From a80f8b64a920e36518098b44976a7194704a1200 Mon Sep 17 00:00:00 2001
From: Nick Wellnhofer <wellnhofer@aevum.de>
Date: Thu, 4 May 2023 15:59:31 +0200
Subject: [PATCH] html: Allow attributes in end tags

Attribute are syntactically allowed in HTML5 end tags but otherwise
ignored.
---
 HTMLparser.c               | 35 +++++++++++++++++++++++++++--------
 result/HTML/doc3.htm.err   |  3 ---
 result/HTML/doc3.htm.sax   |  1 -
 result/HTML/wired.html.err |  3 ---
 result/HTML/wired.html.sax |  1 -
 5 files changed, 27 insertions(+), 16 deletions(-)

diff --git a/HTMLparser.c b/HTMLparser.c
index 6a21d5a0..5ced00ee 100644
--- a/HTMLparser.c
+++ b/HTMLparser.c
@@ -3980,20 +3980,39 @@ htmlParseEndTag(htmlParserCtxtPtr ctxt)
     name = htmlParseHTMLName(ctxt, 0);
     if (name == NULL)
         return (0);
+
     /*
-     * We should definitely be at the ending "S? '>'" part
+     * Parse and ignore attributes.
      */
     SKIP_BLANKS;
-    if (CUR != '>') {
+    while ((CUR != 0) &&
+           (CUR != '>') &&
+	   ((CUR != '/') || (NXT(1) != '>')) &&
+           (ctxt->instate != XML_PARSER_EOF)) {
+        xmlChar *attvalue = NULL;
+
+        /*  unexpected-solidus-in-tag */
+        if (CUR == '/') {
+            NEXT;
+            SKIP_BLANKS;
+            continue;
+        }
+	GROW;
+	htmlParseAttribute(ctxt, &attvalue);
+        if (attvalue != NULL)
+            xmlFree(attvalue);
+
+	SKIP_BLANKS;
+    }
+
+    if (CUR == '>') {
+        NEXT;
+    } else if ((CUR == '/') && (NXT(1) == '>')) {
+        SKIP(2);
+    } else {
         htmlParseErr(ctxt, XML_ERR_GT_REQUIRED,
 	             "End tag : expected '>'\n", NULL, NULL);
-        /* Skip to next '>' */
-        while ((PARSER_STOPPED(ctxt) == 0) &&
-               (CUR != 0) && (CUR != '>'))
-            NEXT;
     }
-    if (CUR == '>')
-        NEXT;
 
     /*
      * if we ignored misplaced tags in htmlParseStartTag don't pop them
diff --git a/result/HTML/doc3.htm.err b/result/HTML/doc3.htm.err
index bb9204eb..7b7c5749 100644
--- a/result/HTML/doc3.htm.err
+++ b/result/HTML/doc3.htm.err
@@ -40,9 +40,6 @@ om/ad_static.asp?pid=2097&sid=1881&asid=7708"></a></IFRAME></CENTER></LI></FONT>
 ./test/HTML/doc3.htm:795: HTML parser error : Unexpected end tag : iframe
 							document.write("42DF8478957377></IFRAME>");
 							                                        ^
-./test/HTML/doc3.htm:804: HTML parser error : End tag : expected '>'
-							document.write("RIPT>");
-							^
 ./test/HTML/doc3.htm:804: HTML parser error : Unexpected end tag : sc");
 							document.write("RIPT>");
 							                     ^
diff --git a/result/HTML/doc3.htm.sax b/result/HTML/doc3.htm.sax
index f2009a95..5545f2d7 100644
--- a/result/HTML/doc3.htm.sax
+++ b/result/HTML/doc3.htm.sax
@@ -2696,7 +2696,6 @@ SAX.cdata(
 SAX.error: Unexpected end tag : iframe
 SAX.cdata(");
 							} else if ((parseI, 463)
-SAX.error: End tag : expected '>'
 SAX.error: Unexpected end tag : sc");
 SAX.cdata(");
 							} else {
diff --git a/result/HTML/wired.html.err b/result/HTML/wired.html.err
index 6b8d8184..244b98ed 100644
--- a/result/HTML/wired.html.err
+++ b/result/HTML/wired.html.err
@@ -217,9 +217,6 @@ wired.com&BANNER=Sprint" style="text-decoration:none"><font color="#000000">Spri
 ./test/HTML/wired.html:408: HTML parser error : Opening and ending tag mismatch: a and font
 com&BANNER=Sprint" style="text-decoration:none"><font color="#000000">Sprint</a>
                                                                                ^
-./test/HTML/wired.html:408: HTML parser error : End tag : expected '>'
-Sprint" style="text-decoration:none"><font color="#000000">Sprint</a></i></font<
-                                                                               ^
 ./test/HTML/wired.html:408: HTML parser error : Unexpected end tag : font<
 " style="text-decoration:none"><font color="#000000">Sprint</a></i></font</font>
                                                                                ^
diff --git a/result/HTML/wired.html.sax b/result/HTML/wired.html.sax
index ce51ed60..54f4a9a6 100644
--- a/result/HTML/wired.html.sax
+++ b/result/HTML/wired.html.sax
@@ -1960,7 +1960,6 @@ SAX.error: Opening and ending tag mismatch: a and font
 SAX.endElement(font)
 SAX.endElement(a)
 SAX.endElement(i)
-SAX.error: End tag : expected '>'
 SAX.error: Unexpected end tag : font<
 SAX.startElement(br)
 SAX.endElement(br)