On Tue, Apr 14, 2015 at 05:43:42PM +0200, Christian Schoenebeck wrote:
> On Tuesday 14 April 2015 17:50:51 you wrote:
> > If anything like this does get put in, it should only be if it is a
> > configurable option that is disabled by default - an xml parser should
> > only accept a strictly-conforming document by default. Adding support for
> > ‘broken’ html because other (weak) parsers allow it is not a good plan as
> > it causes divergence from the standard.
>
> There you go; you find the updated patch attached. It now requires
> HTML_PARSE_RECOVER option to be set for recovering from stand-alone less-than
> characters.
That sounds fine *except* it doesn't raise an error.
The parser knows it's a broken construct that must be pointed out.
thinkpad2:~/XML -> ./xmllint --html tst.html
tst.html:3: HTML parser error : htmlParseStartTag: invalid element name
<p> blah < booh </p>
^
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN"
"http://www.w3.org/TR/REC-html40/loose.dtd";>
<html>
<body>
<p> blah
</p>
</body>
</html>
thinkpad2:~/XML -> ./xmllint --html --recover tst.html
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN"
"http://www.w3.org/TR/REC-html40/loose.dtd";>
<html>
<body>
<p> blah < booh </p>
</body>
</html>
thinkpad2:~/XML ->
the fact that we worked around a broken start tag construct must be reported.
Whether we do that with the recovery option or not is less important IMHO.
It sounds a bit weird to handle that error case as one of the main content
cases, I would still be tempted to go into htmlParseStartTag, get the
error reported, but push corrective data instead in recover mode.
Can we get a v3 ? :-)
thanks
Daniel
> Best regards,
> Christian Schoenebeck
> diff -u libxml2-2.9.1+dfsg1.orig/HTMLparser.c libxml2-2.9.1+dfsg1/HTMLparser.c
> --- libxml2-2.9.1+dfsg1.orig/HTMLparser.c 2015-04-14 13:05:01.000000000
> +0200
> +++ libxml2-2.9.1+dfsg1/HTMLparser.c 2015-04-14 18:22:41.143973776 +0200
> @@ -2948,8 +2948,10 @@
>
>
> /**
> - * htmlParseCharData:
> + * htmlParseCharDataInternal:
> * @ctxt: an HTML parser context
> + * @prep: optional character to be prepended to text, 0 if no character
> + * shall be prepended
> *
> * parse a CharData section.
> * if we are within a CDATA section ']]>' marks an end of section.
> @@ -2958,12 +2960,15 @@
> */
>
> static void
> -htmlParseCharData(htmlParserCtxtPtr ctxt) {
> - xmlChar buf[HTML_PARSER_BIG_BUFFER_SIZE + 5];
> +htmlParseCharDataInternal(htmlParserCtxtPtr ctxt, char prep) {
> + xmlChar buf[HTML_PARSER_BIG_BUFFER_SIZE + 6];
> int nbchar = 0;
> int cur, l;
> int chunk = 0;
>
> + if (prep)
> + buf[nbchar++] = prep;
> +
> SHRINK;
> cur = CUR_CHAR(l);
> while (((cur != '<') || (ctxt->token == '<')) &&
> @@ -3043,6 +3048,21 @@
> }
>
> /**
> + * htmlParseCharData:
> + * @ctxt: an HTML parser context
> + *
> + * parse a CharData section.
> + * if we are within a CDATA section ']]>' marks an end of section.
> + *
> + * [14] CharData ::= [^<&]* - ([^<&]* ']]>' [^<&]*)
> + */
> +
> +static void
> +htmlParseCharData(htmlParserCtxtPtr ctxt) {
> + htmlParseCharDataInternal(ctxt, 0);
> +}
> +
> +/**
> * htmlParseExternalID:
> * @ctxt: an HTML parser context
> * @publicID: a xmlChar** receiving PubidLiteral
> @@ -4157,14 +4177,24 @@
> }
>
> /*
> - * Third case : a sub-element.
> + * Third case : (unescaped) stand-alone less-than character.
> + * Only if HTML_PARSE_RECOVER option is set.
> + */
> + else if (ctxt->recovery && (CUR == '<') &&
> + (IS_BLANK_CH(NXT(1)) || (NXT(1) == '='))) {
> + NEXT;
> + htmlParseCharDataInternal(ctxt, '<');
> + }
> +
> + /*
> + * Fourth case : a sub-element.
> */
> else if (CUR == '<') {
> htmlParseElement(ctxt);
> }
>
> /*
> - * Fourth case : a reference. If if has not been resolved,
> + * Fifth case : a reference. If if has not been resolved,
> * parsing returns it's Name, create the node
> */
> else if (CUR == '&') {
> @@ -4172,7 +4202,7 @@
> }
>
> /*
> - * Fifth case : end of the resource
> + * Sixth case : end of the resource
> */
> else if (CUR == 0) {
> htmlAutoCloseOnEnd(ctxt);
> @@ -4567,7 +4597,17 @@
> }
>
> /*
> - * Third case : a sub-element.
> + * Third case : (unescaped) stand-alone less-than character.
> + * Only if HTML_PARSE_RECOVER option is set.
> + */
> + else if (ctxt->recovery && (CUR == '<') &&
> + (IS_BLANK_CH(NXT(1)) || (NXT(1) == '='))) {
> + NEXT;
> + htmlParseCharDataInternal(ctxt, '<');
> + }
> +
> + /*
> + * Fourth case : a sub-element.
> */
> else if (CUR == '<') {
> htmlParseElementInternal(ctxt);
> @@ -4578,7 +4618,7 @@
> }
>
> /*
> - * Fourth case : a reference. If if has not been resolved,
> + * Fifth case : a reference. If if has not been resolved,
> * parsing returns it's Name, create the node
> */
> else if (CUR == '&') {
> @@ -4586,7 +4626,7 @@
> }
>
> /*
> - * Fifth case : end of the resource
> + * Sixth case : end of the resource
> */
> else if (CUR == 0) {
> htmlAutoCloseOnEnd(ctxt);
> _______________________________________________
> xml mailing list, project page http://xmlsoft.org/
> [email protected]
> https://mail.gnome.org/mailman/listinfo/xml
--
Daniel Veillard | Open Source and Standards, Red Hat
[email protected] | libxml Gnome XML XSLT toolkit http://xmlsoft.org/
http://veillard.com/ | virtualization library http://libvirt.org/
_______________________________________________
xml mailing list, project page http://xmlsoft.org/
[email protected]
https://mail.gnome.org/mailman/listinfo/xml
