kefo
kefo

Reputation: 43

SVC-CODEPOINT error loading triples to MarkLogic

When trying to load an ntriples file with about 7M triples in it, I get the following error from the REST-API:

{"errorResponse":{"statusCode":500, "status":"Internal Server Error", 
"messageCode":"INTERNAL ERROR", 
"message":"SVC-CODEPOINT: (err:FOCH0001) xdmp:nquad($body, $options) -- Codepoint not legal . See the MarkLogic server error log for further detail."}}

The error log is not more helpful here:

2021-01-13 14:24:53.771 Info: Status 500: SVC-CODEPOINT: (err:FOCH0001) xdmp:nquad($body, $options) -- Codepoint not legal

I have run the file through rapper and Jena's riot validation and neither tool kicks up a problem with the file, or any specific triples.

How might I troubleshoot this with MarkLogic?

I can generate a file that employs a different serialization, but that has me rewriting my code when this has worked in the past and works on two other files in the batch. And a different serialization might work or, if it doesn't, may generate the same error. In either case, I've not actually identified the issue. There must be something in the file, but I'd like to find it because we also happen to be the source of the ntriples.

Upvotes: 2

Views: 56

Answers (1)

kefo
kefo

Reputation: 43

I could delete the question, but if it helps someone...

Convert the ntriples to RDF/XML and load that file. MarkLogic will spit out a more valuable error message:

2021-01-13 16:10:43.292 Info: Status 400: XDMP-DOCXMLCHAR: xdmp:get-request-body("xml") -- 
Invalid XML character codepoint 55358 at  line 14398744 -- document contains non-XML character

Upvotes: 2

Related Questions