Update "find_jvm_packages" to include all resources #1921
Conversation
- Ensure extra_data is populated in "find_jvm_packages()" for all resources for mapping purposes - Adjust test logic to update extra_data after map_checksum, matching d2d pipeline sequence Signed-off-by: Chin Yeung Li <[email protected]>
Signed-off-by: Chin Yeung Li <[email protected]>
| resources = ( | ||
| project.codebaseresources.files().no_status().from_codebase().has_no_relation() | ||
| ) | ||
| resources = project.codebaseresources.files().no_status().from_codebase() |
There was a problem hiding this comment.
What issue or problem this change solves ?
There was a problem hiding this comment.
This is for #1854
This update addresses a scenario where both the development and deployment codebases contain the same .java file, but the deployed version also includes the corresponding .class file.
I think the tool relies on the "extra_data" field such as extra_data={"java_package": "org.apache.flume.node"}, (https://github.com/aboutcode-org/scancode.io/blob/main/scanpipe/tests/pipes/test_d2d.py#L402) to perform a mapping from .class to .java. (I may be wrong, but it seems I need this field to trigger the "path" mapping in the test)
However, if a .java file in the development codebase has already been checksum-matched to its counterpart in the deployed codebase, it won't be indexed again, meaning it won’t receive extra_data and won’t be available for mapping.
Since "extra_data" is generated in find_jvm_packages(), this change ensures that all source files in the from_codebase are indexed, even if they’ve already been matched by checksum.
There was a problem hiding this comment.
Makes sense to me, @tdruez please have a look as well. Thanks!
|
@chinyeungli we need unit tests related to those changes. |
Signed-off-by: Chin Yeung Li <[email protected]>
Signed-off-by: Chin Yeung Li [email protected]