feat: add early module import validation for JavaScript/TypeScript

[aseembits93]

Summary

• Adds pre-flight validation to the JS/TS optimizer: before making expensive AI API calls, checks whether the source file is within the configured moduleRoot
• If misconfigured, auto-corrects by inferring the module root from package.json fields and directory conventions (src/, lib/, etc.), and writes the corrected value back to package.json
• Mirrors the existing Python module import validation pattern

Changes

• New: codeflash/code_utils/config_js_validation.py — validate_js_module_resolution() and infer_js_module_root() utilities
• Modified: codeflash/languages/javascript/function_optimizer.py — Added try_correct_module_root() and can_be_optimized() override to JavaScriptFunctionOptimizer
• New: tests/code_utils/test_config_js_validation.py — 10 unit tests for validation utilities
• New: tests/test_languages/test_js_function_optimizer_validation.py — 6 integration tests for optimizer methods

Test plan

• uv run pytest tests/code_utils/test_config_js_validation.py -v — 10/10 pass
• uv run pytest tests/test_languages/test_js_function_optimizer_validation.py -v — 6/6 pass
• uv run pytest tests/test_languages/test_javascript_test_runner.py -v — 29/29 pass (no regressions)
• uv run ruff check and uv run ruff format --check — all clean

🤖 Generated with Claude Code

[claude]

Claude finished @aseembits93's task in 8m 58s —— View job

---

PR Review Summary

• Triage PR scope — substantive code changes, full review proceeding
• Lint and typecheck — prek passed; fixed 2 mypy errors (committed)
• Resolve stale review threads — no prior threads
• Code review — complete
• Duplicate detection — complete
• Test coverage — 97% on validation util, 16/16 tests passing
• Post summary
• Check optimization PRs — closed 28 stale/CI-failing PRs

---

Prek Checks

Ruff check and format both pass. Two mypy errors were found in the new code and auto-fixed in commit cb1492f7:

• codeflash/languages/javascript/function_optimizer.py:103 — wrong import: Language was imported from codeflash.languages.base (which re-exports it from language_enum); changed to from codeflash.languages.language_enum import Language directly.
• function_optimizer.py:45,59,60,93 — self.args is typed Namespace | None but accessed without a null-guard. Added early-return guards (if self.args is None: return False / return super().can_be_optimized()) to both try_correct_module_root() and can_be_optimized(). This is also semantically correct behavior.

Code Review

Bug / Dead code — try_correct_module_root() lines 60–61:

self.args.project_root = project_root  # project_root = self.project_root — unchanged
self.project_root = project_root       # same: self.project_root = self.project_root

project_root is assigned from self.project_root on line 44 and never mutated, so these two lines are pure no-ops. Fix this →

Double validation — can_be_optimized() calls validate_js_module_resolution twice:
try_correct_module_root() calls validate_js_module_resolution internally (line 49) to decide whether correction is needed, then can_be_optimized() calls it again (line 100) to decide whether to return Failure. After try_correct_module_root() successfully corrects the root, the second call is fine (it will pass). But if the file is already valid, validate_js_module_resolution is called unnecessarily twice. Minor inefficiency, not a bug.

Side effect concern — auto-modifying package.json from can_be_optimized():
try_correct_module_root() silently writes the corrected moduleRoot back to package.json as a side effect of can_be_optimized(). The PR description says this is intentional ("auto-corrects … and writes the corrected value back"), but it's worth noting this is a visible-to-user behavior change (modifies a tracked file) triggered deep in an optimization pre-flight check. The Python equivalent (prepare_module) doesn't write to config files. Consider whether the user should be prompted or at least given a distinct log message (the current logger.info is fine but could be more prominent as a logger.warning).

Late import of PACKAGE_JSON_DATA_CACHE inside a function body:

from codeflash.code_utils.config_js import PACKAGE_JSON_DATA_CACHE
PACKAGE_JSON_DATA_CACHE.pop(package_json_path, None)

This import is inside a try block inside try_correct_module_root(). It should be moved to the top of the file with other imports. Fix this →

Duplicate Detection

MEDIUM confidence — infer_js_module_root() in config_js_validation.py is a thin wrapper around detect_module_root() from config_js.py, which itself has nearly identical logic to _detect_js_module_root() in setup/detector.py. This is a pre-existing duplication between config_js.py and setup/detector.py — the new code correctly delegates to the existing detect_module_root() utility rather than reimplementing it, so no new duplication is introduced by this PR.

No other duplicates detected.

Test Coverage

File	Coverage
codeflash/code_utils/config_js_validation.py	97% (35 stmts, 1 miss)
codeflash/languages/javascript/function_optimizer.py	43% (pre-existing code accounts for most uncovered lines)

The one uncovered line is config_js_validation.py:79 — the branch where get_package_json_data() returns None (malformed package.json). Minor gap; consider adding a test with an empty/invalid package.json.

All 16 new tests pass.

---

Last updated: 2026-03-04T03:22Z

[Saga4]

Best to place and dedupe with logic present in file 'init_javascript.py'

[Saga4]

There are functions for this purpose we can strength the logic there.
collect_js_setup_info

[Saga4]

or look into detect_module_root in config_js.py