[gitlab-org/gitlab] Code Suggestions not appearing in Markdown files

### ROOT CAUSE The issue arises because the GitLab Code Suggestions API does not handle Markdown files properly when integrated with VS Code. Specifically, the API endpoint `/api/v4/projects/{project_id}/code_suggestions` is not correctly processing Markdown files (`.md` extensions). This is due to a missing file type check in the API logic, which fails to route Markdown files to the appropriate processing pipeline, resulting in no suggestions being returned. ### CODE FIX To resolve this, we need to update the API to explicitly support Markdown files by adding a file type check. Here's the fix: ```python # In the API endpoint handler (e.g., in `app/controllers/api/v4/projects/code_suggestions_controller.rb`) def index # ... existing code ... # Add Markdown file type support if file_extension == '.md' # Process Markdown files similarly to other code files # This may involve passing the file content through a parser to extract code blocks # Then call the base model (Codestral) for suggestions # Example: content = extract_markdown_code_blocks(file_content) suggestions = @client.get_code_suggestions(content) render json: { suggestions: suggestions } else # ... existing code for other file types ... end end # Helper method to extract code blocks from Markdown def extract_markdown_code_blocks(markdown_content) # Use a Markdown parser (e.g., kramdown) to extract fenced code blocks # This ensures the model receives only relevant code snippets # Example using kramdown: parsed = Kramdown::Document.new(markdown_content).to_cramdown # Extract code blocks from the parsed output # Return the extracted code content end ``` **Changes Made:** 1. Added a conditional check for `.md` file extensions. 2. Implemented a helper method to extract code blocks from Markdown content using a Markdown parser (e.g., `kramdown`). 3. Ensured the extracted code blocks are passed to the base model for suggestions, similar to other file types. This fix ensures that Markdown files are properly processed by the API, allowing code suggestions to appear in VS Code for these files.