ingest constructs illegal mulgara queries if empty elements exist

failed mulgara query on ingest of pgen e10042

article package for pgen e10042

Replying to russ:

it looks like something in the XML, probably around references, is causing a variable expansion to fail, resulting in ingest trying to insert invalid triples such as:

<info:doi/10.1371/journal.pgen.0010042/bibliographicCitation> <http://rdf.plos.org/RDF/hasEditorList> $bn_w383aab2aaac14a
$bn_w383aab2aaac14a <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/1999/02/22-rdf-syntax-ns#Seq>
$bn_w383aab2aaac14a <http://www.w3.org/1999/02/22-rdf-syntax-ns#_1> <info:doi/10.1371/profile/30255e57-c8f6-428b-b60e-70be2d4d9379>

There's nothing wrong with these inserts - they're perfectly valid itql. The $bn_foo are blank-nodes.

i did a query to see what hasEditorList is supposed to look like and got this:

select $s $p $o
from <local:///topazproject#ri>
where $s $p $o
and $p <mulgara:is> <http://rdf.plos.org/RDF/hasEditorList>

<solution>
      <s resource="info:doi/10.1371/journal.pgen.0030029/reference#pgen-0030029-b001"/>
      <p resource="http://rdf.plos.org/RDF/hasEditorList"/>
      <o blank-node="_node2284559"/>
</solution>

Yup, this looks correct too.

So, what is the error you were getting?

i got a similar error on a file that had an empty <caption> element in the <fig> element for one of the figures.

removing the empty caption tag allowed the article to ingest.

i see with 10042 that one of the references has an empty source tag, i'm going to see if removing that will fix.

looks like ingest is getting confused with empty tags and is generating invalid mulgara. i uploaded the entire mulgara query that caused the error - can you see anything else wrong with it?

i'll grab the entire error for the failed 10042 ingest in a sec...

indeed, removing the empty <source/> tag from reference 58 resolved the problem.

these empty tags are valid per DTD (although it's odd semantically) so we might want to handle them without an error as a low priorty.

i'm uploading the entire error we got with the empty <source/> tag, with the actual mulgara query snipped out since it's already uploaded to this ticket.

error trace on ingest with empty source tag in reference 58

Thanks for the info. I see it now: the empty source tag translates to an empty title in the rdf/xml:

<rdf:Description rdf:about="info:doi/10.1371/journal.pgen.0010042/reference#pgen-0010042-b58">
  <dc:title rdf:datatype="http://www.w3.org/1999/02/22-rdf-syntax-ns#XMLLiteral"></dc:title>
</rdf:Description>

This by itself isn't a problem, but the source:head/plos/libs/article-util/src/main/resources/org/plos/article/util/RdfXmlToTriples.xslt script seems to have a bug where it doesn't handle empty elements properly, even though there's explicit code and comments for that case.

(In [4860]) Fix #768: first of all, we use parseType="Literal" instead of converting the content to a string ourselves and setting the datatype to rdf:XMLiteral. The latter causes problems with empty elements because they result in empty literals, and the spec has some really weird rules regarding these when they're a typed literal (see http://www.w3.org/TR/2004/REC-rdf-syntax-grammar-20040210/ section 7.2.16 and 7.2.21, and test014). RdfXmlToTriples? is doing the correct thing here, but not what we want, hence the switch to use parseType="Literal",

Second, RdfXmlToTriples? has been updated to try and preserve (internal) whitespace in literals.