fix(python): trim Layer.source redaction cost and document save_project - #1772

No actionable comments were generated in the recent review. 🎉

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 @coderabbitai help to get the list of available commands.}

🔍 Cloudflare PR preview

Code review

I reviewed the diff (docs/python.md, python/README.md, python/src/geolibre/{authoring.py, geolibre.py, project.py}) against the surrounding source, including _redact_config, _sweep_layer_credentials, redact_layer, and the existing test coverage in python/tests/test_scripting.py (lines 666–731, which exercise Layer.source/Layer.data redaction).

Bugs: None found. redact_layer_field (project.py) is a straight extraction of the per-field sweep _sweep_layer_credentials already performs (layer[field] = _redact_config(layer[field])); _redact_config always rebuilds dicts/lists/leaves into new objects rather than mutating or aliasing its input, so calling it directly on self._layer().get("source") (geolibre.py) is behaviorally identical to the old redact_layer(self._layer()).get("source") — it neither leaks a live reference into the store nor changes what gets redacted. Confidence: high.

Security: No change in redaction coverage — source is still swept by the same credential-name/URL-scrubbing logic before being handed back to the caller. Confidence: high.

Performance: The stated goal (avoid deep-copying the whole layer, including a potentially large inlined GeoJSON blob, just to read source) is achieved correctly. Confidence: high.

Quality: All new/edited docstrings (authoring.py save_project, geolibre.py Layer.source) and the docs/README table additions accurately describe the current code — I verified every method listed in the new docs/python.md table row (set_zoom, set_bearing, set_pitch, fit_project_bounds, center/zoom/bearing/pitch/basemap/name, rename_layer, move_layer, duplicate_layer, show_layer, hide_layer, layer_properties, column_values, describe, remove_layer) exists with matching signatures/behavior. Minor, non-blocking nit: the new Note: docstring section in authoring.py is the only use of that Google-style section header in the package (elsewhere only Args/Returns/Raises appear) — purely stylistic, not worth a fix. Confidence: low.

CLAUDE.md: No applicable guidelines are implicated (change is confined to the Python package and docs, not the JS workspaces/Whitebox catalog/CSP/i18n areas CLAUDE.md calls out). Confidence: high.

No inline comments posted — nothing rose to a level warranting one.

🔍 GitHub Pages PR preview

All accurate. Everything checks out: the redact_layer_field refactor is behaviorally equivalent to the old redact_layer(...).get("source") path but avoids the wasted deep-copy of the whole layer (including any inlined GeoJSON), the docstring updates accurately describe the save_project verbatim-vs-redacted split, and the doc-table syncs match the real method/property signatures in geolibre.py. No test changes were needed since existing test_layer_data_and_source_redact_credentials already exercises layer.source redaction and continues to pass under the new implementation.

I found no bugs, security issues, or CLAUDE.md violations worth flagging as inline comments — this is a small, low-risk refactor + docs PR.

Code review

Bugs: None found. redact_layer_field is a straightforward extraction of the existing _redact_config call that redact_layer/_sweep_layer_credentials already performed on the source field; depth counting, detachment semantics, and the None/missing-field case all match the prior behavior. (Confidence: high)

Security: None found. Credential sweeping still goes through the same _redact_config logic; no new attack surface introduced. (Confidence: high)

Performance: The stated goal is met — Layer.source no longer deep-copies the whole layer record (including any inlined GeoJSON) just to discard everything but source. (Confidence: high)

Quality: Small, well-scoped change with a clear docstring explaining why the new helper exists; existing test (test_layer_data_and_source_redact_credentials) still covers the behavior. No new unit test targets redact_layer_field directly, but coverage via Layer.source is adequate for this size of change. (Confidence: medium)

CLAUDE.md: N/A — this PR touches only python/ and docs/python.md, no areas governed by the repo's other convention rules (Whitebox catalog, i18n, CSP, etc.). (Confidence: high)