Function parse_data_table

openlibrary/data/__init__.py:8–16

Parses the dump of data table and returns an iterator with for all entries.

(filename)

def parse_data_table(filename):
    """Parses the dump of data table and returns an iterator with
    <key, type, revision, json> for all entries.
    """
    with open(filename) as file:
        for line in file:
            _thing_id, _revision, json_data = pgdecode(line).strip().split("\t")
            d = json.loads(json_data)
            yield d["key"], d["type"]["key"], str(d["revision"]), json_data