[JENA-1263] Configure HTTP client to follow 303 redirects

• Type: Improvement
• Status: Closed
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: Jena 3.1.1
• Fix Version/s: Jena 3.2.0
• Component/s: ARQ
• Assignee: Adam Soroka
• Reporter: Aaron Coburn
• Created: 17/Nov/16 15:13
• Updated: 04/Dec/21 13:07
• Resolved: 08/Dec/16 14:25

Description

When calling RDFDataMgr.read(Model model, String uri), the underlying HTTP client does not appear to follow 303 redirects. For example:

Model m = createDefaultModel();
RDFDataMgr.read(m, "http://purl.org/dc/terms/");

org.apache.jena.riot.RiotException: Failed to determine the content type: (URI=http://purl.org/dc/terms/ : stream=text/html)

A work-around is to add a static block with a custom HTTP client like so:

static {
    HttpOp.setDefaultHttpClient(
            HttpClientBuilder.create().setRedirectStrategy(
                    new LaxRedirectStrategy()).build());
}

By default the Apache HTTP client follows 301 and 302 redirects (but not 303 redirects), but the W3C recommends using 303 redirects for publishing RDF vocabularies (https://www.w3.org/TR/swbp-vocab-pub/), which is what the Dublin Core vocabularies use.

This sort of redirect handling worked previously, e.g. Jena 3.1.0; it would be convenient if the underlying HTTP client simply followed the 303 redirects.

Issue Links

• relates to: JENA-2203 303 redirect issues cropped up again in 4.3 snapshot
• links to: GitHub Pull Request #194