Fix split.dictionary defaults for invalid inputs and null values

[ebhills]

Linked issue

Closes #1223

What changes

A split.dictionary step with an explicit default now handles non-dictionary inputs such as Timed Out and applies per-key defaults to null values. With default: {Voltage: {}}, both a missing Voltage key and {"Voltage": null} produce {}; existing Voltage objects remain intact.

Invalid input cells are treated as empty dictionaries only when a default is supplied, including default: {}. Multiple input columns retain their existing precedence, then missing/null keys receive their defaults. The behavior applies to both columns and to_lists, with independent copies of mutable defaults. Schema descriptions explain the fallback, case sensitivity, preserved non-null values, and output-format handling.

Scope: one commit, limited to wrangles/recipe_wrangles/split.py and its existing test file.

How it was verified

• Full credential-free suite: scripts/test-local.ps1 — 2,821 passed, 6 skipped, 139 deselected.
• Focused offline run using pytest-local.ini: the complete split test file plus the wrangle schema-docstring check — 133 passed. Credentials were removed and networking was disabled.
• Generated the recipe schema in a temporary directory and validated it as Draft 7. Three representative recipes passed; three invalid default shapes were rejected.
• git diff --check passed.
• Regression coverage includes the JSON-to-Voltage recipe, invalid inputs with and without defaults, empty/invalid defaults, multi-input precedence, valid falsy values, case-sensitive keys, renaming/filtering, both output formats, and mutable-default isolation.
• No live Excel or Wrangles-service execution was performed. GitHub CI is pending.

Compatibility and risk

Existing recipes that supply defaults intentionally change behavior for invalid input cells and null-valued keys. Recipes without defaults remain strict for invalid inputs. Non-null values, including zero, false, empty strings, lists, and dictionaries, are preserved; defaults do not recursively merge nested objects.

No public parameters or dependencies change. The updated Python package must be deployed to execute-recipes-dev after merge before WranglesXL development uses this behavior. Rollback is to revert this commit and redeploy the previous package.

Ready-for-review checklist

• One human delivery owner is assigned — ebhills
• The linked issue and intended milestone are correct — Apply split.dictionary defaults to invalid inputs and null values #1223; no release milestone specified
• The branch is current with main and has no merge conflicts
• Focused tests pass
• New or changed behavior has direct test coverage
• Documentation/schema/configuration is updated where applicable
• The PR contains no unrelated changes
• The PR description reflects the branch's current scope and latest validation
• One primary reviewer is requested only when this PR is ready

See the pull request workflow.

[Copilot]

🟢 Approval recommended

No unresolved review issues were identified, and all approval-readiness assessments recommend approval.

0 open findings

What changed in this PR

Updates split.dictionary to apply defaults for invalid inputs and null values while preserving valid values and input precedence.

Changes:

• Adds fallback handling for invalid inputs when defaults are supplied.
• Applies per-key defaults to missing and null values.
• Expands schema documentation and regression tests for both output formats.

File	Description
wrangles/​recipe_wrangles/​split.py	Implements fallback handling and updated schema documentation.
tests/​recipes/​wrangles/​test_split.py	Adds comprehensive regression coverage.

🧠 Review effort: Lite

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.