Exemplo n.º 1
0
    def test_put(self):
        idx = MultihashIndex('dataset-spec', self.tmp_dir, self.tmp_dir)
        idx.add('data', self.tmp_dir)

        mf = idx.get_index()
        self.assertTrue(mf.exists('zdj7WgHSKJkoJST5GWGgS53ARqV7oqMGYVvWzEWku3MBfnQ9u'))

        idx.add('image.jpg', self.tmp_dir)
        idx.update_index('zdj7WemKEtQMVL81UU6PSuYaoxvBQ6CiUMq1fMvoXBhPUsCK2', 'image.jpg')
        self.assertTrue(mf.exists('zdj7WemKEtQMVL81UU6PSuYaoxvBQ6CiUMq1fMvoXBhPUsCK2'))
Exemplo n.º 2
0
    def test_push(self):

        mlgit_dir = os.path.join(self.tmp_dir, '.ml-git')

        indexpath = os.path.join(mlgit_dir, 'index-test')
        mdpath = os.path.join(mlgit_dir, 'metadata-test')
        objectpath = os.path.join(mlgit_dir, 'objects-test')
        specpath = os.path.join(mdpath, 'vision-computing/images/dataset-ex')
        ensure_path_exists(specpath)
        ensure_path_exists(indexpath)
        shutil.copy('hdata/dataset-ex.spec', specpath + '/dataset-ex.spec')
        shutil.copy('hdata/config.yaml', mlgit_dir + '/config.yaml')
        manifestpath = os.path.join(specpath, 'MANIFEST.yaml')
        yaml_save(
            {
                'zdj7WjdojNAZN53Wf29rPssZamfbC6MVerzcGwd9tNciMpsQh':
                {'imghires.jpg'}
            }, manifestpath)

        # adds chunks to ml-git Index
        idx = MultihashIndex(specpath, indexpath, objectpath)
        idx.add('data-test-push/', manifestpath)

        fi = yaml_load(os.path.join(specpath, 'INDEX.yaml'))
        self.assertTrue(len(fi) > 0)
        self.assertTrue(os.path.exists(indexpath))

        o = Objects(specpath, objectpath)
        o.commit_index(indexpath, self.tmp_dir)

        self.assertTrue(os.path.exists(objectpath))
        c = yaml_load('hdata/config.yaml')
        r = LocalRepository(c, objectpath)
        r.push(objectpath, specpath + '/dataset-ex.spec')
        s3 = boto3.resource(
            's3',
            region_name='eu-west-1',
            aws_access_key_id='fake_access_key',
            aws_secret_access_key='fake_secret_key',
        )
        for key in idx.get_index():
            self.assertIsNotNone(s3.Object(testbucketname, key))
Exemplo n.º 3
0
    def reset(self, spec, reset_type, head):
        log.info(output_messages['INFO_INITIALIZING_RESET'] %
                 (reset_type, head),
                 class_name=REPOSITORY_CLASS_NAME)
        if (reset_type == '--soft'
                or reset_type == '--mixed') and head == HEAD:
            return
        try:
            repo_type = self.__repo_type
            metadata_path = get_metadata_path(self.__config, repo_type)
            index_path = get_index_path(self.__config, repo_type)
            refs_path = get_refs_path(self.__config, repo_type)
            object_path = get_objects_path(self.__config, repo_type)
            met = Metadata(spec, metadata_path, self.__config, repo_type)
            ref = Refs(refs_path, spec, repo_type)
            idx = MultihashIndex(spec, index_path, object_path)
            fidx = FullIndex(spec, index_path)
        except Exception as e:
            log.error(e, class_name=REPOSITORY_CLASS_NAME)
            return

        # get tag before reset
        tag = met.get_current_tag()
        categories_path = get_path_with_categories(str(tag))
        # current manifest file before reset
        manifest_path = os.path.join(metadata_path, categories_path, spec,
                                     MANIFEST_FILE)
        _manifest = Manifest(manifest_path).load()

        if head == HEAD_1:  # HEAD~1
            try:
                # reset the repo
                met.reset()
            except Exception:
                return

        # get tag after reset
        tag_after_reset = met.get_current_tag()
        sha = met.sha_from_tag(tag_after_reset)

        # update ml-git ref HEAD
        ref.update_head(str(tag_after_reset), sha)

        # # get path to reset workspace in case of --hard
        path, file = None, None
        try:
            path, file = search_spec_file(self.__repo_type, spec,
                                          categories_path)
        except Exception as e:
            log.error(e, class_name=REPOSITORY_CLASS_NAME)

        if reset_type == '--hard' and path is None:
            return

        # get manifest from metadata after reset
        _manifest_changed = Manifest(manifest_path)

        hash_files, file_names = _manifest_changed.get_diff(_manifest)
        idx_mf = idx.get_index().load()

        if reset_type == '--soft':
            # add in index/metadata/<entity-name>/MANIFEST
            idx.update_index_manifest(idx_mf)
            idx.update_index_manifest(hash_files)
            fidx.update_index_status(file_names, Status.a.name)

        else:  # --hard or --mixed
            # remove hash from index/hashsh/store.log
            file_names.update(*idx_mf.values())
            objs = MultihashFS(index_path)
            for key_hash in hash_files:
                objs.remove_hash(key_hash)
            idx.remove_manifest()
            fidx.remove_from_index_yaml(file_names)
            fidx.remove_uncommitted()

        if reset_type == '--hard':  # reset workspace
            remove_from_workspace(file_names, path, spec)