Fix Accept header value sent by document()
This is related to #4748 but a different issue.
As far as I can see, content negotiation is broken as on
document() Saxon-JS makes HTTP requests using an
Accept: */* header. This is a catch-all value that can lead to any media type being returned.
I think there should be an explicit preference for XML formats, which can be achieved using weights.
I would suggest such a header value:
Accept: text/xml, application/xml;q=0.9, application/xhtml+xml;q=0.8, */*;q=0.7. It might be possible to improve/generalize it further. Ideally it should be configurable.
Updated by Debbie Lockett about 1 year ago
- Status changed from New to In Progress
- Assignee set to Debbie Lockett
I believe you are running in the browser rather than on Node.js?
In the browser, for asynchronous loading, you generally will be using
ixsl:schedule-action/@document to load a document before a call to
document(). However there is no mechanism for specifying content negotiation here. (Updating
ixsl:schedule-action to allow this would be complicated; and we can't just amend the internal implementation as suggested because we do want to use this mechanism for loading a variety of file types.)
Another option (available on Node.js as well as the browser) is to use
SaxonJS.getResource, and then supply the preloaded resources for the
SaxonJS.transform using the
textResourcePool) option. Feature #4748: Option to supply Accept header in SaxonJS.getResource() should allow the Accept header to be supplied; but I'm reopening that issue, because it looks like the fix was incomplete.
In the meantime, could you rework your code to use
ixsl:schedule-action/@http-request instead? Here you can specify the full HTTP request, including the headers. I guess a problem might be that the response document will not be in the standard document pool, so you won't be able to just use
document() as before. It is just returned in the
body of the HTTP response map. (See https://www.saxonica.com/saxon-js/documentation/index.html#!development/http)
Updated by Martynas Jusevicius about 1 year ago
Yes running in the browser.
This issue mostly concerns templates shared between the server-side (Saxon-HE) and the client-side (Saxon-JS). Where Saxon-HE succeeds loading XML using
document(), Saxon-JS fails due to the underspecified
ixsl:schedule-action is not applicable if the template has to work on the server-side; the
documentPool helps but some URLs might not be known in advance (at Saxon-JS initialization time).
Would it be hard to change the default
Accept header value? :) Or set it to the same that Saxon-HE sends?
Updated by Debbie Lockett 11 months ago
- Category set to API
- Status changed from In Progress to Resolved
- Fix Committed on JS Branch 2 added
Code updated in BrowserPlatform.js (in
readFile methods), to update the default
Accept header as suggested. We now use
Accept: application/xml, text/xml;q=0.9, application/xhtml+xml;q=0.8, */*;q=0.7.
Browser tests API/document01 and document02 added (testing both synchronous and asynchronous document fetching).
Please register to edit this issue