diff options
author | obrien <obrien@FreeBSD.org> | 2000-01-04 03:25:40 +0000 |
---|---|---|
committer | obrien <obrien@FreeBSD.org> | 2000-01-04 03:25:40 +0000 |
commit | 6a9796d7d65e9c34217a0ad2bd387b12f865327e (patch) | |
tree | c66accfbf4d060e9e088ef8de10261d9fb7d96d8 | |
parent | 8aa7fd43d21fb804431830945d20b9ccab78170f (diff) | |
download | FreeBSD-src-6a9796d7d65e9c34217a0ad2bd387b12f865327e.zip FreeBSD-src-6a9796d7d65e9c34217a0ad2bd387b12f865327e.tar.gz |
Merge FreeBSD chagnes into GNU grep 2.4.
-rw-r--r-- | gnu/usr.bin/grep/dfa.c | 88 | ||||
-rw-r--r-- | gnu/usr.bin/grep/dfa.h | 7 | ||||
-rw-r--r-- | gnu/usr.bin/grep/grep.1 | 252 | ||||
-rw-r--r-- | gnu/usr.bin/grep/grep.c | 486 | ||||
-rw-r--r-- | gnu/usr.bin/grep/grep.h | 6 | ||||
-rw-r--r-- | gnu/usr.bin/grep/search.c | 64 |
6 files changed, 557 insertions, 346 deletions
diff --git a/gnu/usr.bin/grep/dfa.c b/gnu/usr.bin/grep/dfa.c index 9f20c37..1fcf549 100644 --- a/gnu/usr.bin/grep/dfa.c +++ b/gnu/usr.bin/grep/dfa.c @@ -353,15 +353,20 @@ static reg_syntax_t syntax_bits, syntax_bits_set; /* Flag for case-folding letters into sets. */ static int case_fold; +/* End-of-line byte in data. */ +static unsigned char eolbyte; + /* Entry point to set syntax options. */ void -dfasyntax(bits, fold) +dfasyntax(bits, fold, eol) reg_syntax_t bits; int fold; + int eol; { syntax_bits_set = 1; syntax_bits = bits; case_fold = fold; + eolbyte = eol; } /* Lexical analyzer. All the dross that deals with the obnoxious @@ -580,11 +585,32 @@ lex() goto normal_char; if (backslash != ((syntax_bits & RE_NO_BK_BRACES) == 0)) goto normal_char; - minrep = maxrep = 0; + if (!(syntax_bits & RE_CONTEXT_INDEP_OPS) && laststart) + goto normal_char; + + if (syntax_bits & RE_NO_BK_BRACES) + { + /* Scan ahead for a valid interval; if it's not valid, + treat it as a literal '{'. */ + int lo = -1, hi = -1; + char const *p = lexptr; + char const *lim = p + lexleft; + for (; p != lim && ISDIGIT (*p); p++) + lo = (lo < 0 ? 0 : lo * 10) + *p - '0'; + if (p != lim && *p == ',') + while (++p != lim && ISDIGIT (*p)) + hi = (hi < 0 ? 0 : hi * 10) + *p - '0'; + else + hi = lo; + if (p == lim || *p != '}' + || lo < 0 || RE_DUP_MAX < hi || (0 <= hi && hi < lo)) + goto normal_char; + } + + minrep = 0; /* Cases: {M} - exact count {M,} - minimum count, maximum is infinity - {,M} - 0 through M {M,N} - M through N */ FETCH(c, _("unfinished repeat count")); if (ISDIGIT(c)) @@ -598,16 +624,27 @@ lex() minrep = 10 * minrep + c - '0'; } } - else if (c != ',') + else dfaerror(_("malformed repeat count")); if (c == ',') - for (;;) - { - FETCH(c, _("unfinished repeat count")); - if (!ISDIGIT(c)) - break; - maxrep = 10 * maxrep + c - '0'; - } + { + FETCH (c, _("unfinished repeat count")); + if (! ISDIGIT (c)) + maxrep = -1; + else + { + maxrep = c - '0'; + for (;;) + { + FETCH (c, _("unfinished repeat count")); + if (! ISDIGIT (c)) + break; + maxrep = 10 * maxrep + c - '0'; + } + if (0 <= maxrep && maxrep < minrep) + dfaerror (_("malformed repeat count")); + } + } else maxrep = minrep; if (!(syntax_bits & RE_NO_BK_BRACES)) @@ -659,7 +696,7 @@ lex() zeroset(ccl); notset(ccl); if (!(syntax_bits & RE_DOT_NEWLINE)) - clrbit('\n', ccl); + clrbit(eolbyte, ccl); if (syntax_bits & RE_DOT_NOT_NULL) clrbit('\0', ccl); laststart = 0; @@ -776,7 +813,7 @@ lex() { notset(ccl); if (syntax_bits & RE_HAT_LISTS_NOT_NEWLINE) - clrbit('\n', ccl); + clrbit(eolbyte, ccl); } laststart = 0; return lasttok = CSET + charclass_index(ccl); @@ -942,7 +979,7 @@ closure() { ntokens = nsubtoks(dfa->tindex); tindex = dfa->tindex - ntokens; - if (maxrep == 0) + if (maxrep < 0) addtok(PLUS); if (minrep == 0) addtok(QMARK); @@ -1605,7 +1642,7 @@ dfastate(s, d, trans) for (i = 0; i < NOTCHAR; ++i) if (IS_WORD_CONSTITUENT(i)) setbit(i, letters); - setbit('\n', newline); + setbit(eolbyte, newline); } zeroset(matches); @@ -1626,7 +1663,7 @@ dfastate(s, d, trans) { if (! MATCHES_NEWLINE_CONTEXT(pos.constraint, d->states[s].newline, 1)) - clrbit('\n', matches); + clrbit(eolbyte, matches); if (! MATCHES_NEWLINE_CONTEXT(pos.constraint, d->states[s].newline, 0)) for (j = 0; j < CHARCLASS_INTS; ++j) @@ -1737,7 +1774,7 @@ dfastate(s, d, trans) state_letter = state; for (i = 0; i < NOTCHAR; ++i) trans[i] = (IS_WORD_CONSTITUENT(i)) ? state_letter : state; - trans['\n'] = state_newline; + trans[eolbyte] = state_newline; } else for (i = 0; i < NOTCHAR; ++i) @@ -1761,7 +1798,7 @@ dfastate(s, d, trans) /* Find out if the new state will want any context information. */ wants_newline = 0; - if (tstbit('\n', labels[i])) + if (tstbit(eolbyte, labels[i])) for (j = 0; j < follows.nelem; ++j) if (PREV_NEWLINE_DEPENDENT(follows.elems[j].constraint)) wants_newline = 1; @@ -1793,7 +1830,7 @@ dfastate(s, d, trans) { int c = j * INTBITS + k; - if (c == '\n') + if (c == eolbyte) trans[c] = state_newline; else if (IS_WORD_CONSTITUENT(c)) trans[c] = state_letter; @@ -1884,8 +1921,8 @@ build_state(s, d) /* Keep the newline transition in a special place so we can use it as a sentinel. */ - d->newlines[s] = trans['\n']; - trans['\n'] = -1; + d->newlines[s] = trans[eolbyte]; + trans[eolbyte] = -1; if (ACCEPTING(s, *d)) d->fails[s] = trans; @@ -1933,6 +1970,7 @@ dfaexec(d, begin, end, newline, count, backref) register unsigned char *p; /* Current input character. */ register int **trans, *t; /* Copy of d->trans so it can be optimized into a register. */ + register unsigned char eol = eolbyte; /* Likewise for eolbyte. */ static int sbit[NOTCHAR]; /* Table for anding with d->success. */ static int sbit_init; @@ -1943,7 +1981,7 @@ dfaexec(d, begin, end, newline, count, backref) sbit_init = 1; for (i = 0; i < NOTCHAR; ++i) sbit[i] = (IS_WORD_CONSTITUENT(i)) ? 2 : 1; - sbit['\n'] = 4; + sbit[eol] = 4; } if (! d->tralloc) @@ -1952,7 +1990,7 @@ dfaexec(d, begin, end, newline, count, backref) s = s1 = 0; p = (unsigned char *) begin; trans = d->trans; - *end = '\n'; + *end = eol; for (;;) { @@ -1980,7 +2018,7 @@ dfaexec(d, begin, end, newline, count, backref) } /* If the previous character was a newline, count it. */ - if (count && (char *) p <= end && p[-1] == '\n') + if (count && (char *) p <= end && p[-1] == eol) ++*count; /* Check if we've run off the end of the buffer. */ @@ -1994,7 +2032,7 @@ dfaexec(d, begin, end, newline, count, backref) continue; } - if (p[-1] == '\n' && newline) + if (p[-1] == eol && newline) { s = d->newlines[s1]; continue; diff --git a/gnu/usr.bin/grep/dfa.h b/gnu/usr.bin/grep/dfa.h index 1e12515..2a7fed2 100644 --- a/gnu/usr.bin/grep/dfa.h +++ b/gnu/usr.bin/grep/dfa.h @@ -322,9 +322,10 @@ struct dfa /* Entry points. */ -/* dfasyntax() takes two arguments; the first sets the syntax bits described - earlier in this file, and the second sets the case-folding flag. */ -extern void dfasyntax PARAMS ((reg_syntax_t, int)); +/* dfasyntax() takes three arguments; the first sets the syntax bits described + earlier in this file, the second sets the case-folding flag, and the + third specifies the line terminator. */ +extern void dfasyntax PARAMS ((reg_syntax_t, int, int)); /* Compile the given string of the given length into the given struct dfa. Final argument is a flag specifying whether to build a searching or an diff --git a/gnu/usr.bin/grep/grep.1 b/gnu/usr.bin/grep/grep.1 index 75361a9..6fbcb99 100644 --- a/gnu/usr.bin/grep/grep.1 +++ b/gnu/usr.bin/grep/grep.1 @@ -1,26 +1,68 @@ .\" grep man page .\" $FreeBSD$ +.if !\n(.g \{\ +. if !\w|\*(lq| \{\ +. ds lq `` +. if \w'\(lq' .ds lq "\(lq +. \} +. if !\w|\*(rq| \{\ +. ds rq '' +. if \w'\(rq' .ds rq "\(rq +. \} +.\} .de Id .ds Dt \\$4 .. -.Id $Id: grep.1,v 1.1 1998/11/22 06:45:20 alainm Exp $ +.Id $Id: grep.1,v 1.7 1999/10/12 20:41:01 alainm Exp $ .TH GREP 1 \*(Dt "GNU Project" .SH NAME grep, egrep, fgrep, zgrep \- print lines matching a pattern .SH SYNOPSIS .B grep -[-[AB] NUM] [-CEFGVZabchiLlnqrsvwxyUu] [-e PATTERN | -f FILE] -[-d ACTION] [--directories=ACTION] -[--extended-regexp] [--fixed-strings] [--basic-regexp] -[--regexp=PATTERN] [--file=FILE] [--ignore-case] [--word-regexp] -[--line-regexp] [--line-regexp] [--no-messages] [--revert-match] -[--version] [--help] [--byte-offset] [--line-number] -[--with-filename] [--no-filename] [--quiet] [--silent] [--text] -[--files-without-match] [--files-with-matcces] [--count] -[--before-context=NUM] [--after-context=NUM] [--context] -[--binary] [--unix-byte-offsets] [--recursive] -[--decompress] -.I files... +.RB [ \- [ ABC ] +.IR NUM ] +.RB [ \-EFGHLUVZabchilnqrsuvwxyuz ] +.RB [ \-e +.I PATTERN +| +.B \-f +.IR FILE ] +.RB [ \-d +.IR ACTION ] +.RB [ \-\^\-directories=\fIACTION\fP ] +.RB [ \-\^\-extended-regexp ] +.RB [ \-\^\-fixed-strings ] +.RB [ \-\^\-basic-regexp ] +.RB [ \-\^\-regexp=\fIPATTERN\fP ] +.RB [ \-\^\-file=\fIFILE\fP ] +.RB [ \-\^\-ignore-case ] +.RB [ \-\^\-word-regexp ] +.RB [ \-\^\-line-regexp ] +.RB [ \-\^\-line-regexp ] +.RB [ \-\^\-no-messages ] +.RB [ \-\^\-invert-match ] +.RB [ \-\^\-version ] +.RB [ \-\^\-help ] +.RB [ \-\^\-byte-offset ] +.RB [ \-\^\-line-number ] +.RB [ \-\^\-with-filename ] +.RB [ \-\^\-no-filename ] +.RB [ \-\^\-quiet ] +.RB [ \-\^\-silent ] +.RB [ \-\^\-text ] +.RB [ \-\^\-files-without-match ] +.RB [ \-\^\-files-with-matches ] +.RB [ \-\^\-count ] +.RB [ \-\^\-before-context=\fINUM\fP ] +.RB [ \-\^\-after-context=\fINUM\fP ] +.RB [ \-\^\-context [ =\fINUM\fP ]] +.RB [ \-\^\-binary ] +.RB [ \-\^\-unix-byte-offsets ] +.RB [ \-\^\-mmap ] +.RB [ \-\^\-null ] +.RB [ \-\^\-recursive ] +.RB [ \-\^\-decompress ] +.RI [ file .\|.\|.] .SH DESCRIPTION .PP .B grep @@ -41,83 +83,83 @@ There are three major variants of controlled by the following options. .PD 0 .TP -.B \-G, --basic-regexp +.BR \-G ", " \-\^\-basic-regexp Interpret .I pattern as a basic regular expression (see below). This is the default. .TP -.B \-E, --extended-regexp +.BR \-E ", " \-\^\-extended-regexp Interpret .I pattern as an extended regular expression (see below). .TP -.B \-F, --fixed-strings +.BR \-F ", " \-\^\-fixed-strings Interpret .I pattern as a list of fixed strings, separated by newlines, any of which is to be matched. -.LP +.PP In addition, two variant programs .B egrep and .B fgrep are available. .B egrep -is similar (but not identical) to -.BR "grep\ \-E" , -and is compatible with the historical Unix -.BR egrep . +is the same as +.BR "grep\ \-E" . .B fgrep is the same as .BR "grep\ \-F" . .B zgrep is the same as -.BR "grep\ \-Z" . +.BR "grep\ \-z" . .PD -.LP +.PP All variants of .B grep understand the following options: .PD 0 .TP -.BI \-A " NUM" ", --after-context=" NUM +.BI \-A " NUM" "\fR,\fP \-\^\-after-context=" NUM Print .I NUM lines of trailing context after matching lines. .TP -.BI \-B " NUM" ", --before-context=" NUM +.BI \-B " NUM" "\fR,\fP \-\^\-before-context=" NUM Print .I NUM lines of leading context before matching lines. .TP -.BI \-C ,\ --context"[=NUM]" -Print +.BI \-C " \fR[\fPNUM\fR]\fP" "\fR,\fP \-\^\-context\fR[\fP=" NUM\fR]\fP +Print .I NUM lines (default 2) of output context. .TP -.BI \- NUM \ -Same as --context=NUM lines of leading and trailing context. However, +.BI \- NUM +Same as +.BI \-\^\-context= NUM +lines of leading and trailing context. However, .B grep will never print any given line more than once. .TP -.B \-V, --version +.BR \-V ", " \-\^\-version Print the version number of .B grep to standard error. This version number should be included in all bug reports (see below). .TP -.B \-b, --byte-offset +.BR \-b ", " \-\^\-byte-offset Print the byte offset within the input file before each line of output. .TP -.B \-c, --count +.BR \-c ", " \-\^\-count Suppress normal output; instead print a count of matching lines for each input file. With the -.B \-v, --revert-match +.BR \-v ", " \-\^\-invert-match option (see below), count non-matching lines. .TP -.BI \-d " ACTION" ", --directories=" ACTION +.BI \-d " ACTION" "\fR,\fP \-\^\-directories=" ACTION If an input file is a directory, use .I ACTION to process it. By default, @@ -140,75 +182,78 @@ this is equivalent to the .B \-r option. .TP -.BI \-e " PATTERN" ", --regexp=" PATTERN +.BI \-e " PATTERN" "\fR,\fP \-\^\-regexp=" PATTERN Use .I PATTERN as the pattern; useful to protect patterns beginning with .BR \- . .TP -.BI \-f " FILE" ", --file=" FILE +.BI \-f " FILE" "\fR,\fP \-\^\-file=" FILE Obtain patterns from .IR FILE , one per line. The empty file contains zero patterns, and therfore matches nothing. .TP -.B \-h, --no-filename +.BR \-H ", " \-\^\-with-filename +Print the filename for each match. +.TP +.BR \-h ", " \-\^\-no-filename Suppress the prefixing of filenames on output when multiple files are searched. .TP -.B \-i, --ignore-case +.BR \-i ", " \-\^\-ignore-case Ignore case distinctions in both the .I pattern and the input files. .TP -.B \-L, --files-without-match +.BR \-L ", " \-\^\-files-without-match Suppress normal output; instead print the name of each input file from which no output would -normally have been printed. The scanning will stop +normally have been printed. The scanning will stop on the first match. .TP -.B \-l, --files-with-matches +.BR \-l ", " \-\^\-files-with-matches Suppress normal output; instead print the name of each input file from which output -would normally have been printed. The scanning will +would normally have been printed. The scanning will stop on the first match. .TP -.B \-n, --line-number +.BR \-n ", " \-\^\-line-number Prefix each line of output with the line number within its input file. .TP -.B \-q, --quiet, --silent -Quiet; suppress normal output. The scanning will stop +.BR \-q ", " \-\^\-quiet ", " \-\^\-silent +Quiet; suppress normal output. The scanning will stop on the first match. Also see the .B \-s or -.B --no-messages +.B \-\^\-no-messages option below. .TP -.B \-r, --recursive +.BR \-r ", " \-\^\-recursive Read all files under each directory, recursively; this is equivalent to the .B "\-d recurse" option. .TP -.B \-s, --no-messages +.BR \-s ", " \-\^\-no-messages Suppress error messages about nonexistent or unreadable files. -Portability note: unlike GNU +Portability note: unlike \s-1GNU\s0 .BR grep , -BSD +traditional .B grep -does not comply with POSIX.2, because BSD +did not conform to \s-1POSIX.2\s0, because traditional .B grep -lacks a +lacked a .B \-q option and its .B \-s -option behaves like GNU +option behaved like \s-1GNU\s0 .BR grep 's .B \-q option. -Shell scripts intended to be portable to BSD +Shell scripts intended to be portable to traditional .B grep should avoid both .B \-q @@ -216,7 +261,7 @@ and .B \-s and should redirect output to /dev/null instead. .TP -.B \-a, --text +.BR \-a ", " \-\^\-text Do not suppress output lines that contain binary data. Normally, if the first few bytes of a file indicate that the file contains binary data, @@ -227,10 +272,10 @@ This option causes to act as if the file is a text file, even if it would otherwise be treated as binary. .TP -.B \-v, --revert-match +.BR \-v ", " \-\^\-invert-match Invert the sense of matching, to select non-matching lines. .TP -.B \-w, --word-regexp +.BR \-w ", " \-\^\-word-regexp Select only those lines containing matches that form whole words. The test is that the matching substring must either be at the beginning of the line, or preceded by a non-word constituent @@ -238,14 +283,14 @@ character. Similarly, it must be either at the end of the line or followed by a non-word constituent character. Word-constituent characters are letters, digits, and the underscore. .TP -.B \-x, --line-regexp +.BR \-x ", " \-\^\-line-regexp Select only those matches that exactly match the whole line. .TP .B \-y Obsolete synonym for .BR \-i . .TP -.B \-U, --binary +.BR \-U ", " \-\^\-binary Treat the file(s) as binary. By default, under MS-DOS and MS-Windows, .BR grep guesses the file type by looking at the contents of the first 32KB @@ -261,10 +306,11 @@ work correctly). Specifying overrules this guesswork, causing all files to be read and passed to the matching mechanism verbatim; if the file is a text file with CR/LF pairs at the end of each line, this will cause some regular -expressions to fail. This option is only supported on MS-DOS and +expressions to fail. +This option has no effect on platforms other than MS-DOS and MS-Windows. .TP -.B \-u, --unix-byte-offsets +.BR \-u ", " \-\^\-unix-byte-offsets Report Unix-style byte offsets. This switch causes .B grep to report byte offsets as if the file were Unix-style text file, i.e. with @@ -272,13 +318,46 @@ CR characters stripped off. This will produce results identical to running .B grep on a Unix machine. This option has no effect unless .B \-b -option is also used; it is only supported on MS-DOS and MS-Windows. -.PD +option is also used; +it has no effect on platforms other than MS-DOS and MS-Windows. +.TP +.B \-\^\-mmap +If possible, use the +.BR mmap (2) +system call to read input, instead of +the default +.BR read (2) +system call. In some situations, +.B -\^-mmap +yields better performance. However, +.B -\^-mmap +can cause undefined behavior (including core dumps) +if an input file shrinks while +.B grep +is operating, or if an I/O error occurs. +.TP +.BR \-Z ", " \-\^\-null +Output a zero byte (the \s-1ASCII\s0 +.B NUL +character) instead of the character that normally follows a file name. +For example, +.B "grep \-lZ" +outputs a zero byte after each file name instead of the usual newline. +This option makes the output unambiguous, even in the presence of file +names containing unusual characters like newlines. This option can be +used with commands like +.BR "find \-print0" , +.BR "perl \-0" , +.BR "sort \-z" , +and +.B "xargs \-0" +to process arbitrary file names, +even those that contain newline characters. .LP Following option is only available if compiled with zlib(3) library: .PD 0 .TP -.B \-Z, --decompress +.B \-z, --decompress Decompress the input data before searching. .PD .SH "REGULAR EXPRESSIONS" @@ -289,8 +368,8 @@ expressions, by using various operators to combine smaller expressions. .PP .B grep understands two different versions of regular expression syntax: -``basic'' and ``extended.'' In -.RB "GNU\ " grep , +\*(lqbasic\*(rq and \*(lqextended.\*(rq In +.RB "\s-1GNU\s0\ " grep , there is no difference in available functionality using either syntax. In other implementations, basic regular expressions are less powerful. The following description applies to extended regular expressions; @@ -402,11 +481,6 @@ The preceding item is matched .I n or more times. .TP -.BI {, m } -The preceding item is optional and is matched at most -.I m -times. -.TP .BI { n , m } The preceding item is matched at least .I n @@ -456,12 +530,35 @@ versions and .BR \e) . .PP -In +Traditional +.B egrep +did not support the +.B { +metacharacter, and some +.B egrep +implementations support +.B \e{ +instead, so portable scripts should avoid +.B { +in +.B egrep +patterns and should use +.B [{] +to match a literal +.BR { . +.PP +\s-1GNU\s0 .B egrep -the metacharacter +attempts to support traditional usage by assuming that .B { -loses its special meaning; instead use -.BR \e{ . +is not special if it would be the start of an invalid interval +specification. For example, the shell command +.B "egrep '{1'" +searches for the two-character string +.B {1 +instead of reporting a syntax error in the regular expression. +\s-1POSIX.2\s0 allows this behavior as an extension, but portable scripts +should avoid it. .SH ENVIRONMENT The environment variable .B GREP_OPTIONS @@ -483,7 +580,8 @@ other system errors. .PP Email bug reports to .BR bug-gnu-utils@gnu.org . -Be sure to include the word ``grep'' somewhere in the ``Subject:'' field. +Be sure to include the word \*(lqgrep\*(rq somewhere in the +\*(lqSubject:\*(rq field. .PP Large repetition counts in the .BI { m , n } @@ -495,3 +593,5 @@ and space, and may cause to run out of memory. .PP Backreferences are very slow, and may require exponential time. +.\" Work around problems with some troff -man implementations. +.br diff --git a/gnu/usr.bin/grep/grep.c b/gnu/usr.bin/grep/grep.c index 569f43a..79c7c48 100644 --- a/gnu/usr.bin/grep/grep.c +++ b/gnu/usr.bin/grep/grep.c @@ -58,6 +58,17 @@ static int show_help; /* If non-zero, print the version on standard output and exit. */ static int show_version; +/* If nonzero, use mmap if possible. */ +static int mmap_option; + +/* Short options. */ +static char const short_options[] = +#if HAVE_LIBZ > 0 +"0123456789A:B:C::EFGHUVX:abcd:e:f:hiLlnqrsuvwxyZz"; +#else +"0123456789A:B:C::EFGHUVX:abcd:e:f:hiLlnqrsuvwxyZz"; +#endif + /* Long options equivalences. */ static struct option long_options[] = { @@ -78,18 +89,23 @@ static struct option long_options[] = {"ignore-case", no_argument, NULL, 'i'}, {"line-number", no_argument, NULL, 'n'}, {"line-regexp", no_argument, NULL, 'x'}, + {"mmap", no_argument, &mmap_option, 1}, {"no-filename", no_argument, NULL, 'h'}, {"no-messages", no_argument, NULL, 's'}, +#if HAVE_LIBZ > 0 + {"null", no_argument, NULL, /*'Z'*/ 1}, +#else + {"null", no_argument, NULL, 'Z'}, +#endif + {"null-data", no_argument, NULL, 'z'}, {"quiet", no_argument, NULL, 'q'}, {"recursive", no_argument, NULL, 'r'}, {"regexp", required_argument, NULL, 'e'}, - {"revert-match", no_argument, NULL, 'v'}, + {"invert-match", no_argument, NULL, 'v'}, {"silent", no_argument, NULL, 'q'}, {"text", no_argument, NULL, 'a'}, -#if O_BINARY {"binary", no_argument, NULL, 'U'}, {"unix-byte-offsets", no_argument, NULL, 'u'}, -#endif {"version", no_argument, NULL, 'V'}, {"with-filename", no_argument, NULL, 'H'}, {"word-regexp", no_argument, NULL, 'w'}, @@ -100,10 +116,13 @@ static struct option long_options[] = }; /* Define flags declared in grep.h. */ +/* I do not know why we need this decl, while if you build GNU grep 2.4 by + hand you don't... */ char const *matcher; int match_icase; int match_words; int match_lines; +unsigned char eolbyte; /* For error messages. */ static char *prog; @@ -121,7 +140,10 @@ static enum static int ck_atoi PARAMS ((char const *, int *)); static void usage PARAMS ((int)) __attribute__((noreturn)); static void error PARAMS ((const char *, int)); -static int setmatcher PARAMS ((char const *)); +static void setmatcher PARAMS ((char const *)); +static int install_matcher PARAMS ((char const *)); +static int prepend_args PARAMS ((char const *, char *, char **)); +static void prepend_default_options PARAMS ((char const *, int *, char ***)); static char *page_alloc PARAMS ((size_t, char **)); static int reset PARAMS ((int, char const *, struct stats *)); static int fillbuf PARAMS ((size_t, struct stats *)); @@ -221,14 +243,15 @@ static char *ubuffer; /* Unaligned base of buffer. */ static char *buffer; /* Base of buffer. */ static size_t bufsalloc; /* Allocated size of buffer save region. */ static size_t bufalloc; /* Total buffer size. */ +#define PREFERRED_SAVE_FACTOR 5 /* Preferred value of bufalloc / bufsalloc. */ static int bufdesc; /* File descriptor. */ static char *bufbeg; /* Beginning of user-visible stuff. */ static char *buflim; /* Limit of user-visible stuff. */ static size_t pagesize; /* alignment of memory pages */ +static off_t bufoffset; /* Read offset; defined on regular files. */ #if defined(HAVE_MMAP) -static int bufmapped; /* True for ordinary files. */ -static off_t bufoffset; /* What read() normally remembers. */ +static int bufmapped; /* True if buffer is memory-mapped. */ static off_t initial_bufoffset; /* Initial value of bufoffset. */ #endif @@ -245,32 +268,26 @@ static int Zflag; /* uncompress before searching. */ ? (val) \ : (val) + ((alignment) - (size_t) (val) % (alignment))) -/* Return the address of a new page-aligned buffer of size SIZE. Set - *UP to the newly allocated (but possibly unaligned) buffer used to - *build the aligned buffer. To free the buffer, free (*UP). */ +/* Return the address of a page-aligned buffer of size SIZE, + reallocating it from *UP. Set *UP to the newly allocated (but + possibly unaligned) buffer used to build the aligned buffer. To + free the buffer, free (*UP). */ static char * page_alloc (size, up) size_t size; char **up; { - /* HAVE_WORKING_VALLOC means that valloc is properly declared, and - you can free the result of valloc. This symbol is not (yet) - autoconfigured. It can be useful to define HAVE_WORKING_VALLOC - while debugging, since some debugging memory allocators might - catch more bugs if this symbol is enabled. */ -#if HAVE_WORKING_VALLOC - *up = valloc (size); - return *up; -#else size_t asize = size + pagesize - 1; if (size <= asize) { - *up = malloc (asize); - if (*up) - return ALIGN_TO (*up, pagesize); + char *p = *up ? realloc (*up, asize) : malloc (asize); + if (p) + { + *up = p; + return ALIGN_TO (p, pagesize); + } } return NULL; -#endif } /* Reset the buffer for a new file, returning zero if we should skip it. @@ -281,7 +298,9 @@ reset (fd, file, stats) char const *file; struct stats *stats; { - if (pagesize == 0) + if (pagesize) + bufsalloc = ALIGN_TO (bufalloc / PREFERRED_SAVE_FACTOR, pagesize); + else { size_t ubufsalloc; pagesize = getpagesize (); @@ -293,162 +312,212 @@ reset (fd, file, stats) ubufsalloc = BUFSALLOC; #endif bufsalloc = ALIGN_TO (ubufsalloc, pagesize); - bufalloc = 5 * bufsalloc; + bufalloc = PREFERRED_SAVE_FACTOR * bufsalloc; /* The 1 byte of overflow is a kludge for dfaexec(), which inserts a sentinel newline at the end of the buffer being searched. There's gotta be a better way... */ if (bufsalloc < ubufsalloc - || bufalloc / 5 != bufsalloc || bufalloc + 1 < bufalloc + || bufalloc / PREFERRED_SAVE_FACTOR != bufsalloc + || bufalloc + 1 < bufalloc || ! (buffer = page_alloc (bufalloc + 1, &ubuffer))) fatal (_("memory exhausted"), 0); - bufbeg = buffer; - buflim = buffer; } #if HAVE_LIBZ > 0 - if (Zflag) { + if (Zflag) + { gzbufdesc = gzdopen(fd, "r"); if (gzbufdesc == NULL) fatal(_("memory exhausted"), 0); - } + } #endif + + buflim = buffer; bufdesc = fd; - if ( -#if defined(HAVE_MMAP) - 1 -#else - directories != READ_DIRECTORIES -#endif - ) - if (fstat (fd, &stats->stat) != 0) - { - error ("fstat", errno); - return 0; - } + if (fstat (fd, &stats->stat) != 0) + { + error ("fstat", errno); + return 0; + } if (directories == SKIP_DIRECTORIES && S_ISDIR (stats->stat.st_mode)) return 0; -#if defined(HAVE_MMAP) if ( #if HAVE_LIBZ > 0 Zflag || #endif - !S_ISREG (stats->stat.st_mode)) - bufmapped = 0; - else + S_ISREG (stats->stat.st_mode)) { - bufmapped = 1; - bufoffset = initial_bufoffset = file ? 0 : lseek (fd, 0, 1); + if (file) + bufoffset = 0; + else + { + bufoffset = lseek (fd, 0, SEEK_CUR); + if (bufoffset < 0) + { + error ("lseek", errno); + return 0; + } + } +#ifdef HAVE_MMAP + initial_bufoffset = bufoffset; + bufmapped = mmap_option && bufoffset % pagesize == 0; +#endif } + else + { +#ifdef HAVE_MMAP + bufmapped = 0; #endif + } return 1; } /* Read new stuff into the buffer, saving the specified amount of old stuff. When we're done, 'bufbeg' points to the beginning of the buffer contents, and 'buflim' - points just after the end. Return count of new stuff. */ + points just after the end. Return zero if there's an error. */ static int fillbuf (save, stats) size_t save; struct stats *stats; { - int cc; -#if defined(HAVE_MMAP) - caddr_t maddr; -#endif + size_t fillsize = 0; + int cc = 1; + size_t readsize; - if (save > bufsalloc) - { - char *nubuffer; - char *nbuffer; - - while (save > bufsalloc) - bufsalloc *= 2; - bufalloc = 5 * bufsalloc; - if (bufalloc / 5 != bufsalloc || bufalloc + 1 < bufalloc - || ! (nbuffer = page_alloc (bufalloc + 1, &nubuffer))) - fatal (_("memory exhausted"), 0); + /* Offset from start of unaligned buffer to start of old stuff + that we want to save. */ + size_t saved_offset = buflim - ubuffer - save; - bufbeg = nbuffer + bufsalloc - save; - memcpy (bufbeg, buflim - save, save); - free (ubuffer); - ubuffer = nubuffer; - buffer = nbuffer; - } - else + if (bufsalloc < save) { - bufbeg = buffer + bufsalloc - save; - memcpy (bufbeg, buflim - save, save); + size_t aligned_save = ALIGN_TO (save, pagesize); + size_t maxalloc = (size_t) -1; + size_t newalloc; + + if (S_ISREG (stats->stat.st_mode)) + { + /* Calculate an upper bound on how much memory we should allocate. + We can't use ALIGN_TO here, since off_t might be longer than + size_t. Watch out for arithmetic overflow. */ + off_t to_be_read = stats->stat.st_size - bufoffset; + size_t slop = to_be_read % pagesize; + off_t aligned_to_be_read = to_be_read + (slop ? pagesize - slop : 0); + off_t maxalloc_off = aligned_save + aligned_to_be_read; + if (0 <= maxalloc_off && maxalloc_off == (size_t) maxalloc_off) + maxalloc = maxalloc_off; + } + + /* Grow bufsalloc until it is at least as great as `save'; but + if there is an overflow, just grow it to the next page boundary. */ + while (bufsalloc < save) + if (bufsalloc < bufsalloc * 2) + bufsalloc *= 2; + else + { + bufsalloc = aligned_save; + break; + } + + /* Grow the buffer size to be PREFERRED_SAVE_FACTOR times + bufsalloc.... */ + newalloc = PREFERRED_SAVE_FACTOR * bufsalloc; + if (maxalloc < newalloc) + { + /* ... except don't grow it more than a pagesize past the + file size, as that might cause unnecessary memory + exhaustion if the file is large. */ + newalloc = maxalloc; + bufsalloc = aligned_save; + } + + /* Check that the above calculations made progress, which might + not occur if there is arithmetic overflow. If there's no + progress, or if the new buffer size is larger than the old + and buffer reallocation fails, report memory exhaustion. */ + if (bufsalloc < save || newalloc < save + || (newalloc == save && newalloc != maxalloc) + || (bufalloc < newalloc + && ! (buffer + = page_alloc ((bufalloc = newalloc) + 1, &ubuffer)))) + fatal (_("memory exhausted"), 0); } + bufbeg = buffer + bufsalloc - save; + memmove (bufbeg, ubuffer + saved_offset, save); + readsize = bufalloc - bufsalloc; + #if defined(HAVE_MMAP) - if (bufmapped && bufoffset % pagesize == 0 - && stats->stat.st_size - bufoffset >= bufalloc - bufsalloc) + if (bufmapped) { - maddr = buffer + bufsalloc; - maddr = mmap (maddr, bufalloc - bufsalloc, PROT_READ | PROT_WRITE, - MAP_PRIVATE | MAP_FIXED, bufdesc, bufoffset); - if (maddr == (caddr_t) -1) + size_t mmapsize = readsize; + + /* Don't mmap past the end of the file; some hosts don't allow this. + Use `read' on the last page. */ + if (stats->stat.st_size - bufoffset < mmapsize) { - /* This used to issue a warning, but on some hosts - (e.g. Solaris 2.5) mmap can fail merely because some - other process has an advisory read lock on the file. - There's no point alarming the user about this misfeature. */ -#if 0 - fprintf (stderr, _("%s: warning: %s: %s\n"), prog, filename, - strerror (errno)); -#endif - goto tryread; + mmapsize = stats->stat.st_size - bufoffset; + mmapsize -= mmapsize % pagesize; } -#if 0 - /* You might thing this (or MADV_WILLNEED) would help, - but it doesn't, at least not on a Sun running 4.1. - In fact, it actually slows us down about 30%! */ - madvise (maddr, bufalloc - bufsalloc, MADV_SEQUENTIAL); -#endif - cc = bufalloc - bufsalloc; - bufoffset += cc; - } - else - { - tryread: - /* We come here when we're not going to use mmap() any more. - Note that we need to synchronize the file offset the - first time through. */ - if (bufmapped) + + if (mmapsize + && (mmap ((caddr_t) (buffer + bufsalloc), mmapsize, + PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_FIXED, + bufdesc, bufoffset) + != (caddr_t) -1)) { - bufmapped = 0; - if (bufoffset != initial_bufoffset) - lseek (bufdesc, bufoffset, 0); + /* Do not bother to use madvise with MADV_SEQUENTIAL or + MADV_WILLNEED on the mmapped memory. One might think it + would help, but it slows us down about 30% on SunOS 4.1. */ + fillsize = mmapsize; } -#if HAVE_LIBZ > 0 - if (Zflag) - cc = gzread (gzbufdesc, buffer + bufsalloc, bufalloc - bufsalloc); else -#endif - cc = read (bufdesc, buffer + bufsalloc, bufalloc - bufsalloc); + { + /* Stop using mmap on this file. Synchronize the file + offset. Do not warn about mmap failures. On some hosts + (e.g. Solaris 2.5) mmap can fail merely because some + other process has an advisory read lock on the file. + There's no point alarming the user about this misfeature. */ + bufmapped = 0; + if (bufoffset != initial_bufoffset + && lseek (bufdesc, bufoffset, SEEK_SET) < 0) + { + error ("lseek", errno); + cc = 0; + } + } } -#else +#endif /*HAVE_MMAP*/ + + if (! fillsize) + { + ssize_t bytesread; + do #if HAVE_LIBZ > 0 - if (Zflag) - cc = gzread (gzbufdesc, buffer + bufsalloc, bufalloc - bufsalloc); - else + if (Zflag) + bytesread = gzread (gzbufdesc, buffer + bufsalloc, readsize); + else #endif - cc = read (bufdesc, buffer + bufsalloc, bufalloc - bufsalloc); -#endif /*HAVE_MMAP*/ + bytesread = read (bufdesc, buffer + bufsalloc, readsize); + while (bytesread < 0 && errno == EINTR); + if (bytesread < 0) + cc = 0; + else + fillsize = bytesread; + } + + bufoffset += fillsize; #if O_BINARY - if (cc > 0) - cc = undossify_input (buffer + bufsalloc, cc); + if (fillsize) + fillsize = undossify_input (buffer + bufsalloc, fillsize); #endif - if (cc > 0) - buflim = buffer + bufsalloc + cc; - else - buflim = buffer + bufsalloc; + buflim = buffer + bufsalloc + fillsize; return cc; } /* Flags controlling the style of output. */ static int always_text; /* Assume the input is always text. */ +static int filename_mask; /* If zero, output nulls after filenames. */ static int out_quiet; /* Suppress all normal output. */ static int out_invert; /* Print nonmatching stuff. */ static int out_file; /* Print filenames. */ @@ -480,11 +549,9 @@ nlscan (lim) char *lim; { char *beg; - - for (beg = lastnl; beg < lim; ++beg) - if (*beg == '\n') - ++totalnl; - lastnl = beg; + for (beg = lastnl; (beg = memchr (beg, eolbyte, lim - beg)); beg++) + totalnl++; + lastnl = lim; } static void @@ -513,7 +580,7 @@ prline (beg, lim, sep) int sep; { if (out_file) - printf ("%s%c", filename, sep); + printf ("%s%c", filename, sep & filename_mask); if (out_line) { nlscan (beg); @@ -546,7 +613,7 @@ prpending (lim) while (pending > 0 && lastout < lim) { --pending; - if ((nl = memchr (lastout, '\n', lim - lastout)) != 0) + if ((nl = memchr (lastout, eolbyte, lim - lastout)) != 0) ++nl; else nl = lim; @@ -564,6 +631,7 @@ prtext (beg, lim, nlinesp) { static int used; /* avoid printing "--" before any output */ char *bp, *p, *nl; + char eol = eolbyte; int i, n; if (!out_quiet && pending > 0) @@ -580,7 +648,7 @@ prtext (beg, lim, nlinesp) if (p > bp) do --p; - while (p > bp && p[-1] != '\n'); + while (p > bp && p[-1] != eol); /* We only print the "--" separator if our output is discontiguous from the last output in the file. */ @@ -589,7 +657,7 @@ prtext (beg, lim, nlinesp) while (p < beg) { - nl = memchr (p, '\n', beg - p); + nl = memchr (p, eol, beg - p); prline (p, nl + 1, '-'); p = nl + 1; } @@ -600,7 +668,7 @@ prtext (beg, lim, nlinesp) /* Caller wants a line count. */ for (n = 0; p < lim; ++n) { - if ((nl = memchr (p, '\n', lim - p)) != 0) + if ((nl = memchr (p, eol, lim - p)) != 0) ++nl; else nl = lim; @@ -614,7 +682,7 @@ prtext (beg, lim, nlinesp) if (!out_quiet) prline (beg, lim, ':'); - pending = out_after; + pending = out_quiet ? 0 : out_after; used = 1; } @@ -629,13 +697,14 @@ grepbuf (beg, lim) int nlines, n; register char *p, *b; char *endp; + char eol = eolbyte; nlines = 0; p = beg; while ((b = (*execute)(p, lim - p, &endp)) != 0) { /* Avoid matching the empty line at the end of the buffer. */ - if (b == lim && ((b > beg && b[-1] == '\n') || b == beg)) + if (b == lim && ((b > beg && b[-1] == eol) || b == beg)) break; if (!out_invert) { @@ -672,6 +741,7 @@ grep (fd, file, stats) int not_text; size_t residue, save; char *beg, *lim; + char eol = eolbyte; if (!reset (fd, file, stats)) return 0; @@ -700,7 +770,7 @@ grep (fd, file, stats) residue = 0; save = 0; - if (fillbuf (save, stats) < 0) + if (! fillbuf (save, stats)) { if (! (is_EISDIR (errno, file) && suppress_errors)) error (filename, errno); @@ -708,7 +778,7 @@ grep (fd, file, stats) } not_text = (! (always_text | out_quiet) - && memchr (bufbeg, '\0', buflim - bufbeg)); + && memchr (bufbeg, eol ? '\0' : '\200', buflim - bufbeg)); done_on_match += not_text; out_quiet += not_text; @@ -720,7 +790,7 @@ grep (fd, file, stats) if (buflim - bufbeg == save) break; beg = bufbeg + save - residue; - for (lim = buflim; lim > beg && lim[-1] != '\n'; --lim) + for (lim = buflim; lim > beg && lim[-1] != eol; --lim) ; residue = buflim - lim; if (beg < lim) @@ -738,7 +808,7 @@ grep (fd, file, stats) ++i; do --beg; - while (beg > bufbeg && beg[-1] != '\n'); + while (beg > bufbeg && beg[-1] != eol); } if (beg != lastout) lastout = 0; @@ -746,7 +816,7 @@ grep (fd, file, stats) totalcc += buflim - bufbeg - save; if (out_line) nlscan (beg); - if (fillbuf (save, stats) < 0) + if (! fillbuf (save, stats)) { if (! (is_EISDIR (errno, file) && suppress_errors)) error (filename, errno); @@ -784,7 +854,8 @@ grepfile (file, stats) } else { - desc = open (file, O_RDONLY); + while ((desc = open (file, O_RDONLY)) < 0 && errno == EINTR) + continue; if (desc < 0) { @@ -843,30 +914,26 @@ grepfile (file, stats) if (count_matches) { if (out_file) - printf ("%s:", filename); + printf ("%s%c", filename, ':' & filename_mask); printf ("%d\n", count); } - if (count) - { - status = 0; - if (list_files == 1) - printf ("%s\n", filename); - } - else - { - status = 1; - if (list_files == -1) - printf ("%s\n", filename); - } + status = !count; + if (list_files == 1 - 2 * status) + printf ("%s%c", filename, '\n' & filename_mask); #if HAVE_LIBZ > 0 if (Zflag) gzclose(gzbufdesc); else #endif - if (file && close (desc) != 0) - error (file, errno); + if (file) + while (close (desc) != 0) + if (errno != EINTR) + { + error (file, errno); + break; + } } return status; @@ -882,8 +949,8 @@ grepdir (dir, stats) char *name_space; for (ancestor = stats; (ancestor = ancestor->parent) != 0; ) - if (! ((ancestor->stat.st_ino ^ stats->stat.st_ino) - | (ancestor->stat.st_dev ^ stats->stat.st_dev))) + if (ancestor->stat.st_ino == stats->stat.st_ino + && ancestor->stat.st_dev == stats->stat.st_dev) { if (!suppress_errors) fprintf (stderr, _("%s: warning: %s: %s\n"), prog, dir, @@ -946,24 +1013,29 @@ int status; printf (_("Usage: %s [OPTION]... PATTERN [FILE] ...\n"), prog); printf (_("\ Search for PATTERN in each FILE or standard input.\n\ +Example: %s -i 'hello.*world' menu.h main.c\n\ \n\ -Regexp selection and interpretation:\n\ +Regexp selection and interpretation:\n"), prog); + printf (_("\ -E, --extended-regexp PATTERN is an extended regular expression\n\ - -F, --fixed-regexp PATTERN is a fixed string separated by newlines\n\ - -G, --basic-regexp PATTERN is a basic regular expression\n\ + -F, --fixed-strings PATTERN is a set of newline-separated strings\n\ + -G, --basic-regexp PATTERN is a basic regular expression\n")); + printf (_("\ -e, --regexp=PATTERN use PATTERN as a regular expression\n\ -f, --file=FILE obtain PATTERN from FILE\n\ -i, --ignore-case ignore case distinctions\n\ -w, --word-regexp force PATTERN to match only whole words\n\ - -x, --line-regexp force PATTERN to match only whole lines\n")); + -x, --line-regexp force PATTERN to match only whole lines\n\ + -z, --null-data a data line ends in 0 byte, not newline\n")); printf (_("\ \n\ Miscellaneous:\n\ -s, --no-messages suppress error messages\n\ - -v, --revert-match select non-matching lines\n\ + -v, --invert-match select non-matching lines\n\ -V, --version print version information and exit\n\ + --help display this help and exit\n\ -Z, --decompress decompress input before searching (HAVE_LIBZ=1)\n\ - --help display this help and exit\n")); + --mmap use memory-mapped input if possible\n")); printf (_("\ \n\ Output control:\n\ @@ -978,31 +1050,42 @@ Output control:\n\ -r, --recursive equivalent to --directories=recurse.\n\ -L, --files-without-match only print FILE names containing no match\n\ -l, --files-with-matches only print FILE names containing matches\n\ - -c, --count only print a count of matching lines per FILE\n")); + -c, --count only print a count of matching lines per FILE\n\ + --null print 0 byte after FILE name\n")); printf (_("\ \n\ Context control:\n\ -B, --before-context=NUM print NUM lines of leading context\n\ -A, --after-context=NUM print NUM lines of trailing context\n\ -C, --context[=NUM] print NUM (default 2) lines of output context\n\ - unless overriden by -A or -B\n\ + unless overridden by -A or -B\n\ -NUM same as --context=NUM\n\ -U, --binary do not strip CR characters at EOL (MSDOS)\n\ -u, --unix-byte-offsets report offsets as if CRs were not there (MSDOS)\n\ \n\ -If no -[GEF], then `egrep' assumes -E, `fgrep' -F, else -G.\n\ -With no FILE, or when FILE is -, read standard input. If less than\n\ -two FILEs given, assume -h. Exit with 0 if matches, with 1 if none.\n\ -Exit with 2 if syntax errors or system errors.\n")); +`egrep' means `grep -E'. `fgrep' means `grep -F'.\n\ +With no FILE, or when FILE is -, read standard input. If less than\n\ +two FILEs given, assume -h. Exit status is 0 if match, 1 if no match,\n\ +and 2 if trouble.\n")); printf (_("\nReport bugs to <bug-gnu-utils@gnu.org>.\n")); } exit (status); } +/* Set the matcher to M, reporting any conflicts. */ +static void +setmatcher (m) + char const *m; +{ + if (matcher && strcmp (matcher, m) != 0) + fatal (_("conflicting matchers specified"), 0); + matcher = m; +} + /* Go through the matchers vector and look for the specified matcher. If we find it, install it in compile and execute, and return 1. */ static int -setmatcher (name) +install_matcher (name) char const *name; { int i; @@ -1158,7 +1241,8 @@ main (argc, argv) keys = NULL; keycc = 0; with_filenames = 0; - matcher = NULL; + eolbyte = '\n'; + filename_mask = ~0; /* The value -1 means to use DEFAULT_CONTEXT. */ out_after = out_before = -1; @@ -1179,15 +1263,8 @@ main (argc, argv) prepend_default_options (getenv ("GREP_OPTIONS"), &argc, &argv); - while ((opt = getopt_long (argc, argv, -#if O_BINARY - "0123456789A:B:C::EFGHVX:abcd:e:f:hiLlnqrsvwxyUu", -#elif HAVE_LIBZ > 0 - "0123456789A:B:C::EFGHRVX:Zabcd:e:f:hiLlnqrsvwxy", -#else - "0123456789A:B:C::EFGHRVX:abcd:e:f:hiLlnqrsvwxy", -#endif - long_options, NULL)) != EOF) + while ((opt = getopt_long (argc, argv, short_options, long_options, NULL)) + != -1) switch (opt) { case '0': @@ -1229,44 +1306,33 @@ main (argc, argv) default_context = 2; break; case 'E': - if (matcher && strcmp (matcher, "posix-egrep") != 0) - fatal (_("you may specify only one of -E, -F, or -G"), 0); - matcher = "posix-egrep"; + setmatcher ("egrep"); break; case 'F': - if (matcher && strcmp(matcher, "fgrep") != 0) - fatal(_("you may specify only one of -E, -F, or -G"), 0);; - matcher = "fgrep"; + setmatcher ("fgrep"); break; case 'G': - if (matcher && strcmp (matcher, "grep") != 0) - fatal (_("you may specify only one of -E, -F, or -G"), 0); - matcher = "grep"; + setmatcher ("grep"); break; case 'H': with_filenames = 1; break; -#if O_BINARY case 'U': +#if O_BINARY dos_use_file_type = DOS_BINARY; +#endif break; case 'u': +#if O_BINARY dos_report_unix_offset = 1; - break; #endif + break; case 'V': show_version = 1; break; case 'X': - if (matcher) - fatal (_("matcher already specified"), 0); - matcher = optarg; + setmatcher (optarg); break; -#if HAVE_LIBZ > 0 - case 'Z': - Zflag = 1; - break; -#endif case 'a': always_text = 1; break; @@ -1357,6 +1423,16 @@ main (argc, argv) case 'x': match_lines = 1; break; + case 'Z': +#if HAVE_LIBZ > 0 + Zflag = 1; +#else + filename_mask = 0; +#endif + break; + case 'z': + eolbyte = '\0'; + break; case 0: /* long options */ break; @@ -1370,9 +1446,12 @@ main (argc, argv) if (out_before < 0) out_before = default_context; + if (! matcher) + matcher = prog; + if (show_version) { - printf (_("grep (GNU grep) %s\n"), VERSION); + printf (_("%s (GNU grep) %s\n"), matcher, VERSION); printf ("\n"); printf (_("\ Copyright (C) 1988, 1992-1998, 1999 Free Software Foundation, Inc.\n")); @@ -1404,10 +1483,7 @@ warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.\n")) else usage (2); - if (! matcher) - matcher = prog; - - if (!setmatcher (matcher) && !setmatcher ("default")) + if (!install_matcher (matcher) && !install_matcher ("default")) abort (); (*compile)(keys, keycc); diff --git a/gnu/usr.bin/grep/grep.h b/gnu/usr.bin/grep/grep.h index e12cff5..264ad20 100644 --- a/gnu/usr.bin/grep/grep.h +++ b/gnu/usr.bin/grep/grep.h @@ -37,14 +37,12 @@ extern struct matcher char *(*execute) PARAMS ((char *, size_t, char **)); } matchers[]; -/* Exported from grep.c. */ -extern char const *matcher; - /* Exported from fgrepmat.c, egrepmat.c, grepmat.c. */ -extern char const default_matcher[]; +extern char const *matcher; /* The following flags are exported from grep for the matchers to look at. */ extern int match_icase; /* -i */ extern int match_words; /* -w */ extern int match_lines; /* -x */ +extern unsigned char eolbyte; /* -z */ diff --git a/gnu/usr.bin/grep/search.c b/gnu/usr.bin/grep/search.c index a3ca56c..4e3244a 100644 --- a/gnu/usr.bin/grep/search.c +++ b/gnu/usr.bin/grep/search.c @@ -48,7 +48,6 @@ struct matcher matchers[] = { { "default", Gcompile, EGexecute }, { "grep", Gcompile, EGexecute }, { "egrep", Ecompile, EGexecute }, - { "posix-egrep", Ecompile, EGexecute }, { "awk", Ecompile, EGexecute }, { "fgrep", Fcompile, Fexecute }, { 0, 0, 0 }, @@ -61,7 +60,7 @@ struct matcher matchers[] = { static struct dfa dfa; /* Regex compiled regexp. */ -static struct re_pattern_buffer regex; +static struct re_pattern_buffer regexbuf; /* KWset compiled pattern. For Ecompile and Gcompile, we compile a list of strings, at least one of which is known to occur in @@ -140,9 +139,9 @@ Gcompile(pattern, size) const char *err; re_set_syntax(RE_SYNTAX_GREP | RE_HAT_LISTS_NOT_NEWLINE); - dfasyntax(RE_SYNTAX_GREP | RE_HAT_LISTS_NOT_NEWLINE, match_icase); + dfasyntax(RE_SYNTAX_GREP | RE_HAT_LISTS_NOT_NEWLINE, match_icase, eolbyte); - if ((err = re_compile_pattern(pattern, size, ®ex)) != 0) + if ((err = re_compile_pattern(pattern, size, ®exbuf)) != 0) fatal(err, 0); /* In the match_words and match_lines cases, we use a different pattern @@ -155,7 +154,8 @@ Gcompile(pattern, size) (^|[^A-Za-z_])(userpattern)([^A-Za-z_]|$). In the whole-line case, we use the pattern: ^(userpattern)$. - BUG: Using [A-Za-z_] is locale-dependent! */ + BUG: Using [A-Za-z_] is locale-dependent! + So will use [:alnum:] */ char *n = malloc(size + 50); int i = 0; @@ -165,14 +165,14 @@ Gcompile(pattern, size) if (match_lines) strcpy(n, "^\\("); if (match_words) - strcpy(n, "\\(^\\|[^0-9A-Za-z_]\\)\\("); + strcpy(n, "\\(^\\|[^[:alnum:]_]\\)\\("); i = strlen(n); memcpy(n + i, pattern, size); i += size; if (match_words) - strcpy(n + i, "\\)\\([^0-9A-Za-z_]\\|$\\)"); + strcpy(n + i, "\\)\\([^[:alnum:]_]\\|$\\)"); if (match_lines) strcpy(n + i, "\\)$"); @@ -192,23 +192,18 @@ Ecompile(pattern, size) { const char *err; - if (strcmp(matcher, "posix-egrep") == 0) - { - re_set_syntax(RE_SYNTAX_POSIX_EGREP); - dfasyntax(RE_SYNTAX_POSIX_EGREP, match_icase); - } - else if (strcmp(matcher, "awk") == 0) + if (strcmp(matcher, "awk") == 0) { re_set_syntax(RE_SYNTAX_AWK); - dfasyntax(RE_SYNTAX_AWK, match_icase); + dfasyntax(RE_SYNTAX_AWK, match_icase, eolbyte); } else { - re_set_syntax(RE_SYNTAX_EGREP); - dfasyntax(RE_SYNTAX_EGREP, match_icase); + re_set_syntax (RE_SYNTAX_POSIX_EGREP); + dfasyntax (RE_SYNTAX_POSIX_EGREP, match_icase, eolbyte); } - if ((err = re_compile_pattern(pattern, size, ®ex)) != 0) + if ((err = re_compile_pattern(pattern, size, ®exbuf)) != 0) fatal(err, 0); /* In the match_words and match_lines cases, we use a different pattern @@ -221,7 +216,8 @@ Ecompile(pattern, size) (^|[^A-Za-z_])(userpattern)([^A-Za-z_]|$). In the whole-line case, we use the pattern: ^(userpattern)$. - BUG: Using [A-Za-z_] is locale-dependent! */ + BUG: Using [A-Za-z_] is locale-dependent! + so will use the char class */ char *n = malloc(size + 50); int i = 0; @@ -231,14 +227,14 @@ Ecompile(pattern, size) if (match_lines) strcpy(n, "^("); if (match_words) - strcpy(n, "(^|[^0-9A-Za-z_])("); + strcpy(n, "(^|[^[:alnum:]_])("); i = strlen(n); memcpy(n + i, pattern, size); i += size; if (match_words) - strcpy(n + i, ")([^0-9A-Za-z_]|$)"); + strcpy(n + i, ")([^[:alnum:]_]|$)"); if (match_lines) strcpy(n + i, ")$"); @@ -258,6 +254,7 @@ EGexecute(buf, size, endp) char **endp; { register char *buflim, *beg, *end, save; + char eol = eolbyte; int backref, start, len; struct kwsmatch kwsm; static struct re_registers regs; /* This is static on account of a BRAIN-DEAD @@ -275,10 +272,10 @@ EGexecute(buf, size, endp) goto failure; /* Narrow down to the line containing the candidate, and run it through DFA. */ - end = memchr(beg, '\n', buflim - beg); + end = memchr(beg, eol, buflim - beg); if (!end) end = buflim; - while (beg > buf && beg[-1] != '\n') + while (beg > buf && beg[-1] != eol) --beg; save = *end; if (kwsm.index < lastexact) @@ -302,10 +299,10 @@ EGexecute(buf, size, endp) if (!beg) goto failure; /* Narrow down to the line we've found. */ - end = memchr(beg, '\n', buflim - beg); + end = memchr(beg, eol, buflim - beg); if (!end) end = buflim; - while (beg > buf && beg[-1] != '\n') + while (beg > buf && beg[-1] != eol) --beg; /* Successful, no backreferences encountered! */ if (!backref) @@ -313,8 +310,8 @@ EGexecute(buf, size, endp) } /* If we've made it to this point, this means DFA has seen a probable match, and we need to run it through Regex. */ - regex.not_eol = 0; - if ((start = re_search(®ex, beg, end - beg, 0, end - beg, ®s)) >= 0) + regexbuf.not_eol = 0; + if ((start = re_search(®exbuf, beg, end - beg, 0, end - beg, ®s)) >= 0) { len = regs.end[0] - start; if ((!match_lines && !match_words) @@ -337,8 +334,8 @@ EGexecute(buf, size, endp) { /* Try a shorter length anchored at the same place. */ --len; - regex.not_eol = 1; - len = re_match(®ex, beg, start + len, start, ®s); + regexbuf.not_eol = 1; + len = re_match(®exbuf, beg, start + len, start, ®s); } if (len <= 0) { @@ -346,8 +343,8 @@ EGexecute(buf, size, endp) if (start == end - beg) break; ++start; - regex.not_eol = 0; - start = re_search(®ex, beg, end - beg, + regexbuf.not_eol = 0; + start = re_search(®exbuf, beg, end - beg, start, end - beg - start, ®s); len = regs.end[0] - start; } @@ -396,6 +393,7 @@ Fexecute(buf, size, endp) { register char *beg, *try, *end; register size_t len; + char eol = eolbyte; struct kwsmatch kwsmatch; for (beg = buf; beg <= buf + size; ++beg) @@ -405,9 +403,9 @@ Fexecute(buf, size, endp) len = kwsmatch.size[0]; if (match_lines) { - if (beg > buf && beg[-1] != '\n') + if (beg > buf && beg[-1] != eol) continue; - if (beg + len < buf + size && beg[len] != '\n') + if (beg + len < buf + size && beg[len] != eol) continue; goto success; } @@ -431,7 +429,7 @@ Fexecute(buf, size, endp) return 0; success: - if ((end = memchr(beg + len, '\n', (buf + size) - (beg + len))) != 0) + if ((end = memchr(beg + len, eol, (buf + size) - (beg + len))) != 0) ++end; else end = buf + size; |