PHP5 and LIBXML2 XML Parse Bug

[sundog_ak]

It took me some time to figure this out, but there is a major bug with XML parsing when using PHP 5.2.6 and libxml2 (versions 2.7.0 to 2.7.2). Basically, when using the libxml library, it will strip out all the HTML entities so that all the "<" and "&" characters are stripped out. This renders the output as gibberish when using products like Live Writer and ScribeFire to upload content to sites like Word Press or Joomla. Other descriptions of issue (ARGH: I cannot post links do to the minimum 5 post limit):

Seach for php bug #45996
Typo3 bugtracker 00093359

The main problem is there doesn't seem to be any rush by either PHP or the libxml developers to fix this bug. There seems to be a little bit of "debate" on who needs to fix it from what I have read. At any rate, the issue is still present and no immediate patches in site. There seems to be two fixes at the moment:

1) Go back to the libxml versions less than 2.7.0 (which has potential issues), or

2) Recompile PHP5 using libexpat

Option two seems to be the best solution, but I am on Glowhost shared server (Advanced 5 Star) and do not have ability to recompile. Is option 2 something Glowhost would be able to do if I submitted a trouble ticket? I have waited many months for this bug to get resolved and I do not see anything in the PHP or libxml bug listings to indicate it will be.

Cheers,
Chris

[charlesh]

What about the built-in php xml parsing?

[sundog_ak]

Well, the issue is with the built-in php xml parsing when php is compiled using the libxml library rather than the expat library. Most hosts, including Glowhost (at least on the Deb server I am on) compile php with the libxml library which when using the php xml parsing functions results in the bug. If you go to that same link you provided and go to the "installation" heading, it describes (again I cannot provide link). You can test this problem with a simple set of code on your server:
=================
<?php
$simple = "<root><node><strong>Teststring</strong></node></root>";
$p = xml_parser_create();
xml_parse_into_struct($p, $simple, $vals, $index);
xml_parser_free($p);
echo $vals[1]['value'];
?>
=======================
You should get output of a bolded text of "Teststring" as formatted like:

<strong>Teststring</strong>

instead, with php compiled to use libxml for xml php parsing, it will strip out the "<" resulting in:

strongTeststring/strong

Thus, produces gibberish and loss of formatting and is not something I can fix on a shared server since I cannot recompile php to use the expat libary rather than libxml for parsing. When I get to 5 posts I can post some more links on the issue.

Thanks,
Chris

[sundog_ak]

Well, I just saw at php.net that they fixed in CVS as long as the libxml2 2.7.3 or newer is used. I guess I will just have to wait for Glowhost to update to newest versions of php and libxml.

http : // bugs.php.net/bug.php?id=45996

[charlesh]

Have you submitted a ticket on this to Glowhost?

[sundog_ak]

Thanks, I will do that once PHP 5.2.9 and LibXML2 2.7.3 are released stable (hopefully soon).

Cheers,
Chris

[sundog_ak]

I see the Deb server I am on has been updated with LibXML2 2.7.3 (half way there! and PHP 5.2.8. Hopefully, PHP will release 5.2.9 soon and this will be a resolved issue.

I saw a couple posts that 5.2.9 would be released around February. It looks like Glowhost updates relatively quickly once things go stable (unlike my prior host where it took them forever to update).