[JENA-1161] riot cmdline uses wrong base when parsing Turtle over http

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: Jena 3.0.1
• Fix Version/s: Jena 3.1.0
• Component/s: Cmd line tools
• Assignee: Andy Seaborne
• Reporter: Stian Soiland-Reyes
• Created: 07/Apr/16 15:04
• Updated: 14/May/16 09:51
• Resolved: 07/Apr/16 21:39

Description

Parsing a Turtle file served over http:// or https:// which has got no @base, and uses relative IRI references, wrongly uses the current directory in file:/// as a base.

The command line

stain@biggie:/tmp$ riot https://cdn.rawgit.com/stain/3d49908d790c5678faee302ba17f4a43/raw/bceb24ee6bfcaa60c4508779bc5f09ec367876f1/nobase.ttl

where that URL returns Content-Type: text/turtle;charset=utf-8 with the body:

@prefix : <#> .
<> :a <#test> .

is wrongly parsed by the riot command line tool to be relative to the current directory:

<file:///tmp/https:/cdn.rawgit.com/stain/3d49908d790c5678faee302ba17f4a43/raw/bceb24ee6bfcaa60c4508779bc5f09ec367876f1/nobase.ttl> <file:///tmp/https:/cdn.rawgit.com/stain/3d49908d790c5678faee302ba17f4a43/raw/bceb24ee6bfcaa60c4508779bc5f09ec367876f1/nobase.ttl#a> <file:///tmp/https:/cdn.rawgit.com/stain/3d49908d790c5678faee302ba17f4a43/raw/bceb24ee6bfcaa60c4508779bc5f09ec367876f1/nobase.ttl#test> .

The expected output would be the same as supplying the same URI as a --base:

stain@biggie:/tmp$ riot --base https://cdn.rawgit.com/stain/3d49908d790c5678faee302ba17f4a43/raw/bceb24ee6bfcaa60c4508779bc5f09ec367876f1/nobase.ttl https://cdn.rawgit.com/stain/3d49908d790c5678faee302ba17f4a43/raw/bceb24ee6bfcaa60c4508779bc5f09ec367876f1/nobase.ttl
<https://cdn.rawgit.com/stain/3d49908d790c5678faee302ba17f4a43/raw/bceb24ee6bfcaa60c4508779bc5f09ec367876f1/nobase.ttl> <https://cdn.rawgit.com/stain/3d49908d790c5678faee302ba17f4a43/raw/bceb24ee6bfcaa60c4508779bc5f09ec367876f1/nobase.ttl#a> <https://cdn.rawgit.com/stain/3d49908d790c5678faee302ba17f4a43/raw/bceb24ee6bfcaa60c4508779bc5f09ec367876f1/nobase.ttl#test> .

(except if a Content-Location header is provided, or HTTP redirection has been followed - in which case the result of that should be used as base)

Relevant specs:

https://www.w3.org/TR/turtle/#sec-iri-references
https://www.ietf.org/rfc/rfc3986 section 5.1 and 5.2