def task_cns_core_stat(args=None): name = "cns-core" version = "3.2" items = read_cns_core_excel(version, path="data") #logging.info(items) stat(items, [ "wikidataUrl", "wikidataName", "wikipediaUrl", "nameZh", "descriptionZh" ]) dup(items, "wikidataName") dup(items, "wikidataUrl") dup(items, "wikipediaUrl")
def task_cns_core_excel2json(args=None): name = "cns-core" version = cns_config["version"] items = read_cns_core_excel(version, path="data") #logging.info(items) stat(items, [ "wikidataUrl", "wikidataName", "wikipediaUrl", "nameZh", "descriptionZh" ]) dup(items, "wikidataName") dup(items, "wikidataUrl") dup(items, "wikipediaUrl") write_cns_core(items, version, formats=["jsonld"])
def task_init(args=None): so = Schemaorg("3.2") data = so.load_data() stat(data.values(), [], ["_group", "@type"]) logging.info(len(data))