github_searcher/github_searcher.py

import os
import argparse
import logging
import json
from github import Github, Auth, GithubException

# Configure logging
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')

# authentication
def auth():
    access_token = os.getenv("GITHUB_ACCESS_TOKEN")
    if not access_token:
        raise ValueError("GITHUB_ACCESS_TOKEN environment variable not set")
    auth = Auth.Token(access_token)
    g = Github(auth=auth)
    return g

"""
This class is used to search GitHub repositories to gather information about repos and users.
- search_repo: Search for repositories based on the query.
- search_users: Search for users based on the query.
- search_in_repo_name: Search for repositories based on the query in the repository name.
- search_by_repo_name: Search for repositories based on the query in the repository name.
- get_repo_open_issues: Get the open issues of a repository.
- get_repo_stars: Get the number of stars of a repository.
- get_repo_open_pull_requests: Get the open pull requests of a repository.
- get_repo_contents: Get the contents of a repository.
- get_result: Get the search result.
- query: The search query.
- result: The search result.
- g: The GitHub object.

Example usage:
# Searching for a CVE-ID in repositories:
searcher = GithubSearcher("CVE-2021-1234") # Initialize the searcher with the query
searcher.search_repo() # Search for repositories based on the query. This would search github for repositories with the query "CVE-2021-1234"
result = searcher.get_result() # Get the search result

# Searching for a specific repo by name:
searcher = GithubSearcher("php/php-src") # Initialize the searcher with the query
searcher.search_by_repo_name() # Search for repositories based on the query in the repository name. This would search github for the repository "php/php-src"
result = searcher.get_result() # Get the search result

# Getting open issues of a repository:
repo = result[0] # Get the first repository from the search result
open_issues = searcher.get_repo_open_issues(repo) # Get the open issues of the repository
for issue in open_issues:
    print(issue.title) # Print the title of the issue

# Getting the number of stars of a repository:
repo = result[0] # Get the first repository from the search result
stars = searcher.get_repo_stars(repo) # Get the number of stars of the repository

"""
class GithubSearcher():
    def __init__(self, query):
        self.g = auth()
        self.query = query
        self.result = None

    def search_repo(self):
        try:
            self.result = self.g.search_repositories(self.query)
        except GithubException as e:
            logging.error(f"Error searching repositories: {e}")
            self.result = None

    def search_users(self):
        try:
            self.result = self.g.search_users(self.query)
        except GithubException as e:
            logging.error(f"Error searching users: {e}")
            self.result = None

    def search_in_repo_name(self):
        try:
            self.result = self.g.search_repositories('in:name ' + self.query)
        except GithubException as e:
            logging.error(f"Error searching in name: {e}")
            self.result = None

    def search_by_repo_name(self):
        try:
            self.result = self.g.search_repositories(f'repo:{self.query}')
        except GithubException as e:
            logging.error(f"Error searching by repo name: {e}")
            self.result = None

    def get_repo_open_issues(self, repo):
        try:
            return repo.get_issues(state='open')
        except GithubException as e:
            logging.error(f"Error getting open issues: {e}")
            return None

    def get_repo_stars(self, repo):
        try:
            return repo.stargazers_count
        except GithubException as e:
            logging.error(f"Error getting stars: {e}")
            return None

    def get_repo_open_pull_requests(self, repo):
        try:
            return repo.get_pulls(state='open')
        except GithubException as e:
            logging.error(f"Error getting open pull requests: {e}")
            return None

    def get_repo_contents(self, repo):
        try:
            return repo.get_contents("")
        except GithubException as e:
            logging.error(f"Error getting repository contents: {e}")
            return None

    def get_result(self):
        return self.result

def main():
    parser = argparse.ArgumentParser(description="Search GitHub repositories and users for PoC exploits and CVEs.")
    parser.add_argument("--query", type=str, required=True, help="The search query.")
    parser.add_argument("--search_type", type=str, required=True, choices=["repo", "users", "in-repo-name", "by-repo-name"], help="The type of search to perform: 'repo', 'users', 'in-repo-name', or 'by-repo-name'.")
    parser.add_argument("--get_file_contents", action="store_true", help="Get the contents of repo results.")
    parser.add_argument("--get_open_issues", action="store_true", help="Get the open issues of repo results.")
    parser.add_argument("--get_open_pull_requests", action="store_true", help="Get the open pull requests of repo results.")
    parser.add_argument("--json", action="store_true", help="Output results in JSON format.")

    args = parser.parse_args()

    searcher = GithubSearcher(args.query)

    if args.search_type == "repo":
        searcher.search_repo()
    elif args.search_type == "users":
        searcher.search_users()
    elif args.search_type == "in-repo-name":
        searcher.search_in_repo_name()
    elif args.search_type == "by-repo-name":
        searcher.search_by_repo_name()

    result = searcher.get_result()
    if result is None:
        print("No results found.")
        return

    output = []
    for item in result:
        if args.search_type == "users":
            user_repos = item.get_repos()

            for repo in user_repos:
                repo_info = {"repo_url": repo.html_url, "repo_name": repo.name, "repo_description": repo.description}
                output.append(repo_info)
        else:
            repo_info = {"repo_url": item.html_url, "repo_name": item.name, "repo_description": item.description}

            if args.get_open_issues:
                open_issues = searcher.get_repo_open_issues(item)
                if open_issues:
                    repo_info["open_issues"] = [{"title": issue.title, "url": issue.html_url} for issue in open_issues]
                    repo_info['total_open_issues'] = item.open_issues_count
            if args.get_open_pull_requests:
                open_prs = searcher.get_repo_open_pull_requests(item)
                if open_prs:
                    repo_info["open_pull_requests"] = [{"title": pr.title, "url": pr.html_url} for pr in open_prs]
                    repo_info["total_open_pull_requests"] = item.open_pulls_count
            if args.get_file_contents:
                contents = searcher.get_repo_contents(item)
                if contents:
                    repo_info["contents"] = [
                        {
                            "name": content_file.name,
                            "sha": content_file.sha,
                            "size": content_file.size,
                            "encoding": content_file.encoding,
                            "html_url": content_file.html_url
                        } for content_file in contents
                    ]

            output.append(repo_info)

    if args.json:
        print(json.dumps(output, indent=4))
    else:
        for item in output:
            print(item)

if __name__ == "__main__":
    main()