[Dev6] fix XSLT imports by line-o · Pull Request #6544 · eXist-db/exist

line-o · 2026-07-04T12:56:24Z

Description:

Fixes several long-standing issues with imports and URI resolution in fn:transform

Reference:

Type of tests:

XQSuite tests added

In order to build develop-6.x.x we need to downgrade IzPack to the latest version that does not have the transitive dependency on org.pushing-pixels:radiance-theming(-extras) v7.5.0 Because that requires Java 9 to compile. This is likely an unintentional change or bug upstream.

Passing persistent nodes from XQuery to fn:transform as parameters works now. - dereference NodeProxy objects in Convert.of(Sequence) - match both persistent and in-memory Documents in Convert.ofNode(Node) Backport of fddbfd3

refs eXist-db#6065 With this PR applied the shared document builder from the parent XqueryContext is never used. As a result two calls to fn:transform within the same context no longer corrupt each others return values. The actual root-cause of what is described in the linked gist: https://gist.github.com/line-o/4731ed367f437be9cd036b152a18d340

fn:transform did not resolve relative URIs against the database: relative xsl:include/xsl:import hrefs in stored stylesheets, and a relative stylesheet-location, would not resolve the way fn:doc and transform:transform already do. In particular a declared base-uri pointing at a collection without a trailing slash could lose its last path segment during resolution. - Add Transform.newFnTransformURIResolver, leveraging the existing XsltURIResolverHelper.getXsltURIResolver. It is installed in two places: on the XsltCompiler (resolves relative xsl:include/ xsl:import at compile time) and on the Xslt30Transformer's controller (resolves runtime lookups such as fn:doc()/document() inside the stylesheet). - Transform.databaseBaseURI derives the base to resolve against: the stylesheet's own resolved base URI if it's a database location (isDatabaseURI), otherwise falling back to the querying module's location in the database when the stylesheet was built in-memory (stylesheet-node/stylesheet-text with no explicit stylesheet-base-uri). - New error path: stylesheet-text with a relative href and no stylesheet-base-uri now raises XTSE0165 instead of resolving incorrectly. - Options.resolveStylesheetLocation now encapsulates discerning relative/absolute/external stylesheet-location values and returns the Tuple directly, rather than that logic living in Options.getStylesheet. XQSuite tests added in fnTransform5052.xqm. refs eXist-db#5052 NOTE: This fixes the relative imports in XSLT for develop-6; the issue will be closed once the patch is forward-ported to develop.

dizzzz

LGTM .

duncdrum · 2026-07-04T20:03:52Z

@line-o we discovered a bug in izpack 5.2.3 during the beta release, not present in 5.2.2 if you have to downgrade we might want to go down two versions.

I m a bit at a loss how this pr relates to the in-flight upstream PRs touching upon this code area.

dizzzz · 2026-07-04T21:20:31Z

-        <izpack.version>5.2.4</izpack.version>
+        <!-- NOTE: IzPack 5.2.4 pulls in radiance-theming 7.5.0 which requires Java 9+,
+             5.2.3 is the last release whose dependencies are all Java 8 compatible -->
+        <izpack.version>5.2.3</izpack.version>


@reinhapa ?

I will open an issue upstream

line-o · 2026-07-05T05:31:58Z

I m a bit at a loss how this pr relates to the in-flight upstream PRs touching upon this code area.

@duncdrum

This targets develop-6.x.x to and aims to fix fn:transform for that branch
Do you have a particular PR in mind I need to look at?

line-o · 2026-07-05T06:06:18Z

exist 6.4.1 was released with Izpack 5.2.3 -> I think you are referring to 5.2.5

duncdrum · 2026-07-05T07:23:51Z

I mean #4957 and #6530. I mixed up izpack versions, 5.2.3 should work.

line-o · 2026-07-05T11:29:29Z

I also found your PRs #6530 and #6531 - I will also check #4957 but this is likely stale

line-o · 2026-07-05T11:30:06Z

I will base my forward port on #6531 @duncdrum alright?

line-o · 2026-07-05T11:37:11Z

#6531 adresses

✅ Compile-time URI resolution for xsl:import/xsl:include via URIResolution.CompileTimeURIResolver
✅ Saxon catalog resolver integration

Missing from #6531

❌ Runtime URI resolution for document() calls within the XSLT (at transform execution time)
❌ Fixes to Delivery.java for document builder handling
❌ Fixes to Convert.java for persistent node handling

What I will closely look at is the URI resolution logic in Options.java.

duncdrum · 2026-07-05T12:05:11Z

@line-o however you want to thread the needle, the forward port will likely touch #5682 in addition to #5051.

… builder isolation Fixes eXist-db#5052 and eXist-db#6065 by porting key changes from develop-6.x.x: 1. Document builder isolation (Delivery.java): - Always use fresh MemTreeBuilder for transform results - Prevents corruption when multiple fn:transform calls occur in same context - Fixes issue eXist-db#6065 where concurrent transforms corrupted each other's output 2. Persistent node handling (Convert.java): - Handle persistent NodeProxy nodes by dereferencing to underlying DOM - Enables passing database-stored nodes as stylesheet parameters - Checks Document interface instead of specific implementation 3. Enhanced stylesheet location resolution (Options.java): - Return Tuple2<String, Source> to track actual resolved location - Sets SystemId on DOMSource for proper relative URI resolution - Fallback resolution for relative URIs using RFC 3986 + database resolution 4. Runtime URI resolution (Transform.java): - Install runtime URI resolver on XSLT controller - Enables fn:document() calls within stylesheets to resolve against database - Integrates with existing URIResolution.CompileTimeURIResolver - Adds helper methods: newFnTransformURIResolver(), databaseBaseURI(), isDatabaseURI() - Supports xsl:include/xsl:import resolution for stylesheets stored in database 5. Build configuration: - Downgrade IzPack to 5.2.3 (5.2.4+ requires Java 9+) Test coverage: - fnTransform5052.xqm: Tests relative xsl:include/import and stylesheet-location resolution - fnTransform6065.xqm: Tests persistent nodes as stylesheet parameters Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>

line-o · 2026-07-06T06:25:17Z

Develop port is #6546 @duncdrum let's move that discussion over there

duncdrum

LGTM, one minor nitpick about the inlined xquery lines in fnTransform6065.xq,

duncdrum · 2026-07-06T11:54:00Z

+        "stylesheet-params": map {
+            xs:QName("v"): doc($t6065:collection || "/" || $t6065:source-doc-name)
+        }
+(:        ,"delivery-format": "serialized":)


oh right - leftovers from testing

line-o added 4 commits July 4, 2026 09:57

[bugfix] passing persistent nodes to fn:transform

4fcf7c9

Passing persistent nodes from XQuery to fn:transform as parameters works now. - dereference NodeProxy objects in Convert.of(Sequence) - match both persistent and in-memory Documents in Convert.ofNode(Node) Backport of fddbfd3

line-o requested review from a team and duncdrum July 4, 2026 12:56

line-o added xquery issue is related to xquery implementation needs develop port labels Jul 4, 2026

line-o changed the title ~~Dev6 fix xslt imports~~ Dev6 fix XSLT imports Jul 4, 2026

line-o changed the title ~~Dev6 fix XSLT imports~~ [Dev6] fix XSLT imports Jul 4, 2026

line-o requested a review from wolfgangmm July 4, 2026 13:40

line-o self-assigned this Jul 4, 2026

dizzzz approved these changes Jul 4, 2026

View reviewed changes

dizzzz reviewed Jul 4, 2026

View reviewed changes

line-o mentioned this pull request Jul 5, 2026

[bugfix] Forward port PR #6544: fix XSLT imports and document builder isolation #6546

Merged

line-o removed the needs develop port label Jul 6, 2026

duncdrum approved these changes Jul 6, 2026

View reviewed changes

dizzzz merged commit 450f1fb into eXist-db:develop-6.x.x Jul 6, 2026
12 of 13 checks passed

line-o mentioned this pull request Jul 9, 2026

memtree DocumentImpl.getBaseURI() always returns "" when the document element has no xml:base #6566

Open

Uh oh!

Uh oh!

Conversation

line-o commented Jul 4, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description:

Reference:

Type of tests:

Uh oh!

dizzzz left a comment

Choose a reason for hiding this comment

Uh oh!

duncdrum commented Jul 4, 2026

Uh oh!

dizzzz Jul 4, 2026

Choose a reason for hiding this comment

Uh oh!

line-o Jul 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

line-o commented Jul 5, 2026

Uh oh!

line-o commented Jul 5, 2026

Uh oh!

duncdrum commented Jul 5, 2026

Uh oh!

line-o commented Jul 5, 2026

Uh oh!

line-o commented Jul 5, 2026

Uh oh!

line-o commented Jul 5, 2026

Uh oh!

duncdrum commented Jul 5, 2026

Uh oh!

line-o commented Jul 6, 2026

Uh oh!

duncdrum left a comment

Choose a reason for hiding this comment

Uh oh!

duncdrum Jul 6, 2026

Choose a reason for hiding this comment

Uh oh!

line-o Jul 6, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

line-o commented Jul 4, 2026 •

edited

Loading

line-o Jul 5, 2026 •

edited

Loading