Source Links Fix#440
Conversation
License Check Results🚀 The license check job ran with the Bazel command: bazel run --lockfile_mode=error //src:license-checkStatus: Click to expand output |
|
The created documentation from the pull request is available at: docu-html |
There was a problem hiding this comment.
Pull request overview
This PR reworks how “source code links” are generated and injected into Sphinx needs, aiming to support both local builds (derive Git hash from the current repo) and combo/ref-integration builds (use per-module metadata such as repo URL + commit hash).
Changes:
- Add metadata support (module name / repo URL / commit hash) to NeedLink JSON handling and Bazel CLI generation/merge scripts.
- Introduce grouping of needs by module and a new module-grouped cache JSON.
- Update Sphinx extension flow to read the new caches and generate GitHub links using either git-derived or metadata-derived refs.
Reviewed changes
Copilot reviewed 10 out of 10 changed files in this pull request and generated 15 comments.
Show a summary per file
| File | Description |
|---|---|
| src/helper_lib/additional_functions.py | Changes GitHub link generation API to require module metadata. |
| src/extensions/score_source_code_linker/needlinks.py | Extends NeedLink with metadata fields; adds new metadata-aware JSON format/load path. |
| src/extensions/score_source_code_linker/need_source_links.py | Moves group_by_need into this module. |
| src/extensions/score_source_code_linker/module_source_links.py | New module-grouped cache format and grouping logic. |
| src/extensions/score_source_code_linker/metadata.py | New TypedDict + TypeGuard for metadata records. |
| src/extensions/score_source_code_linker/generate_source_code_links_json.py | Refactors extraction helper signature and logging (currently inconsistent call sites). |
| src/extensions/score_source_code_linker/init.py | Adds module-linker stage and changes injection to use module-grouped cache + metadata-based link generation. |
| scripts_bazel/merge_sourcelinks.py | Merges per-module sourcelinks and enriches with known-good repo/hash metadata. |
| scripts_bazel/generate_sourcelinks_cli.py | Emits sourcelinks JSON with a leading metadata dict and uses updated extraction helper signature. |
| docs.bzl | Adds optional known_good wiring into the sourcelinks merge genrule and the public docs() macro. |
You can also share your feedback on Copilot code review. Take the survey.
| store_source_code_links_with_metadata_json( | ||
| file=args.output, metadata=metadata, needlist=all_need_references | ||
| ) |
There was a problem hiding this comment.
This switches the generated JSON format from a plain list of NeedLinks to a list whose first element is a metadata dict. Any existing consumers/tests that expect the old schema will now fail. Consider either keeping the old format as default (with an opt-in flag for metadata), or updating all in-repo consumers and tests in the same PR to avoid a partially-migrated state.
There was a problem hiding this comment.
@a-zw Valid issue here .
Do you think it would be better to rename this specific cache that comes from here a bit so it is clear it is with metadata?
Like 'scl_metadata_cache.json' or whatever? SO that the name makes it clear to use the metadata reader?
| metadata: moduleInfo, | ||
| link: NeedLink | DataForTestLink | DataOfTestCase | None = None, | ||
| ) -> str: | ||
| if link is None: | ||
| link = DefaultNeedLink() | ||
| if not metadata.hash: | ||
| # Local path (//:docs) | ||
| return get_github_link_from_git(link) | ||
| # Ref-Integration path (//:docs_combo..) | ||
| return get_github_link_from_json(metadata, link) |
There was a problem hiding this comment.
get_github_link now requires a metadata argument, but there are existing call sites in the repo that still call get_github_link(link) (e.g. in tests and xml_parser.py). This is a breaking API change that will raise TypeError. Consider keeping backwards compatibility by making metadata optional with a sensible default (e.g. infer local/git behavior when metadata is omitted) or providing a new function name for the metadata-based behavior.
| metadata: moduleInfo, | |
| link: NeedLink | DataForTestLink | DataOfTestCase | None = None, | |
| ) -> str: | |
| if link is None: | |
| link = DefaultNeedLink() | |
| if not metadata.hash: | |
| # Local path (//:docs) | |
| return get_github_link_from_git(link) | |
| # Ref-Integration path (//:docs_combo..) | |
| return get_github_link_from_json(metadata, link) | |
| metadata: moduleInfo | |
| | NeedLink | |
| | DataForTestLink | |
| | DataOfTestCase | |
| | None = None, | |
| link: NeedLink | DataForTestLink | DataOfTestCase | None = None, | |
| ) -> str: | |
| """ | |
| Return a GitHub link for the given location. | |
| Backwards compatible calling conventions: | |
| - New style: get_github_link(metadata, link) | |
| - Old style: get_github_link(link) # metadata omitted | |
| """ | |
| # Distinguish between new-style and old-style calls. | |
| if isinstance(metadata, moduleInfo): | |
| actual_metadata: moduleInfo | None = metadata | |
| actual_link = link | |
| else: | |
| # Old-style: first argument is actually the link (or None) | |
| actual_metadata = None | |
| # If both metadata and link are provided but metadata is not a moduleInfo, | |
| # prefer the explicitly provided link argument. | |
| actual_link = link if link is not None else metadata | |
| if actual_link is None: | |
| actual_link = DefaultNeedLink() | |
| # If no metadata is available or metadata.hash is falsy, fall back to git-based link. | |
| if actual_metadata is None or not getattr(actual_metadata, "hash", None): | |
| # Local path (//:docs) | |
| return get_github_link_from_git(actual_link) | |
| # Ref-Integration path (//:docs_combo..) | |
| return get_github_link_from_json(actual_metadata, actual_link) |
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 9 out of 9 changed files in this pull request and generated 8 comments.
You can also share your feedback on Copilot code review. Take the survey.
src/extensions/score_source_code_linker/generate_source_code_links_json.py
Outdated
Show resolved
Hide resolved
|
Just realized that testlinks also need the module name earlier than source_code_linker extension end state. I need to re-think this approach a bit and see what I can adapt to make this more convenient. I think I have an idea, but unsure. |
|
Still a bit to do here, but the architecture is 'working', though not pretty. |
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 18 out of 18 changed files in this pull request and generated 14 comments.
You can also share your feedback on Copilot code review. Take the survey.
src/extensions/score_source_code_linker/generate_source_code_links_json.py
Outdated
Show resolved
Hide resolved
src/extensions/score_source_code_linker/generate_source_code_links_json.py
Show resolved
Hide resolved
| logger = logging.getLogger(__name__) | ||
|
|
||
|
|
||
| def clean_external_prefix(path: Path) -> Path: |
There was a problem hiding this comment.
Why is this still here? With our new design, this should not be necessary.
There was a problem hiding this comment.
This is neccessary. Otherwise the filename will be wrong that is passed to the extraction function.
| return f"{base_url}/blob/{current_hash}/{link.file}#L{link.line}" | ||
|
|
||
|
|
||
| def parse_module_name_from_path(path: Path) -> str: |
There was a problem hiding this comment.
We should not need this anymore.
There was a problem hiding this comment.
We also need this.
The only place we get the actual model this sourcelink file belongs to is via the path.
And we need that to make the correct json in the end.
Is there a different idea that I missed where we would have the module somehow somewhere else?
| "url": "", | ||
|
|
||
| Else it will parse the module_name e.g. `score_docs_as_code` | ||
| match this in the known_good_json and grab the accompanying |
There was a problem hiding this comment.
This describes the old way of doing it. Why is it still here?
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 29 out of 29 changed files in this pull request and generated 6 comments.
You can also share your feedback on Copilot code review. Take the survey.
| known_good_json = os.environ.get("KNOWN_GOOD_JSON") | ||
| module_name = parse_module_name_from_path(filepath) | ||
| md: MetaData = { | ||
| "module_name": module_name, | ||
| "hash": "", | ||
| "url": "", | ||
| } | ||
| if module_name != "local_module" and known_good_json: | ||
| md["hash"], md["url"] = parse_info_from_known_good( | ||
| Path(known_good_json), module_name | ||
| ) | ||
| return md |
pyproject.toml
Outdated
| ".venv*/**", | ||
| ] | ||
| [tool.pytest.ini_options] | ||
| addopts = ["-v", "-s"] |
src/extensions/score_source_code_linker/generate_source_code_links_json.py
Outdated
Show resolved
Hide resolved
| assert result == expected | ||
|
|
||
|
|
||
| # Done to appease the LSP Gods |
There was a problem hiding this comment.
Change comment:
| # Done to appease the LSP Gods | |
| # Done to ensure type hinting throughout. |
| test_type="requirements-based", | ||
| derivation_technique="requirements-analysis", | ||
| ) | ||
| def test_get_github_link_with_real_repo(git_repo: Path) -> None: |
There was a problem hiding this comment.
This test seems correct & useful.
Must have been accidentally delted:
=> reinstate this.
| @@ -1,5 +1,5 @@ | |||
| # ******************************************************************************* | |||
There was a problem hiding this comment.
Deleted all / most property decorators as I wanted to understand how they should be used (content wise).
Once this is understood, will reinstate & add decorators to all test functions where applicable.
|
|
||
| # ─────────────────[ NeedLink Dataclass Tests ]───────────────── | ||
|
|
||
|
|
There was a problem hiding this comment.
As we 'custom made' these two functions and rely on them working correctly during encoding / decoding I figured we should probably test them.
|
|
||
|
|
||
| def test_store_and_load_source_code_links(tmp_path: Path): | ||
| """Happy path: Store and load without metadata""" |
There was a problem hiding this comment.
Probably should add an actual comparision of the two jsons in every roundtrip tests that was made.
To fully test if saved => loaded
| module_name="mod", | ||
| url="url", | ||
| hash="hash", |
There was a problem hiding this comment.
Double check if it would be nicer in implementation to have the default be 'local_module' instead.
Incremental was passed into every build target regardless if it was correct or not
5673faa to
8c7aee6
Compare
|
Just for informations sake . PR was reviewed multiple times with @a-zw , the latest changes were gone over in a call to ensure understanding and clearing up questions etc. |
src/extensions/score_source_code_linker/tests/test_repo_source_link_integration.py
Outdated
Show resolved
Hide resolved
Fixing assertion Fixing wrong split in path Fixing pytest global arguments
|
Consumer tests are failing with the expected issue only:
This issue is hinged upon this PR being merged: eclipse-score/tooling#130 Once that is the case and a git override is added to the newest main commit of tooling, I would expect all consumer tests to pass without issue |
|
I addressed & fixed the copilot issues as well (those that were actual issues). It should now be good to go from my side. |
6763560
into
eclipse-score:main




Changes the working of source_code_linker.
It now correctly identifies the repository the source link & test links came from, either in local or combo build.
Should redirect to the correct place.
=> KNOWN_GOOD_JSON is required in combo builds now
=> In combo builds it will take the URL from the KNOWN_GOOD_JSON.
Done:
📌 Description
🚨 Impact Analysis
✅ Checklist