petab1to2: allow skipping validation

[FFroehlich]

petab1to2 refuses to convert any PEtab v1 problem that does not pass v1.lint_problem, even when the lint errors have nothing to do with the conversion. Examples are a visualization table that references a datasetId missing from the measurement table (Isensee_JCB2018, Schwen_PONE2014) and a model that is incomplete on purpose.

This PR adds a keyword-only validate: bool = True argument to petab1to2 and petab_files_1to2:

• validate=True (default): behaviour is unchanged. The v1 problem is linted before the conversion and the generated v2 problem after it, and either failure raises ValueError.
• validate=False: both lint steps are skipped. The v1 YAML file is still checked against the v1 schema, since the converter relies on its structure. The docstring recommends checking the result separately, e.g. with petablint.

I skipped the post-conversion v2 lint as well, rather than only downgrading it. A problem that fails v1 lint usually fails v2 lint for the same reason (e.g. an incomplete model), so keeping that check would defeat the opt-out.

Because the new parameter would shadow the imported petab.v1.yaml.validate, that import is now aliased as validate_yaml.

There is no CLI entry point for the converter, so no CLI flag was needed.

Not done: only warning about visualization-table errors by default. The issue also suggests this alternative. It would be simple, since the converter drops the visualization table anyway (PEtab v2 has no visualization_files). However, it changes the default behaviour, so I left it out of this PR. It could be added on top if wanted.

Tests

• New test_petab1to2_validate: a v1 problem whose only lint error is a stale visualization datasetId still raises by default. With validate=False it converts, both to a Problem and to an output directory.
• pytest tests/v2: 78 passed, 1 skipped (benchmark_models_petab not installed locally).
• The repro script from the issue converts with validate=False. A v1 problem with an empty SBML model also converts with validate=False.

Closes #528

🤖 Generated with Claude Code

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 76.63%. Comparing base (78dabf4) to head (bbed14b).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #534      +/-   ##
==========================================
+ Coverage   76.50%   76.63%   +0.12%     
==========================================
  Files          67       67              
  Lines        7526     7529       +3     
  Branches     1341     1342       +1     
==========================================
+ Hits         5758     5770      +12     
+ Misses       1269     1262       -7     
+ Partials      499      497       -2     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Copilot]

🟡 Changes recommended

Regression coverage does not verify that disabling validation also skips post-conversion linting.

1 open finding

• Strengthen tests to verify validate=False skips v2 linting · New

What changed in this PR

Adds an opt-out for linting during PEtab v1-to-v2 conversion, addressing #528 while preserving default validation.

Changes:

• Adds keyword-only validate=True to both conversion functions.
• Skips both lint steps when disabled, while retaining YAML schema validation.
• Adds conversion tests using invalid visualization data.

File	Description
tests/​v2/​test_conversion.py	Tests default rejection and opt-out conversion in both output modes.
petab/​v2/​petab1to2.py	Implements and documents optional linting.

🧠 Review effort: Balanced

---

Give feedback about Copilot approvals in this survey to enter a drawing for a $150 gift card.

[dilpath]

👍