More optimizations of the melm matching.

diff --git a/data1/d1_absyn.c b/data1/d1_absyn.c
index 3edbb6a..a45d51c 100644 (file)
--- a/data1/d1_absyn.c
+++ b/data1/d1_absyn.c
@@ -1,4 +1,4 @@
-/* $Id: d1_absyn.c,v 1.9.2.9 2006-09-28 18:38:41 adam Exp $
+/* $Id: d1_absyn.c,v 1.9.2.10 2006-09-29 10:02:42 adam Exp $
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
    Index Data Aps
 
@@ -543,7 +543,7 @@ data1_absyn *data1_read_absyn (data1_handle dh, const char *file,
                                int file_must_exist)
 {
     data1_sub_elements *cur_elements = NULL;
-    data1_xpelement *cur_xpelement = NULL;
+    data1_xpelement **cur_xpelement = NULL;
 
     data1_absyn *res = 0;
     FILE *f;
@@ -590,6 +590,7 @@ data1_absyn *data1_read_absyn (data1_handle dh, const char *file,
     res->sub_elements = NULL;
     res->main_elements = NULL;
     res->xp_elements = NULL;
+    cur_xpelement = &res->xp_elements;
     
     while (f && (argc = read_absyn_line(f, &lineno, line, 512, argv, 50)))
     {
@@ -737,7 +738,9 @@ data1_absyn *data1_read_absyn (data1_handle dh, const char *file,
            struct DFA *dfa = 0;
            data1_termlist **tp;
            char melm_xpath[128];
-           data1_xpelement *xp_old = 0;
+            data1_xpelement *xp_ele = 0;
+            data1_xpelement *last_match = 0;
+
             
            if (argc < 3)
            {
@@ -755,11 +758,12 @@ data1_absyn *data1_read_absyn (data1_handle dh, const char *file,
            termlists = argv[2];
             regexp = mk_xpath_regexp(dh, xpath_expr);
 #if OPTIMIZE_MELM
-            for (xp_old = res->xp_elements; xp_old; xp_old = xp_old->next)
-                if (!strcmp(xp_old->regexp, regexp))
-                    break;
+            /* get last of existing regulars with same regexp */
+            for (xp_ele = res->xp_elements; xp_ele; xp_ele = xp_ele->next)
+                if (!strcmp(xp_ele->regexp, regexp))
+                    last_match = xp_ele;
 #endif
-            if (!xp_old)
+            if (!last_match)
             {
                const char *regexp_ptr = regexp;
                 dfa = dfa_init();
@@ -771,37 +775,30 @@ data1_absyn *data1_read_absyn (data1_handle dh, const char *file,
                     continue;
                 }
            }           
-           if (!cur_xpelement)
-           {
-                cur_xpelement = (data1_xpelement *)
-                   nmem_malloc(data1_nmem_get(dh), sizeof(*cur_xpelement));
-               res->xp_elements = cur_xpelement;
-            } else {
-                cur_xpelement->next = (data1_xpelement *)
-                    nmem_malloc(data1_nmem_get(dh), sizeof(*cur_xpelement));
-                cur_xpelement = cur_xpelement->next;
-           }
+            *cur_xpelement = (data1_xpelement *)
+                nmem_malloc(data1_nmem_get(dh), sizeof(**cur_xpelement));
+            (*cur_xpelement)->next = 0;
+            (*cur_xpelement)->match_next = 0;
+            if (last_match)
+                last_match->match_next = *cur_xpelement;
 #if OPTIMIZE_MELM
-            cur_xpelement->regexp = regexp;
+            (*cur_xpelement)->regexp = regexp;
 #endif
-           cur_xpelement->next = NULL;
-           cur_xpelement->xpath_expr = nmem_strdup(data1_nmem_get (dh), 
+           (*cur_xpelement)->next = NULL;
+           (*cur_xpelement)->xpath_expr = nmem_strdup(data1_nmem_get (dh), 
                                                    xpath_expr); 
            if (dfa)
                dfa_mkstate (dfa);
-           cur_xpelement->dfa = dfa;
+           (*cur_xpelement)->dfa = dfa;
 #ifdef ENHANCED_XELM 
-            cur_xpelement->xpath_len =
+            (*cur_xpelement)->xpath_len =
                 zebra_parse_xpath_str(xpath_expr, 
-                                      cur_xpelement->xpath, XPATH_STEP_COUNT,
+                                      (*cur_xpelement)->xpath, XPATH_STEP_COUNT,
                                       data1_nmem_get(dh));
             
-           /*
-           dump_xp_steps(cur_xpelement->xpath,cur_xpelement->xpath_len);
-           */
 #endif
-           cur_xpelement->termlists = 0;
-           tp = &cur_xpelement->termlists;
+           (*cur_xpelement)->termlists = 0;
+           tp = &(*cur_xpelement)->termlists;
             
            /* parse termList definitions */
            p = termlists;
@@ -817,6 +814,7 @@ data1_absyn *data1_read_absyn (data1_handle dh, const char *file,
                }
                *tp = all; /* append any ALL entries to the list */
            }
+            cur_xpelement = &(*cur_xpelement)->next;
        }
        else if (!strcmp(cmd, "section"))
        {

diff --git a/include/data1.h b/include/data1.h
index ff1cc1c..cce6457 100644 (file)
--- a/include/data1.h
+++ b/include/data1.h
@@ -1,4 +1,4 @@
-/* $Id: data1.h,v 1.9.2.3 2006-09-28 18:38:42 adam Exp $
+/* $Id: data1.h,v 1.9.2.4 2006-09-29 10:02:43 adam Exp $
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
    Index Data Aps
 
@@ -208,6 +208,7 @@ typedef struct data1_xpelement
     const char *regexp;
 #endif
     int match_state;
+    struct data1_xpelement *match_next;
 } data1_xpelement;
 
 typedef struct data1_xattr {

diff --git a/recctrl/recgrs.c b/recctrl/recgrs.c
index b36b206..7b624cf 100644 (file)
--- a/recctrl/recgrs.c
+++ b/recctrl/recgrs.c
@@ -1,4 +1,4 @@
-/* $Id: recgrs.c,v 1.86.2.11 2006-09-28 18:38:42 adam Exp $
+/* $Id: recgrs.c,v 1.86.2.12 2006-09-29 10:02:44 adam Exp $
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003
    Index Data Aps
 
@@ -466,11 +466,8 @@ data1_termlist *xpath_termlist_by_tagpath(char *tagpath, data1_node *n)
             ok = dfa_match_first(xpe->dfa->states, pexpr);
 #if OPTIMIZE_MELM
             /* mark this and following ones with same regexp */
-            for (xpe1 = xpe; xpe1; xpe1 = xpe1->next)
-            {
-                if (!strcmp(xpe1->regexp, xpe->regexp))
-                    xpe1->match_state = ok;
-            }
+            for (xpe1 = xpe; xpe1; xpe1 = xpe1->match_next)
+                xpe1->match_state = ok;
 #endif
        }
        assert (ok == 0 || ok == 1);