Overhaul hexdump's od syntax code to handle the -s -A -j -N -t options that

SUSv3 requires and give od a proper manual page.

PR:		36783

5 files changed, 425 insertions, 87 deletions

diff --git a/usr.bin/hexdump/conv.c b/usr.bin/hexdump/conv.c
index 9cf3e87..eea03be 100644
--- a/usr.bin/hexdump/conv.c
+++ b/usr.bin/hexdump/conv.c
@@ -57,8 +57,6 @@ conv_c(pr, p)
 		goto strpr;
 	/* case '\a': */
 	case '\007':
-		if (odmode)		/* od didn't know about \a */
-			break;
 		str = "\\a";
 		goto strpr;
 	case '\b':
@@ -77,8 +75,6 @@ conv_c(pr, p)
 		str = "\\t";
 		goto strpr;
 	case '\v':
-		if (odmode)
-			break;
 		str = "\\v";
 		goto strpr;
 	default:
diff --git a/usr.bin/hexdump/display.c b/usr.bin/hexdump/display.c
index ec957b7..d2e32fc 100644
--- a/usr.bin/hexdump/display.c
+++ b/usr.bin/hexdump/display.c
@@ -150,9 +150,11 @@ print(pr, bp)
 			bcopy(bp, &f8, sizeof(f8));
 			(void)printf(pr->fmt, f8);
 			break;
-		case sizeof(long double):
-			bcopy(bp, &ldbl, sizeof(ldbl));
-			(void)printf(pr->fmt, ldbl);
+		default:
+			if (pr->bcnt == sizeof(long double)) {
+				bcopy(bp, &ldbl, sizeof(ldbl));
+				(void)printf(pr->fmt, ldbl);
+			}
 			break;
 		}
 		break;
@@ -259,6 +261,8 @@ get()
 		 * block and set the end flag.
 		 */
 		if (!length || (ateof && !next((char **)NULL))) {
+			if (odmode && address < skip)
+				errx(1, "cannot skip past end of input");
 			if (need == blocksize)
 				return((u_char *)NULL);
 			if (vflag != ALL && 
diff --git a/usr.bin/hexdump/od.1 b/usr.bin/hexdump/od.1
index ac46aab..f135469 100644
--- a/usr.bin/hexdump/od.1
+++ b/usr.bin/hexdump/od.1
@@ -32,7 +32,7 @@
 .\"	@(#)od.1	8.1 (Berkeley) 6/6/93
 .\" $FreeBSD$
 .\"
-.Dd May 27, 1994
+.Dd April 17, 2002
 .Os
 .Dt OD 1
 .Sh NAME
@@ -40,7 +40,11 @@
 .Nd octal, decimal, hex, ASCII dump
 .Sh SYNOPSIS
 .Nm
-.Op Fl aBbcDdeFfHhIiLlOovXx
+.Op Fl aBbcDdeFfHhIiLlOosvXx
+.Op Fl A Ar base
+.Op Fl j Ar skip
+.Op Fl N Ar length
+.Op Fl t Ar type
 .Sm off
 .Oo
 .Op Cm \&+
@@ -49,32 +53,183 @@
 .Op Cm Bb
 .Oc
 .Sm on
-.Ar file
+.Op Ar
 .Sh DESCRIPTION
 The
-.Nm hexdump
-utility, if called as
-.Nm ,
-provides compatibility for the options listed above.
+.Nm
+utility is a filter which displays the specified files, or standard
+input if no files are specified, in a user specified format.
 .Pp
-It does not provide compatibility for the
-.Fl s
-option (see
-.Xr strings 1 )
-or the
-.Fl P ,
-.Fl p ,
+The options are as follows:
+.Bl -tag -width Fl
+.It Fl A Ar base
+Specify the input address base.
+.Ar base
+may be one of
+.Ql d ,
+.Ql o ,
+.Ql x
+or
+.Ql n ,
+which specify decimal, octal, hexadecimal
+addresses or no address, respectively.
+.It Fl a
+Output named characters.
+Equivalent to
+.Fl t Ar a .
+.It Fl B , Fl o
+Output octal shorts.
+Equivalent to
+.Fl t Ar o2 .
+.It Fl b
+Output octal bytes.
+Equivalent to
+.Fl t Ar o1 .
+.It Fl c
+Output C-style escaped characters.
+Equivalent to
+.Fl t Ar c .
+.It Fl D
+Output unsigned decimal ints.
+Equivalent to
+.Fl t Ar u4 .
+.It Fl e , Fl F
+Output double-precision floating point numbers.
+Equivalent to
+.Fl t Ar fD .
+.It Fl f
+Output single-precision floating point numbers.
+Equivalent to
+.Fl t Ar fF .
+.It Fl H , Fl X
+Output hexadecimal ints.
+Equivalent to
+.Fl t Ar x4 .
+.It Fl h , Fl x
+Output hexadecimal shorts.
+Equivalent to
+.Fl t Ar x2 .
+.It Fl I , Fl L , Fl l
+Output signed decimal longs.
+Equivalent to
+.Fl t Ar dL .
+.It Fl i
+Output signed decimal ints.
+Equivalent to
+.Fl t Ar dI .
+.It Fl j Ar skip
+Skip
+.Ar skip
+bytes of the combined input before dumping. The number may be followed by one
+of
+.Ql b ,
+.Ql k
 or
-.Fl w
-options, nor is compatibility provided for the ``label'' component
-of the offset syntax.
+.Ql m
+which specify the units of the number as blocks (512 bytes), kilobytes and
+megabytes, respectively.
+.It Fl N Ar length
+Dump at most
+.Ar length
+bytes of input.
+.It Fl O
+Output octal ints.
+Equivalent to
+.Fl t Ar o4 .
+.It Fl s
+Output signed decimal shorts.
+Equivalent to
+.Fl t Ar d2 .
+.It Fl t Ar type
+Specify the output format.
+.Ar type
+is a string containing one or more of the following kinds of type specifiers:
+.Bl -tag -width indent
+.It Cm a
+Named characters
+.Pq Sq ASCII .
+Control characters are displayed using the following names:
+.Bl -column \&000_nu \&001_so \&002_st \&003_et \&004_eo
+.It "\&000\ nul\t001\ soh\t002\ stx\t003\ etx\t004\ eot\t005\ enq
+.It "\&006\ ack\t007\ bel\t008\ bs\t009\ ht\t00A\ nl\t00B\ vt
+.It "\&00C\ ff\t00D\ cr\t00E\ so\t00F\ si\t010\ dle\t011\ dc1
+.It "\&012\ dc2\t013\ dc3\t014\ dc4\t015\ nak\t016\ syn\t017\ etb
+.It "\&018\ can\t019\ em\t01A\ sub\t01B\ esc\t01C\ fs\t01D\ gs
+.It "\&01E\ rs\t01F\ us\t020\ sp\t0FF\ del
+.El
+.It Cm c
+Characters in the default character set. Non-printing characters are
+represented as 3-digit octal character codes, except the following
+characters, which are represented as C escapes:
+.Bl -column carriage-return \er
+.It NUL Ta \e0
+.It alert Ta \ea
+.It backspace Ta \eb
+.It newline Ta \en
+.It carriage-return Ta \er
+.It tab Ta \et
+.It vertical tab Ta \ev
+.El
+.It Cm [d|o|u|x][C|S|I|L| Ns Ar n Ns ]
+Signed decimal
+.Pq Ql d ,
+octal
+.Pq Ql o ,
+unsigned decimal
+.Pq Ql u
+or
+hexadecimal
+.Pq Ql x .
+Followed by an optional size specifier, which may be either
+.Ql C
+.Pq "char" ,
+.Ql S
+.Pq "short" ,
+.Ql I
+.Pq "int" ,
+.Ql L
+.Pq "long" ,
+or a byte count as a decimal integer.
+.It Cm f[F|D|L| Ns Ar n Ns ]
+Floating-point number.
+Followed by an optional size specifier, which may be either
+.Ql F
+.Pq "float" ,
+.Ql D
+.Pq "double"
+or
+.Ql L
+.Pq "long double" .
+.El
+.It Fl v
+Write all input data, instead of replacing lines of duplicate values with a
+.Ql * .
+.El
+.Pp
+Multiple options that specify output format may be used; the output will
+contain one line for each format.
+.Pp
+If no output format is specified,
+.Fl t Ar oS
+is assumed.
+.Sh DIAGNOSTICS
+.Ex -std
+.Sh COMPATIBILITY
+The traditional
+.Fl s
+option to extract string constants is not supported; consider using
+.Xr strings 1
+instead.
 .Sh SEE ALSO
 .Xr hexdump 1 ,
 .Xr strings 1
-.Sh BUGS
-Quite a few.
+.Sh STANDARDS
+The
+.Nm
+utility conforms to
+.St -p1003.1-2001 .
 .Sh HISTORY
-A
+An
 .Nm
 command appeared in
 .At v1 .
diff --git a/usr.bin/hexdump/odsyntax.c b/usr.bin/hexdump/odsyntax.c
index 214b949..082a0a0 100644
--- a/usr.bin/hexdump/odsyntax.c
+++ b/usr.bin/hexdump/odsyntax.c
@@ -43,107 +43,136 @@ static const char rcsid[] =
 
 #include <ctype.h>
 #include <err.h>
+#include <errno.h>
+#include <float.h>
 #include <stdio.h>
 #include <stdlib.h>
+#include <string.h>
 #include <unistd.h>
 
 #include "hexdump.h"
 
+#define PADDING	"         "
+
 int odmode;
 
+static void odadd(const char *);
+static void odformat(const char *);
+static const char *odformatfp(char, const char *);
+static const char *odformatint(char, const char *);
 static void odoffset(int, char ***);
-static void odprecede(void);
+static void odusage(void);
 
 void
 oldsyntax(argc, argvp)
 	int argc;
 	char ***argvp;
 {
+	static char empty[] = "", padding[] = PADDING;
 	int ch;
-	char **argv;
+	char **argv, *end;
+
+	/* Add initial (default) address format. -A may change it later. */
+#define	TYPE_OFFSET	7
+	add("\"%07.7_Ao\n\"");
+	add("\"%07.7_ao  \"");
 
 	odmode = 1;
 	argv = *argvp;
-	while ((ch = getopt(argc, argv, "aBbcDdeFfHhIiLlOoPpswvXx")) != -1)
+	while ((ch = getopt(argc, argv, "A:aBbcDdeFfHhIij:LlN:Oost:vXx")) != -1)
 		switch (ch) {
+		case 'A':
+			switch (*optarg) {
+			case 'd': case 'o': case 'x':
+				fshead->nextfu->fmt[TYPE_OFFSET] = *optarg;
+				fshead->nextfs->nextfu->fmt[TYPE_OFFSET] =
+				    *optarg;
+				break;
+			case 'n':
+				fshead->nextfu->fmt = empty;
+				fshead->nextfs->nextfu->fmt = padding;
+				break;
+			default:
+				errx(1, "%s: invalid address base", optarg);
+			}
+			break;
 		case 'a':
-			odprecede();
-			add("16/1 \"%3_u \" \"\\n\"");
+			odformat("a");
 			break;
 		case 'B':
 		case 'o':
-			odprecede();
-			add("8/2 \" %06o \" \"\\n\"");
+			odformat("o2");
 			break;
 		case 'b':
-			odprecede();
-			add("16/1 \"%03o \" \"\\n\"");
+			odformat("o1");
 			break;
 		case 'c':
-			odprecede();
-			add("16/1 \"%3_c \" \"\\n\"");
+			odformat("c");
 			break;
 		case 'd':
-			odprecede();
-			add("8/2 \"  %05u \" \"\\n\"");
+			odformat("u2");
 			break;
 		case 'D':
-			odprecede();
-			add("4/4 \"     %010u \" \"\\n\"");
+			odformat("u4");
 			break;
 		case 'e':		/* undocumented in od */
 		case 'F':
-			odprecede();
-			add("2/8 \"          %21.14e \" \"\\n\"");
+			odformat("fD");
 			break;
-
 		case 'f':
-			odprecede();
-			add("4/4 \" %14.7e \" \"\\n\"");
+			odformat("fF");
 			break;
 		case 'H':
 		case 'X':
-			odprecede();
-			add("4/4 \"       %08x \" \"\\n\"");
+			odformat("x4");
 			break;
 		case 'h':
 		case 'x':
-			odprecede();
-			add("8/2 \"   %04x \" \"\\n\"");
+			odformat("x2");
 			break;
 		case 'I':
 		case 'L':
 		case 'l':
-			odprecede();
-			add("4/4 \"    %11d \" \"\\n\"");
+			odformat("dL");
 			break;
 		case 'i':
-			odprecede();
-			add("8/2 \" %6d \" \"\\n\"");
+			odformat("dI");
+			break;
+		case 'j':
+			errno = 0;
+			skip = strtoll(optarg, &end, 0);
+			if (*end == 'b')
+				skip *= 512;
+			else if (*end == 'k')
+				skip *= 1024;
+			else if (*end == 'm')
+				skip *= 1048576L;
+			if (errno != 0 || skip < 0 || strlen(end) > 1)
+				errx(1, "%s: invalid skip amount", optarg);
+			break;
+		case 'N':
+			if ((length = atoi(optarg)) <= 0)
+				errx(1, "%s: invalid length", optarg);
 			break;
 		case 'O':
-			odprecede();
-			add("4/4 \"    %011o \" \"\\n\"");
+			odformat("o4");
+			break;
+		case 's':
+			odformat("d2");
+			break;
+		case 't':
+			odformat(optarg);
 			break;
 		case 'v':
 			vflag = ALL;
 			break;
-		case 'P':
-		case 'p':
-		case 's':
-		case 'w':
 		case '?':
 		default:
-			if (ch != '?')
-				warnx("hexdump(1) compatibility doesn't support the -%c option%s",
-				    ch, ch == 's' ? "; see strings(1)" : "");
-			usage();
+			odusage();
 		}
 
-	if (!fshead) {
-		add("\"%07.7_Ao\n\"");
-		add("\"%07.7_ao  \" 8/2 \"%06o \" \"\\n\"");
-	}
+	if (fshead->nextfs->nextfs == NULL)
+		odformat("oS");
 
 	argc -= optind;
 	*argvp += optind;
@@ -153,6 +182,17 @@ oldsyntax(argc, argvp)
 }
 
 static void
+odusage(void)
+{
+
+	fprintf(stderr,
+"usage: od [-aBbcDdeFfHhIiLlOosvXx] [-A base] [-j skip] [-N length] [-t type]\n");
+	fprintf(stderr,
+"          [[+]offset[.][Bb]] [file ...]\n");
+	exit(1);
+}
+
+static void
 odoffset(argc, argvp)
 	int argc;
 	char ***argvp;
@@ -237,7 +277,6 @@ odoffset(argc, argvp)
 	 * If the offset uses a non-octal base, the base of the offset
 	 * is changed as well.  This isn't pretty, but it's easy.
 	 */
-#define	TYPE_OFFSET	7
 	if (base == 16) {
 		fshead->nextfu->fmt[TYPE_OFFSET] = 'x';
 		fshead->nextfs->nextfu->fmt[TYPE_OFFSET] = 'x';
@@ -251,14 +290,157 @@ odoffset(argc, argvp)
 }
 
 static void
-odprecede()
+odformat(const char *fmt)
+{
+	char fchar;
+
+	while (*fmt != '\0') {
+		switch ((fchar = *fmt++)) {
+		case 'a':
+			odadd("16/1 \"%3_u \" \"\\n\"");
+			break;
+		case 'c':
+			odadd("16/1 \"%3_c \" \"\\n\"");
+			break;
+		case 'o': case 'u': case 'd': case 'x':
+			fmt = odformatint(fchar, fmt);
+			break;
+		case 'f':
+			fmt = odformatfp(fchar, fmt);
+			break;
+		default:
+			errx(1, "%c: unrecognised format character", fchar);
+		}
+	}
+}
+
+static const char *
+odformatfp(char fchar __unused, const char *fmt)
+{
+	size_t isize;
+	int digits;
+	char *end, *hdfmt;
+
+	isize = sizeof(double);
+	switch (*fmt) {
+	case 'F':
+		isize = sizeof(float);
+		fmt++;
+		break;
+	case 'D':
+		isize = sizeof(double);
+		fmt++;
+		break;
+	case 'L':
+		isize = sizeof(long double);
+		fmt++;
+		break;
+	default:
+		if (isdigit((unsigned char)*fmt)) {
+			errno = 0;
+			isize = (size_t)strtoul(fmt, &end, 10);
+			if (errno != 0 || isize == 0)
+				errx(1, "%s: invalid size", fmt);
+			fmt = (const char *)end;
+		}
+	}
+	switch (isize) {
+	case sizeof(float):
+		digits = FLT_DIG;
+		break;
+	case sizeof(double):
+		digits = DBL_DIG;
+		break;
+	default:
+		if (isize == sizeof(long double))
+			digits = LDBL_DIG;
+		else
+			errx(1, "unsupported floating point size %lu",
+			    (u_long)isize);
+	}
+
+	asprintf(&hdfmt, "%lu/%lu \" %%%d.%de \" \"\\n\"",
+	    16UL / (u_long)isize, (u_long)isize, digits + 8, digits);
+	if (hdfmt == NULL)
+		err(1, NULL);
+	odadd(hdfmt);
+	free(hdfmt);
+
+	return (fmt);
+}
+
+static const char *
+odformatint(char fchar, const char *fmt)
+{
+	unsigned long long n;
+	size_t isize;
+	int digits;
+	char *end, *hdfmt;
+
+	isize = sizeof(int);
+	switch (*fmt) {
+	case 'C':
+		isize = sizeof(char);
+		fmt++;
+		break;
+	case 'I':
+		isize = sizeof(int);
+		fmt++;
+		break;
+	case 'L':
+		isize = sizeof(long);
+		fmt++;
+		break;
+	case 'S':
+		isize = sizeof(short);
+		fmt++;
+		break;
+	default:
+		if (isdigit((unsigned char)*fmt)) {
+			errno = 0;
+			isize = (size_t)strtoul(fmt, &end, 10);
+			if (errno != 0 || isize == 0)
+				errx(1, "%s: invalid size", fmt);
+			if (isize != sizeof(char) && isize != sizeof(short) &&
+			    isize != sizeof(int) && isize != sizeof(long))
+				errx(1, "unsupported int size %lu",
+				    (u_long)isize);
+			fmt = (const char *)end;
+		}
+	}
+
+	/*
+	 * Calculate the maximum number of digits we need to
+	 * fit the number. Overestimate for decimal with log
+	 * base 8. We need one extra space for signed numbers
+	 * to store the sign.
+	 */
+	n = (1ULL << (8 * isize)) - 1;
+	digits = 0;
+	while (n != 0) {
+		digits++;
+		n >>= (fchar == 'x') ? 4 : 3;
+	}
+	if (fchar == 'd')
+		digits++;
+	asprintf(&hdfmt, "%lu/%lu \"%%%s%d%c \" \"\\n\"",
+	    16UL / (u_long)isize, (u_long)isize,
+	    (fchar == 'd' || fchar == 'u') ? "" : "0", digits, fchar);
+	if (hdfmt == NULL)
+		err(1, NULL);
+	odadd(hdfmt);
+	free(hdfmt);
+
+	return (fmt);
+}
+
+static void
+odadd(const char *fmt)
 {
-	static int first = 1;
-
-	if (first) {
-		first = 0;
-		add("\"%07.7_Ao\n\"");
-		add("\"%07.7_ao  \"");
-	} else
-		add("\"         \"");
+	static int needpad;
+
+	if (needpad)
+		add("\""PADDING"\"");
+	add(fmt);
+	needpad = 1;
 }
diff --git a/usr.bin/hexdump/parse.c b/usr.bin/hexdump/parse.c
index da93e7b..85e035e 100644
--- a/usr.bin/hexdump/parse.c
+++ b/usr.bin/hexdump/parse.c
@@ -313,15 +313,16 @@ isint:				cs[2] = '\0';
 				case 4:
 					pr->bcnt = 4;
 					break;
-				case sizeof(long double):
-					cs[2] = '\0';
-					cs[1] = cs[0];
-					cs[0] = 'L';
-					pr->bcnt = sizeof(long double);
-					break;
 				default:
-					p1[1] = '\0';
-					badcnt(p1);
+					if (fu->bcnt == sizeof(long double)) {
+						cs[2] = '\0';
+						cs[1] = cs[0];
+						cs[0] = 'L';
+						pr->bcnt = sizeof(long double);
+					} else {
+						p1[1] = '\0';
+						badcnt(p1);
+					}
 				}
 				break;
 			case 's':