dono@killer.Dallas.TX.US (Don OConnell) (03/18/89)
------------------------------Cut Here--------------------------------------
#! /bin/sh
# This is a shell archive. Remove anything before this line, then unpack
# it by saving it into a file and typing "sh file". To overwrite existing
# files, type "sh file -c". You can also feed this as standard input via
# unshar, or by typing "sh <file", e.g.. If this archive is complete, you
# will see the following message at the end:
# "End of archive 1 (of 4)."
# Contents: README Manifest alloca.c regex.c tests
# Wrapped by dono@killer on Sat Mar 18 00:05:26 1989
PATH=/bin:/usr/bin:/usr/ucb ; export PATH
if test -f README -a "${1}" != "-c" ; then
echo shar: Will not over-write existing file \"README\"
else
echo shar: Extracting \"README\" \(7427 characters\)
sed "s/^X//" >README <<'END_OF_README'
XThis README documents GNU e?grep version 1.3.
X
XChanges needed to the makefile under various perversions of Unix are
Xdescribed therein.
X
XIf the type "char" is unsigned on your machine, you will have to fix
Xthe definition of the macro SIGN_EXTEND_CHAR() in regex.c. A reasonable
Xdefinition might be:
X #define SIGN_EXTEND_CHAR(c) ((c)>(char)127?(c)-256:(c))
X
XGNU e?grep is provided "as is" with no warranty. The exact terms
Xunder which you may use and (re)distribute this program are detailed
Xin a comment at the top of grep.c.
X
XGNU e?grep is based on a fast lazy-state deterministic matcher (about
Xtwice as fast as stock Unix egrep) hybridized with a Boyer-Moore-Gosper
Xsearch for a fixed string that eliminates impossible text from being
Xconsidered by the full regexp matcher without necessarily having to
Xlook at every character. The result is typically many times faster
Xthan Unix grep or egrep. (Regular expressions containing backreferencing
Xmay run more slowly, however.)
X
XGNU e?grep attempts, as closely as possible, to understand compatibly
Xthe regexp syntaxes of the Unix programs it replaces. The following table
Xdetails the various special characters understood in both the grep and
Xegrep incarnations:
X
X(grep) (egrep) (explanation)
X . . matches any single character except newline
X \? ? postfix operator; preceeding item is optional
X * * postfix operator; preceeding item 0 or more times
X \+ + postfix operator; preceeding item 1 or more times
X \| | infix operator; matches either argument
X ^ ^ matches the empty string at the beginning of a line
X $ $ matches the empty string at the end of a line
X \< \< matches the empty string at the beginning of a word
X \> \> matches the empty string at the end of a word
X [chars] [chars] match any character in the given class; if the
X first character after [ is ^, match any character
X not in the given class; a range of characters may
X be specified by <first>-<last>; for example, \W
X (below) is equivalent to the class [^A-Za-z0-9]
X \( \) ( ) parentheses are used to override operator precedence
X \<1-9> \<1-9> \<n> matches a repeat of the text matched earlier
X in the regexp by the subexpression inside the
X nth opening parenthesis
X \ \ any special character may be preceded by a backslash
X to match it literally
X
X(the following are for compatibility with GNU Emacs)
X \b \b matches the empty string at the edge of a word
X \B \B matches the empty string if not at the edge of a word
X \w \w matches word-constituent characters (letters & digits)
X \W \W matches characters that are not word-constituent
X
XOperator precedence is (highest to lowest) ?, *, and +, concatenation,
Xand finally |. All other constructs are syntactically identical to
Xnormal characters. For the truly interested, a comment in dfa.c describes
Xthe exact grammar understood by the parser.
X
XGNU e?grep understands the following command line options:
X -A <num> print <num> lines of context after every matching line
X -B <num> print <num> lines of context before every matching line
X -C print 2 lines of context on each side of every match
X -<num> print <num> lines of context on each side
X -V print the version number on stderr
X -b print every match preceded by its byte offset
X -c print a total count of matching lines only
X -e <expr> search for <expr>; useful if <expr> begins with -
X -f <file> take <expr> from the given <file>
X -h don't display filenames on matches
X -i ignore case difference when comparing strings
X -l list files containing matches only
X -n print each match preceded by its line number
X -s run silently producing no output except error messages
X -v print only lines that contain no matches for the <expr>
X -w print only lines where the match is a complete word
X -x print only lines where the match is a whole line
X
XThe options understood by GNU e?grep are meant to be (nearly) compatible
Xwith both the BSD and System V versions of grep and egrep.
X
XThe following incompatibilities with other versions of grep exist:
X the context-dependent meaning of * is not quite the same (grep only)
X -b prints a byte offset instead of a block offset
X the \{m,n\} construct of System V grep is not implemented
X
XGNU e?grep has been thoroughly debugged and tested by several people
Xover a period of several months; we think it's a reliable beast or we
Xwouldn't distribute it. If by some fluke of the universe you discover
Xa bug, send a detailed description (including options, regular
Xexpressions, and a copy of an input file that can reproduce it) to me,
Xmike@wheaties.ai.mit.edu.
X
XGNU e?grep is brought to you by the efforts of several people:
X
X Mike Haertel wrote the deterministic regexp code and the bulk
X of the program.
X
X James A. Woods is responsible for the hybridized search strategy
X of using Boyer-Moore-Gosper fixed-string search as a filter
X before calling the general regexp matcher.
X
X Arthur David Olson contributed code that finds fixed strings for
X the aforementioned BMG search for a large class of regexps.
X
X Richard Stallman wrote the backtracking regexp matcher that is
X used for \<digit> backreferences, as well as the getopt that
X is provided for 4.2BSD sites. The backtracking matcher was
X originally written for GNU Emacs.
X
X D. A. Gwyn wrote the C alloca emulation that is provided so
X System V machines can run this program. (Alloca is used only
X by RMS' backtracking matcher, and then only rarely, so there
X is no loss if your machine doesn't have a "real" alloca.)
X
X Scott Anderson and Henry Spencer designed the regression tests
X used in the "regress" script.
X
X Paul Placeway wrote the manual page, based on this README.
X
XIf you are interested in improving this program, you may wish to try
Xany of the following:
X
X1. Make backreferencing \<digit> faster. Right now, backreferencing is
X handled by calling the Emacs backtracking matcher to verify the partial
X match. This is slow; if the DFA routines could handle backreferencing
X themselves a speedup on the order of three to four times might occur
X in those cases where the backtracking matcher is called to verify nearly
X every line. Also, some portability problems due to the inclusion of the
X emacs matcher would be solved because it could then be eliminated.
X Note that expressions with backreferencing are not true regular
X expressions, and thus are not equivalent to any DFA. So this is hard.
X
X2. There is a bug in the backtracking matcher, regex.c, such that the |
X operator is not properly commutative. Let x and y be arbitrary
X regular expressions, and suppose both x and y have matches at
X some point in the target text. Then the regexp x|y should select
X the longest of the two matches. With the backtracking matcher, if the
X first match succeeds it does not even try the second, even though
X the second may be a longer match. This is obviously of no concern
X for grep, which does not care exactly where or how long a match is,
X so long as it knows it is there. On the other hand, the backtracking
X matcher is used in GNU AWK, wherein its behavior can only be considered
X a bug.
X
X3. Handle POSIX style regexps. I'm not sure if this could be called an
X improvement; some of the things on regexps in the POSIX draft I have
X seen are pretty sickening. But it would be useful in the interests of
X conforming to the standard.
END_OF_README
if test 7427 -ne `wc -c <README`; then
echo shar: \"README\" unpacked with wrong size!
fi
# end of overwriting check
fi
if test -f Manifest -a "${1}" != "-c" ; then
echo shar: Will not over-write existing file \"Manifest\"
else
echo shar: Extracting \"Manifest\" \(516 characters\)
sed "s/^X//" >Manifest <<'END_OF_Manifest'
X File Name Archive # Description
X----------------------------------------------------------
X README 1
X Makefile 4
X Manifest 1 This shipping list
X alloca.c 1
X dfa.c 2
X dfa.h 3
X getopt.c 4
X grep.1m 4
X grep.c 3
X grep.man 4
X regex.c 1
X regex.h 3
X tests 1
END_OF_Manifest
if test 516 -ne `wc -c <Manifest`; then
echo shar: \"Manifest\" unpacked with wrong size!
fi
# end of overwriting check
fi
if test -f alloca.c -a "${1}" != "-c" ; then
echo shar: Will not over-write existing file \"alloca.c\"
else
echo shar: Extracting \"alloca.c\" \(5090 characters\)
sed "s/^X//" >alloca.c <<'END_OF_alloca.c'
X/*
X alloca -- (mostly) portable public-domain implementation -- D A Gwyn
X
X This implementation of the PWB library alloca() function,
X which is used to allocate space off the run-time stack so
X that it is automatically reclaimed upon procedure exit,
X was inspired by discussions with J. Q. Johnson of Cornell.
X
X It should work under any C implementation that uses an
X actual procedure stack (as opposed to a linked list of
X frames). There are some preprocessor constants that can
X be defined when compiling for your specific system, for
X improved efficiency; however, the defaults should be okay.
X
X The general concept of this implementation is to keep
X track of all alloca()-allocated blocks, and reclaim any
X that are found to be deeper in the stack than the current
X invocation. This heuristic does not reclaim storage as
X soon as it becomes invalid, but it will do so eventually.
X
X As a special case, alloca(0) reclaims storage without
X allocating any. It is a good idea to use alloca(0) in
X your main control loop, etc. to force garbage collection.
X*/
X#ifndef lint
Xstatic char SCCSid[] = "@(#)alloca.c 1.1"; /* for the "what" utility */
X#endif
X
X#ifdef emacs
X#include "config.h"
X#ifdef static
X/* actually, only want this if static is defined as ""
X -- this is for usg, in which emacs must undefine static
X in order to make unexec workable
X */
X#ifndef STACK_DIRECTION
Xyou
Xlose
X-- must know STACK_DIRECTION at compile-time
X#endif /* STACK_DIRECTION undefined */
X#endif static
X#endif emacs
X
X#ifdef X3J11
Xtypedef void *pointer; /* generic pointer type */
X#else
Xtypedef char *pointer; /* generic pointer type */
X#endif
X
X#define NULL 0 /* null pointer constant */
X
Xextern void free();
Xextern pointer malloc();
X
X/*
X Define STACK_DIRECTION if you know the direction of stack
X growth for your system; otherwise it will be automatically
X deduced at run-time.
X
X STACK_DIRECTION > 0 => grows toward higher addresses
X STACK_DIRECTION < 0 => grows toward lower addresses
X STACK_DIRECTION = 0 => direction of growth unknown
X*/
X
X#ifndef STACK_DIRECTION
X#define STACK_DIRECTION 0 /* direction unknown */
X#endif
X
X#if STACK_DIRECTION != 0
X
X#define STACK_DIR STACK_DIRECTION /* known at compile-time */
X
X#else /* STACK_DIRECTION == 0; need run-time code */
X
Xstatic int stack_dir; /* 1 or -1 once known */
X#define STACK_DIR stack_dir
X
Xstatic void
Xfind_stack_direction (/* void */)
X{
X static char *addr = NULL; /* address of first
X `dummy', once known */
X auto char dummy; /* to get stack address */
X
X if (addr == NULL)
X { /* initial entry */
X addr = &dummy;
X
X find_stack_direction (); /* recurse once */
X }
X else /* second entry */
X if (&dummy > addr)
X stack_dir = 1; /* stack grew upward */
X else
X stack_dir = -1; /* stack grew downward */
X}
X
X#endif /* STACK_DIRECTION == 0 */
X
X/*
X An "alloca header" is used to:
X (a) chain together all alloca()ed blocks;
X (b) keep track of stack depth.
X
X It is very important that sizeof(header) agree with malloc()
X alignment chunk size. The following default should work okay.
X*/
X
X#ifndef ALIGN_SIZE
X#define ALIGN_SIZE sizeof(double)
X#endif
X
Xtypedef union hdr
X{
X char align[ALIGN_SIZE]; /* to force sizeof(header) */
X struct
X {
X union hdr *next; /* for chaining headers */
X char *deep; /* for stack depth measure */
X } h;
X} header;
X
X/*
X alloca( size ) returns a pointer to at least `size' bytes of
X storage which will be automatically reclaimed upon exit from
X the procedure that called alloca(). Originally, this space
X was supposed to be taken from the current stack frame of the
X caller, but that method cannot be made to work for some
X implementations of C, for example under Gould's UTX/32.
X*/
X
Xstatic header *last_alloca_header = NULL; /* -> last alloca header */
X
Xpointer
Xalloca (size) /* returns pointer to storage */
X unsigned size; /* # bytes to allocate */
X{
X auto char probe; /* probes stack depth: */
X register char *depth = &probe;
X
X#if STACK_DIRECTION == 0
X if (STACK_DIR == 0) /* unknown growth direction */
X find_stack_direction ();
X#endif
X
X /* Reclaim garbage, defined as all alloca()ed storage that
X was allocated from deeper in the stack than currently. */
X
X {
X register header *hp; /* traverses linked list */
X
X for (hp = last_alloca_header; hp != NULL;)
X if (STACK_DIR > 0 && hp->h.deep > depth
X || STACK_DIR < 0 && hp->h.deep < depth)
X {
X register header *np = hp->h.next;
X
X free ((pointer) hp); /* collect garbage */
X
X hp = np; /* -> next header */
X }
X else
X break; /* rest are not deeper */
X
X last_alloca_header = hp; /* -> last valid storage */
X }
X
X if (size == 0)
X return NULL; /* no allocation required */
X
X /* Allocate combined header + user data storage. */
X
X {
X register pointer new = malloc (sizeof (header) + size);
X /* address of header */
X
X ((header *)new)->h.next = last_alloca_header;
X ((header *)new)->h.deep = depth;
X
X last_alloca_header = (header *)new;
X
X /* User storage begins just after header. */
X
X return (pointer)((char *)new + sizeof(header));
X }
X}
END_OF_alloca.c
if test 5090 -ne `wc -c <alloca.c`; then
echo shar: \"alloca.c\" unpacked with wrong size!
fi
# end of overwriting check
fi
if test -f regex.c -a "${1}" != "-c" ; then
echo shar: Will not over-write existing file \"regex.c\"
else
echo shar: Extracting \"regex.c\" \(44778 characters\)
sed "s/^X//" >regex.c <<'END_OF_regex.c'
X/* Extended regular expression matching and search library.
X Copyright (C) 1985, 1989 Free Software Foundation, Inc.
X
X This program is free software; you can redistribute it and/or modify
X it under the terms of the GNU General Public License as published by
X the Free Software Foundation; either version 1, or (at your option)
X any later version.
X
X This program is distributed in the hope that it will be useful,
X but WITHOUT ANY WARRANTY; without even the implied warranty of
X MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
X GNU General Public License for more details.
X
X You should have received a copy of the GNU General Public License
X along with this program; if not, write to the Free Software
X Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
X
X
X In other words, you are welcome to use, share and improve this program.
X You are forbidden to forbid anyone else to use, share and improve
X what you give them. Help stamp out software-hoarding! */
X
X
X /* To test, compile with -Dtest.
X This Dtestable feature turns this into a self-contained program
X which reads a pattern, describes how it compiles,
X then reads a string and searches for it. */
X
X#ifdef emacs
X
X/* The `emacs' switch turns on certain special matching commands
X that make sense only in emacs. */
X
X#include "config.h"
X#include "lisp.h"
X#include "buffer.h"
X#include "syntax.h"
X
X#else /* not emacs */
X
X#ifdef USG
X#define bcopy(s,d,n) memcpy((d),(s),(n))
X#define bcmp(s1,s2,n) memcmp((s1),(s2),(n))
X#define bzero(s,n) memset((s),0,(n))
X#endif
X
X/* Make alloca work the best possible way. */
X#ifdef __GNUC__
X#define alloca __builtin_alloca
X#else
X#ifdef sparc
X#include <alloca.h>
X#endif
X#endif
X
X/*
X * Define the syntax stuff, so we can do the \<...\> things.
X */
X
X#ifndef Sword /* must be non-zero in some of the tests below... */
X#define Sword 1
X#endif
X
X#define SYNTAX(c) re_syntax_table[c]
X
X#ifdef SYNTAX_TABLE
X
Xchar *re_syntax_table;
X
X#else
X
Xstatic char re_syntax_table[256];
X
Xstatic void
Xinit_syntax_once ()
X{
X register int c;
X static int done = 0;
X
X if (done)
X return;
X
X bzero (re_syntax_table, sizeof re_syntax_table);
X
X for (c = 'a'; c <= 'z'; c++)
X re_syntax_table[c] = Sword;
X
X for (c = 'A'; c <= 'Z'; c++)
X re_syntax_table[c] = Sword;
X
X for (c = '0'; c <= '9'; c++)
X re_syntax_table[c] = Sword;
X
X done = 1;
X}
X
X#endif /* SYNTAX_TABLE */
X#endif /* not emacs */
X
X#include "regex.h"
X
X/* Number of failure points to allocate space for initially,
X when matching. If this number is exceeded, more space is allocated,
X so it is not a hard limit. */
X
X#ifndef NFAILURES
X#define NFAILURES 80
X#endif /* NFAILURES */
X
X/* width of a byte in bits */
X
X#define BYTEWIDTH 8
X
X#ifndef SIGN_EXTEND_CHAR
X#define SIGN_EXTEND_CHAR(x) (x)
X#endif
X
Xstatic int obscure_syntax = 0;
X
X/* Specify the precise syntax of regexp for compilation.
X This provides for compatibility for various utilities
X which historically have different, incompatible syntaxes.
X
X The argument SYNTAX is a bit-mask containing the two bits
X RE_NO_BK_PARENS and RE_NO_BK_VBAR. */
X
Xint
Xre_set_syntax (syntax)
X{
X int ret;
X
X ret = obscure_syntax;
X obscure_syntax = syntax;
X return ret;
X}
X
X/* re_compile_pattern takes a regular-expression string
X and converts it into a buffer full of byte commands for matching.
X
X PATTERN is the address of the pattern string
X SIZE is the length of it.
X BUFP is a struct re_pattern_buffer * which points to the info
X on where to store the byte commands.
X This structure contains a char * which points to the
X actual space, which should have been obtained with malloc.
X re_compile_pattern may use realloc to grow the buffer space.
X
X The number of bytes of commands can be found out by looking in
X the struct re_pattern_buffer that bufp pointed to,
X after re_compile_pattern returns.
X*/
X
X#define PATPUSH(ch) (*b++ = (char) (ch))
X
X#define PATFETCH(c) \
X {if (p == pend) goto end_of_pattern; \
X c = * (unsigned char *) p++; \
X if (translate) c = translate[c]; }
X
X#define PATFETCH_RAW(c) \
X {if (p == pend) goto end_of_pattern; \
X c = * (unsigned char *) p++; }
X
X#define PATUNFETCH p--
X
X#define EXTEND_BUFFER \
X { char *old_buffer = bufp->buffer; \
X if (bufp->allocated == (1<<14)) goto too_big; \
X bufp->allocated *= 2; \
X if (bufp->allocated > (1<<14)) bufp->allocated = (1<<14); \
X if (!(bufp->buffer = (char *) realloc (bufp->buffer, bufp->allocated))) \
X goto memory_exhausted; \
X c = bufp->buffer - old_buffer; \
X b += c; \
X if (fixup_jump) \
X fixup_jump += c; \
X if (laststart) \
X laststart += c; \
X begalt += c; \
X if (pending_exact) \
X pending_exact += c; \
X }
X
Xstatic int store_jump (), insert_jump ();
X
Xchar *
Xre_compile_pattern (pattern, size, bufp)
X char *pattern;
X int size;
X struct re_pattern_buffer *bufp;
X{
X register char *b = bufp->buffer;
X register char *p = pattern;
X char *pend = pattern + size;
X register unsigned c, c1;
X char *p1;
X unsigned char *translate = (unsigned char *) bufp->translate;
X
X /* address of the count-byte of the most recently inserted "exactn" command.
X This makes it possible to tell whether a new exact-match character
X can be added to that command or requires a new "exactn" command. */
X
X char *pending_exact = 0;
X
X /* address of the place where a forward-jump should go
X to the end of the containing expression.
X Each alternative of an "or", except the last, ends with a forward-jump
X of this sort. */
X
X char *fixup_jump = 0;
X
X /* address of start of the most recently finished expression.
X This tells postfix * where to find the start of its operand. */
X
X char *laststart = 0;
X
X /* In processing a repeat, 1 means zero matches is allowed */
X
X char zero_times_ok;
X
X /* In processing a repeat, 1 means many matches is allowed */
X
X char many_times_ok;
X
X /* address of beginning of regexp, or inside of last \( */
X
X char *begalt = b;
X
X /* Stack of information saved by \( and restored by \).
X Four stack elements are pushed by each \(:
X First, the value of b.
X Second, the value of fixup_jump.
X Third, the value of regnum.
X Fourth, the value of begalt. */
X
X int stackb[40];
X int *stackp = stackb;
X int *stacke = stackb + 40;
X int *stackt;
X
X /* Counts \('s as they are encountered. Remembered for the matching \),
X where it becomes the "register number" to put in the stop_memory command */
X
X int regnum = 1;
X
X bufp->fastmap_accurate = 0;
X
X#ifndef emacs
X#ifndef SYNTAX_TABLE
X /*
X * Initialize the syntax table.
X */
X init_syntax_once();
X#endif
X#endif
X
X if (bufp->allocated == 0)
X {
X bufp->allocated = 28;
X if (bufp->buffer)
X /* EXTEND_BUFFER loses when bufp->allocated is 0 */
X bufp->buffer = (char *) realloc (bufp->buffer, 28);
X else
X /* Caller did not allocate a buffer. Do it for him */
X bufp->buffer = (char *) malloc (28);
X if (!bufp->buffer) goto memory_exhausted;
X begalt = b = bufp->buffer;
X }
X
X while (p != pend)
X {
X if (b - bufp->buffer > bufp->allocated - 10)
X /* Note that EXTEND_BUFFER clobbers c */
X EXTEND_BUFFER;
X
X PATFETCH (c);
X
X switch (c)
X {
X case '$':
X if (obscure_syntax & RE_TIGHT_VBAR)
X {
X if (! (obscure_syntax & RE_CONTEXT_INDEP_OPS) && p != pend)
X goto normal_char;
X /* Make operand of last vbar end before this `$'. */
X if (fixup_jump)
X store_jump (fixup_jump, jump, b);
X fixup_jump = 0;
X PATPUSH (endline);
X break;
X }
X
X /* $ means succeed if at end of line, but only in special contexts.
X If randomly in the middle of a pattern, it is a normal character. */
X if (p == pend || *p == '\n'
X || (obscure_syntax & RE_CONTEXT_INDEP_OPS)
X || (obscure_syntax & RE_NO_BK_PARENS
X ? *p == ')'
X : *p == '\\' && p[1] == ')')
X || (obscure_syntax & RE_NO_BK_VBAR
X ? *p == '|'
X : *p == '\\' && p[1] == '|'))
X {
X PATPUSH (endline);
X break;
X }
X goto normal_char;
X
X case '^':
X /* ^ means succeed if at beg of line, but only if no preceding pattern. */
X
X if (laststart && p[-2] != '\n'
X && ! (obscure_syntax & RE_CONTEXT_INDEP_OPS))
X goto normal_char;
X if (obscure_syntax & RE_TIGHT_VBAR)
X {
X if (p != pattern + 1
X && ! (obscure_syntax & RE_CONTEXT_INDEP_OPS))
X goto normal_char;
X PATPUSH (begline);
X begalt = b;
X }
X else
X PATPUSH (begline);
X break;
X
X case '+':
X case '?':
X if (obscure_syntax & RE_BK_PLUS_QM)
X goto normal_char;
X handle_plus:
X case '*':
X /* If there is no previous pattern, char not special. */
X if (!laststart && ! (obscure_syntax & RE_CONTEXT_INDEP_OPS))
X goto normal_char;
X /* If there is a sequence of repetition chars,
X collapse it down to equivalent to just one. */
X zero_times_ok = 0;
X many_times_ok = 0;
X while (1)
X {
X zero_times_ok |= c != '+';
X many_times_ok |= c != '?';
X if (p == pend)
X break;
X PATFETCH (c);
X if (c == '*')
X ;
X else if (!(obscure_syntax & RE_BK_PLUS_QM)
X && (c == '+' || c == '?'))
X ;
X else if ((obscure_syntax & RE_BK_PLUS_QM)
X && c == '\\')
X {
X int c1;
X PATFETCH (c1);
X if (!(c1 == '+' || c1 == '?'))
X {
X PATUNFETCH;
X PATUNFETCH;
X break;
X }
X c = c1;
X }
X else
X {
X PATUNFETCH;
X break;
X }
X }
X
X /* Star, etc. applied to an empty pattern is equivalent
X to an empty pattern. */
X if (!laststart)
X break;
X
X /* Now we know whether 0 matches is allowed,
X and whether 2 or more matches is allowed. */
X if (many_times_ok)
X {
X /* If more than one repetition is allowed,
X put in a backward jump at the end. */
X store_jump (b, maybe_finalize_jump, laststart - 3);
X b += 3;
X }
X insert_jump (on_failure_jump, laststart, b + 3, b);
X pending_exact = 0;
X b += 3;
X if (!zero_times_ok)
X {
X /* At least one repetition required: insert before the loop
X a skip over the initial on-failure-jump instruction */
X insert_jump (dummy_failure_jump, laststart, laststart + 6, b);
X b += 3;
X }
X break;
X
X case '.':
X laststart = b;
X PATPUSH (anychar);
X break;
X
X case '[':
X while (b - bufp->buffer
X > bufp->allocated - 3 - (1 << BYTEWIDTH) / BYTEWIDTH)
X /* Note that EXTEND_BUFFER clobbers c */
X EXTEND_BUFFER;
X
X laststart = b;
X if (*p == '^')
X PATPUSH (charset_not), p++;
X else
X PATPUSH (charset);
X p1 = p;
X
X PATPUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
X /* Clear the whole map */
X bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH);
X /* Read in characters and ranges, setting map bits */
X while (1)
X {
X PATFETCH (c);
X if (c == ']' && p != p1 + 1) break;
X if (*p == '-' && p[1] != ']')
X {
X PATFETCH (c1);
X PATFETCH (c1);
X while (c <= c1)
X b[c / BYTEWIDTH] |= 1 << (c % BYTEWIDTH), c++;
X }
X else
X {
X b[c / BYTEWIDTH] |= 1 << (c % BYTEWIDTH);
X }
X }
X /* Discard any bitmap bytes that are all 0 at the end of the map.
X Decrement the map-length byte too. */
X while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
X b[-1]--;
X b += b[-1];
X break;
X
X case '(':
X if (! (obscure_syntax & RE_NO_BK_PARENS))
X goto normal_char;
X else
X goto handle_open;
X
X case ')':
X if (! (obscure_syntax & RE_NO_BK_PARENS))
X goto normal_char;
X else
X goto handle_close;
X
X case '\n':
X if (! (obscure_syntax & RE_NEWLINE_OR))
X goto normal_char;
X else
X goto handle_bar;
X
X case '|':
X if (! (obscure_syntax & RE_NO_BK_VBAR))
X goto normal_char;
X else
X goto handle_bar;
X
X case '\\':
X if (p == pend) goto invalid_pattern;
X PATFETCH_RAW (c);
X switch (c)
X {
X case '(':
X if (obscure_syntax & RE_NO_BK_PARENS)
X goto normal_backsl;
X handle_open:
X if (stackp == stacke) goto nesting_too_deep;
X if (regnum < RE_NREGS)
X {
X PATPUSH (start_memory);
X PATPUSH (regnum);
X }
X *stackp++ = b - bufp->buffer;
X *stackp++ = fixup_jump ? fixup_jump - bufp->buffer + 1 : 0;
X *stackp++ = regnum++;
X *stackp++ = begalt - bufp->buffer;
X fixup_jump = 0;
X laststart = 0;
X begalt = b;
X break;
X
X case ')':
X if (obscure_syntax & RE_NO_BK_PARENS)
X goto normal_backsl;
X handle_close:
X if (stackp == stackb) goto unmatched_close;
X begalt = *--stackp + bufp->buffer;
X if (fixup_jump)
X store_jump (fixup_jump, jump, b);
X if (stackp[-1] < RE_NREGS)
X {
X PATPUSH (stop_memory);
X PATPUSH (stackp[-1]);
X }
X stackp -= 2;
X fixup_jump = 0;
X if (*stackp)
X fixup_jump = *stackp + bufp->buffer - 1;
X laststart = *--stackp + bufp->buffer;
X break;
X
X case '|':
X if (obscure_syntax & RE_NO_BK_VBAR)
X goto normal_backsl;
X handle_bar:
X insert_jump (on_failure_jump, begalt, b + 6, b);
X pending_exact = 0;
X b += 3;
X if (fixup_jump)
X store_jump (fixup_jump, jump, b);
X fixup_jump = b;
X b += 3;
X laststart = 0;
X begalt = b;
X break;
X
X#ifdef emacs
X case '=':
X PATPUSH (at_dot);
X break;
X
X case 's':
X laststart = b;
X PATPUSH (syntaxspec);
X PATFETCH (c);
X PATPUSH (syntax_spec_code[c]);
X break;
X
X case 'S':
X laststart = b;
X PATPUSH (notsyntaxspec);
X PATFETCH (c);
X PATPUSH (syntax_spec_code[c]);
X break;
X#endif /* emacs */
X
X case 'w':
X laststart = b;
X PATPUSH (wordchar);
X break;
X
X case 'W':
X laststart = b;
X PATPUSH (notwordchar);
X break;
X
X case '<':
X PATPUSH (wordbeg);
X break;
X
X case '>':
X PATPUSH (wordend);
X break;
X
X case 'b':
X PATPUSH (wordbound);
X break;
X
X case 'B':
X PATPUSH (notwordbound);
X break;
X
X case '`':
X PATPUSH (begbuf);
X break;
X
X case '\'':
X PATPUSH (endbuf);
X break;
X
X case '1':
X case '2':
X case '3':
X case '4':
X case '5':
X case '6':
X case '7':
X case '8':
X case '9':
X c1 = c - '0';
X if (c1 >= regnum)
X goto normal_char;
X for (stackt = stackp - 2; stackt > stackb; stackt -= 4)
X if (*stackt == c1)
X goto normal_char;
X laststart = b;
X PATPUSH (duplicate);
X PATPUSH (c1);
X break;
X
X case '+':
X case '?':
X if (obscure_syntax & RE_BK_PLUS_QM)
X goto handle_plus;
X
X default:
X normal_backsl:
X /* You might think it would be useful for \ to mean
X not to translate; but if we don't translate it
X it will never match anything. */
X if (translate) c = translate[c];
X goto normal_char;
X }
X break;
X
X default:
X normal_char:
X if (!pending_exact || pending_exact + *pending_exact + 1 != b
X || *pending_exact == 0177 || *p == '*' || *p == '^'
X || ((obscure_syntax & RE_BK_PLUS_QM)
X ? *p == '\\' && (p[1] == '+' || p[1] == '?')
X : (*p == '+' || *p == '?')))
X {
X laststart = b;
X PATPUSH (exactn);
X pending_exact = b;
X PATPUSH (0);
X }
X PATPUSH (c);
X (*pending_exact)++;
X }
X }
X
X if (fixup_jump)
X store_jump (fixup_jump, jump, b);
X
X if (stackp != stackb) goto unmatched_open;
X
X bufp->used = b - bufp->buffer;
X return 0;
X
X invalid_pattern:
X return "Invalid regular expression";
X
X unmatched_open:
X return "Unmatched \\(";
X
X unmatched_close:
X return "Unmatched \\)";
X
X end_of_pattern:
X return "Premature end of regular expression";
X
X nesting_too_deep:
X return "Nesting too deep";
X
X too_big:
X return "Regular expression too big";
X
X memory_exhausted:
X return "Memory exhausted";
X}
X
X/* Store where `from' points a jump operation to jump to where `to' points.
X `opcode' is the opcode to store. */
X
Xstatic int
Xstore_jump (from, opcode, to)
X char *from, *to;
X char opcode;
X{
X from[0] = opcode;
X from[1] = (to - (from + 3)) & 0377;
X from[2] = (to - (from + 3)) >> 8;
X}
X
X/* Open up space at char FROM, and insert there a jump to TO.
X CURRENT_END gives te end of the storage no in use,
X so we know how much data to copy up.
X OP is the opcode of the jump to insert.
X
X If you call this function, you must zero out pending_exact. */
X
Xstatic int
Xinsert_jump (op, from, to, current_end)
X char op;
X char *from, *to, *current_end;
X{
X register char *pto = current_end + 3;
X register char *pfrom = current_end;
X while (pfrom != from)
X *--pto = *--pfrom;
X store_jump (from, op, to);
X}
X
X/* Given a pattern, compute a fastmap from it.
X The fastmap records which of the (1 << BYTEWIDTH) possible characters
X can start a string that matches the pattern.
X This fastmap is used by re_search to skip quickly over totally implausible text.
X
X The caller must supply the address of a (1 << BYTEWIDTH)-byte data area
X as bufp->fastmap.
X The other components of bufp describe the pattern to be used. */
X
Xvoid
Xre_compile_fastmap (bufp)
X struct re_pattern_buffer *bufp;
X{
X unsigned char *pattern = (unsigned char *) bufp->buffer;
X int size = bufp->used;
X register char *fastmap = bufp->fastmap;
X register unsigned char *p = pattern;
X register unsigned char *pend = pattern + size;
X register int j, k;
X unsigned char *translate = (unsigned char *) bufp->translate;
X
X unsigned char *stackb[NFAILURES];
X unsigned char **stackp = stackb;
X
X bzero (fastmap, (1 << BYTEWIDTH));
X bufp->fastmap_accurate = 1;
X bufp->can_be_null = 0;
X
X while (p)
X {
X if (p == pend)
X {
X bufp->can_be_null = 1;
X break;
X }
X#ifdef SWITCH_ENUM_BUG
X switch ((int) ((enum regexpcode) *p++))
X#else
X switch ((enum regexpcode) *p++)
X#endif
X {
X case exactn:
X if (translate)
X fastmap[translate[p[1]]] = 1;
X else
X fastmap[p[1]] = 1;
X break;
X
X case begline:
X case before_dot:
X case at_dot:
X case after_dot:
X case begbuf:
X case endbuf:
X case wordbound:
X case notwordbound:
X case wordbeg:
X case wordend:
X continue;
X
X case endline:
X if (translate)
X fastmap[translate['\n']] = 1;
X else
X fastmap['\n'] = 1;
X if (bufp->can_be_null != 1)
X bufp->can_be_null = 2;
X break;
X
X case finalize_jump:
X case maybe_finalize_jump:
X case jump:
X case dummy_failure_jump:
X bufp->can_be_null = 1;
X j = *p++ & 0377;
X j += SIGN_EXTEND_CHAR (*(char *)p) << 8;
X p += j + 1; /* The 1 compensates for missing ++ above */
X if (j > 0)
X continue;
X /* Jump backward reached implies we just went through
X the body of a loop and matched nothing.
X Opcode jumped to should be an on_failure_jump.
X Just treat it like an ordinary jump.
X For a * loop, it has pushed its failure point already;
X if so, discard that as redundant. */
X if ((enum regexpcode) *p != on_failure_jump)
X continue;
X p++;
X j = *p++ & 0377;
X j += SIGN_EXTEND_CHAR (*(char *)p) << 8;
X p += j + 1; /* The 1 compensates for missing ++ above */
X if (stackp != stackb && *stackp == p)
X stackp--;
X continue;
X
X case on_failure_jump:
X j = *p++ & 0377;
X j += SIGN_EXTEND_CHAR (*(char *)p) << 8;
X p++;
X *++stackp = p + j;
X continue;
X
X case start_memory:
X case stop_memory:
X p++;
X continue;
X
X case duplicate:
X bufp->can_be_null = 1;
X fastmap['\n'] = 1;
X case anychar:
X for (j = 0; j < (1 << BYTEWIDTH); j++)
X if (j != '\n')
X fastmap[j] = 1;
X if (bufp->can_be_null)
X return;
X /* Don't return; check the alternative paths
X so we can set can_be_null if appropriate. */
X break;
X
X case wordchar:
X for (j = 0; j < (1 << BYTEWIDTH); j++)
X if (SYNTAX (j) == Sword)
X fastmap[j] = 1;
X break;
X
X case notwordchar:
X for (j = 0; j < (1 << BYTEWIDTH); j++)
X if (SYNTAX (j) != Sword)
X fastmap[j] = 1;
X break;
X
X#ifdef emacs
X case syntaxspec:
X k = *p++;
X for (j = 0; j < (1 << BYTEWIDTH); j++)
X if (SYNTAX (j) == (enum syntaxcode) k)
X fastmap[j] = 1;
X break;
X
X case notsyntaxspec:
X k = *p++;
X for (j = 0; j < (1 << BYTEWIDTH); j++)
X if (SYNTAX (j) != (enum syntaxcode) k)
X fastmap[j] = 1;
X break;
X#endif /* emacs */
X
X case charset:
X for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
X if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
X {
X if (translate)
X fastmap[translate[j]] = 1;
X else
X fastmap[j] = 1;
X }
X break;
X
X case charset_not:
X /* Chars beyond end of map must be allowed */
X for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++)
X if (translate)
X fastmap[translate[j]] = 1;
X else
X fastmap[j] = 1;
X
X for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
X if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))))
X {
X if (translate)
X fastmap[translate[j]] = 1;
X else
X fastmap[j] = 1;
X }
X break;
X }
X
X /* Get here means we have successfully found the possible starting characters
X of one path of the pattern. We need not follow this path any farther.
X Instead, look at the next alternative remembered in the stack. */
X if (stackp != stackb)
X p = *stackp--;
X else
X break;
X }
X}
X
X/* Like re_search_2, below, but only one string is specified. */
X
Xint
Xre_search (pbufp, string, size, startpos, range, regs)
X struct re_pattern_buffer *pbufp;
X char *string;
X int size, startpos, range;
X struct re_registers *regs;
X{
X return re_search_2 (pbufp, 0, 0, string, size, startpos, range, regs, size);
X}
X
X/* Like re_match_2 but tries first a match starting at index STARTPOS,
X then at STARTPOS + 1, and so on.
X RANGE is the number of places to try before giving up.
X If RANGE is negative, the starting positions tried are
X STARTPOS, STARTPOS - 1, etc.
X It is up to the caller to make sure that range is not so large
X as to take the starting position outside of the input strings.
X
XThe value returned is the position at which the match was found,
X or -1 if no match was found,
X or -2 if error (such as failure stack overflow). */
X
Xint
Xre_search_2 (pbufp, string1, size1, string2, size2, startpos, range, regs, mstop)
X struct re_pattern_buffer *pbufp;
X char *string1, *string2;
X int size1, size2;
X int startpos;
X register int range;
X struct re_registers *regs;
X int mstop;
X{
X register char *fastmap = pbufp->fastmap;
X register unsigned char *translate = (unsigned char *) pbufp->translate;
X int total = size1 + size2;
X int val;
X
X /* Update the fastmap now if not correct already */
X if (fastmap && !pbufp->fastmap_accurate)
X re_compile_fastmap (pbufp);
X
X /* Don't waste time in a long search for a pattern
X that says it is anchored. */
X if (pbufp->used > 0 && (enum regexpcode) pbufp->buffer[0] == begbuf
X && range > 0)
X {
X if (startpos > 0)
X return -1;
X else
X range = 1;
X }
X
X while (1)
X {
X /* If a fastmap is supplied, skip quickly over characters
X that cannot possibly be the start of a match.
X Note, however, that if the pattern can possibly match
X the null string, we must test it at each starting point
X so that we take the first null string we get. */
X
X if (fastmap && startpos < total && pbufp->can_be_null != 1)
X {
X if (range > 0)
X {
X register int lim = 0;
X register unsigned char *p;
X int irange = range;
X if (startpos < size1 && startpos + range >= size1)
X lim = range - (size1 - startpos);
X
X p = ((unsigned char *)
X &(startpos >= size1 ? string2 - size1 : string1)[startpos]);
X
X if (translate)
X {
X while (range > lim && !fastmap[translate[*p++]])
X range--;
X }
X else
X {
X while (range > lim && !fastmap[*p++])
X range--;
X }
X startpos += irange - range;
X }
X else
X {
X register unsigned char c;
X if (startpos >= size1)
X c = string2[startpos - size1];
X else
X c = string1[startpos];
X c &= 0xff;
X if (translate ? !fastmap[translate[c]] : !fastmap[c])
X goto advance;
X }
X }
X
X if (range >= 0 && startpos == total
X && fastmap && pbufp->can_be_null == 0)
X return -1;
X
X val = re_match_2 (pbufp, string1, size1, string2, size2, startpos, regs, mstop);
X if (0 <= val)
X {
X if (val == -2)
X return -2;
X return startpos;
X }
X
X#ifdef C_ALLOCA
X alloca (0);
X#endif /* C_ALLOCA */
X
X advance:
X if (!range) break;
X if (range > 0) range--, startpos++; else range++, startpos--;
X }
X return -1;
X}
X
X#ifndef emacs /* emacs never uses this */
Xint
Xre_match (pbufp, string, size, pos, regs)
X struct re_pattern_buffer *pbufp;
X char *string;
X int size, pos;
X struct re_registers *regs;
X{
X return re_match_2 (pbufp, 0, 0, string, size, pos, regs, size);
X}
X#endif /* emacs */
X
X/* Maximum size of failure stack. Beyond this, overflow is an error. */
X
Xint re_max_failures = 2000;
X
Xstatic int bcmp_translate();
X/* Match the pattern described by PBUFP
X against data which is the virtual concatenation of STRING1 and STRING2.
X SIZE1 and SIZE2 are the sizes of the two data strings.
X Start the match at position POS.
X Do not consider matching past the position MSTOP.
X
X If pbufp->fastmap is nonzero, then it had better be up to date.
X
X The reason that the data to match are specified as two components
X which are to be regarded as concatenated
X is so this function can be used directly on the contents of an Emacs buffer.
X
X -1 is returned if there is no match. -2 is returned if there is
X an error (such as match stack overflow). Otherwise the value is the length
X of the substring which was matched. */
X
Xint
Xre_match_2 (pbufp, string1, size1, string2, size2, pos, regs, mstop)
X struct re_pattern_buffer *pbufp;
X unsigned char *string1, *string2;
X int size1, size2;
X int pos;
X struct re_registers *regs;
X int mstop;
X{
X register unsigned char *p = (unsigned char *) pbufp->buffer;
X register unsigned char *pend = p + pbufp->used;
X /* End of first string */
X unsigned char *end1;
X /* End of second string */
X unsigned char *end2;
X /* Pointer just past last char to consider matching */
X unsigned char *end_match_1, *end_match_2;
X register unsigned char *d, *dend;
X register int mcnt;
X unsigned char *translate = (unsigned char *) pbufp->translate;
X
X /* Failure point stack. Each place that can handle a failure further down the line
X pushes a failure point on this stack. It consists of two char *'s.
X The first one pushed is where to resume scanning the pattern;
X the second pushed is where to resume scanning the strings.
X If the latter is zero, the failure point is a "dummy".
X If a failure happens and the innermost failure point is dormant,
X it discards that failure point and tries the next one. */
X
X unsigned char *initial_stack[2 * NFAILURES];
X unsigned char **stackb = initial_stack;
X unsigned char **stackp = stackb, **stacke = &stackb[2 * NFAILURES];
X
X /* Information on the "contents" of registers.
X These are pointers into the input strings; they record
X just what was matched (on this attempt) by some part of the pattern.
X The start_memory command stores the start of a register's contents
X and the stop_memory command stores the end.
X
X At that point, regstart[regnum] points to the first character in the register,
X regend[regnum] points to the first character beyond the end of the register,
X regstart_seg1[regnum] is true iff regstart[regnum] points into string1,
X and regend_seg1[regnum] is true iff regend[regnum] points into string1. */
X
X unsigned char *regstart[RE_NREGS];
X unsigned char *regend[RE_NREGS];
X unsigned char regstart_seg1[RE_NREGS], regend_seg1[RE_NREGS];
X
X /* Set up pointers to ends of strings.
X Don't allow the second string to be empty unless both are empty. */
X if (!size2)
X {
X string2 = string1;
X size2 = size1;
X string1 = 0;
X size1 = 0;
X }
X end1 = string1 + size1;
X end2 = string2 + size2;
X
X /* Compute where to stop matching, within the two strings */
X if (mstop <= size1)
X {
X end_match_1 = string1 + mstop;
X end_match_2 = string2;
X }
X else
X {
X end_match_1 = end1;
X end_match_2 = string2 + mstop - size1;
X }
X
X /* Initialize \) text positions to -1
X to mark ones that no \( or \) has been seen for. */
X
X for (mcnt = 0; mcnt < sizeof (regend) / sizeof (*regend); mcnt++)
X regend[mcnt] = (unsigned char *) -1;
X
X /* `p' scans through the pattern as `d' scans through the data.
X `dend' is the end of the input string that `d' points within.
X `d' is advanced into the following input string whenever necessary,
X but this happens before fetching;
X therefore, at the beginning of the loop,
X `d' can be pointing at the end of a string,
X but it cannot equal string2. */
X
X if (pos <= size1)
X d = string1 + pos, dend = end_match_1;
X else
X d = string2 + pos - size1, dend = end_match_2;
X
X/* Write PREFETCH; just before fetching a character with *d. */
X#define PREFETCH \
X while (d == dend) \
X { if (dend == end_match_2) goto fail; /* end of string2 => failure */ \
X d = string2; /* end of string1 => advance to string2. */ \
X dend = end_match_2; }
X
X /* This loop loops over pattern commands.
X It exits by returning from the function if match is complete,
X or it drops through if match fails at this starting point in the input data. */
X
X while (1)
X {
X if (p == pend)
X /* End of pattern means we have succeeded! */
X {
X /* If caller wants register contents data back, convert it to indices */
X if (regs)
X {
X regs->start[0] = pos;
X if (dend == end_match_1)
X regs->end[0] = d - string1;
X else
X regs->end[0] = d - string2 + size1;
X for (mcnt = 1; mcnt < RE_NREGS; mcnt++)
X {
X if (regend[mcnt] == (unsigned char *) -1)
X {
X regs->start[mcnt] = -1;
X regs->end[mcnt] = -1;
X continue;
X }
X if (regstart_seg1[mcnt])
X regs->start[mcnt] = regstart[mcnt] - string1;
X else
X regs->start[mcnt] = regstart[mcnt] - string2 + size1;
X if (regend_seg1[mcnt])
X regs->end[mcnt] = regend[mcnt] - string1;
X else
X regs->end[mcnt] = regend[mcnt] - string2 + size1;
X }
X }
X if (dend == end_match_1)
X return (d - string1 - pos);
X else
X return d - string2 + size1 - pos;
X }
X
X /* Otherwise match next pattern command */
X#ifdef SWITCH_ENUM_BUG
X switch ((int) ((enum regexpcode) *p++))
X#else
X switch ((enum regexpcode) *p++)
X#endif
X {
X
X /* \( is represented by a start_memory, \) by a stop_memory.
X Both of those commands contain a "register number" argument.
X The text matched within the \( and \) is recorded under that number.
X Then, \<digit> turns into a `duplicate' command which
X is followed by the numeric value of <digit> as the register number. */
X
X case start_memory:
X regstart[*p] = d;
X regstart_seg1[*p++] = (dend == end_match_1);
X break;
X
X case stop_memory:
X regend[*p] = d;
X regend_seg1[*p++] = (dend == end_match_1);
X break;
X
X case duplicate:
X {
X int regno = *p++; /* Get which register to match against */
X register unsigned char *d2, *dend2;
X
X d2 = regstart[regno];
X dend2 = ((regstart_seg1[regno] == regend_seg1[regno])
X ? regend[regno] : end_match_1);
X while (1)
X {
X /* Advance to next segment in register contents, if necessary */
X while (d2 == dend2)
X {
X if (dend2 == end_match_2) break;
X if (dend2 == regend[regno]) break;
X d2 = string2, dend2 = regend[regno]; /* end of string1 => advance to string2. */
X }
X /* At end of register contents => success */
X if (d2 == dend2) break;
X
X /* Advance to next segment in data being matched, if necessary */
X PREFETCH;
X
X /* mcnt gets # consecutive chars to compare */
X mcnt = dend - d;
X if (mcnt > dend2 - d2)
X mcnt = dend2 - d2;
X /* Compare that many; failure if mismatch, else skip them. */
X if (translate ? bcmp_translate (d, d2, mcnt, translate) : bcmp (d, d2, mcnt))
X goto fail;
X d += mcnt, d2 += mcnt;
X }
X }
X break;
X
X case anychar:
X /* fetch a data character */
X PREFETCH;
X /* Match anything but a newline. */
X if ((translate ? translate[*d++] : *d++) == '\n')
X goto fail;
X break;
X
X case charset:
X case charset_not:
X {
X /* Nonzero for charset_not */
X int not = 0;
X register int c;
X if (*(p - 1) == (unsigned char) charset_not)
X not = 1;
X
X /* fetch a data character */
X PREFETCH;
X
X if (translate)
X c = translate [*d];
X else
X c = *d;
X
X if (c < *p * BYTEWIDTH
X && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
X not = !not;
X
X p += 1 + *p;
X
X if (!not) goto fail;
X d++;
X break;
X }
X
X case begline:
X if (d == string1 || d[-1] == '\n')
X break;
X goto fail;
X
X case endline:
X if (d == end2
X || (d == end1 ? (size2 == 0 || *string2 == '\n') : *d == '\n'))
X break;
X goto fail;
X
X /* "or" constructs ("|") are handled by starting each alternative
X with an on_failure_jump that points to the start of the next alternative.
X Each alternative except the last ends with a jump to the joining point.
X (Actually, each jump except for the last one really jumps
X to the following jump, because tensioning the jumps is a hassle.) */
X
X /* The start of a stupid repeat has an on_failure_jump that points
X past the end of the repeat text.
X This makes a failure point so that, on failure to match a repetition,
X matching restarts past as many repetitions have been found
X with no way to fail and look for another one. */
X
X /* A smart repeat is similar but loops back to the on_failure_jump
X so that each repetition makes another failure point. */
X
X case on_failure_jump:
X if (stackp == stacke)
X {
X unsigned char **stackx;
X if (stacke - stackb > re_max_failures * 2)
X return -2;
X stackx = (unsigned char **) alloca (2 * (stacke - stackb)
X * sizeof (char *));
X bcopy (stackb, stackx, (stacke - stackb) * sizeof (char *));
X stackp = stackx + (stackp - stackb);
X stacke = stackx + 2 * (stacke - stackb);
X stackb = stackx;
X }
X mcnt = *p++ & 0377;
X mcnt += SIGN_EXTEND_CHAR (*(char *)p) << 8;
X p++;
X *stackp++ = mcnt + p;
X *stackp++ = d;
X break;
X
X /* The end of a smart repeat has an maybe_finalize_jump back.
X Change it either to a finalize_jump or an ordinary jump. */
X
X case maybe_finalize_jump:
X mcnt = *p++ & 0377;
X mcnt += SIGN_EXTEND_CHAR (*(char *)p) << 8;
X p++;
X {
X register unsigned char *p2 = p;
X /* Compare what follows with the begining of the repeat.
X If we can establish that there is nothing that they would
X both match, we can change to finalize_jump */
X while (p2 != pend
X && (*p2 == (unsigned char) stop_memory
X || *p2 == (unsigned char) start_memory))
X p2++;
X if (p2 == pend)
X p[-3] = (unsigned char) finalize_jump;
X else if (*p2 == (unsigned char) exactn
X || *p2 == (unsigned char) endline)
X {
X register int c = *p2 == (unsigned char) endline ? '\n' : p2[2];
X register unsigned char *p1 = p + mcnt;
X /* p1[0] ... p1[2] are an on_failure_jump.
X Examine what follows that */
X if (p1[3] == (unsigned char) exactn && p1[5] != c)
X p[-3] = (unsigned char) finalize_jump;
X else if (p1[3] == (unsigned char) charset
X || p1[3] == (unsigned char) charset_not)
X {
X int not = p1[3] == (unsigned char) charset_not;
X if (c < p1[4] * BYTEWIDTH
X && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
X not = !not;
X /* not is 1 if c would match */
X /* That means it is not safe to finalize */
X if (!not)
X p[-3] = (unsigned char) finalize_jump;
X }
X }
X }
X p -= 2;
X if (p[-1] != (unsigned char) finalize_jump)
X {
X p[-1] = (unsigned char) jump;
X goto nofinalize;
X }
X
X /* The end of a stupid repeat has a finalize-jump
X back to the start, where another failure point will be made
X which will point after all the repetitions found so far. */
X
X case finalize_jump:
X stackp -= 2;
X
X case jump:
X nofinalize:
X mcnt = *p++ & 0377;
X mcnt += SIGN_EXTEND_CHAR (*(char *)p) << 8;
X p += mcnt + 1; /* The 1 compensates for missing ++ above */
X break;
X
X case dummy_failure_jump:
X if (stackp == stacke)
X {
X unsigned char **stackx
X = (unsigned char **) alloca (2 * (stacke - stackb)
X * sizeof (char *));
X bcopy (stackb, stackx, (stacke - stackb) * sizeof (char *));
X stackp = stackx + (stackp - stackb);
X stacke = stackx + 2 * (stacke - stackb);
X stackb = stackx;
X }
X *stackp++ = 0;
X *stackp++ = 0;
X goto nofinalize;
X
X case wordbound:
X if (d == string1 /* Points to first char */
X || d == end2 /* Points to end */
X || (d == end1 && size2 == 0)) /* Points to end */
X break;
X if ((SYNTAX (d[-1]) == Sword)
X != (SYNTAX (d == end1 ? *string2 : *d) == Sword))
X break;
X goto fail;
X
X case notwordbound:
X if (d == string1 /* Points to first char */
X || d == end2 /* Points to end */
X || (d == end1 && size2 == 0)) /* Points to end */
X goto fail;
X if ((SYNTAX (d[-1]) == Sword)
X != (SYNTAX (d == end1 ? *string2 : *d) == Sword))
X goto fail;
X break;
X
X case wordbeg:
X if (d == end2 /* Points to end */
X || (d == end1 && size2 == 0) /* Points to end */
X || SYNTAX (* (d == end1 ? string2 : d)) != Sword) /* Next char not a letter */
X goto fail;
X if (d == string1 /* Points to first char */
X || SYNTAX (d[-1]) != Sword) /* prev char not letter */
X break;
X goto fail;
X
X case wordend:
X if (d == string1 /* Points to first char */
X || SYNTAX (d[-1]) != Sword) /* prev char not letter */
X goto fail;
X if (d == end2 /* Points to end */
X || (d == end1 && size2 == 0) /* Points to end */
X || SYNTAX (d == end1 ? *string2 : *d) != Sword) /* Next char not a letter */
X break;
X goto fail;
X
X#ifdef emacs
X case before_dot:
X if (((d - string2 <= (unsigned) size2)
X ? d - bf_p2 : d - bf_p1)
X <= point)
X goto fail;
X break;
X
X case at_dot:
X if (((d - string2 <= (unsigned) size2)
X ? d - bf_p2 : d - bf_p1)
X == point)
X goto fail;
X break;
X
X case after_dot:
X if (((d - string2 <= (unsigned) size2)
X ? d - bf_p2 : d - bf_p1)
X >= point)
X goto fail;
X break;
X
X case wordchar:
X mcnt = (int) Sword;
X goto matchsyntax;
X
X case syntaxspec:
X mcnt = *p++;
X matchsyntax:
X PREFETCH;
X if (SYNTAX (*d++) != (enum syntaxcode) mcnt) goto fail;
X break;
X
X case notwordchar:
X mcnt = (int) Sword;
X goto matchnotsyntax;
X
X case notsyntaxspec:
X mcnt = *p++;
X matchnotsyntax:
X PREFETCH;
X if (SYNTAX (*d++) == (enum syntaxcode) mcnt) goto fail;
X break;
X#else
X case wordchar:
X PREFETCH;
X if (SYNTAX (*d++) == 0) goto fail;
X break;
X
X case notwordchar:
X PREFETCH;
X if (SYNTAX (*d++) != 0) goto fail;
X break;
X#endif /* not emacs */
X
X case begbuf:
X if (d == string1) /* Note, d cannot equal string2 */
X break; /* unless string1 == string2. */
X goto fail;
X
X case endbuf:
X if (d == end2 || (d == end1 && size2 == 0))
X break;
X goto fail;
X
X case exactn:
X /* Match the next few pattern characters exactly.
X mcnt is how many characters to match. */
X mcnt = *p++;
X if (translate)
X {
X do
X {
X PREFETCH;
X if (translate[*d++] != *p++) goto fail;
X }
X while (--mcnt);
X }
X else
X {
X do
X {
X PREFETCH;
X if (*d++ != *p++) goto fail;
X }
X while (--mcnt);
X }
X break;
X }
X continue; /* Successfully matched one pattern command; keep matching */
X
X /* Jump here if any matching operation fails. */
X fail:
X if (stackp != stackb)
X /* A restart point is known. Restart there and pop it. */
X {
X if (!stackp[-2])
X { /* If innermost failure point is dormant, flush it and keep looking */
X stackp -= 2;
X goto fail;
X }
X d = *--stackp;
X p = *--stackp;
X if (d >= string1 && d <= end1)
X dend = end_match_1;
X }
X else break; /* Matching at this starting point really fails! */
X }
X return -1; /* Failure to match */
X}
X
Xstatic int
Xbcmp_translate (s1, s2, len, translate)
X unsigned char *s1, *s2;
X register int len;
X unsigned char *translate;
X{
X register unsigned char *p1 = s1, *p2 = s2;
X while (len)
X {
X if (translate [*p1++] != translate [*p2++]) return 1;
X len--;
X }
X return 0;
X}
X
X/* Entry points compatible with bsd4.2 regex library */
X
X#ifndef emacs
X
Xstatic struct re_pattern_buffer re_comp_buf;
X
Xchar *
Xre_comp (s)
X char *s;
X{
X if (!s)
X {
X if (!re_comp_buf.buffer)
X return "No previous regular expression";
X return 0;
X }
X
X if (!re_comp_buf.buffer)
X {
X if (!(re_comp_buf.buffer = (char *) malloc (200)))
X return "Memory exhausted";
X re_comp_buf.allocated = 200;
X if (!(re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH)))
X return "Memory exhausted";
X }
X return re_compile_pattern (s, strlen (s), &re_comp_buf);
X}
X
Xint
Xre_exec (s)
X char *s;
X{
X int len = strlen (s);
X return 0 <= re_search (&re_comp_buf, s, len, 0, len, 0);
X}
X
X#endif /* emacs */
X
X#ifdef test
X
X#include <stdio.h>
X
X/* Indexed by a character, gives the upper case equivalent of the character */
X
Xstatic char upcase[0400] =
X { 000, 001, 002, 003, 004, 005, 006, 007,
X 010, 011, 012, 013, 014, 015, 016, 017,
X 020, 021, 022, 023, 024, 025, 026, 027,
X 030, 031, 032, 033, 034, 035, 036, 037,
X 040, 041, 042, 043, 044, 045, 046, 047,
X 050, 051, 052, 053, 054, 055, 056, 057,
X 060, 061, 062, 063, 064, 065, 066, 067,
X 070, 071, 072, 073, 074, 075, 076, 077,
X 0100, 0101, 0102, 0103, 0104, 0105, 0106, 0107,
X 0110, 0111, 0112, 0113, 0114, 0115, 0116, 0117,
X 0120, 0121, 0122, 0123, 0124, 0125, 0126, 0127,
X 0130, 0131, 0132, 0133, 0134, 0135, 0136, 0137,
X 0140, 0101, 0102, 0103, 0104, 0105, 0106, 0107,
X 0110, 0111, 0112, 0113, 0114, 0115, 0116, 0117,
X 0120, 0121, 0122, 0123, 0124, 0125, 0126, 0127,
X 0130, 0131, 0132, 0173, 0174, 0175, 0176, 0177,
X 0200, 0201, 0202, 0203, 0204, 0205, 0206, 0207,
X 0210, 0211, 0212, 0213, 0214, 0215, 0216, 0217,
X 0220, 0221, 0222, 0223, 0224, 0225, 0226, 0227,
X 0230, 0231, 0232, 0233, 0234, 0235, 0236, 0237,
X 0240, 0241, 0242, 0243, 0244, 0245, 0246, 0247,
X 0250, 0251, 0252, 0253, 0254, 0255, 0256, 0257,
X 0260, 0261, 0262, 0263, 0264, 0265, 0266, 0267,
X 0270, 0271, 0272, 0273, 0274, 0275, 0276, 0277,
X 0300, 0301, 0302, 0303, 0304, 0305, 0306, 0307,
X 0310, 0311, 0312, 0313, 0314, 0315, 0316, 0317,
X 0320, 0321, 0322, 0323, 0324, 0325, 0326, 0327,
X 0330, 0331, 0332, 0333, 0334, 0335, 0336, 0337,
X 0340, 0341, 0342, 0343, 0344, 0345, 0346, 0347,
X 0350, 0351, 0352, 0353, 0354, 0355, 0356, 0357,
X 0360, 0361, 0362, 0363, 0364, 0365, 0366, 0367,
X 0370, 0371, 0372, 0373, 0374, 0375, 0376, 0377
X };
X
Xmain (argc, argv)
X int argc;
X char **argv;
X{
X char pat[80];
X struct re_pattern_buffer buf;
X int i;
X char c;
X char fastmap[(1 << BYTEWIDTH)];
X
X /* Allow a command argument to specify the style of syntax. */
X if (argc > 1)
X obscure_syntax = atoi (argv[1]);
X
X buf.allocated = 40;
X buf.buffer = (char *) malloc (buf.allocated);
X buf.fastmap = fastmap;
X buf.translate = upcase;
X
X while (1)
X {
X gets (pat);
X
X if (*pat)
X {
X re_compile_pattern (pat, strlen(pat), &buf);
X
X for (i = 0; i < buf.used; i++)
X printchar (buf.buffer[i]);
X
X putchar ('\n');
X
X printf ("%d allocated, %d used.\n", buf.allocated, buf.used);
X
X re_compile_fastmap (&buf);
X printf ("Allowed by fastmap: ");
X for (i = 0; i < (1 << BYTEWIDTH); i++)
X if (fastmap[i]) printchar (i);
X putchar ('\n');
X }
X
X gets (pat); /* Now read the string to match against */
X
X i = re_match (&buf, pat, strlen (pat), 0, 0);
X printf ("Match value %d.\n", i);
X }
X}
X
X#ifdef NOTDEF
Xprint_buf (bufp)
X struct re_pattern_buffer *bufp;
X{
X int i;
X
X printf ("buf is :\n----------------\n");
X for (i = 0; i < bufp->used; i++)
X printchar (bufp->buffer[i]);
X
X printf ("\n%d allocated, %d used.\n", bufp->allocated, bufp->used);
X
X printf ("Allowed by fastmap: ");
X for (i = 0; i < (1 << BYTEWIDTH); i++)
X if (bufp->fastmap[i])
X printchar (i);
X printf ("\nAllowed by translate: ");
X if (bufp->translate)
X for (i = 0; i < (1 << BYTEWIDTH); i++)
X if (bufp->translate[i])
X printchar (i);
X printf ("\nfastmap is%s accurate\n", bufp->fastmap_accurate ? "" : "n't");
X printf ("can %s be null\n----------", bufp->can_be_null ? "" : "not");
X}
X#endif
X
Xprintchar (c)
X char c;
X{
X if (c < 041 || c >= 0177)
X {
X putchar ('\\');
X putchar (((c >> 6) & 3) + '0');
X putchar (((c >> 3) & 7) + '0');
X putchar ((c & 7) + '0');
X }
X else
X putchar (c);
X}
X
Xerror (string)
X char *string;
X{
X puts (string);
X exit (1);
X}
X
X#endif /* test */
END_OF_regex.c
if test 44778 -ne `wc -c <regex.c`; then
echo shar: \"regex.c\" unpacked with wrong size!
fi
# end of overwriting check
fi
if test ! -d tests ; then
echo shar: Creating directory \"tests\"
mkdir tests
fi
echo shar: End of archive 1 \(of 4\).
cp /dev/null ark1isdone
MISSING=""
for I in 1 2 3 4 ; do
if test ! -f ark${I}isdone ; then
MISSING="${MISSING} ${I}"
fi
done
if test "${MISSING}" = "" ; then
echo You have unpacked all 4 archives.
rm -f ark[1-9]isdone
else
echo You still need to unpack the following archives:
echo " " ${MISSING}
fi
## End of shell archive.
exit 0