Exemple #1
0
    def get_all_asset_keys(self, prefix_path=None):
        if not prefix_path:
            if self.has_secondary_index(SECONDARY_INDEX_ASSET_KEY):
                query = db.select([AssetKeyTable.c.asset_key])
            else:
                query = (db.select([
                    SqlEventLogStorageTable.c.asset_key
                ]).where(
                    SqlEventLogStorageTable.c.asset_key != None).distinct())
        else:
            if self.has_secondary_index(SECONDARY_INDEX_ASSET_KEY):
                query = db.select([AssetKeyTable.c.asset_key]).where(
                    AssetKeyTable.c.asset_key.startswith(
                        AssetKey.get_db_prefix(prefix_path)))
            else:
                query = (db.select([
                    SqlEventLogStorageTable.c.asset_key
                ]).where(SqlEventLogStorageTable.c.asset_key != None).where(
                    SqlEventLogStorageTable.c.asset_key.startswith(
                        AssetKey.get_db_prefix(prefix_path))).distinct())

        with self.connect() as conn:
            results = conn.execute(query).fetchall()
        return [
            AssetKey.from_db_string(asset_key) for (asset_key, ) in results
            if asset_key
        ]
    def get_asset_keys(self, prefix_path=None):
        lazy_migrate = False

        if not prefix_path:
            if self.has_secondary_index(SECONDARY_INDEX_ASSET_KEY):
                query = db.select([AssetKeyTable.c.asset_key])
            else:
                query = (db.select([
                    SqlEventLogStorageTable.c.asset_key
                ]).where(
                    SqlEventLogStorageTable.c.asset_key != None).distinct())

                # This is in place to migrate everyone to using the secondary index table for asset
                # keys.  Performing this migration should result in a big performance boost for
                # any asset-catalog reads.

                # After a sufficient amount of time (>= 0.11.0?), we can remove the checks
                # for has_secondary_index(SECONDARY_INDEX_ASSET_KEY) and always read from the
                # AssetKeyTable, since we are already writing to the table. Tracking the conditional
                # check removal here: https://github.com/dagster-io/dagster/issues/3507
                lazy_migrate = True
        else:
            if self.has_secondary_index(SECONDARY_INDEX_ASSET_KEY):
                query = db.select([AssetKeyTable.c.asset_key]).where(
                    db.or_(
                        AssetKeyTable.c.asset_key.startswith(
                            AssetKey.get_db_prefix(prefix_path)),
                        AssetKeyTable.c.asset_key.startswith(
                            AssetKey.get_db_prefix(prefix_path, legacy=True)),
                    ))
            else:
                query = (db.select([
                    SqlEventLogStorageTable.c.asset_key
                ]).where(SqlEventLogStorageTable.c.asset_key != None).where(
                    db.or_(
                        SqlEventLogStorageTable.c.asset_key.startswith(
                            AssetKey.get_db_prefix(prefix_path)),
                        SqlEventLogStorageTable.c.asset_key.startswith(
                            AssetKey.get_db_prefix(prefix_path, legacy=True)),
                    )).distinct())

        with self.index_connection() as conn:
            results = conn.execute(query).fetchall()

        if lazy_migrate:
            # This is in place to migrate everyone to using the secondary index table for asset
            # keys.  Performing this migration should result in a big performance boost for
            # any subsequent asset-catalog reads.
            self._lazy_migrate_secondary_index_asset_key(
                [asset_key for (asset_key, ) in results if asset_key])
        return list(
            set([
                AssetKey.from_db_string(asset_key) for (asset_key, ) in results
                if asset_key
            ]))
    def get_all_asset_keys(self, prefix_path=None):
        if not prefix_path:
            query = db.select([SqlEventLogStorageTable.c.asset_key]).distinct()
        else:
            query = (db.select([SqlEventLogStorageTable.c.asset_key]).where(
                SqlEventLogStorageTable.c.asset_key.startswith(
                    AssetKey.get_db_prefix(prefix_path))).distinct())

        with self.connect() as conn:
            results = conn.execute(query).fetchall()

        return [
            AssetKey.from_db_string(asset_key) for (asset_key, ) in results
            if asset_key
        ]
Exemple #4
0
    def get_all_asset_keys(self, prefix_path=None):
        asset_records = []
        for records in self._logs.values():
            asset_records += [
                record for record in records if record.is_dagster_event
                and record.dagster_event.asset_key and
                (not prefix_path or record.dagster_event.asset_key.to_string().
                 startswith(AssetKey.get_db_prefix(prefix_path)))
            ]

        asset_events = [
            record.dagster_event for record in sorted(
                asset_records, key=lambda x: x.timestamp, reverse=True)
        ]
        asset_keys = OrderedDict()
        for event in asset_events:
            asset_keys["/".join(event.asset_key.path)] = event.asset_key
        return list(asset_keys.values())