def status(self, spec, full_option, status_directory):
repo_type = self.__repo_type
try:
objects_path = get_objects_path(self.__config, repo_type)
repo = LocalRepository(self.__config, objects_path, repo_type)
log.info('%s: status of ml-git index for [%s]' % (repo_type, spec),
class_name=REPOSITORY_CLASS_NAME)
new_files, deleted_files, untracked_files, corruped_files, changed_files = repo.status(
spec, status_directory)
except Exception as e:
log.error(e, class_name=REPOSITORY_CLASS_NAME)
return
if new_files is not None and deleted_files is not None and untracked_files is not None:
print('Changes to be committed:')
self._print_files(new_files, full_option, 'New file: ')
self._print_files(deleted_files, full_option, 'Deleted: ')
print('\nUntracked files:')
self._print_files(untracked_files, full_option)
print('\nCorrupted files:')
self._print_files(corruped_files, full_option)
if changed_files and len(changed_files) > 0:
print('\nChanges not staged for commit:')
self._print_files(changed_files, full_option)
def add(self, spec, file_path, bump_version=False, run_fsck=False):
repo_type = self.__repo_type
is_shared_objects = 'objects_path' in self.__config[repo_type]
is_shared_cache = 'cache_path' in self.__config[repo_type]
if not validate_config_spec_hash(self.__config):
log.error(
'.ml-git/config.yaml invalid. It should look something like this:\n%s'
% get_yaml_str(
get_sample_config_spec('somebucket', 'someprofile',
'someregion')),
class_name=REPOSITORY_CLASS_NAME)
return None
path, file = None, None
try:
refs_path = get_refs_path(self.__config, repo_type)
index_path = get_index_path(self.__config, repo_type)
metadata_path = get_metadata_path(self.__config, repo_type)
cache_path = get_cache_path(self.__config, repo_type)
objects_path = get_objects_path(self.__config, repo_type)
repo = LocalRepository(self.__config, objects_path, repo_type)
mutability, check_mutability = repo.get_mutability_from_spec(
spec, repo_type)
sampling_flag = os.path.exists(
os.path.join(index_path, 'metadata', spec, 'sampling'))
if sampling_flag:
log.error(
'You cannot add new data to an entity that is based on a checkout with the --sampling option.',
class_name=REPOSITORY_CLASS_NAME)
return
if not mutability:
return
if not check_mutability:
log.error('Spec mutability cannot be changed.',
class_name=REPOSITORY_CLASS_NAME)
return
_, deleted, untracked_files, _, changed_files = repo.status(
spec, log_errors=False)
if deleted is None and untracked_files is None and changed_files is None:
return None
elif len(deleted) == 0 and len(untracked_files) == 0 and len(
changed_files) == 0:
log.info('There is no new data to add',
class_name=REPOSITORY_CLASS_NAME)
return None
ref = Refs(refs_path, spec, repo_type)
tag, sha = ref.branch()
categories_path = get_path_with_categories(tag)
path, file = search_spec_file(self.__repo_type, spec,
categories_path)
except Exception as e:
log.error(e, class_name=REPOSITORY_CLASS_NAME)
return
if path is None:
return
spec_path = os.path.join(path, file)
spec_file = yaml_load(spec_path)
if not validate_spec_hash(spec_file, self.__repo_type):
log.error(
'Invalid %s spec in %s. It should look something like this:\n%s'
% (self.__repo_type, spec_path,
get_sample_spec_doc('somebucket', self.__repo_type)),
class_name=REPOSITORY_CLASS_NAME)
return None
if not validate_bucket_name(spec_file[self.__repo_type],
self.__config):
return None
# Check tag before anything to avoid creating unstable state
log.debug('Repository: check if tag already exists',
class_name=REPOSITORY_CLASS_NAME)
m = Metadata(spec, metadata_path, self.__config, repo_type)
if not m.check_exists():
log.error('The %s has not been initialized' % self.__repo_type,
class_name=REPOSITORY_CLASS_NAME)
return
try:
m.update()
except Exception:
pass
# get version of current manifest file
manifest = ''
if tag is not None:
m.checkout(tag)
md_metadata_path = m.get_metadata_path(tag)
manifest = os.path.join(md_metadata_path, 'MANIFEST.yaml')
m.checkout('master')
try:
# adds chunks to ml-git Index
log.info('%s adding path [%s] to ml-git index' % (repo_type, path),
class_name=REPOSITORY_CLASS_NAME)
with change_mask_for_routine(is_shared_objects):
idx = MultihashIndex(spec, index_path, objects_path,
mutability, cache_path)
idx.add(path, manifest, file_path)
# create hard links in ml-git Cache
self.create_hard_links_in_cache(cache_path, index_path,
is_shared_cache, mutability, path,
spec)
except Exception as e:
log.error(e, class_name=REPOSITORY_CLASS_NAME)
return None
if bump_version and not increment_version_in_spec(
spec_path, self.__repo_type):
return None
idx.add_metadata(path, file)
self._check_corrupted_files(spec, repo)
# Run file check
if run_fsck:
self.fsck()