Hello.
There are quite a long-standing problem in the code xinclude.c,
described in libxml2 bugzilla:
https://bugzilla.gnome.org/show_bug.cgi?id=633166
When you xinclude a text file, reading portions (by 4000 bytes) of the
buffer incorrectly handled the situation when the end comes across
portions of the bytes in a multibyte character.
I enclose a patch (also attached to the bugzilla bug report).
Please review the patch.
--
Regards, Vyt
diff --git a/xinclude.c b/xinclude.c
index 2916ffa..b58dc43 100644
--- a/xinclude.c
+++ b/xinclude.c
@@ -1880,6 +1880,8 @@ xmlXIncludeLoadTxt(xmlXIncludeCtxtPtr ctxt, const xmlChar *url, int nr) {
/*
* Scan all chars from the resource and add the to the node
*/
+ int xinclude_multibyte_fallback_used = 0;
+xinclude_multibyte_fallback:
while (xmlParserInputBufferRead(buf, 128) > 0) {
int len;
const xmlChar *content;
@@ -1892,13 +1894,21 @@ xmlXIncludeLoadTxt(xmlXIncludeCtxtPtr ctxt, const xmlChar *url, int nr) {
cur = xmlStringCurrentChar(NULL, &content[i], &l);
if (!IS_CHAR(cur)) {
+ // Handle splitted multibyte char at buffer boundary
+ if ((len - i) < 4 && !xinclude_multibyte_fallback_used) {
+ xinclude_multibyte_fallback_used = 1;
+ xmlBufferShrink(buf->buffer, i);
+ goto xinclude_multibyte_fallback;
+ } else {
xmlXIncludeErr(ctxt, ctxt->incTab[nr]->ref,
XML_XINCLUDE_INVALID_CHAR,
"%s contains invalid char\n", URL);
xmlFreeParserInputBuffer(buf);
xmlFree(URL);
return(-1);
+ }
} else {
+ xinclude_multibyte_fallback_used = 0;
xmlNodeAddContentLen(node, &content[i], l);
}
i += l;
_______________________________________________
xml mailing list, project page http://xmlsoft.org/
[email protected]
https://mail.gnome.org/mailman/listinfo/xml
