2025: Code climate

[stanislaw]

Previous: #1912
Next: TBD

---

General

• confluence_html_table_import.py:129: DeprecationWarning: Call to deprecated method findChildren. (Replaced by find_all) -- Deprecated since version 3.0.0.
• Code climate: end2end tests: fix get_requirement deprecation warnings #2217
• code-climate: fix string-literal-comments #2219
• Field Type -> field type.
• code-climate: re-enable some disabled tests under windows #2222
• Remove FREETEXT. backend: remove FREETEXT from Document and Section #2088
• Introduce end-to-end tests for static HTML export #1020
• .is_requirement -> .is_requirement().
• Bug: Parallellization edge case on Windows #2083
• Make SDoc's grammar require STATEMENT/DESCRIPTION/CONTENT to have REQUIRED: True.
• 01_main_page_serves_304 end2end test does not work on macOS GitHub Actions.
• WARNING: Current configuration will not reload as not all conditions are met, please refer to documentation..
• Code climate: Some/all source coverage end2end tests include the files from build/ folder #1468
• 1 - 667 | None for non-language-aware file links. fix(backend/sdoc_source_code): general reader: always set function marker to entire file #2518.
• Drop Python 3.8 as it is not supported by tree-sitter.
• Resolve FIXME with incorrect debug information when parsing source code markers. chore(backend/sdoc_source_code): remove legacy @sdoc/nosdoc #2522.
• INFO: 127.0.0.1:53134 - "HEAD /_static/stimulus_umd.min.js HTTP/1.1" 405 Method Not Allowed
• Make SOURCE_FILE_LANGUAGE_PARSERS the default behavior.
• Rename all WARNING: to warning: for consistency.
• Replace all instances of sys.exit(...) with raised Exception.
  • fix(RST): do not exit(1) but raise RuntimeError #2454
• 📌 [Windows] Fix forward slashes on Windows when opening source files with server: # FIXME: path_to_source_file must not enter this function with forward slashes.
• https://github.com/search?type=code&q=NOT+is%3Afork+path%3A*.sdoc+-path%3A*.js.*+-path%3A*.sh.*+-path%3A*.h.*+-path%3A*js*+-path%3A*jquery*+-user%3As31tz+-user%3Aaptana&p=1
• Double-check types and conversions: prefix=assert_cast(element_prefix, str),
• context.title_number_string needs to be refactored to be stateless when iterating over document content. Remove self.context = SectionContext() entirely.
• FIXME: Fill in the document meta correctly.
• Task: Revise the naming convention for node parameters: type, style, view etc. #2239
• Update the available Ruff checks: ruff linter | sed 's/^ *//' | sort
• ExportAction/TraceabilityIndexBuild could be improved. Also their handling in main_router. See unfinished: [DISCARDED] Code climate: export_action: simplify building of traceability index #2298.
• The is_standalone vs is_running_on_server logic is messed up in several places: {% set is_not_standalone = view_object.project_config.is_running_on_server and not view_object.standalone %}. Remove is_standalone check from several HTML templates.
• enumerate_meta_field_titles - # FIXME: currently only enumerating a single element ([0])
• D:/a/strictdoc/strictdoc/tests/integration/features/document_grammar/01_basic_grammar_declaration_and_valid_fields/test.itest:2:8: error: CHECK: expected string not found in input
• render_url() and friends should internally do URL escaping with something like urllib.parse.urlencode, then mark as safe.
• View Objects: double-check if they all could return Markup.
• requirement_type -> node_type.
• do_open_form_edit_requirement -> do_open_form_edit_node.
• # TODO: Check if this line is still needed (might be some older workaround).
• test_20_single_document_add_second_child_relation_with_role – Why does it work when the grammar does not have "Child", "IsImplementedBy"?
• There is some duplication in how the Tree Cycle Detector is used. The parent_cycle_traverse_-functions could be reduced.
• Technical debt: Reduce the number of path-related conversions when finding all sorts of files #1502
• SDoc: Graph database: Unify all entities.
• self.ng_document_reference: Optional[DocumentReference] = None – is there a way to make it non-optional?
• node_controls partial is extremely complicated.
• Finish migration to streams rendered from view objects.
• FileEntry: make it use SDocRelativePath instead of separate POSIX path field.
• strictdoc/export/rst/templates/requirement.jinja.rst has funny rstrip(). Could not be removed just yet.

Flaky end2end tests

• tests/end2end/screens/document/_cross_cutting/included_documents/update_included_document_delete_section/test_case.py
• Flaky copy_stable_link. Seen multiple times.
• Flaky tests/end2end/screens/document/update_requirement/_relations/update_requirement_add_three_links/test_case.py Expected attribute {active} of element {//*[@data-testid='form-tab-Relations']} was not present after 7 seconds!
• Flaky create_section/create_section_after_requirement.
• Rarely but flaky: update_section_add_link_to_existing_section_anchor_from_clipboard. Not always copied to clipboard -> update_node_add_link_to_existing_section_anchor_from_clipboard.
• Flaky test: create_requirement_validate_relation_parent_does_not_exist.
• Flaky test: _validations\delete_node_with_anchor_that_has_incoming_links.
• create_requirement_generate_autouid_when_requirement_with_another_prefix_exists.
• export_to_pdf_basic, https://github.com/strictdoc-project/strictdoc/actions/runs/10130508488/job/28011977903?pr=1929, https://github.com/strictdoc-project/strictdoc/actions/runs/10539860537/job/29203900243?pr=1941.
• delete_requirement_delete_newly_created, https://github.com/strictdoc-project/strictdoc/actions/runs/10256651383/job/28376187324.

Type system epic

• Resolve circular references between SDocDocument and DocumentReference and other classes. Make the overall type system stronger.
• Code climate: improve assert_cast to tell mypy about the casted type #2141
• mypy: fix remaining no-redef issues. code-climate: mypy: address no-redef #2262
• mypy: fix remaining type-arg issues. code-climate: mypy: address type-arg #2264
• mypy: fix remaining var-annotated issues. code-climate: address remaining var-annotated #2245
• mypy: fix remaining attr-defined issues. code-climate: mypy, address attr-defined #2375
• mypy: fix remaining no-any-return issues.
• mypy: fix remaining no-untyped-def issues. Code climate mypy address no untyped def #2377, code-climate: address no-untyped-def review findings #2378
• mypy: fix remaining arg-type issues. Code climate: mypy: address remaining arg-type issues #2399
• mypy: fix remaining no-untyped-call issues. Code climate: mypy: fix some instances of no-untyped-call #2406, Code climate: mypy: fix some more instances of no-untyped-call #2407
• mypy: fix remaining union-attr issues.
• mypy issue with pygmented_source_file_lines (see the description below).

Code coverage epic

• Windows-related code lines, e.g., strictdoc/server/app.py.
• Must never reach here errors.
• end2end for --reload is missing. strictdoc/server/server.py.

---

        pygmented_source_file_lines: List[Union[str, RangeMarkerTuple]] = list(

This is for mypy... Honestly I don't understand why we have to make a list()-copy of pygmented_source_file_content.split("\n"). Without list(), mypy would claim

error: Incompatible types in assignment (expression has type "List[str]", variable has type "List[Union[str, RangeMarkerTuple]]")

I know assignment would be a problem without explicit annotation of pygmented_source_file_lines because of list invariance. But the annotation is there... why do we still need list()?

[stanislaw]

@thseiler @haxtibal

I have finally fixed the most important issues related to collecting code coverage from StrictDoc's child processes and got a good number of 94%, see the comment above ☝ .

Several thoughts/observations:

• The code coverage is collected from all 5 test groups as per the test_all Invoke task. Running the tests with code coverage enabled makes them considerably slower. I am still to measure it but I am feeling like this combined coverage calculation should not be part of the normal CI. I still think that it is worth to set a custom link check that the combined coverage must not fall below a given limit, e.g., 95%, 96%, etc., over time.
• I will be working through some of the most obvious fixes from now on. The next good goal will be to reach 95%.
• Pushing to higher coverage is mostly related to resolving all sorts of non-covered edge cases which opens an interesting trade-off: if the higher coverage is set as a goal, then all future contributions must become 100% covered as well which increases the amount of work every contributor like you will have to do. Previously, we only required mostly green-case integration/end2end test but now much more effort by any contributor would be required.

Do you have any thoughts regarding this topic?