Skip to content

Fall back to un-optimized filter chain for unknown filters#1284

Open
hs-lsong wants to merge 5 commits intomasterfrom
local-dt-test
Open

Fall back to un-optimized filter chain for unknown filters#1284
hs-lsong wants to merge 5 commits intomasterfrom
local-dt-test

Conversation

@hs-lsong
Copy link
Collaborator

@hs-lsong hs-lsong commented Feb 5, 2026

Fall back to un-optimized filter chain for unknown filters

Summary

  • When AstFilterChain optimization is enabled and a filter chain contains an unknown filter (e.g. local_dt), fall back to the standard nested AstMethod evaluation at parse time
  • Previously, the optimized path would report an "Unknown filter" error and return null, aborting the entire chain. Now it builds the un-optimized AST instead, matching the behavior of the non-optimized path

Why

The optimized AstFilterChain handles unknown filters differently from the standard EL evaluation path. For example, module | local_dt|unixtimestamp | pprint | md5 would produce no output with the optimization enabled, but works (with a
warning) in the un-optimized path. The optimization should be transparent — falling back when it can't handle a filter ensures parity.

When AstFilterChain optimization is enabled and a filter chain contains
an unknown filter (e.g. local_dt), fall back to the standard nested
AstMethod evaluation at parse time instead of failing with an
"Unknown filter" error and returning null.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
@hs-lsong hs-lsong marked this pull request as ready for review February 5, 2026 21:45
return null;
}
if (filter == null) {
interpreter.addError(
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This does seem more complicated than it needs to be, and doesn't look to handle disabled filters the same as the non-chained logic.

I think for both instances they will make value null. Could that work instead of building the custom error?

value = null;
continue;

hs-lsong and others added 4 commits February 18, 2026 14:02
…ehavior

The non-chained path in JinjavaInterpreterResolver skips disabled filters
and passes the value through to the next filter. The chained path was
incorrectly returning null for the entire expression, discarding the input
and aborting remaining filters.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The previous fix used `continue` alone, but the non-chained path
propagates null through subsequent filters rather than skipping.
Setting value = null before continuing matches that behavior exactly.

Adds parity test confirming optimized and non-chained paths produce
identical output for disabled filters in a chain.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Unknown filters are now handled the same as disabled filters in
AstFilterChain: set value to null and continue. This matches the
non-chained behavior where null propagates through subsequent filters.

The parse-time hasUnknownFilter check and buildUnoptimizedFromSpecs
fallback are no longer needed and are removed.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants

Comments