[5882] don't use regexp search to parse a node

[Gavin D. Smith]

Revision: 5882
          http://svn.sv.gnu.org/viewvc/?view=rev&root=texinfo&revision=5882
Author:   gavin
Date:     2014-10-22 20:21:51 +0000 (Wed, 22 Oct 2014)
Log Message:
-----------
don't use regexp search to parse a node

Modified Paths:
--------------
    trunk/ChangeLog
    trunk/info/info-utils.c

Modified: trunk/ChangeLog
===================================================================
--- trunk/ChangeLog     2014-10-21 17:07:12 UTC (rev 5881)
+++ trunk/ChangeLog     2014-10-22 20:21:51 UTC (rev 5882)
@@ -1,3 +1,10 @@
+2014-10-22  Gavin Smith  <address@hidden>
+
+       * info/info-utils.c (forward_to_info_syntax): New function.
+       (scan_node_contents): Call it instead of using regexp_search.  
+       This produces a speed improvement, noticable when using M-x 
+       index-apropos or opening a long index node.
+
 2014-10-21  Karl Berry  <address@hidden>
 
        * tp/texi2any.pl (makeinfo_help),

Modified: trunk/info/info-utils.c
===================================================================
--- trunk/info/info-utils.c     2014-10-21 17:07:12 UTC (rev 5881)
+++ trunk/info/info-utils.c     2014-10-22 20:21:51 UTC (rev 5882)
@@ -1580,6 +1580,26 @@
   free (expansion);
 }
 
+#define looking_at_string(contents, string) \
+  (!memcmp (contents, string, strlen (string)))
+
+static char *
+forward_to_info_syntax (char *contents)
+{
+  while (contents < input_start + input_length)
+    {
+      /* Menu entry comes first to optimize for the case of looking through a 
+         long index node. */
+      if (looking_at_string (contents, INFO_MENU_ENTRY_LABEL)
+          || looking_at_string (contents, INFO_MENU_LABEL)
+          || looking_at_string (contents, INFO_XREF_LABEL)
+          || !memcmp (contents, " \b[", 3))
+        return contents;
+      contents++;
+    }
+  return 0;
+}
+
 /* Scan (*NODE_PTR)->contents and record location and contents of
    cross-references and menu items.  Convert character encoding of
    node contents to that of the user if the two are known to be
@@ -1590,12 +1610,8 @@
 void
 scan_node_contents (FILE_BUFFER *fb, NODE **node_ptr)
 {
-  char *search_string;
-  long position;
-  regmatch_t *matches;
-  size_t match_count;
-  size_t i;
   int in_menu = 0;
+  char *match;
 
   NODE *node = *node_ptr;
 
@@ -1654,24 +1670,12 @@
 
   parse_top_node_line (node);
 
-  search_string = INFO_MENU_REGEXP "|" INFO_MENU_ENTRY_REGEXP
-    "|" INFO_XREF_REGEXP "|" INFO_TAG_REGEXP;
-
-  if (regexp_search (search_string, 0, 1, node->contents, node->nodelen,
-                     &matches, &match_count)
-      == search_success)
-  for (i = 0; i < match_count; i++)
+  while ((match = forward_to_info_syntax (inptr))
+          && match < node->contents + node->nodelen)
     {
       int in_parentheses = 0;
       REFERENCE *entry;
-      char *match;
-      position = matches[i].rm_so;
 
-      match = node->contents + position;
-
-      if (match < inptr)
-        continue;
-
       /* Write out up to match */
       copy_input_to_output (match - inptr); 
 
@@ -1715,6 +1719,8 @@
 
           add_pointer_to_array (entry, refs_index, refs, refs_slots, 50);
         }
+      else
+        copy_input_to_output (1);
     }
 
   /* If we haven't accidentally gone past the end of the node, write
@@ -1729,8 +1735,6 @@
   /* Free resources used in character encoding conversion. */
   close_conversion ();
   
-  free (matches);
-
   node->references = refs;
 
   if (rewrite_p)