"""Lambda GitHub repositories backup function module."""
import logging
import os
import re
import time
from os.path import exists

import boto3
from github3 import login
from requests.exceptions import ConnectionError
from requests.exceptions import ReadTimeout
import sentry_sdk

import config

sentry_sdk.init(
    dsn=config.SENTRY_DSN,
    traces_sample_rate=1.0,
)

# Initialize s3 client
s3 = boto3.client('s3')


def backup_repo(repo, repo_name, bucket_name):
    """Archive a single GitHub repo and upload it to S3, retrying on transient network errors."""
    for attempt in range(1, config.ARCHIVE_MAX_RETRIES + 1):
        try:
            repo.archive('zipball', path=repo_name, ref='master')

            if exists(repo_name):
                s3.upload_file(repo_name,
                               bucket_name,
                               f'{repo_name}/{repo_name}',
                               ExtraArgs={'ServerSideEncryption': 'aws:kms',
                                          'SSEKMSKeyId': config.KMS_KEY_ID})
                logging.info(f'The {repo_name} repository was archived and uploaded to S3')
                os.remove(repo_name)
            else:
                logging.info(f'The {repo_name} repository is empty')
            return

        except (ReadTimeout, ConnectionError) as e:
            if exists(repo_name):
                os.remove(repo_name)
            if attempt == config.ARCHIVE_MAX_RETRIES:
                raise
            wait = config.ARCHIVE_RETRY_WAIT * (2 ** (attempt - 1))
            logging.warning(
                f'Attempt {attempt}/{config.ARCHIVE_MAX_RETRIES} failed for {repo_name} '
                f'({type(e).__name__}), retrying in {wait}s'
            )
            time.sleep(wait)


def main():
    """Replicate all GitHub repos to s3 bucket."""
    access_token = config.GITHUB_TOKEN
    github = login(token=access_token)
    github.session.default_read_timeout = config.ARCHIVE_READ_TIMEOUT
    org = github.organization(config.GITHUB_ORGANIZATION)

    bucket_name = config.S3_BUCKET_NAME

    logging.basicConfig(
        level=logging.INFO,
        format='%(asctime)s%(levelname)s%(message)s',
        datefmt='%Y-%m-d% %H:%M:%S'
    )

    # Download and upload each repository to S3
    for i in org.repositories(type='all'):
        repo_name = re.sub('^theorchard/', '', str(i))
        backup_repo(i, repo_name, bucket_name)


if __name__ == '__main__':
    main()