/* String search routines for GNU Emacs.
- Copyright (C) 1985, 86, 87, 93, 94, 97, 1998 Free Software Foundation, Inc.
+ Copyright (C) 1985, 86,87,93,94,97,98, 1999 Free Software Foundation, Inc.
This file is part of GNU Emacs.
#include <sys/types.h>
#include "regex.h"
+#define min(a, b) ((a) < (b) ? (a) : (b))
+#define max(a, b) ((a) > (b) ? (a) : (b))
+
#define REGEXP_CACHE_SIZE 20
/* If the regexp is non-nil, then the buffer contains the compiled form
static void set_search_regs ();
static void save_search_regs ();
-
+static int simple_search ();
+static int boyer_moore ();
static int search_buffer ();
static void
error ("Stack overflow in regexp matcher");
}
-#ifdef __STDC__
-#define CONST const
-#else
-#define CONST
-#endif
-
/* Compile a regexp and signal a Lisp error if anything goes wrong.
PATTERN is the pattern to compile.
CP is the place to put the result.
- TRANSLATE is a translation table for ignoring case, or NULL for none.
+ TRANSLATE is a translation table for ignoring case, or nil for none.
REGP is the structure that says where to store the "register"
values that will result from matching this pattern.
If it is 0, we should compile the pattern not to record any
compile_pattern_1 (cp, pattern, translate, regp, posix, multibyte)
struct regexp_cache *cp;
Lisp_Object pattern;
- Lisp_Object *translate;
+ Lisp_Object translate;
struct re_registers *regp;
int posix;
int multibyte;
{
- char *raw_pattern;
+ unsigned char *raw_pattern;
int raw_pattern_size;
char *val;
reg_syntax_t old;
if (multibyte == STRING_MULTIBYTE (pattern))
{
- raw_pattern = (char *) XSTRING (pattern)->data;
- raw_pattern_size = XSTRING (pattern)->size_byte;
+ raw_pattern = (unsigned char *) XSTRING (pattern)->data;
+ raw_pattern_size = STRING_BYTES (XSTRING (pattern));
}
else if (multibyte)
{
raw_pattern_size = count_size_as_multibyte (XSTRING (pattern)->data,
XSTRING (pattern)->size);
- raw_pattern = (char *) alloca (raw_pattern_size + 1);
+ raw_pattern = (unsigned char *) alloca (raw_pattern_size + 1);
copy_text (XSTRING (pattern)->data, raw_pattern,
XSTRING (pattern)->size, 0, 1);
}
so that only the multibyte chars which really correspond to
the chosen single-byte character set can possibly match. */
raw_pattern_size = XSTRING (pattern)->size;
- raw_pattern = (char *) alloca (raw_pattern_size + 1);
+ raw_pattern = (unsigned char *) alloca (raw_pattern_size + 1);
copy_text (XSTRING (pattern)->data, raw_pattern,
- XSTRING (pattern)->size, 1, 0);
+ STRING_BYTES (XSTRING (pattern)), 1, 0);
}
cp->regexp = Qnil;
- cp->buf.translate = translate;
+ cp->buf.translate = (! NILP (translate) ? translate : make_number (0));
cp->posix = posix;
cp->buf.multibyte = multibyte;
BLOCK_INPUT;
old = re_set_syntax (RE_SYNTAX_EMACS
| (posix ? 0 : RE_NO_POSIX_BACKTRACKING));
- val = (char *) re_compile_pattern (raw_pattern, raw_pattern_size, &cp->buf);
+ val = (char *) re_compile_pattern ((char *)raw_pattern,
+ raw_pattern_size, &cp->buf);
re_set_syntax (old);
UNBLOCK_INPUT;
if (val)
cp->regexp = Fcopy_sequence (pattern);
}
+/* Shrink each compiled regexp buffer in the cache
+ to the size actually used right now.
+ This is called from garbage collection. */
+
+void
+shrink_regexp_cache ()
+{
+ struct regexp_cache *cp, **cpp;
+
+ for (cp = searchbuf_head; cp != 0; cp = cp->next)
+ {
+ cp->buf.allocated = cp->buf.used;
+ cp->buf.buffer
+ = (unsigned char *) realloc (cp->buf.buffer, cp->buf.used);
+ }
+}
+
/* Compile a regexp if necessary, but first check to see if there's one in
the cache.
PATTERN is the pattern to compile.
- TRANSLATE is a translation table for ignoring case, or NULL for none.
+ TRANSLATE is a translation table for ignoring case, or nil for none.
REGP is the structure that says where to store the "register"
values that will result from matching this pattern.
If it is 0, we should compile the pattern not to record any
compile_pattern (pattern, regp, translate, posix, multibyte)
Lisp_Object pattern;
struct re_registers *regp;
- Lisp_Object *translate;
+ Lisp_Object translate;
int posix, multibyte;
{
struct regexp_cache *cp, **cpp;
for (cpp = &searchbuf_head; ; cpp = &cp->next)
{
cp = *cpp;
+ /* Entries are initialized to nil, and may be set to nil by
+ compile_pattern_1 if the pattern isn't valid. Don't apply
+ XSTRING in those cases. However, compile_pattern_1 is only
+ applied to the cache entry we pick here to reuse. So nil
+ should never appear before a non-nil entry. */
+ if (NILP (cp->regexp))
+ goto compile_it;
if (XSTRING (cp->regexp)->size == XSTRING (pattern)->size
+ && STRING_MULTIBYTE (cp->regexp) == STRING_MULTIBYTE (pattern)
&& !NILP (Fstring_equal (cp->regexp, pattern))
- && cp->buf.translate == translate
+ && EQ (cp->buf.translate, (! NILP (translate) ? translate : make_number (0)))
&& cp->posix == posix
&& cp->buf.multibyte == multibyte)
break;
- /* If we're at the end of the cache, compile into the last cell. */
+ /* If we're at the end of the cache, compile into the nil cell
+ we found, or the last (least recently used) cell with a
+ string value. */
if (cp->next == 0)
{
+ compile_it:
compile_pattern_1 (cp, pattern, translate, regp, posix, multibyte);
break;
}
CHECK_STRING (string, 0);
bufp = compile_pattern (string, &search_regs,
(!NILP (current_buffer->case_fold_search)
- ? XCHAR_TABLE (DOWNCASE_TABLE)->contents : 0),
+ ? DOWNCASE_TABLE : Qnil),
posix,
!NILP (current_buffer->enable_multibyte_characters));
i = re_match_2 (bufp, (char *) p1, s1, (char *) p2, s2,
PT_BYTE - BEGV_BYTE, &search_regs,
ZV_BYTE - BEGV_BYTE);
+ immediate_quit = 0;
+
if (i == -2)
matcher_overflow ();
= BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE);
}
XSETBUFFER (last_thing_searched, current_buffer);
- immediate_quit = 0;
return val;
}
bufp = compile_pattern (regexp, &search_regs,
(!NILP (current_buffer->case_fold_search)
- ? XCHAR_TABLE (DOWNCASE_TABLE)->contents : 0),
+ ? DOWNCASE_TABLE : Qnil),
posix,
STRING_MULTIBYTE (string));
immediate_quit = 1;
re_match_object = string;
val = re_search (bufp, (char *) XSTRING (string)->data,
- XSTRING (string)->size_byte, pos_byte,
- XSTRING (string)->size_byte - pos_byte,
+ STRING_BYTES (XSTRING (string)), pos_byte,
+ STRING_BYTES (XSTRING (string)) - pos_byte,
&search_regs);
immediate_quit = 0;
last_thing_searched = Qt;
DEFUN ("string-match", Fstring_match, Sstring_match, 2, 3, 0,
"Return index of start of first match for REGEXP in STRING, or nil.\n\
+Case is ignored if `case-fold-search' is non-nil in the current buffer.\n\
If third arg START is non-nil, start search at that index in STRING.\n\
For index of first char beyond the match, do (match-end 0).\n\
`match-end' and `match-beginning' also give indices of substrings\n\
DEFUN ("posix-string-match", Fposix_string_match, Sposix_string_match, 2, 3, 0,
"Return index of start of first match for REGEXP in STRING, or nil.\n\
Find the longest match, in accord with Posix regular expression rules.\n\
+Case is ignored if `case-fold-search' is non-nil in the current buffer.\n\
If third arg START is non-nil, start search at that index in STRING.\n\
For index of first char beyond the match, do (match-end 0).\n\
`match-end' and `match-beginning' also give indices of substrings\n\
int val;
struct re_pattern_buffer *bufp;
- bufp = compile_pattern (regexp, 0, 0, 0, STRING_MULTIBYTE (string));
+ bufp = compile_pattern (regexp, 0, Qnil,
+ 0, STRING_MULTIBYTE (string));
immediate_quit = 1;
re_match_object = string;
val = re_search (bufp, (char *) XSTRING (string)->data,
- XSTRING (string)->size, 0, XSTRING (string)->size,
- 0);
+ STRING_BYTES (XSTRING (string)), 0,
+ STRING_BYTES (XSTRING (string)), 0);
immediate_quit = 0;
return val;
}
regexp = string_make_unibyte (regexp);
re_match_object = Qt;
bufp = compile_pattern (regexp, 0,
- XCHAR_TABLE (Vascii_downcase_table)->contents, 0,
+ Vascii_downcase_table, 0,
0);
immediate_quit = 1;
val = re_search (bufp, string, len, 0, len, 0);
return val;
}
\f
-/* max and min. */
-
-static int
-max (a, b)
- int a, b;
-{
- return ((a > b) ? a : b);
-}
-
-static int
-min (a, b)
- int a, b;
-{
- return ((a < b) ? a : b);
-}
-
-\f
/* The newline cache: remembering which sections of text have no newlines. */
/* If the user has requested newline caching, make sure it's on.
If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do
except when inside redisplay. */
+int
scan_buffer (target, start, end, count, shortage, allow_quit)
register int target;
int start, end;
examine. */
int ceiling_byte = CHAR_TO_BYTE (end) - 1;
int start_byte = CHAR_TO_BYTE (start);
+ int tem;
/* If we're looking for a newline, consult the newline cache
to see where we can avoid some scanning. */
bytes. BUFFER_CEILING_OF returns the last character
position that is contiguous, so the ceiling is the
position after that. */
- ceiling_byte = min (BUFFER_CEILING_OF (start_byte), ceiling_byte);
+ tem = BUFFER_CEILING_OF (start_byte);
+ ceiling_byte = min (tem, ceiling_byte);
{
/* The termination address of the dumb loop. */
/* The last character to check before the next obstacle. */
int ceiling_byte = CHAR_TO_BYTE (end);
int start_byte = CHAR_TO_BYTE (start);
+ int tem;
/* Consult the newline cache, if appropriate. */
if (target == '\n' && newline_cache)
}
/* Stop scanning before the gap. */
- ceiling_byte = max (BUFFER_FLOOR_OF (start_byte - 1), ceiling_byte);
+ tem = BUFFER_FLOOR_OF (start_byte - 1);
+ ceiling_byte = max (tem, ceiling_byte);
{
/* The termination address of the dumb loop. */
}
else
{
- int start_byte = CHAR_TO_BYTE (start);
while (start_byte > limit_byte)
{
ceiling = BUFFER_FLOOR_OF (start_byte - 1);
int posix;
{
register int np;
- int lim;
+ int lim, lim_byte;
int n = direction;
if (!NILP (count))
CHECK_STRING (string, 0);
if (NILP (bound))
- lim = n > 0 ? ZV : BEGV;
+ {
+ if (n > 0)
+ lim = ZV, lim_byte = ZV_BYTE;
+ else
+ lim = BEGV, lim_byte = BEGV_BYTE;
+ }
else
{
CHECK_NUMBER_COERCE_MARKER (bound, 1);
if (n > 0 ? lim < PT : lim > PT)
error ("Invalid search bound (wrong side of point)");
if (lim > ZV)
- lim = ZV;
- if (lim < BEGV)
- lim = BEGV;
+ lim = ZV, lim_byte = ZV_BYTE;
+ else if (lim < BEGV)
+ lim = BEGV, lim_byte = BEGV_BYTE;
+ else
+ lim_byte = CHAR_TO_BYTE (lim);
}
- np = search_buffer (string, PT, lim, n, RE,
+ np = search_buffer (string, PT, PT_BYTE, lim, lim_byte, n, RE,
(!NILP (current_buffer->case_fold_search)
- ? XCHAR_TABLE (current_buffer->case_canon_table)->contents
- : 0),
+ ? current_buffer->case_canon_table
+ : Qnil),
(!NILP (current_buffer->case_fold_search)
- ? XCHAR_TABLE (current_buffer->case_eqv_table)->contents
- : 0),
+ ? current_buffer->case_eqv_table
+ : Qnil),
posix);
if (np <= 0)
{
{
if (lim < BEGV || lim > ZV)
abort ();
- SET_PT (lim);
+ SET_PT_BOTH (lim, lim_byte);
return Qnil;
#if 0 /* This would be clean, but maybe programs depend on
a value of nil here. */
trivial_regexp_p (regexp)
Lisp_Object regexp;
{
- int len = XSTRING (regexp)->size;
+ int len = STRING_BYTES (XSTRING (regexp));
unsigned char *s = XSTRING (regexp)->data;
unsigned char c;
while (--len >= 0)
If N is positive, searching is forward and LIM must be greater than POS.
If N is negative, searching is backward and LIM must be less than POS.
- Returns -x if only N-x occurrences found (x > 0),
+ Returns -x if x occurrences remain to be found (x > 0),
or else the position at the beginning of the Nth occurrence
(if searching backward) or the end (if searching forward).
POSIX is nonzero if we want full backtracking (POSIX style)
for this pattern. 0 means backtrack only enough to get a valid match. */
+#define TRANSLATE(out, trt, d) \
+do \
+ { \
+ if (! NILP (trt)) \
+ { \
+ Lisp_Object temp; \
+ temp = Faref (trt, make_number (d)); \
+ if (INTEGERP (temp)) \
+ out = XINT (temp); \
+ else \
+ out = d; \
+ } \
+ else \
+ out = d; \
+ } \
+while (0)
+
static int
-search_buffer (string, pos, lim, n, RE, trt, inverse_trt, posix)
+search_buffer (string, pos, pos_byte, lim, lim_byte, n,
+ RE, trt, inverse_trt, posix)
Lisp_Object string;
int pos;
+ int pos_byte;
int lim;
+ int lim_byte;
int n;
int RE;
- Lisp_Object *trt;
- Lisp_Object *inverse_trt;
+ Lisp_Object trt;
+ Lisp_Object inverse_trt;
int posix;
{
int len = XSTRING (string)->size;
- int len_byte = XSTRING (string)->size_byte;
- unsigned char *base_pat = XSTRING (string)->data;
- register int *BM_tab;
- int *BM_tab_base;
- register int direction = ((n > 0) ? 1 : -1);
- register int dirlen;
- int infinity, limit, k, stride_for_teases;
- register unsigned char *pat, *cursor, *p_limit;
- register int i, j;
- unsigned char *p1, *p2;
- int s1, s2;
+ int len_byte = STRING_BYTES (XSTRING (string));
+ register int i;
if (running_asynch_code)
save_search_regs ();
+ /* Searching 0 times means don't move. */
/* Null string is found at starting position. */
- if (len == 0)
+ if (len == 0 || n == 0)
{
set_search_regs (pos, 0);
return pos;
}
- /* Searching 0 times means don't move. */
- if (n == 0)
- return pos;
-
if (RE && !trivial_regexp_p (string))
{
+ unsigned char *p1, *p2;
+ int s1, s2;
struct re_pattern_buffer *bufp;
bufp = compile_pattern (string, &search_regs, trt, posix,
{
int val;
val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2,
- pos - BEGV, lim - pos, &search_regs,
+ pos_byte - BEGV_BYTE, lim_byte - pos_byte,
+ &search_regs,
/* Don't allow match past current point */
- pos - BEGV);
+ pos_byte - BEGV_BYTE);
if (val == -2)
{
matcher_overflow ();
}
if (val >= 0)
{
+ pos_byte = search_regs.start[0] + BEGV_BYTE;
for (i = 0; i < search_regs.num_regs; i++)
if (search_regs.start[i] >= 0)
{
{
int val;
val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2,
- pos - BEGV, lim - pos, &search_regs,
- lim - BEGV);
+ pos_byte - BEGV_BYTE, lim_byte - pos_byte,
+ &search_regs,
+ lim_byte - BEGV_BYTE);
if (val == -2)
{
matcher_overflow ();
}
if (val >= 0)
{
+ pos_byte = search_regs.end[0] + BEGV_BYTE;
for (i = 0; i < search_regs.num_regs; i++)
if (search_regs.start[i] >= 0)
{
}
else /* non-RE case */
{
- int pos_byte = CHAR_TO_BYTE (pos);
- int lim_byte = CHAR_TO_BYTE (lim);
-#ifdef C_ALLOCA
- int BM_tab_space[0400];
- BM_tab = &BM_tab_space[0];
-#else
- BM_tab = (int *) alloca (0400 * sizeof (int));
-#endif
- {
- unsigned char *raw_pattern;
- int raw_pattern_size;
- unsigned char *patbuf;
- int multibyte = !NILP (current_buffer->enable_multibyte_characters);
+ unsigned char *raw_pattern, *pat;
+ int raw_pattern_size;
+ int raw_pattern_size_byte;
+ unsigned char *patbuf;
+ int multibyte = !NILP (current_buffer->enable_multibyte_characters);
+ unsigned char *base_pat = XSTRING (string)->data;
+ int charset_base = -1;
+ int boyer_moore_ok = 1;
+
+ /* MULTIBYTE says whether the text to be searched is multibyte.
+ We must convert PATTERN to match that, or we will not really
+ find things right. */
+
+ if (multibyte == STRING_MULTIBYTE (string))
+ {
+ raw_pattern = (unsigned char *) XSTRING (string)->data;
+ raw_pattern_size = XSTRING (string)->size;
+ raw_pattern_size_byte = STRING_BYTES (XSTRING (string));
+ }
+ else if (multibyte)
+ {
+ raw_pattern_size = XSTRING (string)->size;
+ raw_pattern_size_byte
+ = count_size_as_multibyte (XSTRING (string)->data,
+ raw_pattern_size);
+ raw_pattern = (unsigned char *) alloca (raw_pattern_size_byte + 1);
+ copy_text (XSTRING (string)->data, raw_pattern,
+ XSTRING (string)->size, 0, 1);
+ }
+ else
+ {
+ /* Converting multibyte to single-byte.
+
+ ??? Perhaps this conversion should be done in a special way
+ by subtracting nonascii-insert-offset from each non-ASCII char,
+ so that only the multibyte chars which really correspond to
+ the chosen single-byte character set can possibly match. */
+ raw_pattern_size = XSTRING (string)->size;
+ raw_pattern_size_byte = XSTRING (string)->size;
+ raw_pattern = (unsigned char *) alloca (raw_pattern_size + 1);
+ copy_text (XSTRING (string)->data, raw_pattern,
+ STRING_BYTES (XSTRING (string)), 1, 0);
+ }
+
+ /* Copy and optionally translate the pattern. */
+ len = raw_pattern_size;
+ len_byte = raw_pattern_size_byte;
+ patbuf = (unsigned char *) alloca (len_byte);
+ pat = patbuf;
+ base_pat = raw_pattern;
+ if (multibyte)
+ {
+ while (--len >= 0)
+ {
+ unsigned char str[MAX_MULTIBYTE_LENGTH];
+ int c, translated, inverse;
+ int in_charlen, charlen;
+
+ /* If we got here and the RE flag is set, it's because we're
+ dealing with a regexp known to be trivial, so the backslash
+ just quotes the next character. */
+ if (RE && *base_pat == '\\')
+ {
+ len--;
+ len_byte--;
+ base_pat++;
+ }
+
+ c = STRING_CHAR_AND_LENGTH (base_pat, len_byte, in_charlen);
+
+ /* Translate the character, if requested. */
+ TRANSLATE (translated, trt, c);
+ /* If translation changed the byte-length, go back
+ to the original character. */
+ charlen = CHAR_STRING (translated, str);
+ if (in_charlen != charlen)
+ {
+ translated = c;
+ charlen = CHAR_STRING (c, str);
+ }
+
+ /* If we are searching for something strange,
+ an invalid multibyte code, don't use boyer-moore. */
+ if (! ASCII_BYTE_P (translated)
+ && (charlen == 1 /* 8bit code */
+ || charlen != in_charlen /* invalid multibyte code */
+ ))
+ boyer_moore_ok = 0;
+
+ TRANSLATE (inverse, inverse_trt, c);
+
+ /* Did this char actually get translated?
+ Would any other char get translated into it? */
+ if (translated != c || inverse != c)
+ {
+ /* Keep track of which character set row
+ contains the characters that need translation. */
+ int charset_base_code = c & ~CHAR_FIELD3_MASK;
+ if (charset_base == -1)
+ charset_base = charset_base_code;
+ else if (charset_base != charset_base_code)
+ /* If two different rows appear, needing translation,
+ then we cannot use boyer_moore search. */
+ boyer_moore_ok = 0;
+ }
+
+ /* Store this character into the translated pattern. */
+ bcopy (str, pat, charlen);
+ pat += charlen;
+ base_pat += in_charlen;
+ len_byte -= in_charlen;
+ }
+ }
+ else
+ {
+ /* Unibyte buffer. */
+ charset_base = 0;
+ while (--len >= 0)
+ {
+ int c, translated;
+
+ /* If we got here and the RE flag is set, it's because we're
+ dealing with a regexp known to be trivial, so the backslash
+ just quotes the next character. */
+ if (RE && *base_pat == '\\')
+ {
+ len--;
+ base_pat++;
+ }
+ c = *base_pat++;
+ TRANSLATE (translated, trt, c);
+ *pat++ = translated;
+ }
+ }
+
+ len_byte = pat - patbuf;
+ len = raw_pattern_size;
+ pat = base_pat = patbuf;
+
+ if (boyer_moore_ok)
+ return boyer_moore (n, pat, len, len_byte, trt, inverse_trt,
+ pos, pos_byte, lim, lim_byte,
+ charset_base);
+ else
+ return simple_search (n, pat, len, len_byte, trt,
+ pos, pos_byte, lim, lim_byte);
+ }
+}
+\f
+/* Do a simple string search N times for the string PAT,
+ whose length is LEN/LEN_BYTE,
+ from buffer position POS/POS_BYTE until LIM/LIM_BYTE.
+ TRT is the translation table.
+
+ Return the character position where the match is found.
+ Otherwise, if M matches remained to be found, return -M.
- /* MULTIBYTE says whether the text to be searched is multibyte.
- We must convert PATTERN to match that, or we will not really
- find things right. */
+ This kind of search works regardless of what is in PAT and
+ regardless of what is in TRT. It is used in cases where
+ boyer_moore cannot work. */
- if (multibyte == STRING_MULTIBYTE (string))
+static int
+simple_search (n, pat, len, len_byte, trt, pos, pos_byte, lim, lim_byte)
+ int n;
+ unsigned char *pat;
+ int len, len_byte;
+ Lisp_Object trt;
+ int pos, pos_byte;
+ int lim, lim_byte;
+{
+ int multibyte = ! NILP (current_buffer->enable_multibyte_characters);
+ int forward = n > 0;
+
+ if (lim > pos && multibyte)
+ while (n > 0)
+ {
+ while (1)
{
- raw_pattern = (char *) XSTRING (string)->data;
- raw_pattern_size = XSTRING (string)->size_byte;
+ /* Try matching at position POS. */
+ int this_pos = pos;
+ int this_pos_byte = pos_byte;
+ int this_len = len;
+ int this_len_byte = len_byte;
+ unsigned char *p = pat;
+ if (pos + len > lim)
+ goto stop;
+
+ while (this_len > 0)
+ {
+ int charlen, buf_charlen;
+ int pat_ch, buf_ch;
+
+ pat_ch = STRING_CHAR_AND_LENGTH (p, this_len_byte, charlen);
+ buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte),
+ ZV_BYTE - this_pos_byte,
+ buf_charlen);
+ TRANSLATE (buf_ch, trt, buf_ch);
+
+ if (buf_ch != pat_ch)
+ break;
+
+ this_len_byte -= charlen;
+ this_len--;
+ p += charlen;
+
+ this_pos_byte += buf_charlen;
+ this_pos++;
+ }
+
+ if (this_len == 0)
+ {
+ pos += len;
+ pos_byte += len_byte;
+ break;
+ }
+
+ INC_BOTH (pos, pos_byte);
}
- else if (multibyte)
+
+ n--;
+ }
+ else if (lim > pos)
+ while (n > 0)
+ {
+ while (1)
{
- raw_pattern_size = count_size_as_multibyte (XSTRING (string)->data,
- XSTRING (string)->size);
- raw_pattern = (char *) alloca (raw_pattern_size + 1);
- copy_text (XSTRING (string)->data, raw_pattern,
- XSTRING (string)->size, 0, 1);
+ /* Try matching at position POS. */
+ int this_pos = pos;
+ int this_len = len;
+ unsigned char *p = pat;
+
+ if (pos + len > lim)
+ goto stop;
+
+ while (this_len > 0)
+ {
+ int pat_ch = *p++;
+ int buf_ch = FETCH_BYTE (this_pos);
+ TRANSLATE (buf_ch, trt, buf_ch);
+
+ if (buf_ch != pat_ch)
+ break;
+
+ this_len--;
+ this_pos++;
+ }
+
+ if (this_len == 0)
+ {
+ pos += len;
+ break;
+ }
+
+ pos++;
}
- else
+
+ n--;
+ }
+ /* Backwards search. */
+ else if (lim < pos && multibyte)
+ while (n < 0)
+ {
+ while (1)
{
- /* Converting multibyte to single-byte.
-
- ??? Perhaps this conversion should be done in a special way
- by subtracting nonascii-insert-offset from each non-ASCII char,
- so that only the multibyte chars which really correspond to
- the chosen single-byte character set can possibly match. */
- raw_pattern_size = XSTRING (string)->size;
- raw_pattern = (char *) alloca (raw_pattern_size + 1);
- copy_text (XSTRING (string)->data, raw_pattern,
- XSTRING (string)->size, 1, 0);
+ /* Try matching at position POS. */
+ int this_pos = pos - len;
+ int this_pos_byte = pos_byte - len_byte;
+ int this_len = len;
+ int this_len_byte = len_byte;
+ unsigned char *p = pat;
+
+ if (pos - len < lim)
+ goto stop;
+
+ while (this_len > 0)
+ {
+ int charlen, buf_charlen;
+ int pat_ch, buf_ch;
+
+ pat_ch = STRING_CHAR_AND_LENGTH (p, this_len_byte, charlen);
+ buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte),
+ ZV_BYTE - this_pos_byte,
+ buf_charlen);
+ TRANSLATE (buf_ch, trt, buf_ch);
+
+ if (buf_ch != pat_ch)
+ break;
+
+ this_len_byte -= charlen;
+ this_len--;
+ p += charlen;
+ this_pos_byte += buf_charlen;
+ this_pos++;
+ }
+
+ if (this_len == 0)
+ {
+ pos -= len;
+ pos_byte -= len_byte;
+ break;
+ }
+
+ DEC_BOTH (pos, pos_byte);
}
- len_byte = raw_pattern_size;
- patbuf = (unsigned char *) alloca (len_byte);
- pat = patbuf;
- base_pat = raw_pattern;
- while (--len_byte >= 0)
+ n++;
+ }
+ else if (lim < pos)
+ while (n < 0)
+ {
+ while (1)
{
- /* If we got here and the RE flag is set, it's because we're
- dealing with a regexp known to be trivial, so the backslash
- just quotes the next character. */
- if (RE && *base_pat == '\\')
+ /* Try matching at position POS. */
+ int this_pos = pos - len;
+ int this_len = len;
+ unsigned char *p = pat;
+
+ if (pos - len < lim)
+ goto stop;
+
+ while (this_len > 0)
{
- len_byte--;
- base_pat++;
+ int pat_ch = *p++;
+ int buf_ch = FETCH_BYTE (this_pos);
+ TRANSLATE (buf_ch, trt, buf_ch);
+
+ if (buf_ch != pat_ch)
+ break;
+ this_len--;
+ this_pos++;
}
- *pat++ = (trt ? XINT (trt[*base_pat++]) : *base_pat++);
+
+ if (this_len == 0)
+ {
+ pos -= len;
+ break;
+ }
+
+ pos--;
}
- len_byte = pat - patbuf;
- pat = base_pat = patbuf;
+
+ n++;
}
- /* The general approach is that we are going to maintain that we know */
- /* the first (closest to the present position, in whatever direction */
- /* we're searching) character that could possibly be the last */
- /* (furthest from present position) character of a valid match. We */
- /* advance the state of our knowledge by looking at that character */
- /* and seeing whether it indeed matches the last character of the */
- /* pattern. If it does, we take a closer look. If it does not, we */
- /* move our pointer (to putative last characters) as far as is */
- /* logically possible. This amount of movement, which I call a */
- /* stride, will be the length of the pattern if the actual character */
- /* appears nowhere in the pattern, otherwise it will be the distance */
- /* from the last occurrence of that character to the end of the */
- /* pattern. */
- /* As a coding trick, an enormous stride is coded into the table for */
- /* characters that match the last character. This allows use of only */
- /* a single test, a test for having gone past the end of the */
- /* permissible match region, to test for both possible matches (when */
- /* the stride goes past the end immediately) and failure to */
- /* match (where you get nudged past the end one stride at a time). */
-
- /* Here we make a "mickey mouse" BM table. The stride of the search */
- /* is determined only by the last character of the putative match. */
- /* If that character does not match, we will stride the proper */
- /* distance to propose a match that superimposes it on the last */
- /* instance of a character that matches it (per trt), or misses */
- /* it entirely if there is none. */
-
- dirlen = len_byte * direction;
- infinity = dirlen - (lim_byte + pos_byte + len_byte + len_byte) * direction;
- if (direction < 0)
- pat = (base_pat += len_byte - 1);
- BM_tab_base = BM_tab;
- BM_tab += 0400;
- j = dirlen; /* to get it in a register */
- /* A character that does not appear in the pattern induces a */
- /* stride equal to the pattern length. */
- while (BM_tab_base != BM_tab)
- {
- *--BM_tab = j;
- *--BM_tab = j;
- *--BM_tab = j;
- *--BM_tab = j;
- }
- i = 0;
- while (i != infinity)
+
+ stop:
+ if (n == 0)
+ {
+ if (forward)
+ set_search_regs ((multibyte ? pos_byte : pos) - len_byte, len_byte);
+ else
+ set_search_regs (multibyte ? pos_byte : pos, len_byte);
+
+ return pos;
+ }
+ else if (n > 0)
+ return -n;
+ else
+ return n;
+}
+\f
+/* Do Boyer-Moore search N times for the string PAT,
+ whose length is LEN/LEN_BYTE,
+ from buffer position POS/POS_BYTE until LIM/LIM_BYTE.
+ DIRECTION says which direction we search in.
+ TRT and INVERSE_TRT are translation tables.
+
+ This kind of search works if all the characters in PAT that have
+ nontrivial translation are the same aside from the last byte. This
+ makes it possible to translate just the last byte of a character,
+ and do so after just a simple test of the context.
+
+ If that criterion is not satisfied, do not call this function. */
+
+static int
+boyer_moore (n, base_pat, len, len_byte, trt, inverse_trt,
+ pos, pos_byte, lim, lim_byte, charset_base)
+ int n;
+ unsigned char *base_pat;
+ int len, len_byte;
+ Lisp_Object trt;
+ Lisp_Object inverse_trt;
+ int pos, pos_byte;
+ int lim, lim_byte;
+ int charset_base;
+{
+ int direction = ((n > 0) ? 1 : -1);
+ register int dirlen;
+ int infinity, limit, k, stride_for_teases = 0;
+ register int *BM_tab;
+ int *BM_tab_base;
+ register unsigned char *cursor, *p_limit;
+ register int i, j;
+ unsigned char *pat, *pat_end;
+ int multibyte = ! NILP (current_buffer->enable_multibyte_characters);
+
+ unsigned char simple_translate[0400];
+ int translate_prev_byte = 0;
+ int translate_anteprev_byte = 0;
+
+#ifdef C_ALLOCA
+ int BM_tab_space[0400];
+ BM_tab = &BM_tab_space[0];
+#else
+ BM_tab = (int *) alloca (0400 * sizeof (int));
+#endif
+ /* The general approach is that we are going to maintain that we know */
+ /* the first (closest to the present position, in whatever direction */
+ /* we're searching) character that could possibly be the last */
+ /* (furthest from present position) character of a valid match. We */
+ /* advance the state of our knowledge by looking at that character */
+ /* and seeing whether it indeed matches the last character of the */
+ /* pattern. If it does, we take a closer look. If it does not, we */
+ /* move our pointer (to putative last characters) as far as is */
+ /* logically possible. This amount of movement, which I call a */
+ /* stride, will be the length of the pattern if the actual character */
+ /* appears nowhere in the pattern, otherwise it will be the distance */
+ /* from the last occurrence of that character to the end of the */
+ /* pattern. */
+ /* As a coding trick, an enormous stride is coded into the table for */
+ /* characters that match the last character. This allows use of only */
+ /* a single test, a test for having gone past the end of the */
+ /* permissible match region, to test for both possible matches (when */
+ /* the stride goes past the end immediately) and failure to */
+ /* match (where you get nudged past the end one stride at a time). */
+
+ /* Here we make a "mickey mouse" BM table. The stride of the search */
+ /* is determined only by the last character of the putative match. */
+ /* If that character does not match, we will stride the proper */
+ /* distance to propose a match that superimposes it on the last */
+ /* instance of a character that matches it (per trt), or misses */
+ /* it entirely if there is none. */
+
+ dirlen = len_byte * direction;
+ infinity = dirlen - (lim_byte + pos_byte + len_byte + len_byte) * direction;
+
+ /* Record position after the end of the pattern. */
+ pat_end = base_pat + len_byte;
+ /* BASE_PAT points to a character that we start scanning from.
+ It is the first character in a forward search,
+ the last character in a backward search. */
+ if (direction < 0)
+ base_pat = pat_end - 1;
+
+ BM_tab_base = BM_tab;
+ BM_tab += 0400;
+ j = dirlen; /* to get it in a register */
+ /* A character that does not appear in the pattern induces a */
+ /* stride equal to the pattern length. */
+ while (BM_tab_base != BM_tab)
+ {
+ *--BM_tab = j;
+ *--BM_tab = j;
+ *--BM_tab = j;
+ *--BM_tab = j;
+ }
+
+ /* We use this for translation, instead of TRT itself.
+ We fill this in to handle the characters that actually
+ occur in the pattern. Others don't matter anyway! */
+ bzero (simple_translate, sizeof simple_translate);
+ for (i = 0; i < 0400; i++)
+ simple_translate[i] = i;
+
+ i = 0;
+ while (i != infinity)
+ {
+ unsigned char *ptr = base_pat + i;
+ i += direction;
+ if (i == dirlen)
+ i = infinity;
+ if (! NILP (trt))
{
- j = pat[i]; i += direction;
- if (i == dirlen) i = infinity;
- if (trt != 0)
+ int ch;
+ int untranslated;
+ int this_translated = 1;
+
+ if (multibyte
+ /* Is *PTR the last byte of a character? */
+ && (pat_end - ptr == 1 || CHAR_HEAD_P (ptr[1])))
{
- k = (j = XINT (trt[j]));
- if (i == infinity)
- stride_for_teases = BM_tab[j];
- BM_tab[j] = dirlen - i;
- /* A translation table is accompanied by its inverse -- see */
- /* comment following downcase_table for details */
- while ((j = (unsigned char) XINT (inverse_trt[j])) != k)
- BM_tab[j] = dirlen - i;
+ unsigned char *charstart = ptr;
+ while (! CHAR_HEAD_P (*charstart))
+ charstart--;
+ untranslated = STRING_CHAR (charstart, ptr - charstart + 1);
+ if (charset_base == (untranslated & ~CHAR_FIELD3_MASK))
+ {
+ TRANSLATE (ch, trt, untranslated);
+ if (! CHAR_HEAD_P (*ptr))
+ {
+ translate_prev_byte = ptr[-1];
+ if (! CHAR_HEAD_P (translate_prev_byte))
+ translate_anteprev_byte = ptr[-2];
+ }
+ }
+ else
+ {
+ this_translated = 0;
+ ch = *ptr;
+ }
}
+ else if (!multibyte)
+ TRANSLATE (ch, trt, *ptr);
else
{
- if (i == infinity)
- stride_for_teases = BM_tab[j];
- BM_tab[j] = dirlen - i;
+ ch = *ptr;
+ this_translated = 0;
+ }
+
+ if (ch > 0400)
+ j = ((unsigned char) ch) | 0200;
+ else
+ j = (unsigned char) ch;
+
+ if (i == infinity)
+ stride_for_teases = BM_tab[j];
+
+ BM_tab[j] = dirlen - i;
+ /* A translation table is accompanied by its inverse -- see */
+ /* comment following downcase_table for details */
+ if (this_translated)
+ {
+ int starting_ch = ch;
+ int starting_j = j;
+ while (1)
+ {
+ TRANSLATE (ch, inverse_trt, ch);
+ if (ch > 0400)
+ j = ((unsigned char) ch) | 0200;
+ else
+ j = (unsigned char) ch;
+
+ /* For all the characters that map into CH,
+ set up simple_translate to map the last byte
+ into STARTING_J. */
+ simple_translate[j] = starting_j;
+ if (ch == starting_ch)
+ break;
+ BM_tab[j] = dirlen - i;
+ }
}
- /* stride_for_teases tells how much to stride if we get a */
- /* match on the far character but are subsequently */
- /* disappointed, by recording what the stride would have been */
- /* for that character if the last character had been */
- /* different. */
}
- infinity = dirlen - infinity;
- pos_byte += dirlen - ((direction > 0) ? direction : 0);
- /* loop invariant - POS_BYTE points at where last char (first
- char if reverse) of pattern would align in a possible match. */
- while (n != 0)
+ else
{
- /* It's been reported that some (broken) compiler thinks that
- Boolean expressions in an arithmetic context are unsigned.
- Using an explicit ?1:0 prevents this. */
- if ((lim_byte - pos_byte - ((direction > 0) ? 1 : 0)) * direction
- < 0)
- return (n * (0 - direction));
- /* First we do the part we can by pointers (maybe nothing) */
- QUIT;
- pat = base_pat;
- limit = pos_byte - dirlen + direction;
- limit = ((direction > 0)
- ? BUFFER_CEILING_OF (limit)
- : BUFFER_FLOOR_OF (limit));
+ j = *ptr;
+
+ if (i == infinity)
+ stride_for_teases = BM_tab[j];
+ BM_tab[j] = dirlen - i;
+ }
+ /* stride_for_teases tells how much to stride if we get a */
+ /* match on the far character but are subsequently */
+ /* disappointed, by recording what the stride would have been */
+ /* for that character if the last character had been */
+ /* different. */
+ }
+ infinity = dirlen - infinity;
+ pos_byte += dirlen - ((direction > 0) ? direction : 0);
+ /* loop invariant - POS_BYTE points at where last char (first
+ char if reverse) of pattern would align in a possible match. */
+ while (n != 0)
+ {
+ int tail_end;
+ unsigned char *tail_end_ptr;
+
+ /* It's been reported that some (broken) compiler thinks that
+ Boolean expressions in an arithmetic context are unsigned.
+ Using an explicit ?1:0 prevents this. */
+ if ((lim_byte - pos_byte - ((direction > 0) ? 1 : 0)) * direction
+ < 0)
+ return (n * (0 - direction));
+ /* First we do the part we can by pointers (maybe nothing) */
+ QUIT;
+ pat = base_pat;
+ limit = pos_byte - dirlen + direction;
+ if (direction > 0)
+ {
+ limit = BUFFER_CEILING_OF (limit);
/* LIMIT is now the last (not beyond-last!) value POS_BYTE
can take on without hitting edge of buffer or the gap. */
- limit = ((direction > 0)
- ? min (lim_byte - 1, min (limit, pos_byte + 20000))
- : max (lim_byte, max (limit, pos_byte - 20000)));
- if ((limit - pos_byte) * direction > 20)
+ limit = min (limit, pos_byte + 20000);
+ limit = min (limit, lim_byte - 1);
+ }
+ else
+ {
+ limit = BUFFER_FLOOR_OF (limit);
+ /* LIMIT is now the last (not beyond-last!) value POS_BYTE
+ can take on without hitting edge of buffer or the gap. */
+ limit = max (limit, pos_byte - 20000);
+ limit = max (limit, lim_byte);
+ }
+ tail_end = BUFFER_CEILING_OF (pos_byte) + 1;
+ tail_end_ptr = BYTE_POS_ADDR (tail_end);
+
+ if ((limit - pos_byte) * direction > 20)
+ {
+ unsigned char *p2;
+
+ p_limit = BYTE_POS_ADDR (limit);
+ p2 = (cursor = BYTE_POS_ADDR (pos_byte));
+ /* In this loop, pos + cursor - p2 is the surrogate for pos */
+ while (1) /* use one cursor setting as long as i can */
{
- p_limit = BYTE_POS_ADDR (limit);
- p2 = (cursor = BYTE_POS_ADDR (pos_byte));
- /* In this loop, pos + cursor - p2 is the surrogate for pos */
- while (1) /* use one cursor setting as long as i can */
+ if (direction > 0) /* worth duplicating */
{
- if (direction > 0) /* worth duplicating */
- {
- /* Use signed comparison if appropriate
- to make cursor+infinity sure to be > p_limit.
- Assuming that the buffer lies in a range of addresses
- that are all "positive" (as ints) or all "negative",
- either kind of comparison will work as long
- as we don't step by infinity. So pick the kind
- that works when we do step by infinity. */
- if ((EMACS_INT) (p_limit + infinity) > (EMACS_INT) p_limit)
- while ((EMACS_INT) cursor <= (EMACS_INT) p_limit)
- cursor += BM_tab[*cursor];
- else
- while ((EMACS_UINT) cursor <= (EMACS_UINT) p_limit)
- cursor += BM_tab[*cursor];
- }
+ /* Use signed comparison if appropriate
+ to make cursor+infinity sure to be > p_limit.
+ Assuming that the buffer lies in a range of addresses
+ that are all "positive" (as ints) or all "negative",
+ either kind of comparison will work as long
+ as we don't step by infinity. So pick the kind
+ that works when we do step by infinity. */
+ if ((EMACS_INT) (p_limit + infinity) > (EMACS_INT) p_limit)
+ while ((EMACS_INT) cursor <= (EMACS_INT) p_limit)
+ cursor += BM_tab[*cursor];
else
- {
- if ((EMACS_INT) (p_limit + infinity) < (EMACS_INT) p_limit)
- while ((EMACS_INT) cursor >= (EMACS_INT) p_limit)
- cursor += BM_tab[*cursor];
- else
- while ((EMACS_UINT) cursor >= (EMACS_UINT) p_limit)
- cursor += BM_tab[*cursor];
- }
+ while ((EMACS_UINT) cursor <= (EMACS_UINT) p_limit)
+ cursor += BM_tab[*cursor];
+ }
+ else
+ {
+ if ((EMACS_INT) (p_limit + infinity) < (EMACS_INT) p_limit)
+ while ((EMACS_INT) cursor >= (EMACS_INT) p_limit)
+ cursor += BM_tab[*cursor];
+ else
+ while ((EMACS_UINT) cursor >= (EMACS_UINT) p_limit)
+ cursor += BM_tab[*cursor];
+ }
/* If you are here, cursor is beyond the end of the searched region. */
- /* This can happen if you match on the far character of the pattern, */
- /* because the "stride" of that character is infinity, a number able */
- /* to throw you well beyond the end of the search. It can also */
- /* happen if you fail to match within the permitted region and would */
- /* otherwise try a character beyond that region */
- if ((cursor - p_limit) * direction <= len_byte)
- break; /* a small overrun is genuine */
- cursor -= infinity; /* large overrun = hit */
- i = dirlen - direction;
- if (trt != 0)
+/* This can happen if you match on the far character of the pattern, */
+/* because the "stride" of that character is infinity, a number able */
+/* to throw you well beyond the end of the search. It can also */
+/* happen if you fail to match within the permitted region and would */
+/* otherwise try a character beyond that region */
+ if ((cursor - p_limit) * direction <= len_byte)
+ break; /* a small overrun is genuine */
+ cursor -= infinity; /* large overrun = hit */
+ i = dirlen - direction;
+ if (! NILP (trt))
+ {
+ while ((i -= direction) + direction != 0)
{
- while ((i -= direction) + direction != 0)
- if (pat[i] != XINT (trt[*(cursor -= direction)]))
- break;
+ int ch;
+ cursor -= direction;
+ /* Translate only the last byte of a character. */
+ if (! multibyte
+ || ((cursor == tail_end_ptr
+ || CHAR_HEAD_P (cursor[1]))
+ && (CHAR_HEAD_P (cursor[0])
+ || (translate_prev_byte == cursor[-1]
+ && (CHAR_HEAD_P (translate_prev_byte)
+ || translate_anteprev_byte == cursor[-2])))))
+ ch = simple_translate[*cursor];
+ else
+ ch = *cursor;
+ if (pat[i] != ch)
+ break;
}
- else
+ }
+ else
+ {
+ while ((i -= direction) + direction != 0)
{
- while ((i -= direction) + direction != 0)
- if (pat[i] != *(cursor -= direction))
- break;
+ cursor -= direction;
+ if (pat[i] != *cursor)
+ break;
}
- cursor += dirlen - i - direction; /* fix cursor */
- if (i + direction == 0)
- {
- int position;
+ }
+ cursor += dirlen - i - direction; /* fix cursor */
+ if (i + direction == 0)
+ {
+ int position;
- cursor -= direction;
+ cursor -= direction;
- position = pos_byte + cursor - p2 + ((direction > 0)
- ? 1 - len_byte : 0);
- set_search_regs (position, len_byte);
+ position = pos_byte + cursor - p2 + ((direction > 0)
+ ? 1 - len_byte : 0);
+ set_search_regs (position, len_byte);
- if ((n -= direction) != 0)
- cursor += dirlen; /* to resume search */
- else
- return ((direction > 0)
- ? search_regs.end[0] : search_regs.start[0]);
- }
+ if ((n -= direction) != 0)
+ cursor += dirlen; /* to resume search */
else
- cursor += stride_for_teases; /* <sigh> we lose - */
+ return ((direction > 0)
+ ? search_regs.end[0] : search_regs.start[0]);
}
- pos_byte += cursor - p2;
+ else
+ cursor += stride_for_teases; /* <sigh> we lose - */
}
- else
- /* Now we'll pick up a clump that has to be done the hard */
- /* way because it covers a discontinuity */
+ pos_byte += cursor - p2;
+ }
+ else
+ /* Now we'll pick up a clump that has to be done the hard */
+ /* way because it covers a discontinuity */
+ {
+ limit = ((direction > 0)
+ ? BUFFER_CEILING_OF (pos_byte - dirlen + 1)
+ : BUFFER_FLOOR_OF (pos_byte - dirlen - 1));
+ limit = ((direction > 0)
+ ? min (limit + len_byte, lim_byte - 1)
+ : max (limit - len_byte, lim_byte));
+ /* LIMIT is now the last value POS_BYTE can have
+ and still be valid for a possible match. */
+ while (1)
{
- limit = ((direction > 0)
- ? BUFFER_CEILING_OF (pos_byte - dirlen + 1)
- : BUFFER_FLOOR_OF (pos_byte - dirlen - 1));
- limit = ((direction > 0)
- ? min (limit + len_byte, lim_byte - 1)
- : max (limit - len_byte, lim_byte));
- /* LIMIT is now the last value POS_BYTE can have
- and still be valid for a possible match. */
- while (1)
+ /* This loop can be coded for space rather than */
+ /* speed because it will usually run only once. */
+ /* (the reach is at most len + 21, and typically */
+ /* does not exceed len) */
+ while ((limit - pos_byte) * direction >= 0)
+ pos_byte += BM_tab[FETCH_BYTE (pos_byte)];
+ /* now run the same tests to distinguish going off the */
+ /* end, a match or a phony match. */
+ if ((pos_byte - limit) * direction <= len_byte)
+ break; /* ran off the end */
+ /* Found what might be a match.
+ Set POS_BYTE back to last (first if reverse) pos. */
+ pos_byte -= infinity;
+ i = dirlen - direction;
+ while ((i -= direction) + direction != 0)
{
- /* This loop can be coded for space rather than */
- /* speed because it will usually run only once. */
- /* (the reach is at most len + 21, and typically */
- /* does not exceed len) */
- while ((limit - pos_byte) * direction >= 0)
- pos_byte += BM_tab[FETCH_BYTE (pos_byte)];
- /* now run the same tests to distinguish going off the */
- /* end, a match or a phony match. */
- if ((pos_byte - limit) * direction <= len_byte)
- break; /* ran off the end */
- /* Found what might be a match.
- Set POS_BYTE back to last (first if reverse) pos. */
- pos_byte -= infinity;
- i = dirlen - direction;
- while ((i -= direction) + direction != 0)
- {
- pos_byte -= direction;
- if (pat[i] != (trt != 0
- ? XINT (trt[FETCH_BYTE (pos_byte)])
- : FETCH_BYTE (pos_byte)))
- break;
- }
- /* Above loop has moved POS_BYTE part or all the way
- back to the first pos (last pos if reverse).
- Set it once again at the last (first if reverse) char. */
- pos_byte += dirlen - i- direction;
- if (i + direction == 0)
- {
- int position;
- pos_byte -= direction;
+ int ch;
+ unsigned char *ptr;
+ pos_byte -= direction;
+ ptr = BYTE_POS_ADDR (pos_byte);
+ /* Translate only the last byte of a character. */
+ if (! multibyte
+ || ((ptr == tail_end_ptr
+ || CHAR_HEAD_P (ptr[1]))
+ && (CHAR_HEAD_P (ptr[0])
+ || (translate_prev_byte == ptr[-1]
+ && (CHAR_HEAD_P (translate_prev_byte)
+ || translate_anteprev_byte == ptr[-2])))))
+ ch = simple_translate[*ptr];
+ else
+ ch = *ptr;
+ if (pat[i] != ch)
+ break;
+ }
+ /* Above loop has moved POS_BYTE part or all the way
+ back to the first pos (last pos if reverse).
+ Set it once again at the last (first if reverse) char. */
+ pos_byte += dirlen - i- direction;
+ if (i + direction == 0)
+ {
+ int position;
+ pos_byte -= direction;
- position = pos_byte + ((direction > 0) ? 1 - len_byte : 0);
+ position = pos_byte + ((direction > 0) ? 1 - len_byte : 0);
- set_search_regs (position, len_byte);
+ set_search_regs (position, len_byte);
- if ((n -= direction) != 0)
- pos_byte += dirlen; /* to resume search */
- else
- return ((direction > 0)
- ? search_regs.end[0] : search_regs.start[0]);
- }
+ if ((n -= direction) != 0)
+ pos_byte += dirlen; /* to resume search */
else
- pos_byte += stride_for_teases;
+ return ((direction > 0)
+ ? search_regs.end[0] : search_regs.start[0]);
}
- }
- /* We have done one clump. Can we continue? */
- if ((lim_byte - pos_byte) * direction < 0)
- return ((0 - n) * direction);
- }
- return BYTE_TO_CHAR (pos_byte);
+ else
+ pos_byte += stride_for_teases;
+ }
+ }
+ /* We have done one clump. Can we continue? */
+ if ((lim_byte - pos_byte) * direction < 0)
+ return ((0 - n) * direction);
}
+ return BYTE_TO_CHAR (pos_byte);
}
/* Record beginning BEG_BYTE and end BEG_BYTE + NBYTES
- for a match just found in the current buffer. */
+ for the overall match just found in the current buffer.
+ Also clear out the match data for registers 1 and up. */
static void
set_search_regs (beg_byte, nbytes)
int beg_byte, nbytes;
{
+ int i;
+
/* Make sure we have registers in which to store
the match position. */
if (search_regs.num_regs == 0)
search_regs.num_regs = 2;
}
+ /* Clear out the other registers. */
+ for (i = 1; i < search_regs.num_regs; i++)
+ {
+ search_regs.start[i] = -1;
+ search_regs.end[i] = -1;
+ }
+
search_regs.start[0] = BYTE_TO_CHAR (beg_byte);
search_regs.end[0] = BYTE_TO_CHAR (beg_byte + nbytes);
XSETBUFFER (last_thing_searched, current_buffer);
{
int c;
- if (STRING_MULTIBYTE (string))
- FETCH_STRING_CHAR_ADVANCE (c, string, i, i_byte);
- else
- c = XSTRING (string)->data[i++];
+ FETCH_STRING_CHAR_ADVANCE (c, string, i, i_byte);
if (SYNTAX (c) != Sword)
{
return build_string ("");
adjust = - punct_count + 5 * (word_count - 1) + 4;
- val = make_uninit_multibyte_string (len + adjust,
- XSTRING (string)->size_byte + adjust);
+ if (STRING_MULTIBYTE (string))
+ val = make_uninit_multibyte_string (len + adjust,
+ STRING_BYTES (XSTRING (string))
+ + adjust);
+ else
+ val = make_uninit_string (len + adjust);
o = XSTRING (val)->data;
*o++ = '\\';
*o++ = 'b';
+ prev_c = 0;
- for (i = 0; i < XSTRING (val)->size_byte; i++)
- if (SYNTAX (p[i]) == Sword)
- *o++ = p[i];
- else if (i > 0 && SYNTAX (p[i-1]) == Sword && --word_count)
- {
- *o++ = '\\';
- *o++ = 'W';
- *o++ = '\\';
- *o++ = 'W';
- *o++ = '*';
- }
+ for (i = 0, i_byte = 0; i < len; )
+ {
+ int c;
+ int i_byte_orig = i_byte;
+
+ FETCH_STRING_CHAR_ADVANCE (c, string, i, i_byte);
+
+ if (SYNTAX (c) == Sword)
+ {
+ bcopy (&XSTRING (string)->data[i_byte_orig], o,
+ i_byte - i_byte_orig);
+ o += i_byte - i_byte_orig;
+ }
+ else if (i > 0 && SYNTAX (prev_c) == Sword && --word_count)
+ {
+ *o++ = '\\';
+ *o++ = 'W';
+ *o++ = '\\';
+ *o++ = 'W';
+ *o++ = '*';
+ }
+
+ prev_c = c;
+ }
*o++ = '\\';
*o++ = 'b';
Optional third argument, if t, means if fail just return nil (no error).\n\
If not nil and not t, position at limit of search and return nil.\n\
Optional fourth argument is repeat count--search for successive occurrences.\n\
+\n\
+Search case-sensitivity is determined by the value of the variable\n\
+`case-fold-search', which see.\n\
+\n\
See also the functions `match-beginning', `match-end' and `replace-match'.")
(string, bound, noerror, count)
Lisp_Object string, bound, noerror, count;
Optional third argument, if t, means if fail just return nil (no error).\n\
If not nil and not t, move to limit of search and return nil.\n\
Optional fourth argument is repeat count--search for successive occurrences.\n\
+\n\
+Search case-sensitivity is determined by the value of the variable\n\
+`case-fold-search', which see.\n\
+\n\
See also the functions `match-beginning', `match-end' and `replace-match'.")
(string, bound, noerror, count)
Lisp_Object string, bound, noerror, count;
Optional third argument, if t, means if fail just return nil (no error).\n\
If not nil and not t, move to limit of search and return nil.\n\
Optional fourth argument is repeat count--search for successive occurrences.\n\
-See also the functions `match-beginning', `match-end' and `replace-match'.")
+See also the functions `match-beginning', `match-end', `match-string',\n\
+and `replace-match'.")
(regexp, bound, noerror, count)
Lisp_Object regexp, bound, noerror, count;
{
Optional third argument, if t, means if fail just return nil (no error).\n\
If not nil and not t, move to limit of search and return nil.\n\
Optional fourth argument is repeat count--search for successive occurrences.\n\
-See also the functions `match-beginning', `match-end' and `replace-match'.")
+See also the functions `match-beginning', `match-end', `match-string',\n\
+and `replace-match'.")
(regexp, bound, noerror, count)
Lisp_Object regexp, bound, noerror, count;
{
Optional third argument, if t, means if fail just return nil (no error).\n\
If not nil and not t, move to limit of search and return nil.\n\
Optional fourth argument is repeat count--search for successive occurrences.\n\
-See also the functions `match-beginning', `match-end' and `replace-match'.")
+See also the functions `match-beginning', `match-end', `match-string',\n\
+and `replace-match'.")
(regexp, bound, noerror, count)
Lisp_Object regexp, bound, noerror, count;
{
Optional third argument, if t, means if fail just return nil (no error).\n\
If not nil and not t, move to limit of search and return nil.\n\
Optional fourth argument is repeat count--search for successive occurrences.\n\
-See also the functions `match-beginning', `match-end' and `replace-match'.")
+See also the functions `match-beginning', `match-end', `match-string',\n\
+and `replace-match'.")
(regexp, bound, noerror, count)
Lisp_Object regexp, bound, noerror, count;
{
Lisp_Object newtext, fixedcase, literal, string, subexp;
{
enum { nochange, all_caps, cap_initial } case_action;
- register int pos, last;
+ register int pos, pos_byte;
int some_multiletter_word;
int some_lowercase;
int some_uppercase;
if (NILP (fixedcase))
{
- int beg;
/* Decide how to casify by examining the matched text. */
+ int last;
- if (NILP (string))
- last = CHAR_TO_BYTE (search_regs.end[sub]);
- else
- last = search_regs.end[sub];
+ pos = search_regs.start[sub];
+ last = search_regs.end[sub];
if (NILP (string))
- beg = CHAR_TO_BYTE (search_regs.start[sub]);
+ pos_byte = CHAR_TO_BYTE (pos);
else
- beg = search_regs.start[sub];
+ pos_byte = string_char_to_byte (string, pos);
prevc = '\n';
case_action = all_caps;
some_nonuppercase_initial = 0;
some_uppercase = 0;
- for (pos = beg; pos < last; pos++)
+ while (pos < last)
{
if (NILP (string))
- c = FETCH_BYTE (pos);
+ {
+ c = FETCH_CHAR (pos_byte);
+ INC_BOTH (pos, pos_byte);
+ }
else
- c = XSTRING (string)->data[pos];
+ FETCH_STRING_CHAR_ADVANCE (c, string, pos, pos_byte);
if (LOWERCASEP (c))
{
if desired. */
if (NILP (literal))
{
- int lastpos = -1;
- int lastpos_byte = -1;
+ int lastpos = 0;
+ int lastpos_byte = 0;
/* We build up the substituted string in ACCUM. */
Lisp_Object accum;
Lisp_Object middle;
- int pos_byte;
+ int length = STRING_BYTES (XSTRING (newtext));
accum = Qnil;
- for (pos_byte = 0, pos = 0; pos_byte < XSTRING (newtext)->size_byte;)
+ for (pos_byte = 0, pos = 0; pos_byte < length;)
{
int substart = -1;
- int subend;
+ int subend = 0;
int delbackslash = 0;
FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte);
if (c == '\\')
{
FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte);
+
if (c == '&')
{
substart = search_regs.start[sub];
}
if (substart >= 0)
{
- if (pos - 1 != lastpos + 1)
- middle = substring_both (newtext, lastpos + 1,
- lastpos_byte + 1,
- pos - 1, pos_byte - 1);
+ if (pos - 2 != lastpos)
+ middle = substring_both (newtext, lastpos,
+ lastpos_byte,
+ pos - 2, pos_byte - 2);
else
middle = Qnil;
accum = concat3 (accum, middle,
}
else if (delbackslash)
{
- middle = substring_both (newtext, lastpos + 1,
- lastpos_byte + 1,
- pos, pos_byte);
+ middle = substring_both (newtext, lastpos,
+ lastpos_byte,
+ pos - 1, pos_byte - 1);
accum = concat2 (accum, middle);
lastpos = pos;
}
}
- if (pos != lastpos + 1)
- middle = substring_both (newtext, lastpos + 1,
- lastpos_byte + 1,
+ if (pos != lastpos)
+ middle = substring_both (newtext, lastpos,
+ lastpos_byte,
pos, pos_byte);
else
middle = Qnil;
}
/* Record point, the move (quietly) to the start of the match. */
- if (PT > search_regs.start[sub])
+ if (PT >= search_regs.end[sub])
opoint = PT - ZV;
+ else if (PT > search_regs.start[sub])
+ opoint = search_regs.end[sub] - ZV;
else
opoint = PT;
Finsert_and_inherit (1, &newtext);
else
{
- struct gcpro gcpro1;
- GCPRO1 (newtext);
-
- for (pos = 0; pos < XSTRING (newtext)->size; pos++)
+ int length = STRING_BYTES (XSTRING (newtext));
+ unsigned char *substed;
+ int substed_alloc_size, substed_len;
+ int buf_multibyte = !NILP (current_buffer->enable_multibyte_characters);
+ int str_multibyte = STRING_MULTIBYTE (newtext);
+ Lisp_Object rev_tbl;
+
+ rev_tbl= (!buf_multibyte && CHAR_TABLE_P (Vnonascii_translation_table)
+ ? Fchar_table_extra_slot (Vnonascii_translation_table,
+ make_number (0))
+ : Qnil);
+
+ substed_alloc_size = length * 2 + 100;
+ substed = (unsigned char *) xmalloc (substed_alloc_size + 1);
+ substed_len = 0;
+
+ /* Go thru NEWTEXT, producing the actual text to insert in
+ SUBSTED while adjusting multibyteness to that of the current
+ buffer. */
+
+ for (pos_byte = 0, pos = 0; pos_byte < length;)
{
- int offset = PT - search_regs.start[sub];
+ unsigned char str[MAX_MULTIBYTE_LENGTH];
+ unsigned char *add_stuff = NULL;
+ int add_len = 0;
+ int idx = -1;
+
+ if (str_multibyte)
+ {
+ FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, pos, pos_byte);
+ if (!buf_multibyte)
+ c = multibyte_char_to_unibyte (c, rev_tbl);
+ }
+ else
+ {
+ /* Note that we don't have to increment POS. */
+ c = XSTRING (newtext)->data[pos_byte++];
+ if (buf_multibyte)
+ c = unibyte_char_to_multibyte (c);
+ }
+
+ /* Either set ADD_STUFF and ADD_LEN to the text to put in SUBSTED,
+ or set IDX to a match index, which means put that part
+ of the buffer text into SUBSTED. */
- c = XSTRING (newtext)->data[pos];
if (c == '\\')
{
- c = XSTRING (newtext)->data[++pos];
+ if (str_multibyte)
+ {
+ FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext,
+ pos, pos_byte);
+ if (!buf_multibyte && !SINGLE_BYTE_CHAR_P (c))
+ c = multibyte_char_to_unibyte (c, rev_tbl);
+ }
+ else
+ {
+ c = XSTRING (newtext)->data[pos_byte++];
+ if (buf_multibyte)
+ c = unibyte_char_to_multibyte (c);
+ }
+
if (c == '&')
- Finsert_buffer_substring
- (Fcurrent_buffer (),
- make_number (search_regs.start[sub] + offset),
- make_number (search_regs.end[sub] + offset));
+ idx = sub;
else if (c >= '1' && c <= '9' && c <= search_regs.num_regs + '0')
{
if (search_regs.start[c - '0'] >= 1)
- Finsert_buffer_substring
- (Fcurrent_buffer (),
- make_number (search_regs.start[c - '0'] + offset),
- make_number (search_regs.end[c - '0'] + offset));
+ idx = c - '0';
}
else if (c == '\\')
- insert_char (c);
+ add_len = 1, add_stuff = "\\";
else
- error ("Invalid use of `\\' in replacement text");
+ {
+ xfree (substed);
+ error ("Invalid use of `\\' in replacement text");
+ }
}
else
- insert_char (c);
+ {
+ add_len = CHAR_STRING (c, str);
+ add_stuff = str;
+ }
+
+ /* If we want to copy part of a previous match,
+ set up ADD_STUFF and ADD_LEN to point to it. */
+ if (idx >= 0)
+ {
+ int begbyte = CHAR_TO_BYTE (search_regs.start[idx]);
+ add_len = CHAR_TO_BYTE (search_regs.end[idx]) - begbyte;
+ if (search_regs.start[idx] < GPT && GPT < search_regs.end[idx])
+ move_gap (search_regs.start[idx]);
+ add_stuff = BYTE_POS_ADDR (begbyte);
+ }
+
+ /* Now the stuff we want to add to SUBSTED
+ is invariably ADD_LEN bytes starting at ADD_STUFF. */
+
+ /* Make sure SUBSTED is big enough. */
+ if (substed_len + add_len >= substed_alloc_size)
+ {
+ substed_alloc_size = substed_len + add_len + 500;
+ substed = (unsigned char *) xrealloc (substed,
+ substed_alloc_size + 1);
+ }
+
+ /* Now add to the end of SUBSTED. */
+ if (add_stuff)
+ {
+ bcopy (add_stuff, substed + substed_len, add_len);
+ substed_len += add_len;
+ }
}
- UNGCPRO;
+
+ /* Now insert what we accumulated. */
+ insert_and_inherit (substed, substed_len);
+
+ xfree (substed);
}
inslen = PT - (search_regs.start[sub]);
if (NILP (last_thing_searched))
return Qnil;
+ prev = Qnil;
+
data = (Lisp_Object *) alloca ((2 * search_regs.num_regs)
* sizeof (Lisp_Object));
/* If REUSE is a list, store as many value elements as will fit
into the elements of REUSE. */
for (i = 0, tail = reuse; CONSP (tail);
- i++, tail = XCONS (tail)->cdr)
+ i++, tail = XCDR (tail))
{
if (i < 2 * len + 2)
- XCONS (tail)->car = data[i];
+ XCAR (tail) = data[i];
else
- XCONS (tail)->car = Qnil;
+ XCAR (tail) = Qnil;
prev = tail;
}
/* If we couldn't fit all value elements into REUSE,
cons up the rest of them and add them to the end of REUSE. */
if (i < 2 * len + 2)
- XCONS (prev)->cdr = Flist (2 * len + 2 - i, data + i);
+ XCDR (prev) = Flist (2 * len + 2 - i, data + i);
return reuse;
}
-DEFUN ("store-match-data", Fstore_match_data, Sstore_match_data, 1, 1, 0,
+DEFUN ("set-match-data", Fset_match_data, Sset_match_data, 1, 1, 0,
"Set internal data on last search match from elements of LIST.\n\
LIST should have been created by calling `match-data' previously.")
(list)
length * sizeof (regoff_t));
}
+ for (i = search_regs.num_regs; i < length; i++)
+ search_regs.start[i] = -1;
+
search_regs.num_regs = length;
}
}
}
else
{
+ int from;
+
if (MARKERP (marker))
{
if (XMARKER (marker)->buffer == 0)
}
CHECK_NUMBER_COERCE_MARKER (marker, 0);
- search_regs.start[i] = XINT (marker);
+ from = XINT (marker);
list = Fcdr (list);
marker = Fcar (list);
XSETFASTINT (marker, 0);
CHECK_NUMBER_COERCE_MARKER (marker, 0);
+ search_regs.start[i] = from;
search_regs.end[i] = XINT (marker);
}
list = Fcdr (list);
CHECK_STRING (string, 0);
- temp = (unsigned char *) alloca (XSTRING (string)->size_byte * 2);
+ temp = (unsigned char *) alloca (STRING_BYTES (XSTRING (string)) * 2);
/* Now copy the data into the new string, inserting escapes. */
in = XSTRING (string)->data;
- end = in + XSTRING (string)->size_byte;
+ end = in + STRING_BYTES (XSTRING (string));
out = temp;
for (; in != end; in++)
*out++ = *in;
}
- return make_multibyte_string (temp,
+ return make_specified_string (temp,
XSTRING (string)->size + backslashes_added,
- out - temp);
+ out - temp,
+ STRING_MULTIBYTE (string));
}
\f
+void
syms_of_search ()
{
register int i;
defsubr (&Smatch_beginning);
defsubr (&Smatch_end);
defsubr (&Smatch_data);
- defsubr (&Sstore_match_data);
+ defsubr (&Sset_match_data);
defsubr (&Sregexp_quote);
}
HCoop / bpt / emacs.git / blobdiff