From 2cbf596c7f6d4ef21798bb06efd7542781124300 Mon Sep 17 00:00:00 2001
From: Daniel Veillard <veillard@src.gnome.org>
Date: Wed, 31 Mar 2004 15:50:43 +0000
Subject: [PATCH] patched a bug in parsing production 1 and 2 of xmlschemas
 regexp that

* xmlregexp.c: patched a bug in parsing production 1 and 2 of
  xmlschemas regexp that William pointed out while working on
  #134120
* test/regexp/branch result/regexp/branch: added a specific
  regression test
Daniel
---
 ChangeLog            |  8 +++++
 result/regexp/branch | 14 +++++++++
 test/regexp/branch   | 14 +++++++++
 xmlregexp.c          | 71 +++++++++++++++++++++++---------------------
 4 files changed, 73 insertions(+), 34 deletions(-)
 create mode 100644 result/regexp/branch
 create mode 100644 test/regexp/branch

diff --git a/ChangeLog b/ChangeLog
index cb3505a9..13615a8f 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,11 @@
+Wed Mar 31 17:47:28 CEST 2004 Daniel Veillard <daniel@veillard.com>
+
+	* xmlregexp.c: patched a bug in parsing production 1 and 2 of
+	  xmlschemas regexp that William pointed out while working on
+	  #134120
+	* test/regexp/branch result/regexp/branch: added a specific
+	  regression test
+
 Wed Mar 31 09:50:32 HKT 2004 William Brack <wbrack@mmm.com.hk>
 
 	* Makefile.am: added PYTHONPATH to python tests for Schemas
diff --git a/result/regexp/branch b/result/regexp/branch
new file mode 100644
index 00000000..dd703ac2
--- /dev/null
+++ b/result/regexp/branch
@@ -0,0 +1,14 @@
+Regexp: a|b(d|e(g|h|i)|f)|c
+a: Ok
+c: Ok
+bd: Ok
+bf: Ok
+beg: Ok
+beh: Ok
+bei: Ok
+b: Fail
+be: Fail
+bi: Fail
+f: Fail
+ab: Fail
+ac: Fail
diff --git a/test/regexp/branch b/test/regexp/branch
new file mode 100644
index 00000000..6a728e67
--- /dev/null
+++ b/test/regexp/branch
@@ -0,0 +1,14 @@
+=>a|b(d|e(g|h|i)|f)|c
+a
+c
+bd
+bf
+beg
+beh
+bei
+b
+be
+bi
+f
+ab
+ac
diff --git a/xmlregexp.c b/xmlregexp.c
index 057458b2..d1e6f38d 100644
--- a/xmlregexp.c
+++ b/xmlregexp.c
@@ -1214,14 +1214,15 @@ xmlRegStateAddTrans(xmlRegParserCtxtPtr ctxt, xmlRegStatePtr state,
 #ifdef DEBUG_REGEXP_GRAPH
     printf("Add trans from %d to %d ", state->no, target->no);
     if (count == REGEXP_ALL_COUNTER)
-	printf("all transition");
+	printf("all transition\n");
     else if (count >= 0)
-	printf("count based %d", count);
+	printf("count based %d\n", count);
     else if (counter >= 0)
-	printf("counted %d", counter);
+	printf("counted %d\n", counter);
     else if (atom == NULL)
-	printf("epsilon transition");
-    printf("\n");
+	printf("epsilon transition\n");
+    else if (atom != NULL) 
+        xmlRegPrintAtom(stdout, atom);
 #endif
 
     state->trans[state->nbTrans].atom = atom;
@@ -3862,50 +3863,33 @@ xmlFAParsePiece(xmlRegParserCtxtPtr ctxt) {
 /**
  * xmlFAParseBranch:
  * @ctxt:  a regexp parser context
- * @first:  is taht the first
  *
  * [2]   branch   ::=   piece*
  8
  */
 static int
-xmlFAParseBranch(xmlRegParserCtxtPtr ctxt, int first) {
+xmlFAParseBranch(xmlRegParserCtxtPtr ctxt) {
     xmlRegStatePtr previous;
-    xmlRegAtomPtr prevatom = NULL;
     int ret;
 
     previous = ctxt->state;
     ret = xmlFAParsePiece(ctxt);
     if (ret != 0) {
-	if (first) {
-	    if (xmlFAGenerateTransitions(ctxt, previous, NULL, ctxt->atom) < 0)
-	        return(-1);
-	    previous = ctxt->state;
-	} else {
-	    prevatom = ctxt->atom;
-	}
+	if (xmlFAGenerateTransitions(ctxt, previous, NULL, ctxt->atom) < 0)
+	    return(-1);
+	previous = ctxt->state;
 	ctxt->atom = NULL;
     }
     while ((ret != 0) && (ctxt->error == 0)) {
 	ret = xmlFAParsePiece(ctxt);
 	if (ret != 0) {
-	    if (first) {
-		if (xmlFAGenerateTransitions(ctxt, previous, NULL,
-		                             ctxt->atom) < 0)
+	    if (xmlFAGenerateTransitions(ctxt, previous, NULL,
+					 ctxt->atom) < 0)
 		    return(-1);
-	    } else {
-		if (xmlFAGenerateTransitions(ctxt, previous, NULL,
-		                             prevatom) < 0)
-		    return(-1);
-		prevatom = ctxt->atom;
-	    }
 	    previous = ctxt->state;
 	    ctxt->atom = NULL;
 	}
     }
-    if (!first) {
-	if (xmlFAGenerateTransitions(ctxt, previous, ctxt->end, prevatom) < 0)
-	    return(-1);
-    }
     return(0);
 }
 
@@ -3918,12 +3902,21 @@ xmlFAParseBranch(xmlRegParserCtxtPtr ctxt, int first) {
  */
 static void
 xmlFAParseRegExp(xmlRegParserCtxtPtr ctxt, int top) {
-    xmlRegStatePtr start, end, oldend;
+    xmlRegStatePtr start, end, oldend, oldstart;
 
     oldend = ctxt->end;
 
+    oldstart = ctxt->state;
+    /* if not top start should have been generated by an epsilon trans */
     start = ctxt->state;
-    xmlFAParseBranch(ctxt, (ctxt->end == NULL));
+    ctxt->end = NULL;
+    xmlFAParseBranch(ctxt);
+    if (top) {
+#ifdef DEBUG_REGEXP_GRAPH
+	printf("State %d is final\n", ctxt->state->no);
+#endif
+	ctxt->state->type = XML_REGEXP_FINAL_STATE;
+    }
     if (CUR != '|') {
 	ctxt->end = ctxt->state;
 	return;
@@ -3932,11 +3925,21 @@ xmlFAParseRegExp(xmlRegParserCtxtPtr ctxt, int top) {
     while ((CUR == '|') && (ctxt->error == 0)) {
 	NEXT;
 	ctxt->state = start;
-	ctxt->end = end;
-	xmlFAParseBranch(ctxt, 0);
+	ctxt->end = NULL;
+	xmlFAParseBranch(ctxt);
+	if (top) {
+	    ctxt->state->type = XML_REGEXP_FINAL_STATE;
+#ifdef DEBUG_REGEXP_GRAPH
+	    printf("State %d is final\n", ctxt->state->no);
+#endif
+	} else {
+	    xmlFAGenerateEpsilonTransition(ctxt, ctxt->state, end);
+	}
+    }
+    if (!top) {
+	ctxt->state = end;
+	ctxt->end = end;
     }
-    if (!top)
-	ctxt->end = oldend;
 }
 
 /************************************************************************