update with better documentation and ability to search by repo name

This commit is contained in:
kenna-bmcdevitt 2025-01-21 10:10:12 -06:00
parent 2eb947bb0f
commit a0c81591eb
3 changed files with 58 additions and 3 deletions

View file

@ -81,5 +81,5 @@ python github_searcher.py --query QUERY --search_type SEARCH_TYPE [--get_file_co
6. Use Docker Compose to search for a specific CVE-ID and output results to a file:
```sh
docker-compose run --rm app python3 /usr/src/app/github_searcher.py --query=CVE-2024-5932 --search_type=in-repo-name --get_open_issues --get_open_pull_requests --get_file_contents --json > cve-2024-5932
docker-compose run --rm github_searcher python3 /usr/src/app/github_searcher.py --query=CVE-2024-5932 --search_type=in-repo-name --get_open_issues --get_open_pull_requests --get_file_contents --json > cve-2024-5932
```

View file

@ -1,4 +1,4 @@
services:
app:
github_searcher:
build: .
env_file: ".env"

View file

@ -15,6 +15,43 @@ def auth():
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()
@ -42,6 +79,13 @@ class GithubSearcher():
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')
@ -49,6 +93,13 @@ class GithubSearcher():
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')
@ -69,7 +120,7 @@ class GithubSearcher():
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"], help="The type of search to perform: 'repo', 'users', or 'in-repo-name'.")
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.")
@ -85,6 +136,8 @@ def main():
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:
@ -106,10 +159,12 @@ def main():
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: