From c7c496434205bb51aa8873e434ca282bbb8d1fe4 Mon Sep 17 00:00:00 2001
From: Nick Wellnhofer <wellnhofer@aevum.de>
Date: Fri, 9 May 2025 15:26:15 +0200
Subject: [PATCH] html: Move DTD creation to endDocument SAX callback

---
 HTMLparser.c | 27 ---------------------------
 SAX2.c       | 17 ++++++++++++++++-
 2 files changed, 16 insertions(+), 28 deletions(-)

diff --git a/HTMLparser.c b/HTMLparser.c
index f3e04834..1161b2b3 100644
--- a/HTMLparser.c
+++ b/HTMLparser.c
@@ -4331,8 +4331,6 @@ htmlCtxtParseContentInternal(htmlParserCtxtPtr ctxt, xmlParserInputPtr input) {
  */
 int
 htmlParseDocument(htmlParserCtxtPtr ctxt) {
-    xmlDtdPtr dtd;
-
     if ((ctxt == NULL) || (ctxt->input == NULL))
 	return(-1);
 
@@ -4429,17 +4427,6 @@ htmlParseDocument(htmlParserCtxtPtr ctxt) {
     if ((ctxt->sax) && (ctxt->sax->endDocument != NULL))
         ctxt->sax->endDocument(ctxt->userData);
 
-    if ((!(ctxt->options & HTML_PARSE_NODEFDTD)) && (ctxt->myDoc != NULL)) {
-	dtd = xmlGetIntSubset(ctxt->myDoc);
-	if (dtd == NULL) {
-	    ctxt->myDoc->intSubset =
-		xmlCreateIntSubset(ctxt->myDoc, BAD_CAST "html",
-		    BAD_CAST "-//W3C//DTD HTML 4.0 Transitional//EN",
-		    BAD_CAST "http://www.w3.org/TR/REC-html40/loose.dtd");
-            if (ctxt->myDoc->intSubset == NULL)
-                htmlErrMemory(ctxt);
-        }
-    }
     if (! ctxt->wellFormed) return(-1);
     return(0);
 }
@@ -5158,20 +5145,6 @@ htmlParseChunk(htmlParserCtxtPtr ctxt, const char *chunk, int size,
         if ((ctxt->sax) && (ctxt->sax->endDocument != NULL))
             ctxt->sax->endDocument(ctxt->userData);
 
-        if ((!(ctxt->options & HTML_PARSE_NODEFDTD)) &&
-            (ctxt->myDoc != NULL)) {
-            xmlDtdPtr dtd;
-            dtd = xmlGetIntSubset(ctxt->myDoc);
-            if (dtd == NULL) {
-                ctxt->myDoc->intSubset =
-                    xmlCreateIntSubset(ctxt->myDoc, BAD_CAST "html",
-                        BAD_CAST "-//W3C//DTD HTML 4.0 Transitional//EN",
-                        BAD_CAST "http://www.w3.org/TR/REC-html40/loose.dtd");
-                if (ctxt->myDoc->intSubset == NULL)
-                    htmlErrMemory(ctxt);
-            }
-        }
-
 	ctxt->instate = XML_PARSER_EOF;
     }
 
diff --git a/SAX2.c b/SAX2.c
index 6da5123c..f454bb83 100644
--- a/SAX2.c
+++ b/SAX2.c
@@ -837,7 +837,10 @@ xmlSAX2EndDocument(void *ctx)
 #endif /* LIBXML_VALID_ENABLED */
 
     doc = ctxt->myDoc;
-    if ((doc != NULL) && (doc->encoding == NULL)) {
+    if (doc == NULL)
+        return;
+
+    if (doc->encoding == NULL) {
         const xmlChar *encoding = xmlGetActualEncoding(ctxt);
 
         if (encoding != NULL) {
@@ -846,6 +849,18 @@ xmlSAX2EndDocument(void *ctx)
                 xmlSAX2ErrMemory(ctxt);
         }
     }
+
+#ifdef LIBXML_HTML_ENABLED
+    if ((ctxt->html) &&
+        ((ctxt->options & HTML_PARSE_NODEFDTD) == 0) &&
+        (doc->intSubset == NULL)) {
+        doc->intSubset = xmlCreateIntSubset(doc, BAD_CAST "html",
+                BAD_CAST "-//W3C//DTD HTML 4.0 Transitional//EN",
+                BAD_CAST "http://www.w3.org/TR/REC-html40/loose.dtd");
+        if (doc->intSubset == NULL)
+            xmlSAX2ErrMemory(ctxt);
+    }
+#endif /* LIBXML_HTML_ENABLED */
 }
 
 static void