diff --git a/debian/control b/debian/control index fc43b0f..8d94eeb 100644 --- a/debian/control +++ b/debian/control @@ -1,35 +1,36 @@ Source: swh-loader-svn Maintainer: Software Heritage developers Section: python Priority: optional Build-Depends: debhelper (>= 9), dh-python (>= 2), python3-all, python3-click, python3-dateutil, python3-nose, python3-retrying, python3-setuptools, python3-subvertpy (>= 0.9.4~), python3-swh.core (>= 0.0.19~), python3-swh.loader.core (>= 0.0.33~), python3-swh.model (>= 0.0.18~), python3-swh.scheduler (>= 0.0.14~), python3-swh.storage (>= 0.0.83~), python3-vcversioner Standards-Version: 3.9.6 Homepage: https://forge.softwareheritage.org/diffusion/DLDSVN/ Package: python3-swh.loader.svn Architecture: all Depends: gzip, python3-swh.core (>= 0.0.19~), python3-swh.loader.core (>= 0.0.33~), python3-swh.model (>= 0.0.18~), python3-swh.scheduler (>= 0.0.14~), python3-swh.storage (>= 0.0.83~), subversion, + rsvndump, ${misc:Depends}, ${python3:Depends} Description: Software Heritage Loader Svn Module in charge of loading svn repositories into swh storage. diff --git a/swh/loader/svn/loader.py b/swh/loader/svn/loader.py index 36717e5..138f40b 100644 --- a/swh/loader/svn/loader.py +++ b/swh/loader/svn/loader.py @@ -1,573 +1,724 @@ # 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 re import shutil import tempfile +from mmap import mmap, ACCESS_WRITE +from subprocess import run, PIPE + 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 SWHLoader from swh.loader.core.utils import clean_dangling_folders from . import svn, converters -from .utils import init_svn_repo_from_archive_dump +from .utils import ( + init_svn_repo_from_dump, init_svn_repo_from_archive_dump +) from .exception import SvnLoaderUneventful from .exception import SvnLoaderHistoryAltered DEFAULT_BRANCH = b'master' 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. """ return { 'id': None, 'branches': { branch: { 'target': revision_id, 'target_type': 'revision', } } } TEMPORARY_DIR_PREFIX_PATTERN = 'swh.loader.svn.' class SvnLoader(SWHLoader): """Swh svn loader to load an svn repository The repository is either remote or local. The loader deals with update on an already previously loaded repository. Default policy: Keep data as close as possible from the original svn data. We only add information that are needed for update or continuing from last known revision (svn revision and svn repository's uuid). """ CONFIG_BASE_FILENAME = 'loader/svn' ADDITIONAL_CONFIG = { 'check_revision': ('int', 1000), 'temp_directory': ('str', '/tmp'), 'debug': ('bool', False), # NOT FOR PRODUCTION, False is mandatory } def __init__(self): super().__init__(logging_class='swh.loader.svn.SvnLoader') self.check_revision = self.config['check_revision'] self.origin_id = None self.debug = self.config['debug'] self.last_seen_revision = None self.temp_directory = self.config['temp_directory'] self.done = False self.svnrepo = 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): """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 Returns: Instance of :mod:`swh.loader.svn.svn` clients """ return svn.SvnRepo(svn_url, local_dirname=local_dirname, origin_id=origin_id) def swh_latest_snapshot_revision(self, origin_id, previous_swh_revision=None): """Look for latest snapshot revision and returns it if any. Args: origin_id (int): 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: 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 is not 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 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))) # FIXME: Is that still necessary? Rationale: T570 is now closed 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) 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) 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) 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.warn(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: # Potential: svn:external, i/o error... 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 rsvndump utility, + mount it locally and load the repository from it. + """ + def __init__(self, svn_url): + super().__init__() + self.svn_url = svn_url + 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'] + 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 rsvndump tool. + If previous revisions have already been loaded, the dump + file will only contain the new ones (empty revisions will + be created in the dump file to padd them and keep the same + revision numbering). If the rsvndump command failed somehow, + the produced dump file is analyzed to determine if a partial + loading is still feasible. + """ + # Build the rsvndump command line + rsvndump_cmd = ['rsvndump'] + if last_loaded_svn_rev >= 0: + # Previous revisions have already been loaded, so dump + # only those we are interested in while padding already loaded + # ones to keep the original revision numbers in the dump file + rsvndump_cmd += ['-r', '%s:HEAD' % last_loaded_svn_rev, + '--keep-revnums'] + # Use deltas to get a smaller dump file + rsvndump_cmd += ['--deltas', svn_url] + + # Launch the rsvndump 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) + self.log.debug('Executing %s' % ' '.join(rsvndump_cmd)) + with open(dump_path, 'wb') as dump_file: + rsvndump = run(rsvndump_cmd, stdout=dump_file, stderr=PIPE) + + if rsvndump.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 + stderr_lines = rsvndump.stderr.split(b'\n') + 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(b'.*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(('rsvndump 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 occured when running rsvndump 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(self.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(self.svn_url, last_loaded_svn_rev) + + # Finally, mount the dump and load the repository + _, self.repo_path = \ + init_svn_repo_from_dump(dump_path, + prefix=TEMPORARY_DIR_PREFIX_PATTERN, + suffix='-%s' % os.getpid(), + root_dir=self.temp_dir) + 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): + 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/tasks.py b/swh/loader/svn/tasks.py index 3aee408..6509966 100644 --- a/swh/loader/svn/tasks.py +++ b/swh/loader/svn/tasks.py @@ -1,69 +1,95 @@ # 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 + from swh.scheduler.task import Task -from .loader import SvnLoader, SvnLoaderFromDumpArchive +from .loader import ( + SvnLoader, SvnLoaderFromDumpArchive, SvnLoaderFromRemoteDump +) class LoadSvnRepository(Task): """Load an svn repository to Software Heritage. """ task_queue = 'swh_loader_svn' def run_task(self, *, svn_url, destination_path=None, swh_revision=None, origin_url=None, visit_date=None, start_from_scratch=None): """Import a svn repository Args: args: ordered arguments (expected None) kwargs: Dictionary with the following expected keys: - svn_url (str): (mandatory) svn's repository url - destination_path (str): (mandatory) root directory to locally retrieve svn's data - origin_url (str): Optional original url override - swh_revision (dict): (optional) extra revision hex to start from. see swh.loader.svn.SvnLoader.process docstring """ loader = SvnLoader() loader.log = self.log return loader.load( svn_url=svn_url, destination_path=destination_path, origin_url=origin_url, swh_revision=swh_revision, visit_date=visit_date, start_from_scratch=start_from_scratch) class MountAndLoadSvnRepository(Task): """Mount an archive dump into an svn repository, then load the repository to Software Heritage. """ task_queue = 'swh_loader_svn_mount_and_load' def run_task(self, *, archive_path, origin_url=None, visit_date=None, start_from_scratch=False): """1. Mount an svn dump from archive as a local svn repository 2. Load it through the svn loader 3. Clean up mounted svn repository archive """ loader = SvnLoaderFromDumpArchive(archive_path) loader.log = self.log return loader.load(svn_url=None, origin_url=origin_url, visit_date=visit_date, archive_path=archive_path, start_from_scratch=start_from_scratch) + + +class DumpMountAndLoadSvnRepository(Task): + """ + Create a dump of a remote repository through the rsvndump + tool, mount it locally then load the repository into the + Software Heritage archive. + """ + task_queue = 'swh_loader_svn_dump_mount_and_load' + + def run_task(self, *, svn_url, origin_url=None, visit_date=None, + start_from_scratch=False): + """1. Mount an svn dump from archive as a local svn repository. + 2. Load it through the svn loader. + 3. Clean up mounted svn repository archive. + + """ + loader = SvnLoaderFromRemoteDump(svn_url) + loader.log = self.log + return loader.load(svn_url=None, + origin_url=origin_url, + visit_date=visit_date, + start_from_scratch=start_from_scratch) diff --git a/swh/loader/svn/utils.py b/swh/loader/svn/utils.py index 90b749f..634e9f6 100644 --- a/swh/loader/svn/utils.py +++ b/swh/loader/svn/utils.py @@ -1,76 +1,100 @@ # Copyright (C) 2016 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 import os import tempfile import shutil from dateutil import parser from subprocess import PIPE, Popen, call def strdate_to_timestamp(strdate): """Convert a string date to an int timestamp. Args: strdate: A string representing a date with format like 'YYYY-mm-DDTHH:MM:SS.800722Z' Returns: A couple of integers: seconds, microseconds """ if strdate: dt = parser.parse(strdate) ts = { 'seconds': int(dt.timestamp()), 'microseconds': dt.microsecond, } else: # epoch ts = {'seconds': 0, 'microseconds': 0} return ts -def init_svn_repo_from_archive_dump(archive_path, prefix=None, suffix=None, - root_dir='/tmp'): - """Given a path to an archive containing an svn dump. +def init_svn_repo_from_dump(dump_path, prefix=None, suffix=None, + root_dir='/tmp', gzip=False): + """Given a path to a svn dump. Initialize an svn repository with the content of said dump. Returns: A tuple: - temporary folder (str): containing the mounted repository - repo_path (str): path to the mounted repository inside the temporary folder Raises: ValueError in case of failure to run the command to uncompress and load the dump. """ - project_name = os.path.basename(os.path.dirname(archive_path)) + project_name = os.path.basename(os.path.dirname(dump_path)) temp_dir = tempfile.mkdtemp(prefix=prefix, suffix=suffix, dir=root_dir) try: repo_path = os.path.join(temp_dir, project_name) # create the repository that will be loaded with the dump cmd = ['svnadmin', 'create', repo_path] r = call(cmd) if r != 0: raise ValueError( 'Failed to initialize empty svn repo for %s' % project_name) - with Popen(['gzip', '-dc', archive_path], stdout=PIPE) as dump: + read_dump_cmd = ['cat', dump_path] + if gzip: + read_dump_cmd = ['gzip', '-dc', dump_path] + + with Popen(read_dump_cmd, stdout=PIPE) as dump: cmd = ['svnadmin', 'load', '-q', repo_path] r = call(cmd, stdin=dump.stdout) if r != 0: raise ValueError( 'Failed to mount the svn dump for project %s' % project_name) return temp_dir, repo_path except Exception as e: shutil.rmtree(temp_dir) raise e + + +def init_svn_repo_from_archive_dump(archive_path, prefix=None, suffix=None, + root_dir='/tmp'): + """Given a path to an archive containing an svn dump. + Initialize an svn repository with the content of said dump. + + Returns: + A tuple: + - temporary folder (str): containing the mounted repository + - repo_path (str): path to the mounted repository inside the + temporary folder + + Raises: + ValueError in case of failure to run the command to uncompress + and load the dump. + + """ + return init_svn_repo_from_dump(archive_path, prefix=prefix, suffix=suffix, + root_dir=root_dir, gzip=True)