openclipart import fails when descriptions contain entities

[bbyak]

for example search for "lion" - no list is shown, the console shows:

/tmp/ocalfeed.xml:73: parser error : Entity 'uuml' not defined
      <content:encoded>Coat of arms of Russion, Z&uuml;rich, Switzerland</conten
                                                       ^
/tmp/ocalfeed.xml:86: parser error : Entity 'uuml' not defined
      <content:encoded>Coat of arms of Andelfingen, Z&uuml;rich, Switzerland</co
                                                           ^
/tmp/ocalfeed.xml:125: parser error : Entity 'auml' not defined
      <content:encoded>Coat of arms of Pf&auml;ffikon, Z&uuml;rich, Switzerland<
                                               ^
/tmp/ocalfeed.xml:125: parser error : Entity 'uuml' not defined
      <content:encoded>Coat of arms of Pf&auml;ffikon, Z&uuml;rich, Switzerland<
                                                              ^
/tmp/ocalfeed.xml:164: parser error : Entity 'szlig' not defined
      <content:encoded>Wappentier von Gie&szlig;en: roter L&ouml;we mit schwarze
                                                ^
/tmp/ocalfeed.xml:164: parser error : Entity 'ouml' not defined
      <content:encoded>Wappentier von Gie&szlig;en: roter L&ouml;we mit schwarze
                                                                 ^
/tmp/ocalfeed.xml:164: parser error : Entity 'uuml' not defined
tent:encoded>Wappentier von Gie&szlig;en: roter L&ouml;we mit schwarzen Fl&uuml;
                                                                               ^

[bbyak (buliabyak)]

This constitutes a good share of all ocal searches, so it makes this feature almost useless. Since it is one of the important new features in 0.46, raising priority.

[MenTaLguY (mental-deactivatedaccount)]

Unfortunately this is an ocal bug rather than an Inkscape bug; entites like &uuml; etc. are specific to HTML and don't exist in XML. If the OCAL uses them then it needs to define the values for those entities in an internal DTD subset.

[Bryce Harrington (bryce)]

When doing an import from openclipart via Inkscape for example, we see:

search for "lion" - no list is shown, the console shows:

/tmp/ocalfeed.xml:73: parser error : Entity 'uuml' not defined
      <content:encoded>Coat of arms of Russion, Z&uuml;rich, Switzerland</conten
                                                       ^
/tmp/ocalfeed.xml:86: parser error : Entity 'uuml' not defined
      <content:encoded>Coat of arms of Andelfingen, Z&uuml;rich, Switzerland</co
                                                           ^
/tmp/ocalfeed.xml:125: parser error : Entity 'auml' not defined
      <content:encoded>Coat of arms of Pf&auml;ffikon, Z&uuml;rich, Switzerland<
                                               ^
/tmp/ocalfeed.xml:125: parser error : Entity 'uuml' not defined
      <content:encoded>Coat of arms of Pf&auml;ffikon, Z&uuml;rich, Switzerland<
                                                              ^
/tmp/ocalfeed.xml:164: parser error : Entity 'szlig' not defined
      <content:encoded>Wappentier von Gie&szlig;en: roter L&ouml;we mit schwarze
                                                ^
/tmp/ocalfeed.xml:164: parser error : Entity 'ouml' not defined
      <content:encoded>Wappentier von Gie&szlig;en: roter L&ouml;we mit schwarze
                                                                 ^
/tmp/ocalfeed.xml:164: parser error : Entity 'uuml' not defined
tent:encoded>Wappentier von Gie&szlig;en: roter L&ouml;we mit schwarzen Fl&uuml;
                                                                               ^

This constitutes a good share of all ocal searches, so it makes this feature almost useless in Inkscape. It's extremely high priority for Inkscape 0.46.

Unfortunately this is an ocal bug rather than an Inkscape bug; entites like &uuml; etc. are specific to HTML and don't exist in XML. If the OCAL uses them then it needs to define the values for those entities in an internal DTD subset.

For original report, please see: https://bugs.launchpad.net/inkscape/+bug/179326

[Lubomir Rintel (lkundrak)]

The situation is not so tragic or hard to fix or work around. Subtle bugs like this are no good reason to stop processing the document; passing XML_PARSE_RECOVER flag to xmlRead* will do. We don't use the strings with problematic entities at all anyways.

inkscape-0.46pre1-ocal2.patch from bug #191847 does that, passing the flag to xmlReadIO that it uses instead xmlReadFile (for other, security, reason). This is already in Fedora's testing and devel and seems to work well.

[Kees Cook (kees)]

Thanks! This will be in SVN shortly. :)

[Tollef Fog Heen (tfheen)]

Closing all openclipart bugs as openclipart is now on launchpad, as per request from Jon Philips.