LeetCode Solver

A tool to scrape LeetCode problems and generate solutions using LLMs.

Setup

1. Copy .env.example to .env and fill in your credentials:

cp .env.example .env

Required environment variables:

• GOOGLE_API_KEY: Your Google API key for using Gemini
• LEETCODE_SESSION: Your LeetCode session cookie for submitting solutions

Usage

1. Fetch LeetCode Problems

Scrape problems from LeetCode and store them locally:

# Scrape 50 problems starting from the beginning
pdm run python src/scrape_leetcode.py --limit 50

# Skip first 100 problems and scrape next 50
pdm run python src/scrape_leetcode.py --skip 100 --limit 50

# Scrape without using Gemini for test case extraction
pdm run python src/scrape_leetcode.py --no-gemini

Problems will be saved to data/problems/ directory.

2. Generate and Submit Solutions

Generate solutions using LLM and optionally submit them to LeetCode:

# Generate solution for problem #50
pdm run python src/solve_leetcode.py --start 50

# Solve problems 50 through 55
pdm run python src/solve_leetcode.py --start 50 --end 55

# Continue on errors when solving multiple problems
pdm run python src/solve_leetcode.py --start 50 --end 55 --continue-on-error

# Use a different example problem (default is Two Sum, problem #1)
pdm run python src/solve_leetcode.py --start 50 --example-id 2

# Save solution to a custom directory
pdm run python src/solve_leetcode.py --start 50 --save-dir ./my_solutions

# Generate without saving to file
pdm run python src/solve_leetcode.py --start 50 --no-save

# Generate and submit to LeetCode
pdm run python src/solve_leetcode.py --start 50 --submit

Solutions will be saved to data/solutions/ directory by default.

Directory Structure

leetcode-solver/
├── data/
│   ├── problems/     # Scraped problem data
│   └── solutions/    # Generated solutions
├── src/
│   ├── leetcode_solver/
│   │   ├── gemini_parser.py
│   │   ├── llm_manager.py
│   │   ├── llm_solver.py
│   │   └── scraper.py
│   ├── scrape_leetcode.py
│   └── solve_leetcode.py
├── .env
└── README.md

Features

• Scrape LeetCode problems with test cases
• Use Gemini for enhanced test case extraction
• Generate solutions using LLM with example-based learning
• Save solutions with problem descriptions and metadata
• Submit solutions directly to LeetCode
• Configurable LLM models and parameters