feat: add Jaro-Winkler typo suggestions for unknown operations (Fixes #43)

[andreasronge]

Summary

Implements helpful typo suggestions in validation error messages when users provide unknown operation names. Uses the built-in Jaro-Winkler similarity algorithm to suggest the closest matching valid operation when similarity exceeds 0.8.

This directly supports Phase 5 goals for "Error messages optimized for LLM consumption" - typo suggestions help LLMs self-correct their generated programs, reducing roundtrips.

Changes

• lib/ptc_runner/validator.ex:

  • Added @valid_operations module attribute listing all 34 valid operations
  • Implemented suggest_operation/1 helper function using String.jaro_distance/2
  • Updated unknown operation error to include suggestion suffix when similarity > 0.8
  • Handles case-insensitive matching via String.downcase/1

• test/ptc_runner_test.exs:

  • Added 5 unit tests covering common typos:
    • Single character typo ("filer" → "filter")
    • Missing letter ("selct" → "select")
    • Extra letter ("filtter" → "filter")
    • Case insensitivity ("FILTER" → "filter")
    • No suggestion for unrelated names ("xyz")
  • Added E2E test demonstrating typo suggestion in complex pipeline

Test Results

• All 242 tests pass
• No code style issues detected
• No compilation warnings

Examples

```
Unknown operation 'filer'. Did you mean 'filter'?
Unknown operation 'selct'. Did you mean 'select'?
Unknown operation 'xyz' (no suggestion, below threshold)
```

Fixes #43

🤖 Generated with Claude Code

[github-actions]

PR Review: feat: add Jaro-Winkler typo suggestions for unknown operations

Summary

Clean implementation of typo suggestions for unknown operation names using Elixir's built-in String.jaro_distance/2. The PR correctly addresses Issue #43 with a minimal, focused approach that directly supports the Phase 5 goal of LLM-friendly error messages.

What's Good

• Correct algorithm choice: Using String.jaro_distance/2 from stdlib avoids external dependencies
• Sensible threshold: 0.8 threshold is appropriate - high enough to avoid misleading suggestions, low enough to catch common typos
• Case-insensitive matching: String.downcase/1 on the unknown op ensures FILTER → filter works
• Complete @valid_operations list: All 33 defined operations are correctly included (verified against actual validate_operation/2 clauses)
• Comprehensive test coverage:
  • Common typo patterns (missing letter, extra letter, adjacent letter swap via "filer")
  • Case insensitivity
  • No suggestion for unrelated names
  • E2E test showing suggestion in nested context
• Clean integration: The change is minimal (14 lines added) and fits well with existing validator patterns

Issues (Must Fix)

None. The implementation is complete and correct.

Suggestions (Optional)

1. OPTIONAL: Empty string edge case test - Nice-to-have
   • Suggestion: Consider adding a test for {"op": ""} to document behavior
   • Benefit: Documents that empty strings naturally get no suggestion (jaro_distance returns 0.0)
   • Complexity: Trivial (1 test)
   • Recommendation: Not blocking - current validation may reject this earlier anyway

Security

No concerns. The feature only affects validation error messages and uses no external input beyond what's already validated. The @valid_operations list is hardcoded at compile time.

Verdict

Approve - This is a well-implemented, focused enhancement that directly addresses the issue specification. All tests pass, the code follows existing patterns, and the @valid_operations list correctly matches all 33 defined operations. Ready to merge.

[andreasronge]

Auto-Triage Summary

Review Analysis

Reviewed the PR #44 approval comment from github-actions bot. The review gave an APPROVE verdict with zero must-fix issues and one optional suggestion.

Decisions Made

#	Issue	Decision	Reasoning
1	Empty string edge case test	IGNORE	Reviewer explicitly marked as "OPTIONAL", "nice-to-have", and "not blocking". This is a truly optional stylistic preference with no functional benefit. The PR received an APPROVE verdict and is ready to merge as-is.

Status

• FIX_NOW items: 0
• Issues created: 0
• Items ignored: 1

Conclusion

The PR is approved and ready to merge. All critical functionality is implemented correctly with comprehensive test coverage. The single optional suggestion does not warrant blocking or additional work.