Bug #76696 for XML-LibXML: keep_blanks has no effect on parse_balanced

Thu Apr 19 06:01:51 2012 SREZIC [...] cpan.org - Ticket created

Subject:

keep_blanks has no effect on parse_balanced_chunk anymore

Specifying ->keep_blanks(0) has no effect on parse_balanced_chunk anymore. The script below used to pass with XML::LibXML 1.69, but is broken since 1.70 and also with the newest 1.96. Regards, Slaven #!/usr/bin/perl use strict; use warnings; use Test::More 'no_plan'; use XML::LibXML; my $xml = <<EOF; <bla> <foo/> </bla> EOF my $p = XML::LibXML->new; $p->keep_blanks(0); is $p->parse_balanced_chunk($xml)->serialize, "<bla><foo/></bla>\n"; __END__

Thu Apr 19 06:02:46 2012 SREZIC [...] cpan.org - Cc srezic@iconmobile.com added

Sat Apr 21 14:44:33 2012 SHLOMIF [...] cpan.org - Correspondence added

Hi Slaven, thanks for the repot and the test. I'll look into it. Regards, -- Shlomi Fish On Thu Apr 19 06:01:51 2012, SREZIC wrote: Show quoted text

> Specifying ->keep_blanks(0) has no effect on parse_balanced_chunk > anymore. The script below used to pass with XML::LibXML 1.69, but is > broken since 1.70 and also with the newest 1.96. > > Regards, > Slaven > > #!/usr/bin/perl > use strict; > use warnings; > > use Test::More 'no_plan'; > use XML::LibXML; > > my $xml = <<EOF; > <bla> > <foo/> > </bla> > EOF > > my $p = XML::LibXML->new; > $p->keep_blanks(0); > is $p->parse_balanced_chunk($xml)->serialize, "<bla><foo/></bla>\n"; > > __END__

Sat Apr 21 14:44:34 2012 The RT System itself - Status changed from 'new' to 'open'

Tue Apr 24 05:20:27 2012 SREZIC [...] cpan.org - Correspondence added

I assume the problem was caused by changing the keep_blanks/no_blanks support from using xmlKeepBlanksDefault() to the use of the parser option XML_PARSER_NOBLANKS. Maybe the code behind parse_balanced_chunk() cannot deal with the parser options, but still needs xmlKeepBlanksDefault(). The attached patch works for me, but I don't know if this is the correct way to fix this problem. Regards, Slaven

Subject:

parse-balanced-chunk.patch

diff --git a/LibXML.xs b/LibXML.xs index 535302c..35e1ab1 100644 --- a/LibXML.xs +++ b/LibXML.xs @@ -891,6 +891,12 @@ LibXML_init_parser( SV * self, xmlParserCtxtPtr ctxt ) { } if (ctxt) xmlCtxtUseOptions(ctxt, parserOptions ); /* Note: sets ctxt->linenumbers = 1 */ + if (parserOptions & XML_PARSE_NOBLANKS) { + xmlKeepBlanksDefault(0); + } else { + xmlKeepBlanksDefault(1); + } + item = hv_fetch( real_obj, "XML_LIBXML_LINENUMBERS", 22, 0 ); if ( item != NULL && SvTRUE(*item) ) { if (ctxt) ctxt->linenumbers = 1;

Mon Apr 30 13:38:42 2012 SHLOMIF [...] cpan.org - Correspondence added

Hi Slaven, thanks for the report, the test case and the fix. It was applied (with some comments, cleanups and clarifications) in the repository and released to CPAN as release "1.97". Please open a new ticket if there are more problems, as I'm resolving this ticket. Regards, -- Shlomi Fish

Mon Apr 30 13:38:43 2012 SHLOMIF [...] cpan.org - Status changed from 'open' to 'resolved'

Mon May 14 07:49:38 2012 NWELLNHOF [...] cpan.org - Correspondence added

Unfortunately, there is no way to pass parser options to xmlParseBalancedChunkMemory. So if you want to use that function without keeping blanks, you have to change the global setting. But using globals to configure a library is a really bad idea, since there might be several other users of libxml2 in the same process (think of an Apache process with mod_perl, PHP and all kind of other modules). If you set the keep_blanks global, all other users of library will be affected and it's possible that something breaks in a very hard to debug way. That was the motivation behind removing the calls to xmlKeepBlanksDefault in version 1.70.

Mon May 14 07:49:40 2012 The RT System itself - Status changed from 'resolved' to 'open'

Bug #76696 for XML-LibXML: keep_blanks has no effect on parse_balanced_chunk anymore