Merge pull request #8 from obiwancenobi/fix/line-number-highlighting

fix/line number highlighting

Author: obiwancenobi

.github/marketplace/metadata.yml

@@ -39,7 +39,7 @@ long_description: |
     review:
       runs-on: ubuntu-latest
       steps:
-        - uses: obiwancenovi/ai-code-reviewer@v1.0.20
+        - uses: obiwancenovi/ai-code-reviewer@v1.0.21
           with:
             pr-number: ${{ github.event.pull_request.number }}
             repository: ${{ github.repository }}

.github/workflows/ai-review.yml

@@ -100,7 +100,7 @@ jobs:
           fi
 
       - name: AI Code Review
-        uses: obiwancenobi/ai-code-reviewer@v1.0.20
+        uses: obiwancenobi/ai-code-reviewer@v1.0.21
         with:
           pr-number: ${{ inputs.pr-number || github.event.pull_request.number }}
           repository: ${{ inputs.repository || github.repository }}

README.md

@@ -1,5 +1,9 @@
+<div align="center">
+
 # BugBeaver Code Reviewer
 
+</div>
+
 <div align="center">
   <img src=".github/marketplace//bugbeaver.png" width="200" alt="BugBeaver Logo">
 </div>
@@ -106,7 +110,7 @@ Add AI code review to any repository with one simple step:
 
        steps:
          - name: AI Code Review
-           uses: obiwancenobi/ai-code-reviewer@v1.0.20
+           uses: obiwancenobi/ai-code-reviewer@v1.0.21
            with:
              pr-number: ${{ github.event.pull_request.number }}
              repository: ${{ github.repository }}
@@ -142,7 +146,7 @@ Use repository variables for organization-wide settings:
 
 ```yaml
 - name: AI Code Review
-  uses: obiwancenobi/ai-code-reviewer@v1.0.20
+  uses: obiwancenobi/ai-code-reviewer@v1.0.21
   with:
     pr-number: ${{ github.event.pull_request.number }}
     repository: ${{ github.repository }}
@@ -488,7 +492,7 @@ Settings are applied in this priority order (highest to lowest):
 
 **Workflow sets:**
 ```yaml
-- uses: obiwancenobi/ai-code-reviewer@v1.0.20
+- uses: obiwancenobi/ai-code-reviewer@v1.0.21
   with:
     ai-provider: ${{ vars.AI_PROVIDER || 'anthropic' }}
     ai-model: ${{ vars.AI_MODEL || 'claude-3-sonnet' }}
@@ -568,7 +572,7 @@ jobs:
 
     steps:
       - name: AI Code Review
-        uses: obiwancenobi/ai-code-reviewer@v1.0.20
+        uses: obiwancenobi/ai-code-reviewer@v1.0.21
         with:
           pr-number: ${{ github.event.pull_request.number }}
           repository: ${{ github.repository }}
@@ -598,7 +602,7 @@ jobs:
 
     steps:
       - name: AI Code Review
-        uses: obiwancenobi/ai-code-reviewer@v1.0.20
+        uses: obiwancenobi/ai-code-reviewer@v1.0.21
         with:
           pr-number: ${{ github.event.pull_request.number }}
           repository: ${{ github.repository }}
@@ -611,7 +615,7 @@ jobs:
 #### Python Projects
 ```yaml
 - name: AI Code Review
-  uses: obiwancenobi/ai-code-reviewer@v1.0.20
+  uses: obiwancenobi/ai-code-reviewer@v1.0.21
   with:
     pr-number: ${{ github.event.pull_request.number }}
     repository: ${{ github.repository }}
@@ -624,7 +628,7 @@ jobs:
 #### Java/.NET Projects
 ```yaml
 - name: AI Code Review
-  uses: obiwancenobi/ai-code-reviewer@v1.0.20
+  uses: obiwancenobi/ai-code-reviewer@v1.0.21
   with:
     pr-number: ${{ github.event.pull_request.number }}
     repository: ${{ github.repository }}
@@ -646,7 +650,7 @@ Set these in repository Settings → Actions → Variables:
 
 ```yaml
 - name: AI Code Review
-  uses: obiwancenobi/ai-code-reviewer@v1.0.20
+  uses: obiwancenobi/ai-code-reviewer@v1.0.21
   with:
     pr-number: ${{ github.event.pull_request.number }}
     repository: ${{ github.repository }}

index.js

@@ -16,7 +16,7 @@ const program = new Command();
 program
   .name('ai-code-reviewer')
   .description('AI-powered code review for GitHub pull requests')
-  .version('1.0.20');
+  .version('1.0.21');
 
 // Review command for GitHub Actions
 program

line-number-analysis.md

@@ -0,0 +1,119 @@
+# AI Code Review Line Number Mismatch Analysis
+
+## Problem Description
+The AI code review system sometimes highlights comments on wrong line numbers. For example, the changes are on line 4, and the review context is correct, but the highlighting appears on line 3.
+
+## Root Causes Identified
+
+### 1. Content Type Inconsistency
+- **File Source**: In `aiReviewService.js:138`, the system uses `content = file.patch || ''`
+- **Issue**: When `file.patch` exists, it contains GitHub diff format with hunks and line prefixes
+- **Impact**: AI reviews patch content but line numbers are calculated as if it's full file content
+
+### 2. Simplified Chunking Logic
+- **Current Implementation**: `aiReviewService.js:160`
+  ```javascript
+  line_number: comment.line_number ? comment.line_number + (i * Math.floor(content.split('\n').length / chunks.length)) : null
+  ```
+- **Issues**:
+  - Assumes uniform chunk sizes (doesn't account for actual content boundaries)
+  - Doesn't track which specific lines are included in each chunk
+  - Cannot handle variable-length hunks in patch content
+
+### 3. Dual Chunking Systems
+- **FileChunker** (`fileChunker.js`): Sophisticated chunking with proper line tracking
+- **FileProcessor** (`fileProcessor.js`): Simple string-based chunking without line awareness
+- **Current Usage**: System uses FileProcessor's `splitIntoChunks` which lacks line number tracking
+
+### 4. Patch Format Misunderstanding
+- **GitHub Patch Format**: Contains hunks with headers like `@@ -1,3 +1,3 @@`
+- **AI Processing**: AI sees content with `+`/`-` prefixes but produces line numbers for original file
+- **Gap**: No conversion between patch line numbers and original file line numbers
+
+### 5. Line Number Validation Issues
+- **GitHubClient** (`client.js:73-76`): Validates line numbers but uses different calculation
+- **Mismatch**: Review service calculates differently than GitHub API expects
+
+## Detailed Analysis
+
+### Patch Content Handling
+```javascript
+// aiReviewService.js:138
+content = file.patch || '';
+```
+
+When `file.patch` exists:
+1. AI reviews diff hunks with line prefixes (`+`, `-`, ` `)
+2. AI produces line numbers relative to patch content
+3. System tries to map these to original file lines (incorrect approach)
+
+### Chunking Line Number Calculation
+```javascript
+// aiReviewService.js:160
+line_number: comment.line_number ? 
+  comment.line_number + (i * Math.floor(content.split('\n').length / chunks.length)) : 
+  null
+```
+
+**Problems**:
+- Assumes chunks have equal line counts
+- Doesn't consider actual content structure
+- No awareness of where chunks start/end in original file
+
+### FileChunker Usage Gap
+The sophisticated `FileChunker` class exists but isn't used by the main review flow. It has:
+- Proper line number tracking (`startLine`, `endLine`)
+- Chunk overlap management
+- Content validation
+
+## Solution Recommendations
+
+### 1. Content Strategy Selection
+- **When full content available**: Use complete file content for AI review
+- **When only patch available**: Parse patch to extract changed lines and context
+- **Fallback**: If neither available, skip line-specific comments
+
+### 2. Unified Chunking System
+- Use `FileChunker` throughout the application
+- Maintain proper line number mapping for all chunks
+- Remove duplicate chunking logic from `FileProcessor`
+
+### 3. Patch Content Parser
+Implement proper patch parsing to:
+- Extract hunks and their original line ranges
+- Map patch line numbers to original file line numbers
+- Handle context lines in hunks correctly
+
+### 4. Line Number Validation
+- Validate AI line numbers against file content bounds
+- Add logging for line number calculation steps
+- Fall back to general comments when line numbers are invalid
+
+### 5. Testing and Verification
+- Add unit tests for line number calculation
+- Create integration tests with sample PR diffs
+- Add logging to trace line number adjustments
+
+## Impact Assessment
+
+**High Impact Issues**:
+- Incorrect line highlighting confuses developers
+- Reduces trust in AI review comments
+- May cause important issues to be missed
+
+**Medium Impact Issues**:
+- Inconsistent behavior across different file types
+- Increased support requests
+
+**Technical Debt**:
+- Duplicate chunking implementations
+- Lack of comprehensive testing
+- Poor error handling for edge cases
+
+## Next Steps
+
+1. **Immediate Fix**: Use FileChunker instead of FileProcessor for chunking
+2. **Patch Parser**: Implement GitHub patch format parsing
+3. **Line Number Validation**: Add bounds checking and validation
+4. **Testing**: Create comprehensive test suite
+5. **Documentation**: Update code comments explaining line number handling

line-number-fix-summary.md

@@ -0,0 +1,146 @@
+# Line Number Highlighting Fix - Implementation Summary
+
+## Problem Overview
+The AI code review system was highlighting comments on incorrect line numbers. For example, when changes were made on line 4, the review context was correct but the highlighting appeared on line 3.
+
+## Root Causes Identified and Fixed
+
+### 1. Patch Content vs Full Content Mismatch
+**Problem**: The system used `content = file.patch || ''` but treated patch content as if it was full file content.
+**Solution**: Implemented proper patch parsing with `PatchParser` class that:
+- Parses GitHub unified diff format
+- Maps patch line numbers to original file line numbers
+- Extracts reviewable content with proper context
+
+### 2. Inadequate Chunking System
+**Problem**: Used `fileProcessor.splitIntoChunks()` which lacked line number tracking.
+**Solution**: Replaced with `FileChunker` which provides:
+- Proper line number tracking (`startLine`, `endLine`)
+- Chunk overlap management
+- Accurate line number adjustment for chunked content
+
+### 3. Simplified Line Number Calculation
+**Problem**: Line numbers were calculated with oversimplified formula:
+```javascript
+line_number + (i * Math.floor(content.split('\n').length / chunks.length))
+```
+**Solution**: Implemented proper mapping:
+- Use `FileChunker.adjustCommentLineNumbers()` for chunk-relative line numbers
+- Use `PatchParser.mapPatchLineToOriginalFile()` for patch-to-original mapping
+- Add validation with `validateCommentLineNumber()`
+
+## Key Components Implemented
+
+### 1. PatchParser (`src/utils/patchParser.js`)
+- **parsePatch()**: Parses GitHub patch format into structured hunks
+- **mapPatchLineToOriginalFile()**: Maps AI line numbers from patch to original file
+- **extractReviewContent()**: Creates AI-reviewable content from patch
+- **isValidLineNumber()**: Validates line numbers against file boundaries
+
+### 2. Enhanced AIReviewService (`src/services/aiReviewService.js`)
+- **reviewFile()**: Now properly handles both patch and full content
+- **reviewCodeChunk()**: Enhanced validation and context
+- **validateCommentLineNumber()**: New validation method
+- Uses FileChunker for all chunking operations
+- Implements proper line number mapping for patch content
+
+### 3. Comprehensive Testing
+- **Unit Tests**: `tests/unit/utils/patchParser.test.js` - Tests patch parsing logic
+- **Integration Tests**: `tests/integration/lineNumberAccuracy.test.js` - Tests end-to-end line number accuracy
+
+## How the Fix Works
+
+### Before (Problematic Flow):
+```
+GitHub Patch → AI Review → Incorrect Line Numbers → Wrong Highlighting
+```
+
+### After (Fixed Flow):
+```
+GitHub Patch → Parse with PatchParser → AI Review → Map Lines → Correct Highlighting
+```
+
+### Step-by-Step Process:
+1. **Content Detection**: System detects if content is patch or full file
+2. **Patch Parsing**: If patch, `PatchParser` extracts hunks and mappings
+3. **Content Extraction**: Reviewable content is prepared for AI
+4. **AI Review**: AI analyzes content and provides line numbers
+5. **Line Mapping**: Patch line numbers are mapped to original file lines
+6. **Validation**: Line numbers are validated against file boundaries
+7. **Comment Creation**: Final comments have accurate line numbers
+
+## Testing the Fix
+
+### Run Unit Tests:
+```bash
+npm test tests/unit/utils/patchParser.test.js
+```
+
+### Run Integration Tests:
+```bash
+npm test tests/integration/lineNumberAccuracy.test.js
+```
+
+### Manual Testing Scenarios:
+1. **Single Hunk Patch**: Test with `@@ -5,3 +5,3 @@`
+2. **Multiple Hunks**: Test with complex patches
+3. **Edge Cases**: Test with empty patches, large files, etc.
+
+## Key Benefits
+
+### 1. Accurate Line Highlighting
+- Comments now appear on the exact lines where issues were identified
+- Developers can quickly locate the problematic code
+
+### 2. Robust Error Handling
+- Graceful fallback when patch parsing fails
+- Validation prevents invalid line numbers
+
+### 3. Better AI Context
+- AI receives properly formatted content for analysis
+- Enhanced context includes chunk metadata
+
+### 4. Comprehensive Testing
+- Unit tests ensure individual component correctness
+- Integration tests verify end-to-end functionality
+
+## Backward Compatibility
+
+The fix maintains backward compatibility:
+- Existing configurations continue to work
+- Full file content handling is unchanged
+- Error handling gracefully degrades to old behavior if needed
+
+## Performance Impact
+
+- **Minimal Overhead**: Patch parsing is lightweight
+- **Efficient Chunking**: FileChunker is optimized for large files
+- **Caching Opportunities**: Parsed patches can be cached for multiple reviews
+
+## Monitoring and Debugging
+
+The implementation includes extensive logging:
+```javascript
+logger.debug(`Parsed patch with ${result.hunks.length} hunks...`);
+logger.warn(`Could not map patch line ${aiLineNumber} to original file line`);
+```
+
+This helps troubleshoot any remaining edge cases in production.
+
+## Future Enhancements
+
+Potential improvements for future versions:
+1. **Enhanced Context**: Include more file context around changes
+2. **Smart Chunking**: Adaptive chunk sizes based on code structure
+3. **Caching**: Cache parsed patches for repeated reviews
+4. **Machine Learning**: Use historical data to improve line number accuracy
+
+## Conclusion
+
+The line number highlighting issue has been comprehensively addressed through:
+- Proper patch parsing and line number mapping
+- Replacement of inadequate chunking with FileChunker
+- Extensive validation and error handling
+- Comprehensive testing to prevent regressions
+
+This fix ensures that AI code review comments appear on the correct lines, improving developer trust and reducing confusion during code reviews.