fix: [xxx](xxx) render as xxx](xxx)

[tomerqodo]

Benchmark PR from qodo-benchmark#275

[greptile-apps]

Greptile Summary

This PR fixes markdown link rendering by implementing a placeholder-based approach to preserve markdown links and images during URL removal. The fix prevents markdown links like [text](url) from being mangled into text](url) when the remove_urls_emails cleaning rule is enabled.

Key changes:

• Modified CleanProcessor.clean() to use placeholders that protect entire markdown links/images before URL removal
• Updated remove_leading_symbols() to detect and preserve markdown links at the start of text
• Added comprehensive test coverage including edge cases like URL-as-link-text

Critical issues found:

• Regex pattern for markdown images on line 38 missing capture group for alt text, causing match.group(2) on line 51 to fail with IndexError
• Alt text hardcoded as "image" instead of preserving original value
• Unused __init__ method added to CleanProcessor with default_rules parameter that's never used

Confidence Score: 2/5

• This PR has critical runtime bugs that will cause failures when processing markdown images
• Score reflects critical regex bug causing IndexError when match.group(2) is called on a pattern with only one capture group (line 38/51 in clean_processor.py). The alt text preservation is also broken. While the approach is sound and tests are comprehensive, the implementation will fail at runtime when processing markdown images.
• Pay close attention to api/core/rag/cleaner/clean_processor.py - the markdown image pattern needs fixing on line 38, and the placeholder function needs updating on lines 49-54

Important Files Changed

Filename	Overview
api/core/rag/cleaner/clean_processor.py	Markdown link preservation logic added but has critical regex bug - match.group(2) called on pattern with only one capture group
api/core/tools/utils/text_processing_utils.py	Added markdown link preservation check at start of function - straightforward and correct implementation
api/tests/unit_tests/core/rag/cleaner/test_clean_processor.py	Comprehensive test coverage added for markdown preservation, but tests would fail due to bug in implementation

Sequence Diagram

sequenceDiagram
    participant User
    participant CleanProcessor
    participant RegexEngine
    
    User->>CleanProcessor: clean(text, process_rule)
    
    Note over CleanProcessor: Apply default cleaning
    CleanProcessor->>RegexEngine: Remove <| and |>
    CleanProcessor->>RegexEngine: Remove control characters
    
    alt remove_urls_emails enabled
        Note over CleanProcessor: Protect markdown links/images
        CleanProcessor->>RegexEngine: Match markdown links [text](url)
        RegexEngine-->>CleanProcessor: Replace with placeholders
        CleanProcessor->>RegexEngine: Match markdown images ![alt](url)
        RegexEngine-->>CleanProcessor: Replace with placeholders
        
        Note over CleanProcessor: Remove emails and URLs
        CleanProcessor->>RegexEngine: Remove email addresses
        CleanProcessor->>RegexEngine: Remove remaining URLs
        
        Note over CleanProcessor: Restore markdown
        CleanProcessor->>CleanProcessor: Restore links from placeholders
        CleanProcessor->>CleanProcessor: Restore images from placeholders
    end
    
    alt remove_extra_spaces enabled
        CleanProcessor->>RegexEngine: Reduce multiple newlines to 2
        CleanProcessor->>RegexEngine: Reduce multiple spaces to 1
    end
    
    CleanProcessor-->>User: Return cleaned text

Loading

[greptile-apps]

_{5 files reviewed, 3 comments}

_{Edit Code Review Agent Settings | Greptile}

[greptile-apps]

syntax: pattern missing capture group for alt text - line 51 calls match.group(2) but this pattern only has one capture group (the URL)

Suggested change

	markdown_image_pattern = r"!\[.*?\]\((https?://[^)]+)\)"
	markdown_image_pattern = r"!\[(.*?)\]\((https?://[^)]+)\)"

Prompt To Fix With AI

This is a comment left during a code review.
Path: api/core/rag/cleaner/clean_processor.py
Line: 38:38

Comment:
**syntax:** pattern missing capture group for alt text - line 51 calls `match.group(2)` but this pattern only has one capture group (the URL)

```suggestion
                    markdown_image_pattern = r"!\[(.*?)\]\((https?://[^)]+)\)"
```

How can I resolve this? If you propose a fix, please make it concise.

[greptile-apps]

logic: hardcoding "image" as alt text instead of preserving the original - should use match.group(1) after fixing the pattern

Suggested change

	def replace_image_with_placeholder(match, placeholders=placeholders):
	link_type = "image"
	url = match.group(2)
	placeholder = f"__MARKDOWN_PLACEHOLDER_{len(placeholders)}__"
	placeholders.append((link_type, "image", url))
	return placeholder
	def replace_image_with_placeholder(match, placeholders=placeholders):
	link_type = "image"
	alt_text = match.group(1)
	url = match.group(2)
	placeholder = f"__MARKDOWN_PLACEHOLDER_{len(placeholders)}__"
	placeholders.append((link_type, alt_text, url))
	return placeholder

Prompt To Fix With AI

This is a comment left during a code review.
Path: api/core/rag/cleaner/clean_processor.py
Line: 49:54

Comment:
**logic:** hardcoding "image" as alt text instead of preserving the original - should use `match.group(1)` after fixing the pattern

```suggestion
                    def replace_image_with_placeholder(match, placeholders=placeholders):
                        link_type = "image"
                        alt_text = match.group(1)
                        url = match.group(2)
                        placeholder = f"__MARKDOWN_PLACEHOLDER_{len(placeholders)}__"
                        placeholders.append((link_type, alt_text, url))
                        return placeholder
```

How can I resolve this? If you propose a fix, please make it concise.

[greptile-apps]

style: __init__ added but default_rules is never used - the clean method is a classmethod that doesn't access instance attributes

Prompt To Fix With AI

This is a comment left during a code review.
Path: api/core/rag/cleaner/clean_processor.py
Line: 5:7

Comment:
**style:** `__init__` added but `default_rules` is never used - the `clean` method is a classmethod that doesn't access instance attributes

How can I resolve this? If you propose a fix, please make it concise.