diff --git a/swh/loader/svn/loader.py b/swh/loader/svn/loader.py index 3b0c087..8eeddaa 100644 --- a/swh/loader/svn/loader.py +++ b/swh/loader/svn/loader.py @@ -1,749 +1,756 @@ # Copyright (C) 2015-2018 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information """Loader in charge of injecting either new or existing svn mirrors to swh-storage. """ import os import pty import re import shutil import tempfile from mmap import mmap, ACCESS_WRITE from subprocess import Popen from swh.model import hashutil from swh.model.from_disk import Directory from swh.model.identifiers import identifier_to_bytes, revision_identifier from swh.model.identifiers import snapshot_identifier from swh.loader.core.loader import BufferedLoader from swh.loader.core.utils import clean_dangling_folders +from swh.storage.algos.snapshot import snapshot_get_all_branches from . import svn, converters from .utils import ( init_svn_repo_from_dump, init_svn_repo_from_archive_dump, OutputStream ) from .exception import SvnLoaderUneventful from .exception import SvnLoaderHistoryAltered DEFAULT_BRANCH = b'HEAD' def _revision_id(revision): return identifier_to_bytes(revision_identifier(revision)) def build_swh_snapshot(revision_id, branch=DEFAULT_BRANCH): - """Build a swh snapshot from the revision id, origin id, and visit. + """Build a swh snapshot from the revision id, origin url, and visit. """ return { 'id': None, 'branches': { branch: { 'target': revision_id, 'target_type': 'revision', } } } TEMPORARY_DIR_PREFIX_PATTERN = 'swh.loader.svn.' class SvnLoader(BufferedLoader): """Swh svn loader. The repository is either remote or local. The loader deals with update on an already previously loaded repository. """ CONFIG_BASE_FILENAME = 'loader/svn' ADDITIONAL_CONFIG = { 'temp_directory': ('str', '/tmp'), 'debug': ('bool', False), # NOT FOR PRODUCTION, False for production 'check_revision': ('dict', { 'status': False, # do we check the revision? 'limit': 1000, # at which pace do we check it? }), } def __init__(self): super().__init__(logging_class='swh.loader.svn.SvnLoader') - self.origin_id = None + self.origin_url = None self.debug = self.config['debug'] self.last_seen_revision = None self.temp_directory = self.config['temp_directory'] self.done = False self.svnrepo = None # Revision check is configurable check_revision = self.config['check_revision'] if check_revision['status']: self.check_revision = check_revision['limit'] else: self.check_revision = None # internal state used to store swh objects self._contents = [] self._directories = [] self._revisions = [] self._snapshot = None self._last_revision = None self._visit_status = 'full' self._load_status = 'uneventful' def pre_cleanup(self): """Cleanup potential dangling files from prior runs (e.g. OOM killed tasks) """ clean_dangling_folders(self.temp_directory, pattern_check=TEMPORARY_DIR_PREFIX_PATTERN, log=self.log) def cleanup(self): """Clean up the svn repository's working representation on disk. """ if not self.svnrepo: # could happen if `prepare` fails return if self.debug: self.log.error('''NOT FOR PRODUCTION - debug flag activated Local repository not cleaned up for investigation: %s''' % ( self.svnrepo.local_url.decode('utf-8'), )) return self.svnrepo.clean_fs() def swh_revision_hash_tree_at_svn_revision(self, revision): """Compute and return the hash tree at a given svn revision. Args: rev (int): the svn revision we want to check Returns: The hash tree directory as bytes. """ local_dirname, local_url = self.svnrepo.export_temporary(revision) h = Directory.from_disk(path=local_url).hash self.svnrepo.clean_fs(local_dirname) return h - def get_svn_repo(self, svn_url, local_dirname, origin_id): + def get_svn_repo(self, svn_url, local_dirname, origin_url): """Instantiates the needed svnrepo collaborator to permit reading svn repository. Args: svn_url (str): the svn repository url to read from local_dirname (str): the local path on disk to compute data - origin_id (int): the corresponding origin id + origin_url (str): the corresponding origin url Returns: Instance of :mod:`swh.loader.svn.svn` clients """ return svn.SvnRepo(svn_url, - local_dirname=local_dirname, origin_id=origin_id) + local_dirname=local_dirname, origin_url=origin_url) - def swh_latest_snapshot_revision(self, origin_id, + def swh_latest_snapshot_revision(self, origin_url, previous_swh_revision=None): """Look for latest snapshot revision and returns it if any. Args: - origin_id (int): Origin identifier + origin_url (str): Origin identifier previous_swh_revision: (optional) id of a possible previous swh revision Returns: dict: The latest known point in time. Dict with keys: 'revision': latest visited revision 'snapshot': latest snapshot If None is found, return an empty dict. """ storage = self.storage if not previous_swh_revision: # check latest snapshot's revision - latest_snap = storage.snapshot_get_latest(origin_id) - if latest_snap: - branches = latest_snap.get('branches') - if not branches: + visit = storage.origin_visit_get_latest( + origin_url, require_snapshot=True) + if visit: + latest_snap = snapshot_get_all_branches( + storage, visit['snapshot']) + if latest_snap: + branches = latest_snap.get('branches') + if not branches: + return {} + branch = branches.get(DEFAULT_BRANCH) + if not branch: + return {} + target_type = branch['target_type'] + if target_type != 'revision': + return {} + previous_swh_revision = branch['target'] + else: return {} - branch = branches.get(DEFAULT_BRANCH) - if not branch: - return {} - target_type = branch['target_type'] - if target_type != 'revision': - return {} - previous_swh_revision = branch['target'] else: return {} if isinstance(previous_swh_revision, dict): swh_id = previous_swh_revision['id'] else: swh_id = previous_swh_revision revs = list(storage.revision_get([swh_id])) if revs: return { 'snapshot': latest_snap, 'revision': revs[0] } return {} def build_swh_revision(self, rev, commit, dir_id, parents): """Build the swh revision dictionary. This adds: - the `'synthetic`' flag to true - the '`extra_headers`' containing the repository's uuid and the svn revision number. Args: rev (dict): the svn revision commit (dict): the commit metadata dir_id (bytes): the upper tree's hash identifier parents ([bytes]): the parents' identifiers Returns: The swh revision corresponding to the svn revision. """ return converters.build_swh_revision( rev, commit, self.svnrepo.uuid, dir_id, parents) def check_history_not_altered(self, svnrepo, revision_start, swh_rev): """Given a svn repository, check if the history was not tampered with. """ revision_id = swh_rev['id'] parents = swh_rev['parents'] hash_data_per_revs = svnrepo.swh_hash_data_at_revision(revision_start) rev = revision_start rev, _, commit, _, root_dir = list(hash_data_per_revs)[0] dir_id = root_dir.hash swh_revision = self.build_swh_revision(rev, commit, dir_id, parents) swh_revision_id = _revision_id(swh_revision) return swh_revision_id == revision_id def _init_from(self, partial_swh_revision, previous_swh_revision): """Function to determine from where to start from. Args: partial_swh_revision (dict): A known revision from which the previous loading did not finish. known_previous_revision (dict): A known revision from which the previous loading did finish. Returns: The revision from which to start or None if nothing (fresh start). """ if partial_swh_revision and not previous_swh_revision: return partial_swh_revision if not partial_swh_revision and previous_swh_revision: return previous_swh_revision if partial_swh_revision and previous_swh_revision: # will determine from which to start from extra_headers1 = dict( partial_swh_revision['metadata']['extra_headers']) extra_headers2 = dict( previous_swh_revision['metadata']['extra_headers']) rev_start1 = int(extra_headers1['svn_revision']) rev_start2 = int(extra_headers2['svn_revision']) if rev_start1 <= rev_start2: return previous_swh_revision return partial_swh_revision return None def start_from(self, last_known_swh_revision=None, start_from_scratch=False): """Determine from where to start the loading. Args: last_known_swh_revision (dict): Last know swh revision or None start_from_scratch (bool): To start loading from scratch or not Returns: tuple (revision_start, revision_end, revision_parents) Raises: SvnLoaderHistoryAltered: When a hash divergence has been detected (should not happen) SvnLoaderUneventful: Nothing changed since last visit """ revision_head = self.svnrepo.head_revision() if revision_head == 0: # empty repository case revision_start = 0 revision_end = 0 else: # default configuration revision_start = self.svnrepo.initial_revision() revision_end = revision_head revision_parents = { revision_start: [] } if not start_from_scratch: # Check if we already know a previous revision for that origin if self.latest_snapshot: swh_rev = self.latest_snapshot['revision'] else: swh_rev = None # Determine from which known revision to start swh_rev = self._init_from(last_known_swh_revision, previous_swh_revision=swh_rev) if swh_rev: # Yes, we know a previous revision. Try and update it. extra_headers = dict(swh_rev['metadata']['extra_headers']) revision_start = int(extra_headers['svn_revision']) revision_parents = { revision_start: swh_rev['parents'], } self.log.debug('svn export --ignore-keywords %s@%s' % ( self.svnrepo.remote_url, revision_start)) if swh_rev and not self.check_history_not_altered( self.svnrepo, revision_start, swh_rev): msg = 'History of svn %s@%s altered. ' \ 'Skipping...' % ( self.svnrepo.remote_url, revision_start) raise SvnLoaderHistoryAltered(msg) # now we know history is ok, we start at next revision revision_start = revision_start + 1 # and the parent become the latest know revision for # that repository revision_parents[revision_start] = [swh_rev['id']] if revision_start > revision_end and revision_start != 1: msg = '%s@%s already injected.' % (self.svnrepo.remote_url, revision_end) raise SvnLoaderUneventful(msg) self.log.info('Processing revisions [%s-%s] for %s' % ( revision_start, revision_end, self.svnrepo)) return revision_start, revision_end, revision_parents def _check_revision_divergence(self, count, rev, dir_id): """Check for hash revision computation divergence. The Rationale behind this is that svn can trigger unknown edge cases (mixed CRLF, svn properties, etc...). Those are not always easy to spot. Adding a check will help in spotting missing edge cases. Args: count (int): The number of revisions done so far rev (dict): The actual revision we are computing from dir_id (bytes): The actual directory for the given revision Returns: False if no hash divergence detected Raises ValueError if a hash divergence is detected """ if (count % self.check_revision) == 0: # hash computation check self.log.debug('Checking hash computations on revision %s...' % rev) checked_dir_id = self.swh_revision_hash_tree_at_svn_revision( rev) if checked_dir_id != dir_id: err = 'Hash tree computation divergence detected ' \ '(%s != %s), stopping!' % ( hashutil.hash_to_hex(dir_id), hashutil.hash_to_hex(checked_dir_id)) raise ValueError(err) def process_svn_revisions(self, svnrepo, revision_start, revision_end, revision_parents): """Process svn revisions from revision_start to revision_end. At each svn revision, apply new diffs and simultaneously compute swh hashes. This yields those computed swh hashes as a tuple (contents, directories, revision). Note that at every `self.check_revision`, a supplementary check takes place to check for hash-tree divergence (related T570). Yields: tuple (contents, directories, revision) of dict as a dictionary with keys, sha1_git, sha1, etc... Raises: ValueError in case of a hash divergence detection """ gen_revs = svnrepo.swh_hash_data_per_revision( revision_start, revision_end) swh_revision = None count = 0 for rev, nextrev, commit, new_objects, root_directory in gen_revs: count += 1 # Send the associated contents/directories _contents = new_objects.get('content', {}).values() _directories = new_objects.get('directory', {}).values() # compute the fs tree's checksums dir_id = root_directory.hash swh_revision = self.build_swh_revision( rev, commit, dir_id, revision_parents[rev]) swh_revision['id'] = _revision_id(swh_revision) self.log.debug('rev: %s, swhrev: %s, dir: %s' % ( rev, hashutil.hash_to_hex(swh_revision['id']), hashutil.hash_to_hex(dir_id))) if self.check_revision: self._check_revision_divergence(count, rev, dir_id) if nextrev: revision_parents[nextrev] = [swh_revision['id']] yield _contents, _directories, swh_revision def prepare_origin_visit(self, *, svn_url, visit_date=None, origin_url=None, **kwargs): self.origin = { 'url': origin_url if origin_url else svn_url, 'type': 'svn', } self.visit_date = visit_date def prepare(self, *, svn_url, destination_path=None, swh_revision=None, start_from_scratch=False, **kwargs): self.start_from_scratch = start_from_scratch if swh_revision: self.last_known_swh_revision = swh_revision else: self.last_known_swh_revision = None self.latest_snapshot = self.swh_latest_snapshot_revision( - self.origin_id, self.last_known_swh_revision) + self.origin_url, self.last_known_swh_revision) if destination_path: local_dirname = destination_path else: local_dirname = tempfile.mkdtemp( suffix='-%s' % os.getpid(), prefix=TEMPORARY_DIR_PREFIX_PATTERN, dir=self.temp_directory) self.svnrepo = self.get_svn_repo( - svn_url, local_dirname, self.origin_id) + svn_url, local_dirname, self.origin_url) try: revision_start, revision_end, revision_parents = self.start_from( self.last_known_swh_revision, self.start_from_scratch) self.swh_revision_gen = self.process_svn_revisions( self.svnrepo, revision_start, revision_end, revision_parents) except SvnLoaderUneventful as e: self.log.warning(e) if self.latest_snapshot and 'snapshot' in self.latest_snapshot: self._snapshot = self.latest_snapshot['snapshot'] self.done = True except SvnLoaderHistoryAltered as e: self.log.error(e) self.done = True self._visit_status = 'partial' def fetch_data(self): """Fetching svn revision information. This will apply svn revision as patch on disk, and at the same time, compute the swh hashes. In effect, fetch_data fetches those data and compute the necessary swh objects. It's then stored in the internal state instance variables (initialized in `_prepare_state`). This is up to `store_data` to actually discuss with the storage to store those objects. Returns: bool: True to continue fetching data (next svn revision), False to stop. """ data = None if self.done: return False try: data = next(self.swh_revision_gen) self._load_status = 'eventful' except StopIteration: self.done = True self._visit_status = 'full' return False # Stopping iteration except Exception as e: # svn:external, hash divergence, i/o error... self.log.exception(e) self.done = True self._visit_status = 'partial' return False # Stopping iteration self._contents, self._directories, revision = data if revision: self._last_revision = revision self._revisions.append(revision) return True # next svn revision def store_data(self): """We store the data accumulated in internal instance variable. If the iteration over the svn revisions is done, we create the snapshot and flush to storage the data. This also resets the internal instance variable state. """ self.maybe_load_contents(self._contents) self.maybe_load_directories(self._directories) self.maybe_load_revisions(self._revisions) if self.done: # finish line, snapshot! self.generate_and_load_snapshot(revision=self._last_revision, snapshot=self._snapshot) self.flush() self._contents = [] self._directories = [] self._revisions = [] def generate_and_load_snapshot(self, revision=None, snapshot=None): """Create the snapshot either from existing revision or snapshot. Revision (supposedly new) has priority over the snapshot (supposedly existing one). Args: revision (dict): Last revision seen if any (None by default) snapshot (dict): Snapshot to use if any (None by default) """ if revision: # Priority to the revision snap = build_swh_snapshot(revision['id']) snap['id'] = identifier_to_bytes(snapshot_identifier(snap)) elif snapshot: # Fallback to prior snapshot snap = snapshot else: return None self.log.debug('snapshot: %s' % snap) self.maybe_load_snapshot(snap) def load_status(self): return { 'status': self._load_status, } def visit_status(self): return self._visit_status class SvnLoaderFromDumpArchive(SvnLoader): """Uncompress an archive containing an svn dump, mount the svn dump as an svn repository and load said repository. """ def __init__(self, archive_path): super().__init__() self.archive_path = archive_path self.temp_dir = None self.repo_path = None def prepare(self, *, svn_url, destination_path=None, swh_revision=None, start_from_scratch=False, **kwargs): self.log.info('Archive to mount and load %s' % self.archive_path) self.temp_dir, self.repo_path = init_svn_repo_from_archive_dump( self.archive_path, prefix=TEMPORARY_DIR_PREFIX_PATTERN, suffix='-%s' % os.getpid(), root_dir=self.temp_directory) if not svn_url: svn_url = 'file://%s' % self.repo_path super().prepare(svn_url=svn_url, destination_path=destination_path, swh_revision=swh_revision, start_from_scratch=start_from_scratch, **kwargs) def cleanup(self): super().cleanup() if self.temp_dir and os.path.exists(self.temp_dir): msg = 'Clean up temporary directory dump %s for project %s' % ( self.temp_dir, os.path.basename(self.repo_path)) self.log.debug(msg) shutil.rmtree(self.temp_dir) class SvnLoaderFromRemoteDump(SvnLoader): """ Create a subversion repository dump using the svnrdump utility, mount it locally and load the repository from it. """ def __init__(self): super().__init__() self.temp_dir = tempfile.mkdtemp(dir=self.temp_directory) self.repo_path = None self.truncated_dump = False def get_last_loaded_svn_rev(self, svn_url): """ Check if the svn repository has already been visited and return the last loaded svn revision number or -1 otherwise. """ last_loaded_svn_rev = -1 try: origin = \ self.storage.origin_get({'type': 'svn', 'url': svn_url}) last_swh_rev = \ - self.swh_latest_snapshot_revision(origin['id'])['revision'] + self.swh_latest_snapshot_revision(origin['url'])['revision'] last_swh_rev_headers = \ dict(last_swh_rev['metadata']['extra_headers']) last_loaded_svn_rev = int(last_swh_rev_headers['svn_revision']) except Exception: pass return last_loaded_svn_rev def dump_svn_revisions(self, svn_url, last_loaded_svn_rev=-1): """ Generate a subversion dump file using the svnrdump tool. If the svnrdump command failed somehow, the produced dump file is analyzed to determine if a partial loading is still feasible. """ # Build the svnrdump command line svnrdump_cmd = ['svnrdump', 'dump', svn_url] # Launch the svnrdump command while capturing stderr as # successfully dumped revision numbers are printed to it dump_temp_dir = tempfile.mkdtemp(dir=self.temp_dir) dump_name = ''.join(c for c in svn_url if c.isalnum()) dump_path = '%s/%s.svndump' % (dump_temp_dir, dump_name) stderr_lines = [] self.log.debug('Executing %s' % ' '.join(svnrdump_cmd)) with open(dump_path, 'wb') as dump_file: stderr_r, stderr_w = pty.openpty() svnrdump = Popen(svnrdump_cmd, stdout=dump_file, stderr=stderr_w) os.close(stderr_w) stderr_stream = OutputStream(stderr_r) readable = True while readable: lines, readable = stderr_stream.read_lines() stderr_lines += lines for line in lines: self.log.debug(line) svnrdump.wait() os.close(stderr_r) if svnrdump.returncode == 0: return dump_path # There was an error but it does not mean that no revisions # can be loaded. # Get the stderr line with latest dumped revision last_dumped_rev = None if len(stderr_lines) > 1: last_dumped_rev = stderr_lines[-2] if last_dumped_rev: # Get the latest dumped revision number matched_rev = re.search('.*revision ([0-9]+)', last_dumped_rev) last_dumped_rev = int(matched_rev.group(1)) if matched_rev else -1 # Check if revisions inside the dump file can be loaded anyway if last_dumped_rev > last_loaded_svn_rev: self.log.debug(('svnrdump did not dump all expected revisions ' 'but revisions range %s:%s are available in ' 'the generated dump file and will be loaded ' 'into the archive.') % (last_loaded_svn_rev+1, last_dumped_rev)) # Truncate the dump file after the last successfully dumped # revision to avoid the loading of corrupted data self.log.debug(('Truncating dump file after the last ' 'successfully dumped revision (%s) to avoid ' 'the loading of corrupted data') % last_dumped_rev) with open(dump_path, 'r+b') as f: with mmap(f.fileno(), 0, access=ACCESS_WRITE) as s: pattern = ('Revision-number: %s' % (last_dumped_rev+1)).encode() n = s.rfind(pattern) if n != -1: s.resize(n) self.truncated_dump = True return dump_path elif last_dumped_rev != -1: raise Exception(('Last dumped subversion revision (%s) is ' 'lesser than the last one loaded into the ' 'archive (%s).') % (last_dumped_rev, last_loaded_svn_rev)) raise Exception('An error occurred when running svnrdump and ' 'no exploitable dump file has been generated.') def prepare(self, *, svn_url, destination_path=None, swh_revision=None, start_from_scratch=False, **kwargs): # First, check if previous revisions have been loaded for the # subversion origin and get the number of the last one last_loaded_svn_rev = self.get_last_loaded_svn_rev(svn_url) # Then try to generate a dump file containing relevant svn revisions # to load, an exception will be thrown if something wrong happened dump_path = self.dump_svn_revisions(svn_url, last_loaded_svn_rev) # Finally, mount the dump and load the repository self.log.debug('Mounting dump file with "svnadmin load".') _, self.repo_path = \ init_svn_repo_from_dump(dump_path, prefix=TEMPORARY_DIR_PREFIX_PATTERN, suffix='-%s' % os.getpid(), root_dir=self.temp_dir) super().prepare(svn_url='file://%s' % self.repo_path, destination_path=destination_path, swh_revision=swh_revision, start_from_scratch=start_from_scratch, **kwargs) def cleanup(self): super().cleanup() if self.temp_dir and os.path.exists(self.temp_dir): shutil.rmtree(self.temp_dir) def visit_status(self): if self.truncated_dump: return 'partial' else: return super().visit_status() diff --git a/swh/loader/svn/svn.py b/swh/loader/svn/svn.py index 6730faf..34d942e 100644 --- a/swh/loader/svn/svn.py +++ b/swh/loader/svn/svn.py @@ -1,266 +1,266 @@ # Copyright (C) 2015-2018 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information """SVN client in charge of iterating over svn logs and yield commit representations including the hash tree/content computations per svn commit. """ import logging import os import tempfile import shutil from subvertpy.ra import RemoteAccess, Auth, get_username_provider from subvertpy import client, properties from swh.model.from_disk import Directory from . import ra, converters # When log message contains empty data DEFAULT_AUTHOR_MESSAGE = '' class SvnRepo: """Svn repository representation. Args: remote_url (str): - origin_id (int): Associated origin identifier + origin_url (str): Associated origin identifier local_dirname (str): Path to write intermediary svn action results """ - def __init__(self, remote_url, origin_id, local_dirname): + def __init__(self, remote_url, origin_url, local_dirname): self.remote_url = remote_url.rstrip('/') - self.origin_id = origin_id + self.origin_url = origin_url auth = Auth([get_username_provider()]) # one connection for log iteration self.conn_log = RemoteAccess(self.remote_url, auth=auth) # another for replay self.conn = RemoteAccess(self.remote_url, auth=auth) # one client for update operation self.client = client.Client(auth=auth) self.local_dirname = local_dirname local_name = os.path.basename(self.remote_url) self.local_url = os.path.join(self.local_dirname, local_name).encode( 'utf-8') self.uuid = self.conn.get_uuid().encode('utf-8') self.swhreplay = ra.Replay(conn=self.conn, rootpath=self.local_url) def __str__(self): return str({ - 'swh-origin': self.origin_id, + 'swh-origin': self.origin_url, 'remote_url': self.remote_url, 'local_url': self.local_url, 'uuid': self.uuid, }) def head_revision(self): """Retrieve current head revision. """ return self.conn.get_latest_revnum() def initial_revision(self): """Retrieve the initial revision from which the remote url appeared. """ return 1 def convert_commit_message(self, msg): """Simply encode the commit message. Args: msg (str): the commit message to convert. Returns: The transformed message as bytes. """ if isinstance(msg, bytes): return msg return msg.encode('utf-8') def convert_commit_date(self, date): """Convert the message commit date into a timestamp in swh format. The precision is kept. Args: date (str): the commit date to convert. Returns: The transformed date. """ return converters.svn_date_to_swh_date(date) def convert_commit_author(self, author): """Convert the commit author into an swh person. The user becomes a dictionary of the form:: { name: author, email: '', fullname: author } Args: author (str): the commit author to convert. Returns: The transformed author as dict. """ return converters.svn_author_to_swh_person(author) def __to_entry(self, log_entry): changed_paths, rev, revprops, has_children = log_entry author_date = self.convert_commit_date( revprops.get(properties.PROP_REVISION_DATE)) author = self.convert_commit_author( revprops.get(properties.PROP_REVISION_AUTHOR)) message = self.convert_commit_message( revprops.get(properties.PROP_REVISION_LOG, DEFAULT_AUTHOR_MESSAGE)) return { 'rev': rev, 'author_date': author_date, 'author_name': author, 'message': message, } def logs(self, revision_start, revision_end): """Stream svn logs between revision_start and revision_end by chunks of block_size logs. Yields revision and associated revision information between the revision start and revision_end. Args: revision_start: the svn revision starting bound revision_end: the svn revision ending bound Yields: tuple: tuple of revisions and logs: - revisions: list of revisions in order - logs: Dictionary with key revision number and value the log entry. The log entry is a dictionary with the following keys: - author_date: date of the commit - author_name: name of the author - message: commit message """ for log_entry in self.conn_log.iter_log(paths=None, start=revision_start, end=revision_end, discover_changed_paths=False): yield self.__to_entry(log_entry) def export(self, revision): """Export the repository to a given version. """ self.client.export(self.remote_url, to=self.local_url.decode('utf-8'), rev=revision, ignore_keywords=True) def export_temporary(self, revision): """Export the repository to a given revision in a temporary location. This is up to the caller of this function to clean up the temporary location when done (cf. self.clean_fs method) Args: revision: Revision to export at Returns: The tuple local_dirname the temporary location root folder, local_url where the repository was exported. """ local_dirname = tempfile.mkdtemp( prefix='check-revision-%s.' % revision, dir=self.local_dirname) local_name = os.path.basename(self.remote_url) local_url = os.path.join(local_dirname, local_name) self.client.export( self.remote_url, to=local_url, rev=revision, ignore_keywords=True) return local_dirname, os.fsencode(local_url) def swh_hash_data_per_revision(self, start_revision, end_revision): """Compute swh hash data per each revision between start_revision and end_revision. Args: start_revision: starting revision end_revision: ending revision Yields: tuple (rev, nextrev, commit, objects_per_path) - rev: current revision - nextrev: next revision - commit: commit data (author, date, message) for such revision - objects_per_path: dictionary of path, swh hash data with type """ for commit in self.logs(start_revision, end_revision): rev = commit['rev'] objects = self.swhreplay.compute_hashes(rev) if rev == end_revision: nextrev = None else: nextrev = rev + 1 yield rev, nextrev, commit, objects, self.swhreplay.directory def swh_hash_data_at_revision(self, revision): """Compute the hash data at revision. Expected to be used for update only. """ # Update the disk at revision self.export(revision) # Compute the current hashes on disk directory = Directory.from_disk(path=os.fsencode(self.local_url), save_path=True) # Update the replay collaborator with the right state self.swhreplay = ra.Replay( conn=self.conn, rootpath=self.local_url, directory=directory) # Retrieve the commit information for revision commit = list(self.logs(revision, revision))[0] yield revision, revision + 1, commit, {}, directory def clean_fs(self, local_dirname=None): """Clean up the local working copy. Args: local_dirname (str): Path to remove recursively if provided. Otherwise, remove the temporary upper root tree used for svn repository loading. """ dirname = local_dirname if local_dirname else self.local_dirname if os.path.exists(dirname): logging.debug('cleanup %s' % dirname) shutil.rmtree(dirname)