- SAX.c entities.c parser.c: changed completely the way entities

  are handled when running the parser in entity substitution mode.
  This fixes a bug reported by Stephan Kulow and nearly divides
  by 3 the amount of memory required by libxslt to load and process
  DocBook TDG.
Daniel

4 files changed, 64 insertions, 12 deletions

diff --git a/ChangeLog b/ChangeLog
index b00fb4f2..9b9d703c 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,11 @@
+Wed Jul  4 21:42:24 CEST 2001 Daniel Veillard <Daniel.Veillard@imag.fr>
+
+	* SAX.c entities.c parser.c: changed completely the way entities
+	  are handled when running the parser in entity substitution mode.
+	  This fixes a bug reported by Stephan Kulow and nearly divides
+	  by 3 the amount of memory required by libxslt to load and process
+	  DocBook TDG.
+
 Wed Jul  4 18:02:58 CEST 2001 Daniel Veillard <Daniel.Veillard@imag.fr>
 
 	* HTMLparser.c: fixing a too early root closing problem raised
diff --git a/SAX.c b/SAX.c
index 72b82e46..a1079289 100644
--- a/SAX.c
+++ b/SAX.c
@@ -885,7 +885,8 @@ attribute(void *ctx, const xmlChar *fullname, const xmlChar *value)
 	    ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt, ctxt->myDoc,
 					       ctxt->node, ret, value);
 	}
-    } else if (ctxt->external != 2){
+    } else if (((ctxt->replaceEntities == 0) && (ctxt->external != 2)) ||
+	       ((ctxt->replaceEntities != 0) && (ctxt->inSubset == 0))) {
         /*
 	 * when validating, the ID registration is done at the attribute
 	 * validation level. Otherwise we have to do specific handling here.
diff --git a/entities.c b/entities.c
index 94f1f6ff..4ecb98c9 100644
--- a/entities.c
+++ b/entities.c
@@ -46,7 +46,7 @@ xmlHashTablePtr xmlPredefinedEntities = NULL;
 static void xmlFreeEntity(xmlEntityPtr entity) {
     if (entity == NULL) return;
 
-    if (entity->children)
+    if ((entity->children) && (entity->children->parent == entity))
 	xmlFreeNodeList(entity->children);
     if (entity->name != NULL)
 	xmlFree((char *) entity->name);
diff --git a/parser.c b/parser.c
index 8d34e723..bac83573 100644
--- a/parser.c
+++ b/parser.c
@@ -5050,14 +5050,35 @@ xmlParseReference(xmlParserCtxtPtr ctxt) {
 			 (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY))&&
 			    (ent->children == NULL)) {
 			    ent->children = list;
-			    while (list != NULL) {
-				list->parent = (xmlNodePtr) ent;
-				if (list->next == NULL)
-				    ent->last = list;
-				list = list->next;
+			    if (ctxt->replaceEntities) {
+				/*
+				 * Prune it directly in the generated document
+				 * except for single text nodes.
+				 */
+				if ((list->type == XML_TEXT_NODE) &&
+				    (list->next == NULL)) {
+				    list->parent = (xmlNodePtr) ent;
+				    list = NULL;
+				} else {
+				    while (list != NULL) {
+					list->parent = (xmlNodePtr) ctxt->node;
+					if (list->next == NULL)
+					    ent->last = list;
+					list = list->next;
+				    }
+				    list = ent->children;
+				}
+			    } else {
+				while (list != NULL) {
+				    list->parent = (xmlNodePtr) ent;
+				    if (list->next == NULL)
+					ent->last = list;
+				    list = list->next;
+				}
 			    }
 			} else {
 			    xmlFreeNodeList(list);
+			    list = NULL;
 			}
 		    } else if (ret > 0) {
 			ctxt->errNo = ret;
@@ -5068,6 +5089,7 @@ xmlParseReference(xmlParserCtxtPtr ctxt) {
 			ctxt->disableSAX = 1;
 		    } else if (list != NULL) {
 			xmlFreeNodeList(list);
+			list = NULL;
 		    }
 		}
 	    }
@@ -5082,12 +5104,33 @@ xmlParseReference(xmlParserCtxtPtr ctxt) {
 		if ((ctxt->node != NULL) && (ent->children != NULL)) {
 		    /*
 		     * Seems we are generating the DOM content, do
-		     * a simple tree copy
+		     * a simple tree copy for all references except the first
+		     * In the first occurence list contains the replacement
 		     */
-		    xmlNodePtr new;
-		    new = xmlCopyNodeList(ent->children);
-		    
-		    xmlAddChildList(ctxt->node, new);
+		    if (list == NULL) {
+			xmlNodePtr new, cur;
+			cur = ent->children;
+			while (cur != NULL) {
+			    new = xmlCopyNode(cur, 1);
+			    xmlAddChild(ctxt->node, new);
+			    if (cur == ent->last)
+				break;
+			    cur = cur->next;
+			}
+		    } else {
+			/*
+			 * the name change is to avoid coalescing of the
+			 * node with a prossible previous text one which
+			 * would make ent->children a dandling pointer
+			 */
+			if (ent->children->type == XML_TEXT_NODE)
+			    ent->children->name = xmlStrdup(BAD_CAST "nbktext");
+			if ((ent->last != ent->children) &&
+			    (ent->last->type == XML_TEXT_NODE))
+			    ent->last->name = xmlStrdup(BAD_CAST "nbktext");
+			xmlAddChildList(ctxt->node, ent->children);
+		    }
+
 		    /*
 		     * This is to avoid a nasty side effect, see
 		     * characters() in SAX.c