Result), + SizeTmp2, Result); + } else { + Kind = tok::period; + } + break; + case '&': + Char = getCharAndSize(CurPtr, SizeTmp); + if (Char == '&') { + Kind = tok::ampamp; + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + } else if (Char == '=') { + Kind = tok::ampequal; + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + } else { + Kind = tok::amp; + } + break; + case '*': + if (getCharAndSize(CurPtr, SizeTmp) == '=') { + Kind = tok::starequal; + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + } else { + Kind = tok::star; + } + break; + case '+': + Char = getCharAndSize(CurPtr, SizeTmp); + if (Char == '+') { + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + Kind = tok::plusplus; + } else if (Char == '=') { + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + Kind = tok::plusequal; + } else { + Kind = tok::plus; + } + break; + case '-': + Char = getCharAndSize(CurPtr, SizeTmp); + if (Char == '-') { // -- + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + Kind = tok::minusminus; + } else if (Char == '>' && Features.CPlusPlus && + getCharAndSize(CurPtr+SizeTmp, SizeTmp2) == '*') { // C++ ->* + CurPtr = ConsumeChar(ConsumeChar(CurPtr, SizeTmp, Result), + SizeTmp2, Result); + Kind = tok::arrowstar; + } else if (Char == '>') { // -> + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + Kind = tok::arrow; + } else if (Char == '=') { // -= + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + Kind = tok::minusequal; + } else { + Kind = tok::minus; + } + break; + case '~': + Kind = tok::tilde; + break; + case '!': + if (getCharAndSize(CurPtr, SizeTmp) == '=') { + Kind = tok::exclaimequal; + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + } else { + Kind = tok::exclaim; + } + break; + case '/': + // 6.4.9: Comments + Char = getCharAndSize(CurPtr, SizeTmp); + if (Char == '/') { // BCPL comment. + // Even if BCPL comments are disabled (e.g. in C89 mode), we generally + // want to lex this as a comment. There is one problem with this though, + // that in one particular corner case, this can change the behavior of the + // resultant program. For example, In "foo //**/ bar", C89 would lex + // this as "foo / bar" and langauges with BCPL comments would lex it as + // "foo". Check to see if the character after the second slash is a '*'. + // If so, we will lex that as a "/" instead of the start of a comment. + if (Features.BCPLComment || + getCharAndSize(CurPtr+SizeTmp, SizeTmp2) != '*') { + if (SkipBCPLComment(Result, ConsumeChar(CurPtr, SizeTmp, Result))) + return; // KeepCommentMode + + // It is common for the tokens immediately after a // comment to be + // whitespace (indentation for the next line). Instead of going through + // the big switch, handle it efficiently now. + goto SkipIgnoredUnits; + } + } + + if (Char == '*') { // /**/ comment. + if (SkipBlockComment(Result, ConsumeChar(CurPtr, SizeTmp, Result))) + return; // KeepCommentMode + goto LexNextToken; // GCC isn't tail call eliminating. + } + + if (Char == '=') { + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + Kind = tok::slashequal; + } else { + Kind = tok::slash; + } + break; + case '%': + Char = getCharAndSize(CurPtr, SizeTmp); + if (Char == '=') { + Kind = tok::percentequal; + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + } else if (Features.Digraphs && Char == '>') { + Kind = tok::r_brace; // '%>' -> '}' + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + } else if (Features.Digraphs && Char == ':') { + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + Char = getCharAndSize(CurPtr, SizeTmp); + if (Char == '%' && getCharAndSize(CurPtr+SizeTmp, SizeTmp2) == ':') { + Kind = tok::hashhash; // '%:%:' -> '##' + CurPtr = ConsumeChar(ConsumeChar(CurPtr, SizeTmp, Result), + SizeTmp2, Result); + } else if (Char == '@' && Features.Microsoft) { // %:@ -> #@ -> Charize + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + if (!isLexingRawMode()) + Diag(BufferPtr, diag::charize_microsoft_ext); + Kind = tok::hashat; + } else { // '%:' -> '#' + // We parsed a # character. If this occurs at the start of the line, + // it's actually the start of a preprocessing directive. Callback to + // the preprocessor to handle it. + // FIXME: -fpreprocessed mode?? + if (Result.isAtStartOfLine() && !LexingRawMode && !Is_PragmaLexer) { + FormTokenWithChars(Result, CurPtr, tok::hash); + PP->HandleDirective(Result); + + // As an optimization, if the preprocessor didn't switch lexers, tail + // recurse. + if (PP->isCurrentLexer(this)) { + // Start a new token. If this is a #include or something, the PP may + // want us starting at the beginning of the line again. If so, set + // the StartOfLine flag. + if (IsAtStartOfLine) { + Result.setFlag(Token::StartOfLine); + IsAtStartOfLine = false; + } + goto LexNextToken; // GCC isn't tail call eliminating. + } + + return PP->Lex(Result); + } + + Kind = tok::hash; + } + } else { + Kind = tok::percent; + } + break; + case '<': + Char = getCharAndSize(CurPtr, SizeTmp); + if (ParsingFilename) { + return LexAngledStringLiteral(Result, CurPtr); + } else if (Char == '<' && + getCharAndSize(CurPtr+SizeTmp, SizeTmp2) == '=') { + Kind = tok::lesslessequal; + CurPtr = ConsumeChar(ConsumeChar(CurPtr, SizeTmp, Result), + SizeTmp2, Result); + } else if (Char == '<') { + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + Kind = tok::lessless; + } else if (Char == '=') { + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + Kind = tok::lessequal; + } else if (Features.Digraphs && Char == ':') { // '<:' -> '[' + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + Kind = tok::l_square; + } else if (Features.Digraphs && Char == '%') { // '<%' -> '{' + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + Kind = tok::l_brace; + } else { + Kind = tok::less; + } + break; + case '>': + Char = getCharAndSize(CurPtr, SizeTmp); + if (Char == '=') { + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + Kind = tok::greaterequal; + } else if (Char == '>' && + getCharAndSize(CurPtr+SizeTmp, SizeTmp2) == '=') { + CurPtr = ConsumeChar(ConsumeChar(CurPtr, SizeTmp, Result), + SizeTmp2, Result); + Kind = tok::greatergreaterequal; + } else if (Char == '>') { + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + Kind = tok::greatergreater; + } else { + Kind = tok::greater; + } + break; + case '^': + Char = getCharAndSize(CurPtr, SizeTmp); + if (Char == '=') { + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + Kind = tok::caretequal; + } else { + Kind = tok::caret; + } + break; + case '|': + Char = getCharAndSize(CurPtr, SizeTmp); + if (Char == '=') { + Kind = tok::pipeequal; + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + } else if (Char == '|') { + Kind = tok::pipepipe; + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + } else { + Kind = tok::pipe; + } + break; + case ':': + Char = getCharAndSize(CurPtr, SizeTmp); + if (Features.Digraphs && Char == '>') { + Kind = tok::r_square; // ':>' -> ']' + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + } else if (Features.CPlusPlus && Char == ':') { + Kind = tok::coloncolon; + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + } else { + Kind = tok::colon; + } + break; + case ';': + Kind = tok::semi; + break; + case '=': + Char = getCharAndSize(CurPtr, SizeTmp); + if (Char == '=') { + Kind = tok::equalequal; + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + } else { + Kind = tok::equal; + } + break; + case ',': + Kind = tok::comma; + break; + case '#': + Char = getCharAndSize(CurPtr, SizeTmp); + if (Char == '#') { + Kind = tok::hashhash; + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + } else if (Char == '@' && Features.Microsoft) { // #@ -> Charize + Kind = tok::hashat; + if (!isLexingRawMode()) + Diag(BufferPtr, diag::charize_microsoft_ext); + CurPtr = ConsumeChar(CurPtr, SizeTmp, Result); + } else { + // We parsed a # character. If this occurs at the start of the line, + // it's actually the start of a preprocessing directive. Callback to + // the preprocessor to handle it. + // FIXME: -fpreprocessed mode?? + if (Result.isAtStartOfLine() && !LexingRawMode && !Is_PragmaLexer) { + FormTokenWithChars(Result, CurPtr, tok::hash); + PP->HandleDirective(Result); + + // As an optimization, if the preprocessor didn't switch lexers, tail + // recurse. + if (PP->isCurrentLexer(this)) { + // Start a new token. If this is a #include or something, the PP may + // want us starting at the beginning of the line again. If so, set + // the StartOfLine flag. + if (IsAtStartOfLine) { + Result.setFlag(Token::StartOfLine); + IsAtStartOfLine = false; + } + goto LexNextToken; // GCC isn't tail call eliminating. + } + return PP->Lex(Result); + } + + Kind = tok::hash; + } + break; + + case '@': + // Objective C support. + if (CurPtr[-1] == '@' && Features.ObjC1) + Kind = tok::at; + else + Kind = tok::unknown; + break; + + case '\\': + // FIXME: UCN's. + // FALL THROUGH. + default: + Kind = tok::unknown; + break; + } + + // Notify MIOpt that we read a non-whitespace/non-comment token. + MIOpt.ReadToken(); + + // Update the location of token as well as BufferPtr. + FormTokenWithChars(Result, CurPtr, Kind); +} diff --git a/lib/Lex/LiteralSupport.cpp b/lib/Lex/LiteralSupport.cpp new file mode 100644 index 0000000..0324c0b --- /dev/null +++ b/lib/Lex/LiteralSupport.cpp @@ -0,0 +1,929 @@ +//===--- LiteralSupport.cpp - Code to parse and process literals ----------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements the NumericLiteralParser, CharLiteralParser, and +// StringLiteralParser interfaces. +// +//===----------------------------------------------------------------------===// + +#include "clang/Lex/LiteralSupport.h" +#include "clang/Lex/Preprocessor.h" +#include "clang/Lex/LexDiagnostic.h" +#include "clang/Basic/TargetInfo.h" +#include "llvm/ADT/StringExtras.h" +using namespace clang; + +/// HexDigitValue - Return the value of the specified hex digit, or -1 if it's +/// not valid. +static int HexDigitValue(char C) { + if (C >= '0' && C <= '9') return C-'0'; + if (C >= 'a' && C <= 'f') return C-'a'+10; + if (C >= 'A' && C <= 'F') return C-'A'+10; + return -1; +} + +/// ProcessCharEscape - Parse a standard C escape sequence, which can occur in +/// either a character or a string literal. +static unsigned ProcessCharEscape(const char *&ThisTokBuf, + const char *ThisTokEnd, bool &HadError, + SourceLocation Loc, bool IsWide, + Preprocessor &PP) { + // Skip the '\' char. + ++ThisTokBuf; + + // We know that this character can't be off the end of the buffer, because + // that would have been \", which would not have been the end of string. + unsigned ResultChar = *ThisTokBuf++; + switch (ResultChar) { + // These map to themselves. + case '\\': case '\'': case '"': case '?': break; + + // These have fixed mappings. + case 'a': + // TODO: K&R: the meaning of '\\a' is different in traditional C + ResultChar = 7; + break; + case 'b': + ResultChar = 8; + break; + case 'e': + PP.Diag(Loc, diag::ext_nonstandard_escape) << "e"; + ResultChar = 27; + break; + case 'f': + ResultChar = 12; + break; + case 'n': + ResultChar = 10; + break; + case 'r': + ResultChar = 13; + break; + case 't': + ResultChar = 9; + break; + case 'v': + ResultChar = 11; + break; + case 'x': { // Hex escape. + ResultChar = 0; + if (ThisTokBuf == ThisTokEnd || !isxdigit(*ThisTokBuf)) { + PP.Diag(Loc, diag::err_hex_escape_no_digits); + HadError = 1; + break; + } + + // Hex escapes are a maximal series of hex digits. + bool Overflow = false; + for (; ThisTokBuf != ThisTokEnd; ++ThisTokBuf) { + int CharVal = HexDigitValue(ThisTokBuf[0]); + if (CharVal == -1) break; + // About to shift out a digit? + Overflow |= (ResultChar & 0xF0000000) ? true : false; + ResultChar <<= 4; + ResultChar |= CharVal; + } + + // See if any bits will be truncated when evaluated as a character. + unsigned CharWidth = PP.getTargetInfo().getCharWidth(IsWide); + + if (CharWidth != 32 && (ResultChar >> CharWidth) != 0) { + Overflow = true; + ResultChar &= ~0U >> (32-CharWidth); + } + + // Check for overflow. + if (Overflow) // Too many digits to fit in + PP.Diag(Loc, diag::warn_hex_escape_too_large); + break; + } + case '0': case '1': case '2': case '3': + case '4': case '5': case '6': case '7': { + // Octal escapes. + --ThisTokBuf; + ResultChar = 0; + + // Octal escapes are a series of octal digits with maximum length 3. + // "\0123" is a two digit sequence equal to "\012" "3". + unsigned NumDigits = 0; + do { + ResultChar <<= 3; + ResultChar |= *ThisTokBuf++ - '0'; + ++NumDigits; + } while (ThisTokBuf != ThisTokEnd && NumDigits < 3 && + ThisTokBuf[0] >= '0' && ThisTokBuf[0] <= '7'); + + // Check for overflow. Reject '\777', but not L'\777'. + unsigned CharWidth = PP.getTargetInfo().getCharWidth(IsWide); + + if (CharWidth != 32 && (ResultChar >> CharWidth) != 0) { + PP.Diag(Loc, diag::warn_octal_escape_too_large); + ResultChar &= ~0U >> (32-CharWidth); + } + break; + } + + // Otherwise, these are not valid escapes. + case '(': case '{': case '[': case '%': + // GCC accepts these as extensions. We warn about them as such though. + PP.Diag(Loc, diag::ext_nonstandard_escape) + << std::string()+(char)ResultChar; + break; + // FALL THROUGH. + default: + if (isgraph(ThisTokBuf[0])) + PP.Diag(Loc, diag::ext_unknown_escape) << std::string()+(char)ResultChar; + else + PP.Diag(Loc, diag::ext_unknown_escape) << "x"+llvm::utohexstr(ResultChar); + break; + } + + return ResultChar; +} + +/// ProcessUCNEscape - Read the Universal Character Name, check constraints and +/// convert the UTF32 to UTF8. This is a subroutine of StringLiteralParser. +/// When we decide to implement UCN's for character constants and identifiers, +/// we will likely rework our support for UCN's. +static void ProcessUCNEscape(const char *&ThisTokBuf, const char *ThisTokEnd, + char *&ResultBuf, bool &HadError, + SourceLocation Loc, bool IsWide, Preprocessor &PP) +{ + // FIXME: Add a warning - UCN's are only valid in C++ & C99. + // FIXME: Handle wide strings. + + // Save the beginning of the string (for error diagnostics). + const char *ThisTokBegin = ThisTokBuf; + + // Skip the '\u' char's. + ThisTokBuf += 2; + + if (ThisTokBuf == ThisTokEnd || !isxdigit(*ThisTokBuf)) { + PP.Diag(Loc, diag::err_ucn_escape_no_digits); + HadError = 1; + return; + } + typedef uint32_t UTF32; + + UTF32 UcnVal = 0; + unsigned short UcnLen = (ThisTokBuf[-1] == 'u' ? 4 : 8); + for (; ThisTokBuf != ThisTokEnd && UcnLen; ++ThisTokBuf, UcnLen--) { + int CharVal = HexDigitValue(ThisTokBuf[0]); + if (CharVal == -1) break; + UcnVal <<= 4; + UcnVal |= CharVal; + } + // If we didn't consume the proper number of digits, there is a problem. + if (UcnLen) { + PP.Diag(PP.AdvanceToTokenCharacter(Loc, ThisTokBuf-ThisTokBegin), + diag::err_ucn_escape_incomplete); + HadError = 1; + return; + } + // Check UCN constraints (C99 6.4.3p2). + if ((UcnVal < 0xa0 && + (UcnVal != 0x24 && UcnVal != 0x40 && UcnVal != 0x60 )) // $, @, ` + || (UcnVal >= 0xD800 && UcnVal <= 0xDFFF) + || (UcnVal > 0x10FFFF)) /* the maximum legal UTF32 value */ { + PP.Diag(Loc, diag::err_ucn_escape_invalid); + HadError = 1; + return; + } + // Now that we've parsed/checked the UCN, we convert from UTF32->UTF8. + // The conversion below was inspired by: + // http://www.unicode.org/Public/PROGRAMS/CVTUTF/ConvertUTF.c + // First, we determine how many bytes the result will require. + typedef uint8_t UTF8; + + unsigned short bytesToWrite = 0; + if (UcnVal < (UTF32)0x80) + bytesToWrite = 1; + else if (UcnVal < (UTF32)0x800) + bytesToWrite = 2; + else if (UcnVal < (UTF32)0x10000) + bytesToWrite = 3; + else + bytesToWrite = 4; + + const unsigned byteMask = 0xBF; + const unsigned byteMark = 0x80; + + // Once the bits are split out into bytes of UTF8, this is a mask OR-ed + // into the first byte, depending on how many bytes follow. + static const UTF8 firstByteMark[5] = { + 0x00, 0x00, 0xC0, 0xE0, 0xF0 + }; + // Finally, we write the bytes into ResultBuf. + ResultBuf += bytesToWrite; + switch (bytesToWrite) { // note: everything falls through. + case 4: *--ResultBuf = (UTF8)((UcnVal | byteMark) & byteMask); UcnVal >>= 6; + case 3: *--ResultBuf = (UTF8)((UcnVal | byteMark) & byteMask); UcnVal >>= 6; + case 2: *--ResultBuf = (UTF8)((UcnVal | byteMark) & byteMask); UcnVal >>= 6; + case 1: *--ResultBuf = (UTF8) (UcnVal | firstByteMark[bytesToWrite]); + } + // Update the buffer. + ResultBuf += bytesToWrite; +} + + +/// integer-constant: [C99 6.4.4.1] +/// decimal-constant integer-suffix +/// octal-constant integer-suffix +/// hexadecimal-constant integer-suffix +/// decimal-constant: +/// nonzero-digit +/// decimal-constant digit +/// octal-constant: +/// 0 +/// octal-constant octal-digit +/// hexadecimal-constant: +/// hexadecimal-prefix hexadecimal-digit +/// hexadecimal-constant hexadecimal-digit +/// hexadecimal-prefix: one of +/// 0x 0X +/// integer-suffix: +/// unsigned-suffix [long-suffix] +/// unsigned-suffix [long-long-suffix] +/// long-suffix [unsigned-suffix] +/// long-long-suffix [unsigned-sufix] +/// nonzero-digit: +/// 1 2 3 4 5 6 7 8 9 +/// octal-digit: +/// 0 1 2 3 4 5 6 7 +/// hexadecimal-digit: +/// 0 1 2 3 4 5 6 7 8 9 +/// a b c d e f +/// A B C D E F +/// unsigned-suffix: one of +/// u U +/// long-suffix: one of +/// l L +/// long-long-suffix: one of +/// ll LL +/// +/// floating-constant: [C99 6.4.4.2] +/// TODO: add rules... +/// +NumericLiteralParser:: +NumericLiteralParser(const char *begin, const char *end, + SourceLocation TokLoc, Preprocessor &pp) + : PP(pp), ThisTokBegin(begin), ThisTokEnd(end) { + + // This routine assumes that the range begin/end matches the regex for integer + // and FP constants (specifically, the 'pp-number' regex), and assumes that + // the byte at "*end" is both valid and not part of the regex. Because of + // this, it doesn't have to check for 'overscan' in various places. + assert(!isalnum(*end) && *end != '.' && *end != '_' && + "Lexer didn't maximally munch?"); + + s = DigitsBegin = begin; + saw_exponent = false; + saw_period = false; + isLong = false; + isUnsigned = false; + isLongLong = false; + isFloat = false; + isImaginary = false; + hadError = false; + + if (*s == '0') { // parse radix + ParseNumberStartingWithZero(TokLoc); + if (hadError) + return; + } else { // the first digit is non-zero + radix = 10; + s = SkipDigits(s); + if (s == ThisTokEnd) { + // Done. + } else if (isxdigit(*s) && !(*s == 'e' || *s == 'E')) { + PP.Diag(PP.AdvanceToTokenCharacter(TokLoc, s-begin), + diag::err_invalid_decimal_digit) << std::string(s, s+1); + hadError = true; + return; + } else if (*s == '.') { + s++; + saw_period = true; + s = SkipDigits(s); + } + if ((*s == 'e' || *s == 'E')) { // exponent + const char *Exponent = s; + s++; + saw_exponent = true; + if (*s == '+' || *s == '-') s++; // sign + const char *first_non_digit = SkipDigits(s); + if (first_non_digit != s) { + s = first_non_digit; + } else { + PP.Diag(PP.AdvanceToTokenCharacter(TokLoc, Exponent-begin), + diag::err_exponent_has_no_digits); + hadError = true; + return; + } + } + } + + SuffixBegin = s; + + // Parse the suffix. At this point we can classify whether we have an FP or + // integer constant. + bool isFPConstant = isFloatingLiteral(); + + // Loop over all of the characters of the suffix. If we see something bad, + // we break out of the loop. + for (; s != ThisTokEnd; ++s) { + switch (*s) { + case 'f': // FP Suffix for "float" + case 'F': + if (!isFPConstant) break; // Error for integer constant. + if (isFloat || isLong) break; // FF, LF invalid. + isFloat = true; + continue; // Success. + case 'u': + case 'U': + if (isFPConstant) break; // Error for floating constant. + if (isUnsigned) break; // Cannot be repeated. + isUnsigned = true; + continue; // Success. + case 'l': + case 'L': + if (isLong || isLongLong) break; // Cannot be repeated. + if (isFloat) break; // LF invalid. + + // Check for long long. The L's need to be adjacent and the same case. + if (s+1 != ThisTokEnd && s[1] == s[0]) { + if (isFPConstant) break; // long long invalid for floats. + isLongLong = true; + ++s; // Eat both of them. + } else { + isLong = true; + } + continue; // Success. + case 'i': + if (PP.getLangOptions().Microsoft) { + // Allow i8, i16, i32, i64, and i128. + if (++s == ThisTokEnd) break; + switch (*s) { + case '8': + s++; // i8 suffix + break; + case '1': + if (++s == ThisTokEnd) break; + if (*s == '6') s++; // i16 suffix + else if (*s == '2') { + if (++s == ThisTokEnd) break; + if (*s == '8') s++; // i128 suffix + } + break; + case '3': + if (++s == ThisTokEnd) break; + if (*s == '2') s++; // i32 suffix + break; + case '6': + if (++s == ThisTokEnd) break; + if (*s == '4') s++; // i64 suffix + break; + default: + break; + } + break; + } + // fall through. + case 'I': + case 'j': + case 'J': + if (isImaginary) break; // Cannot be repeated. + PP.Diag(PP.AdvanceToTokenCharacter(TokLoc, s-begin), + diag::ext_imaginary_constant); + isImaginary = true; + continue; // Success. + } + // If we reached here, there was an error. + break; + } + + // Report an error if there are any. + if (s != ThisTokEnd) { + PP.Diag(PP.AdvanceToTokenCharacter(TokLoc, s-begin), + isFPConstant ? diag::err_invalid_suffix_float_constant : + diag::err_invalid_suffix_integer_constant) + << std::string(SuffixBegin, ThisTokEnd); + hadError = true; + return; + } +} + +/// ParseNumberStartingWithZero - This method is called when the first character +/// of the number is found to be a zero. This means it is either an octal +/// number (like '04') or a hex number ('0x123a') a binary number ('0b1010') or +/// a floating point number (01239.123e4). Eat the prefix, determining the +/// radix etc. +void NumericLiteralParser::ParseNumberStartingWithZero(SourceLocation TokLoc) { + assert(s[0] == '0' && "Invalid method call"); + s++; + + // Handle a hex number like 0x1234. + if ((*s == 'x' || *s == 'X') && (isxdigit(s[1]) || s[1] == '.')) { + s++; + radix = 16; + DigitsBegin = s; + s = SkipHexDigits(s); + if (s == ThisTokEnd) { + // Done. + } else if (*s == '.') { + s++; + saw_period = true; + s = SkipHexDigits(s); + } + // A binary exponent can appear with or with a '.'. If dotted, the + // binary exponent is required. + if (*s == 'p' || *s == 'P') { + const char *Exponent = s; + s++; + saw_exponent = true; + if (*s == '+' || *s == '-') s++; // sign + const char *first_non_digit = SkipDigits(s); + if (first_non_digit == s) { + PP.Diag(PP.AdvanceToTokenCharacter(TokLoc, Exponent-ThisTokBegin), + diag::err_exponent_has_no_digits); + hadError = true; + return; + } + s = first_non_digit; + + if (!PP.getLangOptions().HexFloats) + PP.Diag(TokLoc, diag::ext_hexconstant_invalid); + } else if (saw_period) { + PP.Diag(PP.AdvanceToTokenCharacter(TokLoc, s-ThisTokBegin), + diag::err_hexconstant_requires_exponent); + hadError = true; + } + return; + } + + // Handle simple binary numbers 0b01010 + if (*s == 'b' || *s == 'B') { + // 0b101010 is a GCC extension. + PP.Diag(TokLoc, diag::ext_binary_literal); + ++s; + radix = 2; + DigitsBegin = s; + s = SkipBinaryDigits(s); + if (s == ThisTokEnd) { + // Done. + } else if (isxdigit(*s)) { + PP.Diag(PP.AdvanceToTokenCharacter(TokLoc, s-ThisTokBegin), + diag::err_invalid_binary_digit) << std::string(s, s+1); + hadError = true; + } + // Other suffixes will be diagnosed by the caller. + return; + } + + // For now, the radix is set to 8. If we discover that we have a + // floating point constant, the radix will change to 10. Octal floating + // point constants are not permitted (only decimal and hexadecimal). + radix = 8; + DigitsBegin = s; + s = SkipOctalDigits(s); + if (s == ThisTokEnd) + return; // Done, simple octal number like 01234 + + // If we have some other non-octal digit that *is* a decimal digit, see if + // this is part of a floating point number like 094.123 or 09e1. + if (isdigit(*s)) { + const char *EndDecimal = SkipDigits(s); + if (EndDecimal[0] == '.' || EndDecimal[0] == 'e' || EndDecimal[0] == 'E') { + s = EndDecimal; + radix = 10; + } + } + + // If we have a hex digit other than 'e' (which denotes a FP exponent) then + // the code is using an incorrect base. + if (isxdigit(*s) && *s != 'e' && *s != 'E') { + PP.Diag(PP.AdvanceToTokenCharacter(TokLoc, s-ThisTokBegin), + diag::err_invalid_octal_digit) << std::string(s, s+1); + hadError = true; + return; + } + + if (*s == '.') { + s++; + radix = 10; + saw_period = true; + s = SkipDigits(s); // Skip suffix. + } + if (*s == 'e' || *s == 'E') { // exponent + const char *Exponent = s; + s++; + radix = 10; + saw_exponent = true; + if (*s == '+' || *s == '-') s++; // sign + const char *first_non_digit = SkipDigits(s); + if (first_non_digit != s) { + s = first_non_digit; + } else { + PP.Diag(PP.AdvanceToTokenCharacter(TokLoc, Exponent-ThisTokBegin), + diag::err_exponent_has_no_digits); + hadError = true; + return; + } + } +} + + +/// GetIntegerValue - Convert this numeric literal value to an APInt that +/// matches Val's input width. If there is an overflow, set Val to the low bits +/// of the result and return true. Otherwise, return false. +bool NumericLiteralParser::GetIntegerValue(llvm::APInt &Val) { + // Fast path: Compute a conservative bound on the maximum number of + // bits per digit in this radix. If we can't possibly overflow a + // uint64 based on that bound then do the simple conversion to + // integer. This avoids the expensive overflow checking below, and + // handles the common cases that matter (small decimal integers and + // hex/octal values which don't overflow). + unsigned MaxBitsPerDigit = 1; + while ((1U << MaxBitsPerDigit) < radix) + MaxBitsPerDigit += 1; + if ((SuffixBegin - DigitsBegin) * MaxBitsPerDigit <= 64) { + uint64_t N = 0; + for (s = DigitsBegin; s != SuffixBegin; ++s) + N = N*radix + HexDigitValue(*s); + + // This will truncate the value to Val's input width. Simply check + // for overflow by comparing. + Val = N; + return Val.getZExtValue() != N; + } + + Val = 0; + s = DigitsBegin; + + llvm::APInt RadixVal(Val.getBitWidth(), radix); + llvm::APInt CharVal(Val.getBitWidth(), 0); + llvm::APInt OldVal = Val; + + bool OverflowOccurred = false; + while (s < SuffixBegin) { + unsigned C = HexDigitValue(*s++); + + // If this letter is out of bound for this radix, reject it. + assert(C < radix && "NumericLiteralParser ctor should have rejected this"); + + CharVal = C; + + // Add the digit to the value in the appropriate radix. If adding in digits + // made the value smaller, then this overflowed. + OldVal = Val; + + // Multiply by radix, did overflow occur on the multiply? + Val *= RadixVal; + OverflowOccurred |= Val.udiv(RadixVal) != OldVal; + + // Add value, did overflow occur on the value? + // (a + b) ult b <=> overflow + Val += CharVal; + OverflowOccurred |= Val.ult(CharVal); + } + return OverflowOccurred; +} + +llvm::APFloat NumericLiteralParser:: +GetFloatValue(const llvm::fltSemantics &Format, bool* isExact) { + using llvm::APFloat; + + llvm::SmallVector<char,256> floatChars; + for (unsigned i = 0, n = ThisTokEnd-ThisTokBegin; i != n; ++i) + floatChars.push_back(ThisTokBegin[i]); + + floatChars.push_back('\0'); + + APFloat V (Format, APFloat::fcZero, false); + APFloat::opStatus status; + + status = V.convertFromString(&floatChars[0],APFloat::rmNearestTiesToEven); + + if (isExact) + *isExact = status == APFloat::opOK; + + return V; +} + + +CharLiteralParser::CharLiteralParser(const char *begin, const char *end, + SourceLocation Loc, Preprocessor &PP) { + // At this point we know that the character matches the regex "L?'.*'". + HadError = false; + + // Determine if this is a wide character. + IsWide = begin[0] == 'L'; + if (IsWide) ++begin; + + // Skip over the entry quote. + assert(begin[0] == '\'' && "Invalid token lexed"); + ++begin; + + // FIXME: The "Value" is an uint64_t so we can handle char literals of + // upto 64-bits. + // FIXME: This extensively assumes that 'char' is 8-bits. + assert(PP.getTargetInfo().getCharWidth() == 8 && + "Assumes char is 8 bits"); + assert(PP.getTargetInfo().getIntWidth() <= 64 && + (PP.getTargetInfo().getIntWidth() & 7) == 0 && + "Assumes sizeof(int) on target is <= 64 and a multiple of char"); + assert(PP.getTargetInfo().getWCharWidth() <= 64 && + "Assumes sizeof(wchar) on target is <= 64"); + + // This is what we will use for overflow detection + llvm::APInt LitVal(PP.getTargetInfo().getIntWidth(), 0); + + unsigned NumCharsSoFar = 0; + while (begin[0] != '\'') { + uint64_t ResultChar; + if (begin[0] != '\\') // If this is a normal character, consume it. + ResultChar = *begin++; + else // Otherwise, this is an escape character. + ResultChar = ProcessCharEscape(begin, end, HadError, Loc, IsWide, PP); + + // If this is a multi-character constant (e.g. 'abc'), handle it. These are + // implementation defined (C99 6.4.4.4p10). + if (NumCharsSoFar) { + if (IsWide) { + // Emulate GCC's (unintentional?) behavior: L'ab' -> L'b'. + LitVal = 0; + } else { + // Narrow character literals act as though their value is concatenated + // in this implementation, but warn on overflow. + if (LitVal.countLeadingZeros() < 8) + PP.Diag(Loc, diag::warn_char_constant_too_large); + LitVal <<= 8; + } + } + + LitVal = LitVal + ResultChar; + ++NumCharsSoFar; + } + + // If this is the second character being processed, do special handling. + if (NumCharsSoFar > 1) { + // Warn about discarding the top bits for multi-char wide-character + // constants (L'abcd'). + if (IsWide) + PP.Diag(Loc, diag::warn_extraneous_wide_char_constant); + else if (NumCharsSoFar != 4) + PP.Diag(Loc, diag::ext_multichar_character_literal); + else + PP.Diag(Loc, diag::ext_four_char_character_literal); + IsMultiChar = true; + } + + // Transfer the value from APInt to uint64_t + Value = LitVal.getZExtValue(); + + // If this is a single narrow character, sign extend it (e.g. '\xFF' is "-1") + // if 'char' is signed for this target (C99 6.4.4.4p10). Note that multiple + // character constants are not sign extended in the this implementation: + // '\xFF\xFF' = 65536 and '\x0\xFF' = 255, which matches GCC. + if (!IsWide && NumCharsSoFar == 1 && (Value & 128) && + PP.getTargetInfo().isCharSigned()) + Value = (signed char)Value; +} + + +/// string-literal: [C99 6.4.5] +/// " [s-char-sequence] " +/// L" [s-char-sequence] " +/// s-char-sequence: +/// s-char +/// s-char-sequence s-char +/// s-char: +/// any source character except the double quote ", +/// backslash \, or newline character +/// escape-character +/// universal-character-name +/// escape-character: [C99 6.4.4.4] +/// \ escape-code +/// universal-character-name +/// escape-code: +/// character-escape-code +/// octal-escape-code +/// hex-escape-code +/// character-escape-code: one of +/// n t b r f v a +/// \ ' " ? +/// octal-escape-code: +/// octal-digit +/// octal-digit octal-digit +/// octal-digit octal-digit octal-digit +/// hex-escape-code: +/// x hex-digit +/// hex-escape-code hex-digit +/// universal-character-name: +/// \u hex-quad +/// \U hex-quad hex-quad +/// hex-quad: +/// hex-digit hex-digit hex-digit hex-digit +/// +StringLiteralParser:: +StringLiteralParser(const Token *StringToks, unsigned NumStringToks, + Preprocessor &pp) : PP(pp) { + // Scan all of the string portions, remember the max individual token length, + // computing a bound on the concatenated string length, and see whether any + // piece is a wide-string. If any of the string portions is a wide-string + // literal, the result is a wide-string literal [C99 6.4.5p4]. + MaxTokenLength = StringToks[0].getLength(); + SizeBound = StringToks[0].getLength()-2; // -2 for "". + AnyWide = StringToks[0].is(tok::wide_string_literal); + + hadError = false; + + // Implement Translation Phase #6: concatenation of string literals + /// (C99 5.1.1.2p1). The common case is only one string fragment. + for (unsigned i = 1; i != NumStringToks; ++i) { + // The string could be shorter than this if it needs cleaning, but this is a + // reasonable bound, which is all we need. + SizeBound += StringToks[i].getLength()-2; // -2 for "". + + // Remember maximum string piece length. + if (StringToks[i].getLength() > MaxTokenLength) + MaxTokenLength = StringToks[i].getLength(); + + // Remember if we see any wide strings. + AnyWide |= StringToks[i].is(tok::wide_string_literal); + } + + // Include space for the null terminator. + ++SizeBound; + + // TODO: K&R warning: "traditional C rejects string constant concatenation" + + // Get the width in bytes of wchar_t. If no wchar_t strings are used, do not + // query the target. As such, wchar_tByteWidth is only valid if AnyWide=true. + wchar_tByteWidth = ~0U; + if (AnyWide) { + wchar_tByteWidth = PP.getTargetInfo().getWCharWidth(); + assert((wchar_tByteWidth & 7) == 0 && "Assumes wchar_t is byte multiple!"); + wchar_tByteWidth /= 8; + } + + // The output buffer size needs to be large enough to hold wide characters. + // This is a worst-case assumption which basically corresponds to L"" "long". + if (AnyWide) + SizeBound *= wchar_tByteWidth; + + // Size the temporary buffer to hold the result string data. + ResultBuf.resize(SizeBound); + + // Likewise, but for each string piece. + llvm::SmallString<512> TokenBuf; + TokenBuf.resize(MaxTokenLength); + + // Loop over all the strings, getting their spelling, and expanding them to + // wide strings as appropriate. + ResultPtr = &ResultBuf[0]; // Next byte to fill in. + + Pascal = false; + + for (unsigned i = 0, e = NumStringToks; i != e; ++i) { + const char *ThisTokBuf = &TokenBuf[0]; + // Get the spelling of the token, which eliminates trigraphs, etc. We know + // that ThisTokBuf points to a buffer that is big enough for the whole token + // and 'spelled' tokens can only shrink. + unsigned ThisTokLen = PP.getSpelling(StringToks[i], ThisTokBuf); + const char *ThisTokEnd = ThisTokBuf+ThisTokLen-1; // Skip end quote. + + // TODO: Input character set mapping support. + + // Skip L marker for wide strings. + bool ThisIsWide = false; + if (ThisTokBuf[0] == 'L') { + ++ThisTokBuf; + ThisIsWide = true; + } + + assert(ThisTokBuf[0] == '"' && "Expected quote, lexer broken?"); + ++ThisTokBuf; + + // Check if this is a pascal string + if (pp.getLangOptions().PascalStrings && ThisTokBuf + 1 != ThisTokEnd && + ThisTokBuf[0] == '\\' && ThisTokBuf[1] == 'p') { + + // If the \p sequence is found in the first token, we have a pascal string + // Otherwise, if we already have a pascal string, ignore the first \p + if (i == 0) { + ++ThisTokBuf; + Pascal = true; + } else if (Pascal) + ThisTokBuf += 2; + } + + while (ThisTokBuf != ThisTokEnd) { + // Is this a span of non-escape characters? + if (ThisTokBuf[0] != '\\') { + const char *InStart = ThisTokBuf; + do { + ++ThisTokBuf; + } while (ThisTokBuf != ThisTokEnd && ThisTokBuf[0] != '\\'); + + // Copy the character span over. + unsigned Len = ThisTokBuf-InStart; + if (!AnyWide) { + memcpy(ResultPtr, InStart, Len); + ResultPtr += Len; + } else { + // Note: our internal rep of wide char tokens is always little-endian. + for (; Len; --Len, ++InStart) { + *ResultPtr++ = InStart[0]; + // Add zeros at the end. + for (unsigned i = 1, e = wchar_tByteWidth; i != e; ++i) + *ResultPtr++ = 0; + } + } + continue; + } + // Is this a Universal Character Name escape? + if (ThisTokBuf[1] == 'u' || ThisTokBuf[1] == 'U') { + ProcessUCNEscape(ThisTokBuf, ThisTokEnd, ResultPtr, + hadError, StringToks[i].getLocation(), ThisIsWide, PP); + continue; + } + // Otherwise, this is a non-UCN escape character. Process it. + unsigned ResultChar = ProcessCharEscape(ThisTokBuf, ThisTokEnd, hadError, + StringToks[i].getLocation(), + ThisIsWide, PP); + + // Note: our internal rep of wide char tokens is always little-endian. + *ResultPtr++ = ResultChar & 0xFF; + + if (AnyWide) { + for (unsigned i = 1, e = wchar_tByteWidth; i != e; ++i) + *ResultPtr++ = ResultChar >> i*8; + } + } + } + + if (Pascal) { + ResultBuf[0] = ResultPtr-&ResultBuf[0]-1; + + // Verify that pascal strings aren't too large. + if (GetStringLength() > 256) { + PP.Diag(StringToks[0].getLocation(), diag::err_pascal_string_too_long) + << SourceRange(StringToks[0].getLocation(), + StringToks[NumStringToks-1].getLocation()); + hadError = 1; + return; + } + } +} + + +/// getOffsetOfStringByte - This function returns the offset of the +/// specified byte of the string data represented by Token. This handles +/// advancing over escape sequences in the string. +unsigned StringLiteralParser::getOffsetOfStringByte(const Token &Tok, + unsigned ByteNo, + Preprocessor &PP) { + // Get the spelling of the token. + llvm::SmallString<16> SpellingBuffer; + SpellingBuffer.resize(Tok.getLength()); + + const char *SpellingPtr = &SpellingBuffer[0]; + unsigned TokLen = PP.getSpelling(Tok, SpellingPtr); + + assert(SpellingPtr[0] != 'L' && "Doesn't handle wide strings yet"); + + + const char *SpellingStart = SpellingPtr; + const char *SpellingEnd = SpellingPtr+TokLen; + + // Skip over the leading quote. + assert(SpellingPtr[0] == '"' && "Should be a string literal!"); + ++SpellingPtr; + + // Skip over bytes until we find the offset we're looking for. + while (ByteNo) { + assert(SpellingPtr < SpellingEnd && "Didn't find byte offset!"); + + // Step over non-escapes simply. + if (*SpellingPtr != '\\') { + ++SpellingPtr; + --ByteNo; + continue; + } + + // Otherwise, this is an escape character. Advance over it. + bool HadError = false; + ProcessCharEscape(SpellingPtr, SpellingEnd, HadError, + Tok.getLocation(), false, PP); + assert(!HadError && "This method isn't valid on erroneous strings"); + --ByteNo; + } + + return SpellingPtr-SpellingStart; +} diff --git a/lib/Lex/MacroArgs.cpp b/lib/Lex/MacroArgs.cpp new file mode 100644 index 0000000..cba69b7 --- /dev/null +++ b/lib/Lex/MacroArgs.cpp @@ -0,0 +1,240 @@ +//===--- TokenLexer.cpp - Lex from a token stream -------------------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements the TokenLexer interface. +// +//===----------------------------------------------------------------------===// + +#include "MacroArgs.h" +#include "clang/Lex/MacroInfo.h" +#include "clang/Lex/Preprocessor.h" +#include "clang/Lex/LexDiagnostic.h" +using namespace clang; + +/// MacroArgs ctor function - This destroys the vector passed in. +MacroArgs *MacroArgs::create(const MacroInfo *MI, + const Token *UnexpArgTokens, + unsigned NumToks, bool VarargsElided) { + assert(MI->isFunctionLike() && + "Can't have args for an object-like macro!"); + + // Allocate memory for the MacroArgs object with the lexer tokens at the end. + MacroArgs *Result = (MacroArgs*)malloc(sizeof(MacroArgs) + + NumToks*sizeof(Token)); + // Construct the macroargs object. + new (Result) MacroArgs(NumToks, VarargsElided); + + // Copy the actual unexpanded tokens to immediately after the result ptr. + if (NumToks) + memcpy(const_cast<Token*>(Result->getUnexpArgument(0)), + UnexpArgTokens, NumToks*sizeof(Token)); + + return Result; +} + +/// destroy - Destroy and deallocate the memory for this object. +/// +void MacroArgs::destroy() { + // Run the dtor to deallocate the vectors. + this->~MacroArgs(); + // Release the memory for the object. + free(this); +} + + +/// getArgLength - Given a pointer to an expanded or unexpanded argument, +/// return the number of tokens, not counting the EOF, that make up the +/// argument. +unsigned MacroArgs::getArgLength(const Token *ArgPtr) { + unsigned NumArgTokens = 0; + for (; ArgPtr->isNot(tok::eof); ++ArgPtr) + ++NumArgTokens; + return NumArgTokens; +} + + +/// getUnexpArgument - Return the unexpanded tokens for the specified formal. +/// +const Token *MacroArgs::getUnexpArgument(unsigned Arg) const { + // The unexpanded argument tokens start immediately after the MacroArgs object + // in memory. + const Token *Start = (const Token *)(this+1); + const Token *Result = Start; + // Scan to find Arg. + for (; Arg; ++Result) { + assert(Result < Start+NumUnexpArgTokens && "Invalid arg #"); + if (Result->is(tok::eof)) + --Arg; + } + assert(Result < Start+NumUnexpArgTokens && "Invalid arg #"); + return Result; +} + + +/// ArgNeedsPreexpansion - If we can prove that the argument won't be affected +/// by pre-expansion, return false. Otherwise, conservatively return true. +bool MacroArgs::ArgNeedsPreexpansion(const Token *ArgTok, + Preprocessor &PP) const { + // If there are no identifiers in the argument list, or if the identifiers are + // known to not be macros, pre-expansion won't modify it. + for (; ArgTok->isNot(tok::eof); ++ArgTok) + if (IdentifierInfo *II = ArgTok->getIdentifierInfo()) { + if (II->hasMacroDefinition() && PP.getMacroInfo(II)->isEnabled()) + // Return true even though the macro could be a function-like macro + // without a following '(' token. + return true; + } + return false; +} + +/// getPreExpArgument - Return the pre-expanded form of the specified +/// argument. +const std::vector<Token> & +MacroArgs::getPreExpArgument(unsigned Arg, Preprocessor &PP) { + assert(Arg < NumUnexpArgTokens && "Invalid argument number!"); + + // If we have already computed this, return it. + if (PreExpArgTokens.empty()) + PreExpArgTokens.resize(NumUnexpArgTokens); + + std::vector<Token> &Result = PreExpArgTokens[Arg]; + if (!Result.empty()) return Result; + + const Token *AT = getUnexpArgument(Arg); + unsigned NumToks = getArgLength(AT)+1; // Include the EOF. + + // Otherwise, we have to pre-expand this argument, populating Result. To do + // this, we set up a fake TokenLexer to lex from the unexpanded argument + // list. With this installed, we lex expanded tokens until we hit the EOF + // token at the end of the unexp list. + PP.EnterTokenStream(AT, NumToks, false /*disable expand*/, + false /*owns tokens*/); + + // Lex all of the macro-expanded tokens into Result. + do { + Result.push_back(Token()); + Token &Tok = Result.back(); + PP.Lex(Tok); + } while (Result.back().isNot(tok::eof)); + + // Pop the token stream off the top of the stack. We know that the internal + // pointer inside of it is to the "end" of the token stream, but the stack + // will not otherwise be popped until the next token is lexed. The problem is + // that the token may be lexed sometime after the vector of tokens itself is + // destroyed, which would be badness. + PP.RemoveTopOfLexerStack(); + return Result; +} + + +/// StringifyArgument - Implement C99 6.10.3.2p2, converting a sequence of +/// tokens into the literal string token that should be produced by the C # +/// preprocessor operator. If Charify is true, then it should be turned into +/// a character literal for the Microsoft charize (#@) extension. +/// +Token MacroArgs::StringifyArgument(const Token *ArgToks, + Preprocessor &PP, bool Charify) { + Token Tok; + Tok.startToken(); + Tok.setKind(tok::string_literal); + + const Token *ArgTokStart = ArgToks; + + // Stringify all the tokens. + llvm::SmallString<128> Result; + Result += "\""; + + bool isFirst = true; + for (; ArgToks->isNot(tok::eof); ++ArgToks) { + const Token &Tok = *ArgToks; + if (!isFirst && (Tok.hasLeadingSpace() || Tok.isAtStartOfLine())) + Result += ' '; + isFirst = false; + + // If this is a string or character constant, escape the token as specified + // by 6.10.3.2p2. + if (Tok.is(tok::string_literal) || // "foo" + Tok.is(tok::wide_string_literal) || // L"foo" + Tok.is(tok::char_constant)) { // 'x' and L'x'. + std::string Str = Lexer::Stringify(PP.getSpelling(Tok)); + Result.append(Str.begin(), Str.end()); + } else { + // Otherwise, just append the token. Do some gymnastics to get the token + // in place and avoid copies where possible. + unsigned CurStrLen = Result.size(); + Result.resize(CurStrLen+Tok.getLength()); + const char *BufPtr = &Result[CurStrLen]; + unsigned ActualTokLen = PP.getSpelling(Tok, BufPtr); + + // If getSpelling returned a pointer to an already uniqued version of the + // string instead of filling in BufPtr, memcpy it onto our string. + if (BufPtr != &Result[CurStrLen]) + memcpy(&Result[CurStrLen], BufPtr, ActualTokLen); + + // If the token was dirty, the spelling may be shorter than the token. + if (ActualTokLen != Tok.getLength()) + Result.resize(CurStrLen+ActualTokLen); + } + } + + // If the last character of the string is a \, and if it isn't escaped, this + // is an invalid string literal, diagnose it as specified in C99. + if (Result.back() == '\\') { + // Count the number of consequtive \ characters. If even, then they are + // just escaped backslashes, otherwise it's an error. + unsigned FirstNonSlash = Result.size()-2; + // Guaranteed to find the starting " if nothing else. + while (Result[FirstNonSlash] == '\\') + --FirstNonSlash; + if ((Result.size()-1-FirstNonSlash) & 1) { + // Diagnose errors for things like: #define F(X) #X / F(\) + PP.Diag(ArgToks[-1], diag::pp_invalid_string_literal); + Result.pop_back(); // remove one of the \'s. + } + } + Result += '"'; + + // If this is the charify operation and the result is not a legal character + // constant, diagnose it. + if (Charify) { + // First step, turn double quotes into single quotes: + Result[0] = '\''; + Result[Result.size()-1] = '\''; + + // Check for bogus character. + bool isBad = false; + if (Result.size() == 3) + isBad = Result[1] == '\''; // ''' is not legal. '\' already fixed above. + else + isBad = (Result.size() != 4 || Result[1] != '\\'); // Not '\x' + + if (isBad) { + PP.Diag(ArgTokStart[0], diag::err_invalid_character_to_charify); + Result = "' '"; // Use something arbitrary, but legal. + } + } + + PP.CreateString(&Result[0], Result.size(), Tok); + return Tok; +} + +/// getStringifiedArgument - Compute, cache, and return the specified argument +/// that has been 'stringified' as required by the # operator. +const Token &MacroArgs::getStringifiedArgument(unsigned ArgNo, + Preprocessor &PP) { + assert(ArgNo < NumUnexpArgTokens && "Invalid argument number!"); + if (StringifiedArgs.empty()) { + StringifiedArgs.resize(getNumArguments()); + memset(&StringifiedArgs[0], 0, + sizeof(StringifiedArgs[0])*getNumArguments()); + } + if (StringifiedArgs[ArgNo].isNot(tok::string_literal)) + StringifiedArgs[ArgNo] = StringifyArgument(getUnexpArgument(ArgNo), PP); + return StringifiedArgs[ArgNo]; +} diff --git a/lib/Lex/MacroArgs.h b/lib/Lex/MacroArgs.h new file mode 100644 index 0000000..4b22fa1 --- /dev/null +++ b/lib/Lex/MacroArgs.h @@ -0,0 +1,109 @@ +//===--- MacroArgs.h - Formal argument info for Macros ----------*- C++ -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file defines the MacroArgs interface. +// +//===----------------------------------------------------------------------===// + +#ifndef LLVM_CLANG_MACROARGS_H +#define LLVM_CLANG_MACROARGS_H + +#include <vector> + +namespace clang { + class MacroInfo; + class Preprocessor; + class Token; + +/// MacroArgs - An instance of this class captures information about +/// the formal arguments specified to a function-like macro invocation. +class MacroArgs { + /// NumUnexpArgTokens - The number of raw, unexpanded tokens for the + /// arguments. All of the actual argument tokens are allocated immediately + /// after the MacroArgs object in memory. This is all of the arguments + /// concatenated together, with 'EOF' markers at the end of each argument. + unsigned NumUnexpArgTokens; + + /// PreExpArgTokens - Pre-expanded tokens for arguments that need them. Empty + /// if not yet computed. This includes the EOF marker at the end of the + /// stream. + std::vector<std::vector<Token> > PreExpArgTokens; + + /// StringifiedArgs - This contains arguments in 'stringified' form. If the + /// stringified form of an argument has not yet been computed, this is empty. + std::vector<Token> StringifiedArgs; + + /// VarargsElided - True if this is a C99 style varargs macro invocation and + /// there was no argument specified for the "..." argument. If the argument + /// was specified (even empty) or this isn't a C99 style varargs function, or + /// if in strict mode and the C99 varargs macro had only a ... argument, this + /// is false. + bool VarargsElided; + + MacroArgs(unsigned NumToks, bool varargsElided) + : NumUnexpArgTokens(NumToks), VarargsElided(varargsElided) {} + ~MacroArgs() {} +public: + /// MacroArgs ctor function - Create a new MacroArgs object with the specified + /// macro and argument info. + static MacroArgs *create(const MacroInfo *MI, + const Token *UnexpArgTokens, + unsigned NumArgTokens, bool VarargsElided); + + /// destroy - Destroy and deallocate the memory for this object. + /// + void destroy(); + + /// ArgNeedsPreexpansion - If we can prove that the argument won't be affected + /// by pre-expansion, return false. Otherwise, conservatively return true. + bool ArgNeedsPreexpansion(const Token *ArgTok, Preprocessor &PP) const; + + /// getUnexpArgument - Return a pointer to the first token of the unexpanded + /// token list for the specified formal. + /// + const Token *getUnexpArgument(unsigned Arg) const; + + /// getArgLength - Given a pointer to an expanded or unexpanded argument, + /// return the number of tokens, not counting the EOF, that make up the + /// argument. + static unsigned getArgLength(const Token *ArgPtr); + + /// getPreExpArgument - Return the pre-expanded form of the specified + /// argument. + const std::vector<Token> & + getPreExpArgument(unsigned Arg, Preprocessor &PP); + + /// getStringifiedArgument - Compute, cache, and return the specified argument + /// that has been 'stringified' as required by the # operator. + const Token &getStringifiedArgument(unsigned ArgNo, Preprocessor &PP); + + /// getNumArguments - Return the number of arguments passed into this macro + /// invocation. + unsigned getNumArguments() const { return NumUnexpArgTokens; } + + + /// isVarargsElidedUse - Return true if this is a C99 style varargs macro + /// invocation and there was no argument specified for the "..." argument. If + /// the argument was specified (even empty) or this isn't a C99 style varargs + /// function, or if in strict mode and the C99 varargs macro had only a ... + /// argument, this returns false. + bool isVarargsElidedUse() const { return VarargsElided; } + + /// StringifyArgument - Implement C99 6.10.3.2p2, converting a sequence of + /// tokens into the literal string token that should be produced by the C # + /// preprocessor operator. If Charify is true, then it should be turned into + /// a character literal for the Microsoft charize (#@) extension. + /// + static Token StringifyArgument(const Token *ArgToks, + Preprocessor &PP, bool Charify = false); +}; + +} // end namespace clang + +#endif diff --git a/lib/Lex/MacroInfo.cpp b/lib/Lex/MacroInfo.cpp new file mode 100644 index 0000000..df89450 --- /dev/null +++ b/lib/Lex/MacroInfo.cpp @@ -0,0 +1,75 @@ +//===--- MacroInfo.cpp - Information about #defined identifiers -----------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements the MacroInfo interface. +// +//===----------------------------------------------------------------------===// + +#include "clang/Lex/MacroInfo.h" +#include "clang/Lex/Preprocessor.h" +using namespace clang; + +MacroInfo::MacroInfo(SourceLocation DefLoc) : Location(DefLoc) { + IsFunctionLike = false; + IsC99Varargs = false; + IsGNUVarargs = false; + IsBuiltinMacro = false; + IsDisabled = false; + IsUsed = true; + + ArgumentList = 0; + NumArguments = 0; +} + +/// isIdenticalTo - Return true if the specified macro definition is equal to +/// this macro in spelling, arguments, and whitespace. This is used to emit +/// duplicate definition warnings. This implements the rules in C99 6.10.3. +/// +bool MacroInfo::isIdenticalTo(const MacroInfo &Other, Preprocessor &PP) const { + // Check # tokens in replacement, number of args, and various flags all match. + if (ReplacementTokens.size() != Other.ReplacementTokens.size() || + getNumArgs() != Other.getNumArgs() || + isFunctionLike() != Other.isFunctionLike() || + isC99Varargs() != Other.isC99Varargs() || + isGNUVarargs() != Other.isGNUVarargs()) + return false; + + // Check arguments. + for (arg_iterator I = arg_begin(), OI = Other.arg_begin(), E = arg_end(); + I != E; ++I, ++OI) + if (*I != *OI) return false; + + // Check all the tokens. + for (unsigned i = 0, e = ReplacementTokens.size(); i != e; ++i) { + const Token &A = ReplacementTokens[i]; + const Token &B = Other.ReplacementTokens[i]; + if (A.getKind() != B.getKind()) + return false; + + // If this isn't the first first token, check that the whitespace and + // start-of-line characteristics match. + if (i != 0 && + (A.isAtStartOfLine() != B.isAtStartOfLine() || + A.hasLeadingSpace() != B.hasLeadingSpace())) + return false; + + // If this is an identifier, it is easy. + if (A.getIdentifierInfo() || B.getIdentifierInfo()) { + if (A.getIdentifierInfo() != B.getIdentifierInfo()) + return false; + continue; + } + + // Otherwise, check the spelling. + if (PP.getSpelling(A) != PP.getSpelling(B)) + return false; + } + + return true; +} diff --git a/lib/Lex/Makefile b/lib/Lex/Makefile new file mode 100644 index 0000000..a2437da --- /dev/null +++ b/lib/Lex/Makefile @@ -0,0 +1,28 @@ +##===- clang/lib/Lex/Makefile ------------------------------*- Makefile -*-===## +# +# The LLVM Compiler Infrastructure +# +# This file is distributed under the University of Illinois Open Source +# License. See LICENSE.TXT for details. +# +##===----------------------------------------------------------------------===## +# +# This implements the Lexer library for the C-Language front-end. +# +##===----------------------------------------------------------------------===## + +LEVEL = ../../../.. +include $(LEVEL)/Makefile.config + +LIBRARYNAME := clangLex +BUILD_ARCHIVE = 1 +CXXFLAGS = -fno-rtti + +ifeq ($(ARCH),PowerPC) +CXXFLAGS += -maltivec +endif + +CPPFLAGS += -I$(PROJ_SRC_DIR)/../../include -I$(PROJ_OBJ_DIR)/../../include + +include $(LEVEL)/Makefile.common + diff --git a/lib/Lex/PPCaching.cpp b/lib/Lex/PPCaching.cpp new file mode 100644 index 0000000..53aa09c --- /dev/null +++ b/lib/Lex/PPCaching.cpp @@ -0,0 +1,113 @@ +//===--- PPCaching.cpp - Handle caching lexed tokens ----------------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements pieces of the Preprocessor interface that manage the +// caching of lexed tokens. +// +//===----------------------------------------------------------------------===// + +#include "clang/Lex/Preprocessor.h" +using namespace clang; + +/// EnableBacktrackAtThisPos - From the point that this method is called, and +/// until CommitBacktrackedTokens() or Backtrack() is called, the Preprocessor +/// keeps track of the lexed tokens so that a subsequent Backtrack() call will +/// make the Preprocessor re-lex the same tokens. +/// +/// Nested backtracks are allowed, meaning that EnableBacktrackAtThisPos can +/// be called multiple times and CommitBacktrackedTokens/Backtrack calls will +/// be combined with the EnableBacktrackAtThisPos calls in reverse order. +void Preprocessor::EnableBacktrackAtThisPos() { + BacktrackPositions.push_back(CachedLexPos); + EnterCachingLexMode(); +} + +/// CommitBacktrackedTokens - Disable the last EnableBacktrackAtThisPos call. +void Preprocessor::CommitBacktrackedTokens() { + assert(!BacktrackPositions.empty() + && "EnableBacktrackAtThisPos was not called!"); + BacktrackPositions.pop_back(); +} + +/// Backtrack - Make Preprocessor re-lex the tokens that were lexed since +/// EnableBacktrackAtThisPos() was previously called. +void Preprocessor::Backtrack() { + assert(!BacktrackPositions.empty() + && "EnableBacktrackAtThisPos was not called!"); + CachedLexPos = BacktrackPositions.back(); + BacktrackPositions.pop_back(); +} + +void Preprocessor::CachingLex(Token &Result) { + if (CachedLexPos < CachedTokens.size()) { + Result = CachedTokens[CachedLexPos++]; + return; + } + + ExitCachingLexMode(); + Lex(Result); + + if (!isBacktrackEnabled()) { + // All cached tokens were consumed. + CachedTokens.clear(); + CachedLexPos = 0; + return; + } + + // We should cache the lexed token. + + EnterCachingLexMode(); + if (Result.isNot(tok::eof)) { + CachedTokens.push_back(Result); + ++CachedLexPos; + } +} + +void Preprocessor::EnterCachingLexMode() { + if (InCachingLexMode()) + return; + + PushIncludeMacroStack(); +} + + +const Token &Preprocessor::PeekAhead(unsigned N) { + assert(CachedLexPos + N > CachedTokens.size() && "Confused caching."); + ExitCachingLexMode(); + for (unsigned C = CachedLexPos + N - CachedTokens.size(); C > 0; --C) { + CachedTokens.push_back(Token()); + Lex(CachedTokens.back()); + } + EnterCachingLexMode(); + return CachedTokens.back(); +} + +void Preprocessor::AnnotatePreviousCachedTokens(const Token &Tok) { + assert(Tok.isAnnotation() && "Expected annotation token"); + assert(CachedLexPos != 0 && "Expected to have some cached tokens"); + assert(CachedTokens[CachedLexPos-1].getLocation() == Tok.getAnnotationEndLoc() + && "The annotation should be until the most recent cached token"); + + // Start from the end of the cached tokens list and look for the token + // that is the beginning of the annotation token. + for (CachedTokensTy::size_type i = CachedLexPos; i != 0; --i) { + CachedTokensTy::iterator AnnotBegin = CachedTokens.begin() + i-1; + if (AnnotBegin->getLocation() == Tok.getLocation()) { + assert((BacktrackPositions.empty() || BacktrackPositions.back() < i) && + "The backtrack pos points inside the annotated tokens!"); + // Replace the cached tokens with the single annotation token. + CachedTokens.erase(AnnotBegin + 1, CachedTokens.begin() + CachedLexPos); + *AnnotBegin = Tok; + CachedLexPos = i; + return; + } + } + + assert(0&&"Didn't find the first token represented by the annotation token!"); +} diff --git a/lib/Lex/PPDirectives.cpp b/lib/Lex/PPDirectives.cpp new file mode 100644 index 0000000..af59ded --- /dev/null +++ b/lib/Lex/PPDirectives.cpp @@ -0,0 +1,1665 @@ +//===--- PPDirectives.cpp - Directive Handling for Preprocessor -----------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements # directive processing for the Preprocessor. +// +//===----------------------------------------------------------------------===// + +#include "clang/Lex/Preprocessor.h" +#include "clang/Lex/LiteralSupport.h" +#include "clang/Lex/HeaderSearch.h" +#include "clang/Lex/MacroInfo.h" +#include "clang/Lex/LexDiagnostic.h" +#include "clang/Basic/SourceManager.h" +#include "llvm/ADT/APInt.h" +using namespace clang; + +//===----------------------------------------------------------------------===// +// Utility Methods for Preprocessor Directive Handling. +//===----------------------------------------------------------------------===// + +MacroInfo *Preprocessor::AllocateMacroInfo(SourceLocation L) { + MacroInfo *MI; + + if (!MICache.empty()) { + MI = MICache.back(); + MICache.pop_back(); + } else + MI = (MacroInfo*) BP.Allocate<MacroInfo>(); + new (MI) MacroInfo(L); + return MI; +} + +/// ReleaseMacroInfo - Release the specified MacroInfo. This memory will +/// be reused for allocating new MacroInfo objects. +void Preprocessor::ReleaseMacroInfo(MacroInfo* MI) { + MICache.push_back(MI); + MI->FreeArgumentList(BP); +} + + +/// DiscardUntilEndOfDirective - Read and discard all tokens remaining on the +/// current line until the tok::eom token is found. +void Preprocessor::DiscardUntilEndOfDirective() { + Token Tmp; + do { + LexUnexpandedToken(Tmp); + } while (Tmp.isNot(tok::eom)); +} + +/// ReadMacroName - Lex and validate a macro name, which occurs after a +/// #define or #undef. This sets the token kind to eom and discards the rest +/// of the macro line if the macro name is invalid. isDefineUndef is 1 if +/// this is due to a a #define, 2 if #undef directive, 0 if it is something +/// else (e.g. #ifdef). +void Preprocessor::ReadMacroName(Token &MacroNameTok, char isDefineUndef) { + // Read the token, don't allow macro expansion on it. + LexUnexpandedToken(MacroNameTok); + + // Missing macro name? + if (MacroNameTok.is(tok::eom)) { + Diag(MacroNameTok, diag::err_pp_missing_macro_name); + return; + } + + IdentifierInfo *II = MacroNameTok.getIdentifierInfo(); + if (II == 0) { + std::string Spelling = getSpelling(MacroNameTok); + const IdentifierInfo &Info = Identifiers.get(Spelling); + if (Info.isCPlusPlusOperatorKeyword()) + // C++ 2.5p2: Alternative tokens behave the same as its primary token + // except for their spellings. + Diag(MacroNameTok, diag::err_pp_operator_used_as_macro_name) << Spelling; + else + Diag(MacroNameTok, diag::err_pp_macro_not_identifier); + // Fall through on error. + } else if (isDefineUndef && II->getPPKeywordID() == tok::pp_defined) { + // Error if defining "defined": C99 6.10.8.4. + Diag(MacroNameTok, diag::err_defined_macro_name); + } else if (isDefineUndef && II->hasMacroDefinition() && + getMacroInfo(II)->isBuiltinMacro()) { + // Error if defining "__LINE__" and other builtins: C99 6.10.8.4. + if (isDefineUndef == 1) + Diag(MacroNameTok, diag::pp_redef_builtin_macro); + else + Diag(MacroNameTok, diag::pp_undef_builtin_macro); + } else { + // Okay, we got a good identifier node. Return it. + return; + } + + // Invalid macro name, read and discard the rest of the line. Then set the + // token kind to tok::eom. + MacroNameTok.setKind(tok::eom); + return DiscardUntilEndOfDirective(); +} + +/// CheckEndOfDirective - Ensure that the next token is a tok::eom token. If +/// not, emit a diagnostic and consume up until the eom. If EnableMacros is +/// true, then we consider macros that expand to zero tokens as being ok. +void Preprocessor::CheckEndOfDirective(const char *DirType, bool EnableMacros) { + Token Tmp; + // Lex unexpanded tokens for most directives: macros might expand to zero + // tokens, causing us to miss diagnosing invalid lines. Some directives (like + // #line) allow empty macros. + if (EnableMacros) + Lex(Tmp); + else + LexUnexpandedToken(Tmp); + + // There should be no tokens after the directive, but we allow them as an + // extension. + while (Tmp.is(tok::comment)) // Skip comments in -C mode. + LexUnexpandedToken(Tmp); + + if (Tmp.isNot(tok::eom)) { + // Add a fixit in GNU/C99/C++ mode. Don't offer a fixit for strict-C89, + // because it is more trouble than it is worth to insert /**/ and check that + // there is no /**/ in the range also. + CodeModificationHint FixItHint; + if (Features.GNUMode || Features.C99 || Features.CPlusPlus) + FixItHint = CodeModificationHint::CreateInsertion(Tmp.getLocation(),"//"); + Diag(Tmp, diag::ext_pp_extra_tokens_at_eol) << DirType << FixItHint; + DiscardUntilEndOfDirective(); + } +} + + + +/// SkipExcludedConditionalBlock - We just read a #if or related directive and +/// decided that the subsequent tokens are in the #if'd out portion of the +/// file. Lex the rest of the file, until we see an #endif. If +/// FoundNonSkipPortion is true, then we have already emitted code for part of +/// this #if directive, so #else/#elif blocks should never be entered. If ElseOk +/// is true, then #else directives are ok, if not, then we have already seen one +/// so a #else directive is a duplicate. When this returns, the caller can lex +/// the first valid token. +void Preprocessor::SkipExcludedConditionalBlock(SourceLocation IfTokenLoc, + bool FoundNonSkipPortion, + bool FoundElse) { + ++NumSkipped; + assert(CurTokenLexer == 0 && CurPPLexer && "Lexing a macro, not a file?"); + + CurPPLexer->pushConditionalLevel(IfTokenLoc, /*isSkipping*/false, + FoundNonSkipPortion, FoundElse); + + if (CurPTHLexer) { + PTHSkipExcludedConditionalBlock(); + return; + } + + // Enter raw mode to disable identifier lookup (and thus macro expansion), + // disabling warnings, etc. + CurPPLexer->LexingRawMode = true; + Token Tok; + while (1) { + if (CurLexer) + CurLexer->Lex(Tok); + else + CurPTHLexer->Lex(Tok); + + // If this is the end of the buffer, we have an error. + if (Tok.is(tok::eof)) { + // Emit errors for each unterminated conditional on the stack, including + // the current one. + while (!CurPPLexer->ConditionalStack.empty()) { + Diag(CurPPLexer->ConditionalStack.back().IfLoc, + diag::err_pp_unterminated_conditional); + CurPPLexer->ConditionalStack.pop_back(); + } + + // Just return and let the caller lex after this #include. + break; + } + + // If this token is not a preprocessor directive, just skip it. + if (Tok.isNot(tok::hash) || !Tok.isAtStartOfLine()) + continue; + + // We just parsed a # character at the start of a line, so we're in + // directive mode. Tell the lexer this so any newlines we see will be + // converted into an EOM token (this terminates the macro). + CurPPLexer->ParsingPreprocessorDirective = true; + if (CurLexer) CurLexer->SetCommentRetentionState(false); + + + // Read the next token, the directive flavor. + LexUnexpandedToken(Tok); + + // If this isn't an identifier directive (e.g. is "# 1\n" or "#\n", or + // something bogus), skip it. + if (Tok.isNot(tok::identifier)) { + CurPPLexer->ParsingPreprocessorDirective = false; + // Restore comment saving mode. + if (CurLexer) CurLexer->SetCommentRetentionState(KeepComments); + continue; + } + + // If the first letter isn't i or e, it isn't intesting to us. We know that + // this is safe in the face of spelling differences, because there is no way + // to spell an i/e in a strange way that is another letter. Skipping this + // allows us to avoid looking up the identifier info for #define/#undef and + // other common directives. + const char *RawCharData = SourceMgr.getCharacterData(Tok.getLocation()); + char FirstChar = RawCharData[0]; + if (FirstChar >= 'a' && FirstChar <= 'z' && + FirstChar != 'i' && FirstChar != 'e') { + CurPPLexer->ParsingPreprocessorDirective = false; + // Restore comment saving mode. + if (CurLexer) CurLexer->SetCommentRetentionState(KeepComments); + continue; + } + + // Get the identifier name without trigraphs or embedded newlines. Note + // that we can't use Tok.getIdentifierInfo() because its lookup is disabled + // when skipping. + // TODO: could do this with zero copies in the no-clean case by using + // strncmp below. + char Directive[20]; + unsigned IdLen; + if (!Tok.needsCleaning() && Tok.getLength() < 20) { + IdLen = Tok.getLength(); + memcpy(Directive, RawCharData, IdLen); + Directive[IdLen] = 0; + } else { + std::string DirectiveStr = getSpelling(Tok); + IdLen = DirectiveStr.size(); + if (IdLen >= 20) { + CurPPLexer->ParsingPreprocessorDirective = false; + // Restore comment saving mode. + if (CurLexer) CurLexer->SetCommentRetentionState(KeepComments); + continue; + } + memcpy(Directive, &DirectiveStr[0], IdLen); + Directive[IdLen] = 0; + FirstChar = Directive[0]; + } + + if (FirstChar == 'i' && Directive[1] == 'f') { + if ((IdLen == 2) || // "if" + (IdLen == 5 && !strcmp(Directive+2, "def")) || // "ifdef" + (IdLen == 6 && !strcmp(Directive+2, "ndef"))) { // "ifndef" + // We know the entire #if/#ifdef/#ifndef block will be skipped, don't + // bother parsing the condition. + DiscardUntilEndOfDirective(); + CurPPLexer->pushConditionalLevel(Tok.getLocation(), /*wasskipping*/true, + /*foundnonskip*/false, + /*fnddelse*/false); + } + } else if (FirstChar == 'e') { + if (IdLen == 5 && !strcmp(Directive+1, "ndif")) { // "endif" + CheckEndOfDirective("endif"); + PPConditionalInfo CondInfo; + CondInfo.WasSkipping = true; // Silence bogus warning. + bool InCond = CurPPLexer->popConditionalLevel(CondInfo); + InCond = InCond; // Silence warning in no-asserts mode. + assert(!InCond && "Can't be skipping if not in a conditional!"); + + // If we popped the outermost skipping block, we're done skipping! + if (!CondInfo.WasSkipping) + break; + } else if (IdLen == 4 && !strcmp(Directive+1, "lse")) { // "else". + // #else directive in a skipping conditional. If not in some other + // skipping conditional, and if #else hasn't already been seen, enter it + // as a non-skipping conditional. + DiscardUntilEndOfDirective(); // C99 6.10p4. + PPConditionalInfo &CondInfo = CurPPLexer->peekConditionalLevel(); + + // If this is a #else with a #else before it, report the error. + if (CondInfo.FoundElse) Diag(Tok, diag::pp_err_else_after_else); + + // Note that we've seen a #else in this conditional. + CondInfo.FoundElse = true; + + // If the conditional is at the top level, and the #if block wasn't + // entered, enter the #else block now. + if (!CondInfo.WasSkipping && !CondInfo.FoundNonSkip) { + CondInfo.FoundNonSkip = true; + break; + } + } else if (IdLen == 4 && !strcmp(Directive+1, "lif")) { // "elif". + PPConditionalInfo &CondInfo = CurPPLexer->peekConditionalLevel(); + + bool ShouldEnter; + // If this is in a skipping block or if we're already handled this #if + // block, don't bother parsing the condition. + if (CondInfo.WasSkipping || CondInfo.FoundNonSkip) { + DiscardUntilEndOfDirective(); + ShouldEnter = false; + } else { + // Restore the value of LexingRawMode so that identifiers are + // looked up, etc, inside the #elif expression. + assert(CurPPLexer->LexingRawMode && "We have to be skipping here!"); + CurPPLexer->LexingRawMode = false; + IdentifierInfo *IfNDefMacro = 0; + ShouldEnter = EvaluateDirectiveExpression(IfNDefMacro); + CurPPLexer->LexingRawMode = true; + } + + // If this is a #elif with a #else before it, report the error. + if (CondInfo.FoundElse) Diag(Tok, diag::pp_err_elif_after_else); + + // If this condition is true, enter it! + if (ShouldEnter) { + CondInfo.FoundNonSkip = true; + break; + } + } + } + + CurPPLexer->ParsingPreprocessorDirective = false; + // Restore comment saving mode. + if (CurLexer) CurLexer->SetCommentRetentionState(KeepComments); + } + + // Finally, if we are out of the conditional (saw an #endif or ran off the end + // of the file, just stop skipping and return to lexing whatever came after + // the #if block. + CurPPLexer->LexingRawMode = false; +} + +void Preprocessor::PTHSkipExcludedConditionalBlock() { + + while(1) { + assert(CurPTHLexer); + assert(CurPTHLexer->LexingRawMode == false); + + // Skip to the next '#else', '#elif', or #endif. + if (CurPTHLexer->SkipBlock()) { + // We have reached an #endif. Both the '#' and 'endif' tokens + // have been consumed by the PTHLexer. Just pop off the condition level. + PPConditionalInfo CondInfo; + bool InCond = CurPTHLexer->popConditionalLevel(CondInfo); + InCond = InCond; // Silence warning in no-asserts mode. + assert(!InCond && "Can't be skipping if not in a conditional!"); + break; + } + + // We have reached a '#else' or '#elif'. Lex the next token to get + // the directive flavor. + Token Tok; + LexUnexpandedToken(Tok); + + // We can actually look up the IdentifierInfo here since we aren't in + // raw mode. + tok::PPKeywordKind K = Tok.getIdentifierInfo()->getPPKeywordID(); + + if (K == tok::pp_else) { + // #else: Enter the else condition. We aren't in a nested condition + // since we skip those. We're always in the one matching the last + // blocked we skipped. + PPConditionalInfo &CondInfo = CurPTHLexer->peekConditionalLevel(); + // Note that we've seen a #else in this conditional. + CondInfo.FoundElse = true; + + // If the #if block wasn't entered then enter the #else block now. + if (!CondInfo.FoundNonSkip) { + CondInfo.FoundNonSkip = true; + + // Scan until the eom token. + CurPTHLexer->ParsingPreprocessorDirective = true; + DiscardUntilEndOfDirective(); + CurPTHLexer->ParsingPreprocessorDirective = false; + + break; + } + + // Otherwise skip this block. + continue; + } + + assert(K == tok::pp_elif); + PPConditionalInfo &CondInfo = CurPTHLexer->peekConditionalLevel(); + + // If this is a #elif with a #else before it, report the error. + if (CondInfo.FoundElse) + Diag(Tok, diag::pp_err_elif_after_else); + + // If this is in a skipping block or if we're already handled this #if + // block, don't bother parsing the condition. We just skip this block. + if (CondInfo.FoundNonSkip) + continue; + + // Evaluate the condition of the #elif. + IdentifierInfo *IfNDefMacro = 0; + CurPTHLexer->ParsingPreprocessorDirective = true; + bool ShouldEnter = EvaluateDirectiveExpression(IfNDefMacro); + CurPTHLexer->ParsingPreprocessorDirective = false; + + // If this condition is true, enter it! + if (ShouldEnter) { + CondInfo.FoundNonSkip = true; + break; + } + + // Otherwise, skip this block and go to the next one. + continue; + } +} + +/// LookupFile - Given a "foo" or <foo> reference, look up the indicated file, +/// return null on failure. isAngled indicates whether the file reference is +/// for system #include's or not (i.e. using <> instead of ""). +const FileEntry *Preprocessor::LookupFile(const char *FilenameStart, + const char *FilenameEnd, + bool isAngled, + const DirectoryLookup *FromDir, + const DirectoryLookup *&CurDir) { + // If the header lookup mechanism may be relative to the current file, pass in + // info about where the current file is. + const FileEntry *CurFileEnt = 0; + if (!FromDir) { + FileID FID = getCurrentFileLexer()->getFileID(); + CurFileEnt = SourceMgr.getFileEntryForID(FID); + + // If there is no file entry associated with this file, it must be the + // predefines buffer. Any other file is not lexed with a normal lexer, so + // it won't be scanned for preprocessor directives. If we have the + // predefines buffer, resolve #include references (which come from the + // -include command line argument) as if they came from the main file, this + // affects file lookup etc. + if (CurFileEnt == 0) { + FID = SourceMgr.getMainFileID(); + CurFileEnt = SourceMgr.getFileEntryForID(FID); + } + } + + // Do a standard file entry lookup. + CurDir = CurDirLookup; + const FileEntry *FE = + HeaderInfo.LookupFile(FilenameStart, FilenameEnd, + isAngled, FromDir, CurDir, CurFileEnt); + if (FE) return FE; + + // Otherwise, see if this is a subframework header. If so, this is relative + // to one of the headers on the #include stack. Walk the list of the current + // headers on the #include stack and pass them to HeaderInfo. + if (IsFileLexer()) { + if ((CurFileEnt = SourceMgr.getFileEntryForID(CurPPLexer->getFileID()))) + if ((FE = HeaderInfo.LookupSubframeworkHeader(FilenameStart, FilenameEnd, + CurFileEnt))) + return FE; + } + + for (unsigned i = 0, e = IncludeMacroStack.size(); i != e; ++i) { + IncludeStackInfo &ISEntry = IncludeMacroStack[e-i-1]; + if (IsFileLexer(ISEntry)) { + if ((CurFileEnt = + SourceMgr.getFileEntryForID(ISEntry.ThePPLexer->getFileID()))) + if ((FE = HeaderInfo.LookupSubframeworkHeader(FilenameStart, + FilenameEnd, CurFileEnt))) + return FE; + } + } + + // Otherwise, we really couldn't find the file. + return 0; +} + + +//===----------------------------------------------------------------------===// +// Preprocessor Directive Handling. +//===----------------------------------------------------------------------===// + +/// HandleDirective - This callback is invoked when the lexer sees a # token +/// at the start of a line. This consumes the directive, modifies the +/// lexer/preprocessor state, and advances the lexer(s) so that the next token +/// read is the correct one. +void Preprocessor::HandleDirective(Token &Result) { + // FIXME: Traditional: # with whitespace before it not recognized by K&R? + + // We just parsed a # character at the start of a line, so we're in directive + // mode. Tell the lexer this so any newlines we see will be converted into an + // EOM token (which terminates the directive). + CurPPLexer->ParsingPreprocessorDirective = true; + + ++NumDirectives; + + // We are about to read a token. For the multiple-include optimization FA to + // work, we have to remember if we had read any tokens *before* this + // pp-directive. + bool ReadAnyTokensBeforeDirective = CurPPLexer->MIOpt.getHasReadAnyTokensVal(); + + // Save the '#' token in case we need to return it later. + Token SavedHash = Result; + + // Read the next token, the directive flavor. This isn't expanded due to + // C99 6.10.3p8. + LexUnexpandedToken(Result); + + // C99 6.10.3p11: Is this preprocessor directive in macro invocation? e.g.: + // #define A(x) #x + // A(abc + // #warning blah + // def) + // If so, the user is relying on non-portable behavior, emit a diagnostic. + if (InMacroArgs) + Diag(Result, diag::ext_embedded_directive); + +TryAgain: + switch (Result.getKind()) { + case tok::eom: + return; // null directive. + case tok::comment: + // Handle stuff like "# /*foo*/ define X" in -E -C mode. + LexUnexpandedToken(Result); + goto TryAgain; + + case tok::numeric_constant: // # 7 GNU line marker directive. + if (getLangOptions().AsmPreprocessor) + break; // # 4 is not a preprocessor directive in .S files. + return HandleDigitDirective(Result); + default: + IdentifierInfo *II = Result.getIdentifierInfo(); + if (II == 0) break; // Not an identifier. + + // Ask what the preprocessor keyword ID is. + switch (II->getPPKeywordID()) { + default: break; + // C99 6.10.1 - Conditional Inclusion. + case tok::pp_if: + return HandleIfDirective(Result, ReadAnyTokensBeforeDirective); + case tok::pp_ifdef: + return HandleIfdefDirective(Result, false, true/*not valid for miopt*/); + case tok::pp_ifndef: + return HandleIfdefDirective(Result, true, ReadAnyTokensBeforeDirective); + case tok::pp_elif: + return HandleElifDirective(Result); + case tok::pp_else: + return HandleElseDirective(Result); + case tok::pp_endif: + return HandleEndifDirective(Result); + + // C99 6.10.2 - Source File Inclusion. + case tok::pp_include: + return HandleIncludeDirective(Result); // Handle #include. + case tok::pp___include_macros: + return HandleIncludeMacrosDirective(Result); // Handle -imacros. + + // C99 6.10.3 - Macro Replacement. + case tok::pp_define: + return HandleDefineDirective(Result); + case tok::pp_undef: + return HandleUndefDirective(Result); + + // C99 6.10.4 - Line Control. + case tok::pp_line: + return HandleLineDirective(Result); + + // C99 6.10.5 - Error Directive. + case tok::pp_error: + return HandleUserDiagnosticDirective(Result, false); + + // C99 6.10.6 - Pragma Directive. + case tok::pp_pragma: + return HandlePragmaDirective(); + + // GNU Extensions. + case tok::pp_import: + return HandleImportDirective(Result); + case tok::pp_include_next: + return HandleIncludeNextDirective(Result); + + case tok::pp_warning: + Diag(Result, diag::ext_pp_warning_directive); + return HandleUserDiagnosticDirective(Result, true); + case tok::pp_ident: + return HandleIdentSCCSDirective(Result); + case tok::pp_sccs: + return HandleIdentSCCSDirective(Result); + case tok::pp_assert: + //isExtension = true; // FIXME: implement #assert + break; + case tok::pp_unassert: + //isExtension = true; // FIXME: implement #unassert + break; + } + break; + } + + // If this is a .S file, treat unknown # directives as non-preprocessor + // directives. This is important because # may be a comment or introduce + // various pseudo-ops. Just return the # token and push back the following + // token to be lexed next time. + if (getLangOptions().AsmPreprocessor) { + Token *Toks = new Token[2](); + // Return the # and the token after it. + Toks[0] = SavedHash; + Toks[1] = Result; + // Enter this token stream so that we re-lex the tokens. Make sure to + // enable macro expansion, in case the token after the # is an identifier + // that is expanded. + EnterTokenStream(Toks, 2, false, true); + return; + } + + // If we reached here, the preprocessing token is not valid! + Diag(Result, diag::err_pp_invalid_directive); + + // Read the rest of the PP line. + DiscardUntilEndOfDirective(); + + // Okay, we're done parsing the directive. +} + +/// GetLineValue - Convert a numeric token into an unsigned value, emitting +/// Diagnostic DiagID if it is invalid, and returning the value in Val. +static bool GetLineValue(Token &DigitTok, unsigned &Val, + unsigned DiagID, Preprocessor &PP) { + if (DigitTok.isNot(tok::numeric_constant)) { + PP.Diag(DigitTok, DiagID); + + if (DigitTok.isNot(tok::eom)) + PP.DiscardUntilEndOfDirective(); + return true; + } + + llvm::SmallString<64> IntegerBuffer; + IntegerBuffer.resize(DigitTok.getLength()); + const char *DigitTokBegin = &IntegerBuffer[0]; + unsigned ActualLength = PP.getSpelling(DigitTok, DigitTokBegin); + + // Verify that we have a simple digit-sequence, and compute the value. This + // is always a simple digit string computed in decimal, so we do this manually + // here. + Val = 0; + for (unsigned i = 0; i != ActualLength; ++i) { + if (!isdigit(DigitTokBegin[i])) { + PP.Diag(PP.AdvanceToTokenCharacter(DigitTok.getLocation(), i), + diag::err_pp_line_digit_sequence); + PP.DiscardUntilEndOfDirective(); + return true; + } + + unsigned NextVal = Val*10+(DigitTokBegin[i]-'0'); + if (NextVal < Val) { // overflow. + PP.Diag(DigitTok, DiagID); + PP.DiscardUntilEndOfDirective(); + return true; + } + Val = NextVal; + } + + // Reject 0, this is needed both by #line numbers and flags. + if (Val == 0) { + PP.Diag(DigitTok, DiagID); + PP.DiscardUntilEndOfDirective(); + return true; + } + + if (DigitTokBegin[0] == '0') + PP.Diag(DigitTok.getLocation(), diag::warn_pp_line_decimal); + + return false; +} + +/// HandleLineDirective - Handle #line directive: C99 6.10.4. The two +/// acceptable forms are: +/// # line digit-sequence +/// # line digit-sequence "s-char-sequence" +void Preprocessor::HandleLineDirective(Token &Tok) { + // Read the line # and string argument. Per C99 6.10.4p5, these tokens are + // expanded. + Token DigitTok; + Lex(DigitTok); + + // Validate the number and convert it to an unsigned. + unsigned LineNo; + if (GetLineValue(DigitTok, LineNo, diag::err_pp_line_requires_integer,*this)) + return; + + // Enforce C99 6.10.4p3: "The digit sequence shall not specify ... a + // number greater than 2147483647". C90 requires that the line # be <= 32767. + unsigned LineLimit = Features.C99 ? 2147483648U : 32768U; + if (LineNo >= LineLimit) + Diag(DigitTok, diag::ext_pp_line_too_big) << LineLimit; + + int FilenameID = -1; + Token StrTok; + Lex(StrTok); + + // If the StrTok is "eom", then it wasn't present. Otherwise, it must be a + // string followed by eom. + if (StrTok.is(tok::eom)) + ; // ok + else if (StrTok.isNot(tok::string_literal)) { + Diag(StrTok, diag::err_pp_line_invalid_filename); + DiscardUntilEndOfDirective(); + return; + } else { + // Parse and validate the string, converting it into a unique ID. + StringLiteralParser Literal(&StrTok, 1, *this); + assert(!Literal.AnyWide && "Didn't allow wide strings in"); + if (Literal.hadError) + return DiscardUntilEndOfDirective(); + if (Literal.Pascal) { + Diag(StrTok, diag::err_pp_linemarker_invalid_filename); + return DiscardUntilEndOfDirective(); + } + FilenameID = SourceMgr.getLineTableFilenameID(Literal.GetString(), + Literal.GetStringLength()); + + // Verify that there is nothing after the string, other than EOM. Because + // of C99 6.10.4p5, macros that expand to empty tokens are ok. + CheckEndOfDirective("line", true); + } + + SourceMgr.AddLineNote(DigitTok.getLocation(), LineNo, FilenameID); + + if (Callbacks) + Callbacks->FileChanged(DigitTok.getLocation(), PPCallbacks::RenameFile, + SrcMgr::C_User); +} + +/// ReadLineMarkerFlags - Parse and validate any flags at the end of a GNU line +/// marker directive. +static bool ReadLineMarkerFlags(bool &IsFileEntry, bool &IsFileExit, + bool &IsSystemHeader, bool &IsExternCHeader, + Preprocessor &PP) { + unsigned FlagVal; + Token FlagTok; + PP.Lex(FlagTok); + if (FlagTok.is(tok::eom)) return false; + if (GetLineValue(FlagTok, FlagVal, diag::err_pp_linemarker_invalid_flag, PP)) + return true; + + if (FlagVal == 1) { + IsFileEntry = true; + + PP.Lex(FlagTok); + if (FlagTok.is(tok::eom)) return false; + if (GetLineValue(FlagTok, FlagVal, diag::err_pp_linemarker_invalid_flag,PP)) + return true; + } else if (FlagVal == 2) { + IsFileExit = true; + + SourceManager &SM = PP.getSourceManager(); + // If we are leaving the current presumed file, check to make sure the + // presumed include stack isn't empty! + FileID CurFileID = + SM.getDecomposedInstantiationLoc(FlagTok.getLocation()).first; + PresumedLoc PLoc = SM.getPresumedLoc(FlagTok.getLocation()); + + // If there is no include loc (main file) or if the include loc is in a + // different physical file, then we aren't in a "1" line marker flag region. + SourceLocation IncLoc = PLoc.getIncludeLoc(); + if (IncLoc.isInvalid() || + SM.getDecomposedInstantiationLoc(IncLoc).first != CurFileID) { + PP.Diag(FlagTok, diag::err_pp_linemarker_invalid_pop); + PP.DiscardUntilEndOfDirective(); + return true; + } + + PP.Lex(FlagTok); + if (FlagTok.is(tok::eom)) return false; + if (GetLineValue(FlagTok, FlagVal, diag::err_pp_linemarker_invalid_flag,PP)) + return true; + } + + // We must have 3 if there are still flags. + if (FlagVal != 3) { + PP.Diag(FlagTok, diag::err_pp_linemarker_invalid_flag); + PP.DiscardUntilEndOfDirective(); + return true; + } + + IsSystemHeader = true; + + PP.Lex(FlagTok); + if (FlagTok.is(tok::eom)) return false; + if (GetLineValue(FlagTok, FlagVal, diag::err_pp_linemarker_invalid_flag, PP)) + return true; + + // We must have 4 if there is yet another flag. + if (FlagVal != 4) { + PP.Diag(FlagTok, diag::err_pp_linemarker_invalid_flag); + PP.DiscardUntilEndOfDirective(); + return true; + } + + IsExternCHeader = true; + + PP.Lex(FlagTok); + if (FlagTok.is(tok::eom)) return false; + + // There are no more valid flags here. + PP.Diag(FlagTok, diag::err_pp_linemarker_invalid_flag); + PP.DiscardUntilEndOfDirective(); + return true; +} + +/// HandleDigitDirective - Handle a GNU line marker directive, whose syntax is +/// one of the following forms: +/// +/// # 42 +/// # 42 "file" ('1' | '2')? +/// # 42 "file" ('1' | '2')? '3' '4'? +/// +void Preprocessor::HandleDigitDirective(Token &DigitTok) { + // Validate the number and convert it to an unsigned. GNU does not have a + // line # limit other than it fit in 32-bits. + unsigned LineNo; + if (GetLineValue(DigitTok, LineNo, diag::err_pp_linemarker_requires_integer, + *this)) + return; + + Token StrTok; + Lex(StrTok); + + bool IsFileEntry = false, IsFileExit = false; + bool IsSystemHeader = false, IsExternCHeader = false; + int FilenameID = -1; + + // If the StrTok is "eom", then it wasn't present. Otherwise, it must be a + // string followed by eom. + if (StrTok.is(tok::eom)) + ; // ok + else if (StrTok.isNot(tok::string_literal)) { + Diag(StrTok, diag::err_pp_linemarker_invalid_filename); + return DiscardUntilEndOfDirective(); + } else { + // Parse and validate the string, converting it into a unique ID. + StringLiteralParser Literal(&StrTok, 1, *this); + assert(!Literal.AnyWide && "Didn't allow wide strings in"); + if (Literal.hadError) + return DiscardUntilEndOfDirective(); + if (Literal.Pascal) { + Diag(StrTok, diag::err_pp_linemarker_invalid_filename); + return DiscardUntilEndOfDirective(); + } + FilenameID = SourceMgr.getLineTableFilenameID(Literal.GetString(), + Literal.GetStringLength()); + + // If a filename was present, read any flags that are present. + if (ReadLineMarkerFlags(IsFileEntry, IsFileExit, + IsSystemHeader, IsExternCHeader, *this)) + return; + } + + // Create a line note with this information. + SourceMgr.AddLineNote(DigitTok.getLocation(), LineNo, FilenameID, + IsFileEntry, IsFileExit, + IsSystemHeader, IsExternCHeader); + + // If the preprocessor has callbacks installed, notify them of the #line + // change. This is used so that the line marker comes out in -E mode for + // example. + if (Callbacks) { + PPCallbacks::FileChangeReason Reason = PPCallbacks::RenameFile; + if (IsFileEntry) + Reason = PPCallbacks::EnterFile; + else if (IsFileExit) + Reason = PPCallbacks::ExitFile; + SrcMgr::CharacteristicKind FileKind = SrcMgr::C_User; + if (IsExternCHeader) + FileKind = SrcMgr::C_ExternCSystem; + else if (IsSystemHeader) + FileKind = SrcMgr::C_System; + + Callbacks->FileChanged(DigitTok.getLocation(), Reason, FileKind); + } +} + + +/// HandleUserDiagnosticDirective - Handle a #warning or #error directive. +/// +void Preprocessor::HandleUserDiagnosticDirective(Token &Tok, + bool isWarning) { + // PTH doesn't emit #warning or #error directives. + if (CurPTHLexer) + return CurPTHLexer->DiscardToEndOfLine(); + + // Read the rest of the line raw. We do this because we don't want macros + // to be expanded and we don't require that the tokens be valid preprocessing + // tokens. For example, this is allowed: "#warning ` 'foo". GCC does + // collapse multiple consequtive white space between tokens, but this isn't + // specified by the standard. + std::string Message = CurLexer->ReadToEndOfLine(); + if (isWarning) + Diag(Tok, diag::pp_hash_warning) << Message; + else + Diag(Tok, diag::err_pp_hash_error) << Message; +} + +/// HandleIdentSCCSDirective - Handle a #ident/#sccs directive. +/// +void Preprocessor::HandleIdentSCCSDirective(Token &Tok) { + // Yes, this directive is an extension. + Diag(Tok, diag::ext_pp_ident_directive); + + // Read the string argument. + Token StrTok; + Lex(StrTok); + + // If the token kind isn't a string, it's a malformed directive. + if (StrTok.isNot(tok::string_literal) && + StrTok.isNot(tok::wide_string_literal)) { + Diag(StrTok, diag::err_pp_malformed_ident); + if (StrTok.isNot(tok::eom)) + DiscardUntilEndOfDirective(); + return; + } + + // Verify that there is nothing after the string, other than EOM. + CheckEndOfDirective("ident"); + + if (Callbacks) + Callbacks->Ident(Tok.getLocation(), getSpelling(StrTok)); +} + +//===----------------------------------------------------------------------===// +// Preprocessor Include Directive Handling. +//===----------------------------------------------------------------------===// + +/// GetIncludeFilenameSpelling - Turn the specified lexer token into a fully +/// checked and spelled filename, e.g. as an operand of #include. This returns +/// true if the input filename was in <>'s or false if it were in ""'s. The +/// caller is expected to provide a buffer that is large enough to hold the +/// spelling of the filename, but is also expected to handle the case when +/// this method decides to use a different buffer. +bool Preprocessor::GetIncludeFilenameSpelling(SourceLocation Loc, + const char *&BufStart, + const char *&BufEnd) { + // Get the text form of the filename. + assert(BufStart != BufEnd && "Can't have tokens with empty spellings!"); + + // Make sure the filename is <x> or "x". + bool isAngled; + if (BufStart[0] == '<') { + if (BufEnd[-1] != '>') { + Diag(Loc, diag::err_pp_expects_filename); + BufStart = 0; + return true; + } + isAngled = true; + } else if (BufStart[0] == '"') { + if (BufEnd[-1] != '"') { + Diag(Loc, diag::err_pp_expects_filename); + BufStart = 0; + return true; + } + isAngled = false; + } else { + Diag(Loc, diag::err_pp_expects_filename); + BufStart = 0; + return true; + } + + // Diagnose #include "" as invalid. + if (BufEnd-BufStart <= 2) { + Diag(Loc, diag::err_pp_empty_filename); + BufStart = 0; + return ""; + } + + // Skip the brackets. + ++BufStart; + --BufEnd; + return isAngled; +} + +/// ConcatenateIncludeName - Handle cases where the #include name is expanded +/// from a macro as multiple tokens, which need to be glued together. This +/// occurs for code like: +/// #define FOO <a/b.h> +/// #include FOO +/// because in this case, "<a/b.h>" is returned as 7 tokens, not one. +/// +/// This code concatenates and consumes tokens up to the '>' token. It returns +/// false if the > was found, otherwise it returns true if it finds and consumes +/// the EOM marker. +static bool ConcatenateIncludeName(llvm::SmallVector<char, 128> &FilenameBuffer, + Preprocessor &PP) { + Token CurTok; + + PP.Lex(CurTok); + while (CurTok.isNot(tok::eom)) { + // Append the spelling of this token to the buffer. If there was a space + // before it, add it now. + if (CurTok.hasLeadingSpace()) + FilenameBuffer.push_back(' '); + + // Get the spelling of the token, directly into FilenameBuffer if possible. + unsigned PreAppendSize = FilenameBuffer.size(); + FilenameBuffer.resize(PreAppendSize+CurTok.getLength()); + + const char *BufPtr = &FilenameBuffer[PreAppendSize]; + unsigned ActualLen = PP.getSpelling(CurTok, BufPtr); + + // If the token was spelled somewhere else, copy it into FilenameBuffer. + if (BufPtr != &FilenameBuffer[PreAppendSize]) + memcpy(&FilenameBuffer[PreAppendSize], BufPtr, ActualLen); + + // Resize FilenameBuffer to the correct size. + if (CurTok.getLength() != ActualLen) + FilenameBuffer.resize(PreAppendSize+ActualLen); + + // If we found the '>' marker, return success. + if (CurTok.is(tok::greater)) + return false; + + PP.Lex(CurTok); + } + + // If we hit the eom marker, emit an error and return true so that the caller + // knows the EOM has been read. + PP.Diag(CurTok.getLocation(), diag::err_pp_expects_filename); + return true; +} + +/// HandleIncludeDirective - The "#include" tokens have just been read, read the +/// file to be included from the lexer, then include it! This is a common +/// routine with functionality shared between #include, #include_next and +/// #import. LookupFrom is set when this is a #include_next directive, it +/// specifies the file to start searching from. +void Preprocessor::HandleIncludeDirective(Token &IncludeTok, + const DirectoryLookup *LookupFrom, + bool isImport) { + + Token FilenameTok; + CurPPLexer->LexIncludeFilename(FilenameTok); + + // Reserve a buffer to get the spelling. + llvm::SmallVector<char, 128> FilenameBuffer; + const char *FilenameStart, *FilenameEnd; + + switch (FilenameTok.getKind()) { + case tok::eom: + // If the token kind is EOM, the error has already been diagnosed. + return; + + case tok::angle_string_literal: + case tok::string_literal: { + FilenameBuffer.resize(FilenameTok.getLength()); + FilenameStart = &FilenameBuffer[0]; + unsigned Len = getSpelling(FilenameTok, FilenameStart); + FilenameEnd = FilenameStart+Len; + break; + } + + case tok::less: + // This could be a <foo/bar.h> file coming from a macro expansion. In this + // case, glue the tokens together into FilenameBuffer and interpret those. + FilenameBuffer.push_back('<'); + if (ConcatenateIncludeName(FilenameBuffer, *this)) + return; // Found <eom> but no ">"? Diagnostic already emitted. + FilenameStart = FilenameBuffer.data(); + FilenameEnd = FilenameStart + FilenameBuffer.size(); + break; + default: + Diag(FilenameTok.getLocation(), diag::err_pp_expects_filename); + DiscardUntilEndOfDirective(); + return; + } + + bool isAngled = GetIncludeFilenameSpelling(FilenameTok.getLocation(), + FilenameStart, FilenameEnd); + // If GetIncludeFilenameSpelling set the start ptr to null, there was an + // error. + if (FilenameStart == 0) { + DiscardUntilEndOfDirective(); + return; + } + + // Verify that there is nothing after the filename, other than EOM. Note that + // we allow macros that expand to nothing after the filename, because this + // falls into the category of "#include pp-tokens new-line" specified in + // C99 6.10.2p4. + CheckEndOfDirective(IncludeTok.getIdentifierInfo()->getName(), true); + + // Check that we don't have infinite #include recursion. + if (IncludeMacroStack.size() == MaxAllowedIncludeStackDepth-1) { + Diag(FilenameTok, diag::err_pp_include_too_deep); + return; + } + + // Search include directories. + const DirectoryLookup *CurDir; + const FileEntry *File = LookupFile(FilenameStart, FilenameEnd, + isAngled, LookupFrom, CurDir); + if (File == 0) { + Diag(FilenameTok, diag::err_pp_file_not_found) + << std::string(FilenameStart, FilenameEnd); + return; + } + + // Ask HeaderInfo if we should enter this #include file. If not, #including + // this file will have no effect. + if (!HeaderInfo.ShouldEnterIncludeFile(File, isImport)) + return; + + // The #included file will be considered to be a system header if either it is + // in a system include directory, or if the #includer is a system include + // header. + SrcMgr::CharacteristicKind FileCharacter = + std::max(HeaderInfo.getFileDirFlavor(File), + SourceMgr.getFileCharacteristic(FilenameTok.getLocation())); + + // Look up the file, create a File ID for it. + FileID FID = SourceMgr.createFileID(File, FilenameTok.getLocation(), + FileCharacter); + if (FID.isInvalid()) { + Diag(FilenameTok, diag::err_pp_file_not_found) + << std::string(FilenameStart, FilenameEnd); + return; + } + + // Finally, if all is good, enter the new file! + EnterSourceFile(FID, CurDir); +} + +/// HandleIncludeNextDirective - Implements #include_next. +/// +void Preprocessor::HandleIncludeNextDirective(Token &IncludeNextTok) { + Diag(IncludeNextTok, diag::ext_pp_include_next_directive); + + // #include_next is like #include, except that we start searching after + // the current found directory. If we can't do this, issue a + // diagnostic. + const DirectoryLookup *Lookup = CurDirLookup; + if (isInPrimaryFile()) { + Lookup = 0; + Diag(IncludeNextTok, diag::pp_include_next_in_primary); + } else if (Lookup == 0) { + Diag(IncludeNextTok, diag::pp_include_next_absolute_path); + } else { + // Start looking up in the next directory. + ++Lookup; + } + + return HandleIncludeDirective(IncludeNextTok, Lookup); +} + +/// HandleImportDirective - Implements #import. +/// +void Preprocessor::HandleImportDirective(Token &ImportTok) { + if (!Features.ObjC1) // #import is standard for ObjC. + Diag(ImportTok, diag::ext_pp_import_directive); + + return HandleIncludeDirective(ImportTok, 0, true); +} + +/// HandleIncludeMacrosDirective - The -imacros command line option turns into a +/// pseudo directive in the predefines buffer. This handles it by sucking all +/// tokens through the preprocessor and discarding them (only keeping the side +/// effects on the preprocessor). +void Preprocessor::HandleIncludeMacrosDirective(Token &IncludeMacrosTok) { + // This directive should only occur in the predefines buffer. If not, emit an + // error and reject it. + SourceLocation Loc = IncludeMacrosTok.getLocation(); + if (strcmp(SourceMgr.getBufferName(Loc), "<built-in>") != 0) { + Diag(IncludeMacrosTok.getLocation(), + diag::pp_include_macros_out_of_predefines); + DiscardUntilEndOfDirective(); + return; + } + + // Treat this as a normal #include for checking purposes. If this is + // successful, it will push a new lexer onto the include stack. + HandleIncludeDirective(IncludeMacrosTok, 0, false); + + Token TmpTok; + do { + Lex(TmpTok); + assert(TmpTok.isNot(tok::eof) && "Didn't find end of -imacros!"); + } while (TmpTok.isNot(tok::hashhash)); +} + +//===----------------------------------------------------------------------===// +// Preprocessor Macro Directive Handling. +//===----------------------------------------------------------------------===// + +/// ReadMacroDefinitionArgList - The ( starting an argument list of a macro +/// definition has just been read. Lex the rest of the arguments and the +/// closing ), updating MI with what we learn. Return true if an error occurs +/// parsing the arg list. +bool Preprocessor::ReadMacroDefinitionArgList(MacroInfo *MI) { + llvm::SmallVector<IdentifierInfo*, 32> Arguments; + + Token Tok; + while (1) { + LexUnexpandedToken(Tok); + switch (Tok.getKind()) { + case tok::r_paren: + // Found the end of the argument list. + if (Arguments.empty()) // #define FOO() + return false; + // Otherwise we have #define FOO(A,) + Diag(Tok, diag::err_pp_expected_ident_in_arg_list); + return true; + case tok::ellipsis: // #define X(... -> C99 varargs + // Warn if use of C99 feature in non-C99 mode. + if (!Features.C99) Diag(Tok, diag::ext_variadic_macro); + + // Lex the token after the identifier. + LexUnexpandedToken(Tok); + if (Tok.isNot(tok::r_paren)) { + Diag(Tok, diag::err_pp_missing_rparen_in_macro_def); + return true; + } + // Add the __VA_ARGS__ identifier as an argument. + Arguments.push_back(Ident__VA_ARGS__); + MI->setIsC99Varargs(); + MI->setArgumentList(&Arguments[0], Arguments.size(), BP); + return false; + case tok::eom: // #define X( + Diag(Tok, diag::err_pp_missing_rparen_in_macro_def); + return true; + default: + // Handle keywords and identifiers here to accept things like + // #define Foo(for) for. + IdentifierInfo *II = Tok.getIdentifierInfo(); + if (II == 0) { + // #define X(1 + Diag(Tok, diag::err_pp_invalid_tok_in_arg_list); + return true; + } + + // If this is already used as an argument, it is used multiple times (e.g. + // #define X(A,A. + if (std::find(Arguments.begin(), Arguments.end(), II) != + Arguments.end()) { // C99 6.10.3p6 + Diag(Tok, diag::err_pp_duplicate_name_in_arg_list) << II; + return true; + } + + // Add the argument to the macro info. + Arguments.push_back(II); + + // Lex the token after the identifier. + LexUnexpandedToken(Tok); + + switch (Tok.getKind()) { + default: // #define X(A B + Diag(Tok, diag::err_pp_expected_comma_in_arg_list); + return true; + case tok::r_paren: // #define X(A) + MI->setArgumentList(&Arguments[0], Arguments.size(), BP); + return false; + case tok::comma: // #define X(A, + break; + case tok::ellipsis: // #define X(A... -> GCC extension + // Diagnose extension. + Diag(Tok, diag::ext_named_variadic_macro); + + // Lex the token after the identifier. + LexUnexpandedToken(Tok); + if (Tok.isNot(tok::r_paren)) { + Diag(Tok, diag::err_pp_missing_rparen_in_macro_def); + return true; + } + + MI->setIsGNUVarargs(); + MI->setArgumentList(&Arguments[0], Arguments.size(), BP); + return false; + } + } + } +} + +/// HandleDefineDirective - Implements #define. This consumes the entire macro +/// line then lets the caller lex the next real token. +void Preprocessor::HandleDefineDirective(Token &DefineTok) { + ++NumDefined; + + Token MacroNameTok; + ReadMacroName(MacroNameTok, 1); + + // Error reading macro name? If so, diagnostic already issued. + if (MacroNameTok.is(tok::eom)) + return; + + Token LastTok = MacroNameTok; + + // If we are supposed to keep comments in #defines, reenable comment saving + // mode. + if (CurLexer) CurLexer->SetCommentRetentionState(KeepMacroComments); + + // Create the new macro. + MacroInfo *MI = AllocateMacroInfo(MacroNameTok.getLocation()); + + Token Tok; + LexUnexpandedToken(Tok); + + // If this is a function-like macro definition, parse the argument list, + // marking each of the identifiers as being used as macro arguments. Also, + // check other constraints on the first token of the macro body. + if (Tok.is(tok::eom)) { + // If there is no body to this macro, we have no special handling here. + } else if (Tok.hasLeadingSpace()) { + // This is a normal token with leading space. Clear the leading space + // marker on the first token to get proper expansion. + Tok.clearFlag(Token::LeadingSpace); + } else if (Tok.is(tok::l_paren)) { + // This is a function-like macro definition. Read the argument list. + MI->setIsFunctionLike(); + if (ReadMacroDefinitionArgList(MI)) { + // Forget about MI. + ReleaseMacroInfo(MI); + // Throw away the rest of the line. + if (CurPPLexer->ParsingPreprocessorDirective) + DiscardUntilEndOfDirective(); + return; + } + + // If this is a definition of a variadic C99 function-like macro, not using + // the GNU named varargs extension, enabled __VA_ARGS__. + + // "Poison" __VA_ARGS__, which can only appear in the expansion of a macro. + // This gets unpoisoned where it is allowed. + assert(Ident__VA_ARGS__->isPoisoned() && "__VA_ARGS__ should be poisoned!"); + if (MI->isC99Varargs()) + Ident__VA_ARGS__->setIsPoisoned(false); + + // Read the first token after the arg list for down below. + LexUnexpandedToken(Tok); + } else if (Features.C99) { + // C99 requires whitespace between the macro definition and the body. Emit + // a diagnostic for something like "#define X+". + Diag(Tok, diag::ext_c99_whitespace_required_after_macro_name); + } else { + // C90 6.8 TC1 says: "In the definition of an object-like macro, if the + // first character of a replacement list is not a character required by + // subclause 5.2.1, then there shall be white-space separation between the + // identifier and the replacement list.". 5.2.1 lists this set: + // "A-Za-z0-9!"#%&'()*+,_./:;<=>?[\]^_{|}~" as well as whitespace, which + // is irrelevant here. + bool isInvalid = false; + if (Tok.is(tok::at)) // @ is not in the list above. + isInvalid = true; + else if (Tok.is(tok::unknown)) { + // If we have an unknown token, it is something strange like "`". Since + // all of valid characters would have lexed into a single character + // token of some sort, we know this is not a valid case. + isInvalid = true; + } + if (isInvalid) + Diag(Tok, diag::ext_missing_whitespace_after_macro_name); + else + Diag(Tok, diag::warn_missing_whitespace_after_macro_name); + } + + if (!Tok.is(tok::eom)) + LastTok = Tok; + + // Read the rest of the macro body. + if (MI->isObjectLike()) { + // Object-like macros are very simple, just read their body. + while (Tok.isNot(tok::eom)) { + LastTok = Tok; + MI->AddTokenToBody(Tok); + // Get the next token of the macro. + LexUnexpandedToken(Tok); + } + + } else { + // Otherwise, read the body of a function-like macro. While we are at it, + // check C99 6.10.3.2p1: ensure that # operators are followed by macro + // parameters in function-like macro expansions. + while (Tok.isNot(tok::eom)) { + LastTok = Tok; + + if (Tok.isNot(tok::hash)) { + MI->AddTokenToBody(Tok); + + // Get the next token of the macro. + LexUnexpandedToken(Tok); + continue; + } + + // Get the next token of the macro. + LexUnexpandedToken(Tok); + + // Check for a valid macro arg identifier. + if (Tok.getIdentifierInfo() == 0 || + MI->getArgumentNum(Tok.getIdentifierInfo()) == -1) { + + // If this is assembler-with-cpp mode, we accept random gibberish after + // the '#' because '#' is often a comment character. However, change + // the kind of the token to tok::unknown so that the preprocessor isn't + // confused. + if (getLangOptions().AsmPreprocessor && Tok.isNot(tok::eom)) { + LastTok.setKind(tok::unknown); + } else { + Diag(Tok, diag::err_pp_stringize_not_parameter); + ReleaseMacroInfo(MI); + + // Disable __VA_ARGS__ again. + Ident__VA_ARGS__->setIsPoisoned(true); + return; + } + } + + // Things look ok, add the '#' and param name tokens to the macro. + MI->AddTokenToBody(LastTok); + MI->AddTokenToBody(Tok); + LastTok = Tok; + + // Get the next token of the macro. + LexUnexpandedToken(Tok); + } + } + + + // Disable __VA_ARGS__ again. + Ident__VA_ARGS__->setIsPoisoned(true); + + // Check that there is no paste (##) operator at the begining or end of the + // replacement list. + unsigned NumTokens = MI->getNumTokens(); + if (NumTokens != 0) { + if (MI->getReplacementToken(0).is(tok::hashhash)) { + Diag(MI->getReplacementToken(0), diag::err_paste_at_start); + ReleaseMacroInfo(MI); + return; + } + if (MI->getReplacementToken(NumTokens-1).is(tok::hashhash)) { + Diag(MI->getReplacementToken(NumTokens-1), diag::err_paste_at_end); + ReleaseMacroInfo(MI); + return; + } + } + + // If this is the primary source file, remember that this macro hasn't been + // used yet. + if (isInPrimaryFile()) + MI->setIsUsed(false); + + MI->setDefinitionEndLoc(LastTok.getLocation()); + + // Finally, if this identifier already had a macro defined for it, verify that + // the macro bodies are identical and free the old definition. + if (MacroInfo *OtherMI = getMacroInfo(MacroNameTok.getIdentifierInfo())) { + // It is very common for system headers to have tons of macro redefinitions + // and for warnings to be disabled in system headers. If this is the case, + // then don't bother calling MacroInfo::isIdenticalTo. + if (!getDiagnostics().getSuppressSystemWarnings() || + !SourceMgr.isInSystemHeader(DefineTok.getLocation())) { + if (!OtherMI->isUsed()) + Diag(OtherMI->getDefinitionLoc(), diag::pp_macro_not_used); + + // Macros must be identical. This means all tokes and whitespace + // separation must be the same. C99 6.10.3.2. + if (!MI->isIdenticalTo(*OtherMI, *this)) { + Diag(MI->getDefinitionLoc(), diag::ext_pp_macro_redef) + << MacroNameTok.getIdentifierInfo(); + Diag(OtherMI->getDefinitionLoc(), diag::note_previous_definition); + } + } + + ReleaseMacroInfo(OtherMI); + } + + setMacroInfo(MacroNameTok.getIdentifierInfo(), MI); + + // If the callbacks want to know, tell them about the macro definition. + if (Callbacks) + Callbacks->MacroDefined(MacroNameTok.getIdentifierInfo(), MI); +} + +/// HandleUndefDirective - Implements #undef. +/// +void Preprocessor::HandleUndefDirective(Token &UndefTok) { + ++NumUndefined; + + Token MacroNameTok; + ReadMacroName(MacroNameTok, 2); + + // Error reading macro name? If so, diagnostic already issued. + if (MacroNameTok.is(tok::eom)) + return; + + // Check to see if this is the last token on the #undef line. + CheckEndOfDirective("undef"); + + // Okay, we finally have a valid identifier to undef. + MacroInfo *MI = getMacroInfo(MacroNameTok.getIdentifierInfo()); + + // If the macro is not defined, this is a noop undef, just return. + if (MI == 0) return; + + if (!MI->isUsed()) + Diag(MI->getDefinitionLoc(), diag::pp_macro_not_used); + + // If the callbacks want to know, tell them about the macro #undef. + if (Callbacks) + Callbacks->MacroUndefined(MacroNameTok.getIdentifierInfo(), MI); + + // Free macro definition. + ReleaseMacroInfo(MI); + setMacroInfo(MacroNameTok.getIdentifierInfo(), 0); +} + + +//===----------------------------------------------------------------------===// +// Preprocessor Conditional Directive Handling. +//===----------------------------------------------------------------------===// + +/// HandleIfdefDirective - Implements the #ifdef/#ifndef directive. isIfndef is +/// true when this is a #ifndef directive. ReadAnyTokensBeforeDirective is true +/// if any tokens have been returned or pp-directives activated before this +/// #ifndef has been lexed. +/// +void Preprocessor::HandleIfdefDirective(Token &Result, bool isIfndef, + bool ReadAnyTokensBeforeDirective) { + ++NumIf; + Token DirectiveTok = Result; + + Token MacroNameTok; + ReadMacroName(MacroNameTok); + + // Error reading macro name? If so, diagnostic already issued. + if (MacroNameTok.is(tok::eom)) { + // Skip code until we get to #endif. This helps with recovery by not + // emitting an error when the #endif is reached. + SkipExcludedConditionalBlock(DirectiveTok.getLocation(), + /*Foundnonskip*/false, /*FoundElse*/false); + return; + } + + // Check to see if this is the last token on the #if[n]def line. + CheckEndOfDirective(isIfndef ? "ifndef" : "ifdef"); + + if (CurPPLexer->getConditionalStackDepth() == 0) { + // If the start of a top-level #ifdef, inform MIOpt. + if (!ReadAnyTokensBeforeDirective) { + assert(isIfndef && "#ifdef shouldn't reach here"); + CurPPLexer->MIOpt.EnterTopLevelIFNDEF(MacroNameTok.getIdentifierInfo()); + } else + CurPPLexer->MIOpt.EnterTopLevelConditional(); + } + + IdentifierInfo *MII = MacroNameTok.getIdentifierInfo(); + MacroInfo *MI = getMacroInfo(MII); + + // If there is a macro, process it. + if (MI) // Mark it used. + MI->setIsUsed(true); + + // Should we include the stuff contained by this directive? + if (!MI == isIfndef) { + // Yes, remember that we are inside a conditional, then lex the next token. + CurPPLexer->pushConditionalLevel(DirectiveTok.getLocation(), /*wasskip*/false, + /*foundnonskip*/true, /*foundelse*/false); + } else { + // No, skip the contents of this block and return the first token after it. + SkipExcludedConditionalBlock(DirectiveTok.getLocation(), + /*Foundnonskip*/false, + /*FoundElse*/false); + } +} + +/// HandleIfDirective - Implements the #if directive. +/// +void Preprocessor::HandleIfDirective(Token &IfToken, + bool ReadAnyTokensBeforeDirective) { + ++NumIf; + + // Parse and evaluation the conditional expression. + IdentifierInfo *IfNDefMacro = 0; + bool ConditionalTrue = EvaluateDirectiveExpression(IfNDefMacro); + + + // If this condition is equivalent to #ifndef X, and if this is the first + // directive seen, handle it for the multiple-include optimization. + if (CurPPLexer->getConditionalStackDepth() == 0) { + if (!ReadAnyTokensBeforeDirective && IfNDefMacro) + CurPPLexer->MIOpt.EnterTopLevelIFNDEF(IfNDefMacro); + else + CurPPLexer->MIOpt.EnterTopLevelConditional(); + } + + // Should we include the stuff contained by this directive? + if (ConditionalTrue) { + // Yes, remember that we are inside a conditional, then lex the next token. + CurPPLexer->pushConditionalLevel(IfToken.getLocation(), /*wasskip*/false, + /*foundnonskip*/true, /*foundelse*/false); + } else { + // No, skip the contents of this block and return the first token after it. + SkipExcludedConditionalBlock(IfToken.getLocation(), /*Foundnonskip*/false, + /*FoundElse*/false); + } +} + +/// HandleEndifDirective - Implements the #endif directive. +/// +void Preprocessor::HandleEndifDirective(Token &EndifToken) { + ++NumEndif; + + // Check that this is the whole directive. + CheckEndOfDirective("endif"); + + PPConditionalInfo CondInfo; + if (CurPPLexer->popConditionalLevel(CondInfo)) { + // No conditionals on the stack: this is an #endif without an #if. + Diag(EndifToken, diag::err_pp_endif_without_if); + return; + } + + // If this the end of a top-level #endif, inform MIOpt. + if (CurPPLexer->getConditionalStackDepth() == 0) + CurPPLexer->MIOpt.ExitTopLevelConditional(); + + assert(!CondInfo.WasSkipping && !CurPPLexer->LexingRawMode && + "This code should only be reachable in the non-skipping case!"); +} + + +void Preprocessor::HandleElseDirective(Token &Result) { + ++NumElse; + + // #else directive in a non-skipping conditional... start skipping. + CheckEndOfDirective("else"); + + PPConditionalInfo CI; + if (CurPPLexer->popConditionalLevel(CI)) { + Diag(Result, diag::pp_err_else_without_if); + return; + } + + // If this is a top-level #else, inform the MIOpt. + if (CurPPLexer->getConditionalStackDepth() == 0) + CurPPLexer->MIOpt.EnterTopLevelConditional(); + + // If this is a #else with a #else before it, report the error. + if (CI.FoundElse) Diag(Result, diag::pp_err_else_after_else); + + // Finally, skip the rest of the contents of this block and return the first + // token after it. + return SkipExcludedConditionalBlock(CI.IfLoc, /*Foundnonskip*/true, + /*FoundElse*/true); +} + +void Preprocessor::HandleElifDirective(Token &ElifToken) { + ++NumElse; + + // #elif directive in a non-skipping conditional... start skipping. + // We don't care what the condition is, because we will always skip it (since + // the block immediately before it was included). + DiscardUntilEndOfDirective(); + + PPConditionalInfo CI; + if (CurPPLexer->popConditionalLevel(CI)) { + Diag(ElifToken, diag::pp_err_elif_without_if); + return; + } + + // If this is a top-level #elif, inform the MIOpt. + if (CurPPLexer->getConditionalStackDepth() == 0) + CurPPLexer->MIOpt.EnterTopLevelConditional(); + + // If this is a #elif with a #else before it, report the error. + if (CI.FoundElse) Diag(ElifToken, diag::pp_err_elif_after_else); + + // Finally, skip the rest of the contents of this block and return the first + // token after it. + return SkipExcludedConditionalBlock(CI.IfLoc, /*Foundnonskip*/true, + /*FoundElse*/CI.FoundElse); +} + diff --git a/lib/Lex/PPExpressions.cpp b/lib/Lex/PPExpressions.cpp new file mode 100644 index 0000000..709e316 --- /dev/null +++ b/lib/Lex/PPExpressions.cpp @@ -0,0 +1,717 @@ +//===--- PPExpressions.cpp - Preprocessor Expression Evaluation -----------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements the Preprocessor::EvaluateDirectiveExpression method, +// which parses and evaluates integer constant expressions for #if directives. +// +//===----------------------------------------------------------------------===// +// +// FIXME: implement testing for #assert's. +// +//===----------------------------------------------------------------------===// + +#include "clang/Lex/Preprocessor.h" +#include "clang/Lex/MacroInfo.h" +#include "clang/Lex/LiteralSupport.h" +#include "clang/Basic/TargetInfo.h" +#include "clang/Lex/LexDiagnostic.h" +#include "llvm/ADT/APSInt.h" +using namespace clang; + +/// PPValue - Represents the value of a subexpression of a preprocessor +/// conditional and the source range covered by it. +class PPValue { + SourceRange Range; +public: + llvm::APSInt Val; + + // Default ctor - Construct an 'invalid' PPValue. + PPValue(unsigned BitWidth) : Val(BitWidth) {} + + unsigned getBitWidth() const { return Val.getBitWidth(); } + bool isUnsigned() const { return Val.isUnsigned(); } + + const SourceRange &getRange() const { return Range; } + + void setRange(SourceLocation L) { Range.setBegin(L); Range.setEnd(L); } + void setRange(SourceLocation B, SourceLocation E) { + Range.setBegin(B); Range.setEnd(E); + } + void setBegin(SourceLocation L) { Range.setBegin(L); } + void setEnd(SourceLocation L) { Range.setEnd(L); } +}; + +static bool EvaluateDirectiveSubExpr(PPValue &LHS, unsigned MinPrec, + Token &PeekTok, bool ValueLive, + Preprocessor &PP); + +/// DefinedTracker - This struct is used while parsing expressions to keep track +/// of whether !defined(X) has been seen. +/// +/// With this simple scheme, we handle the basic forms: +/// !defined(X) and !defined X +/// but we also trivially handle (silly) stuff like: +/// !!!defined(X) and +!defined(X) and !+!+!defined(X) and !(defined(X)). +struct DefinedTracker { + /// Each time a Value is evaluated, it returns information about whether the + /// parsed value is of the form defined(X), !defined(X) or is something else. + enum TrackerState { + DefinedMacro, // defined(X) + NotDefinedMacro, // !defined(X) + Unknown // Something else. + } State; + /// TheMacro - When the state is DefinedMacro or NotDefinedMacro, this + /// indicates the macro that was checked. + IdentifierInfo *TheMacro; +}; + +/// EvaluateValue - Evaluate the token PeekTok (and any others needed) and +/// return the computed value in Result. Return true if there was an error +/// parsing. This function also returns information about the form of the +/// expression in DT. See above for information on what DT means. +/// +/// If ValueLive is false, then this value is being evaluated in a context where +/// the result is not used. As such, avoid diagnostics that relate to +/// evaluation. +static bool EvaluateValue(PPValue &Result, Token &PeekTok, DefinedTracker &DT, + bool ValueLive, Preprocessor &PP) { + DT.State = DefinedTracker::Unknown; + + // If this token's spelling is a pp-identifier, check to see if it is + // 'defined' or if it is a macro. Note that we check here because many + // keywords are pp-identifiers, so we can't check the kind. + if (IdentifierInfo *II = PeekTok.getIdentifierInfo()) { + // If this identifier isn't 'defined' and it wasn't macro expanded, it turns + // into a simple 0, unless it is the C++ keyword "true", in which case it + // turns into "1". + if (!II->isStr("defined")) { + if (ValueLive) + PP.Diag(PeekTok, diag::warn_pp_undef_identifier) << II; + Result.Val = II->getTokenID() == tok::kw_true; + Result.Val.setIsUnsigned(false); // "0" is signed intmax_t 0. + Result.setRange(PeekTok.getLocation()); + PP.LexNonComment(PeekTok); + return false; + } + + // Handle "defined X" and "defined(X)". + Result.setBegin(PeekTok.getLocation()); + + // Get the next token, don't expand it. + PP.LexUnexpandedToken(PeekTok); + + // Two options, it can either be a pp-identifier or a (. + SourceLocation LParenLoc; + if (PeekTok.is(tok::l_paren)) { + // Found a paren, remember we saw it and skip it. + LParenLoc = PeekTok.getLocation(); + PP.LexUnexpandedToken(PeekTok); + } + + // If we don't have a pp-identifier now, this is an error. + if ((II = PeekTok.getIdentifierInfo()) == 0) { + PP.Diag(PeekTok, diag::err_pp_defined_requires_identifier); + return true; + } + + // Otherwise, we got an identifier, is it defined to something? + Result.Val = II->hasMacroDefinition(); + Result.Val.setIsUnsigned(false); // Result is signed intmax_t. + + // If there is a macro, mark it used. + if (Result.Val != 0 && ValueLive) { + MacroInfo *Macro = PP.getMacroInfo(II); + Macro->setIsUsed(true); + } + + // Consume identifier. + Result.setEnd(PeekTok.getLocation()); + PP.LexNonComment(PeekTok); + + // If we are in parens, ensure we have a trailing ). + if (LParenLoc.isValid()) { + if (PeekTok.isNot(tok::r_paren)) { + PP.Diag(PeekTok.getLocation(), diag::err_pp_missing_rparen); + PP.Diag(LParenLoc, diag::note_matching) << "("; + return true; + } + // Consume the ). + Result.setEnd(PeekTok.getLocation()); + PP.LexNonComment(PeekTok); + } + + // Success, remember that we saw defined(X). + DT.State = DefinedTracker::DefinedMacro; + DT.TheMacro = II; + return false; + } + + switch (PeekTok.getKind()) { + default: // Non-value token. + PP.Diag(PeekTok, diag::err_pp_expr_bad_token_start_expr); + return true; + case tok::eom: + case tok::r_paren: + // If there is no expression, report and exit. + PP.Diag(PeekTok, diag::err_pp_expected_value_in_expr); + return true; + case tok::numeric_constant: { + llvm::SmallString<64> IntegerBuffer; + IntegerBuffer.resize(PeekTok.getLength()); + const char *ThisTokBegin = &IntegerBuffer[0]; + unsigned ActualLength = PP.getSpelling(PeekTok, ThisTokBegin); + NumericLiteralParser Literal(ThisTokBegin, ThisTokBegin+ActualLength, + PeekTok.getLocation(), PP); + if (Literal.hadError) + return true; // a diagnostic was already reported. + + if (Literal.isFloatingLiteral() || Literal.isImaginary) { + PP.Diag(PeekTok, diag::err_pp_illegal_floating_literal); + return true; + } + assert(Literal.isIntegerLiteral() && "Unknown ppnumber"); + + // long long is a C99 feature. + if (!PP.getLangOptions().C99 && !PP.getLangOptions().CPlusPlus0x + && Literal.isLongLong) + PP.Diag(PeekTok, diag::ext_longlong); + + // Parse the integer literal into Result. + if (Literal.GetIntegerValue(Result.Val)) { + // Overflow parsing integer literal. + if (ValueLive) PP.Diag(PeekTok, diag::warn_integer_too_large); + Result.Val.setIsUnsigned(true); + } else { + // Set the signedness of the result to match whether there was a U suffix + // or not. + Result.Val.setIsUnsigned(Literal.isUnsigned); + + // Detect overflow based on whether the value is signed. If signed + // and if the value is too large, emit a warning "integer constant is so + // large that it is unsigned" e.g. on 12345678901234567890 where intmax_t + // is 64-bits. + if (!Literal.isUnsigned && Result.Val.isNegative()) { + // Don't warn for a hex literal: 0x8000..0 shouldn't warn. + if (ValueLive && Literal.getRadix() != 16) + PP.Diag(PeekTok, diag::warn_integer_too_large_for_signed); + Result.Val.setIsUnsigned(true); + } + } + + // Consume the token. + Result.setRange(PeekTok.getLocation()); + PP.LexNonComment(PeekTok); + return false; + } + case tok::char_constant: { // 'x' + llvm::SmallString<32> CharBuffer; + CharBuffer.resize(PeekTok.getLength()); + const char *ThisTokBegin = &CharBuffer[0]; + unsigned ActualLength = PP.getSpelling(PeekTok, ThisTokBegin); + CharLiteralParser Literal(ThisTokBegin, ThisTokBegin+ActualLength, + PeekTok.getLocation(), PP); + if (Literal.hadError()) + return true; // A diagnostic was already emitted. + + // Character literals are always int or wchar_t, expand to intmax_t. + TargetInfo &TI = PP.getTargetInfo(); + unsigned NumBits; + if (Literal.isMultiChar()) + NumBits = TI.getIntWidth(); + else + NumBits = TI.getCharWidth(Literal.isWide()); + + // Set the width. + llvm::APSInt Val(NumBits); + // Set the value. + Val = Literal.getValue(); + // Set the signedness. + Val.setIsUnsigned(!TI.isCharSigned()); + + if (Result.Val.getBitWidth() > Val.getBitWidth()) { + Result.Val = Val.extend(Result.Val.getBitWidth()); + } else { + assert(Result.Val.getBitWidth() == Val.getBitWidth() && + "intmax_t smaller than char/wchar_t?"); + Result.Val = Val; + } + + // Consume the token. + Result.setRange(PeekTok.getLocation()); + PP.LexNonComment(PeekTok); + return false; + } + case tok::l_paren: { + SourceLocation Start = PeekTok.getLocation(); + PP.LexNonComment(PeekTok); // Eat the (. + // Parse the value and if there are any binary operators involved, parse + // them. + if (EvaluateValue(Result, PeekTok, DT, ValueLive, PP)) return true; + + // If this is a silly value like (X), which doesn't need parens, check for + // !(defined X). + if (PeekTok.is(tok::r_paren)) { + // Just use DT unmodified as our result. + } else { + // Otherwise, we have something like (x+y), and we consumed '(x'. + if (EvaluateDirectiveSubExpr(Result, 1, PeekTok, ValueLive, PP)) + return true; + + if (PeekTok.isNot(tok::r_paren)) { + PP.Diag(PeekTok.getLocation(), diag::err_pp_expected_rparen) + << Result.getRange(); + PP.Diag(Start, diag::note_matching) << "("; + return true; + } + DT.State = DefinedTracker::Unknown; + } + Result.setRange(Start, PeekTok.getLocation()); + PP.LexNonComment(PeekTok); // Eat the ). + return false; + } + case tok::plus: { + SourceLocation Start = PeekTok.getLocation(); + // Unary plus doesn't modify the value. + PP.LexNonComment(PeekTok); + if (EvaluateValue(Result, PeekTok, DT, ValueLive, PP)) return true; + Result.setBegin(Start); + return false; + } + case tok::minus: { + SourceLocation Loc = PeekTok.getLocation(); + PP.LexNonComment(PeekTok); + if (EvaluateValue(Result, PeekTok, DT, ValueLive, PP)) return true; + Result.setBegin(Loc); + + // C99 6.5.3.3p3: The sign of the result matches the sign of the operand. + Result.Val = -Result.Val; + + // -MININT is the only thing that overflows. Unsigned never overflows. + bool Overflow = !Result.isUnsigned() && Result.Val.isMinSignedValue(); + + // If this operator is live and overflowed, report the issue. + if (Overflow && ValueLive) + PP.Diag(Loc, diag::warn_pp_expr_overflow) << Result.getRange(); + + DT.State = DefinedTracker::Unknown; + return false; + } + + case tok::tilde: { + SourceLocation Start = PeekTok.getLocation(); + PP.LexNonComment(PeekTok); + if (EvaluateValue(Result, PeekTok, DT, ValueLive, PP)) return true; + Result.setBegin(Start); + + // C99 6.5.3.3p4: The sign of the result matches the sign of the operand. + Result.Val = ~Result.Val; + DT.State = DefinedTracker::Unknown; + return false; + } + + case tok::exclaim: { + SourceLocation Start = PeekTok.getLocation(); + PP.LexNonComment(PeekTok); + if (EvaluateValue(Result, PeekTok, DT, ValueLive, PP)) return true; + Result.setBegin(Start); + Result.Val = !Result.Val; + // C99 6.5.3.3p5: The sign of the result is 'int', aka it is signed. + Result.Val.setIsUnsigned(false); + + if (DT.State == DefinedTracker::DefinedMacro) + DT.State = DefinedTracker::NotDefinedMacro; + else if (DT.State == DefinedTracker::NotDefinedMacro) + DT.State = DefinedTracker::DefinedMacro; + return false; + } + + // FIXME: Handle #assert + } +} + + + +/// getPrecedence - Return the precedence of the specified binary operator +/// token. This returns: +/// ~0 - Invalid token. +/// 14 -> 3 - various operators. +/// 0 - 'eom' or ')' +static unsigned getPrecedence(tok::TokenKind Kind) { + switch (Kind) { + default: return ~0U; + case tok::percent: + case tok::slash: + case tok::star: return 14; + case tok::plus: + case tok::minus: return 13; + case tok::lessless: + case tok::greatergreater: return 12; + case tok::lessequal: + case tok::less: + case tok::greaterequal: + case tok::greater: return 11; + case tok::exclaimequal: + case tok::equalequal: return 10; + case tok::amp: return 9; + case tok::caret: return 8; + case tok::pipe: return 7; + case tok::ampamp: return 6; + case tok::pipepipe: return 5; + case tok::question: return 4; + case tok::comma: return 3; + case tok::colon: return 2; + case tok::r_paren: return 0; // Lowest priority, end of expr. + case tok::eom: return 0; // Lowest priority, end of macro. + } +} + + +/// EvaluateDirectiveSubExpr - Evaluate the subexpression whose first token is +/// PeekTok, and whose precedence is PeekPrec. This returns the result in LHS. +/// +/// If ValueLive is false, then this value is being evaluated in a context where +/// the result is not used. As such, avoid diagnostics that relate to +/// evaluation, such as division by zero warnings. +static bool EvaluateDirectiveSubExpr(PPValue &LHS, unsigned MinPrec, + Token &PeekTok, bool ValueLive, + Preprocessor &PP) { + unsigned PeekPrec = getPrecedence(PeekTok.getKind()); + // If this token isn't valid, report the error. + if (PeekPrec == ~0U) { + PP.Diag(PeekTok.getLocation(), diag::err_pp_expr_bad_token_binop) + << LHS.getRange(); + return true; + } + + while (1) { + // If this token has a lower precedence than we are allowed to parse, return + // it so that higher levels of the recursion can parse it. + if (PeekPrec < MinPrec) + return false; + + tok::TokenKind Operator = PeekTok.getKind(); + + // If this is a short-circuiting operator, see if the RHS of the operator is + // dead. Note that this cannot just clobber ValueLive. Consider + // "0 && 1 ? 4 : 1 / 0", which is parsed as "(0 && 1) ? 4 : (1 / 0)". In + // this example, the RHS of the && being dead does not make the rest of the + // expr dead. + bool RHSIsLive; + if (Operator == tok::ampamp && LHS.Val == 0) + RHSIsLive = false; // RHS of "0 && x" is dead. + else if (Operator == tok::pipepipe && LHS.Val != 0) + RHSIsLive = false; // RHS of "1 || x" is dead. + else if (Operator == tok::question && LHS.Val == 0) + RHSIsLive = false; // RHS (x) of "0 ? x : y" is dead. + else + RHSIsLive = ValueLive; + + // Consume the operator, remembering the operator's location for reporting. + SourceLocation OpLoc = PeekTok.getLocation(); + PP.LexNonComment(PeekTok); + + PPValue RHS(LHS.getBitWidth()); + // Parse the RHS of the operator. + DefinedTracker DT; + if (EvaluateValue(RHS, PeekTok, DT, RHSIsLive, PP)) return true; + + // Remember the precedence of this operator and get the precedence of the + // operator immediately to the right of the RHS. + unsigned ThisPrec = PeekPrec; + PeekPrec = getPrecedence(PeekTok.getKind()); + + // If this token isn't valid, report the error. + if (PeekPrec == ~0U) { + PP.Diag(PeekTok.getLocation(), diag::err_pp_expr_bad_token_binop) + << RHS.getRange(); + return true; + } + + // Decide whether to include the next binop in this subexpression. For + // example, when parsing x+y*z and looking at '*', we want to recursively + // handle y*z as a single subexpression. We do this because the precedence + // of * is higher than that of +. The only strange case we have to handle + // here is for the ?: operator, where the precedence is actually lower than + // the LHS of the '?'. The grammar rule is: + // + // conditional-expression ::= + // logical-OR-expression ? expression : conditional-expression + // where 'expression' is actually comma-expression. + unsigned RHSPrec; + if (Operator == tok::question) + // The RHS of "?" should be maximally consumed as an expression. + RHSPrec = getPrecedence(tok::comma); + else // All others should munch while higher precedence. + RHSPrec = ThisPrec+1; + + if (PeekPrec >= RHSPrec) { + if (EvaluateDirectiveSubExpr(RHS, RHSPrec, PeekTok, RHSIsLive, PP)) + return true; + PeekPrec = getPrecedence(PeekTok.getKind()); + } + assert(PeekPrec <= ThisPrec && "Recursion didn't work!"); + + // Usual arithmetic conversions (C99 6.3.1.8p1): result is unsigned if + // either operand is unsigned. + llvm::APSInt Res(LHS.getBitWidth()); + switch (Operator) { + case tok::question: // No UAC for x and y in "x ? y : z". + case tok::lessless: // Shift amount doesn't UAC with shift value. + case tok::greatergreater: // Shift amount doesn't UAC with shift value. + case tok::comma: // Comma operands are not subject to UACs. + case tok::pipepipe: // Logical || does not do UACs. + case tok::ampamp: // Logical && does not do UACs. + break; // No UAC + default: + Res.setIsUnsigned(LHS.isUnsigned()|RHS.isUnsigned()); + // If this just promoted something from signed to unsigned, and if the + // value was negative, warn about it. + if (ValueLive && Res.isUnsigned()) { + if (!LHS.isUnsigned() && LHS.Val.isNegative()) + PP.Diag(OpLoc, diag::warn_pp_convert_lhs_to_positive) + << LHS.Val.toString(10, true) + " to " + + LHS.Val.toString(10, false) + << LHS.getRange() << RHS.getRange(); + if (!RHS.isUnsigned() && RHS.Val.isNegative()) + PP.Diag(OpLoc, diag::warn_pp_convert_rhs_to_positive) + << RHS.Val.toString(10, true) + " to " + + RHS.Val.toString(10, false) + << LHS.getRange() << RHS.getRange(); + } + LHS.Val.setIsUnsigned(Res.isUnsigned()); + RHS.Val.setIsUnsigned(Res.isUnsigned()); + } + + // FIXME: All of these should detect and report overflow?? + bool Overflow = false; + switch (Operator) { + default: assert(0 && "Unknown operator token!"); + case tok::percent: + if (RHS.Val != 0) + Res = LHS.Val % RHS.Val; + else if (ValueLive) { + PP.Diag(OpLoc, diag::err_pp_remainder_by_zero) + << LHS.getRange() << RHS.getRange(); + return true; + } + break; + case tok::slash: + if (RHS.Val != 0) { + Res = LHS.Val / RHS.Val; + if (LHS.Val.isSigned()) // MININT/-1 --> overflow. + Overflow = LHS.Val.isMinSignedValue() && RHS.Val.isAllOnesValue(); + } else if (ValueLive) { + PP.Diag(OpLoc, diag::err_pp_division_by_zero) + << LHS.getRange() << RHS.getRange(); + return true; + } + break; + + case tok::star: + Res = LHS.Val * RHS.Val; + if (Res.isSigned() && LHS.Val != 0 && RHS.Val != 0) + Overflow = Res/RHS.Val != LHS.Val || Res/LHS.Val != RHS.Val; + break; + case tok::lessless: { + // Determine whether overflow is about to happen. + unsigned ShAmt = static_cast<unsigned>(RHS.Val.getLimitedValue()); + if (ShAmt >= LHS.Val.getBitWidth()) + Overflow = true, ShAmt = LHS.Val.getBitWidth()-1; + else if (LHS.isUnsigned()) + Overflow = false; + else if (LHS.Val.isNonNegative()) // Don't allow sign change. + Overflow = ShAmt >= LHS.Val.countLeadingZeros(); + else + Overflow = ShAmt >= LHS.Val.countLeadingOnes(); + + Res = LHS.Val << ShAmt; + break; + } + case tok::greatergreater: { + // Determine whether overflow is about to happen. + unsigned ShAmt = static_cast<unsigned>(RHS.Val.getLimitedValue()); + if (ShAmt >= LHS.getBitWidth()) + Overflow = true, ShAmt = LHS.getBitWidth()-1; + Res = LHS.Val >> ShAmt; + break; + } + case tok::plus: + Res = LHS.Val + RHS.Val; + if (LHS.isUnsigned()) + Overflow = false; + else if (LHS.Val.isNonNegative() == RHS.Val.isNonNegative() && + Res.isNonNegative() != LHS.Val.isNonNegative()) + Overflow = true; // Overflow for signed addition. + break; + case tok::minus: + Res = LHS.Val - RHS.Val; + if (LHS.isUnsigned()) + Overflow = false; + else if (LHS.Val.isNonNegative() != RHS.Val.isNonNegative() && + Res.isNonNegative() != LHS.Val.isNonNegative()) + Overflow = true; // Overflow for signed subtraction. + break; + case tok::lessequal: + Res = LHS.Val <= RHS.Val; + Res.setIsUnsigned(false); // C99 6.5.8p6, result is always int (signed) + break; + case tok::less: + Res = LHS.Val < RHS.Val; + Res.setIsUnsigned(false); // C99 6.5.8p6, result is always int (signed) + break; + case tok::greaterequal: + Res = LHS.Val >= RHS.Val; + Res.setIsUnsigned(false); // C99 6.5.8p6, result is always int (signed) + break; + case tok::greater: + Res = LHS.Val > RHS.Val; + Res.setIsUnsigned(false); // C99 6.5.8p6, result is always int (signed) + break; + case tok::exclaimequal: + Res = LHS.Val != RHS.Val; + Res.setIsUnsigned(false); // C99 6.5.9p3, result is always int (signed) + break; + case tok::equalequal: + Res = LHS.Val == RHS.Val; + Res.setIsUnsigned(false); // C99 6.5.9p3, result is always int (signed) + break; + case tok::amp: + Res = LHS.Val & RHS.Val; + break; + case tok::caret: + Res = LHS.Val ^ RHS.Val; + break; + case tok::pipe: + Res = LHS.Val | RHS.Val; + break; + case tok::ampamp: + Res = (LHS.Val != 0 && RHS.Val != 0); + Res.setIsUnsigned(false); // C99 6.5.13p3, result is always int (signed) + break; + case tok::pipepipe: + Res = (LHS.Val != 0 || RHS.Val != 0); + Res.setIsUnsigned(false); // C99 6.5.14p3, result is always int (signed) + break; + case tok::comma: + // Comma is invalid in pp expressions in c89/c++ mode, but is valid in C99 + // if not being evaluated. + if (!PP.getLangOptions().C99 || ValueLive) + PP.Diag(OpLoc, diag::ext_pp_comma_expr) + << LHS.getRange() << RHS.getRange(); + Res = RHS.Val; // LHS = LHS,RHS -> RHS. + break; + case tok::question: { + // Parse the : part of the expression. + if (PeekTok.isNot(tok::colon)) { + PP.Diag(PeekTok.getLocation(), diag::err_expected_colon) + << LHS.getRange(), RHS.getRange(); + PP.Diag(OpLoc, diag::note_matching) << "?"; + return true; + } + // Consume the :. + PP.LexNonComment(PeekTok); + + // Evaluate the value after the :. + bool AfterColonLive = ValueLive && LHS.Val == 0; + PPValue AfterColonVal(LHS.getBitWidth()); + DefinedTracker DT; + if (EvaluateValue(AfterColonVal, PeekTok, DT, AfterColonLive, PP)) + return true; + + // Parse anything after the : with the same precedence as ?. We allow + // things of equal precedence because ?: is right associative. + if (EvaluateDirectiveSubExpr(AfterColonVal, ThisPrec, + PeekTok, AfterColonLive, PP)) + return true; + + // Now that we have the condition, the LHS and the RHS of the :, evaluate. + Res = LHS.Val != 0 ? RHS.Val : AfterColonVal.Val; + RHS.setEnd(AfterColonVal.getRange().getEnd()); + + // Usual arithmetic conversions (C99 6.3.1.8p1): result is unsigned if + // either operand is unsigned. + Res.setIsUnsigned(RHS.isUnsigned() | AfterColonVal.isUnsigned()); + + // Figure out the precedence of the token after the : part. + PeekPrec = getPrecedence(PeekTok.getKind()); + break; + } + case tok::colon: + // Don't allow :'s to float around without being part of ?: exprs. + PP.Diag(OpLoc, diag::err_pp_colon_without_question) + << LHS.getRange() << RHS.getRange(); + return true; + } + + // If this operator is live and overflowed, report the issue. + if (Overflow && ValueLive) + PP.Diag(OpLoc, diag::warn_pp_expr_overflow) + << LHS.getRange() << RHS.getRange(); + + // Put the result back into 'LHS' for our next iteration. + LHS.Val = Res; + LHS.setEnd(RHS.getRange().getEnd()); + } + + return false; +} + +/// EvaluateDirectiveExpression - Evaluate an integer constant expression that +/// may occur after a #if or #elif directive. If the expression is equivalent +/// to "!defined(X)" return X in IfNDefMacro. +bool Preprocessor:: +EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro) { + // Peek ahead one token. + Token Tok; + Lex(Tok); + + // C99 6.10.1p3 - All expressions are evaluated as intmax_t or uintmax_t. + unsigned BitWidth = getTargetInfo().getIntMaxTWidth(); + + PPValue ResVal(BitWidth); + DefinedTracker DT; + if (EvaluateValue(ResVal, Tok, DT, true, *this)) { + // Parse error, skip the rest of the macro line. + if (Tok.isNot(tok::eom)) + DiscardUntilEndOfDirective(); + return false; + } + + // If we are at the end of the expression after just parsing a value, there + // must be no (unparenthesized) binary operators involved, so we can exit + // directly. + if (Tok.is(tok::eom)) { + // If the expression we parsed was of the form !defined(macro), return the + // macro in IfNDefMacro. + if (DT.State == DefinedTracker::NotDefinedMacro) + IfNDefMacro = DT.TheMacro; + + return ResVal.Val != 0; + } + + // Otherwise, we must have a binary operator (e.g. "#if 1 < 2"), so parse the + // operator and the stuff after it. + if (EvaluateDirectiveSubExpr(ResVal, getPrecedence(tok::question), + Tok, true, *this)) { + // Parse error, skip the rest of the macro line. + if (Tok.isNot(tok::eom)) + DiscardUntilEndOfDirective(); + return false; + } + + // If we aren't at the tok::eom token, something bad happened, like an extra + // ')' token. + if (Tok.isNot(tok::eom)) { + Diag(Tok, diag::err_pp_expected_eol); + DiscardUntilEndOfDirective(); + } + + return ResVal.Val != 0; +} + diff --git a/lib/Lex/PPLexerChange.cpp b/lib/Lex/PPLexerChange.cpp new file mode 100644 index 0000000..2a05ba3 --- /dev/null +++ b/lib/Lex/PPLexerChange.cpp @@ -0,0 +1,345 @@ +//===--- PPLexerChange.cpp - Handle changing lexers in the preprocessor ---===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements pieces of the Preprocessor interface that manage the +// current lexer stack. +// +//===----------------------------------------------------------------------===// + +#include "clang/Lex/Preprocessor.h" +#include "clang/Lex/HeaderSearch.h" +#include "clang/Lex/MacroInfo.h" +#include "clang/Lex/LexDiagnostic.h" +#include "clang/Basic/SourceManager.h" +#include "llvm/Support/MemoryBuffer.h" +using namespace clang; + +PPCallbacks::~PPCallbacks() {} + +//===----------------------------------------------------------------------===// +// Miscellaneous Methods. +//===----------------------------------------------------------------------===// + +/// isInPrimaryFile - Return true if we're in the top-level file, not in a +/// #include. This looks through macro expansions and active _Pragma lexers. +bool Preprocessor::isInPrimaryFile() const { + if (IsFileLexer()) + return IncludeMacroStack.empty(); + + // If there are any stacked lexers, we're in a #include. + assert(IsFileLexer(IncludeMacroStack[0]) && + "Top level include stack isn't our primary lexer?"); + for (unsigned i = 1, e = IncludeMacroStack.size(); i != e; ++i) + if (IsFileLexer(IncludeMacroStack[i])) + return false; + return true; +} + +/// getCurrentLexer - Return the current file lexer being lexed from. Note +/// that this ignores any potentially active macro expansions and _Pragma +/// expansions going on at the time. +PreprocessorLexer *Preprocessor::getCurrentFileLexer() const { + if (IsFileLexer()) + return CurPPLexer; + + // Look for a stacked lexer. + for (unsigned i = IncludeMacroStack.size(); i != 0; --i) { + const IncludeStackInfo& ISI = IncludeMacroStack[i-1]; + if (IsFileLexer(ISI)) + return ISI.ThePPLexer; + } + return 0; +} + + +//===----------------------------------------------------------------------===// +// Methods for Entering and Callbacks for leaving various contexts +//===----------------------------------------------------------------------===// + +/// EnterSourceFile - Add a source file to the top of the include stack and +/// start lexing tokens from it instead of the current buffer. Return true +/// on failure. +void Preprocessor::EnterSourceFile(FileID FID, const DirectoryLookup *CurDir) { + assert(CurTokenLexer == 0 && "Cannot #include a file inside a macro!"); + ++NumEnteredSourceFiles; + + if (MaxIncludeStackDepth < IncludeMacroStack.size()) + MaxIncludeStackDepth = IncludeMacroStack.size(); + + if (PTH) { + if (PTHLexer *PL = PTH->CreateLexer(FID)) + return EnterSourceFileWithPTH(PL, CurDir); + } + EnterSourceFileWithLexer(new Lexer(FID, *this), CurDir); +} + +/// EnterSourceFileWithLexer - Add a source file to the top of the include stack +/// and start lexing tokens from it instead of the current buffer. +void Preprocessor::EnterSourceFileWithLexer(Lexer *TheLexer, + const DirectoryLookup *CurDir) { + + // Add the current lexer to the include stack. + if (CurPPLexer || CurTokenLexer) + PushIncludeMacroStack(); + + CurLexer.reset(TheLexer); + CurPPLexer = TheLexer; + CurDirLookup = CurDir; + + // Notify the client, if desired, that we are in a new source file. + if (Callbacks && !CurLexer->Is_PragmaLexer) { + SrcMgr::CharacteristicKind FileType = + SourceMgr.getFileCharacteristic(CurLexer->getFileLoc()); + + Callbacks->FileChanged(CurLexer->getFileLoc(), + PPCallbacks::EnterFile, FileType); + } +} + +/// EnterSourceFileWithPTH - Add a source file to the top of the include stack +/// and start getting tokens from it using the PTH cache. +void Preprocessor::EnterSourceFileWithPTH(PTHLexer *PL, + const DirectoryLookup *CurDir) { + + if (CurPPLexer || CurTokenLexer) + PushIncludeMacroStack(); + + CurDirLookup = CurDir; + CurPTHLexer.reset(PL); + CurPPLexer = CurPTHLexer.get(); + + // Notify the client, if desired, that we are in a new source file. + if (Callbacks) { + FileID FID = CurPPLexer->getFileID(); + SourceLocation EnterLoc = SourceMgr.getLocForStartOfFile(FID); + SrcMgr::CharacteristicKind FileType = + SourceMgr.getFileCharacteristic(EnterLoc); + Callbacks->FileChanged(EnterLoc, PPCallbacks::EnterFile, FileType); + } +} + +/// EnterMacro - Add a Macro to the top of the include stack and start lexing +/// tokens from it instead of the current buffer. +void Preprocessor::EnterMacro(Token &Tok, SourceLocation ILEnd, + MacroArgs *Args) { + PushIncludeMacroStack(); + CurDirLookup = 0; + + if (NumCachedTokenLexers == 0) { + CurTokenLexer.reset(new TokenLexer(Tok, ILEnd, Args, *this)); + } else { + CurTokenLexer.reset(TokenLexerCache[--NumCachedTokenLexers]); + CurTokenLexer->Init(Tok, ILEnd, Args); + } +} + +/// EnterTokenStream - Add a "macro" context to the top of the include stack, +/// which will cause the lexer to start returning the specified tokens. +/// +/// If DisableMacroExpansion is true, tokens lexed from the token stream will +/// not be subject to further macro expansion. Otherwise, these tokens will +/// be re-macro-expanded when/if expansion is enabled. +/// +/// If OwnsTokens is false, this method assumes that the specified stream of +/// tokens has a permanent owner somewhere, so they do not need to be copied. +/// If it is true, it assumes the array of tokens is allocated with new[] and +/// must be freed. +/// +void Preprocessor::EnterTokenStream(const Token *Toks, unsigned NumToks, + bool DisableMacroExpansion, + bool OwnsTokens) { + // Save our current state. + PushIncludeMacroStack(); + CurDirLookup = 0; + + // Create a macro expander to expand from the specified token stream. + if (NumCachedTokenLexers == 0) { + CurTokenLexer.reset(new TokenLexer(Toks, NumToks, DisableMacroExpansion, + OwnsTokens, *this)); + } else { + CurTokenLexer.reset(TokenLexerCache[--NumCachedTokenLexers]); + CurTokenLexer->Init(Toks, NumToks, DisableMacroExpansion, OwnsTokens); + } +} + +/// HandleEndOfFile - This callback is invoked when the lexer hits the end of +/// the current file. This either returns the EOF token or pops a level off +/// the include stack and keeps going. +bool Preprocessor::HandleEndOfFile(Token &Result, bool isEndOfMacro) { + assert(!CurTokenLexer && + "Ending a file when currently in a macro!"); + + // See if this file had a controlling macro. + if (CurPPLexer) { // Not ending a macro, ignore it. + if (const IdentifierInfo *ControllingMacro = + CurPPLexer->MIOpt.GetControllingMacroAtEndOfFile()) { + // Okay, this has a controlling macro, remember in HeaderFileInfo. + if (const FileEntry *FE = + SourceMgr.getFileEntryForID(CurPPLexer->getFileID())) + HeaderInfo.SetFileControllingMacro(FE, ControllingMacro); + } + } + + // If this is a #include'd file, pop it off the include stack and continue + // lexing the #includer file. + if (!IncludeMacroStack.empty()) { + // We're done with the #included file. + RemoveTopOfLexerStack(); + + // Notify the client, if desired, that we are in a new source file. + if (Callbacks && !isEndOfMacro && CurPPLexer) { + SrcMgr::CharacteristicKind FileType = + SourceMgr.getFileCharacteristic(CurPPLexer->getSourceLocation()); + Callbacks->FileChanged(CurPPLexer->getSourceLocation(), + PPCallbacks::ExitFile, FileType); + } + + // Client should lex another token. + return false; + } + + // If the file ends with a newline, form the EOF token on the newline itself, + // rather than "on the line following it", which doesn't exist. This makes + // diagnostics relating to the end of file include the last file that the user + // actually typed, which is goodness. + if (CurLexer) { + const char *EndPos = CurLexer->BufferEnd; + if (EndPos != CurLexer->BufferStart && + (EndPos[-1] == '\n' || EndPos[-1] == '\r')) { + --EndPos; + + // Handle \n\r and \r\n: + if (EndPos != CurLexer->BufferStart && + (EndPos[-1] == '\n' || EndPos[-1] == '\r') && + EndPos[-1] != EndPos[0]) + --EndPos; + } + + Result.startToken(); + CurLexer->BufferPtr = EndPos; + CurLexer->FormTokenWithChars(Result, EndPos, tok::eof); + + // We're done with the #included file. + CurLexer.reset(); + } else { + assert(CurPTHLexer && "Got EOF but no current lexer set!"); + CurPTHLexer->getEOF(Result); + CurPTHLexer.reset(); + } + + CurPPLexer = 0; + + // This is the end of the top-level file. If the diag::pp_macro_not_used + // diagnostic is enabled, look for macros that have not been used. + if (getDiagnostics().getDiagnosticLevel(diag::pp_macro_not_used) != + Diagnostic::Ignored) { + for (macro_iterator I = macro_begin(), E = macro_end(); I != E; ++I) + if (!I->second->isUsed()) + Diag(I->second->getDefinitionLoc(), diag::pp_macro_not_used); + } + return true; +} + +/// HandleEndOfTokenLexer - This callback is invoked when the current TokenLexer +/// hits the end of its token stream. +bool Preprocessor::HandleEndOfTokenLexer(Token &Result) { + assert(CurTokenLexer && !CurPPLexer && + "Ending a macro when currently in a #include file!"); + + // Delete or cache the now-dead macro expander. + if (NumCachedTokenLexers == TokenLexerCacheSize) + CurTokenLexer.reset(); + else + TokenLexerCache[NumCachedTokenLexers++] = CurTokenLexer.take(); + + // Handle this like a #include file being popped off the stack. + return HandleEndOfFile(Result, true); +} + +/// RemoveTopOfLexerStack - Pop the current lexer/macro exp off the top of the +/// lexer stack. This should only be used in situations where the current +/// state of the top-of-stack lexer is unknown. +void Preprocessor::RemoveTopOfLexerStack() { + assert(!IncludeMacroStack.empty() && "Ran out of stack entries to load"); + + if (CurTokenLexer) { + // Delete or cache the now-dead macro expander. + if (NumCachedTokenLexers == TokenLexerCacheSize) + CurTokenLexer.reset(); + else + TokenLexerCache[NumCachedTokenLexers++] = CurTokenLexer.take(); + } + + PopIncludeMacroStack(); +} + +/// HandleMicrosoftCommentPaste - When the macro expander pastes together a +/// comment (/##/) in microsoft mode, this method handles updating the current +/// state, returning the token on the next source line. +void Preprocessor::HandleMicrosoftCommentPaste(Token &Tok) { + assert(CurTokenLexer && !CurPPLexer && + "Pasted comment can only be formed from macro"); + + // We handle this by scanning for the closest real lexer, switching it to + // raw mode and preprocessor mode. This will cause it to return \n as an + // explicit EOM token. + PreprocessorLexer *FoundLexer = 0; + bool LexerWasInPPMode = false; + for (unsigned i = 0, e = IncludeMacroStack.size(); i != e; ++i) { + IncludeStackInfo &ISI = *(IncludeMacroStack.end()-i-1); + if (ISI.ThePPLexer == 0) continue; // Scan for a real lexer. + + // Once we find a real lexer, mark it as raw mode (disabling macro + // expansions) and preprocessor mode (return EOM). We know that the lexer + // was *not* in raw mode before, because the macro that the comment came + // from was expanded. However, it could have already been in preprocessor + // mode (#if COMMENT) in which case we have to return it to that mode and + // return EOM. + FoundLexer = ISI.ThePPLexer; + FoundLexer->LexingRawMode = true; + LexerWasInPPMode = FoundLexer->ParsingPreprocessorDirective; + FoundLexer->ParsingPreprocessorDirective = true; + break; + } + + // Okay, we either found and switched over the lexer, or we didn't find a + // lexer. In either case, finish off the macro the comment came from, getting + // the next token. + if (!HandleEndOfTokenLexer(Tok)) Lex(Tok); + + // Discarding comments as long as we don't have EOF or EOM. This 'comments + // out' the rest of the line, including any tokens that came from other macros + // that were active, as in: + // #define submacro a COMMENT b + // submacro c + // which should lex to 'a' only: 'b' and 'c' should be removed. + while (Tok.isNot(tok::eom) && Tok.isNot(tok::eof)) + Lex(Tok); + + // If we got an eom token, then we successfully found the end of the line. + if (Tok.is(tok::eom)) { + assert(FoundLexer && "Can't get end of line without an active lexer"); + // Restore the lexer back to normal mode instead of raw mode. + FoundLexer->LexingRawMode = false; + + // If the lexer was already in preprocessor mode, just return the EOM token + // to finish the preprocessor line. + if (LexerWasInPPMode) return; + + // Otherwise, switch out of PP mode and return the next lexed token. + FoundLexer->ParsingPreprocessorDirective = false; + return Lex(Tok); + } + + // If we got an EOF token, then we reached the end of the token stream but + // didn't find an explicit \n. This can only happen if there was no lexer + // active (an active lexer would return EOM at EOF if there was no \n in + // preprocessor directive mode), so just return EOF as our token. + assert(!FoundLexer && "Lexer should return EOM before EOF in PP mode"); +} diff --git a/lib/Lex/PPMacroExpansion.cpp b/lib/Lex/PPMacroExpansion.cpp new file mode 100644 index 0000000..55222c9 --- /dev/null +++ b/lib/Lex/PPMacroExpansion.cpp @@ -0,0 +1,605 @@ +//===--- MacroExpansion.cpp - Top level Macro Expansion -------------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements the top level handling of macro expasion for the +// preprocessor. +// +//===----------------------------------------------------------------------===// + +#include "clang/Lex/Preprocessor.h" +#include "MacroArgs.h" +#include "clang/Lex/MacroInfo.h" +#include "clang/Basic/SourceManager.h" +#include "clang/Basic/FileManager.h" +#include "clang/Lex/LexDiagnostic.h" +#include <cstdio> +#include <ctime> +using namespace clang; + +/// setMacroInfo - Specify a macro for this identifier. +/// +void Preprocessor::setMacroInfo(IdentifierInfo *II, MacroInfo *MI) { + if (MI) { + Macros[II] = MI; + II->setHasMacroDefinition(true); + } else if (II->hasMacroDefinition()) { + Macros.erase(II); + II->setHasMacroDefinition(false); + } +} + +/// RegisterBuiltinMacro - Register the specified identifier in the identifier +/// table and mark it as a builtin macro to be expanded. +IdentifierInfo *Preprocessor::RegisterBuiltinMacro(const char *Name) { + // Get the identifier. + IdentifierInfo *Id = getIdentifierInfo(Name); + + // Mark it as being a macro that is builtin. + MacroInfo *MI = AllocateMacroInfo(SourceLocation()); + MI->setIsBuiltinMacro(); + setMacroInfo(Id, MI); + return Id; +} + + +/// RegisterBuiltinMacros - Register builtin macros, such as __LINE__ with the +/// identifier table. +void Preprocessor::RegisterBuiltinMacros() { + Ident__LINE__ = RegisterBuiltinMacro("__LINE__"); + Ident__FILE__ = RegisterBuiltinMacro("__FILE__"); + Ident__DATE__ = RegisterBuiltinMacro("__DATE__"); + Ident__TIME__ = RegisterBuiltinMacro("__TIME__"); + Ident__COUNTER__ = RegisterBuiltinMacro("__COUNTER__"); + Ident_Pragma = RegisterBuiltinMacro("_Pragma"); + + // GCC Extensions. + Ident__BASE_FILE__ = RegisterBuiltinMacro("__BASE_FILE__"); + Ident__INCLUDE_LEVEL__ = RegisterBuiltinMacro("__INCLUDE_LEVEL__"); + Ident__TIMESTAMP__ = RegisterBuiltinMacro("__TIMESTAMP__"); +} + +/// isTrivialSingleTokenExpansion - Return true if MI, which has a single token +/// in its expansion, currently expands to that token literally. +static bool isTrivialSingleTokenExpansion(const MacroInfo *MI, + const IdentifierInfo *MacroIdent, + Preprocessor &PP) { + IdentifierInfo *II = MI->getReplacementToken(0).getIdentifierInfo(); + + // If the token isn't an identifier, it's always literally expanded. + if (II == 0) return true; + + // If the identifier is a macro, and if that macro is enabled, it may be + // expanded so it's not a trivial expansion. + if (II->hasMacroDefinition() && PP.getMacroInfo(II)->isEnabled() && + // Fast expanding "#define X X" is ok, because X would be disabled. + II != MacroIdent) + return false; + + // If this is an object-like macro invocation, it is safe to trivially expand + // it. + if (MI->isObjectLike()) return true; + + // If this is a function-like macro invocation, it's safe to trivially expand + // as long as the identifier is not a macro argument. + for (MacroInfo::arg_iterator I = MI->arg_begin(), E = MI->arg_end(); + I != E; ++I) + if (*I == II) + return false; // Identifier is a macro argument. + + return true; +} + + +/// isNextPPTokenLParen - Determine whether the next preprocessor token to be +/// lexed is a '('. If so, consume the token and return true, if not, this +/// method should have no observable side-effect on the lexed tokens. +bool Preprocessor::isNextPPTokenLParen() { + // Do some quick tests for rejection cases. + unsigned Val; + if (CurLexer) + Val = CurLexer->isNextPPTokenLParen(); + else if (CurPTHLexer) + Val = CurPTHLexer->isNextPPTokenLParen(); + else + Val = CurTokenLexer->isNextTokenLParen(); + + if (Val == 2) { + // We have run off the end. If it's a source file we don't + // examine enclosing ones (C99 5.1.1.2p4). Otherwise walk up the + // macro stack. + if (CurPPLexer) + return false; + for (unsigned i = IncludeMacroStack.size(); i != 0; --i) { + IncludeStackInfo &Entry = IncludeMacroStack[i-1]; + if (Entry.TheLexer) + Val = Entry.TheLexer->isNextPPTokenLParen(); + else if (Entry.ThePTHLexer) + Val = Entry.ThePTHLexer->isNextPPTokenLParen(); + else + Val = Entry.TheTokenLexer->isNextTokenLParen(); + + if (Val != 2) + break; + + // Ran off the end of a source file? + if (Entry.ThePPLexer) + return false; + } + } + + // Okay, if we know that the token is a '(', lex it and return. Otherwise we + // have found something that isn't a '(' or we found the end of the + // translation unit. In either case, return false. + return Val == 1; +} + +/// HandleMacroExpandedIdentifier - If an identifier token is read that is to be +/// expanded as a macro, handle it and return the next token as 'Identifier'. +bool Preprocessor::HandleMacroExpandedIdentifier(Token &Identifier, + MacroInfo *MI) { + if (Callbacks) Callbacks->MacroExpands(Identifier, MI); + + // If this is a macro exapnsion in the "#if !defined(x)" line for the file, + // then the macro could expand to different things in other contexts, we need + // to disable the optimization in this case. + if (CurPPLexer) CurPPLexer->MIOpt.ExpandedMacro(); + + // If this is a builtin macro, like __LINE__ or _Pragma, handle it specially. + if (MI->isBuiltinMacro()) { + ExpandBuiltinMacro(Identifier); + return false; + } + + /// Args - If this is a function-like macro expansion, this contains, + /// for each macro argument, the list of tokens that were provided to the + /// invocation. + MacroArgs *Args = 0; + + // Remember where the end of the instantiation occurred. For an object-like + // macro, this is the identifier. For a function-like macro, this is the ')'. + SourceLocation InstantiationEnd = Identifier.getLocation(); + + // If this is a function-like macro, read the arguments. + if (MI->isFunctionLike()) { + // C99 6.10.3p10: If the preprocessing token immediately after the the macro + // name isn't a '(', this macro should not be expanded. + if (!isNextPPTokenLParen()) + return true; + + // Remember that we are now parsing the arguments to a macro invocation. + // Preprocessor directives used inside macro arguments are not portable, and + // this enables the warning. + InMacroArgs = true; + Args = ReadFunctionLikeMacroArgs(Identifier, MI, InstantiationEnd); + + // Finished parsing args. + InMacroArgs = false; + + // If there was an error parsing the arguments, bail out. + if (Args == 0) return false; + + ++NumFnMacroExpanded; + } else { + ++NumMacroExpanded; + } + + // Notice that this macro has been used. + MI->setIsUsed(true); + + // If we started lexing a macro, enter the macro expansion body. + + // If this macro expands to no tokens, don't bother to push it onto the + // expansion stack, only to take it right back off. + if (MI->getNumTokens() == 0) { + // No need for arg info. + if (Args) Args->destroy(); + + // Ignore this macro use, just return the next token in the current + // buffer. + bool HadLeadingSpace = Identifier.hasLeadingSpace(); + bool IsAtStartOfLine = Identifier.isAtStartOfLine(); + + Lex(Identifier); + + // If the identifier isn't on some OTHER line, inherit the leading + // whitespace/first-on-a-line property of this token. This handles + // stuff like "! XX," -> "! ," and " XX," -> " ,", when XX is + // empty. + if (!Identifier.isAtStartOfLine()) { + if (IsAtStartOfLine) Identifier.setFlag(Token::StartOfLine); + if (HadLeadingSpace) Identifier.setFlag(Token::LeadingSpace); + } + ++NumFastMacroExpanded; + return false; + + } else if (MI->getNumTokens() == 1 && + isTrivialSingleTokenExpansion(MI, Identifier.getIdentifierInfo(), + *this)) { + // Otherwise, if this macro expands into a single trivially-expanded + // token: expand it now. This handles common cases like + // "#define VAL 42". + + // No need for arg info. + if (Args) Args->destroy(); + + // Propagate the isAtStartOfLine/hasLeadingSpace markers of the macro + // identifier to the expanded token. + bool isAtStartOfLine = Identifier.isAtStartOfLine(); + bool hasLeadingSpace = Identifier.hasLeadingSpace(); + + // Remember where the token is instantiated. + SourceLocation InstantiateLoc = Identifier.getLocation(); + + // Replace the result token. + Identifier = MI->getReplacementToken(0); + + // Restore the StartOfLine/LeadingSpace markers. + Identifier.setFlagValue(Token::StartOfLine , isAtStartOfLine); + Identifier.setFlagValue(Token::LeadingSpace, hasLeadingSpace); + + // Update the tokens location to include both its instantiation and physical + // locations. + SourceLocation Loc = + SourceMgr.createInstantiationLoc(Identifier.getLocation(), InstantiateLoc, + InstantiationEnd,Identifier.getLength()); + Identifier.setLocation(Loc); + + // If this is #define X X, we must mark the result as unexpandible. + if (IdentifierInfo *NewII = Identifier.getIdentifierInfo()) + if (getMacroInfo(NewII) == MI) + Identifier.setFlag(Token::DisableExpand); + + // Since this is not an identifier token, it can't be macro expanded, so + // we're done. + ++NumFastMacroExpanded; + return false; + } + + // Start expanding the macro. + EnterMacro(Identifier, InstantiationEnd, Args); + + // Now that the macro is at the top of the include stack, ask the + // preprocessor to read the next token from it. + Lex(Identifier); + return false; +} + +/// ReadFunctionLikeMacroArgs - After reading "MACRO" and knowing that the next +/// token is the '(' of the macro, this method is invoked to read all of the +/// actual arguments specified for the macro invocation. This returns null on +/// error. +MacroArgs *Preprocessor::ReadFunctionLikeMacroArgs(Token &MacroName, + MacroInfo *MI, + SourceLocation &MacroEnd) { + // The number of fixed arguments to parse. + unsigned NumFixedArgsLeft = MI->getNumArgs(); + bool isVariadic = MI->isVariadic(); + + // Outer loop, while there are more arguments, keep reading them. + Token Tok; + + // Read arguments as unexpanded tokens. This avoids issues, e.g., where + // an argument value in a macro could expand to ',' or '(' or ')'. + LexUnexpandedToken(Tok); + assert(Tok.is(tok::l_paren) && "Error computing l-paren-ness?"); + + // ArgTokens - Build up a list of tokens that make up each argument. Each + // argument is separated by an EOF token. Use a SmallVector so we can avoid + // heap allocations in the common case. + llvm::SmallVector<Token, 64> ArgTokens; + + unsigned NumActuals = 0; + while (Tok.isNot(tok::r_paren)) { + assert((Tok.is(tok::l_paren) || Tok.is(tok::comma)) && + "only expect argument separators here"); + + unsigned ArgTokenStart = ArgTokens.size(); + SourceLocation ArgStartLoc = Tok.getLocation(); + + // C99 6.10.3p11: Keep track of the number of l_parens we have seen. Note + // that we already consumed the first one. + unsigned NumParens = 0; + + while (1) { + // Read arguments as unexpanded tokens. This avoids issues, e.g., where + // an argument value in a macro could expand to ',' or '(' or ')'. + LexUnexpandedToken(Tok); + + if (Tok.is(tok::eof) || Tok.is(tok::eom)) { // "#if f(<eof>" & "#if f(\n" + Diag(MacroName, diag::err_unterm_macro_invoc); + // Do not lose the EOF/EOM. Return it to the client. + MacroName = Tok; + return 0; + } else if (Tok.is(tok::r_paren)) { + // If we found the ) token, the macro arg list is done. + if (NumParens-- == 0) { + MacroEnd = Tok.getLocation(); + break; + } + } else if (Tok.is(tok::l_paren)) { + ++NumParens; + } else if (Tok.is(tok::comma) && NumParens == 0) { + // Comma ends this argument if there are more fixed arguments expected. + // However, if this is a variadic macro, and this is part of the + // variadic part, then the comma is just an argument token. + if (!isVariadic) break; + if (NumFixedArgsLeft > 1) + break; + } else if (Tok.is(tok::comment) && !KeepMacroComments) { + // If this is a comment token in the argument list and we're just in + // -C mode (not -CC mode), discard the comment. + continue; + } else if (Tok.getIdentifierInfo() != 0) { + // Reading macro arguments can cause macros that we are currently + // expanding from to be popped off the expansion stack. Doing so causes + // them to be reenabled for expansion. Here we record whether any + // identifiers we lex as macro arguments correspond to disabled macros. + // If so, we mark the token as noexpand. This is a subtle aspect of + // C99 6.10.3.4p2. + if (MacroInfo *MI = getMacroInfo(Tok.getIdentifierInfo())) + if (!MI->isEnabled()) + Tok.setFlag(Token::DisableExpand); + } + ArgTokens.push_back(Tok); + } + + // If this was an empty argument list foo(), don't add this as an empty + // argument. + if (ArgTokens.empty() && Tok.getKind() == tok::r_paren) + break; + + // If this is not a variadic macro, and too many args were specified, emit + // an error. + if (!isVariadic && NumFixedArgsLeft == 0) { + if (ArgTokens.size() != ArgTokenStart) + ArgStartLoc = ArgTokens[ArgTokenStart].getLocation(); + + // Emit the diagnostic at the macro name in case there is a missing ). + // Emitting it at the , could be far away from the macro name. + Diag(ArgStartLoc, diag::err_too_many_args_in_macro_invoc); + return 0; + } + + // Empty arguments are standard in C99 and supported as an extension in + // other modes. + if (ArgTokens.size() == ArgTokenStart && !Features.C99) + Diag(Tok, diag::ext_empty_fnmacro_arg); + + // Add a marker EOF token to the end of the token list for this argument. + Token EOFTok; + EOFTok.startToken(); + EOFTok.setKind(tok::eof); + EOFTok.setLocation(Tok.getLocation()); + EOFTok.setLength(0); + ArgTokens.push_back(EOFTok); + ++NumActuals; + assert(NumFixedArgsLeft != 0 && "Too many arguments parsed"); + --NumFixedArgsLeft; + } + + // Okay, we either found the r_paren. Check to see if we parsed too few + // arguments. + unsigned MinArgsExpected = MI->getNumArgs(); + + // See MacroArgs instance var for description of this. + bool isVarargsElided = false; + + if (NumActuals < MinArgsExpected) { + // There are several cases where too few arguments is ok, handle them now. + if (NumActuals == 0 && MinArgsExpected == 1) { + // #define A(X) or #define A(...) ---> A() + + // If there is exactly one argument, and that argument is missing, + // then we have an empty "()" argument empty list. This is fine, even if + // the macro expects one argument (the argument is just empty). + isVarargsElided = MI->isVariadic(); + } else if (MI->isVariadic() && + (NumActuals+1 == MinArgsExpected || // A(x, ...) -> A(X) + (NumActuals == 0 && MinArgsExpected == 2))) {// A(x,...) -> A() + // Varargs where the named vararg parameter is missing: ok as extension. + // #define A(x, ...) + // A("blah") + Diag(Tok, diag::ext_missing_varargs_arg); + + // Remember this occurred, allowing us to elide the comma when used for + // cases like: + // #define A(x, foo...) blah(a, ## foo) + // #define B(x, ...) blah(a, ## __VA_ARGS__) + // #define C(...) blah(a, ## __VA_ARGS__) + // A(x) B(x) C() + isVarargsElided = true; + } else { + // Otherwise, emit the error. + Diag(Tok, diag::err_too_few_args_in_macro_invoc); + return 0; + } + + // Add a marker EOF token to the end of the token list for this argument. + SourceLocation EndLoc = Tok.getLocation(); + Tok.startToken(); + Tok.setKind(tok::eof); + Tok.setLocation(EndLoc); + Tok.setLength(0); + ArgTokens.push_back(Tok); + + // If we expect two arguments, add both as empty. + if (NumActuals == 0 && MinArgsExpected == 2) + ArgTokens.push_back(Tok); + + } else if (NumActuals > MinArgsExpected && !MI->isVariadic()) { + // Emit the diagnostic at the macro name in case there is a missing ). + // Emitting it at the , could be far away from the macro name. + Diag(MacroName, diag::err_too_many_args_in_macro_invoc); + return 0; + } + + return MacroArgs::create(MI, ArgTokens.data(), ArgTokens.size(), + isVarargsElided); +} + +/// ComputeDATE_TIME - Compute the current time, enter it into the specified +/// scratch buffer, then return DATELoc/TIMELoc locations with the position of +/// the identifier tokens inserted. +static void ComputeDATE_TIME(SourceLocation &DATELoc, SourceLocation &TIMELoc, + Preprocessor &PP) { + time_t TT = time(0); + struct tm *TM = localtime(&TT); + + static const char * const Months[] = { + "Jan","Feb","Mar","Apr","May","Jun","Jul","Aug","Sep","Oct","Nov","Dec" + }; + + char TmpBuffer[100]; + sprintf(TmpBuffer, "\"%s %2d %4d\"", Months[TM->tm_mon], TM->tm_mday, + TM->tm_year+1900); + + Token TmpTok; + TmpTok.startToken(); + PP.CreateString(TmpBuffer, strlen(TmpBuffer), TmpTok); + DATELoc = TmpTok.getLocation(); + + sprintf(TmpBuffer, "\"%02d:%02d:%02d\"", TM->tm_hour, TM->tm_min, TM->tm_sec); + PP.CreateString(TmpBuffer, strlen(TmpBuffer), TmpTok); + TIMELoc = TmpTok.getLocation(); +} + +/// ExpandBuiltinMacro - If an identifier token is read that is to be expanded +/// as a builtin macro, handle it and return the next token as 'Tok'. +void Preprocessor::ExpandBuiltinMacro(Token &Tok) { + // Figure out which token this is. + IdentifierInfo *II = Tok.getIdentifierInfo(); + assert(II && "Can't be a macro without id info!"); + + // If this is an _Pragma directive, expand it, invoke the pragma handler, then + // lex the token after it. + if (II == Ident_Pragma) + return Handle_Pragma(Tok); + + ++NumBuiltinMacroExpanded; + + char TmpBuffer[100]; + + // Set up the return result. + Tok.setIdentifierInfo(0); + Tok.clearFlag(Token::NeedsCleaning); + + if (II == Ident__LINE__) { + // C99 6.10.8: "__LINE__: The presumed line number (within the current + // source file) of the current source line (an integer constant)". This can + // be affected by #line. + SourceLocation Loc = Tok.getLocation(); + + // Advance to the location of the first _, this might not be the first byte + // of the token if it starts with an escaped newline. + Loc = AdvanceToTokenCharacter(Loc, 0); + + // One wrinkle here is that GCC expands __LINE__ to location of the *end* of + // a macro instantiation. This doesn't matter for object-like macros, but + // can matter for a function-like macro that expands to contain __LINE__. + // Skip down through instantiation points until we find a file loc for the + // end of the instantiation history. + Loc = SourceMgr.getInstantiationRange(Loc).second; + PresumedLoc PLoc = SourceMgr.getPresumedLoc(Loc); + + // __LINE__ expands to a simple numeric value. + sprintf(TmpBuffer, "%u", PLoc.getLine()); + Tok.setKind(tok::numeric_constant); + CreateString(TmpBuffer, strlen(TmpBuffer), Tok, Tok.getLocation()); + } else if (II == Ident__FILE__ || II == Ident__BASE_FILE__) { + // C99 6.10.8: "__FILE__: The presumed name of the current source file (a + // character string literal)". This can be affected by #line. + PresumedLoc PLoc = SourceMgr.getPresumedLoc(Tok.getLocation()); + + // __BASE_FILE__ is a GNU extension that returns the top of the presumed + // #include stack instead of the current file. + if (II == Ident__BASE_FILE__) { + Diag(Tok, diag::ext_pp_base_file); + SourceLocation NextLoc = PLoc.getIncludeLoc(); + while (NextLoc.isValid()) { + PLoc = SourceMgr.getPresumedLoc(NextLoc); + NextLoc = PLoc.getIncludeLoc(); + } + } + + // Escape this filename. Turn '\' -> '\\' '"' -> '\"' + std::string FN = PLoc.getFilename(); + FN = '"' + Lexer::Stringify(FN) + '"'; + Tok.setKind(tok::string_literal); + CreateString(&FN[0], FN.size(), Tok, Tok.getLocation()); + } else if (II == Ident__DATE__) { + if (!DATELoc.isValid()) + ComputeDATE_TIME(DATELoc, TIMELoc, *this); + Tok.setKind(tok::string_literal); + Tok.setLength(strlen("\"Mmm dd yyyy\"")); + Tok.setLocation(SourceMgr.createInstantiationLoc(DATELoc, Tok.getLocation(), + Tok.getLocation(), + Tok.getLength())); + } else if (II == Ident__TIME__) { + if (!TIMELoc.isValid()) + ComputeDATE_TIME(DATELoc, TIMELoc, *this); + Tok.setKind(tok::string_literal); + Tok.setLength(strlen("\"hh:mm:ss\"")); + Tok.setLocation(SourceMgr.createInstantiationLoc(TIMELoc, Tok.getLocation(), + Tok.getLocation(), + Tok.getLength())); + } else if (II == Ident__INCLUDE_LEVEL__) { + Diag(Tok, diag::ext_pp_include_level); + + // Compute the presumed include depth of this token. This can be affected + // by GNU line markers. + unsigned Depth = 0; + + PresumedLoc PLoc = SourceMgr.getPresumedLoc(Tok.getLocation()); + PLoc = SourceMgr.getPresumedLoc(PLoc.getIncludeLoc()); + for (; PLoc.isValid(); ++Depth) + PLoc = SourceMgr.getPresumedLoc(PLoc.getIncludeLoc()); + + // __INCLUDE_LEVEL__ expands to a simple numeric value. + sprintf(TmpBuffer, "%u", Depth); + Tok.setKind(tok::numeric_constant); + CreateString(TmpBuffer, strlen(TmpBuffer), Tok, Tok.getLocation()); + } else if (II == Ident__TIMESTAMP__) { + // MSVC, ICC, GCC, VisualAge C++ extension. The generated string should be + // of the form "Ddd Mmm dd hh::mm::ss yyyy", which is returned by asctime. + Diag(Tok, diag::ext_pp_timestamp); + + // Get the file that we are lexing out of. If we're currently lexing from + // a macro, dig into the include stack. + const FileEntry *CurFile = 0; + PreprocessorLexer *TheLexer = getCurrentFileLexer(); + + if (TheLexer) + CurFile = SourceMgr.getFileEntryForID(TheLexer->getFileID()); + + // If this file is older than the file it depends on, emit a diagnostic. + const char *Result; + if (CurFile) { + time_t TT = CurFile->getModificationTime(); + struct tm *TM = localtime(&TT); + Result = asctime(TM); + } else { + Result = "??? ??? ?? ??:??:?? ????\n"; + } + TmpBuffer[0] = '"'; + strcpy(TmpBuffer+1, Result); + unsigned Len = strlen(TmpBuffer); + TmpBuffer[Len] = '"'; // Replace the newline with a quote. + Tok.setKind(tok::string_literal); + CreateString(TmpBuffer, Len+1, Tok, Tok.getLocation()); + } else if (II == Ident__COUNTER__) { + Diag(Tok, diag::ext_pp_counter); + + // __COUNTER__ expands to a simple numeric value. + sprintf(TmpBuffer, "%u", CounterValue++); + Tok.setKind(tok::numeric_constant); + CreateString(TmpBuffer, strlen(TmpBuffer), Tok, Tok.getLocation()); + } else { + assert(0 && "Unknown identifier!"); + } +} diff --git a/lib/Lex/PTHLexer.cpp b/lib/Lex/PTHLexer.cpp new file mode 100644 index 0000000..916bdef --- /dev/null +++ b/lib/Lex/PTHLexer.cpp @@ -0,0 +1,701 @@ +//===--- PTHLexer.cpp - Lex from a token stream ---------------------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements the PTHLexer interface. +// +//===----------------------------------------------------------------------===// + +#include "clang/Basic/TokenKinds.h" +#include "clang/Basic/FileManager.h" +#include "clang/Basic/IdentifierTable.h" +#include "clang/Basic/OnDiskHashTable.h" +#include "clang/Lex/PTHLexer.h" +#include "clang/Lex/Preprocessor.h" +#include "clang/Lex/PTHManager.h" +#include "clang/Lex/Token.h" +#include "clang/Lex/Preprocessor.h" +#include "llvm/ADT/StringMap.h" +#include "llvm/ADT/OwningPtr.h" +#include "llvm/Support/MemoryBuffer.h" +#include <sys/stat.h> +using namespace clang; +using namespace clang::io; + +#define DISK_TOKEN_SIZE (1+1+2+4+4) + +//===----------------------------------------------------------------------===// +// PTHLexer methods. +//===----------------------------------------------------------------------===// + +PTHLexer::PTHLexer(Preprocessor &PP, FileID FID, const unsigned char *D, + const unsigned char *ppcond, PTHManager &PM) + : PreprocessorLexer(&PP, FID), TokBuf(D), CurPtr(D), LastHashTokPtr(0), + PPCond(ppcond), CurPPCondPtr(ppcond), PTHMgr(PM) { + + FileStartLoc = PP.getSourceManager().getLocForStartOfFile(FID); +} + +void PTHLexer::Lex(Token& Tok) { +LexNextToken: + + //===--------------------------------------==// + // Read the raw token data. + //===--------------------------------------==// + + // Shadow CurPtr into an automatic variable. + const unsigned char *CurPtrShadow = CurPtr; + + // Read in the data for the token. + unsigned Word0 = ReadLE32(CurPtrShadow); + uint32_t IdentifierID = ReadLE32(CurPtrShadow); + uint32_t FileOffset = ReadLE32(CurPtrShadow); + + tok::TokenKind TKind = (tok::TokenKind) (Word0 & 0xFF); + Token::TokenFlags TFlags = (Token::TokenFlags) ((Word0 >> 8) & 0xFF); + uint32_t Len = Word0 >> 16; + + CurPtr = CurPtrShadow; + + //===--------------------------------------==// + // Construct the token itself. + //===--------------------------------------==// + + Tok.startToken(); + Tok.setKind(TKind); + Tok.setFlag(TFlags); + assert(!LexingRawMode); + Tok.setLocation(FileStartLoc.getFileLocWithOffset(FileOffset)); + Tok.setLength(Len); + + // Handle identifiers. + if (Tok.isLiteral()) { + Tok.setLiteralData((const char*) (PTHMgr.SpellingBase + IdentifierID)); + } + else if (IdentifierID) { + MIOpt.ReadToken(); + IdentifierInfo *II = PTHMgr.GetIdentifierInfo(IdentifierID-1); + + Tok.setIdentifierInfo(II); + + // Change the kind of this identifier to the appropriate token kind, e.g. + // turning "for" into a keyword. + Tok.setKind(II->getTokenID()); + + if (II->isHandleIdentifierCase()) + PP->HandleIdentifier(Tok); + return; + } + + //===--------------------------------------==// + // Process the token. + //===--------------------------------------==// +#if 0 + SourceManager& SM = PP->getSourceManager(); + llvm::cerr << SM.getFileEntryForID(FileID)->getName() + << ':' << SM.getLogicalLineNumber(Tok.getLocation()) + << ':' << SM.getLogicalColumnNumber(Tok.getLocation()) + << '\n'; +#endif + + if (TKind == tok::eof) { + // Save the end-of-file token. + EofToken = Tok; + + Preprocessor *PPCache = PP; + + assert(!ParsingPreprocessorDirective); + assert(!LexingRawMode); + + // FIXME: Issue diagnostics similar to Lexer. + if (PP->HandleEndOfFile(Tok, false)) + return; + + assert(PPCache && "Raw buffer::LexEndOfFile should return a token"); + return PPCache->Lex(Tok); + } + + if (TKind == tok::hash && Tok.isAtStartOfLine()) { + LastHashTokPtr = CurPtr - DISK_TOKEN_SIZE; + assert(!LexingRawMode); + PP->HandleDirective(Tok); + + if (PP->isCurrentLexer(this)) + goto LexNextToken; + + return PP->Lex(Tok); + } + + if (TKind == tok::eom) { + assert(ParsingPreprocessorDirective); + ParsingPreprocessorDirective = false; + return; + } + + MIOpt.ReadToken(); +} + +// FIXME: We can just grab the last token instead of storing a copy +// into EofToken. +void PTHLexer::getEOF(Token& Tok) { + assert(EofToken.is(tok::eof)); + Tok = EofToken; +} + +void PTHLexer::DiscardToEndOfLine() { + assert(ParsingPreprocessorDirective && ParsingFilename == false && + "Must be in a preprocessing directive!"); + + // We assume that if the preprocessor wishes to discard to the end of + // the line that it also means to end the current preprocessor directive. + ParsingPreprocessorDirective = false; + + // Skip tokens by only peeking at their token kind and the flags. + // We don't need to actually reconstruct full tokens from the token buffer. + // This saves some copies and it also reduces IdentifierInfo* lookup. + const unsigned char* p = CurPtr; + while (1) { + // Read the token kind. Are we at the end of the file? + tok::TokenKind x = (tok::TokenKind) (uint8_t) *p; + if (x == tok::eof) break; + + // Read the token flags. Are we at the start of the next line? + Token::TokenFlags y = (Token::TokenFlags) (uint8_t) p[1]; + if (y & Token::StartOfLine) break; + + // Skip to the next token. + p += DISK_TOKEN_SIZE; + } + + CurPtr = p; +} + +/// SkipBlock - Used by Preprocessor to skip the current conditional block. +bool PTHLexer::SkipBlock() { + assert(CurPPCondPtr && "No cached PP conditional information."); + assert(LastHashTokPtr && "No known '#' token."); + + const unsigned char* HashEntryI = 0; + uint32_t Offset; + uint32_t TableIdx; + + do { + // Read the token offset from the side-table. + Offset = ReadLE32(CurPPCondPtr); + + // Read the target table index from the side-table. + TableIdx = ReadLE32(CurPPCondPtr); + + // Compute the actual memory address of the '#' token data for this entry. + HashEntryI = TokBuf + Offset; + + // Optmization: "Sibling jumping". #if...#else...#endif blocks can + // contain nested blocks. In the side-table we can jump over these + // nested blocks instead of doing a linear search if the next "sibling" + // entry is not at a location greater than LastHashTokPtr. + if (HashEntryI < LastHashTokPtr && TableIdx) { + // In the side-table we are still at an entry for a '#' token that + // is earlier than the last one we saw. Check if the location we would + // stride gets us closer. + const unsigned char* NextPPCondPtr = + PPCond + TableIdx*(sizeof(uint32_t)*2); + assert(NextPPCondPtr >= CurPPCondPtr); + // Read where we should jump to. + uint32_t TmpOffset = ReadLE32(NextPPCondPtr); + const unsigned char* HashEntryJ = TokBuf + TmpOffset; + + if (HashEntryJ <= LastHashTokPtr) { + // Jump directly to the next entry in the side table. + HashEntryI = HashEntryJ; + Offset = TmpOffset; + TableIdx = ReadLE32(NextPPCondPtr); + CurPPCondPtr = NextPPCondPtr; + } + } + } + while (HashEntryI < LastHashTokPtr); + assert(HashEntryI == LastHashTokPtr && "No PP-cond entry found for '#'"); + assert(TableIdx && "No jumping from #endifs."); + + // Update our side-table iterator. + const unsigned char* NextPPCondPtr = PPCond + TableIdx*(sizeof(uint32_t)*2); + assert(NextPPCondPtr >= CurPPCondPtr); + CurPPCondPtr = NextPPCondPtr; + + // Read where we should jump to. + HashEntryI = TokBuf + ReadLE32(NextPPCondPtr); + uint32_t NextIdx = ReadLE32(NextPPCondPtr); + + // By construction NextIdx will be zero if this is a #endif. This is useful + // to know to obviate lexing another token. + bool isEndif = NextIdx == 0; + + // This case can occur when we see something like this: + // + // #if ... + // /* a comment or nothing */ + // #elif + // + // If we are skipping the first #if block it will be the case that CurPtr + // already points 'elif'. Just return. + + if (CurPtr > HashEntryI) { + assert(CurPtr == HashEntryI + DISK_TOKEN_SIZE); + // Did we reach a #endif? If so, go ahead and consume that token as well. + if (isEndif) + CurPtr += DISK_TOKEN_SIZE*2; + else + LastHashTokPtr = HashEntryI; + + return isEndif; + } + + // Otherwise, we need to advance. Update CurPtr to point to the '#' token. + CurPtr = HashEntryI; + + // Update the location of the last observed '#'. This is useful if we + // are skipping multiple blocks. + LastHashTokPtr = CurPtr; + + // Skip the '#' token. + assert(((tok::TokenKind)*CurPtr) == tok::hash); + CurPtr += DISK_TOKEN_SIZE; + + // Did we reach a #endif? If so, go ahead and consume that token as well. + if (isEndif) { CurPtr += DISK_TOKEN_SIZE*2; } + + return isEndif; +} + +SourceLocation PTHLexer::getSourceLocation() { + // getSourceLocation is not on the hot path. It is used to get the location + // of the next token when transitioning back to this lexer when done + // handling a #included file. Just read the necessary data from the token + // data buffer to construct the SourceLocation object. + // NOTE: This is a virtual function; hence it is defined out-of-line. + const unsigned char *OffsetPtr = CurPtr + (DISK_TOKEN_SIZE - 4); + uint32_t Offset = ReadLE32(OffsetPtr); + return FileStartLoc.getFileLocWithOffset(Offset); +} + +//===----------------------------------------------------------------------===// +// PTH file lookup: map from strings to file data. +//===----------------------------------------------------------------------===// + +/// PTHFileLookup - This internal data structure is used by the PTHManager +/// to map from FileEntry objects managed by FileManager to offsets within +/// the PTH file. +namespace { +class VISIBILITY_HIDDEN PTHFileData { + const uint32_t TokenOff; + const uint32_t PPCondOff; +public: + PTHFileData(uint32_t tokenOff, uint32_t ppCondOff) + : TokenOff(tokenOff), PPCondOff(ppCondOff) {} + + uint32_t getTokenOffset() const { return TokenOff; } + uint32_t getPPCondOffset() const { return PPCondOff; } +}; + + +class VISIBILITY_HIDDEN PTHFileLookupCommonTrait { +public: + typedef std::pair<unsigned char, const char*> internal_key_type; + + static unsigned ComputeHash(internal_key_type x) { + return BernsteinHash(x.second); + } + + static std::pair<unsigned, unsigned> + ReadKeyDataLength(const unsigned char*& d) { + unsigned keyLen = (unsigned) ReadUnalignedLE16(d); + unsigned dataLen = (unsigned) *(d++); + return std::make_pair(keyLen, dataLen); + } + + static internal_key_type ReadKey(const unsigned char* d, unsigned) { + unsigned char k = *(d++); // Read the entry kind. + return std::make_pair(k, (const char*) d); + } +}; + +class VISIBILITY_HIDDEN PTHFileLookupTrait : public PTHFileLookupCommonTrait { +public: + typedef const FileEntry* external_key_type; + typedef PTHFileData data_type; + + static internal_key_type GetInternalKey(const FileEntry* FE) { + return std::make_pair((unsigned char) 0x1, FE->getName()); + } + + static bool EqualKey(internal_key_type a, internal_key_type b) { + return a.first == b.first && strcmp(a.second, b.second) == 0; + } + + static PTHFileData ReadData(const internal_key_type& k, + const unsigned char* d, unsigned) { + assert(k.first == 0x1 && "Only file lookups can match!"); + uint32_t x = ::ReadUnalignedLE32(d); + uint32_t y = ::ReadUnalignedLE32(d); + return PTHFileData(x, y); + } +}; + +class VISIBILITY_HIDDEN PTHStringLookupTrait { +public: + typedef uint32_t + data_type; + + typedef const std::pair<const char*, unsigned> + external_key_type; + + typedef external_key_type internal_key_type; + + static bool EqualKey(const internal_key_type& a, + const internal_key_type& b) { + return (a.second == b.second) ? memcmp(a.first, b.first, a.second) == 0 + : false; + } + + static unsigned ComputeHash(const internal_key_type& a) { + return BernsteinHash(a.first, a.second); + } + + // This hopefully will just get inlined and removed by the optimizer. + static const internal_key_type& + GetInternalKey(const external_key_type& x) { return x; } + + static std::pair<unsigned, unsigned> + ReadKeyDataLength(const unsigned char*& d) { + return std::make_pair((unsigned) ReadUnalignedLE16(d), sizeof(uint32_t)); + } + + static std::pair<const char*, unsigned> + ReadKey(const unsigned char* d, unsigned n) { + assert(n >= 2 && d[n-1] == '\0'); + return std::make_pair((const char*) d, n-1); + } + + static uint32_t ReadData(const internal_key_type& k, const unsigned char* d, + unsigned) { + return ::ReadUnalignedLE32(d); + } +}; + +} // end anonymous namespace + +typedef OnDiskChainedHashTable<PTHFileLookupTrait> PTHFileLookup; +typedef OnDiskChainedHashTable<PTHStringLookupTrait> PTHStringIdLookup; + +//===----------------------------------------------------------------------===// +// PTHManager methods. +//===----------------------------------------------------------------------===// + +PTHManager::PTHManager(const llvm::MemoryBuffer* buf, void* fileLookup, + const unsigned char* idDataTable, + IdentifierInfo** perIDCache, + void* stringIdLookup, unsigned numIds, + const unsigned char* spellingBase, + const char* originalSourceFile) +: Buf(buf), PerIDCache(perIDCache), FileLookup(fileLookup), + IdDataTable(idDataTable), StringIdLookup(stringIdLookup), + NumIds(numIds), PP(0), SpellingBase(spellingBase), + OriginalSourceFile(originalSourceFile) {} + +PTHManager::~PTHManager() { + delete Buf; + delete (PTHFileLookup*) FileLookup; + delete (PTHStringIdLookup*) StringIdLookup; + free(PerIDCache); +} + +static void InvalidPTH(Diagnostic *Diags, Diagnostic::Level level, + const char* Msg = 0) { + if (!Diags) return; + if (!Msg) Msg = "Invalid or corrupted PTH file"; + unsigned DiagID = Diags->getCustomDiagID(level, Msg); + Diags->Report(FullSourceLoc(), DiagID); +} + +PTHManager* PTHManager::Create(const std::string& file, Diagnostic* Diags, + Diagnostic::Level level) { + // Memory map the PTH file. + llvm::OwningPtr<llvm::MemoryBuffer> + File(llvm::MemoryBuffer::getFile(file.c_str())); + + if (!File) { + if (Diags) { + unsigned DiagID = Diags->getCustomDiagID(level, + "PTH file %0 could not be read"); + Diags->Report(FullSourceLoc(), DiagID) << file; + } + + return 0; + } + + // Get the buffer ranges and check if there are at least three 32-bit + // words at the end of the file. + const unsigned char* BufBeg = (unsigned char*)File->getBufferStart(); + const unsigned char* BufEnd = (unsigned char*)File->getBufferEnd(); + + // Check the prologue of the file. + if ((BufEnd - BufBeg) < (signed) (sizeof("cfe-pth") + 3 + 4) || + memcmp(BufBeg, "cfe-pth", sizeof("cfe-pth") - 1) != 0) { + InvalidPTH(Diags, level); + return 0; + } + + // Read the PTH version. + const unsigned char *p = BufBeg + (sizeof("cfe-pth") - 1); + unsigned Version = ReadLE32(p); + + if (Version != PTHManager::Version) { + InvalidPTH(Diags, level, + Version < PTHManager::Version + ? "PTH file uses an older PTH format that is no longer supported" + : "PTH file uses a newer PTH format that cannot be read"); + return 0; + } + + // Compute the address of the index table at the end of the PTH file. + const unsigned char *PrologueOffset = p; + + if (PrologueOffset >= BufEnd) { + InvalidPTH(Diags, level); + return 0; + } + + // Construct the file lookup table. This will be used for mapping from + // FileEntry*'s to cached tokens. + const unsigned char* FileTableOffset = PrologueOffset + sizeof(uint32_t)*2; + const unsigned char* FileTable = BufBeg + ReadLE32(FileTableOffset); + + if (!(FileTable > BufBeg && FileTable < BufEnd)) { + InvalidPTH(Diags, level); + return 0; // FIXME: Proper error diagnostic? + } + + llvm::OwningPtr<PTHFileLookup> FL(PTHFileLookup::Create(FileTable, BufBeg)); + + // Warn if the PTH file is empty. We still want to create a PTHManager + // as the PTH could be used with -include-pth. + if (FL->isEmpty()) + InvalidPTH(Diags, level, "PTH file contains no cached source data"); + + // Get the location of the table mapping from persistent ids to the + // data needed to reconstruct identifiers. + const unsigned char* IDTableOffset = PrologueOffset + sizeof(uint32_t)*0; + const unsigned char* IData = BufBeg + ReadLE32(IDTableOffset); + + if (!(IData >= BufBeg && IData < BufEnd)) { + InvalidPTH(Diags, level); + return 0; + } + + // Get the location of the hashtable mapping between strings and + // persistent IDs. + const unsigned char* StringIdTableOffset = PrologueOffset + sizeof(uint32_t)*1; + const unsigned char* StringIdTable = BufBeg + ReadLE32(StringIdTableOffset); + if (!(StringIdTable >= BufBeg && StringIdTable < BufEnd)) { + InvalidPTH(Diags, level); + return 0; + } + + llvm::OwningPtr<PTHStringIdLookup> SL(PTHStringIdLookup::Create(StringIdTable, + BufBeg)); + + // Get the location of the spelling cache. + const unsigned char* spellingBaseOffset = PrologueOffset + sizeof(uint32_t)*3; + const unsigned char* spellingBase = BufBeg + ReadLE32(spellingBaseOffset); + if (!(spellingBase >= BufBeg && spellingBase < BufEnd)) { + InvalidPTH(Diags, level); + return 0; + } + + // Get the number of IdentifierInfos and pre-allocate the identifier cache. + uint32_t NumIds = ReadLE32(IData); + + // Pre-allocate the peristent ID -> IdentifierInfo* cache. We use calloc() + // so that we in the best case only zero out memory once when the OS returns + // us new pages. + IdentifierInfo** PerIDCache = 0; + + if (NumIds) { + PerIDCache = (IdentifierInfo**)calloc(NumIds, sizeof(*PerIDCache)); + if (!PerIDCache) { + InvalidPTH(Diags, level, + "Could not allocate memory for processing PTH file"); + return 0; + } + } + + // Compute the address of the original source file. + const unsigned char* originalSourceBase = PrologueOffset + sizeof(uint32_t)*4; + unsigned len = ReadUnalignedLE16(originalSourceBase); + if (!len) originalSourceBase = 0; + + // Create the new PTHManager. + return new PTHManager(File.take(), FL.take(), IData, PerIDCache, + SL.take(), NumIds, spellingBase, + (const char*) originalSourceBase); +} + +IdentifierInfo* PTHManager::LazilyCreateIdentifierInfo(unsigned PersistentID) { + // Look in the PTH file for the string data for the IdentifierInfo object. + const unsigned char* TableEntry = IdDataTable + sizeof(uint32_t)*PersistentID; + const unsigned char* IDData = + (const unsigned char*)Buf->getBufferStart() + ReadLE32(TableEntry); + assert(IDData < (const unsigned char*)Buf->getBufferEnd()); + + // Allocate the object. + std::pair<IdentifierInfo,const unsigned char*> *Mem = + Alloc.Allocate<std::pair<IdentifierInfo,const unsigned char*> >(); + + Mem->second = IDData; + assert(IDData[0] != '\0'); + IdentifierInfo *II = new ((void*) Mem) IdentifierInfo(); + + // Store the new IdentifierInfo in the cache. + PerIDCache[PersistentID] = II; + assert(II->getName() && II->getName()[0] != '\0'); + return II; +} + +IdentifierInfo* PTHManager::get(const char *NameStart, const char *NameEnd) { + PTHStringIdLookup& SL = *((PTHStringIdLookup*)StringIdLookup); + // Double check our assumption that the last character isn't '\0'. + assert(NameEnd==NameStart || NameStart[NameEnd-NameStart-1] != '\0'); + PTHStringIdLookup::iterator I = SL.find(std::make_pair(NameStart, + NameEnd - NameStart)); + if (I == SL.end()) // No identifier found? + return 0; + + // Match found. Return the identifier! + assert(*I > 0); + return GetIdentifierInfo(*I-1); +} + +PTHLexer *PTHManager::CreateLexer(FileID FID) { + const FileEntry *FE = PP->getSourceManager().getFileEntryForID(FID); + if (!FE) + return 0; + + // Lookup the FileEntry object in our file lookup data structure. It will + // return a variant that indicates whether or not there is an offset within + // the PTH file that contains cached tokens. + PTHFileLookup& PFL = *((PTHFileLookup*)FileLookup); + PTHFileLookup::iterator I = PFL.find(FE); + + if (I == PFL.end()) // No tokens available? + return 0; + + const PTHFileData& FileData = *I; + + const unsigned char *BufStart = (const unsigned char *)Buf->getBufferStart(); + // Compute the offset of the token data within the buffer. + const unsigned char* data = BufStart + FileData.getTokenOffset(); + + // Get the location of pp-conditional table. + const unsigned char* ppcond = BufStart + FileData.getPPCondOffset(); + uint32_t Len = ReadLE32(ppcond); + if (Len == 0) ppcond = 0; + + assert(PP && "No preprocessor set yet!"); + return new PTHLexer(*PP, FID, data, ppcond, *this); +} + +//===----------------------------------------------------------------------===// +// 'stat' caching. +//===----------------------------------------------------------------------===// + +namespace { +class VISIBILITY_HIDDEN PTHStatData { +public: + const bool hasStat; + const ino_t ino; + const dev_t dev; + const mode_t mode; + const time_t mtime; + const off_t size; + + PTHStatData(ino_t i, dev_t d, mode_t mo, time_t m, off_t s) + : hasStat(true), ino(i), dev(d), mode(mo), mtime(m), size(s) {} + + PTHStatData() + : hasStat(false), ino(0), dev(0), mode(0), mtime(0), size(0) {} +}; + +class VISIBILITY_HIDDEN PTHStatLookupTrait : public PTHFileLookupCommonTrait { +public: + typedef const char* external_key_type; // const char* + typedef PTHStatData data_type; + + static internal_key_type GetInternalKey(const char *path) { + // The key 'kind' doesn't matter here because it is ignored in EqualKey. + return std::make_pair((unsigned char) 0x0, path); + } + + static bool EqualKey(internal_key_type a, internal_key_type b) { + // When doing 'stat' lookups we don't care about the kind of 'a' and 'b', + // just the paths. + return strcmp(a.second, b.second) == 0; + } + + static data_type ReadData(const internal_key_type& k, const unsigned char* d, + unsigned) { + + if (k.first /* File or Directory */) { + if (k.first == 0x1 /* File */) d += 4 * 2; // Skip the first 2 words. + ino_t ino = (ino_t) ReadUnalignedLE32(d); + dev_t dev = (dev_t) ReadUnalignedLE32(d); + mode_t mode = (mode_t) ReadUnalignedLE16(d); + time_t mtime = (time_t) ReadUnalignedLE64(d); + return data_type(ino, dev, mode, mtime, (off_t) ReadUnalignedLE64(d)); + } + + // Negative stat. Don't read anything. + return data_type(); + } +}; + +class VISIBILITY_HIDDEN PTHStatCache : public StatSysCallCache { + typedef OnDiskChainedHashTable<PTHStatLookupTrait> CacheTy; + CacheTy Cache; + +public: + PTHStatCache(PTHFileLookup &FL) : + Cache(FL.getNumBuckets(), FL.getNumEntries(), FL.getBuckets(), + FL.getBase()) {} + + ~PTHStatCache() {} + + int stat(const char *path, struct stat *buf) { + // Do the lookup for the file's data in the PTH file. + CacheTy::iterator I = Cache.find(path); + + // If we don't get a hit in the PTH file just forward to 'stat'. + if (I == Cache.end()) return ::stat(path, buf); + + const PTHStatData& Data = *I; + + if (!Data.hasStat) + return 1; + + buf->st_ino = Data.ino; + buf->st_dev = Data.dev; + buf->st_mtime = Data.mtime; + buf->st_mode = Data.mode; + buf->st_size = Data.size; + return 0; + } +}; +} // end anonymous namespace + +StatSysCallCache *PTHManager::createStatCache() { + return new PTHStatCache(*((PTHFileLookup*) FileLookup)); +} diff --git a/lib/Lex/Pragma.cpp b/lib/Lex/Pragma.cpp new file mode 100644 index 0000000..ce59341 --- /dev/null +++ b/lib/Lex/Pragma.cpp @@ -0,0 +1,699 @@ +//===--- Pragma.cpp - Pragma registration and handling --------------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements the PragmaHandler/PragmaTable interfaces and implements +// pragma related methods of the Preprocessor class. +// +//===----------------------------------------------------------------------===// + +#include "clang/Lex/Pragma.h" +#include "clang/Lex/HeaderSearch.h" +#include "clang/Lex/LiteralSupport.h" +#include "clang/Lex/Preprocessor.h" +#include "clang/Lex/LexDiagnostic.h" +#include "clang/Basic/FileManager.h" +#include "clang/Basic/SourceManager.h" +using namespace clang; + +// Out-of-line destructor to provide a home for the class. +PragmaHandler::~PragmaHandler() { +} + +//===----------------------------------------------------------------------===// +// PragmaNamespace Implementation. +//===----------------------------------------------------------------------===// + + +PragmaNamespace::~PragmaNamespace() { + for (unsigned i = 0, e = Handlers.size(); i != e; ++i) + delete Handlers[i]; +} + +/// FindHandler - Check to see if there is already a handler for the +/// specified name. If not, return the handler for the null identifier if it +/// exists, otherwise return null. If IgnoreNull is true (the default) then +/// the null handler isn't returned on failure to match. +PragmaHandler *PragmaNamespace::FindHandler(const IdentifierInfo *Name, + bool IgnoreNull) const { + PragmaHandler *NullHandler = 0; + for (unsigned i = 0, e = Handlers.size(); i != e; ++i) { + if (Handlers[i]->getName() == Name) + return Handlers[i]; + + if (Handlers[i]->getName() == 0) + NullHandler = Handlers[i]; + } + return IgnoreNull ? 0 : NullHandler; +} + +void PragmaNamespace::RemovePragmaHandler(PragmaHandler *Handler) { + for (unsigned i = 0, e = Handlers.size(); i != e; ++i) { + if (Handlers[i] == Handler) { + Handlers[i] = Handlers.back(); + Handlers.pop_back(); + return; + } + } + assert(0 && "Handler not registered in this namespace"); +} + +void PragmaNamespace::HandlePragma(Preprocessor &PP, Token &Tok) { + // Read the 'namespace' that the directive is in, e.g. STDC. Do not macro + // expand it, the user can have a STDC #define, that should not affect this. + PP.LexUnexpandedToken(Tok); + + // Get the handler for this token. If there is no handler, ignore the pragma. + PragmaHandler *Handler = FindHandler(Tok.getIdentifierInfo(), false); + if (Handler == 0) { + PP.Diag(Tok, diag::warn_pragma_ignored); + return; + } + + // Otherwise, pass it down. + Handler->HandlePragma(PP, Tok); +} + +//===----------------------------------------------------------------------===// +// Preprocessor Pragma Directive Handling. +//===----------------------------------------------------------------------===// + +/// HandlePragmaDirective - The "#pragma" directive has been parsed. Lex the +/// rest of the pragma, passing it to the registered pragma handlers. +void Preprocessor::HandlePragmaDirective() { + ++NumPragma; + + // Invoke the first level of pragma handlers which reads the namespace id. + Token Tok; + PragmaHandlers->HandlePragma(*this, Tok); + + // If the pragma handler didn't read the rest of the line, consume it now. + if (CurPPLexer->ParsingPreprocessorDirective) + DiscardUntilEndOfDirective(); +} + +/// Handle_Pragma - Read a _Pragma directive, slice it up, process it, then +/// return the first token after the directive. The _Pragma token has just +/// been read into 'Tok'. +void Preprocessor::Handle_Pragma(Token &Tok) { + // Remember the pragma token location. + SourceLocation PragmaLoc = Tok.getLocation(); + + // Read the '('. + Lex(Tok); + if (Tok.isNot(tok::l_paren)) { + Diag(PragmaLoc, diag::err__Pragma_malformed); + return; + } + + // Read the '"..."'. + Lex(Tok); + if (Tok.isNot(tok::string_literal) && Tok.isNot(tok::wide_string_literal)) { + Diag(PragmaLoc, diag::err__Pragma_malformed); + return; + } + + // Remember the string. + std::string StrVal = getSpelling(Tok); + + // Read the ')'. + Lex(Tok); + if (Tok.isNot(tok::r_paren)) { + Diag(PragmaLoc, diag::err__Pragma_malformed); + return; + } + + SourceLocation RParenLoc = Tok.getLocation(); + + // The _Pragma is lexically sound. Destringize according to C99 6.10.9.1: + // "The string literal is destringized by deleting the L prefix, if present, + // deleting the leading and trailing double-quotes, replacing each escape + // sequence \" by a double-quote, and replacing each escape sequence \\ by a + // single backslash." + if (StrVal[0] == 'L') // Remove L prefix. + StrVal.erase(StrVal.begin()); + assert(StrVal[0] == '"' && StrVal[StrVal.size()-1] == '"' && + "Invalid string token!"); + + // Remove the front quote, replacing it with a space, so that the pragma + // contents appear to have a space before them. + StrVal[0] = ' '; + + // Replace the terminating quote with a \n. + StrVal[StrVal.size()-1] = '\n'; + + // Remove escaped quotes and escapes. + for (unsigned i = 0, e = StrVal.size(); i != e-1; ++i) { + if (StrVal[i] == '\\' && + (StrVal[i+1] == '\\' || StrVal[i+1] == '"')) { + // \\ -> '\' and \" -> '"'. + StrVal.erase(StrVal.begin()+i); + --e; + } + } + + // Plop the string (including the newline and trailing null) into a buffer + // where we can lex it. + Token TmpTok; + TmpTok.startToken(); + CreateString(&StrVal[0], StrVal.size(), TmpTok); + SourceLocation TokLoc = TmpTok.getLocation(); + + // Make and enter a lexer object so that we lex and expand the tokens just + // like any others. + Lexer *TL = Lexer::Create_PragmaLexer(TokLoc, PragmaLoc, RParenLoc, + StrVal.size(), *this); + + EnterSourceFileWithLexer(TL, 0); + + // With everything set up, lex this as a #pragma directive. + HandlePragmaDirective(); + + // Finally, return whatever came after the pragma directive. + return Lex(Tok); +} + + + +/// HandlePragmaOnce - Handle #pragma once. OnceTok is the 'once'. +/// +void Preprocessor::HandlePragmaOnce(Token &OnceTok) { + if (isInPrimaryFile()) { + Diag(OnceTok, diag::pp_pragma_once_in_main_file); + return; + } + + // Get the current file lexer we're looking at. Ignore _Pragma 'files' etc. + // Mark the file as a once-only file now. + HeaderInfo.MarkFileIncludeOnce(getCurrentFileLexer()->getFileEntry()); +} + +void Preprocessor::HandlePragmaMark() { + assert(CurPPLexer && "No current lexer?"); + if (CurLexer) CurLexer->ReadToEndOfLine(); + else CurPTHLexer->DiscardToEndOfLine(); +} + + +/// HandlePragmaPoison - Handle #pragma GCC poison. PoisonTok is the 'poison'. +/// +void Preprocessor::HandlePragmaPoison(Token &PoisonTok) { + Token Tok; + + while (1) { + // Read the next token to poison. While doing this, pretend that we are + // skipping while reading the identifier to poison. + // This avoids errors on code like: + // #pragma GCC poison X + // #pragma GCC poison X + if (CurPPLexer) CurPPLexer->LexingRawMode = true; + LexUnexpandedToken(Tok); + if (CurPPLexer) CurPPLexer->LexingRawMode = false; + + // If we reached the end of line, we're done. + if (Tok.is(tok::eom)) return; + + // Can only poison identifiers. + if (Tok.isNot(tok::identifier)) { + Diag(Tok, diag::err_pp_invalid_poison); + return; + } + + // Look up the identifier info for the token. We disabled identifier lookup + // by saying we're skipping contents, so we need to do this manually. + IdentifierInfo *II = LookUpIdentifierInfo(Tok); + + // Already poisoned. + if (II->isPoisoned()) continue; + + // If this is a macro identifier, emit a warning. + if (II->hasMacroDefinition()) + Diag(Tok, diag::pp_poisoning_existing_macro); + + // Finally, poison it! + II->setIsPoisoned(); + } +} + +/// HandlePragmaSystemHeader - Implement #pragma GCC system_header. We know +/// that the whole directive has been parsed. +void Preprocessor::HandlePragmaSystemHeader(Token &SysHeaderTok) { + if (isInPrimaryFile()) { + Diag(SysHeaderTok, diag::pp_pragma_sysheader_in_main_file); + return; + } + + // Get the current file lexer we're looking at. Ignore _Pragma 'files' etc. + PreprocessorLexer *TheLexer = getCurrentFileLexer(); + + // Mark the file as a system header. + HeaderInfo.MarkFileSystemHeader(TheLexer->getFileEntry()); + + // Notify the client, if desired, that we are in a new source file. + if (Callbacks) + Callbacks->FileChanged(SysHeaderTok.getLocation(), + PPCallbacks::SystemHeaderPragma, SrcMgr::C_System); +} + +/// HandlePragmaDependency - Handle #pragma GCC dependency "foo" blah. +/// +void Preprocessor::HandlePragmaDependency(Token &DependencyTok) { + Token FilenameTok; + CurPPLexer->LexIncludeFilename(FilenameTok); + + // If the token kind is EOM, the error has already been diagnosed. + if (FilenameTok.is(tok::eom)) + return; + + // Reserve a buffer to get the spelling. + llvm::SmallVector<char, 128> FilenameBuffer; + FilenameBuffer.resize(FilenameTok.getLength()); + + const char *FilenameStart = &FilenameBuffer[0]; + unsigned Len = getSpelling(FilenameTok, FilenameStart); + const char *FilenameEnd = FilenameStart+Len; + bool isAngled = GetIncludeFilenameSpelling(FilenameTok.getLocation(), + FilenameStart, FilenameEnd); + // If GetIncludeFilenameSpelling set the start ptr to null, there was an + // error. + if (FilenameStart == 0) + return; + + // Search include directories for this file. + const DirectoryLookup *CurDir; + const FileEntry *File = LookupFile(FilenameStart, FilenameEnd, + isAngled, 0, CurDir); + if (File == 0) { + Diag(FilenameTok, diag::err_pp_file_not_found) + << std::string(FilenameStart, FilenameEnd); + return; + } + + const FileEntry *CurFile = getCurrentFileLexer()->getFileEntry(); + + // If this file is older than the file it depends on, emit a diagnostic. + if (CurFile && CurFile->getModificationTime() < File->getModificationTime()) { + // Lex tokens at the end of the message and include them in the message. + std::string Message; + Lex(DependencyTok); + while (DependencyTok.isNot(tok::eom)) { + Message += getSpelling(DependencyTok) + " "; + Lex(DependencyTok); + } + + Message.erase(Message.end()-1); + Diag(FilenameTok, diag::pp_out_of_date_dependency) << Message; + } +} + +/// HandlePragmaComment - Handle the microsoft #pragma comment extension. The +/// syntax is: +/// #pragma comment(linker, "foo") +/// 'linker' is one of five identifiers: compiler, exestr, lib, linker, user. +/// "foo" is a string, which is fully macro expanded, and permits string +/// concatenation, embedded escape characters etc. See MSDN for more details. +void Preprocessor::HandlePragmaComment(Token &Tok) { + SourceLocation CommentLoc = Tok.getLocation(); + Lex(Tok); + if (Tok.isNot(tok::l_paren)) { + Diag(CommentLoc, diag::err_pragma_comment_malformed); + return; + } + + // Read the identifier. + Lex(Tok); + if (Tok.isNot(tok::identifier)) { + Diag(CommentLoc, diag::err_pragma_comment_malformed); + return; + } + + // Verify that this is one of the 5 whitelisted options. + // FIXME: warn that 'exestr' is deprecated. + const IdentifierInfo *II = Tok.getIdentifierInfo(); + if (!II->isStr("compiler") && !II->isStr("exestr") && !II->isStr("lib") && + !II->isStr("linker") && !II->isStr("user")) { + Diag(Tok.getLocation(), diag::err_pragma_comment_unknown_kind); + return; + } + + // Read the optional string if present. + Lex(Tok); + std::string ArgumentString; + if (Tok.is(tok::comma)) { + Lex(Tok); // eat the comma. + + // We need at least one string. + if (Tok.isNot(tok::string_literal)) { + Diag(Tok.getLocation(), diag::err_pragma_comment_malformed); + return; + } + + // String concatenation allows multiple strings, which can even come from + // macro expansion. + // "foo " "bar" "Baz" + llvm::SmallVector<Token, 4> StrToks; + while (Tok.is(tok::string_literal)) { + StrToks.push_back(Tok); + Lex(Tok); + } + + // Concatenate and parse the strings. + StringLiteralParser Literal(&StrToks[0], StrToks.size(), *this); + assert(!Literal.AnyWide && "Didn't allow wide strings in"); + if (Literal.hadError) + return; + if (Literal.Pascal) { + Diag(StrToks[0].getLocation(), diag::err_pragma_comment_malformed); + return; + } + + ArgumentString = std::string(Literal.GetString(), + Literal.GetString()+Literal.GetStringLength()); + } + + // FIXME: If the kind is "compiler" warn if the string is present (it is + // ignored). + // FIXME: 'lib' requires a comment string. + // FIXME: 'linker' requires a comment string, and has a specific list of + // things that are allowable. + + if (Tok.isNot(tok::r_paren)) { + Diag(Tok.getLocation(), diag::err_pragma_comment_malformed); + return; + } + Lex(Tok); // eat the r_paren. + + if (Tok.isNot(tok::eom)) { + Diag(Tok.getLocation(), diag::err_pragma_comment_malformed); + return; + } + + // If the pragma is lexically sound, notify any interested PPCallbacks. + if (Callbacks) + Callbacks->PragmaComment(CommentLoc, II, ArgumentString); +} + + + + +/// AddPragmaHandler - Add the specified pragma handler to the preprocessor. +/// If 'Namespace' is non-null, then it is a token required to exist on the +/// pragma line before the pragma string starts, e.g. "STDC" or "GCC". +void Preprocessor::AddPragmaHandler(const char *Namespace, + PragmaHandler *Handler) { + PragmaNamespace *InsertNS = PragmaHandlers; + + // If this is specified to be in a namespace, step down into it. + if (Namespace) { + IdentifierInfo *NSID = getIdentifierInfo(Namespace); + + // If there is already a pragma handler with the name of this namespace, + // we either have an error (directive with the same name as a namespace) or + // we already have the namespace to insert into. + if (PragmaHandler *Existing = PragmaHandlers->FindHandler(NSID)) { + InsertNS = Existing->getIfNamespace(); + assert(InsertNS != 0 && "Cannot have a pragma namespace and pragma" + " handler with the same name!"); + } else { + // Otherwise, this namespace doesn't exist yet, create and insert the + // handler for it. + InsertNS = new PragmaNamespace(NSID); + PragmaHandlers->AddPragma(InsertNS); + } + } + + // Check to make sure we don't already have a pragma for this identifier. + assert(!InsertNS->FindHandler(Handler->getName()) && + "Pragma handler already exists for this identifier!"); + InsertNS->AddPragma(Handler); +} + +/// RemovePragmaHandler - Remove the specific pragma handler from the +/// preprocessor. If \arg Namespace is non-null, then it should be the +/// namespace that \arg Handler was added to. It is an error to remove +/// a handler that has not been registered. +void Preprocessor::RemovePragmaHandler(const char *Namespace, + PragmaHandler *Handler) { + PragmaNamespace *NS = PragmaHandlers; + + // If this is specified to be in a namespace, step down into it. + if (Namespace) { + IdentifierInfo *NSID = getIdentifierInfo(Namespace); + PragmaHandler *Existing = PragmaHandlers->FindHandler(NSID); + assert(Existing && "Namespace containing handler does not exist!"); + + NS = Existing->getIfNamespace(); + assert(NS && "Invalid namespace, registered as a regular pragma handler!"); + } + + NS->RemovePragmaHandler(Handler); + + // If this is a non-default namespace and it is now empty, remove + // it. + if (NS != PragmaHandlers && NS->IsEmpty()) + PragmaHandlers->RemovePragmaHandler(NS); +} + +namespace { +/// PragmaOnceHandler - "#pragma once" marks the file as atomically included. +struct PragmaOnceHandler : public PragmaHandler { + PragmaOnceHandler(const IdentifierInfo *OnceID) : PragmaHandler(OnceID) {} + virtual void HandlePragma(Preprocessor &PP, Token &OnceTok) { + PP.CheckEndOfDirective("pragma once"); + PP.HandlePragmaOnce(OnceTok); + } +}; + +/// PragmaMarkHandler - "#pragma mark ..." is ignored by the compiler, and the +/// rest of the line is not lexed. +struct PragmaMarkHandler : public PragmaHandler { + PragmaMarkHandler(const IdentifierInfo *MarkID) : PragmaHandler(MarkID) {} + virtual void HandlePragma(Preprocessor &PP, Token &MarkTok) { + PP.HandlePragmaMark(); + } +}; + +/// PragmaPoisonHandler - "#pragma poison x" marks x as not usable. +struct PragmaPoisonHandler : public PragmaHandler { + PragmaPoisonHandler(const IdentifierInfo *ID) : PragmaHandler(ID) {} + virtual void HandlePragma(Preprocessor &PP, Token &PoisonTok) { + PP.HandlePragmaPoison(PoisonTok); + } +}; + +/// PragmaSystemHeaderHandler - "#pragma system_header" marks the current file +/// as a system header, which silences warnings in it. +struct PragmaSystemHeaderHandler : public PragmaHandler { + PragmaSystemHeaderHandler(const IdentifierInfo *ID) : PragmaHandler(ID) {} + virtual void HandlePragma(Preprocessor &PP, Token &SHToken) { + PP.HandlePragmaSystemHeader(SHToken); + PP.CheckEndOfDirective("pragma"); + } +}; +struct PragmaDependencyHandler : public PragmaHandler { + PragmaDependencyHandler(const IdentifierInfo *ID) : PragmaHandler(ID) {} + virtual void HandlePragma(Preprocessor &PP, Token &DepToken) { + PP.HandlePragmaDependency(DepToken); + } +}; + +/// PragmaDiagnosticHandler - e.g. '#pragma GCC diagnostic ignored "-Wformat"' +struct PragmaDiagnosticHandler : public PragmaHandler { + PragmaDiagnosticHandler(const IdentifierInfo *ID) : PragmaHandler(ID) {} + virtual void HandlePragma(Preprocessor &PP, Token &DiagToken) { + Token Tok; + PP.LexUnexpandedToken(Tok); + if (Tok.isNot(tok::identifier)) { + PP.Diag(Tok, diag::warn_pragma_diagnostic_invalid); + return; + } + IdentifierInfo *II = Tok.getIdentifierInfo(); + + diag::Mapping Map; + if (II->isStr("warning")) + Map = diag::MAP_WARNING; + else if (II->isStr("error")) + Map = diag::MAP_ERROR; + else if (II->isStr("ignored")) + Map = diag::MAP_IGNORE; + else if (II->isStr("fatal")) + Map = diag::MAP_FATAL; + else { + PP.Diag(Tok, diag::warn_pragma_diagnostic_invalid); + return; + } + + PP.LexUnexpandedToken(Tok); + + // We need at least one string. + if (Tok.isNot(tok::string_literal)) { + PP.Diag(Tok.getLocation(), diag::warn_pragma_diagnostic_invalid_token); + return; + } + + // String concatenation allows multiple strings, which can even come from + // macro expansion. + // "foo " "bar" "Baz" + llvm::SmallVector<Token, 4> StrToks; + while (Tok.is(tok::string_literal)) { + StrToks.push_back(Tok); + PP.LexUnexpandedToken(Tok); + } + + if (Tok.isNot(tok::eom)) { + PP.Diag(Tok.getLocation(), diag::warn_pragma_diagnostic_invalid_token); + return; + } + + // Concatenate and parse the strings. + StringLiteralParser Literal(&StrToks[0], StrToks.size(), PP); + assert(!Literal.AnyWide && "Didn't allow wide strings in"); + if (Literal.hadError) + return; + if (Literal.Pascal) { + PP.Diag(StrToks[0].getLocation(), diag::warn_pragma_diagnostic_invalid); + return; + } + + std::string WarningName(Literal.GetString(), + Literal.GetString()+Literal.GetStringLength()); + + if (WarningName.size() < 3 || WarningName[0] != '-' || + WarningName[1] != 'W') { + PP.Diag(StrToks[0].getLocation(), + diag::warn_pragma_diagnostic_invalid_option); + return; + } + + if (PP.getDiagnostics().setDiagnosticGroupMapping(WarningName.c_str()+2, + Map)) + PP.Diag(StrToks[0].getLocation(), + diag::warn_pragma_diagnostic_unknown_warning) << WarningName; + } +}; + +/// PragmaCommentHandler - "#pragma comment ...". +struct PragmaCommentHandler : public PragmaHandler { + PragmaCommentHandler(const IdentifierInfo *ID) : PragmaHandler(ID) {} + virtual void HandlePragma(Preprocessor &PP, Token &CommentTok) { + PP.HandlePragmaComment(CommentTok); + } +}; + +// Pragma STDC implementations. + +enum STDCSetting { + STDC_ON, STDC_OFF, STDC_DEFAULT, STDC_INVALID +}; + +static STDCSetting LexOnOffSwitch(Preprocessor &PP) { + Token Tok; + PP.LexUnexpandedToken(Tok); + + if (Tok.isNot(tok::identifier)) { + PP.Diag(Tok, diag::ext_stdc_pragma_syntax); + return STDC_INVALID; + } + IdentifierInfo *II = Tok.getIdentifierInfo(); + STDCSetting Result; + if (II->isStr("ON")) + Result = STDC_ON; + else if (II->isStr("OFF")) + Result = STDC_OFF; + else if (II->isStr("DEFAULT")) + Result = STDC_DEFAULT; + else { + PP.Diag(Tok, diag::ext_stdc_pragma_syntax); + return STDC_INVALID; + } + + // Verify that this is followed by EOM. + PP.LexUnexpandedToken(Tok); + if (Tok.isNot(tok::eom)) + PP.Diag(Tok, diag::ext_stdc_pragma_syntax_eom); + return Result; +} + +/// PragmaSTDC_FP_CONTRACTHandler - "#pragma STDC FP_CONTRACT ...". +struct PragmaSTDC_FP_CONTRACTHandler : public PragmaHandler { + PragmaSTDC_FP_CONTRACTHandler(const IdentifierInfo *ID) : PragmaHandler(ID) {} + virtual void HandlePragma(Preprocessor &PP, Token &Tok) { + // We just ignore the setting of FP_CONTRACT. Since we don't do contractions + // at all, our default is OFF and setting it to ON is an optimization hint + // we can safely ignore. When we support -ffma or something, we would need + // to diagnose that we are ignoring FMA. + LexOnOffSwitch(PP); + } +}; + +/// PragmaSTDC_FENV_ACCESSHandler - "#pragma STDC FENV_ACCESS ...". +struct PragmaSTDC_FENV_ACCESSHandler : public PragmaHandler { + PragmaSTDC_FENV_ACCESSHandler(const IdentifierInfo *ID) : PragmaHandler(ID) {} + virtual void HandlePragma(Preprocessor &PP, Token &Tok) { + if (LexOnOffSwitch(PP) == STDC_ON) + PP.Diag(Tok, diag::warn_stdc_fenv_access_not_supported); + } +}; + +/// PragmaSTDC_CX_LIMITED_RANGEHandler - "#pragma STDC CX_LIMITED_RANGE ...". +struct PragmaSTDC_CX_LIMITED_RANGEHandler : public PragmaHandler { + PragmaSTDC_CX_LIMITED_RANGEHandler(const IdentifierInfo *ID) + : PragmaHandler(ID) {} + virtual void HandlePragma(Preprocessor &PP, Token &Tok) { + LexOnOffSwitch(PP); + } +}; + +/// PragmaSTDC_UnknownHandler - "#pragma STDC ...". +struct PragmaSTDC_UnknownHandler : public PragmaHandler { + PragmaSTDC_UnknownHandler() : PragmaHandler(0) {} + virtual void HandlePragma(Preprocessor &PP, Token &UnknownTok) { + // C99 6.10.6p2, unknown forms are not allowed. + PP.Diag(UnknownTok, diag::ext_stdc_pragma_ignored); + } +}; + +} // end anonymous namespace + + +/// RegisterBuiltinPragmas - Install the standard preprocessor pragmas: +/// #pragma GCC poison/system_header/dependency and #pragma once. +void Preprocessor::RegisterBuiltinPragmas() { + AddPragmaHandler(0, new PragmaOnceHandler(getIdentifierInfo("once"))); + AddPragmaHandler(0, new PragmaMarkHandler(getIdentifierInfo("mark"))); + + // #pragma GCC ... + AddPragmaHandler("GCC", new PragmaPoisonHandler(getIdentifierInfo("poison"))); + AddPragmaHandler("GCC", new PragmaSystemHeaderHandler( + getIdentifierInfo("system_header"))); + AddPragmaHandler("GCC", new PragmaDependencyHandler( + getIdentifierInfo("dependency"))); + AddPragmaHandler("GCC", new PragmaDiagnosticHandler( + getIdentifierInfo("diagnostic"))); + // #pragma clang ... + AddPragmaHandler("clang", new PragmaPoisonHandler( + getIdentifierInfo("poison"))); + AddPragmaHandler("clang", new PragmaSystemHeaderHandler( + getIdentifierInfo("system_header"))); + AddPragmaHandler("clang", new PragmaDependencyHandler( + getIdentifierInfo("dependency"))); + AddPragmaHandler("clang", new PragmaDiagnosticHandler( + getIdentifierInfo("diagnostic"))); + + AddPragmaHandler("STDC", new PragmaSTDC_FP_CONTRACTHandler( + getIdentifierInfo("FP_CONTRACT"))); + AddPragmaHandler("STDC", new PragmaSTDC_FENV_ACCESSHandler( + getIdentifierInfo("FENV_ACCESS"))); + AddPragmaHandler("STDC", new PragmaSTDC_CX_LIMITED_RANGEHandler( + getIdentifierInfo("CX_LIMITED_RANGE"))); + AddPragmaHandler("STDC", new PragmaSTDC_UnknownHandler()); + + // MS extensions. + if (Features.Microsoft) + AddPragmaHandler(0, new PragmaCommentHandler(getIdentifierInfo("comment"))); +} diff --git a/lib/Lex/Preprocessor.cpp b/lib/Lex/Preprocessor.cpp new file mode 100644 index 0000000..0a7d92e --- /dev/null +++ b/lib/Lex/Preprocessor.cpp @@ -0,0 +1,478 @@ +//===--- Preprocess.cpp - C Language Family Preprocessor Implementation ---===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements the Preprocessor interface. +// +//===----------------------------------------------------------------------===// +// +// Options to support: +// -H - Print the name of each header file used. +// -d[DNI] - Dump various things. +// -fworking-directory - #line's with preprocessor's working dir. +// -fpreprocessed +// -dependency-file,-M,-MM,-MF,-MG,-MP,-MT,-MQ,-MD,-MMD +// -W* +// -w +// +// Messages to emit: +// "Multiple include guards may be useful for:\n" +// +//===----------------------------------------------------------------------===// + +#include "clang/Lex/Preprocessor.h" +#include "clang/Lex/HeaderSearch.h" +#include "clang/Lex/MacroInfo.h" +#include "clang/Lex/Pragma.h" +#include "clang/Lex/ScratchBuffer.h" +#include "clang/Lex/LexDiagnostic.h" +#include "clang/Basic/SourceManager.h" +#include "clang/Basic/FileManager.h" +#include "clang/Basic/TargetInfo.h" +#include "llvm/ADT/APFloat.h" +#include "llvm/ADT/SmallVector.h" +#include "llvm/Support/MemoryBuffer.h" +#include "llvm/Support/Streams.h" +#include <cstdio> +using namespace clang; + +//===----------------------------------------------------------------------===// + +PreprocessorFactory::~PreprocessorFactory() {} + +Preprocessor::Preprocessor(Diagnostic &diags, const LangOptions &opts, + TargetInfo &target, SourceManager &SM, + HeaderSearch &Headers, + IdentifierInfoLookup* IILookup) + : Diags(&diags), Features(opts), Target(target),FileMgr(Headers.getFileMgr()), + SourceMgr(SM), HeaderInfo(Headers), Identifiers(opts, IILookup), + CurPPLexer(0), CurDirLookup(0), Callbacks(0) { + ScratchBuf = new ScratchBuffer(SourceMgr); + CounterValue = 0; // __COUNTER__ starts at 0. + + // Clear stats. + NumDirectives = NumDefined = NumUndefined = NumPragma = 0; + NumIf = NumElse = NumEndif = 0; + NumEnteredSourceFiles = 0; + NumMacroExpanded = NumFnMacroExpanded = NumBuiltinMacroExpanded = 0; + NumFastMacroExpanded = NumTokenPaste = NumFastTokenPaste = 0; + MaxIncludeStackDepth = 0; + NumSkipped = 0; + + // Default to discarding comments. + KeepComments = false; + KeepMacroComments = false; + + // Macro expansion is enabled. + DisableMacroExpansion = false; + InMacroArgs = false; + NumCachedTokenLexers = 0; + + CachedLexPos = 0; + + // "Poison" __VA_ARGS__, which can only appear in the expansion of a macro. + // This gets unpoisoned where it is allowed. + (Ident__VA_ARGS__ = getIdentifierInfo("__VA_ARGS__"))->setIsPoisoned(); + + // Initialize the pragma handlers. + PragmaHandlers = new PragmaNamespace(0); + RegisterBuiltinPragmas(); + + // Initialize builtin macros like __LINE__ and friends. + RegisterBuiltinMacros(); +} + +Preprocessor::~Preprocessor() { + assert(BacktrackPositions.empty() && "EnableBacktrack/Backtrack imbalance!"); + + while (!IncludeMacroStack.empty()) { + delete IncludeMacroStack.back().TheLexer; + delete IncludeMacroStack.back().TheTokenLexer; + IncludeMacroStack.pop_back(); + } + + // Free any macro definitions. + for (llvm::DenseMap<IdentifierInfo*, MacroInfo*>::iterator I = + Macros.begin(), E = Macros.end(); I != E; ++I) { + // We don't need to free the MacroInfo objects directly. These + // will be released when the BumpPtrAllocator 'BP' object gets + // destroyed. We still need to run the dstor, however, to free + // memory alocated by MacroInfo. + I->second->Destroy(BP); + I->first->setHasMacroDefinition(false); + } + + // Free any cached macro expanders. + for (unsigned i = 0, e = NumCachedTokenLexers; i != e; ++i) + delete TokenLexerCache[i]; + + // Release pragma information. + delete PragmaHandlers; + + // Delete the scratch buffer info. + delete ScratchBuf; + + delete Callbacks; +} + +void Preprocessor::setPTHManager(PTHManager* pm) { + PTH.reset(pm); + FileMgr.setStatCache(PTH->createStatCache()); +} + +void Preprocessor::DumpToken(const Token &Tok, bool DumpFlags) const { + llvm::cerr << tok::getTokenName(Tok.getKind()) << " '" + << getSpelling(Tok) << "'"; + + if (!DumpFlags) return; + + llvm::cerr << "\t"; + if (Tok.isAtStartOfLine()) + llvm::cerr << " [StartOfLine]"; + if (Tok.hasLeadingSpace()) + llvm::cerr << " [LeadingSpace]"; + if (Tok.isExpandDisabled()) + llvm::cerr << " [ExpandDisabled]"; + if (Tok.needsCleaning()) { + const char *Start = SourceMgr.getCharacterData(Tok.getLocation()); + llvm::cerr << " [UnClean='" << std::string(Start, Start+Tok.getLength()) + << "']"; + } + + llvm::cerr << "\tLoc=<"; + DumpLocation(Tok.getLocation()); + llvm::cerr << ">"; +} + +void Preprocessor::DumpLocation(SourceLocation Loc) const { + Loc.dump(SourceMgr); +} + +void Preprocessor::DumpMacro(const MacroInfo &MI) const { + llvm::cerr << "MACRO: "; + for (unsigned i = 0, e = MI.getNumTokens(); i != e; ++i) { + DumpToken(MI.getReplacementToken(i)); + llvm::cerr << " "; + } + llvm::cerr << "\n"; +} + +void Preprocessor::PrintStats() { + llvm::cerr << "\n*** Preprocessor Stats:\n"; + llvm::cerr << NumDirectives << " directives found:\n"; + llvm::cerr << " " << NumDefined << " #define.\n"; + llvm::cerr << " " << NumUndefined << " #undef.\n"; + llvm::cerr << " #include/#include_next/#import:\n"; + llvm::cerr << " " << NumEnteredSourceFiles << " source files entered.\n"; + llvm::cerr << " " << MaxIncludeStackDepth << " max include stack depth\n"; + llvm::cerr << " " << NumIf << " #if/#ifndef/#ifdef.\n"; + llvm::cerr << " " << NumElse << " #else/#elif.\n"; + llvm::cerr << " " << NumEndif << " #endif.\n"; + llvm::cerr << " " << NumPragma << " #pragma.\n"; + llvm::cerr << NumSkipped << " #if/#ifndef#ifdef regions skipped\n"; + + llvm::cerr << NumMacroExpanded << "/" << NumFnMacroExpanded << "/" + << NumBuiltinMacroExpanded << " obj/fn/builtin macros expanded, " + << NumFastMacroExpanded << " on the fast path.\n"; + llvm::cerr << (NumFastTokenPaste+NumTokenPaste) + << " token paste (##) operations performed, " + << NumFastTokenPaste << " on the fast path.\n"; +} + +//===----------------------------------------------------------------------===// +// Token Spelling +//===----------------------------------------------------------------------===// + + +/// getSpelling() - Return the 'spelling' of this token. The spelling of a +/// token are the characters used to represent the token in the source file +/// after trigraph expansion and escaped-newline folding. In particular, this +/// wants to get the true, uncanonicalized, spelling of things like digraphs +/// UCNs, etc. +std::string Preprocessor::getSpelling(const Token &Tok) const { + assert((int)Tok.getLength() >= 0 && "Token character range is bogus!"); + + // If this token contains nothing interesting, return it directly. + const char* TokStart = SourceMgr.getCharacterData(Tok.getLocation()); + if (!Tok.needsCleaning()) + return std::string(TokStart, TokStart+Tok.getLength()); + + std::string Result; + Result.reserve(Tok.getLength()); + + // Otherwise, hard case, relex the characters into the string. + for (const char *Ptr = TokStart, *End = TokStart+Tok.getLength(); + Ptr != End; ) { + unsigned CharSize; + Result.push_back(Lexer::getCharAndSizeNoWarn(Ptr, CharSize, Features)); + Ptr += CharSize; + } + assert(Result.size() != unsigned(Tok.getLength()) && + "NeedsCleaning flag set on something that didn't need cleaning!"); + return Result; +} + +/// getSpelling - This method is used to get the spelling of a token into a +/// preallocated buffer, instead of as an std::string. The caller is required +/// to allocate enough space for the token, which is guaranteed to be at least +/// Tok.getLength() bytes long. The actual length of the token is returned. +/// +/// Note that this method may do two possible things: it may either fill in +/// the buffer specified with characters, or it may *change the input pointer* +/// to point to a constant buffer with the data already in it (avoiding a +/// copy). The caller is not allowed to modify the returned buffer pointer +/// if an internal buffer is returned. +unsigned Preprocessor::getSpelling(const Token &Tok, + const char *&Buffer) const { + assert((int)Tok.getLength() >= 0 && "Token character range is bogus!"); + + // If this token is an identifier, just return the string from the identifier + // table, which is very quick. + if (const IdentifierInfo *II = Tok.getIdentifierInfo()) { + Buffer = II->getName(); + return II->getLength(); + } + + // Otherwise, compute the start of the token in the input lexer buffer. + const char *TokStart = 0; + + if (Tok.isLiteral()) + TokStart = Tok.getLiteralData(); + + if (TokStart == 0) + TokStart = SourceMgr.getCharacterData(Tok.getLocation()); + + // If this token contains nothing interesting, return it directly. + if (!Tok.needsCleaning()) { + Buffer = TokStart; + return Tok.getLength(); + } + + // Otherwise, hard case, relex the characters into the string. + char *OutBuf = const_cast<char*>(Buffer); + for (const char *Ptr = TokStart, *End = TokStart+Tok.getLength(); + Ptr != End; ) { + unsigned CharSize; + *OutBuf++ = Lexer::getCharAndSizeNoWarn(Ptr, CharSize, Features); + Ptr += CharSize; + } + assert(unsigned(OutBuf-Buffer) != Tok.getLength() && + "NeedsCleaning flag set on something that didn't need cleaning!"); + + return OutBuf-Buffer; +} + +/// CreateString - Plop the specified string into a scratch buffer and return a +/// location for it. If specified, the source location provides a source +/// location for the token. +void Preprocessor::CreateString(const char *Buf, unsigned Len, Token &Tok, + SourceLocation InstantiationLoc) { + Tok.setLength(Len); + + const char *DestPtr; + SourceLocation Loc = ScratchBuf->getToken(Buf, Len, DestPtr); + + if (InstantiationLoc.isValid()) + Loc = SourceMgr.createInstantiationLoc(Loc, InstantiationLoc, + InstantiationLoc, Len); + Tok.setLocation(Loc); + + // If this is a literal token, set the pointer data. + if (Tok.isLiteral()) + Tok.setLiteralData(DestPtr); +} + + +/// AdvanceToTokenCharacter - Given a location that specifies the start of a +/// token, return a new location that specifies a character within the token. +SourceLocation Preprocessor::AdvanceToTokenCharacter(SourceLocation TokStart, + unsigned CharNo) { + // Figure out how many physical characters away the specified instantiation + // character is. This needs to take into consideration newlines and + // trigraphs. + const char *TokPtr = SourceMgr.getCharacterData(TokStart); + + // If they request the first char of the token, we're trivially done. + if (CharNo == 0 && Lexer::isObviouslySimpleCharacter(*TokPtr)) + return TokStart; + + unsigned PhysOffset = 0; + + // The usual case is that tokens don't contain anything interesting. Skip + // over the uninteresting characters. If a token only consists of simple + // chars, this method is extremely fast. + while (Lexer::isObviouslySimpleCharacter(*TokPtr)) { + if (CharNo == 0) + return TokStart.getFileLocWithOffset(PhysOffset); + ++TokPtr, --CharNo, ++PhysOffset; + } + + // If we have a character that may be a trigraph or escaped newline, use a + // lexer to parse it correctly. + for (; CharNo; --CharNo) { + unsigned Size; + Lexer::getCharAndSizeNoWarn(TokPtr, Size, Features); + TokPtr += Size; + PhysOffset += Size; + } + + // Final detail: if we end up on an escaped newline, we want to return the + // location of the actual byte of the token. For example foo\<newline>bar + // advanced by 3 should return the location of b, not of \\. One compounding + // detail of this is that the escape may be made by a trigraph. + if (!Lexer::isObviouslySimpleCharacter(*TokPtr)) + PhysOffset = Lexer::SkipEscapedNewLines(TokPtr)-TokPtr; + + return TokStart.getFileLocWithOffset(PhysOffset); +} + +/// \brief Computes the source location just past the end of the +/// token at this source location. +/// +/// This routine can be used to produce a source location that +/// points just past the end of the token referenced by \p Loc, and +/// is generally used when a diagnostic needs to point just after a +/// token where it expected something different that it received. If +/// the returned source location would not be meaningful (e.g., if +/// it points into a macro), this routine returns an invalid +/// source location. +SourceLocation Preprocessor::getLocForEndOfToken(SourceLocation Loc) { + if (Loc.isInvalid() || !Loc.isFileID()) + return SourceLocation(); + + unsigned Len = Lexer::MeasureTokenLength(Loc, getSourceManager(), Features); + return AdvanceToTokenCharacter(Loc, Len); +} + + + +//===----------------------------------------------------------------------===// +// Preprocessor Initialization Methods +//===----------------------------------------------------------------------===// + + +/// EnterMainSourceFile - Enter the specified FileID as the main source file, +/// which implicitly adds the builtin defines etc. +void Preprocessor::EnterMainSourceFile() { + // We do not allow the preprocessor to reenter the main file. Doing so will + // cause FileID's to accumulate information from both runs (e.g. #line + // information) and predefined macros aren't guaranteed to be set properly. + assert(NumEnteredSourceFiles == 0 && "Cannot reenter the main file!"); + FileID MainFileID = SourceMgr.getMainFileID(); + + // Enter the main file source buffer. + EnterSourceFile(MainFileID, 0); + + // Tell the header info that the main file was entered. If the file is later + // #imported, it won't be re-entered. + if (const FileEntry *FE = SourceMgr.getFileEntryForID(MainFileID)) + HeaderInfo.IncrementIncludeCount(FE); + + std::vector<char> PrologFile; + PrologFile.reserve(4080); + + // FIXME: Don't make a copy. + PrologFile.insert(PrologFile.end(), Predefines.begin(), Predefines.end()); + + // Memory buffer must end with a null byte! + PrologFile.push_back(0); + + // Now that we have emitted the predefined macros, #includes, etc into + // PrologFile, preprocess it to populate the initial preprocessor state. + llvm::MemoryBuffer *SB = + llvm::MemoryBuffer::getMemBufferCopy(&PrologFile.front(),&PrologFile.back(), + "<built-in>"); + assert(SB && "Cannot fail to create predefined source buffer"); + FileID FID = SourceMgr.createFileIDForMemBuffer(SB); + assert(!FID.isInvalid() && "Could not create FileID for predefines?"); + + // Start parsing the predefines. + EnterSourceFile(FID, 0); +} + + +//===----------------------------------------------------------------------===// +// Lexer Event Handling. +//===----------------------------------------------------------------------===// + +/// LookUpIdentifierInfo - Given a tok::identifier token, look up the +/// identifier information for the token and install it into the token. +IdentifierInfo *Preprocessor::LookUpIdentifierInfo(Token &Identifier, + const char *BufPtr) { + assert(Identifier.is(tok::identifier) && "Not an identifier!"); + assert(Identifier.getIdentifierInfo() == 0 && "Identinfo already exists!"); + + // Look up this token, see if it is a macro, or if it is a language keyword. + IdentifierInfo *II; + if (BufPtr && !Identifier.needsCleaning()) { + // No cleaning needed, just use the characters from the lexed buffer. + II = getIdentifierInfo(BufPtr, BufPtr+Identifier.getLength()); + } else { + // Cleaning needed, alloca a buffer, clean into it, then use the buffer. + llvm::SmallVector<char, 64> IdentifierBuffer; + IdentifierBuffer.resize(Identifier.getLength()); + const char *TmpBuf = &IdentifierBuffer[0]; + unsigned Size = getSpelling(Identifier, TmpBuf); + II = getIdentifierInfo(TmpBuf, TmpBuf+Size); + } + Identifier.setIdentifierInfo(II); + return II; +} + + +/// HandleIdentifier - This callback is invoked when the lexer reads an +/// identifier. This callback looks up the identifier in the map and/or +/// potentially macro expands it or turns it into a named token (like 'for'). +/// +/// Note that callers of this method are guarded by checking the +/// IdentifierInfo's 'isHandleIdentifierCase' bit. If this method changes, the +/// IdentifierInfo methods that compute these properties will need to change to +/// match. +void Preprocessor::HandleIdentifier(Token &Identifier) { + assert(Identifier.getIdentifierInfo() && + "Can't handle identifiers without identifier info!"); + + IdentifierInfo &II = *Identifier.getIdentifierInfo(); + + // If this identifier was poisoned, and if it was not produced from a macro + // expansion, emit an error. + if (II.isPoisoned() && CurPPLexer) { + if (&II != Ident__VA_ARGS__) // We warn about __VA_ARGS__ with poisoning. + Diag(Identifier, diag::err_pp_used_poisoned_id); + else + Diag(Identifier, diag::ext_pp_bad_vaargs_use); + } + + // If this is a macro to be expanded, do it. + if (MacroInfo *MI = getMacroInfo(&II)) { + if (!DisableMacroExpansion && !Identifier.isExpandDisabled()) { + if (MI->isEnabled()) { + if (!HandleMacroExpandedIdentifier(Identifier, MI)) + return; + } else { + // C99 6.10.3.4p2 says that a disabled macro may never again be + // expanded, even if it's in a context where it could be expanded in the + // future. + Identifier.setFlag(Token::DisableExpand); + } + } + } + + // C++ 2.11p2: If this is an alternative representation of a C++ operator, + // then we act as if it is the actual operator and not the textual + // representation of it. + if (II.isCPlusPlusOperatorKeyword()) + Identifier.setIdentifierInfo(0); + + // If this is an extension token, diagnose its use. + // We avoid diagnosing tokens that originate from macro definitions. + // FIXME: This warning is disabled in cases where it shouldn't be, + // like "#define TY typeof", "TY(1) x". + if (II.isExtensionToken() && !DisableMacroExpansion) + Diag(Identifier, diag::ext_token_used); +} diff --git a/lib/Lex/PreprocessorLexer.cpp b/lib/Lex/PreprocessorLexer.cpp new file mode 100644 index 0000000..f9dfad9 --- /dev/null +++ b/lib/Lex/PreprocessorLexer.cpp @@ -0,0 +1,45 @@ +//===--- PreprocessorLexer.cpp - C Language Family Lexer ------------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements the PreprocessorLexer and Token interfaces. +// +//===----------------------------------------------------------------------===// + +#include "clang/Lex/PreprocessorLexer.h" +#include "clang/Lex/Preprocessor.h" +#include "clang/Lex/LexDiagnostic.h" +#include "clang/Basic/SourceManager.h" +using namespace clang; + +/// LexIncludeFilename - After the preprocessor has parsed a #include, lex and +/// (potentially) macro expand the filename. +void PreprocessorLexer::LexIncludeFilename(Token &FilenameTok) { + assert(ParsingPreprocessorDirective && + ParsingFilename == false && + "Must be in a preprocessing directive!"); + + // We are now parsing a filename! + ParsingFilename = true; + + // Lex the filename. + IndirectLex(FilenameTok); + + // We should have obtained the filename now. + ParsingFilename = false; + + // No filename? + if (FilenameTok.is(tok::eom)) + PP->Diag(FilenameTok.getLocation(), diag::err_pp_expects_filename); +} + +/// getFileEntry - Return the FileEntry corresponding to this FileID. Like +/// getFileID(), this only works for lexers with attached preprocessors. +const FileEntry *PreprocessorLexer::getFileEntry() const { + return PP->getSourceManager().getFileEntryForID(getFileID()); +} diff --git a/lib/Lex/ScratchBuffer.cpp b/lib/Lex/ScratchBuffer.cpp new file mode 100644 index 0000000..28f3d7f --- /dev/null +++ b/lib/Lex/ScratchBuffer.cpp @@ -0,0 +1,73 @@ +//===--- ScratchBuffer.cpp - Scratch space for forming tokens -------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements the ScratchBuffer interface. +// +//===----------------------------------------------------------------------===// + +#include "clang/Lex/ScratchBuffer.h" +#include "clang/Basic/SourceManager.h" +#include "llvm/Support/MemoryBuffer.h" +#include <cstring> +using namespace clang; + +// ScratchBufSize - The size of each chunk of scratch memory. Slightly less +//than a page, almost certainly enough for anything. :) +static const unsigned ScratchBufSize = 4060; + +ScratchBuffer::ScratchBuffer(SourceManager &SM) : SourceMgr(SM), CurBuffer(0) { + // Set BytesUsed so that the first call to getToken will require an alloc. + BytesUsed = ScratchBufSize; +} + +/// getToken - Splat the specified text into a temporary MemoryBuffer and +/// return a SourceLocation that refers to the token. This is just like the +/// method below, but returns a location that indicates the physloc of the +/// token. +SourceLocation ScratchBuffer::getToken(const char *Buf, unsigned Len, + const char *&DestPtr) { + if (BytesUsed+Len+2 > ScratchBufSize) + AllocScratchBuffer(Len+2); + + // Prefix the token with a \n, so that it looks like it is the first thing on + // its own virtual line in caret diagnostics. + CurBuffer[BytesUsed++] = '\n'; + + // Return a pointer to the character data. + DestPtr = CurBuffer+BytesUsed; + + // Copy the token data into the buffer. + memcpy(CurBuffer+BytesUsed, Buf, Len); + + // Remember that we used these bytes. + BytesUsed += Len+1; + + // Add a NUL terminator to the token. This keeps the tokens separated, in + // case they get relexed, and puts them on their own virtual lines in case a + // diagnostic points to one. + CurBuffer[BytesUsed-1] = '\0'; + + return BufferStartLoc.getFileLocWithOffset(BytesUsed-Len-1); +} + +void ScratchBuffer::AllocScratchBuffer(unsigned RequestLen) { + // Only pay attention to the requested length if it is larger than our default + // page size. If it is, we allocate an entire chunk for it. This is to + // support gigantic tokens, which almost certainly won't happen. :) + if (RequestLen < ScratchBufSize) + RequestLen = ScratchBufSize; + + llvm::MemoryBuffer *Buf = + llvm::MemoryBuffer::getNewMemBuffer(RequestLen, "<scratch space>"); + FileID FID = SourceMgr.createFileIDForMemBuffer(Buf); + BufferStartLoc = SourceMgr.getLocForStartOfFile(FID); + CurBuffer = const_cast<char*>(Buf->getBufferStart()); + BytesUsed = 1; + CurBuffer[0] = '0'; // Start out with a \0 for cleanliness. +} diff --git a/lib/Lex/TokenConcatenation.cpp b/lib/Lex/TokenConcatenation.cpp new file mode 100644 index 0000000..ab989ca --- /dev/null +++ b/lib/Lex/TokenConcatenation.cpp @@ -0,0 +1,219 @@ +//===--- TokenConcatenation.cpp - Token Concatenation Avoidance -----------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements the TokenConcatenation class. +// +//===----------------------------------------------------------------------===// + +#include "clang/Lex/TokenConcatenation.h" +#include "clang/Lex/Preprocessor.h" +using namespace clang; + + +/// StartsWithL - Return true if the spelling of this token starts with 'L'. +bool TokenConcatenation::StartsWithL(const Token &Tok) const { + if (!Tok.needsCleaning()) { + SourceManager &SM = PP.getSourceManager(); + return *SM.getCharacterData(SM.getSpellingLoc(Tok.getLocation())) == 'L'; + } + + if (Tok.getLength() < 256) { + char Buffer[256]; + const char *TokPtr = Buffer; + PP.getSpelling(Tok, TokPtr); + return TokPtr[0] == 'L'; + } + + return PP.getSpelling(Tok)[0] == 'L'; +} + +/// IsIdentifierL - Return true if the spelling of this token is literally +/// 'L'. +bool TokenConcatenation::IsIdentifierL(const Token &Tok) const { + if (!Tok.needsCleaning()) { + if (Tok.getLength() != 1) + return false; + SourceManager &SM = PP.getSourceManager(); + return *SM.getCharacterData(SM.getSpellingLoc(Tok.getLocation())) == 'L'; + } + + if (Tok.getLength() < 256) { + char Buffer[256]; + const char *TokPtr = Buffer; + if (PP.getSpelling(Tok, TokPtr) != 1) + return false; + return TokPtr[0] == 'L'; + } + + return PP.getSpelling(Tok) == "L"; +} + +TokenConcatenation::TokenConcatenation(Preprocessor &pp) : PP(pp) { + memset(TokenInfo, 0, sizeof(TokenInfo)); + + // These tokens have custom code in AvoidConcat. + TokenInfo[tok::identifier ] |= aci_custom; + TokenInfo[tok::numeric_constant] |= aci_custom_firstchar; + TokenInfo[tok::period ] |= aci_custom_firstchar; + TokenInfo[tok::amp ] |= aci_custom_firstchar; + TokenInfo[tok::plus ] |= aci_custom_firstchar; + TokenInfo[tok::minus ] |= aci_custom_firstchar; + TokenInfo[tok::slash ] |= aci_custom_firstchar; + TokenInfo[tok::less ] |= aci_custom_firstchar; + TokenInfo[tok::greater ] |= aci_custom_firstchar; + TokenInfo[tok::pipe ] |= aci_custom_firstchar; + TokenInfo[tok::percent ] |= aci_custom_firstchar; + TokenInfo[tok::colon ] |= aci_custom_firstchar; + TokenInfo[tok::hash ] |= aci_custom_firstchar; + TokenInfo[tok::arrow ] |= aci_custom_firstchar; + + // These tokens change behavior if followed by an '='. + TokenInfo[tok::amp ] |= aci_avoid_equal; // &= + TokenInfo[tok::plus ] |= aci_avoid_equal; // += + TokenInfo[tok::minus ] |= aci_avoid_equal; // -= + TokenInfo[tok::slash ] |= aci_avoid_equal; // /= + TokenInfo[tok::less ] |= aci_avoid_equal; // <= + TokenInfo[tok::greater ] |= aci_avoid_equal; // >= + TokenInfo[tok::pipe ] |= aci_avoid_equal; // |= + TokenInfo[tok::percent ] |= aci_avoid_equal; // %= + TokenInfo[tok::star ] |= aci_avoid_equal; // *= + TokenInfo[tok::exclaim ] |= aci_avoid_equal; // != + TokenInfo[tok::lessless ] |= aci_avoid_equal; // <<= + TokenInfo[tok::greaterequal] |= aci_avoid_equal; // >>= + TokenInfo[tok::caret ] |= aci_avoid_equal; // ^= + TokenInfo[tok::equal ] |= aci_avoid_equal; // == +} + +/// GetFirstChar - Get the first character of the token \arg Tok, +/// avoiding calls to getSpelling where possible. +static char GetFirstChar(Preprocessor &PP, const Token &Tok) { + if (IdentifierInfo *II = Tok.getIdentifierInfo()) { + // Avoid spelling identifiers, the most common form of token. + return II->getName()[0]; + } else if (!Tok.needsCleaning()) { + if (Tok.isLiteral() && Tok.getLiteralData()) { + return *Tok.getLiteralData(); + } else { + SourceManager &SM = PP.getSourceManager(); + return *SM.getCharacterData(SM.getSpellingLoc(Tok.getLocation())); + } + } else if (Tok.getLength() < 256) { + char Buffer[256]; + const char *TokPtr = Buffer; + PP.getSpelling(Tok, TokPtr); + return TokPtr[0]; + } else { + return PP.getSpelling(Tok)[0]; + } +} + +/// AvoidConcat - If printing PrevTok immediately followed by Tok would cause +/// the two individual tokens to be lexed as a single token, return true +/// (which causes a space to be printed between them). This allows the output +/// of -E mode to be lexed to the same token stream as lexing the input +/// directly would. +/// +/// This code must conservatively return true if it doesn't want to be 100% +/// accurate. This will cause the output to include extra space characters, +/// but the resulting output won't have incorrect concatenations going on. +/// Examples include "..", which we print with a space between, because we +/// don't want to track enough to tell "x.." from "...". +bool TokenConcatenation::AvoidConcat(const Token &PrevTok, + const Token &Tok) const { + // First, check to see if the tokens were directly adjacent in the original + // source. If they were, it must be okay to stick them together: if there + // were an issue, the tokens would have been lexed differently. + if (PrevTok.getLocation().isFileID() && Tok.getLocation().isFileID() && + PrevTok.getLocation().getFileLocWithOffset(PrevTok.getLength()) == + Tok.getLocation()) + return false; + + tok::TokenKind PrevKind = PrevTok.getKind(); + if (PrevTok.getIdentifierInfo()) // Language keyword or named operator. + PrevKind = tok::identifier; + + // Look up information on when we should avoid concatenation with prevtok. + unsigned ConcatInfo = TokenInfo[PrevKind]; + + // If prevtok never causes a problem for anything after it, return quickly. + if (ConcatInfo == 0) return false; + + if (ConcatInfo & aci_avoid_equal) { + // If the next token is '=' or '==', avoid concatenation. + if (Tok.is(tok::equal) || Tok.is(tok::equalequal)) + return true; + ConcatInfo &= ~aci_avoid_equal; + } + + if (ConcatInfo == 0) return false; + + // Basic algorithm: we look at the first character of the second token, and + // determine whether it, if appended to the first token, would form (or + // would contribute) to a larger token if concatenated. + char FirstChar = 0; + if (ConcatInfo & aci_custom) { + // If the token does not need to know the first character, don't get it. + } else { + FirstChar = GetFirstChar(PP, Tok); + } + + switch (PrevKind) { + default: assert(0 && "InitAvoidConcatTokenInfo built wrong"); + case tok::identifier: // id+id or id+number or id+L"foo". + // id+'.'... will not append. + if (Tok.is(tok::numeric_constant)) + return GetFirstChar(PP, Tok) != '.'; + + if (Tok.getIdentifierInfo() || Tok.is(tok::wide_string_literal) /* || + Tok.is(tok::wide_char_literal)*/) + return true; + + // If this isn't identifier + string, we're done. + if (Tok.isNot(tok::char_constant) && Tok.isNot(tok::string_literal)) + return false; + + // FIXME: need a wide_char_constant! + + // If the string was a wide string L"foo" or wide char L'f', it would + // concat with the previous identifier into fooL"bar". Avoid this. + if (StartsWithL(Tok)) + return true; + + // Otherwise, this is a narrow character or string. If the *identifier* + // is a literal 'L', avoid pasting L "foo" -> L"foo". + return IsIdentifierL(PrevTok); + case tok::numeric_constant: + return isalnum(FirstChar) || Tok.is(tok::numeric_constant) || + FirstChar == '+' || FirstChar == '-' || FirstChar == '.'; + case tok::period: // ..., .*, .1234 + return FirstChar == '.' || isdigit(FirstChar) || FirstChar == '*'; + case tok::amp: // && + return FirstChar == '&'; + case tok::plus: // ++ + return FirstChar == '+'; + case tok::minus: // --, ->, ->* + return FirstChar == '-' || FirstChar == '>'; + case tok::slash: //, /*, // + return FirstChar == '*' || FirstChar == '/'; + case tok::less: // <<, <<=, <:, <% + return FirstChar == '<' || FirstChar == ':' || FirstChar == '%'; + case tok::greater: // >>, >>= + return FirstChar == '>'; + case tok::pipe: // || + return FirstChar == '|'; + case tok::percent: // %>, %: + return FirstChar == '>' || FirstChar == ':'; + case tok::colon: // ::, :> + return FirstChar == ':' ||FirstChar == '>'; + case tok::hash: // ##, #@, %:%: + return FirstChar == '#' || FirstChar == '@' || FirstChar == '%'; + case tok::arrow: // ->* + return FirstChar == '*'; + } +} diff --git a/lib/Lex/TokenLexer.cpp b/lib/Lex/TokenLexer.cpp new file mode 100644 index 0000000..f9f9386 --- /dev/null +++ b/lib/Lex/TokenLexer.cpp @@ -0,0 +1,542 @@ +//===--- TokenLexer.cpp - Lex from a token stream -------------------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements the TokenLexer interface. +// +//===----------------------------------------------------------------------===// + +#include "clang/Lex/TokenLexer.h" +#include "MacroArgs.h" +#include "clang/Lex/MacroInfo.h" +#include "clang/Lex/Preprocessor.h" +#include "clang/Basic/SourceManager.h" +#include "clang/Lex/LexDiagnostic.h" +#include "llvm/ADT/SmallVector.h" +using namespace clang; + + +/// Create a TokenLexer for the specified macro with the specified actual +/// arguments. Note that this ctor takes ownership of the ActualArgs pointer. +void TokenLexer::Init(Token &Tok, SourceLocation ILEnd, MacroArgs *Actuals) { + // If the client is reusing a TokenLexer, make sure to free any memory + // associated with it. + destroy(); + + Macro = PP.getMacroInfo(Tok.getIdentifierInfo()); + ActualArgs = Actuals; + CurToken = 0; + + InstantiateLocStart = Tok.getLocation(); + InstantiateLocEnd = ILEnd; + AtStartOfLine = Tok.isAtStartOfLine(); + HasLeadingSpace = Tok.hasLeadingSpace(); + Tokens = &*Macro->tokens_begin(); + OwnsTokens = false; + DisableMacroExpansion = false; + NumTokens = Macro->tokens_end()-Macro->tokens_begin(); + + // If this is a function-like macro, expand the arguments and change + // Tokens to point to the expanded tokens. + if (Macro->isFunctionLike() && Macro->getNumArgs()) + ExpandFunctionArguments(); + + // Mark the macro as currently disabled, so that it is not recursively + // expanded. The macro must be disabled only after argument pre-expansion of + // function-like macro arguments occurs. + Macro->DisableMacro(); +} + + + +/// Create a TokenLexer for the specified token stream. This does not +/// take ownership of the specified token vector. +void TokenLexer::Init(const Token *TokArray, unsigned NumToks, + bool disableMacroExpansion, bool ownsTokens) { + // If the client is reusing a TokenLexer, make sure to free any memory + // associated with it. + destroy(); + + Macro = 0; + ActualArgs = 0; + Tokens = TokArray; + OwnsTokens = ownsTokens; + DisableMacroExpansion = disableMacroExpansion; + NumTokens = NumToks; + CurToken = 0; + InstantiateLocStart = InstantiateLocEnd = SourceLocation(); + AtStartOfLine = false; + HasLeadingSpace = false; + + // Set HasLeadingSpace/AtStartOfLine so that the first token will be + // returned unmodified. + if (NumToks != 0) { + AtStartOfLine = TokArray[0].isAtStartOfLine(); + HasLeadingSpace = TokArray[0].hasLeadingSpace(); + } +} + + +void TokenLexer::destroy() { + // If this was a function-like macro that actually uses its arguments, delete + // the expanded tokens. + if (OwnsTokens) { + delete [] Tokens; + Tokens = 0; + OwnsTokens = false; + } + + // TokenLexer owns its formal arguments. + if (ActualArgs) ActualArgs->destroy(); +} + +/// Expand the arguments of a function-like macro so that we can quickly +/// return preexpanded tokens from Tokens. +void TokenLexer::ExpandFunctionArguments() { + llvm::SmallVector<Token, 128> ResultToks; + + // Loop through 'Tokens', expanding them into ResultToks. Keep + // track of whether we change anything. If not, no need to keep them. If so, + // we install the newly expanded sequence as the new 'Tokens' list. + bool MadeChange = false; + + // NextTokGetsSpace - When this is true, the next token appended to the + // output list will get a leading space, regardless of whether it had one to + // begin with or not. This is used for placemarker support. + bool NextTokGetsSpace = false; + + for (unsigned i = 0, e = NumTokens; i != e; ++i) { + // If we found the stringify operator, get the argument stringified. The + // preprocessor already verified that the following token is a macro name + // when the #define was parsed. + const Token &CurTok = Tokens[i]; + if (CurTok.is(tok::hash) || CurTok.is(tok::hashat)) { + int ArgNo = Macro->getArgumentNum(Tokens[i+1].getIdentifierInfo()); + assert(ArgNo != -1 && "Token following # is not an argument?"); + + Token Res; + if (CurTok.is(tok::hash)) // Stringify + Res = ActualArgs->getStringifiedArgument(ArgNo, PP); + else { + // 'charify': don't bother caching these. + Res = MacroArgs::StringifyArgument(ActualArgs->getUnexpArgument(ArgNo), + PP, true); + } + + // The stringified/charified string leading space flag gets set to match + // the #/#@ operator. + if (CurTok.hasLeadingSpace() || NextTokGetsSpace) + Res.setFlag(Token::LeadingSpace); + + ResultToks.push_back(Res); + MadeChange = true; + ++i; // Skip arg name. + NextTokGetsSpace = false; + continue; + } + + // Otherwise, if this is not an argument token, just add the token to the + // output buffer. + IdentifierInfo *II = CurTok.getIdentifierInfo(); + int ArgNo = II ? Macro->getArgumentNum(II) : -1; + if (ArgNo == -1) { + // This isn't an argument, just add it. + ResultToks.push_back(CurTok); + + if (NextTokGetsSpace) { + ResultToks.back().setFlag(Token::LeadingSpace); + NextTokGetsSpace = false; + } + continue; + } + + // An argument is expanded somehow, the result is different than the + // input. + MadeChange = true; + + // Otherwise, this is a use of the argument. Find out if there is a paste + // (##) operator before or after the argument. + bool PasteBefore = + !ResultToks.empty() && ResultToks.back().is(tok::hashhash); + bool PasteAfter = i+1 != e && Tokens[i+1].is(tok::hashhash); + + // If it is not the LHS/RHS of a ## operator, we must pre-expand the + // argument and substitute the expanded tokens into the result. This is + // C99 6.10.3.1p1. + if (!PasteBefore && !PasteAfter) { + const Token *ResultArgToks; + + // Only preexpand the argument if it could possibly need it. This + // avoids some work in common cases. + const Token *ArgTok = ActualArgs->getUnexpArgument(ArgNo); + if (ActualArgs->ArgNeedsPreexpansion(ArgTok, PP)) + ResultArgToks = &ActualArgs->getPreExpArgument(ArgNo, PP)[0]; + else + ResultArgToks = ArgTok; // Use non-preexpanded tokens. + + // If the arg token expanded into anything, append it. + if (ResultArgToks->isNot(tok::eof)) { + unsigned FirstResult = ResultToks.size(); + unsigned NumToks = MacroArgs::getArgLength(ResultArgToks); + ResultToks.append(ResultArgToks, ResultArgToks+NumToks); + + // If any tokens were substituted from the argument, the whitespace + // before the first token should match the whitespace of the arg + // identifier. + ResultToks[FirstResult].setFlagValue(Token::LeadingSpace, + CurTok.hasLeadingSpace() || + NextTokGetsSpace); + NextTokGetsSpace = false; + } else { + // If this is an empty argument, and if there was whitespace before the + // formal token, make sure the next token gets whitespace before it. + NextTokGetsSpace = CurTok.hasLeadingSpace(); + } + continue; + } + + // Okay, we have a token that is either the LHS or RHS of a paste (##) + // argument. It gets substituted as its non-pre-expanded tokens. + const Token *ArgToks = ActualArgs->getUnexpArgument(ArgNo); + unsigned NumToks = MacroArgs::getArgLength(ArgToks); + if (NumToks) { // Not an empty argument? + // If this is the GNU ", ## __VA_ARG__" extension, and we just learned + // that __VA_ARG__ expands to multiple tokens, avoid a pasting error when + // the expander trys to paste ',' with the first token of the __VA_ARG__ + // expansion. + if (PasteBefore && ResultToks.size() >= 2 && + ResultToks[ResultToks.size()-2].is(tok::comma) && + (unsigned)ArgNo == Macro->getNumArgs()-1 && + Macro->isVariadic()) { + // Remove the paste operator, report use of the extension. + PP.Diag(ResultToks.back().getLocation(), diag::ext_paste_comma); + ResultToks.pop_back(); + } + + ResultToks.append(ArgToks, ArgToks+NumToks); + + // If this token (the macro argument) was supposed to get leading + // whitespace, transfer this information onto the first token of the + // expansion. + // + // Do not do this if the paste operator occurs before the macro argument, + // as in "A ## MACROARG". In valid code, the first token will get + // smooshed onto the preceding one anyway (forming AMACROARG). In + // assembler-with-cpp mode, invalid pastes are allowed through: in this + // case, we do not want the extra whitespace to be added. For example, + // we want ". ## foo" -> ".foo" not ". foo". + if ((CurTok.hasLeadingSpace() || NextTokGetsSpace) && + !PasteBefore) + ResultToks[ResultToks.size()-NumToks].setFlag(Token::LeadingSpace); + + NextTokGetsSpace = false; + continue; + } + + // If an empty argument is on the LHS or RHS of a paste, the standard (C99 + // 6.10.3.3p2,3) calls for a bunch of placemarker stuff to occur. We + // implement this by eating ## operators when a LHS or RHS expands to + // empty. + NextTokGetsSpace |= CurTok.hasLeadingSpace(); + if (PasteAfter) { + // Discard the argument token and skip (don't copy to the expansion + // buffer) the paste operator after it. + NextTokGetsSpace |= Tokens[i+1].hasLeadingSpace(); + ++i; + continue; + } + + // If this is on the RHS of a paste operator, we've already copied the + // paste operator to the ResultToks list. Remove it. + assert(PasteBefore && ResultToks.back().is(tok::hashhash)); + NextTokGetsSpace |= ResultToks.back().hasLeadingSpace(); + ResultToks.pop_back(); + + // If this is the __VA_ARGS__ token, and if the argument wasn't provided, + // and if the macro had at least one real argument, and if the token before + // the ## was a comma, remove the comma. + if ((unsigned)ArgNo == Macro->getNumArgs()-1 && // is __VA_ARGS__ + ActualArgs->isVarargsElidedUse() && // Argument elided. + !ResultToks.empty() && ResultToks.back().is(tok::comma)) { + // Never add a space, even if the comma, ##, or arg had a space. + NextTokGetsSpace = false; + // Remove the paste operator, report use of the extension. + PP.Diag(ResultToks.back().getLocation(), diag::ext_paste_comma); + ResultToks.pop_back(); + } + continue; + } + + // If anything changed, install this as the new Tokens list. + if (MadeChange) { + assert(!OwnsTokens && "This would leak if we already own the token list"); + // This is deleted in the dtor. + NumTokens = ResultToks.size(); + llvm::BumpPtrAllocator &Alloc = PP.getPreprocessorAllocator(); + Token *Res = + static_cast<Token *>(Alloc.Allocate(sizeof(Token)*ResultToks.size(), + llvm::alignof<Token>())); + if (NumTokens) + memcpy(Res, &ResultToks[0], NumTokens*sizeof(Token)); + Tokens = Res; + + // The preprocessor bump pointer owns these tokens, not us. + OwnsTokens = false; + } +} + +/// Lex - Lex and return a token from this macro stream. +/// +void TokenLexer::Lex(Token &Tok) { + // Lexing off the end of the macro, pop this macro off the expansion stack. + if (isAtEnd()) { + // If this is a macro (not a token stream), mark the macro enabled now + // that it is no longer being expanded. + if (Macro) Macro->EnableMacro(); + + // Pop this context off the preprocessors lexer stack and get the next + // token. This will delete "this" so remember the PP instance var. + Preprocessor &PPCache = PP; + if (PP.HandleEndOfTokenLexer(Tok)) + return; + + // HandleEndOfTokenLexer may not return a token. If it doesn't, lex + // whatever is next. + return PPCache.Lex(Tok); + } + + // If this is the first token of the expanded result, we inherit spacing + // properties later. + bool isFirstToken = CurToken == 0; + + // Get the next token to return. + Tok = Tokens[CurToken++]; + + bool TokenIsFromPaste = false; + + // If this token is followed by a token paste (##) operator, paste the tokens! + if (!isAtEnd() && Tokens[CurToken].is(tok::hashhash)) { + if (PasteTokens(Tok)) { + // When handling the microsoft /##/ extension, the final token is + // returned by PasteTokens, not the pasted token. + return; + } else { + TokenIsFromPaste = true; + } + } + + // The token's current location indicate where the token was lexed from. We + // need this information to compute the spelling of the token, but any + // diagnostics for the expanded token should appear as if they came from + // InstantiationLoc. Pull this information together into a new SourceLocation + // that captures all of this. + if (InstantiateLocStart.isValid()) { // Don't do this for token streams. + SourceManager &SM = PP.getSourceManager(); + Tok.setLocation(SM.createInstantiationLoc(Tok.getLocation(), + InstantiateLocStart, + InstantiateLocEnd, + Tok.getLength())); + } + + // If this is the first token, set the lexical properties of the token to + // match the lexical properties of the macro identifier. + if (isFirstToken) { + Tok.setFlagValue(Token::StartOfLine , AtStartOfLine); + Tok.setFlagValue(Token::LeadingSpace, HasLeadingSpace); + } + + // Handle recursive expansion! + if (!Tok.isAnnotation() && Tok.getIdentifierInfo() != 0) { + // Change the kind of this identifier to the appropriate token kind, e.g. + // turning "for" into a keyword. + IdentifierInfo *II = Tok.getIdentifierInfo(); + Tok.setKind(II->getTokenID()); + + // If this identifier was poisoned and from a paste, emit an error. This + // won't be handled by Preprocessor::HandleIdentifier because this is coming + // from a macro expansion. + if (II->isPoisoned() && TokenIsFromPaste) { + // We warn about __VA_ARGS__ with poisoning. + if (II->isStr("__VA_ARGS__")) + PP.Diag(Tok, diag::ext_pp_bad_vaargs_use); + else + PP.Diag(Tok, diag::err_pp_used_poisoned_id); + } + + if (!DisableMacroExpansion && II->isHandleIdentifierCase()) + PP.HandleIdentifier(Tok); + } + + // Otherwise, return a normal token. +} + +/// PasteTokens - Tok is the LHS of a ## operator, and CurToken is the ## +/// operator. Read the ## and RHS, and paste the LHS/RHS together. If there +/// are more ## after it, chomp them iteratively. Return the result as Tok. +/// If this returns true, the caller should immediately return the token. +bool TokenLexer::PasteTokens(Token &Tok) { + llvm::SmallVector<char, 128> Buffer; + const char *ResultTokStrPtr = 0; + do { + // Consume the ## operator. + SourceLocation PasteOpLoc = Tokens[CurToken].getLocation(); + ++CurToken; + assert(!isAtEnd() && "No token on the RHS of a paste operator!"); + + // Get the RHS token. + const Token &RHS = Tokens[CurToken]; + + // Allocate space for the result token. This is guaranteed to be enough for + // the two tokens. + Buffer.resize(Tok.getLength() + RHS.getLength()); + + // Get the spelling of the LHS token in Buffer. + const char *BufPtr = &Buffer[0]; + unsigned LHSLen = PP.getSpelling(Tok, BufPtr); + if (BufPtr != &Buffer[0]) // Really, we want the chars in Buffer! + memcpy(&Buffer[0], BufPtr, LHSLen); + + BufPtr = &Buffer[LHSLen]; + unsigned RHSLen = PP.getSpelling(RHS, BufPtr); + if (BufPtr != &Buffer[LHSLen]) // Really, we want the chars in Buffer! + memcpy(&Buffer[LHSLen], BufPtr, RHSLen); + + // Trim excess space. + Buffer.resize(LHSLen+RHSLen); + + // Plop the pasted result (including the trailing newline and null) into a + // scratch buffer where we can lex it. + Token ResultTokTmp; + ResultTokTmp.startToken(); + + // Claim that the tmp token is a string_literal so that we can get the + // character pointer back from CreateString. + ResultTokTmp.setKind(tok::string_literal); + PP.CreateString(&Buffer[0], Buffer.size(), ResultTokTmp); + SourceLocation ResultTokLoc = ResultTokTmp.getLocation(); + ResultTokStrPtr = ResultTokTmp.getLiteralData(); + + // Lex the resultant pasted token into Result. + Token Result; + + if (Tok.is(tok::identifier) && RHS.is(tok::identifier)) { + // Common paste case: identifier+identifier = identifier. Avoid creating + // a lexer and other overhead. + PP.IncrementPasteCounter(true); + Result.startToken(); + Result.setKind(tok::identifier); + Result.setLocation(ResultTokLoc); + Result.setLength(LHSLen+RHSLen); + } else { + PP.IncrementPasteCounter(false); + + assert(ResultTokLoc.isFileID() && + "Should be a raw location into scratch buffer"); + SourceManager &SourceMgr = PP.getSourceManager(); + FileID LocFileID = SourceMgr.getFileID(ResultTokLoc); + + const char *ScratchBufStart = SourceMgr.getBufferData(LocFileID).first; + + // Make a lexer to lex this string from. Lex just this one token. + // Make a lexer object so that we lex and expand the paste result. + Lexer TL(SourceMgr.getLocForStartOfFile(LocFileID), + PP.getLangOptions(), ScratchBufStart, + ResultTokStrPtr, ResultTokStrPtr+LHSLen+RHSLen); + + // Lex a token in raw mode. This way it won't look up identifiers + // automatically, lexing off the end will return an eof token, and + // warnings are disabled. This returns true if the result token is the + // entire buffer. + bool isInvalid = !TL.LexFromRawLexer(Result); + + // If we got an EOF token, we didn't form even ONE token. For example, we + // did "/ ## /" to get "//". + isInvalid |= Result.is(tok::eof); + + // If pasting the two tokens didn't form a full new token, this is an + // error. This occurs with "x ## +" and other stuff. Return with Tok + // unmodified and with RHS as the next token to lex. + if (isInvalid) { + // Test for the Microsoft extension of /##/ turning into // here on the + // error path. + if (PP.getLangOptions().Microsoft && Tok.is(tok::slash) && + RHS.is(tok::slash)) { + HandleMicrosoftCommentPaste(Tok); + return true; + } + + // Do not emit the warning when preprocessing assembler code. + if (!PP.getLangOptions().AsmPreprocessor) { + // Explicitly convert the token location to have proper instantiation + // information so that the user knows where it came from. + SourceManager &SM = PP.getSourceManager(); + SourceLocation Loc = + SM.createInstantiationLoc(PasteOpLoc, InstantiateLocStart, + InstantiateLocEnd, 2); + PP.Diag(Loc, diag::err_pp_bad_paste) + << std::string(Buffer.begin(), Buffer.end()); + } + + // Do not consume the RHS. + --CurToken; + } + + // Turn ## into 'unknown' to avoid # ## # from looking like a paste + // operator. + if (Result.is(tok::hashhash)) + Result.setKind(tok::unknown); + } + + // Transfer properties of the LHS over the the Result. + Result.setFlagValue(Token::StartOfLine , Tok.isAtStartOfLine()); + Result.setFlagValue(Token::LeadingSpace, Tok.hasLeadingSpace()); + + // Finally, replace LHS with the result, consume the RHS, and iterate. + ++CurToken; + Tok = Result; + } while (!isAtEnd() && Tokens[CurToken].is(tok::hashhash)); + + // Now that we got the result token, it will be subject to expansion. Since + // token pasting re-lexes the result token in raw mode, identifier information + // isn't looked up. As such, if the result is an identifier, look up id info. + if (Tok.is(tok::identifier)) { + // Look up the identifier info for the token. We disabled identifier lookup + // by saying we're skipping contents, so we need to do this manually. + IdentifierInfo *II = PP.LookUpIdentifierInfo(Tok, ResultTokStrPtr); + Tok.setIdentifierInfo(II); + } + return false; +} + +/// isNextTokenLParen - If the next token lexed will pop this macro off the +/// expansion stack, return 2. If the next unexpanded token is a '(', return +/// 1, otherwise return 0. +unsigned TokenLexer::isNextTokenLParen() const { + // Out of tokens? + if (isAtEnd()) + return 2; + return Tokens[CurToken].is(tok::l_paren); +} + + +/// HandleMicrosoftCommentPaste - In microsoft compatibility mode, /##/ pastes +/// together to form a comment that comments out everything in the current +/// macro, other active macros, and anything left on the current physical +/// source line of the instantiated buffer. Handle this by returning the +/// first token on the next line. +void TokenLexer::HandleMicrosoftCommentPaste(Token &Tok) { + // We 'comment out' the rest of this macro by just ignoring the rest of the + // tokens that have not been lexed yet, if any. + + // Since this must be a macro, mark the macro enabled now that it is no longer + // being expanded. + assert(Macro && "Token streams can't paste comments"); + Macro->EnableMacro(); + + PP.HandleMicrosoftCommentPaste(Tok); +} |
