arrfab / rpms / glibc

Forked from rpms/glibc 5 years ago
Clone

Blame SOURCES/glibc-rh989862.patch

147e83
commit e91a48408662e591e1320b2cab6d3fcda7793b18
147e83
Author: Siddhesh Poyarekar <siddhesh@redhat.com>
147e83
Date:   Fri Jun 28 16:24:16 2013 +0530
147e83
147e83
    Simplify strcoll implementation
147e83
147e83
diff --git glibc-2.17-c758a686/string/strcoll_l.c glibc-2.17-c758a686/string/strcoll_l.c
147e83
index ecda08f..1bb9e23 100644
147e83
--- glibc-2.17-c758a686/string/strcoll_l.c
147e83
+++ glibc-2.17-c758a686/string/strcoll_l.c
147e83
@@ -41,11 +41,244 @@
147e83
 
147e83
 #include "../locale/localeinfo.h"
147e83
 
147e83
+/* Track status while looking for sequences in a string.  */
147e83
+typedef struct
147e83
+{
147e83
+  int len;			/* Length of the current sequence.  */
147e83
+  int val;			/* Position of the sequence relative to the
147e83
+				   previous non-ignored sequence.  */
147e83
+  size_t idxnow;		/* Current index in sequences.  */
147e83
+  size_t idxmax;		/* Maximum index in sequences.  */
147e83
+  size_t idxcnt;		/* Current count of indeces.  */
147e83
+  size_t backw;			/* Current Backward sequence index.  */
147e83
+  size_t backw_stop;		/* Index where the backward sequences stop.  */
147e83
+  const USTRING_TYPE *us;	/* The string.  */
147e83
+  int32_t *idxarr;		/* Array to cache weight indeces.  */
147e83
+  unsigned char *rulearr;	/* Array to cache rules.  */
147e83
+} coll_seq;
147e83
+
147e83
+/* Get next sequence.  The weight indeces are cached, so we don't need to
147e83
+   traverse the string.  */
147e83
+static void
147e83
+get_next_seq_cached (coll_seq *seq, int nrules, int pass,
147e83
+		     const unsigned char *rulesets,
147e83
+		     const USTRING_TYPE *weights)
147e83
+{
147e83
+  int val = seq->val = 0;
147e83
+  int len = seq->len;
147e83
+  size_t backw_stop = seq->backw_stop;
147e83
+  size_t backw = seq->backw;
147e83
+  size_t idxcnt = seq->idxcnt;
147e83
+  size_t idxmax = seq->idxmax;
147e83
+  size_t idxnow = seq->idxnow;
147e83
+  unsigned char *rulearr = seq->rulearr;
147e83
+  int32_t *idxarr = seq->idxarr;
147e83
+
147e83
+  while (len == 0)
147e83
+    {
147e83
+      ++val;
147e83
+      if (backw_stop != ~0ul)
147e83
+	{
147e83
+	  /* The is something pushed.  */
147e83
+	  if (backw == backw_stop)
147e83
+	    {
147e83
+	      /* The last pushed character was handled.  Continue
147e83
+		 with forward characters.  */
147e83
+	      if (idxcnt < idxmax)
147e83
+		{
147e83
+		  idxnow = idxcnt;
147e83
+		  backw_stop = ~0ul;
147e83
+		}
147e83
+	      else
147e83
+		{
147e83
+		  /* Nothing anymore.  The backward sequence
147e83
+		     ended with the last sequence in the string.  */
147e83
+		  idxnow = ~0ul;
147e83
+		  break;
147e83
+		}
147e83
+	    }
147e83
+	  else
147e83
+	    idxnow = --backw;
147e83
+	}
147e83
+      else
147e83
+	{
147e83
+	  backw_stop = idxcnt;
147e83
+
147e83
+	  while (idxcnt < idxmax)
147e83
+	    {
147e83
+	      if ((rulesets[rulearr[idxcnt] * nrules + pass]
147e83
+		   & sort_backward) == 0)
147e83
+		/* No more backward characters to push.  */
147e83
+		break;
147e83
+	      ++idxcnt;
147e83
+	    }
147e83
+
147e83
+	  if (backw_stop == idxcnt)
147e83
+	    {
147e83
+	      /* No sequence at all or just one.  */
147e83
+	      if (idxcnt == idxmax)
147e83
+		/* Note that seq1len is still zero.  */
147e83
+		break;
147e83
+
147e83
+	      backw_stop = ~0ul;
147e83
+	      idxnow = idxcnt++;
147e83
+	    }
147e83
+	  else
147e83
+	    /* We pushed backward sequences.  */
147e83
+	    idxnow = backw = idxcnt - 1;
147e83
+	}
147e83
+      len = weights[idxarr[idxnow]++];
147e83
+    }
147e83
+
147e83
+  /* Update the structure.  */
147e83
+  seq->val = val;
147e83
+  seq->len = len;
147e83
+  seq->backw_stop = backw_stop;
147e83
+  seq->backw = backw;
147e83
+  seq->idxcnt = idxcnt;
147e83
+  seq->idxnow = idxnow;
147e83
+}
147e83
+
147e83
+/* Get next sequence.  Traverse the string as required.  */
147e83
+static void
147e83
+get_next_seq (coll_seq *seq, int nrules, const unsigned char *rulesets,
147e83
+	      const USTRING_TYPE *weights, const int32_t *table,
147e83
+	      const USTRING_TYPE *extra, const int32_t *indirect)
147e83
+{
147e83
+#include WEIGHT_H
147e83
+  int val = seq->val = 0;
147e83
+  int len = seq->len;
147e83
+  size_t backw_stop = seq->backw_stop;
147e83
+  size_t backw = seq->backw;
147e83
+  size_t idxcnt = seq->idxcnt;
147e83
+  size_t idxmax = seq->idxmax;
147e83
+  size_t idxnow = seq->idxnow;
147e83
+  unsigned char *rulearr = seq->rulearr;
147e83
+  int32_t *idxarr = seq->idxarr;
147e83
+  const USTRING_TYPE *us = seq->us;
147e83
+
147e83
+  while (len == 0)
147e83
+    {
147e83
+      ++val;
147e83
+      if (backw_stop != ~0ul)
147e83
+	{
147e83
+	  /* The is something pushed.  */
147e83
+	  if (backw == backw_stop)
147e83
+	    {
147e83
+	      /* The last pushed character was handled.  Continue
147e83
+		 with forward characters.  */
147e83
+	      if (idxcnt < idxmax)
147e83
+		{
147e83
+		  idxnow = idxcnt;
147e83
+		  backw_stop = ~0ul;
147e83
+		}
147e83
+	      else
147e83
+		/* Nothing anymore.  The backward sequence ended with
147e83
+		   the last sequence in the string.  Note that seq2len
147e83
+		   is still zero.  */
147e83
+		break;
147e83
+	    }
147e83
+	  else
147e83
+	    idxnow = --backw;
147e83
+	}
147e83
+      else
147e83
+	{
147e83
+	  backw_stop = idxmax;
147e83
+
147e83
+	  while (*us != L('\0'))
147e83
+	    {
147e83
+	      int32_t tmp = findidx (&us, -1);
147e83
+	      rulearr[idxmax] = tmp >> 24;
147e83
+	      idxarr[idxmax] = tmp & 0xffffff;
147e83
+	      idxcnt = idxmax++;
147e83
+
147e83
+	      if ((rulesets[rulearr[idxcnt] * nrules]
147e83
+		   & sort_backward) == 0)
147e83
+		/* No more backward characters to push.  */
147e83
+		break;
147e83
+	      ++idxcnt;
147e83
+	    }
147e83
+
147e83
+	  if (backw_stop >= idxcnt)
147e83
+	    {
147e83
+	      /* No sequence at all or just one.  */
147e83
+	      if (idxcnt == idxmax || backw_stop > idxcnt)
147e83
+		/* Note that seq1len is still zero.  */
147e83
+		break;
147e83
+
147e83
+	      backw_stop = ~0ul;
147e83
+	      idxnow = idxcnt;
147e83
+	    }
147e83
+	  else
147e83
+	    /* We pushed backward sequences.  */
147e83
+	    idxnow = backw = idxcnt - 1;
147e83
+	}
147e83
+      len = weights[idxarr[idxnow]++];
147e83
+    }
147e83
+
147e83
+  /* Update the structure.  */
147e83
+  seq->val = val;
147e83
+  seq->len = len;
147e83
+  seq->backw_stop = backw_stop;
147e83
+  seq->backw = backw;
147e83
+  seq->idxcnt = idxcnt;
147e83
+  seq->idxmax = idxmax;
147e83
+  seq->idxnow = idxnow;
147e83
+  seq->us = us;
147e83
+}
147e83
+
147e83
+/* Compare two sequences.  */
147e83
+static int
147e83
+do_compare (coll_seq *seq1, coll_seq *seq2, int position,
147e83
+	    const USTRING_TYPE *weights)
147e83
+{
147e83
+  int seq1len = seq1->len;
147e83
+  int seq2len = seq2->len;
147e83
+  int val1 = seq1->val;
147e83
+  int val2 = seq2->val;
147e83
+  int32_t *idx1arr = seq1->idxarr;
147e83
+  int32_t *idx2arr = seq2->idxarr;
147e83
+  int idx1now = seq1->idxnow;
147e83
+  int idx2now = seq2->idxnow;
147e83
+  int result = 0;
147e83
+
147e83
+  /* Test for position if necessary.  */
147e83
+  if (position && val1 != val2)
147e83
+    {
147e83
+      result = val1 - val2;
147e83
+      goto out;
147e83
+    }
147e83
+
147e83
+  /* Compare the two sequences.  */
147e83
+  do
147e83
+    {
147e83
+      if (weights[idx1arr[idx1now]] != weights[idx2arr[idx2now]])
147e83
+	{
147e83
+	  /* The sequences differ.  */
147e83
+	  result = weights[idx1arr[idx1now]] - weights[idx2arr[idx2now]];
147e83
+	  goto out;
147e83
+	}
147e83
+
147e83
+      /* Increment the offsets.  */
147e83
+      ++idx1arr[idx1now];
147e83
+      ++idx2arr[idx2now];
147e83
+
147e83
+      --seq1len;
147e83
+      --seq2len;
147e83
+    }
147e83
+  while (seq1len > 0 && seq2len > 0);
147e83
+
147e83
+  if (position && seq1len != seq2len)
147e83
+    result = seq1len - seq2len;
147e83
+
147e83
+out:
147e83
+  seq1->len = seq1len;
147e83
+  seq2->len = seq2len;
147e83
+  return result;
147e83
+}
147e83
+
147e83
 int
147e83
-STRCOLL (s1, s2, l)
147e83
-     const STRING_TYPE *s1;
147e83
-     const STRING_TYPE *s2;
147e83
-     __locale_t l;
147e83
+STRCOLL (const STRING_TYPE *s1, const STRING_TYPE *s2, __locale_t l)
147e83
 {
147e83
   struct __locale_data *current = l->__locales[LC_COLLATE];
147e83
   uint_fast32_t nrules = current->values[_NL_ITEM_INDEX (_NL_COLLATE_NRULES)].word;
147e83
@@ -56,34 +289,6 @@ STRCOLL (s1, s2, l)
147e83
   const USTRING_TYPE *weights;
147e83
   const USTRING_TYPE *extra;
147e83
   const int32_t *indirect;
147e83
-  uint_fast32_t pass;
147e83
-  int result = 0;
147e83
-  const USTRING_TYPE *us1;
147e83
-  const USTRING_TYPE *us2;
147e83
-  size_t s1len;
147e83
-  size_t s2len;
147e83
-  int32_t *idx1arr;
147e83
-  int32_t *idx2arr;
147e83
-  unsigned char *rule1arr;
147e83
-  unsigned char *rule2arr;
147e83
-  size_t idx1max;
147e83
-  size_t idx2max;
147e83
-  size_t idx1cnt;
147e83
-  size_t idx2cnt;
147e83
-  size_t idx1now;
147e83
-  size_t idx2now;
147e83
-  size_t backw1_stop;
147e83
-  size_t backw2_stop;
147e83
-  size_t backw1;
147e83
-  size_t backw2;
147e83
-  int val1;
147e83
-  int val2;
147e83
-  int position;
147e83
-  int seq1len;
147e83
-  int seq2len;
147e83
-  int use_malloc;
147e83
-
147e83
-#include WEIGHT_H
147e83
 
147e83
   if (nrules == 0)
147e83
     return STRCMP (s1, s2);
147e83
@@ -98,7 +303,6 @@ STRCOLL (s1, s2, l)
147e83
     current->values[_NL_ITEM_INDEX (CONCAT(_NL_COLLATE_EXTRA,SUFFIX))].string;
147e83
   indirect = (const int32_t *)
147e83
     current->values[_NL_ITEM_INDEX (CONCAT(_NL_COLLATE_INDIRECT,SUFFIX))].string;
147e83
-  use_malloc = 0;
147e83
 
147e83
   assert (((uintptr_t) table) % __alignof__ (table[0]) == 0);
147e83
   assert (((uintptr_t) weights) % __alignof__ (weights[0]) == 0);
147e83
@@ -106,18 +310,13 @@ STRCOLL (s1, s2, l)
147e83
   assert (((uintptr_t) indirect) % __alignof__ (indirect[0]) == 0);
147e83
 
147e83
   /* We need this a few times.  */
147e83
-  s1len = STRLEN (s1);
147e83
-  s2len = STRLEN (s2);
147e83
+  size_t s1len = STRLEN (s1);
147e83
+  size_t s2len = STRLEN (s2);
147e83
 
147e83
   /* Catch empty strings.  */
147e83
-  if (__builtin_expect (s1len == 0, 0) || __builtin_expect (s2len == 0, 0))
147e83
+  if (__glibc_unlikely (s1len == 0) || __glibc_unlikely (s2len == 0))
147e83
     return (s1len != 0) - (s2len != 0);
147e83
 
147e83
-  /* We need the elements of the strings as unsigned values since they
147e83
-     are used as indeces.  */
147e83
-  us1 = (const USTRING_TYPE *) s1;
147e83
-  us2 = (const USTRING_TYPE *) s2;
147e83
-
147e83
   /* Perform the first pass over the string and while doing this find
147e83
      and store the weights for each character.  Since we want this to
147e83
      be as fast as possible we are using `alloca' to store the temporary
147e83
@@ -127,14 +326,27 @@ STRCOLL (s1, s2, l)
147e83
 
147e83
      Please note that the localedef programs makes sure that `position'
147e83
      is not used at the first level.  */
147e83
+
147e83
+  coll_seq seq1, seq2;
147e83
+  bool use_malloc = false;
147e83
+  int result = 0;
147e83
+
147e83
+  memset (&seq1, 0, sizeof (seq1));
147e83
+  seq2 = seq1;
147e83
+
147e83
+  /* We need the elements of the strings as unsigned values since they
147e83
+     are used as indeces.  */
147e83
+  seq1.us = (const USTRING_TYPE *) s1;
147e83
+  seq2.us = (const USTRING_TYPE *) s2;
147e83
+
147e83
   if (! __libc_use_alloca ((s1len + s2len) * (sizeof (int32_t) + 1)))
147e83
     {
147e83
-      idx1arr = (int32_t *) malloc ((s1len + s2len) * (sizeof (int32_t) + 1));
147e83
-      idx2arr = &idx1arr[s1len];
147e83
-      rule1arr = (unsigned char *) &idx2arr[s2len];
147e83
-      rule2arr = &rule1arr[s1len];
147e83
+      seq1.idxarr = (int32_t *) malloc ((s1len + s2len) * (sizeof (int32_t) + 1));
147e83
+      seq2.idxarr = &seq1.idxarr[s1len];
147e83
+      seq1.rulearr = (unsigned char *) &seq2.idxarr[s2len];
147e83
+      seq2.rulearr = &seq1.rulearr[s1len];
147e83
 
147e83
-      if (idx1arr == NULL)
147e83
+      if (seq1.idxarr == NULL)
147e83
 	/* No memory.  Well, go with the stack then.
147e83
 
147e83
 	   XXX Once this implementation is stable we will handle this
147e83
@@ -142,396 +354,73 @@ STRCOLL (s1, s2, l)
147e83
 	   do this in time.  This means, though, that this happens for
147e83
 	   every pass again.  */
147e83
 	goto try_stack;
147e83
-      use_malloc = 1;
147e83
+      use_malloc = true;
147e83
     }
147e83
   else
147e83
     {
147e83
     try_stack:
147e83
-      idx1arr = (int32_t *) alloca (s1len * sizeof (int32_t));
147e83
-      idx2arr = (int32_t *) alloca (s2len * sizeof (int32_t));
147e83
-      rule1arr = (unsigned char *) alloca (s1len);
147e83
-      rule2arr = (unsigned char *) alloca (s2len);
147e83
+      seq1.idxarr = (int32_t *) alloca (s1len * sizeof (int32_t));
147e83
+      seq2.idxarr = (int32_t *) alloca (s2len * sizeof (int32_t));
147e83
+      seq1.rulearr = (unsigned char *) alloca (s1len);
147e83
+      seq2.rulearr = (unsigned char *) alloca (s2len);
147e83
     }
147e83
 
147e83
-  idx1cnt = 0;
147e83
-  idx2cnt = 0;
147e83
-  idx1max = 0;
147e83
-  idx2max = 0;
147e83
-  idx1now = 0;
147e83
-  idx2now = 0;
147e83
-  backw1_stop = ~0ul;
147e83
-  backw2_stop = ~0ul;
147e83
-  backw1 = ~0ul;
147e83
-  backw2 = ~0ul;
147e83
-  seq1len = 0;
147e83
-  seq2len = 0;
147e83
-  position = rulesets[0] & sort_position;
147e83
-  while (1)
147e83
-    {
147e83
-      val1 = 0;
147e83
-      val2 = 0;
147e83
-
147e83
-      /* Get the next non-IGNOREd element for string `s1'.  */
147e83
-      if (seq1len == 0)
147e83
-	do
147e83
-	  {
147e83
-	    ++val1;
147e83
-
147e83
-	    if (backw1_stop != ~0ul)
147e83
-	      {
147e83
-		/* The is something pushed.  */
147e83
-		if (backw1 == backw1_stop)
147e83
-		  {
147e83
-		    /* The last pushed character was handled.  Continue
147e83
-		       with forward characters.  */
147e83
-		    if (idx1cnt < idx1max)
147e83
-		      {
147e83
-			idx1now = idx1cnt;
147e83
-			backw1_stop = ~0ul;
147e83
-		      }
147e83
-		    else
147e83
-		      /* Nothing anymore.  The backward sequence ended with
147e83
-			 the last sequence in the string.  Note that seq1len
147e83
-			 is still zero.  */
147e83
-		      break;
147e83
-		  }
147e83
-		else
147e83
-		  idx1now = --backw1;
147e83
-	      }
147e83
-	    else
147e83
-	      {
147e83
-		backw1_stop = idx1max;
147e83
-
147e83
-		while (*us1 != L('\0'))
147e83
-		  {
147e83
-		    int32_t tmp = findidx (&us1, -1);
147e83
-		    rule1arr[idx1max] = tmp >> 24;
147e83
-		    idx1arr[idx1max] = tmp & 0xffffff;
147e83
-		    idx1cnt = idx1max++;
147e83
-
147e83
-		    if ((rulesets[rule1arr[idx1cnt] * nrules]
147e83
-			 & sort_backward) == 0)
147e83
-		      /* No more backward characters to push.  */
147e83
-		      break;
147e83
-		    ++idx1cnt;
147e83
-		  }
147e83
-
147e83
-		if (backw1_stop >= idx1cnt)
147e83
-		  {
147e83
-		    /* No sequence at all or just one.  */
147e83
-		    if (idx1cnt == idx1max || backw1_stop > idx1cnt)
147e83
-		      /* Note that seq1len is still zero.  */
147e83
-		      break;
147e83
-
147e83
-		    backw1_stop = ~0ul;
147e83
-		    idx1now = idx1cnt;
147e83
-		  }
147e83
-		else
147e83
-		  /* We pushed backward sequences.  */
147e83
-		  idx1now = backw1 = idx1cnt - 1;
147e83
-	      }
147e83
-	  }
147e83
-	while ((seq1len = weights[idx1arr[idx1now]++]) == 0);
147e83
-
147e83
-      /* And the same for string `s2'.  */
147e83
-      if (seq2len == 0)
147e83
-	do
147e83
-	  {
147e83
-	    ++val2;
147e83
-
147e83
-	    if (backw2_stop != ~0ul)
147e83
-	      {
147e83
-		/* The is something pushed.  */
147e83
-		if (backw2 == backw2_stop)
147e83
-		  {
147e83
-		    /* The last pushed character was handled.  Continue
147e83
-		       with forward characters.  */
147e83
-		    if (idx2cnt < idx2max)
147e83
-		      {
147e83
-			idx2now = idx2cnt;
147e83
-			backw2_stop = ~0ul;
147e83
-		      }
147e83
-		    else
147e83
-		      /* Nothing anymore.  The backward sequence ended with
147e83
-			 the last sequence in the string.  Note that seq2len
147e83
-			 is still zero.  */
147e83
-		      break;
147e83
-		  }
147e83
-		else
147e83
-		  idx2now = --backw2;
147e83
-	      }
147e83
-	    else
147e83
-	      {
147e83
-		backw2_stop = idx2max;
147e83
-
147e83
-		while (*us2 != L('\0'))
147e83
-		  {
147e83
-		    int32_t tmp = findidx (&us2, -1);
147e83
-		    rule2arr[idx2max] = tmp >> 24;
147e83
-		    idx2arr[idx2max] = tmp & 0xffffff;
147e83
-		    idx2cnt = idx2max++;
147e83
-
147e83
-		    if ((rulesets[rule2arr[idx2cnt] * nrules]
147e83
-			 & sort_backward) == 0)
147e83
-		      /* No more backward characters to push.  */
147e83
-		      break;
147e83
-		    ++idx2cnt;
147e83
-		  }
147e83
-
147e83
-		if (backw2_stop >= idx2cnt)
147e83
-		  {
147e83
-		    /* No sequence at all or just one.  */
147e83
-		    if (idx2cnt == idx2max || backw2_stop > idx2cnt)
147e83
-		      /* Note that seq1len is still zero.  */
147e83
-		      break;
147e83
-
147e83
-		    backw2_stop = ~0ul;
147e83
-		    idx2now = idx2cnt;
147e83
-		  }
147e83
-		else
147e83
-		  /* We pushed backward sequences.  */
147e83
-		  idx2now = backw2 = idx2cnt - 1;
147e83
-	      }
147e83
-	  }
147e83
-	while ((seq2len = weights[idx2arr[idx2now]++]) == 0);
147e83
-
147e83
-      /* See whether any or both strings are empty.  */
147e83
-      if (seq1len == 0 || seq2len == 0)
147e83
-	{
147e83
-	  if (seq1len == seq2len)
147e83
-	    /* Both ended.  So far so good, both strings are equal at the
147e83
-	       first level.  */
147e83
-	    break;
147e83
-
147e83
-	  /* This means one string is shorter than the other.  Find out
147e83
-	     which one and return an appropriate value.  */
147e83
-	  result = seq1len == 0 ? -1 : 1;
147e83
-	  goto free_and_return;
147e83
-	}
147e83
+  seq1.rulearr[0] = 0;
147e83
 
147e83
-      /* Test for position if necessary.  */
147e83
-      if (position && val1 != val2)
147e83
-	{
147e83
-	  result = val1 - val2;
147e83
-	  goto free_and_return;
147e83
-	}
147e83
-
147e83
-      /* Compare the two sequences.  */
147e83
-      do
147e83
-	{
147e83
-	  if (weights[idx1arr[idx1now]] != weights[idx2arr[idx2now]])
147e83
-	    {
147e83
-	      /* The sequences differ.  */
147e83
-	      result = weights[idx1arr[idx1now]] - weights[idx2arr[idx2now]];
147e83
-	      goto free_and_return;
147e83
-	    }
147e83
-
147e83
-	  /* Increment the offsets.  */
147e83
-	  ++idx1arr[idx1now];
147e83
-	  ++idx2arr[idx2now];
147e83
-
147e83
-	  --seq1len;
147e83
-	  --seq2len;
147e83
-	}
147e83
-      while (seq1len > 0 && seq2len > 0);
147e83
-
147e83
-      if (position && seq1len != seq2len)
147e83
-	{
147e83
-	  result = seq1len - seq2len;
147e83
-	  goto free_and_return;
147e83
-	}
147e83
-    }
147e83
-
147e83
-  /* Now the remaining passes over the weights.  We now use the
147e83
-     indeces we found before.  */
147e83
-  for (pass = 1; pass < nrules; ++pass)
147e83
+  /* Cache values in the first pass and if needed, use them in subsequent
147e83
+     passes.  */
147e83
+  for (int pass = 0; pass < nrules; ++pass)
147e83
     {
147e83
+      seq1.idxcnt = 0;
147e83
+      seq1.backw_stop = ~0ul;
147e83
+      seq1.backw = ~0ul;
147e83
+      seq2.idxcnt = 0;
147e83
+      seq2.backw_stop = ~0ul;
147e83
+      seq2.backw = ~0ul;
147e83
+
147e83
       /* We assume that if a rule has defined `position' in one section
147e83
 	 this is true for all of them.  */
147e83
-      idx1cnt = 0;
147e83
-      idx2cnt = 0;
147e83
-      backw1_stop = ~0ul;
147e83
-      backw2_stop = ~0ul;
147e83
-      backw1 = ~0ul;
147e83
-      backw2 = ~0ul;
147e83
-      position = rulesets[rule1arr[0] * nrules + pass] & sort_position;
147e83
+      int position = rulesets[seq1.rulearr[0] * nrules + pass] & sort_position;
147e83
 
147e83
       while (1)
147e83
 	{
147e83
-	  val1 = 0;
147e83
-	  val2 = 0;
147e83
-
147e83
-	  /* Get the next non-IGNOREd element for string `s1'.  */
147e83
-	  if (seq1len == 0)
147e83
-	    do
147e83
-	      {
147e83
-		++val1;
147e83
-
147e83
-		if (backw1_stop != ~0ul)
147e83
-		  {
147e83
-		    /* The is something pushed.  */
147e83
-		    if (backw1 == backw1_stop)
147e83
-		      {
147e83
-			/* The last pushed character was handled.  Continue
147e83
-			   with forward characters.  */
147e83
-			if (idx1cnt < idx1max)
147e83
-			  {
147e83
-			    idx1now = idx1cnt;
147e83
-			    backw1_stop = ~0ul;
147e83
-			  }
147e83
-			else
147e83
-			  {
147e83
-			    /* Nothing anymore.  The backward sequence
147e83
-			       ended with the last sequence in the string.  */
147e83
-			    idx1now = ~0ul;
147e83
-			    break;
147e83
-			  }
147e83
-		      }
147e83
-		    else
147e83
-		      idx1now = --backw1;
147e83
-		  }
147e83
-		else
147e83
-		  {
147e83
-		    backw1_stop = idx1cnt;
147e83
-
147e83
-		    while (idx1cnt < idx1max)
147e83
-		      {
147e83
-			if ((rulesets[rule1arr[idx1cnt] * nrules + pass]
147e83
-			     & sort_backward) == 0)
147e83
-			  /* No more backward characters to push.  */
147e83
-			  break;
147e83
-			++idx1cnt;
147e83
-		      }
147e83
-
147e83
-		    if (backw1_stop == idx1cnt)
147e83
-		      {
147e83
-			/* No sequence at all or just one.  */
147e83
-			if (idx1cnt == idx1max)
147e83
-			  /* Note that seq1len is still zero.  */
147e83
-			  break;
147e83
-
147e83
-			backw1_stop = ~0ul;
147e83
-			idx1now = idx1cnt++;
147e83
-		      }
147e83
-		    else
147e83
-		      /* We pushed backward sequences.  */
147e83
-		      idx1now = backw1 = idx1cnt - 1;
147e83
-		  }
147e83
-	      }
147e83
-	    while ((seq1len = weights[idx1arr[idx1now]++]) == 0);
147e83
-
147e83
-	  /* And the same for string `s2'.  */
147e83
-	  if (seq2len == 0)
147e83
-	    do
147e83
-	      {
147e83
-		++val2;
147e83
-
147e83
-		if (backw2_stop != ~0ul)
147e83
-		  {
147e83
-		    /* The is something pushed.  */
147e83
-		    if (backw2 == backw2_stop)
147e83
-		      {
147e83
-			/* The last pushed character was handled.  Continue
147e83
-			   with forward characters.  */
147e83
-			if (idx2cnt < idx2max)
147e83
-			  {
147e83
-			    idx2now = idx2cnt;
147e83
-			    backw2_stop = ~0ul;
147e83
-			  }
147e83
-			else
147e83
-			  {
147e83
-			    /* Nothing anymore.  The backward sequence
147e83
-			       ended with the last sequence in the string.  */
147e83
-			    idx2now = ~0ul;
147e83
-			    break;
147e83
-			  }
147e83
-		      }
147e83
-		    else
147e83
-		      idx2now = --backw2;
147e83
-		  }
147e83
-		else
147e83
-		  {
147e83
-		    backw2_stop = idx2cnt;
147e83
-
147e83
-		    while (idx2cnt < idx2max)
147e83
-		      {
147e83
-			if ((rulesets[rule2arr[idx2cnt] * nrules + pass]
147e83
-			     & sort_backward) == 0)
147e83
-			  /* No more backward characters to push.  */
147e83
-			  break;
147e83
-			++idx2cnt;
147e83
-		      }
147e83
-
147e83
-		    if (backw2_stop == idx2cnt)
147e83
-		      {
147e83
-			/* No sequence at all or just one.  */
147e83
-			if (idx2cnt == idx2max)
147e83
-			  /* Note that seq2len is still zero.  */
147e83
-			  break;
147e83
-
147e83
-			backw2_stop = ~0ul;
147e83
-			idx2now = idx2cnt++;
147e83
-		      }
147e83
-		    else
147e83
-		      /* We pushed backward sequences.  */
147e83
-		      idx2now = backw2 = idx2cnt - 1;
147e83
-		  }
147e83
-	      }
147e83
-	    while ((seq2len = weights[idx2arr[idx2now]++]) == 0);
147e83
+	  if (pass == 0)
147e83
+	    {
147e83
+	      get_next_seq (&seq1, nrules, rulesets, weights, table, extra,
147e83
+			    indirect);
147e83
+	      get_next_seq (&seq2, nrules, rulesets, weights, table, extra,
147e83
+			    indirect);
147e83
+	    }
147e83
+	  else
147e83
+	    {
147e83
+	      get_next_seq_cached (&seq1, nrules, pass, rulesets, weights);
147e83
+	      get_next_seq_cached (&seq2, nrules, pass, rulesets, weights);
147e83
+	    }
147e83
 
147e83
 	  /* See whether any or both strings are empty.  */
147e83
-	  if (seq1len == 0 || seq2len == 0)
147e83
+	  if (seq1.len == 0 || seq2.len == 0)
147e83
 	    {
147e83
-	      if (seq1len == seq2len)
147e83
+	      if (seq1.len == seq2.len)
147e83
 		/* Both ended.  So far so good, both strings are equal
147e83
 		   at this level.  */
147e83
 		break;
147e83
 
147e83
 	      /* This means one string is shorter than the other.  Find out
147e83
 		 which one and return an appropriate value.  */
147e83
-	      result = seq1len == 0 ? -1 : 1;
147e83
+	      result = seq1.len == 0 ? -1 : 1;
147e83
 	      goto free_and_return;
147e83
 	    }
147e83
 
147e83
-	  /* Test for position if necessary.  */
147e83
-	  if (position && val1 != val2)
147e83
-	    {
147e83
-	      result = val1 - val2;
147e83
-	      goto free_and_return;
147e83
-	    }
147e83
-
147e83
-	  /* Compare the two sequences.  */
147e83
-	  do
147e83
-	    {
147e83
-	      if (weights[idx1arr[idx1now]] != weights[idx2arr[idx2now]])
147e83
-		{
147e83
-		  /* The sequences differ.  */
147e83
-		  result = (weights[idx1arr[idx1now]]
147e83
-			    - weights[idx2arr[idx2now]]);
147e83
-		  goto free_and_return;
147e83
-		}
147e83
-
147e83
-	      /* Increment the offsets.  */
147e83
-	      ++idx1arr[idx1now];
147e83
-	      ++idx2arr[idx2now];
147e83
-
147e83
-	      --seq1len;
147e83
-	      --seq2len;
147e83
-	    }
147e83
-	  while (seq1len > 0 && seq2len > 0);
147e83
-
147e83
-	  if (position && seq1len != seq2len)
147e83
-	    {
147e83
-	      result = seq1len - seq2len;
147e83
-	      goto free_and_return;
147e83
-	    }
147e83
+	  result = do_compare (&seq1, &seq2, position, weights);
147e83
+	  if (result != 0)
147e83
+	    goto free_and_return;
147e83
 	}
147e83
     }
147e83
 
147e83
   /* Free the memory if needed.  */
147e83
  free_and_return:
147e83
   if (use_malloc)
147e83
-    free (idx1arr);
147e83
+    free (seq1.idxarr);
147e83
 
147e83
   return result;
147e83
 }