Add rubric evaluation engine with crJSON and untrusted-cert support#7
Open
erik-sv wants to merge 3 commits into
Open
Add rubric evaluation engine with crJSON and untrusted-cert support#7erik-sv wants to merge 3 commits into
erik-sv wants to merge 3 commits into
Conversation
Add full rubric evaluation pipeline for the C2PA asset conformance program's composable rubric framework. json-formula-rs: - Add normalize_expression() for bare true/false/null keyword rewriting - Add arg_count() helper and $argN parameterized named expression support in register_expression() with globals injection and save/restore pattern profile-evaluator-rs: - Add evaluate_rubric_conformance() for whole-crJSON conformance evaluation with failIfMatched support and true/false trait bucketing - Add evaluate_rubric_signals() for per-manifest signal detection with inception/transformation grouping, ingredient index resolution, assertedBy extraction, and mimeType derivation - Support both report_text (profiles) and reportText (rubrics) field names - 36 golden fixture tests matching upstream Python reference evaluator output (1 documented deviation: startsWith array projection in ii2i) c2pa-validate CLI: - Add -rubric, -rubric-dir, -rubric-mode (conformance/signals), -emit-crjson, -crjson, -rubric-strict flags - Remove crJSON evaluation bail that blocked rubric eval on crJSON inputs - Add rubric_results to CrJsonValidationReport for structured output Test fixtures: - 5 rubric YAML files from c2pa-org/conformance PR #324 - 18 golden test scenarios (54 files) from upstream test suite
When --rubric is used with binary assets, the CLI now extracts crJSON and runs rubric evaluation even when trust verification fails. This supports the conformance program onboarding workflow where products use self-signed certificates before receiving program-issued certs. The pipeline attempts trust verification first, then falls back to reading with verify_trust disabled. The crJSON validationResults still reflect the untrusted state, so the trusted_success trait fails as expected while all other conformance traits can be evaluated. Also fix structured JSON output for CrJsonValidation report items so rubric results are properly serialized instead of null.
…andling Move $argN injection and bare-keyword normalization from json-formula-rs to profile-evaluator-rs. This keeps json-formula-rs upstream-compatible while supporting parameterized named expressions in rubric evaluation. json-formula-rs retains only register_function() and globals_mut() as additions over upstream. All expression preprocessing now runs in the evaluator layer at registration time.
This was referenced May 5, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
The conformance tool can extract crJSON from binary assets but has no way to evaluate that crJSON against a rubric. This PR adds a rubric evaluation engine with two modes: conformance (whole-document) and signals (per-manifest), plus support for evaluating pre-existing crJSON files directly.
Three capabilities:
Rubric evaluation (
--rubric,--rubric-dir,--rubric-mode). Evaluate assets or crJSON against YAML rubric files. Conformance mode checks whole-document traits (spec version, signature, trust). Signals mode iterates per-manifest to detect provenance patterns across ingredient chains.crJSON read path (
--crjson). Treat input files as pre-existing crJSON rather than binary assets. This supports preliminary conformance testing for formats not yet supported by c2pa-rs - the submitter extracts crJSON through their own tooling, then evaluates manifest-level conformance without requiring native container support.Untrusted certificate fallback. When
--rubricis used with binary assets, the CLI attempts trust verification first, then falls back to reading withverify_trustdisabled. Conformance program applicants using self-signed certificates can evaluate all rubric traits excepttrusted_successwithout requiring a program-issued certificate. The crJSONvalidationResultsstill reflects the untrusted state.Includes five rubric YAML files (conformance 0.1 and 0.2 for spec 2.2, conformance 0.2 for spec 2.4, integrity, and signals) with golden-file tests covering 14 provenance scenarios.
Also updates
json-formula-rsandprofile-evaluator-rssubmodules. Expression parameter injection ($argN) and bare-keyword normalization (true/false/nullto function form) now run in the evaluator layer rather than the formula engine, keepingjson-formula-rsupstream-compatible.Usage
Test plan
cargo testpasses for all new integration tests (crjson_validation, asset_validation, schema_validation)--crjsonmode correctly evaluates pre-existing crJSON without attempting binary asset read--emit-crjsonextracts crJSON from binary assets without rubric evaluation