X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=recctrl%2Fregxread.c;h=7c9919ff0367fe5348fc26ba4bd2995a78560bb1;hb=b5eb8cd7ae4134d915ed8b4aec25ab90c8bd0dcb;hp=f56738d21d8bbe7f2fb6c4adea3fe9d9387cec84;hpb=bd7b3a96753d67f123e61c3ec7ce6a9b5ac2ea73;p=idzebra-moved-to-github.git

diff --git a/recctrl/regxread.c b/recctrl/regxread.c
index f56738d..7c9919f 100644
--- a/recctrl/regxread.c
+++ b/recctrl/regxread.c
@@ -1,5 +1,5 @@
-/* $Id: regxread.c,v 1.46 2002-09-24 19:41:00 adam Exp $
-   Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
+/* $Id: regxread.c,v 1.51 2004-08-04 08:35:25 adam Exp $
+   Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
    Index Data Aps
 
 This file is part of the Zebra server.
@@ -473,6 +473,8 @@ static int actionListMk (struct lexSpec *spec, const char *s,
                 logf (LOG_WARN, "regular expression error '%.*s'", s-s0, s0);
                 return -1;
             }
+	    if (debug_dfa_tran)
+		printf ("pattern: %.*s\n", s-s0, s0);
             dfa_mkstate ((*ap)->u.pattern.dfa);
             s++;
             break;
@@ -590,6 +592,14 @@ int readFileSpec (struct lexSpec *spec)
     if (spec->tcl_interp)
 	logf (LOG_LOG, "Tcl enabled");
 #endif
+
+#if 0
+    debug_dfa_trav = 0;
+    debug_dfa_tran = 1;
+    debug_dfa_followpos = 0;
+    dfa_verbose = 1;
+#endif
+
     lineBuf = wrbuf_alloc();
     spec->lineNo = 0;
     c = getc (spec_inf);
@@ -632,12 +642,6 @@ int readFileSpec (struct lexSpec *spec)
     fclose (spec_inf);
     wrbuf_free(lineBuf, 1);
 
-#if 0
-    debug_dfa_trav = 1;
-    debug_dfa_tran = 1;
-    debug_dfa_followpos = 1;
-    dfa_verbose = 1;
-#endif
     for (lc = spec->context; lc; lc = lc->next)
     {
 	struct lexRule *rp;
@@ -668,13 +672,17 @@ static void execData (struct lexSpec *spec,
     if (elen == 0) /* shouldn't happen, but it does! */
 	return ;
 #if REGX_DEBUG
-    if (elen > 40)
-        logf (LOG_LOG, "data (%d bytes) %.15s ... %.*s", elen,
-	      ebuf, 15, ebuf + elen-15);
+    if (elen > 80)
+        logf (LOG_LOG, "data(%d bytes) %.40s ... %.*s", elen,
+	      ebuf, 40, ebuf + elen-40);
+    else if (elen == 1 && ebuf[0] == '\n')
+    {
+        logf (LOG_LOG, "data(new line)");
+    }
     else if (elen > 0)
-        logf (LOG_LOG, "data (%d bytes) %.*s", elen, elen, ebuf);
+        logf (LOG_LOG, "data(%d bytes) %.*s", elen, elen, ebuf);
     else 
-        logf (LOG_LOG, "data (%d bytes)", elen);
+        logf (LOG_LOG, "data(%d bytes)", elen);
 #endif
         
     if (spec->d1_level <= 1)
@@ -768,7 +776,7 @@ static void variantBegin (struct lexSpec *spec,
     ttype[type_len] = '\0';
 
 #if REGX_DEBUG 
-    logf (LOG_LOG, "variant begin %s %s (%d)", tclass, ttype,
+    logf (LOG_LOG, "variant begin(%s,%s,%d)", tclass, ttype,
 	  spec->d1_level);
 #endif
 
@@ -793,7 +801,7 @@ static void variantBegin (struct lexSpec *spec,
 	}
 
 #if REGX_DEBUG 
-    logf (LOG_LOG, "variant node (%d)", spec->d1_level);
+    logf (LOG_LOG, "variant node(%d)", spec->d1_level);
 #endif
     parent = spec->d1_stack[spec->d1_level-1];
     res = data1_mk_node2 (spec->dh, spec->m, DATA1N_variant, parent);
@@ -838,7 +846,7 @@ static void tagBegin (struct lexSpec *spec,
 	tagDataRelease (spec);
 
 #if REGX_DEBUG 
-    logf (LOG_LOG, "begin tag %s (%d)", tag, spec->d1_level);
+    logf (LOG_LOG, "begin tag(%.*s, %d)", len, tag, spec->d1_level);
 #endif
 
     spec->d1_stack[spec->d1_level] = data1_mk_tag_n (
@@ -864,27 +872,39 @@ static void tagEnd (struct lexSpec *spec, int min_level,
             break;
     }
 #if REGX_DEBUG
-    logf (LOG_LOG, "end tag (%d)", spec->d1_level);
+    logf (LOG_LOG, "end tag(%d)", spec->d1_level);
 #endif
 }
 
 
 static int tryMatch (struct lexSpec *spec, int *pptr, int *mptr,
-                     struct DFA *dfa)
+                     struct DFA *dfa, int greedy)
 {
     struct DFA_state *state = dfa->states[0];
     struct DFA_tran *t;
-    unsigned char c;
+    unsigned char c = 0;
     unsigned char c_prev = 0;
     int ptr = *pptr;          /* current pointer */
     int start_ptr = *pptr;    /* first char of match */
     int last_ptr = 0;         /* last char of match */
     int last_rule = 0;        /* rule number of current match */
+    int restore_ptr = 0;
     int i;
 
+    if (ptr)
+    {
+	--ptr;
+        c = f_win_advance (spec, &ptr);
+    }
     while (1)
     {
+	if (dfa->states[0] == state)
+	{
+	    c_prev = c;
+	    restore_ptr = ptr;
+	}
         c = f_win_advance (spec, &ptr);
+
         if (ptr == F_WIN_EOF)
         {
             if (last_rule)
@@ -895,10 +915,11 @@ static int tryMatch (struct lexSpec *spec, int *pptr, int *mptr,
             }
             break;
         }
+
         t = state->trans;
         i = state->tran_no;
         while (1)
-            if (--i < 0)
+            if (--i < 0)    /* no transition for character c */
             {
                 if (last_rule)
                 {
@@ -907,27 +928,28 @@ static int tryMatch (struct lexSpec *spec, int *pptr, int *mptr,
                     return 1;
                 }
                 state = dfa->states[0];
+
+		ptr = restore_ptr;
+		c = f_win_advance (spec, &ptr);
+
                 start_ptr = ptr;
-                c_prev = c;
+
                 break;
             }
             else if (c >= t->ch[0] && c <= t->ch[1])
             {
                 state = dfa->states[t->to];
-                if (state->rule_no)
-                {
-                    if (c_prev == '\n')
-                    {
-                        last_rule = state->rule_no;
-                        last_ptr = ptr;
-                    }
-                    else
-                    {
-                        last_rule = state->rule_nno;
-                        last_ptr = ptr;
-                    }
-                }
-                break;
+                if (state->rule_no && c_prev == '\n')
+		{
+		    last_rule = state->rule_no;
+		    last_ptr = ptr;
+		}
+		else if (state->rule_nno)
+		{
+		    last_rule = state->rule_nno;
+		    last_ptr = ptr;
+		}
+		break;
             }
             else
                 t++;
@@ -1011,14 +1033,14 @@ static char *regxStrz (const char *src, int len, char *str)
 
 #if HAVE_TCL_H
 static int cmd_tcl_begin (ClientData clientData, Tcl_Interp *interp,
-			  int argc, char **argv)
+			  int argc, const char **argv)
 {
     struct lexSpec *spec = (struct lexSpec *) clientData;
     if (argc < 2)
 	return TCL_ERROR;
     if (!strcmp(argv[1], "record") && argc == 3)
     {
-	char *absynName = argv[2];
+	const char *absynName = argv[2];
         data1_node *res;
 
 #if REGX_DEBUG
@@ -1026,6 +1048,8 @@ static int cmd_tcl_begin (ClientData clientData, Tcl_Interp *interp,
 #endif
         res = data1_mk_root (spec->dh, spec->m, absynName);
         
+	spec->d1_level = 0;
+
         spec->d1_stack[spec->d1_level++] = res;
 
         res = data1_mk_tag (spec->dh, spec->m, absynName, 0, res);
@@ -1065,7 +1089,7 @@ static int cmd_tcl_begin (ClientData clientData, Tcl_Interp *interp,
 }
 
 static int cmd_tcl_end (ClientData clientData, Tcl_Interp *interp,
-			int argc, char **argv)
+			int argc, const char **argv)
 {
     struct lexSpec *spec = (struct lexSpec *) clientData;
     if (argc < 2)
@@ -1085,8 +1109,8 @@ static int cmd_tcl_end (ClientData clientData, Tcl_Interp *interp,
     }
     else if (!strcmp (argv[1], "element"))
     {
-	int min_level = 1;
-	char *element = 0;
+	int min_level = 2;
+	const char *element = 0;
 	if (argc >= 3 && !strcmp(argv[2], "-record"))
 	{
 	    min_level = 0;
@@ -1097,7 +1121,7 @@ static int cmd_tcl_end (ClientData clientData, Tcl_Interp *interp,
 	    if (argc == 3)
 		element = argv[2];
 	tagEnd (spec, min_level, element, (element ? strlen(element) : 0));
-	if (spec->d1_level == 0)
+	if (spec->d1_level <= 1)
 	{
 #if REGX_DEBUG
 	    logf (LOG_LOG, "end element end records");
@@ -1119,7 +1143,7 @@ static int cmd_tcl_end (ClientData clientData, Tcl_Interp *interp,
 }
 
 static int cmd_tcl_data (ClientData clientData, Tcl_Interp *interp,
-			 int argc, char **argv)
+			 int argc, const char **argv)
 {
     int argi = 1;
     int textFlag = 0;
@@ -1158,12 +1182,12 @@ static int cmd_tcl_data (ClientData clientData, Tcl_Interp *interp,
 	argi++;
     }
     if (element)
-	tagEnd (spec, 1, NULL, 0);
+	tagEnd (spec, 2, NULL, 0);
     return TCL_OK;
 }
 
 static int cmd_tcl_unread (ClientData clientData, Tcl_Interp *interp,
-			   int argc, char **argv)
+			   int argc, const char **argv)
 {
     struct lexSpec *spec = (struct lexSpec *) clientData;
     int argi = 1;
@@ -1261,7 +1285,7 @@ static void execCode (struct lexSpec *spec, struct regxCode *code)
                 r = execTok (spec, &s, &cmd_str, &cmd_len);
                 if (r < 2)
                     continue;
-                if (spec->d1_level == 0)
+                if (spec->d1_level <= 1)
                 {
                     static char absynName[64];
                     data1_node *res;
@@ -1275,6 +1299,8 @@ static void execCode (struct lexSpec *spec, struct regxCode *code)
 #endif
                     res = data1_mk_root (spec->dh, spec->m, absynName);
                     
+		    spec->d1_level = 0;
+
                     spec->d1_stack[spec->d1_level++] = res;
 
                     res = data1_mk_tag (spec->dh, spec->m, absynName, 0, res);
@@ -1373,7 +1399,7 @@ static void execCode (struct lexSpec *spec, struct regxCode *code)
 	    }
 	    else if (!strcmp (p, "element"))
 	    {
-                int min_level = 1;
+                int min_level = 2;
                 while ((r = execTok (spec, &s, &cmd_str, &cmd_len)) == 3)
                 {
                     if (cmd_len==7 && !memcmp ("-record", cmd_str, cmd_len))
@@ -1386,7 +1412,7 @@ static void execCode (struct lexSpec *spec, struct regxCode *code)
 		}
 		else
 		    tagEnd (spec, min_level, NULL, 0);
-                if (spec->d1_level == 0)
+                if (spec->d1_level <= 1)
                 {
 #if REGX_DEBUG
 		    logf (LOG_LOG, "end element end records");
@@ -1440,7 +1466,7 @@ static void execCode (struct lexSpec *spec, struct regxCode *code)
                 r = execTok (spec, &s, &cmd_str, &cmd_len);
             } while (r > 1);
             if (element_str)
-                tagEnd (spec, 1, NULL, 0);
+                tagEnd (spec, 2, NULL, 0);
         }
         else if (!strcmp (p, "unread"))
         {
@@ -1537,13 +1563,14 @@ static int execAction (struct lexSpec *spec, struct lexRuleAction *ap,
             if (ap->u.pattern.body)
             {
                 arg_start[arg_no] = *pptr;
-                if (!tryMatch (spec, pptr, &sptr, ap->u.pattern.dfa))
+                if (!tryMatch (spec, pptr, &sptr, ap->u.pattern.dfa, 0))
                 {
                     arg_end[arg_no] = F_WIN_EOF;
                     arg_no++;
                     arg_start[arg_no] = F_WIN_EOF;
                     arg_end[arg_no] = F_WIN_EOF;
-/* return 1*/
+		    yaz_log(LOG_DEBUG, "Pattern match rest of record");
+		    *pptr = F_WIN_EOF;
                 }
                 else
                 {
@@ -1556,7 +1583,7 @@ static int execAction (struct lexSpec *spec, struct lexRuleAction *ap,
             else
             {
                 arg_start[arg_no] = *pptr;
-                if (!tryMatch (spec, pptr, &sptr, ap->u.pattern.dfa))
+                if (!tryMatch (spec, pptr, &sptr, ap->u.pattern.dfa, 1))
                     return 1;
                 if (sptr != arg_start[arg_no])
                     return 1;