| Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 1 | #-*- coding: utf-8 -*- | 
|  | 2 |  | 
|  | 3 | import datetime | 
|  | 4 | import os | 
|  | 5 |  | 
|  | 6 | import utils | 
|  | 7 |  | 
|  | 8 | from rsync_remote import RsyncRemote | 
|  | 9 | from utils import singleton | 
|  | 10 |  | 
|  | 11 |  | 
|  | 12 | @singleton | 
|  | 13 | class TimeStamp(object): | 
|  | 14 | def __init__(self): | 
|  | 15 | self.now = datetime.datetime.utcnow() | 
|  | 16 | self.staging_snapshot_stamp_format = r'%Y-%m-%d-%H%M%S' | 
|  | 17 | self.staging_snapshot_stamp_regexp = \ | 
|  | 18 | r'[0-9]{4}-[0-9]{2}-[0-9]{2}-[0-9]{6}' | 
|  | 19 | self.staging_snapshot_stamp = \ | 
|  | 20 | self.now.strftime(self.staging_snapshot_stamp_format) | 
|  | 21 |  | 
|  | 22 | def __str__(self): | 
|  | 23 | return self.staging_snapshot_stamp | 
|  | 24 |  | 
|  | 25 |  | 
|  | 26 | class RsyncVersioned(RsyncRemote): | 
|  | 27 | # retry and other function with mirror | 
|  | 28 | # add all the needed directory functions here, like mkdir, ls, rm etc | 
|  | 29 | # possible check that rsync url is exists | 
|  | 30 | def __init__(self, | 
|  | 31 | rsync_url, | 
| Max Rasskazov | 0127162 | 2015-06-17 02:35:09 +0300 | [diff] [blame] | 32 | snapshot_dir='snapshots', | 
|  | 33 | latest_successful_postfix='latest', | 
| Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 34 | save_latest_days=14, | 
|  | 35 | init_directory_structure=True, | 
|  | 36 | ): | 
|  | 37 | super(RsyncVersioned, self).__init__(rsync_url) | 
|  | 38 | self.logger = utils.logger.getChild('RsyncVersioned.' + rsync_url) | 
|  | 39 | self.timestamp = TimeStamp() | 
|  | 40 | self.logger.info('Using timestamp {}'.format(self.timestamp)) | 
| Max Rasskazov | 46cc073 | 2015-06-05 19:23:24 +0300 | [diff] [blame] | 41 | self.snapshot_dir = self.url.a_dir(snapshot_dir) | 
| Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 42 | self.latest_successful_postfix = latest_successful_postfix | 
|  | 43 | self.save_latest_days = save_latest_days | 
|  | 44 |  | 
| Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 45 | if init_directory_structure is True: | 
|  | 46 | self.init_directory_structure() | 
|  | 47 |  | 
|  | 48 | def init_directory_structure(self): | 
|  | 49 | # TODO: self.rsyncRemote.mkdir | 
| Max Rasskazov | 0127162 | 2015-06-17 02:35:09 +0300 | [diff] [blame] | 50 | if self.url.url_type != 'path': | 
|  | 51 | server_root = RsyncRemote(self.url.root) | 
|  | 52 | return server_root.mkdir( | 
|  | 53 | self.url.a_dir(self.url.path, self.snapshot_dir) | 
|  | 54 | ) | 
| Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 55 |  | 
|  | 56 | def push(self, source, repo_name, extra=None): | 
| Max Rasskazov | 46cc073 | 2015-06-05 19:23:24 +0300 | [diff] [blame] | 57 | latest_path = self.url.a_file( | 
| Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 58 | self.snapshot_dir, | 
| Max Rasskazov | 46cc073 | 2015-06-05 19:23:24 +0300 | [diff] [blame] | 59 | '{}-{}'.format(self.url.a_file(repo_name), | 
| Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 60 | self.latest_successful_postfix) | 
|  | 61 | ) | 
| Max Rasskazov | 46cc073 | 2015-06-05 19:23:24 +0300 | [diff] [blame] | 62 | snapshot_name = self.url.a_file( | 
|  | 63 | '{}-{}'.format(self.url.a_file(repo_name), self.timestamp) | 
| Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 64 | ) | 
| Max Rasskazov | 46cc073 | 2015-06-05 19:23:24 +0300 | [diff] [blame] | 65 | repo_path = self.url.a_file(self.snapshot_dir, snapshot_name) | 
| Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 66 |  | 
| Max Rasskazov | 854399e | 2015-06-05 16:35:17 +0300 | [diff] [blame] | 67 | extra = '--link-dest={}'.format( | 
| Max Rasskazov | 46cc073 | 2015-06-05 19:23:24 +0300 | [diff] [blame] | 68 | self.url.a_file(self.url.path, latest_path) | 
| Max Rasskazov | 854399e | 2015-06-05 16:35:17 +0300 | [diff] [blame] | 69 | ) | 
| Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 70 | result = super(RsyncVersioned, self).push(source, repo_path, extra) | 
| Max Rasskazov | 17f4a6c | 2015-06-17 02:40:59 +0300 | [diff] [blame^] | 71 | self.symlink(repo_name, repo_path) | 
|  | 72 | self.symlink(latest_path, snapshot_name) | 
| Max Rasskazov | 452138b | 2015-06-17 02:37:34 +0300 | [diff] [blame] | 73 | self._remove_old_snapshots(repo_name) | 
| Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 74 | return result | 
| Max Rasskazov | 452138b | 2015-06-17 02:37:34 +0300 | [diff] [blame] | 75 |  | 
|  | 76 | def _remove_old_snapshots(self, repo_name, save_latest_days=None): | 
|  | 77 | if save_latest_days is None: | 
|  | 78 | save_latest_days = self.save_latest_days | 
|  | 79 | if save_latest_days is None or save_latest_days is False: | 
|  | 80 | # delete all snapshots | 
|  | 81 | self.logger.info('Deletion all of the old snapshots ' | 
|  | 82 | '(save_latest_days == {})' | 
|  | 83 | ''.format(save_latest_days)) | 
|  | 84 | save_latest_days = -1 | 
|  | 85 | elif save_latest_days == 0: | 
|  | 86 | # skipping deletion | 
|  | 87 | self.logger.info('Skip deletion of old snapshots ' | 
|  | 88 | '(save_latest_days == {})' | 
|  | 89 | ''.format(save_latest_days)) | 
|  | 90 | return | 
|  | 91 | else: | 
|  | 92 | # delete snapshots older than | 
|  | 93 | self.logger.info('Deletion all of the unlinked snapshots older ' | 
|  | 94 | 'than {0} days (save_latest_days == {0})' | 
|  | 95 | ''.format(save_latest_days)) | 
|  | 96 | warn_date = \ | 
|  | 97 | self.timestamp.now - datetime.timedelta(days=save_latest_days) | 
|  | 98 | warn_date = datetime.datetime.combine(warn_date, datetime.time(0)) | 
|  | 99 | snapshots = self.ls_dirs( | 
|  | 100 | self.url.a_dir(self.snapshot_dir), | 
|  | 101 | pattern=r'^{}-{}$'.format( | 
|  | 102 | repo_name, | 
|  | 103 | self.timestamp.staging_snapshot_stamp_regexp | 
|  | 104 | ) | 
|  | 105 | ) | 
|  | 106 | links = self.ls_symlinks(self.url.a_dir()) | 
|  | 107 | links += self.ls_symlinks(self.url.a_dir(self.snapshot_dir)) | 
|  | 108 | for s in snapshots: | 
|  | 109 | s_date = datetime.datetime.strptime( | 
|  | 110 | s, | 
|  | 111 | '{}-{}'.format(repo_name, | 
|  | 112 | self.timestamp.staging_snapshot_stamp_format) | 
|  | 113 | ) | 
|  | 114 | s_date = datetime.datetime.combine(s_date, datetime.time(0)) | 
|  | 115 | s_path = self.url.a_dir(self.snapshot_dir, s) | 
|  | 116 | if s_date < warn_date: | 
|  | 117 | s_links = [_[0] for _ in links | 
|  | 118 | if _[1] == s | 
|  | 119 | or _[1].endswith('/{}'.format(s)) | 
|  | 120 | ] | 
|  | 121 | if not s_links: | 
|  | 122 | self.rmdir(s_path) | 
|  | 123 | else: | 
|  | 124 | self.logger.info('Skip deletion of "{}" because there are ' | 
|  | 125 | 'symlinks found: {}'.format(s, s_links)) | 
|  | 126 | else: | 
|  | 127 | self.logger.info('Skip deletion of "{}" because it newer than ' | 
|  | 128 | '{} days'.format(s, save_latest_days)) |