[
https://issues.apache.org/jira/browse/JENA-127?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13123422#comment-13123422
]
Andy Seaborne commented on JENA-127:
------------------------------------
The RIOT output framework is less defined; "undefined" would be
accurate. For now, writing into Jena core is best as a Jena writer then
go via a static function like writeRDFJSON(graph) and it'll get adapted
when/if.
Input efficiency is more important because it more directly affects
users experience with large data.
Rob - the tokenizer is doing minimal lookahead. Bytes->chars conversion
is done in large chunks (by the Java library - I tried to short circuit
it with a non-codepoint-checking version but it was not faster).
The tokens are quite simple - only one character look ahead is need
except in the case of prefix names [*]. There's non geneal rexexps
going on so it should be fast and I've profiled it heavily for N-triples
and others - don't see any obvious hot spots or inefficiencies.
[*] The end of local name is (CHARS|'.')* CHARS so there is a little
dance to handle ".". Does not affect RDF/JSON, and it's only at most
one charactser pushback code, not general backtracking.
> Add RDF/JSON Parsing Support to RIOT
> ------------------------------------
>
> Key: JENA-127
> URL: https://issues.apache.org/jira/browse/JENA-127
> Project: Jena
> Issue Type: New Feature
> Components: ARQ, Jena, RIOT
> Environment: All
> Reporter: Rob Vesse
> Assignee: Paolo Castagna
> Priority: Minor
> Labels: patch, rdf/json, riot
> Attachments: ARQ-RDF-JSON-tests_r1179639.patch,
> ARQ_JENA-127_r1179358.patch, JenaReaderRdfJson.java, LangRDFJSON.java,
> RdfJsonRiotPatch-ApacheSVN.patch, RdfJsonRiotPatch.patch,
> RdfJsonRiotPatch.patch, TestLangRdfJson.java, TestLangRdfJson.java
>
>
> The attached patch provides a RDF/JSON (Talis Specification) parser for RIOT,
> the patch is against ARQ trunk from the Jena SourceForge SVN repository
> It plugs in as an implementation of LangRIOT (named LangRDFJSON) and uses the
> existing TokenizerJSON from the atlas package to do the tokenisation. There
> is also a JenaReaderRdfJson added as part of this patch which does what the
> name suggests.
> I have also included in this patch a set of unit tests which verify the
> parsers behaviour with a variety of valid and invalid inputs.
> There are still some things to be addressed:
> - The patch includes registration of the Jena reader when
> SysRiot.writeIntoJena() is called but does not unregister itself when
> resetJenaReaders() is called, should this be done?
> - Add a RDF/JSON writer - a separate patch will be submitted at a later date
> (likely next week) for this
> Otherwise the patch is fairly comprehensive and I hope can be reviewed and
> included in future releases
> EDIT - I have now redone the patch against Apache SVN as well and attached
> that as a separate file since there are some differences in the structure of
> the two repos and some minor code changes that mean the SourceForge SVN patch
> cannot be applied directly against Apache SVN
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
