0.12. (Implements POSIX draft P10003.2/D11.2, except for
internationalization features.)
- Copyright (C) 1993, 1994, 1995, 1996, 1997 Free Software Foundation, Inc.
+ Copyright (C) 1993, 1994, 1995, 1996, 1997, 1998 Free Software Foundation, Inc.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
#undef _GNU_SOURCE
#define _GNU_SOURCE
+#ifdef emacs
/* Converts the pointer to the char to BEG-based offset from the start. */
#define PTR_TO_OFFSET(d) \
POS_AS_IN_BUFFER (MATCHING_IN_FIRST_STRING \
? (d) - string1 : (d) - (string2 - size1))
-#define POS_AS_IN_BUFFER(p) ((p) + 1)
+#define POS_AS_IN_BUFFER(p) ((p) + (NILP (re_match_object) || BUFFERP (re_match_object)))
+#else
+#define PTR_TO_OFFSET(d) 0
+#endif
#ifdef HAVE_CONFIG_H
#include <config.h>
#include "category.h"
#define malloc xmalloc
+#define realloc xrealloc
#define free xfree
#else /* not emacs */
#define SYNTAX(c) re_syntax_table[c]
-/* Dummy macro for non emacs environments. */
+/* Dummy macros for non-Emacs environments. */
#define BASE_LEADING_CODE_P(c) (0)
#define WORD_BOUNDARY_P(c1, c2) (0)
#define CHAR_HEAD_P(p) (1)
REGEX_ALLOCATE_STACK. */
-/* Number of failure points for which to initially allocate space
+/* Approximate number of failure points for which to initially allocate space
when matching. If this number is exceeded, we allocate more
space, so it is not a hard limit. */
#ifndef INIT_FAILURE_ALLOC
-#define INIT_FAILURE_ALLOC 5
+#define INIT_FAILURE_ALLOC 20
#endif
/* Roughly the maximum number of failure points on the stack. Would be
- exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
+ exactly that if always used TYPICAL_FAILURE_SIZE items each time we failed.
This is a variable only so users of regex can assign to it; we never
change it ourselves. */
#if defined (MATCH_MAY_ALLOCATE)
-/* 4400 was enough to cause a crash on Alpha OSF/1,
- whose default stack limit is 2mb. */
-int re_max_failures = 20000;
+/* Note that 4400 is enough to cause a crash on Alpha OSF/1,
+ whose default stack limit is 2mb. In order for a larger
+ value to work reliably, you have to try to make it accord
+ with the process stack limit. */
+int re_max_failures = 40000;
#else
-int re_max_failures = 2000;
+int re_max_failures = 4000;
#endif
union fail_stack_elt
#define INIT_FAIL_STACK() \
do { \
fail_stack.stack = (fail_stack_elt_t *) \
- REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (fail_stack_elt_t)); \
+ REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * TYPICAL_FAILURE_SIZE \
+ * sizeof (fail_stack_elt_t)); \
\
if (fail_stack.stack == NULL) \
return -2; \
#endif
-/* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
+/* Double the size of FAIL_STACK, up to a limit
+ which allows approximately `re_max_failures' items.
Return 1 if succeeds, and 0 if either ran out of memory
allocating space for it or it was already too large.
REGEX_REALLOCATE_STACK requires `destination' be declared. */
-#define DOUBLE_FAIL_STACK(fail_stack) \
- ((fail_stack).size > re_max_failures * MAX_FAILURE_ITEMS \
+/* Factor to increase the failure stack size by
+ when we increase it.
+ This used to be 2, but 2 was too wasteful
+ because the old discarded stacks added up to as much space
+ were as ultimate, maximum-size stack. */
+#define FAIL_STACK_GROWTH_FACTOR 4
+
+#define GROW_FAIL_STACK(fail_stack) \
+ (((fail_stack).size * sizeof (fail_stack_elt_t) \
+ >= re_max_failures * TYPICAL_FAILURE_SIZE) \
? 0 \
- : ((fail_stack).stack = (fail_stack_elt_t *) \
+ : ((fail_stack).stack \
+ = (fail_stack_elt_t *) \
REGEX_REALLOCATE_STACK ((fail_stack).stack, \
(fail_stack).size * sizeof (fail_stack_elt_t), \
- ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \
+ MIN (re_max_failures * TYPICAL_FAILURE_SIZE, \
+ ((fail_stack).size * sizeof (fail_stack_elt_t) \
+ * FAIL_STACK_GROWTH_FACTOR))), \
\
(fail_stack).stack == NULL \
? 0 \
- : ((fail_stack).size <<= 1, \
+ : ((fail_stack).size \
+ = (MIN (re_max_failures * TYPICAL_FAILURE_SIZE, \
+ ((fail_stack).size * sizeof (fail_stack_elt_t) \
+ * FAIL_STACK_GROWTH_FACTOR)) \
+ / sizeof (fail_stack_elt_t)), \
1)))
space to do so. */
#define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
((FAIL_STACK_FULL () \
- && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
+ && !GROW_FAIL_STACK (FAIL_STACK)) \
? 0 \
: ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1))
if we ever fail back to it.
Requires variables fail_stack, regstart, regend, reg_info, and
- num_regs be declared. DOUBLE_FAIL_STACK requires `destination' be
+ num_regs be declared. GROW_FAIL_STACK requires `destination' be
declared.
Does `return FAILURE_CODE' if runs out of memory. */
/* Ensure we have enough space allocated for what we will push. */ \
while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
{ \
- if (!DOUBLE_FAIL_STACK (fail_stack)) \
+ if (!GROW_FAIL_STACK (fail_stack)) \
return failure_code; \
\
DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \
#define NUM_NONREG_ITEMS 4
#endif
-/* We push at most this many items on the stack. */
-/* We used to use (num_regs - 1), which is the number of registers
- this regexp will save; but that was changed to 5
- to avoid stack overflow for a regexp with lots of parens. */
-#define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
+/* Estimate the size of data pushed by a typical failure stack entry.
+ An estimate is all we need, because all we use this for
+ is to choose a limit for how big to make the failure stack. */
-/* We actually push this many items. */
+#define TYPICAL_FAILURE_SIZE 20
+
+/* This is how many items we actually use for a failure point.
+ It depends on the regexp. */
#define NUM_FAILURE_ITEMS \
(((0 \
? 0 : highest_active_reg - lowest_active_reg + 1) \
#define PATFETCH(c) \
do {if (p == pend) return REG_EEND; \
c = (unsigned char) *p++; \
- if (translate) c = (unsigned char) translate[c]; \
+ if (RE_TRANSLATE_P (translate)) c = RE_TRANSLATE (translate, c); \
} while (0)
#endif
when we use a character as a subscript we must make it unsigned. */
#ifndef TRANSLATE
#define TRANSLATE(d) \
- (translate ? (unsigned char) translate[(unsigned char) (d)] : (d))
+ (RE_TRANSLATE_P (translate) \
+ ? (unsigned) RE_TRANSLATE (translate, (unsigned) (d)) : (d))
#endif
compile_stack_type compile_stack;
/* Points to the current (ending) position in the pattern. */
+#ifdef AIX
+ /* `const' makes AIX compiler fail. */
+ char *p = pattern;
+#else
const char *p = pattern;
+#endif
const char *pend = pattern + size;
/* How to translate the characters in the pattern. */
incremented `p', by the way, to be the character after
the `*'. Do we have to do something analogous here
for null bytes, because of RE_DOT_NOT_NULL? */
- if (TRANSLATE (*(p - 2)) == TRANSLATE ('.')
+ if (TRANSLATE ((unsigned char)*(p - 2)) == TRANSLATE ('.')
&& zero_times_ok
- && p < pend && TRANSLATE (*p) == TRANSLATE ('\n')
+ && p < pend
+ && TRANSLATE ((unsigned char)*p) == TRANSLATE ('\n')
&& !(syntax & RE_DOT_NEWLINE))
{ /* We have .*\n. */
STORE_JUMP (jump, b, laststart);
}
else
{
- /* Could be the end of the bracket expression. If it's
- not (i.e., when the bracket expression is `[]' so
- far), the ']' character bit gets set way below. */
- if (c == ']' && p != p1 + 1)
- break;
+ /* Could be the end of the bracket expression. If it's
+ not (i.e., when the bracket expression is `[]' so
+ far), the ']' character bit gets set way below. */
+ if (c == ']' && p != p1 + 1)
+ break;
}
/* If C indicates start of multibyte char, get the
else if (!escaped_char &&
syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
- { /* Leave room for the null. */
+ {
+ /* Leave room for the null. */
char str[CHAR_CLASS_MAX_LENGTH + 1];
PATFETCH (c);
p += len;
}
- if (!SAME_CHARSET_P (c, c1))
+ if (SINGLE_BYTE_CHAR_P (c)
+ && ! SINGLE_BYTE_CHAR_P (c1))
+ {
+ /* Handle a range such as \177-\377 in multibyte mode.
+ Split that into two ranges,,
+ the low one ending at 0237, and the high one
+ starting at ...040. */
+ int c1_base = (c1 & ~0177) | 040;
+ SET_RANGE_TABLE_WORK_AREA (range_table_work, c, c1);
+ c1 = 0237;
+ }
+ else if (!SAME_CHARSET_P (c, c1))
FREE_STACK_RETURN (REG_ERANGE);
}
else
for (this_char = range_start; this_char <= range_end;
this_char++)
SET_LIST_BIT (TRANSLATE (this_char));
+ }
}
- }
else
/* ... into range table. */
SET_RANGE_TABLE_WORK_AREA (range_table_work, c, c1);
p1 = p - 1; /* P1 points the head of C. */
#ifdef emacs
if (bufp->multibyte)
- /* Set P to the next character boundary. */
- p += MULTIBYTE_FORM_LENGTH (p1, pend - p1) - 1;
+ {
+ c = STRING_CHAR (p1, pend - p1);
+ c = TRANSLATE (c);
+ /* Set P to the next character boundary. */
+ p += MULTIBYTE_FORM_LENGTH (p1, pend - p1) - 1;
+ }
#endif
/* If no exactn currently being built. */
if (!pending_exact
|| *pending_exact >= (1 << BYTEWIDTH) - (p - p1)
/* If followed by a repetition operator. */
- || *p == '*' || *p == '^'
+ || (p != pend && (*p == '*' || *p == '^'))
|| ((syntax & RE_BK_PLUS_QM)
- ? *p == '\\' && (p[1] == '+' || p[1] == '?')
- : (*p == '+' || *p == '?'))
+ ? p + 1 < pend && *p == '\\' && (p[1] == '+' || p[1] == '?')
+ : p != pend && (*p == '+' || *p == '?'))
|| ((syntax & RE_INTERVALS)
&& ((syntax & RE_NO_BK_BRACES)
- ? *p == '{'
- : (p[0] == '\\' && p[1] == '{'))))
+ ? p != pend && *p == '{'
+ : p + 1 < pend && p[0] == '\\' && p[1] == '{')))
{
/* Start building a new exactn. */
pending_exact = b - 1;
}
- /* Here, C may translated, therefore C may not equal to *P1. */
- while (1)
+#ifdef emacs
+ if (! SINGLE_BYTE_CHAR_P (c))
{
- BUF_PUSH (c);
- (*pending_exact)++;
- if (++p1 == p)
- break;
-
- /* Rest of multibyte form should be copied literally. */
- c = *(unsigned char *)p1;
+ unsigned char work[4], *str;
+ int i = CHAR_STRING (c, work, str);
+ int j;
+ for (j = 0; j < i; j++)
+ {
+ BUF_PUSH (str[j]);
+ (*pending_exact)++;
+ }
+ }
+ else
+#endif
+ {
+ BUF_PUSH (c);
+ (*pending_exact)++;
}
break;
} /* switch (c) */
{
int num_regs = bufp->re_nsub + 1;
- /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
- is strictly greater than re_max_failures, the largest possible stack
- is 2 * re_max_failures failure points. */
- if (fail_stack.size < (2 * re_max_failures * MAX_FAILURE_ITEMS))
+ if (fail_stack.size < re_max_failures * TYPICAL_FAILURE_SIZE)
{
- fail_stack.size = (2 * re_max_failures * MAX_FAILURE_ITEMS);
+ fail_stack.size = re_max_failures * TYPICAL_FAILURE_SIZE;
#ifdef emacs
if (! fail_stack.stack)
return false;
}
-
-
-/* Read the ending character of a range (in a bracket expression) from the
- uncompiled pattern *P_PTR (which ends at PEND). We assume the
- starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
- Then we set the translation of all bits between the starting and
- ending characters (inclusive) in the compiled pattern B.
-
- Return an error code.
-
- We use these short variable names so we can use the same macros as
- `regex_compile' itself. */
-
-static reg_errcode_t
-compile_range (p_ptr, pend, translate, syntax, b)
- const char **p_ptr, *pend;
- RE_TRANSLATE_TYPE translate;
- reg_syntax_t syntax;
- unsigned char *b;
-{
- unsigned this_char;
-
- const char *p = *p_ptr;
- int range_start, range_end;
-
- if (p == pend)
- return REG_ERANGE;
-
- /* Even though the pattern is a signed `char *', we need to fetch
- with unsigned char *'s; if the high bit of the pattern character
- is set, the range endpoints will be negative if we fetch using a
- signed char *.
-
- We also want to fetch the endpoints without translating them; the
- appropriate translation is done in the bit-setting loop below. */
- /* The SVR4 compiler on the 3B2 had trouble with unsigned const char *. */
- range_start = ((const unsigned char *) p)[-2];
- range_end = ((const unsigned char *) p)[0];
-
- /* Have to increment the pointer into the pattern string, so the
- caller isn't still at the ending character. */
- (*p_ptr)++;
-
- /* If the start is after the end, the range is empty. */
- if (range_start > range_end)
- return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
-
- /* Here we see why `this_char' has to be larger than an `unsigned
- char' -- the range is inclusive, so if `range_end' == 0xff
- (assuming 8-bit characters), we would otherwise go into an infinite
- loop, since all characters <= 0xff. */
- for (this_char = range_start; this_char <= range_end; this_char++)
- {
- SET_LIST_BIT (TRANSLATE (this_char));
- }
-
- return REG_NOERROR;
-}
\f
/* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
case charset_not:
- /* Chars beyond end of map must be allowed. End of map is
- `127' if bufp->multibyte is nonzero. */
- simple_char_max = bufp->multibyte ? 0x80 : (1 << BYTEWIDTH);
+ /* Chars beyond end of bitmap are possible matches.
+ All the single-byte codes can occur in multibyte buffers.
+ So any that are not listed in the charset
+ are possible matches, even in multibyte buffers. */
+ simple_char_max = (1 << BYTEWIDTH);
for (j = CHARSET_BITMAP_SIZE (&p[-1]) * BYTEWIDTH;
j < simple_char_max; j++)
fastmap[j] = 1;
case wordchar:
- simple_char_max = bufp->multibyte ? 0x80 : (1 << BYTEWIDTH);
+ /* All the single-byte codes can occur in multibyte buffers,
+ and they may have word syntax. So do consider them. */
+ simple_char_max = (1 << BYTEWIDTH);
for (j = 0; j < simple_char_max; j++)
if (SYNTAX (j) == Sword)
fastmap[j] = 1;
case notwordchar:
- simple_char_max = bufp->multibyte ? 0x80 : (1 << BYTEWIDTH);
+ /* All the single-byte codes can occur in multibyte buffers,
+ and they may not have word syntax. So do consider them. */
+ simple_char_max = (1 << BYTEWIDTH);
for (j = 0; j < simple_char_max; j++)
if (SYNTAX (j) != Sword)
fastmap[j] = 1;
{
int fastmap_newline = fastmap['\n'];
- /* `.' matches anything (but if bufp->multibyte is
- nonzero, matches `\000' .. `\127' and possible multibyte
- character) ... */
+ /* `.' matches anything, except perhaps newline.
+ Even in a multibyte buffer, it should match any
+ conceivable byte value for the fastmap. */
if (bufp->multibyte)
- {
- simple_char_max = 0x80;
-
- for (j = 0x80; j < 0xA0; j++)
- if (BASE_LEADING_CODE_P (j))
- fastmap[j] = 1;
- match_any_multibyte_characters = true;
- }
- else
- simple_char_max = (1 << BYTEWIDTH);
+ match_any_multibyte_characters = true;
+ simple_char_max = (1 << BYTEWIDTH);
for (j = 0; j < simple_char_max; j++)
fastmap[j] = 1;
case categoryspec:
k = *p++;
- simple_char_max = bufp->multibyte ? 0x80 : (1 << BYTEWIDTH);
+ simple_char_max = (1 << BYTEWIDTH);
for (j = 0; j < simple_char_max; j++)
if (CHAR_HAS_CATEGORY (j, k))
fastmap[j] = 1;
case notcategoryspec:
k = *p++;
- simple_char_max = bufp->multibyte ? 0x80 : (1 << BYTEWIDTH);
+ simple_char_max = (1 << BYTEWIDTH);
for (j = 0; j < simple_char_max; j++)
if (!CHAR_HAS_CATEGORY (j, k))
fastmap[j] = 1;
range = total_size - startpos;
/* If the search isn't to be a backwards one, don't waste time in a
- search for a pattern that must be anchored. */
+ search for a pattern anchored at beginning of buffer. */
if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == begbuf && range > 0)
{
if (startpos > 0)
return -1;
else
- range = 1;
+ range = 0;
}
#ifdef emacs
don't keep searching past point. */
if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == at_dot && range > 0)
{
- range = PT - startpos;
- if (range <= 0)
+ range = PT_BYTE - BEGV_BYTE - startpos;
+ if (range < 0)
return -1;
}
#endif /* emacs */
anchored_start = 1;
#ifdef emacs
- SETUP_SYNTAX_TABLE_FOR_OBJECT (re_match_object,
- POS_AS_IN_BUFFER (startpos > 0
- ? startpos - 1 : startpos),
- 1);
+ gl_state.object = re_match_object;
+ {
+ int adjpos = NILP (re_match_object) || BUFFERP (re_match_object);
+ int charpos = SYNTAX_TABLE_BYTE_TO_CHAR (startpos + adjpos);
+
+ SETUP_SYNTAX_TABLE_FOR_OBJECT (re_match_object, charpos, 1);
+ }
#endif
/* Loop through the string, looking for a place to start matching. */
the first null string. */
if (fastmap && startpos < total_size && !bufp->can_be_null)
{
+ register const char *d;
+ register unsigned int buf_ch;
+
+ d = POS_ADDR_VSTRING (startpos);
+
if (range > 0) /* Searching forwards. */
{
- register const char *d;
register int lim = 0;
int irange = range;
if (startpos < size1 && startpos + range >= size1)
lim = range - (size1 - startpos);
- d = POS_ADDR_VSTRING (startpos);
-
/* Written out as an if-else to avoid testing `translate'
inside the loop. */
- if (translate)
- while (range > lim
- && !fastmap[(unsigned char)
- translate[(unsigned char) *d++]])
- range--;
+ if (RE_TRANSLATE_P (translate))
+ {
+ if (multibyte)
+ while (range > lim)
+ {
+ int buf_charlen;
+
+ buf_ch = STRING_CHAR_AND_LENGTH (d, range - lim,
+ buf_charlen);
+
+ buf_ch = RE_TRANSLATE (translate, buf_ch);
+ if (buf_ch >= 0400
+ || fastmap[buf_ch])
+ break;
+
+ range -= buf_charlen;
+ d += buf_charlen;
+ }
+ else
+ while (range > lim
+ && !fastmap[(unsigned char)
+ RE_TRANSLATE (translate, (unsigned char) *d)])
+ {
+ d++;
+ range--;
+ }
+ }
else
- while (range > lim && !fastmap[(unsigned char) *d++])
- range--;
+ while (range > lim && !fastmap[(unsigned char) *d])
+ {
+ d++;
+ range--;
+ }
startpos += irange - range;
}
else /* Searching backwards. */
{
- register char c = (size1 == 0 || startpos >= size1
- ? string2[startpos - size1]
- : string1[startpos]);
+ int room = (size1 == 0 || startpos >= size1
+ ? size2 + size1 - startpos
+ : size1 - startpos);
+
+ buf_ch = STRING_CHAR (d, room);
+ if (RE_TRANSLATE_P (translate))
+ buf_ch = RE_TRANSLATE (translate, buf_ch);
- if (!fastmap[(unsigned char) TRANSLATE (c)])
+ if (! (buf_ch >= 0400
+ || fastmap[buf_ch]))
goto advance;
}
}
/* Update STARTPOS to the next character boundary. */
if (multibyte)
{
- const unsigned char *p = POS_ADDR_VSTRING (startpos);
- const unsigned char *pend = STOP_ADDR_VSTRING (startpos);
+ const unsigned char *p
+ = (const unsigned char *) POS_ADDR_VSTRING (startpos);
+ const unsigned char *pend
+ = (const unsigned char *) STOP_ADDR_VSTRING (startpos);
int len = MULTIBYTE_FORM_LENGTH (p, pend - p);
range -= len;
}
else
{
- range--;
- startpos++;
- }
+ range--;
+ startpos++;
+ }
}
else
{
/* Update STARTPOS to the previous character boundary. */
if (multibyte)
{
- const unsigned char *p = POS_ADDR_VSTRING (startpos);
+ const unsigned char *p
+ = (const unsigned char *) POS_ADDR_VSTRING (startpos);
int len = 0;
/* Find the head of multibyte form. */
- while (!CHAR_HEAD_P (p))
+ while (!CHAR_HEAD_P (*p))
p--, len++;
/* Adjust it. */
/* Free everything we malloc. */
#ifdef MATCH_MAY_ALLOCATE
-#define FREE_VAR(var) if (var) then { REGEX_FREE (var); var = NULL; } else
+#define FREE_VAR(var) if (var) { REGEX_FREE (var); var = NULL; } else
#define FREE_VARIABLES() \
do { \
REGEX_FREE_STACK (fail_stack.stack); \
int result;
#ifdef emacs
- SETUP_SYNTAX_TABLE_FOR_OBJECT (re_match_object,
- POS_AS_IN_BUFFER (pos > 0 ? pos - 1 : pos),
- 1);
+ int charpos;
+ int adjpos = NILP (re_match_object) || BUFFERP (re_match_object);
+ gl_state.object = re_match_object;
+ charpos = SYNTAX_TABLE_BYTE_TO_CHAR (pos + adjpos);
+ SETUP_SYNTAX_TABLE_FOR_OBJECT (re_match_object, charpos, 1);
#endif
result = re_match_2_internal (bufp, string1, size1, string2, size2,
- pos, regs, stop);
+ pos, regs, stop);
alloca (0);
return result;
}
/* This is written out as an if-else so we don't waste time
testing `translate' inside the loop. */
- if (translate)
+ if (RE_TRANSLATE_P (translate))
{
- do
- {
- PREFETCH ();
- if ((unsigned char) translate[(unsigned char) *d++]
- != (unsigned char) *p++)
- goto fail;
- }
- while (--mcnt);
+#ifdef emacs
+ if (multibyte)
+ do
+ {
+ int pat_charlen, buf_charlen;
+ unsigned int pat_ch, buf_ch;
+
+ PREFETCH ();
+ pat_ch = STRING_CHAR_AND_LENGTH (p, pend - p, pat_charlen);
+ buf_ch = STRING_CHAR_AND_LENGTH (d, dend - d, buf_charlen);
+
+ if (RE_TRANSLATE (translate, buf_ch)
+ != pat_ch)
+ goto fail;
+
+ p += pat_charlen;
+ d += buf_charlen;
+ mcnt -= pat_charlen;
+ }
+ while (mcnt > 0);
+ else
+#endif /* not emacs */
+ do
+ {
+ PREFETCH ();
+ if ((unsigned char) RE_TRANSLATE (translate, (unsigned char) *d)
+ != (unsigned char) *p++)
+ goto fail;
+ d++;
+ }
+ while (--mcnt);
}
else
{
/* Match any character except possibly a newline or a null. */
case anychar:
- DEBUG_PRINT1 ("EXECUTING anychar.\n");
+ {
+ int buf_charlen;
+ unsigned int buf_ch;
- PREFETCH ();
+ DEBUG_PRINT1 ("EXECUTING anychar.\n");
- if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n')
- || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000'))
- goto fail;
+ PREFETCH ();
- SET_REGS_MATCHED ();
- DEBUG_PRINT2 (" Matched `%d'.\n", *d);
- d += multibyte ? MULTIBYTE_FORM_LENGTH (d, dend - d) : 1;
+#ifdef emacs
+ if (multibyte)
+ buf_ch = STRING_CHAR_AND_LENGTH (d, dend - d, buf_charlen);
+ else
+#endif /* not emacs */
+ {
+ buf_ch = (unsigned char) *d;
+ buf_charlen = 1;
+ }
+
+ buf_ch = TRANSLATE (buf_ch);
+
+ if ((!(bufp->syntax & RE_DOT_NEWLINE)
+ && buf_ch == '\n')
+ || ((bufp->syntax & RE_DOT_NOT_NULL)
+ && buf_ch == '\000'))
+ goto fail;
+
+ SET_REGS_MATCHED ();
+ DEBUG_PRINT2 (" Matched `%d'.\n", *d);
+ d += buf_charlen;
+ }
break;
/* Compare that many; failure if mismatch, else move
past them. */
- if (translate
+ if (RE_TRANSLATE_P (translate)
? bcmp_translate (d, d2, mcnt, translate)
: bcmp (d, d2, mcnt))
goto fail;
on_failure:
DEBUG_PRINT1 ("EXECUTING on_failure_jump");
+#if defined (WINDOWSNT) && defined (emacs)
+ QUIT;
+#endif
+
EXTRACT_NUMBER_AND_INCR (mcnt, p);
DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt);
/* A smart repeat ends with `maybe_pop_jump'.
We change it to either `pop_failure_jump' or `jump'. */
case maybe_pop_jump:
+#if defined (WINDOWSNT) && defined (emacs)
+ QUIT;
+#endif
EXTRACT_NUMBER_AND_INCR (mcnt, p);
DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt);
{
/* Unconditionally jump (without popping any failure points). */
case jump:
unconditional_jump:
+#if defined (WINDOWSNT) && defined (emacs)
+ QUIT;
+#endif
EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
p += mcnt; /* Do the jump. */
is the character at D, and S2 is the syntax of C2. */
int c1, c2, s1, s2;
int pos1 = PTR_TO_OFFSET (d - 1);
+ int charpos;
GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2);
GET_CHAR_AFTER_2 (c2, d, string1, end1, string2, end2);
#ifdef emacs
- UPDATE_SYNTAX_TABLE (pos1 ? pos1 : 1);
+ charpos = SYNTAX_TABLE_BYTE_TO_CHAR (pos1);
+ UPDATE_SYNTAX_TABLE (charpos);
#endif
s1 = SYNTAX (c1);
#ifdef emacs
- UPDATE_SYNTAX_TABLE_FORWARD (pos1 + 1);
+ UPDATE_SYNTAX_TABLE_FORWARD (charpos + 1);
#endif
s2 = SYNTAX (c2);
is the character at D, and S2 is the syntax of C2. */
int c1, c2, s1, s2;
int pos1 = PTR_TO_OFFSET (d - 1);
+ int charpos;
GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2);
GET_CHAR_AFTER_2 (c2, d, string1, end1, string2, end2);
#ifdef emacs
- UPDATE_SYNTAX_TABLE (pos1);
+ charpos = SYNTAX_TABLE_BYTE_TO_CHAR (pos1);
+ UPDATE_SYNTAX_TABLE (charpos);
#endif
s1 = SYNTAX (c1);
#ifdef emacs
- UPDATE_SYNTAX_TABLE_FORWARD (pos1 + 1);
+ UPDATE_SYNTAX_TABLE_FORWARD (charpos + 1);
#endif
s2 = SYNTAX (c2);
is the character at D, and S2 is the syntax of C2. */
int c1, c2, s1, s2;
int pos1 = PTR_TO_OFFSET (d);
+ int charpos;
GET_CHAR_AFTER_2 (c2, d, string1, end1, string2, end2);
#ifdef emacs
- UPDATE_SYNTAX_TABLE (pos1);
+ charpos = SYNTAX_TABLE_BYTE_TO_CHAR (pos1);
+ UPDATE_SYNTAX_TABLE (charpos);
#endif
s2 = SYNTAX (c2);
{
GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2);
#ifdef emacs
- UPDATE_SYNTAX_TABLE_BACKWARD (pos1 - 1);
+ UPDATE_SYNTAX_TABLE_BACKWARD (charpos - 1);
#endif
s1 = SYNTAX (c1);
/* C1 is the character before D, S1 is the syntax of C1, C2
is the character at D, and S2 is the syntax of C2. */
int c1, c2, s1, s2;
+ int pos1 = PTR_TO_OFFSET (d);
+ int charpos;
GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2);
+#ifdef emacs
+ charpos = SYNTAX_TABLE_BYTE_TO_CHAR (pos1 - 1);
+ UPDATE_SYNTAX_TABLE (charpos);
+#endif
s1 = SYNTAX (c1);
/* Case 2: S1 is not Sword. */
if (!AT_STRINGS_END (d))
{
GET_CHAR_AFTER_2 (c2, d, string1, end1, string2, end2);
+#ifdef emacs
+ UPDATE_SYNTAX_TABLE_FORWARD (charpos);
+#endif
s2 = SYNTAX (c2);
/* ... and S2 is Sword, and WORD_BOUNDARY_P (C1, C2)
#ifdef emacs
case before_dot:
DEBUG_PRINT1 ("EXECUTING before_dot.\n");
- if (PTR_CHAR_POS ((unsigned char *) d) >= PT)
+ if (PTR_BYTE_POS ((unsigned char *) d) >= PT_BYTE)
goto fail;
break;
case at_dot:
DEBUG_PRINT1 ("EXECUTING at_dot.\n");
- if (PTR_CHAR_POS ((unsigned char *) d) != PT)
+ if (PTR_BYTE_POS ((unsigned char *) d) != PT_BYTE)
goto fail;
break;
case after_dot:
DEBUG_PRINT1 ("EXECUTING after_dot.\n");
- if (PTR_CHAR_POS ((unsigned char *) d) <= PT)
+ if (PTR_BYTE_POS ((unsigned char *) d) <= PT_BYTE)
goto fail;
break;
PREFETCH ();
#ifdef emacs
{
- int pos1 = PTR_TO_OFFSET (d);
+ int pos1 = SYNTAX_TABLE_BYTE_TO_CHAR (PTR_TO_OFFSET (d));
UPDATE_SYNTAX_TABLE (pos1);
}
#endif
PREFETCH ();
#ifdef emacs
{
- int pos1 = PTR_TO_OFFSET (d);
+ int pos1 = SYNTAX_TABLE_BYTE_TO_CHAR (PTR_TO_OFFSET (d));
UPDATE_SYNTAX_TABLE (pos1);
}
#endif
/* We goto here if a matching operation fails. */
fail:
+#if defined (WINDOWSNT) && defined (emacs)
+ QUIT;
+#endif
if (!FAIL_STACK_EMPTY ())
{ /* A restart point is known. Restore to that state. */
DEBUG_PRINT1 ("\nFAIL:\n");
RE_TRANSLATE_TYPE translate;
{
register unsigned char *p1 = s1, *p2 = s2;
- while (len)
+ unsigned char *p1_end = s1 + len;
+ unsigned char *p2_end = s2 + len;
+
+ while (p1 != p1_end && p2 != p2_end)
{
- if (translate[*p1++] != translate[*p2++]) return 1;
- len--;
+ int p1_charlen, p2_charlen;
+ int p1_ch, p2_ch;
+
+ p1_ch = STRING_CHAR_AND_LENGTH (p1, p1_end - p1, p1_charlen);
+ p2_ch = STRING_CHAR_AND_LENGTH (p2, p2_end - p2, p2_charlen);
+
+ if (RE_TRANSLATE (translate, p1_ch)
+ != RE_TRANSLATE (translate, p2_ch))
+ return 1;
+
+ p1 += p1_charlen, p2 += p2_charlen;
}
+
+ if (p1 != p1_end || p2 != p2_end)
+ return 1;
+
return 0;
}
\f
projects / gnulib.git / blobdiff