fix(grep): handle CRLF line endings in ripgrep output parsing

[zhoufanscut]

Summary

• Fix parseOutput and parseCountOutput in src/tools/grep/cli.ts to handle CRLF (\r\n) line endings from ripgrep output
• Add unit tests for both parsers covering CRLF, LF, and empty-output cases

Problem

When the Grep tool searches files with Windows-style CRLF line endings, ripgrep preserves the \r in its stdout. The parsers split on "\n" only, leaving a trailing \r on every line. This causes the content-mode regex /^(.+?):(\d+):(.*)$/ to silently fail — the $ anchor does not match before \r in JavaScript — so all matches are dropped and the tool returns "No matches found".

The files_with_matches mode was unaffected because it uses line.trim(), which strips \r. The count mode has the same latent bug in parseCountOutput.

Reproduction

# Search a CRLF file in content mode → returns "No matches found"
grep(pattern="proxy", path="/path/to/crlf_file.py", output_mode="content")

# Same file in files_with_matches mode → works fine
grep(pattern="proxy", path="/path/to/crlf_file.py", output_mode="files_with_matches")

Fix

Change output.split("\n") → output.split(/\r?\n/) in both parseOutput (line 103) and parseCountOutput (line 135). After proper splitting, the existing regexes work correctly with no further changes needed.

Changes

File	Change
src/tools/grep/cli.ts	split("\n") → split(/\r?\n/) in parseOutput and parseCountOutput; export both functions for testability
src/tools/grep/cli.test.ts	New — 7 unit tests covering CRLF content, CRLF files-only, CRLF no-trailing-\r, CRLF count, LF content, LF files-only, and empty output

Testing

• Existing result-formatter.test.ts — 5/5 pass (no regressions)
• New cli.test.ts — all assertions verified (import currently blocked by unrelated missing js-yaml dependency in the test runner, logic verified independently)
• Manual reproduction confirmed: CRLF file now returns correct matches in content mode

---

Summary by cubic

Fix parsing of ripgrep output with CRLF line endings so content and count modes return correct results on Windows. Adds tests covering CRLF, LF, and empty-output cases.

• Bug Fixes
  • Split lines with /\r?\n/ in parseOutput and parseCountOutput in src/tools/grep/cli.ts to remove stray \r.
  • Export both functions and add src/tools/grep/cli.test.ts covering content, files-only, and count modes for CRLF/LF and empty output.
  • Resolves "No matches found" in content mode on CRLF files; count mode handles CRLF correctly.

^{Written for commit 697e44a. Summary will update on new commits.}

[github-actions]

All contributors have signed the CLA. Thank you! ✅
_{Posted by the CLA Assistant Lite bot.}

[cubic-dev-ai]

No issues found across 2 files

Confidence score: 5/5

• Automated review surfaced no issues in the provided summaries.
• No files require special attention.

_{Auto-approved: Fixes CRLF parsing in ripgrep output using standard regex split. Safe change with comprehensive tests and no impact on existing LF logic.}

[Copilot]

Pull request overview

This PR fixes ripgrep output parsing to correctly handle Windows-style CRLF (\r\n) line endings so matches aren’t dropped in content and count modes.

Changes:

• Update parseOutput and parseCountOutput to split stdout with /\r?\n/ instead of "\n".
• Export both parser functions to enable direct unit testing.
• Add unit tests covering LF, CRLF, and empty-output cases for both parsers.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
src/tools/grep/cli.ts	Makes stdout splitting CRLF-tolerant in both parsers; exports parsers for testability.
src/tools/grep/cli.test.ts	Adds Bun unit tests validating parsing behavior for LF/CRLF and empty output.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[zhoufanscut]

I have read the CLA Document and I hereby sign the CLA