comparison src/search.c @ 103001:6b3176775d00

* search.c: Use EMACS_INT for buffer positions. Add prototypes. * lisp.h: Adjust prototypes.

103000:f5a6ea840fe4

/* If non-nil, the match data will not be changed during call to
   searching or matching functions.  This variable is for internal use
   only.  */
Lisp_Object Vinhibit_changing_match_data;

static void set_search_regs ();
static void save_search_regs ();
static int simple_search ();
static int boyer_moore ();
static int search_buffer ();
static void matcher_overflow () NO_RETURN;

static void
matcher_overflow ()
{

     Lisp_Object string;
     int posix;
{
  Lisp_Object val;
  unsigned char *p1, *p2;
  int s1, s2;
  register int i;
  struct re_pattern_buffer *bufp;

  if (running_asynch_code)
    save_search_regs ();

     Lisp_Object regexp, string, start;
     int posix;
{
  int val;
  struct re_pattern_buffer *bufp;
  int pos, pos_byte;
  int i;

  if (running_asynch_code)
    save_search_regs ();

     Lisp_Object string;
{
  int multibyte;
  struct re_pattern_buffer *buf;
  unsigned char *p1, *p2;
  int s1, s2;
  EMACS_INT len;
  
  if (STRINGP (string))
    {
      if (pos_byte < 0)

   except when inside redisplay.  */

int
scan_buffer (target, start, end, count, shortage, allow_quit)
     register int target;
     int start, end;
     int count;
     int *shortage;
     int allow_quit;
{
  struct region_cache *newline_cache;

        /* Our innermost scanning loop is very simple; it doesn't know
           about gaps, buffer ends, or the newline cache.  ceiling is
           the position of the last character before the next such
           obstacle --- the last character the dumb search loop should
           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.  */
        if (target == '\n' && newline_cache)
          {

      }
  else
    while (start > end)
      {
        /* 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)
          {
            int next_change;

   If ALLOW_QUIT is non-zero, set immediate_quit.  That's good to do
   except in special cases.  */

int
scan_newline (start, start_byte, limit, limit_byte, count, allow_quit)
     int start, start_byte;
     int limit, limit_byte;
     register int count;
     int allow_quit;
{
  int direction = ((count > 0) ? 1 : -1);

  register unsigned char *cursor;
  unsigned char *base;

  register int ceiling;
  register unsigned char *ceiling_addr;

  int old_immediate_quit = immediate_quit;

  /* The code that follows is like scan_buffer

  return count * direction;
}

int
find_next_newline_no_quit (from, cnt)
     register int from, cnt;
{
  return scan_buffer ('\n', from, 0, cnt, (int *) 0, 0);
}

/* Like find_next_newline, but returns position before the newline,
   not after, and only search up to TO.  This isn't just
   find_next_newline (...)-1, because you might hit TO.  */

int
find_before_next_newline (from, to, cnt)
     int from, to, cnt;
{
  int shortage;
  int pos = scan_buffer ('\n', from, to, cnt, &shortage, 1);

  if (shortage == 0)

/* Only used in search_buffer, to record the end position of the match
   when searching regexps and SEARCH_REGS should not be changed
   (i.e. Vinhibit_changing_match_data is non-nil).  */
static struct re_registers search_regs_1;

static int
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;
     int posix;

   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.  */

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;
  /* Number of buffer bytes matched.  Note that this may be different
     from len_byte in a multibyte buffer.  */

    while (n > 0)
      {
	while (1)
	  {
	    /* 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 || pos_byte + len_byte > lim_byte)
	      goto stop;

    while (n > 0)
      {
	while (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 (n < 0)
      {
	while (1)
	  {
	    /* Try matching at position POS.  */
	    int this_pos = pos - len;
	    int this_pos_byte;
	    int this_len = len;
	    int this_len_byte = len_byte;
	    unsigned char *p = pat;

	    if (this_pos < lim || (pos_byte - len_byte) < lim_byte)

    while (n < 0)
      {
	while (1)
	  {
	    /* Try matching at position POS.  */
	    int this_pos = pos - len;
	    int this_len = len;
	    unsigned char *p = pat;

	    if (this_pos < lim)
	      goto stop;

   character, and do so after just a simple test of the context.
   CHAR_BASE is nonzero if there is such a non-ASCII character.

   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, char_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 char_base;
{
  int direction = ((n > 0) ? 1 : -1);
  register int dirlen;
  int limit, stride_for_teases = 0;
  int BM_tab[0400];
  register unsigned char *cursor, *p_limit;
  register int i, j;
  unsigned char *pat, *pat_end;
  int multibyte = ! NILP (current_buffer->enable_multibyte_characters);

  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.  */

		    }
		}
	      cursor += dirlen - i - direction;	/* fix cursor */
	      if (i + direction == 0)
		{
		  int position, start, end;

		  cursor -= direction;

		  position = pos_byte + cursor - p2 + ((direction > 0)
						       ? 1 - len_byte : 0);

		 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, start, end;
		  pos_byte -= direction;

		  position = pos_byte + ((direction > 0) ? 1 - len_byte : 0);
		  set_search_regs (position, len_byte);

   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;

  if (!NILP (Vinhibit_changing_match_data))
    return;

  int some_lowercase;
  int some_uppercase;
  int some_nonuppercase_initial;
  register int c, prevc;
  int sub;
  int opoint, newpoint;

  CHECK_STRING (newtext);

  if (! NILP (string))
    CHECK_STRING (string);

    }

  if (NILP (fixedcase))
    {
      /* Decide how to casify by examining the matched text. */
      int last;

      pos = search_regs.start[sub];
      last = search_regs.end[sub];

      if (NILP (string))

      /* Substitute parts of the match into NEWTEXT
	 if desired.  */
      if (NILP (literal))
	{
	  int lastpos = 0;
	  int lastpos_byte = 0;
	  /* We build up the substituted string in ACCUM.  */
	  Lisp_Object accum;
	  Lisp_Object middle;
	  int length = SBYTES (newtext);

	  /* 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);
	    }

    Fupcase_initials_region (make_number (search_regs.start[sub]),
			     make_number (newpoint));

  /* Adjust search data for this change.  */
  {
    int oldend = search_regs.end[sub];
    int oldstart = search_regs.start[sub];
    int change = newpoint - search_regs.end[sub];
    int i;

    for (i = 0; i < search_regs.num_regs; i++)
      {
	if (search_regs.start[i] >= oldend)

	    search_regs.start[i] = -1;
	    list = XCDR (list);
	  }
	else
	  {
	    int from;
	    Lisp_Object m;

	    m = marker;
	    if (MARKERP (marker))
	      {

103001:6b3176775d00

/* If non-nil, the match data will not be changed during call to
   searching or matching functions.  This variable is for internal use
   only.  */
Lisp_Object Vinhibit_changing_match_data;

static void set_search_regs P_ ((EMACS_INT, EMACS_INT));
static void save_search_regs P_ ((void));
static EMACS_INT simple_search P_ ((int, unsigned char *, int, int, 
				    Lisp_Object, EMACS_INT, EMACS_INT,
				    EMACS_INT, EMACS_INT));
static EMACS_INT boyer_moore P_ ((int, unsigned char *, int, int, 
				  Lisp_Object, Lisp_Object,
				  EMACS_INT, EMACS_INT,
				  EMACS_INT, EMACS_INT, int));
static EMACS_INT search_buffer P_ ((Lisp_Object, EMACS_INT, EMACS_INT,
				    EMACS_INT, EMACS_INT, int, int,
				    Lisp_Object, Lisp_Object, int));
static void matcher_overflow () NO_RETURN;

static void
matcher_overflow ()
{

     Lisp_Object string;
     int posix;
{
  Lisp_Object val;
  unsigned char *p1, *p2;
  EMACS_INT s1, s2;
  register int i;
  struct re_pattern_buffer *bufp;

  if (running_asynch_code)
    save_search_regs ();

     Lisp_Object regexp, string, start;
     int posix;
{
  int val;
  struct re_pattern_buffer *bufp;
  EMACS_INT pos, pos_byte;
  int i;

  if (running_asynch_code)
    save_search_regs ();

     Lisp_Object string;
{
  int multibyte;
  struct re_pattern_buffer *buf;
  unsigned char *p1, *p2;
  EMACS_INT s1, s2;
  EMACS_INT len;
  
  if (STRINGP (string))
    {
      if (pos_byte < 0)

   except when inside redisplay.  */

int
scan_buffer (target, start, end, count, shortage, allow_quit)
     register int target;
     EMACS_INT start, end;
     int count;
     int *shortage;
     int allow_quit;
{
  struct region_cache *newline_cache;

        /* Our innermost scanning loop is very simple; it doesn't know
           about gaps, buffer ends, or the newline cache.  ceiling is
           the position of the last character before the next such
           obstacle --- the last character the dumb search loop should
           examine.  */
	EMACS_INT ceiling_byte = CHAR_TO_BYTE (end) - 1;
	EMACS_INT start_byte = CHAR_TO_BYTE (start);
	EMACS_INT tem;

        /* If we're looking for a newline, consult the newline cache
           to see where we can avoid some scanning.  */
        if (target == '\n' && newline_cache)
          {

      }
  else
    while (start > end)
      {
        /* The last character to check before the next obstacle.  */
	EMACS_INT ceiling_byte = CHAR_TO_BYTE (end);
	EMACS_INT start_byte = CHAR_TO_BYTE (start);
	EMACS_INT tem;

        /* Consult the newline cache, if appropriate.  */
        if (target == '\n' && newline_cache)
          {
            int next_change;

   If ALLOW_QUIT is non-zero, set immediate_quit.  That's good to do
   except in special cases.  */

int
scan_newline (start, start_byte, limit, limit_byte, count, allow_quit)
     EMACS_INT start, start_byte;
     EMACS_INT limit, limit_byte;
     register int count;
     int allow_quit;
{
  int direction = ((count > 0) ? 1 : -1);

  register unsigned char *cursor;
  unsigned char *base;

  EMACS_INT ceiling;
  register unsigned char *ceiling_addr;

  int old_immediate_quit = immediate_quit;

  /* The code that follows is like scan_buffer

  return count * direction;
}

int
find_next_newline_no_quit (from, cnt)
     EMACS_INT from;
     int cnt;
{
  return scan_buffer ('\n', from, 0, cnt, (int *) 0, 0);
}

/* Like find_next_newline, but returns position before the newline,
   not after, and only search up to TO.  This isn't just
   find_next_newline (...)-1, because you might hit TO.  */

int
find_before_next_newline (from, to, cnt)
     EMACS_INT from, to;
     int cnt;
{
  int shortage;
  int pos = scan_buffer ('\n', from, to, cnt, &shortage, 1);

  if (shortage == 0)

/* Only used in search_buffer, to record the end position of the match
   when searching regexps and SEARCH_REGS should not be changed
   (i.e. Vinhibit_changing_match_data is non-nil).  */
static struct re_registers search_regs_1;

static EMACS_INT
search_buffer (string, pos, pos_byte, lim, lim_byte, n,
	       RE, trt, inverse_trt, posix)
     Lisp_Object string;
     EMACS_INT pos;
     EMACS_INT pos_byte;
     EMACS_INT lim;
     EMACS_INT lim_byte;
     int n;
     int RE;
     Lisp_Object trt;
     Lisp_Object inverse_trt;
     int posix;

   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.  */

static EMACS_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;
     EMACS_INT pos, pos_byte;
     EMACS_INT lim, lim_byte;
{
  int multibyte = ! NILP (current_buffer->enable_multibyte_characters);
  int forward = n > 0;
  /* Number of buffer bytes matched.  Note that this may be different
     from len_byte in a multibyte buffer.  */

    while (n > 0)
      {
	while (1)
	  {
	    /* Try matching at position POS.  */
	    EMACS_INT this_pos = pos;
	    EMACS_INT this_pos_byte = pos_byte;
	    int this_len = len;
	    int this_len_byte = len_byte;
	    unsigned char *p = pat;
	    if (pos + len > lim || pos_byte + len_byte > lim_byte)
	      goto stop;

    while (n > 0)
      {
	while (1)
	  {
	    /* Try matching at position POS.  */
	    EMACS_INT this_pos = pos;
	    int this_len = len;
	    unsigned char *p = pat;

	    if (pos + len > lim)
	      goto stop;

    while (n < 0)
      {
	while (1)
	  {
	    /* Try matching at position POS.  */
	    EMACS_INT this_pos = pos - len;
	    EMACS_INT this_pos_byte;
	    int this_len = len;
	    int this_len_byte = len_byte;
	    unsigned char *p = pat;

	    if (this_pos < lim || (pos_byte - len_byte) < lim_byte)

    while (n < 0)
      {
	while (1)
	  {
	    /* Try matching at position POS.  */
	    EMACS_INT this_pos = pos - len;
	    int this_len = len;
	    unsigned char *p = pat;

	    if (this_pos < lim)
	      goto stop;

   character, and do so after just a simple test of the context.
   CHAR_BASE is nonzero if there is such a non-ASCII character.

   If that criterion is not satisfied, do not call this function.  */

static EMACS_INT
boyer_moore (n, base_pat, len, len_byte, trt, inverse_trt,
	     pos, pos_byte, lim, lim_byte, char_base)
     int n;
     unsigned char *base_pat;
     int len, len_byte;
     Lisp_Object trt;
     Lisp_Object inverse_trt;
     EMACS_INT pos, pos_byte;
     EMACS_INT lim, lim_byte;
     int char_base;
{
  int direction = ((n > 0) ? 1 : -1);
  register int dirlen;
  EMACS_INT limit;
  int stride_for_teases = 0;
  int BM_tab[0400];
  register unsigned char *cursor, *p_limit;
  register int i, j;
  unsigned char *pat, *pat_end;
  int multibyte = ! NILP (current_buffer->enable_multibyte_characters);

  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)
    {
      EMACS_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.  */

		    }
		}
	      cursor += dirlen - i - direction;	/* fix cursor */
	      if (i + direction == 0)
		{
		  EMACS_INT position, start, end;

		  cursor -= direction;

		  position = pos_byte + cursor - p2 + ((direction > 0)
						       ? 1 - len_byte : 0);

		 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)
		{
		  EMACS_INT position, start, end;
		  pos_byte -= direction;

		  position = pos_byte + ((direction > 0) ? 1 - len_byte : 0);
		  set_search_regs (position, len_byte);

   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)
     EMACS_INT beg_byte, nbytes;
{
  int i;

  if (!NILP (Vinhibit_changing_match_data))
    return;

  int some_lowercase;
  int some_uppercase;
  int some_nonuppercase_initial;
  register int c, prevc;
  int sub;
  EMACS_INT opoint, newpoint;

  CHECK_STRING (newtext);

  if (! NILP (string))
    CHECK_STRING (string);

    }

  if (NILP (fixedcase))
    {
      /* Decide how to casify by examining the matched text. */
      EMACS_INT last;

      pos = search_regs.start[sub];
      last = search_regs.end[sub];

      if (NILP (string))

      /* Substitute parts of the match into NEWTEXT
	 if desired.  */
      if (NILP (literal))
	{
	  EMACS_INT lastpos = 0;
	  EMACS_INT lastpos_byte = 0;
	  /* We build up the substituted string in ACCUM.  */
	  Lisp_Object accum;
	  Lisp_Object middle;
	  int length = SBYTES (newtext);

	  /* If we want to copy part of a previous match,
	     set up ADD_STUFF and ADD_LEN to point to it.  */
	  if (idx >= 0)
	    {
	      EMACS_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);
	    }

    Fupcase_initials_region (make_number (search_regs.start[sub]),
			     make_number (newpoint));

  /* Adjust search data for this change.  */
  {
    EMACS_INT oldend = search_regs.end[sub];
    EMACS_INT oldstart = search_regs.start[sub];
    EMACS_INT change = newpoint - search_regs.end[sub];
    int i;

    for (i = 0; i < search_regs.num_regs; i++)
      {
	if (search_regs.start[i] >= oldend)

	    search_regs.start[i] = -1;
	    list = XCDR (list);
	  }
	else
	  {
	    EMACS_INT from;
	    Lisp_Object m;

	    m = marker;
	    if (MARKERP (marker))
	      {