SynPrune-Python / raw /negative /collect_script.py
Sheerio's picture
init
24e9bb7
import requests
import json
import ast
from datetime import datetime
import time
import os
from requests.adapters import HTTPAdapter
from requests.packages.urllib3.util.retry import Retry
# Configuration Parameters
GITHUB_TOKEN = 'YOUR-GITHUB-TOKEN'
HEADERS = {
"Authorization": f"token {GITHUB_TOKEN}",
"Accept": "application/vnd.github.v3+json"
}
# Search for active Python repositories created after 2024
SEARCH_QUERY = "language:python created:>=2024-01-01"
OUTPUT_FILE = "negative_raw.jsonl"
STATE_FILE = "crawler_state.json" # File to save crawling state
FUNCTIONS_PER_REPO = 10 # Maximum number of functions to collect per repository
MAX_RETRIES = 10 # Maximum number of retries for requests
RETRY_BACKOFF_FACTOR = 2 # Multiplier for retry waiting time
NETWORK_ERROR_SLEEP = 30 # Waiting time (seconds) after network errors
def create_session():
"""Create a requests session with retry mechanism"""
session = requests.Session()
# Set up retry strategy
retry_strategy = Retry(
total=MAX_RETRIES,
backoff_factor=RETRY_BACKOFF_FACTOR,
status_forcelist=[429, 500, 502, 503, 504, 408],
allowed_methods=["HEAD", "GET", "OPTIONS"]
)
# Apply retry strategy to HTTP and HTTPS connections
adapter = HTTPAdapter(max_retries=retry_strategy)
session.mount("https://", adapter)
session.mount("http://", adapter)
return session
def save_state(state):
"""Save current crawling state to file"""
with open(STATE_FILE, 'w') as f:
json.dump(state, f)
def load_state():
"""Load crawling state from file"""
if os.path.exists(STATE_FILE):
with open(STATE_FILE, 'r') as f:
return json.load(f)
return {
'current_page': 1,
'processed_repos': [],
'processed_files': set(),
'repo_function_counts': {}
}
def parse_functions(source_code):
"""Parse Python code and return list of functions"""
try:
tree = ast.parse(source_code)
except Exception as e:
print(f"Failed to parse code: {str(e)}")
return []
functions = []
source_lines = source_code.split('\n')
for node in ast.walk(tree):
if isinstance(node, (ast.FunctionDef, ast.AsyncFunctionDef)):
if not hasattr(node, 'end_lineno'):
continue # Skip functions where end line number cannot be obtained
start_line = node.lineno - 1
end_line = node.end_lineno
function_code = '\n'.join(source_lines[start_line:end_line])
functions.append({
'name': node.name,
'code': function_code
})
return functions
def fetch_github_api(url, params=None, session=None):
"""Send GitHub API request and handle rate limits and network exceptions"""
if not session:
session = create_session()
retries = 0
while retries < MAX_RETRIES:
try:
response = session.get(url, headers=HEADERS, params=params, timeout=60, verify=True)
if response.status_code == 403 and 'rate limit' in response.text.lower():
reset_time = int(response.headers.get('X-RateLimit-Reset', 0))
sleep_time = max(reset_time - time.time(), 0) + 5
print(f"Rate limited, waiting {sleep_time:.1f} seconds")
time.sleep(sleep_time)
continue
return response
except requests.exceptions.SSLError as e:
print(f"SSL Error: {str(e)}, attempting retry {retries+1}/{MAX_RETRIES}")
retries += 1
time.sleep(NETWORK_ERROR_SLEEP) # Wait longer after network errors
except requests.exceptions.RequestException as e:
print(f"Request Exception: {str(e)}, attempting retry {retries+1}/{MAX_RETRIES}")
retries += 1
time.sleep(NETWORK_ERROR_SLEEP) # Wait longer after network errors
print(f"Reached maximum retry attempts, skipping URL: {url}")
return None
def get_file_creation_date(repo_full_name, file_path, commit_sha, session=None):
"""Get file creation date (first commit date)"""
# Get file commit history
commits_url = f"https://api.github.com/repos/{repo_full_name}/commits?path={file_path}&per_page=100"
commits_response = fetch_github_api(commits_url, session=session)
if not commits_response or commits_response.status_code != 200:
return None
commits = commits_response.json()
if not commits:
return None
# The last commit is the first commit of the file
first_commit = commits[-1]
return first_commit['commit']['committer']['date']
def main():
# Load saved state
state = load_state()
page = state['current_page']
processed_repos = state['processed_repos']
processed_files = set(state.get('processed_files', []))
repo_function_counts = state.get('repo_function_counts', {})
session = create_session()
with open(OUTPUT_FILE, 'a') as out_file: # Open in append mode
while True:
# Search for eligible repositories
search_url = f"https://api.github.com/search/repositories?q={SEARCH_QUERY}&per_page=100&page={page}"
response = fetch_github_api(search_url, session=session)
if not response or response.status_code != 200:
print(f"Search failed, waiting {NETWORK_ERROR_SLEEP} seconds before retrying")
time.sleep(NETWORK_ERROR_SLEEP)
continue
data = response.json()
if not data.get('items'):
print(f"No search results on page {page}, exiting")
break
print(f"Processing page {page}, total {len(data['items'])} repositories")
for repo_item in data['items']:
repo_full_name = repo_item['full_name']
# Skip already processed repositories
if repo_full_name in processed_repos:
print(f"Skipping processed repository: {repo_full_name}")
continue
stars = repo_item['stargazers_count']
print(f"\nProcessing repository: {repo_full_name} (Stars: {stars})")
# Get or initialize repository function counter
repo_function_count = repo_function_counts.get(repo_full_name, 0)
# Get repository's default branch
repo_url = f"https://api.github.com/repos/{repo_full_name}"
repo_response = fetch_github_api(repo_url, session=session)
if not repo_response or repo_response.status_code != 200:
print(f"Failed to get repository information, skipping {repo_full_name}")
continue
default_branch = repo_response.json().get('default_branch', 'main')
# Get list of Python files in the repository (recursive retrieval)
contents_url = f"https://api.github.com/repos/{repo_full_name}/contents?ref={default_branch}"
stack = [contents_url]
try:
while stack:
# Check if function count limit is reached
if repo_function_count >= FUNCTIONS_PER_REPO:
print(f"Collected {repo_function_count} functions from {repo_full_name}, limit reached, moving to next repository")
break
current_url = stack.pop()
contents_response = fetch_github_api(current_url, session=session)
if not contents_response or contents_response.status_code != 200:
print(f"Failed to get directory contents, skipping {current_url}")
continue
items = contents_response.json()
for item in items:
# Check if function count limit is reached
if repo_function_count >= FUNCTIONS_PER_REPO:
print(f"Collected {repo_function_count} functions from {repo_full_name}, limit reached, moving to next repository")
break
if item['type'] == 'dir':
# Recursively process subdirectories
stack.append(item['url'])
elif item['name'].endswith('.py'):
file_path = item['path']
file_id = f"{repo_full_name}/{file_path}"
if file_id in processed_files:
continue
# Get file content
download_url = item['download_url']
file_response = fetch_github_api(download_url, session=session)
if not file_response or file_response.status_code != 200:
print(f"Failed to get file content, skipping {file_id}")
continue
# Get SHA while retrieving file content (for querying commit history)
content_data = file_response.json() if 'json' in file_response.headers.get('Content-Type', '') else {}
sha = content_data.get('sha', '')
# Get file creation date
creation_date_str = get_file_creation_date(repo_full_name, file_path, sha, session=session)
if not creation_date_str:
print(f"Failed to get file creation date, skipping {file_id}")
continue
creation_date = datetime.strptime(creation_date_str, "%Y-%m-%dT%H:%M:%SZ")
# Strictly check file creation date
if creation_date < datetime(2024, 1, 1):
print(f"File {file_id} was created on {creation_date_str}, which is before 2024-01-01, skipping")
continue
# Parse functions in the file
source_code = file_response.text
functions = parse_functions(source_code)
if not functions:
print(f"No functions found in file {file_id}")
continue
print(f"Parsed {len(functions)} functions from file {file_id}")
# Write function records and update counter
for func in functions:
if repo_function_count >= FUNCTIONS_PER_REPO:
break
record = {
'function': func['code'],
'creation_date': creation_date_str,
'repo': repo_full_name,
'file_path': file_path,
'stars': stars,
'label': 0
}
out_file.write(json.dumps(record) + '\n')
repo_function_count += 1
# Save state after every 10 functions processed
if repo_function_count % 10 == 0:
state = {
'current_page': page,
'processed_repos': processed_repos,
'processed_files': list(processed_files),
'repo_function_counts': repo_function_counts
}
save_state(state)
print(f"State saved: Collected {repo_function_count} functions from repository {repo_full_name}")
# Mark file as processed
processed_files.add(file_id)
# Save state after processing each directory
state = {
'current_page': page,
'processed_repos': processed_repos,
'processed_files': list(processed_files),
'repo_function_counts': repo_function_counts
}
save_state(state)
except KeyboardInterrupt:
print("User interruption detected, saving current state...")
state = {
'current_page': page,
'processed_repos': processed_repos,
'processed_files': list(processed_files),
'repo_function_counts': repo_function_counts
}
save_state(state)
print("State saved, program exiting. Next run will resume from the last interruption point.")
return
except Exception as e:
print(f"Unexpected error occurred while processing repository {repo_full_name}: {str(e)}")
print("Saving current state and skipping this repository...")
state = {
'current_page': page,
'processed_repos': processed_repos,
'processed_files': list(processed_files),
'repo_function_counts': repo_function_counts
}
save_state(state)
continue
# Update repository function count
repo_function_counts[repo_full_name] = repo_function_count
# Mark repository as processed
processed_repos.append(repo_full_name)
# Save state
state = {
'current_page': page,
'processed_repos': processed_repos,
'processed_files': list(processed_files),
'repo_function_counts': repo_function_counts
}
save_state(state)
print(f"Completed processing repository {repo_full_name}, collected {repo_function_count} functions in total")
# Check if there is a next page
if 'next' in response.links:
page += 1
# Save page number state
state = {
'current_page': page,
'processed_repos': processed_repos,
'processed_files': list(processed_files),
'repo_function_counts': repo_function_counts
}
save_state(state)
print(f"State saved: About to process page {page}")
else:
print(f"No next page, processing completed")
break
if __name__ == "__main__":
main()
'''
Usage Instructions
1. Run the script normally:
```bash
python collect_script.py
```
2. To start over from scratch, delete the state file:
```bash
rm crawler_state.json
```
3. After network recovery, simply re-run the script to resume from the interruption point
4. Adjust parameters:
```python
# Increase retry count or extend waiting time
MAX_RETRIES = 15
NETWORK_ERROR_SLEEP = 60
```
'''