FWIW, I have some initial code so I can start testing it against the RDFa testsuite.
I’ve used JSoup since it is a well-maintained library with a nice API
The only really annoying part seems to be the lack of line/column indication when an error occurs.
(I guess I could first use Jsoup to create well-formed X(HT)ML, and then use SAX to iterate over the result,
but it seems to be a bit of an overkill to include a new dependency just to only do tag balancing…)
Technically, attoparser fits the bill (smaller, line/column indication), but there seems to be only 1 maintainer and one other contributor.
Which does not say anything about the quality of the project of course 😊
Best regards
Bart
From: Bart Hanssens (BOSA)
Sent: dinsdag 30 april 2019 9:57
To: rdf4j developer discussions <rdf4j-dev@xxxxxxxxxxx>
Subject: RE: [rdf4j-dev] thoughts on RDFa ?
Hi Håvard,
Well, I’m mainly looking into RDFa because of the (somewhat basic) support for RDFa in Drupal CMS.
We’re running quite a few Drupal-websites, so this could come in handy…
But “perfect syntax” and “website” is a rare combo, so I’ll use Jsoup or attoparser
😊
Best regards
Bart
Hi Bart,
I have not used RDFa for anything. I do know that the metadata in images is rdf, and also that google is pushing for more jsonld on webpages.
My experience with both SAX and jsoup are good. I usually use jsoup when I need to crawl webpages, and for this it is the best library I have used. Very robust and simple to use.
SAX I use in my XmlToRdf converter for performance. I can convert 100 mb of XML to turtle with only 20 mb of ram in less than 2 seconds on my laptop. It even works all the way down to 3 mb of ram, but then the parsing time jumps to around
10 seconds because of GC.
I would recommend SAX for pure XML, perfect syntax, usecases. JAXB for when you want java objects, and jsoup for everything else.
Hi,
for scraping purposes, I'm looking into RDFa/RDFa-Lite and I'm thinking about writing a RIO parser (see also issue #512).
IIRC James did some experimental work on RDFa as well, but I think it was based on SAX,
so probably assuming that the source would be perfectly formatted XHTML... which is rarely the case
So currently I'm looking at using either attoparser (smaller, event-driven) or jsoup (more frequently updated, DOM-interface),
and there is a wonderful test suite available at
http://rdfa.info/test-suite/
So I was wondering
- are there other HTML parser I'd should look into (Jodd Lagarto ? NekoHTML ?)
- where should the testsuite go (if it gets CQ approval): I remember some emails about moving the rdf4j-testsuite back into the main repo, but I'm not sure what the conclusion
was
Thanks
Bart
