MohamedHamed19m/capl-docs-parser-mcp

Vector Documentation Parser

A Python package for extracting structured function information from Vector (CAPL) documentation in Markdown format and MCP Server to utilize this parser and offering tools for AI agents.

Overview

This tool parses Vector documentation files (such as CAPL function md docs) and extracts detailed function information into structured Python class objects. It automates the process of converting Markdown-based documentation into data that can be further processed, analyzed, or integrated into other systems.

Setup

This project includes a setup script (setup_mcp.py) to manage the MCP server configuration, allowing for both local (project-specific) and global installations in gemini settings.json (in future will support other agents like claude).

Usage:

# Install globally (checks existing config, appends or overwrites)
python setup_mcp.py install --global

# Install locally for this project only
python setup_mcp.py install --local

# Check current status of both local and global configs
python setup_mcp.py status

# Remove from global config (keeps other servers)
python setup_mcp.py uninstall --global

# Remove local config completely
python setup_mcp.py uninstall --local

Run python setup_mcp.py --help for more details on commands and options.

Environment Setup with uv

This project uses uv for efficient dependency management and virtual environment creation.

1. Create Virtual Environment:

   uv venv
   

2. Activate Environment:

   • macOS/Linux:

     source .venv/bin/activate
     

   • Windows (PowerShell):

     .venv\Scripts\Activate.ps1
     

3. Synchronize Dependencies and Create Lock File:

   uv sync
   

   This command reads pyproject.toml, creates uv.lock (if it doesn't exist), and installs dependencies into .venv. Commit uv.lock to Git.

4. Regenerate Lock File: If pyproject.toml is manually changed or branches are switched:

   uv lock
   

5. Deactivate Environment:

   deactivate
   

Features

• Comprehensive Markdown Parsing: Extracts function names, multiple syntax forms, descriptions, parameters, return values, code examples, and "Valid for" information.
• Structured Output: Stores all extracted data in well-defined Python dataclasses (FunctionInfo, Parameter).
• Handles Complex Docs: Supports multiple syntax forms, preserves code examples, and robustly parses various Markdown section formats.
• Batch Processing: Can parse a single file or all Markdown files in a directory.
• Pretty Printing: Includes __str__ methods for easy inspection of parsed data.
• Error Handling: Designed to handle malformed or incomplete documentation gracefully.

Main Components

• FunctionInfo class: Stores all extracted information for a function, including:
  • Function name
  • Multiple syntax forms
  • Description
  • Parameters (with names and descriptions)
  • Return values
  • Example code
  • "Valid for" information
• Parameter class: Represents individual parameters with name and description.
• VectorDocParser class: Handles the actual parsing logic for files and directories.

Usage

Parse a single file

parser = VectorDocParser()
func_info = parser.parse_file("path/to/file.md")

Parse all markdown files in a directory

functions = parser.parse_directory("./vector_docs")

Command-line usage

Run the script directly to parse a file or directory:

python main.py <file_path_or_directory>

• To parse a single file: python main.py path/to/file.md
• To parse a directory: python main.py path/to/directory/

Requirements

• Python 3.7+
• No external dependencies (uses only the Python standard library)

Input Format

• Markdown files documenting Vector (CAPL) functions, with sections for syntax, parameters, return values, examples, etc.

Output

• Structured Python objects containing all extracted information, ready for further processing or integration.

Project Structure

• main.py — Entry point for the parser (CLI and importable)
• setup_mcp.py — Setup script to configure the MCP server.
• mcp_app/ — Contains the MCP server application.
  • MCP_Server.py — Hosts the parsing logic as an MCP server for Gemini.
• src/ — Core Python source code for parsing and searching.
• docs/ — Detailed documentation files.
• inputs/ — Example input Markdown files.
• README.md — This file.

Documentation

This project includes an MCP (Model Context Protocol) server that exposes the documentation parsing and search functionality as a set of tools for AI agents.

For more details, see the documentation below:

• : Explains when and why to use this MCP server.
• : Provides detailed information on the server's tools and how to connect to it.

📘 Example CAPL Documentation

If you want to explore additional CAPL function documentation examples or test Markdown files, you can refer to the following public repository:

🔗 Sylphith/capl_docs — AUTOSAR Ethernet IL CAPL Functions [https://github.com/Sylphith/capl_docs/blob/main/IP/AUTOSARethIL/CAPLfunctionsAREthILOverview.md].

This repo contains real-world CAPL .md documentation files that you can use to test or validate the parser.

This project is maintained by MohamedHamed19m.