backslash characters in clinicaltrials

[micheldumontier]

Hi everyone,

I was trying to extend the ClinicalTrials.gov bio2rdf conversion script (see jvsoest@ce09d83), and test it before creating a pull request.

Generation of triples runs fine, and creates the clinicaltrials.nq.gz file. However some URIs in this file are containing strange backslash characters at random locations (see example below, "clnicaltrials_voca\bulary").

Does anyone know what could cause this issue?

Thanks in advance!

Kind regards,
Johan

Example:
http://bio2rdf.org/clinicaltrials_vocabulary:Clinical-Study http://bio2rdf.org/bio2rdf_vocabulary:namespace "clinicaltrials_voca\bulary"^^http://www.w3.org/2001/XMLSchema#string http://bio2rdf.org/clinicaltrials_resource:bio2rdf.dataset.clinicaltrials.R4 .

[micheldumontier]

what platform are you using? can you share a (smallish) generated file?

[jvsoest]

Hi Michel,

I'm trying this on Ubuntu 16.04, with PHP version 7.0.15.
I've attached the conversion for one trial, where it sometimes escapes the "b" character.
clinicaltrials.nq.gz

Maybe it's in the specialEscape function of rdfapi, I'll give it a try as well.

Regards,
Johan

[jvsoest]

As suspected, specialEscape and safeLiteral in rdfapi.php did the trick for me (currently). I've removed "\b" (first two characters) from the addcslashes function. I think it handles the b as a character, instead of \b as backspace. But I'm not sure if this creates problems for other bio2rdf scripts.

[micheldumontier]

ok, i replaced the string definition with a hex list. see micheldumontier/php-lib@252b8ef