baseimage-tutorial-framework/lib/tfw/components/snapshot_provider.py

194 lines
5.3 KiB
Python
Raw Normal View History

# Copyright (C) 2018 Avatao.com Innovative Learning Kft.
# All Rights Reserved. See LICENSE file for details.
import re
from subprocess import run, CalledProcessError, PIPE
from getpass import getuser
from os.path import isdir
from datetime import datetime
from uuid import uuid4
from dateutil import parser as dateparser
class SnapshotProvider:
def __init__(self, directory, git_dir):
2018-07-19 13:05:43 +00:00
self._classname = self.__class__.__name__
author = f'{getuser()} via TFW {self._classname}'
self.gitenv = {
'GIT_DIR': git_dir,
'GIT_WORK_TREE': directory,
'GIT_AUTHOR_NAME': author,
'GIT_AUTHOR_EMAIL': '',
'GIT_COMMITTER_NAME': author,
'GIT_COMMITTER_EMAIL': '',
'GIT_PAGER': 'cat'
}
self._init_repo()
self.__last_valid_branch = self._branch
def _init_repo(self):
self._check_environment()
if not self._repo_is_initialized:
self._run(('git', 'init'))
if self._number_of_commits == 0:
2018-07-19 13:05:43 +00:00
try:
self._snapshot()
except CalledProcessError:
raise EnvironmentError(f'{self._classname} cannot init on empty directories!')
self._check_head_not_detached()
def _check_environment(self):
if not isdir(self.gitenv['GIT_DIR']) or not isdir(self.gitenv['GIT_WORK_TREE']):
2018-07-19 13:05:43 +00:00
raise EnvironmentError(f'{self._classname}: "directory" and "git_dir" must exist!')
@property
def _repo_is_initialized(self):
return self._run(
('git', 'status'),
check=False
).returncode == 0
@property
def _number_of_commits(self):
return int(
self._get_stdout((
'git', 'rev-list',
'--all',
'--count'
))
)
def _snapshot(self):
self._run((
'git', 'add',
'-A'
))
try:
self._get_stdout((
'git', 'commit',
'-m', 'Snapshot'
))
except CalledProcessError as err:
if b'nothing to commit, working tree clean' not in err.output:
raise
def _check_head_not_detached(self):
if self._head_detached:
2018-07-19 13:05:43 +00:00
raise EnvironmentError(f'{self._classname} cannot init from detached HEAD state!')
@property
def _head_detached(self):
return self._branch == 'HEAD'
@property
def _branch(self):
return self._get_stdout((
'git', 'rev-parse',
'--abbrev-ref', 'HEAD'
))
def _get_stdout(self, *args, **kwargs):
kwargs['stdout'] = PIPE
kwargs['stderr'] = PIPE
stdout_bytes = self._run(*args, **kwargs).stdout
return stdout_bytes.decode().rstrip('\n')
def _run(self, *args, **kwargs):
if 'check' not in kwargs:
kwargs['check'] = True
if 'env' not in kwargs:
kwargs['env'] = self.gitenv
return run(*args, **kwargs)
def take_snapshot(self):
if self._head_detached:
self._checkout_new_branch_from_head()
self._snapshot()
def _checkout_new_branch_from_head(self):
branch_name = str(uuid4())
self._run((
2018-07-19 09:31:31 +00:00
'git', 'branch',
branch_name
2018-07-19 09:11:56 +00:00
))
self._checkout(branch_name)
2018-07-19 14:15:54 +00:00
def _checkout(self, what):
self._run((
'git', 'checkout',
what
))
def restore_snapshot(self, date):
commit = self._get_commit_from_timestamp(date)
self._checkout(commit)
def _get_commit_from_timestamp(self, date):
commit = self._get_stdout((
'git', 'rev-list',
'--date=iso',
'-n', '1',
f'--before="{date.isoformat()}"',
self._last_valid_branch
))
if not commit:
commit = self._get_oldest_parent_of_head()
return commit
def _get_oldest_parent_of_head(self):
return self._get_stdout((
'git',
'rev-list',
'--max-parents=0',
'HEAD'
))
@property
def _last_valid_branch(self):
if not self._head_detached:
self.__last_valid_branch = self._branch
return self.__last_valid_branch
@property
def all_timelines(self):
return self._branches
@property
def _branches(self):
git_branch_output = self._get_stdout(('git', 'branch'))
regex_pattern = re.compile(r'(?:[^\S\n]|[*])') # matches '*' and non-newline whitespace chars
return re.sub(regex_pattern, '', git_branch_output).splitlines()
@property
def timeline(self):
2018-07-19 12:29:14 +00:00
return self._last_valid_branch
@timeline.setter
def timeline(self, value):
self._checkout(value)
@property
def snapshots(self):
return self._pretty_log_branch()
def _pretty_log_branch(self):
git_log_output = self._get_stdout((
'git', 'log',
'--pretty=%H@%aI'
))
commits = []
for line in git_log_output.splitlines():
commit_hash, timestamp = line.split('@')
commits.append({
'hash': commit_hash,
'timestamp': dateparser.parse(timestamp)
})
return commits