fix: mark malformed Keras H5 configs inconclusive#917
Open
mldangelo-oai wants to merge 2 commits intomainfrom
Open
fix: mark malformed Keras H5 configs inconclusive#917mldangelo-oai wants to merge 2 commits intomainfrom
mldangelo-oai wants to merge 2 commits intomainfrom
Conversation
Contributor
WalkthroughKeras H5 scanner now treats malformed or unparseable Changes
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~25 minutes Poem
🚥 Pre-merge checks | ✅ 3✅ Passed checks (3 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Contributor
Performance BenchmarksCompared
|
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 3
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@modelaudit/scanners/keras_h5_scanner.py`:
- Around line 523-537: The code only validates layer_config type when
layer_class is in self.suspicious_layer_types, which lets malformed wrappers
(e.g., TimeDistributed with config as a string) bypass checks; move or duplicate
the dict-type validation for layer_config so every layer runs the check before
any further inspection: for each layer obtained in the scanner, call
layer_config = layer.get("config", {}), validate isinstance(layer_config, dict)
and on failure call self._mark_inconclusive_scan_result(result,
"keras_h5_layer_config_invalid_type") and result.add_check(...) (same payload
currently used), then set layer_config = {} and continue; keep subsequent logic
that inspects wrapped/inner layers (and uses self.suspicious_layer_types)
unchanged but rely on the earlier universal validation to prevent silent skips
(reference symbols: layer_config, layer.get, self.suspicious_layer_types,
_mark_inconclusive_scan_result, result.add_check, self.current_file_path).
In `@tests/scanners/test_keras_h5_scanner.py`:
- Around line 294-309: The helper _assert_inconclusive_keras_h5_scan currently
only asserts the check name and status; update it to also assert the specific
"fail-closed" message appears where expected (e.g., in result.checks'
message/text field and in metadata["scan_outcome_reasons"] or
audit_result.file_metadata[str(model_path)]), so that the test verifies the
operational fail-closed message alongside success/exit-code/cache behavior;
locate checks via KerasH5Scanner().scan(...) and audit_result from
scan_model_directory_or_file(...) and add assertions that the check with name ==
expected_check_name has the expected fail-closed message and that the same
message is present in the metadata scan_outcome_reasons.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Repository UI
Review profile: ASSERTIVE
Plan: Pro
Run ID: 9678ece1-c35d-47d1-a591-8a2b54199d64
📒 Files selected for processing (3)
CHANGELOG.mdmodelaudit/scanners/keras_h5_scanner.pytests/scanners/test_keras_h5_scanner.py
Contributor
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@modelaudit/scanners/keras_h5_scanner.py`:
- Around line 734-749: The nested-container branch only runs when "config" in
layer, so layers with a container class (self._MODEL_CONTAINER_CLASSES) but
missing "config" (e.g. {"class_name":"Functional"}) silently pass; update the
logic in the block that handles layer_class in self._MODEL_CONTAINER_CLASSES to
explicitly handle the case where "config" is absent by calling
self._mark_inconclusive_scan_result(result,
"keras_h5_nested_model_layers_missing") and adding the same
result.add_check(...) entry (use the same name "Nested Model Layers Presence
Validation", passed=False, rule_code="S902", severity=IssueSeverity.INFO,
location=self.current_file_path and details={"layer_class": layer_class,
"expected_key":"config.layers"}) so malformed nested containers are marked
inconclusive rather than scanning clean.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Repository UI
Review profile: ASSERTIVE
Plan: Pro
Run ID: 521d20e8-8aeb-4d40-9e8c-06da88e9e6fe
📒 Files selected for processing (2)
modelaudit/scanners/keras_h5_scanner.pytests/scanners/test_keras_h5_scanner.py
Comment on lines
+734
to
749
| if layer_class in self._MODEL_CONTAINER_CLASSES and "config" in layer: | ||
| nested_config = layer.get("config") | ||
| if isinstance(nested_config, dict) and "layers" in nested_config: | ||
| self._scan_model_config(layer, result) | ||
| elif isinstance(nested_config, dict): | ||
| self._mark_inconclusive_scan_result(result, "keras_h5_nested_model_layers_missing") | ||
| result.add_check( | ||
| name="Nested Model Layers Presence Validation", | ||
| passed=False, | ||
| message=f"Nested model config for {layer_class} is missing required layers list", | ||
| rule_code="S902", | ||
| severity=IssueSeverity.INFO, | ||
| location=self.current_file_path, | ||
| details={"layer_class": layer_class, "expected_key": "config.layers"}, | ||
| ) | ||
|
|
Contributor
There was a problem hiding this comment.
Nested container layers without config can still scan clean.
At Line 734, nested container validation only runs when "config" in layer. A malformed nested layer like {"class_name": "Functional"} skips this block and does not get marked inconclusive, which reopens a false-clean path.
🔧 Suggested fix
- if layer_class in self._MODEL_CONTAINER_CLASSES and "config" in layer:
- nested_config = layer.get("config")
- if isinstance(nested_config, dict) and "layers" in nested_config:
- self._scan_model_config(layer, result)
- elif isinstance(nested_config, dict):
+ if layer_class in self._MODEL_CONTAINER_CLASSES:
+ nested_config = layer.get("config")
+ if isinstance(nested_config, dict) and "layers" in nested_config:
+ self._scan_model_config(layer, result)
+ else:
self._mark_inconclusive_scan_result(result, "keras_h5_nested_model_layers_missing")
result.add_check(
name="Nested Model Layers Presence Validation",
passed=False,
message=f"Nested model config for {layer_class} is missing required layers list",
rule_code="S902",
severity=IssueSeverity.INFO,
location=self.current_file_path,
details={"layer_class": layer_class, "expected_key": "config.layers"},
)Based on learnings: "When a scan intentionally fails closed, make behavior operationally explicit and test the message plus success, exit-code, and cache semantics".
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@modelaudit/scanners/keras_h5_scanner.py` around lines 734 - 749, The
nested-container branch only runs when "config" in layer, so layers with a
container class (self._MODEL_CONTAINER_CLASSES) but missing "config" (e.g.
{"class_name":"Functional"}) silently pass; update the logic in the block that
handles layer_class in self._MODEL_CONTAINER_CLASSES to explicitly handle the
case where "config" is absent by calling
self._mark_inconclusive_scan_result(result,
"keras_h5_nested_model_layers_missing") and adding the same
result.add_check(...) entry (use the same name "Nested Model Layers Presence
Validation", passed=False, rule_code="S902", severity=IssueSeverity.INFO,
location=self.current_file_path and details={"layer_class": layer_class,
"expected_key":"config.layers"}) so malformed nested containers are marked
inconclusive rather than scanning clean.
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
Validation
Summary by CodeRabbit
Bug Fixes
Tests