codearchiver/codearchiver/modules/git.py

import codearchiver.core
import codearchiver.subprocess
import datetime
import functools
import logging
import os.path
import shutil
import subprocess


_logger = logging.getLogger(__name__)


class GitIndex(codearchiver.core.Index):
	fields = [
		codearchiver.core.IndexField(key = 'Based on bundle', required = False, repeatable = True),
		codearchiver.core.IndexField(key = 'Ref', required = True, repeatable = True),
		codearchiver.core.IndexField(key = 'Root commit', required = True, repeatable = True),
		codearchiver.core.IndexField(key = 'Commit', required = True, repeatable = True),
	]


class Git(codearchiver.core.Module):
	name = 'git'

	@staticmethod
	def matches(inputUrl):
		return inputUrl.url.endswith('.git')

	def __init__(self, *args, extraBranches = {}, **kwargs):
		super().__init__(*args, **kwargs)
		self._extraBranches = extraBranches

	def process(self):
		directory = self._url.rsplit('/', 1)[1]
		if os.path.exists(directory):
			_logger.fatal(f'{directory!r} already exists')
			raise FileExistsError(f'{directory!r} already exists')
		startTime = datetime.datetime.utcnow()
		if self._id is None:
			self._id = f'git_{self._url.replace("/", "_")}_{startTime:%Y%m%dT%H%M%SZ}'
		bundle = f'{self._id}.bundle'
		if os.path.exists(bundle):
			_logger.fatal(f'{bundle!r} already exists')
			raise FileExistsError(f'{bundle!r} already exists')

		_logger.info(f'Cloning {self._url} into {directory}')
		codearchiver.subprocess.run_with_log(['git', 'clone', '--verbose', '--progress', '--mirror', self._url, directory], env = {**os.environ, 'GIT_TERMINAL_PROMPT': '0'})

		if self._extraBranches:
			for branch, commit in self._extraBranches.items():
				_logger.info(f'Fetching commit {commit} as {branch}')
				r, _ = codearchiver.subprocess.run_with_log(['git', 'fetch', '--verbose', '--progress', 'origin', commit], cwd = directory, check = False)
				if r == 0:
					r2, _ = codearchiver.subprocess.run_with_log(['git', 'update-ref', f'refs/codearchiver/{branch}', commit, ''], cwd = directory, check = False)
					if r2 != 0:
						_logger.error(f'Failed to update-ref refs/codearchiver/{branch} to {commit}')
				else:
					_logger.error(f'Failed to fetch {commit}')

		_logger.info(f'Collecting repository metadata for index')
		_, refs = codearchiver.subprocess.run_with_log(['git', 'show-ref'], cwd = directory)
		_, commits = codearchiver.subprocess.run_with_log(['git', 'log', '--reflog', '--all', '--format=format:%H% P'], cwd = directory)
		commits = list(map(functools.partial(str.split, sep = ' '), commits.splitlines()))
		rootCommits = [c[0] for c in commits if len(c) == 1]

		# Check whether there are relevant prior bundles to create an incremental one
		# Collect their commits shared with this clone (else `git bundle` complains about 'bad object')
		commitSet = set(c[0] for c in commits)  # For fast lookup
		oldCommits = {}  # dict to keep the order reasonable
		basedOnBundles = {}  # ditto
		if self._storage:
			for oldBundle in self._storage.search_indices([('Root commit', c) for c in rootCommits]):
				if not oldBundle.startswith('git_'):  #TODO Is there a more generic and elegant approach?
					continue
				_logger.info(f'Previous bundle: {oldBundle!r}')
				with self._storage.open_index(oldBundle) as fp:
					idx = GitIndex.deserialise(fp)
				for key, value in idx:
					if key == 'Commit' and value in commitSet:
						oldCommits[value] = True
						basedOnBundles[oldBundle] = True

		_logger.info(f'Bundling into {bundle}')
		codearchiver.subprocess.run_with_log(['git', 'bundle', 'create', '--progress', f'../{bundle}', '--stdin', '--reflog', '--all'], cwd = directory, input = ''.join(f'^{commit}\n' for commit in oldCommits).encode('ascii'))

		_logger.info(f'Removing clone')
		shutil.rmtree(directory)

		index = GitIndex()
		for oldBundle in basedOnBundles:
			index.append('Based on bundle', oldBundle)
		for line in refs.splitlines():
			index.append('Ref', line)
		for commitHash, *parents in commits:
			if commitHash not in oldCommits:
				index.append('Commit', commitHash)
			if not parents:
				index.append('Root commit', commitHash)

		return codearchiver.core.Result(id = self._id, files = [(bundle, index)])

	def __repr__(self):
		return f'{type(self).__module__}.{type(self).__name__}({self._inputUrl!r}, extraBranches = {self._extraBranches!r})'