starting to cleanup some of the problems exposed by the W3C/NIST

* SAX2.c parser.c valid.c: starting to cleanup some of the
  problems exposed by the W3C/NIST regression suite.
* result/ent7.sax result/xml2.sax: small fixes.
Daniel

ChangeLog

@@ -1,3 +1,9 @@
+Mon Sep 15 14:54:42 CEST 2003 Daniel Veillard <daniel@veillard.com>
+
+	* SAX2.c parser.c valid.c: starting to cleanup some of the
+	  problems exposed by the W3C/NIST regression suite.
+	* result/ent7.sax result/xml2.sax: small fixes.
+
 Mon Sep 15 11:46:47 CEST 2003 Daniel Veillard <daniel@veillard.com>
 
 	* parser.c: more parser error factoring

SAX2.c

@@ -1596,9 +1596,9 @@ decode:
 /**
  * xmlSAX2AttributeNs:
  * @ctx: the user data (XML parser context)
- * @localname:  the local name of the element
+ * @localname:  the local name of the attribute
- * @prefix:  the element namespace prefix if available
+ * @prefix:  the attribute namespace prefix if available
- * @URI:  the element namespace name if available
+ * @URI:  the attribute namespace name if available
  * @value:  Start of the attribute value
  * @valueend: end of the attribute value
  *
@@ -1618,11 +1618,6 @@ xmlSAX2AttributeNs(xmlParserCtxtPtr ctxt,
     xmlNsPtr namespace = NULL;
     xmlChar *dup = NULL;
 
-#if 0
-    TODO, check taht CDATA normalization is done at the
-    parser level !!!!!
-#endif
-
     /*
      * Note: if prefix == NULL, the attribute is not in the default namespace
      */
@@ -1662,31 +1657,50 @@ xmlSAX2AttributeNs(xmlParserCtxtPtr ctxt,
         if (!ctxt->replaceEntities) {
 	    dup = xmlSAX2DecodeAttrEntities(ctxt, value, valueend);
 	    if (dup == NULL) {
-	        /*
+	        if (*valueend == 0) {
-		 * cheaper to finally allocate here than duplicate
+		    ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt,
-		 * entry points in the full validation code
+				    ctxt->myDoc, ctxt->node, ret, value);
-		 */
+		} else {
-	        dup = xmlStrndup(value, valueend - value);
+		    /*
+		     * That should already be normalized.
+		     * cheaper to finally allocate here than duplicate
+		     * entry points in the full validation code
+		     */
+		    dup = xmlStrndup(value, valueend - value);
 
-		ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt,
+		    ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt,
-				ctxt->myDoc, ctxt->node, ret, dup);
+				    ctxt->myDoc, ctxt->node, ret, dup);
+		}
 	    } else {
-#if 0
+	        /*
-	        TODO
+		 * dup now contains a string of the flattened attribute
-		xmlChar *nvalnorm;
+		 * content with entities substitued. Check if we need to
-
+		 * apply an extra layer of normalization.
-		/*
-		 * Do the last stage of the attribute normalization
 		 * It need to be done twice ... it's an extra burden related
 		 * to the ability to keep references in attributes
 		 */
-		nvalnorm = xmlValidNormalizeAttributeValue(ctxt->myDoc,
+		if (ctxt->attsSpecial != NULL) {
-					    ctxt->node, fullname, dup);
+		    xmlChar *nvalnorm;
-		if (nvalnorm != NULL) {
+		    xmlChar fn[50];
-		    xmlFree(dup);
+		    xmlChar *fullname;
-		    dup = nvalnorm;
+		    
+		    fullname = xmlBuildQName(localname, prefix, fn, 50);
+		    if (fullname != NULL) {
+			ctxt->vctxt.valid = 1;
+		        nvalnorm = xmlValidCtxtNormalizeAttributeValue(
+			                 &ctxt->vctxt, ctxt->myDoc,
+					 ctxt->node, fullname, dup);
+			if (ctxt->vctxt.valid != 1)
+			    ctxt->valid = 0;
+
+			if ((fullname != fn) && (fullname != localname))
+			    xmlFree(fullname);
+			if (nvalnorm != NULL) {
+			    xmlFree(dup);
+			    dup = nvalnorm;
+			}
+		    }
 		}
-#endif
 
 		ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt,
 			        ctxt->myDoc, ctxt->node, ret, dup);
@@ -2221,7 +2235,7 @@ xmlSAX2CDataBlock(void *ctx, const xmlChar *value, int len)
     }
 }
 
-static xmlSAX2DefaultVersionValue = 2;
+static int xmlSAX2DefaultVersionValue = 2;
 
 /**
  * xmlSAXDefaultVersion:

parser.c

@@ -3024,10 +3024,10 @@ xmlParseAttValueComplex(xmlParserCtxtPtr ctxt, int *attlen, int normalize) {
 	   (c != '<')) {
 	if (c == 0) break;
 	if (c == '&') {
+	    in_space = 0;
 	    if (NXT(1) == '#') {
 		int val = xmlParseCharRef(ctxt);
 
-		in_space = 0;
 		if (val == '&') {
 		    if (ctxt->replaceEntities) {
 			if (len > buf_size - 10) {
@@ -3071,7 +3071,6 @@ xmlParseAttValueComplex(xmlParserCtxtPtr ctxt, int *attlen, int normalize) {
 		    } else {
 			buf[len++] = ent->content[0];
 		    }
-		    in_space = 0;
 		} else if ((ent != NULL) && 
 		           (ctxt->replaceEntities != 0)) {
 		    xmlChar *rep;

result/ent7.sax

@@ -1,4 +1,3 @@
-xmlSAXUserParseFile returned error 27
 SAX.setDocumentLocator()
 SAX.startDocument()
 SAX.internalSubset(item, , )

result/xml2.sax

@@ -1,4 +1,3 @@
-xmlSAXUserParseFile returned error 27
 SAX.setDocumentLocator()
 SAX.startDocument()
 SAX.internalSubset(test, , )

valid.c

@@ -3319,17 +3319,10 @@ xmlValidateAttributeValue2(xmlValidCtxtPtr ctxt, xmlDocPtr doc,
 	    xmlEntityPtr ent;
 
 	    ent = xmlGetDocEntity(doc, value);
+	    /* yeah it's a bit messy... */
 	    if ((ent == NULL) && (doc->standalone == 1)) {
 		doc->standalone = 0;
 		ent = xmlGetDocEntity(doc, value);
-		if (ent != NULL) {
-		    VERROR(ctxt->userData, 
-"standalone problem: attribute %s reference entity \"%s\" in external subset\n",
-			   name, value);
-		    /* WAIT to get answer from the Core WG on this 
-		    ret = 0;
-		     */
-		}
 	    } 
 	    if (ent == NULL) {
 		VERROR(ctxt->userData,