Context

Summary By MatterAI

🔄 What Changed

Optimized the mapNormalizedIndexToOriginal function in fileEditTool.ts to support offset-based resumption. This allows the tool to calculate the end position of a match by starting from the previously found start position, rather than re-scanning the entire file from the beginning. This PR also incorporates previous release refinements for CRLF handling and search feedback.

🔍 Impact of the Change

Significantly improves performance for large file edits, especially those with many CRLF sequences. By reducing the complexity of index mapping from O(N) to O(M) for subsequent lookups, the tool becomes more efficient during automated code modifications and prevents "off-by-one" errors on Windows systems.

📁 Total Files Changed

🧪 Test Added/Recommended

Recommended

• Performance benchmarks for large files (>1MB) with mixed line endings to quantify the gain.
• Unit tests verifying the accuracy of originalEnd when using non-zero start offsets.

🔒 Security Vulnerabilities

N/A

Implementation

The mapNormalizedIndexToOriginal function was updated to accept startOriginalIndex and startNormalizedPos as optional parameters. The flexibleSubstringReplacer logic now passes the originalStart and normalizedIndex into the second call to mapNormalizedIndexToOriginal, ensuring the loop resumes from the correct position. This avoids redundant iterations over the file content.

Screenshots

How to Test

1. Run fileEditTool on a file with CRLF line endings.
2. Verify that the substring replacement correctly identifies the original content range.
3. Check for performance improvements on files larger than 1MB.
4. Run a ripgrep search that returns zero results to verify the new hint message.

Get in Touch

N/A

/matterai fix

✅ I've created a fix PR to address the review comments!

🔗 matterai-fix: Address review comments for PR #21

Files Changed:

• src/core/tools/fileEditTool.ts

Please review the changes and merge them if they look correct.

✅ Reviewed the changes: The changes correctly implement the performance optimizations for mapNormalizedIndexToOriginal by allowing resumed scanning. The logic appears sound and handles the CRLF mapping correctly.