def validate_corpus(args):
corpus = args.corpus
if not corpus:
corpus = default_corpus
logging.info("corpus: {}".format(corpus))
error_file_name = "".join(["validation-", corpus, ".txt"])
error_file_path = os.path.join(os.path.dirname(__file__), os.pardir, "log", error_file_name)
logging.info("error_file_path: {}".format(error_file_path))
corpus_service.validate_corpus(corpus, error_file_path)
def validate_corpus(args):
corpus = args.corpus
if not corpus:
corpus = default_corpus
logging.info("corpus: {}".format(corpus))
error_file_name = "".join(["validation-", corpus, ".txt"])
error_file_path = os.path.join(os.path.dirname(__file__), os.pardir, "log",
error_file_name)
logging.info("error_file_path: {}".format(error_file_path))
corpus_service.validate_corpus(corpus, error_file_path)
]
harvest_by_ids(corpus, target, ids)
#harvest_by_set(corpus, target, "SHS:STAT")
# path
data_result_dir = os.path.join(os.path.dirname(__file__), os.pardir, "data", "result")
logging.info("data_result_dir: {}".format(data_result_dir))
error_file_path = os.path.join(data_result_dir, "result_validation_errors.txt")
metajson_file_path = os.path.join(data_result_dir, "result_didl_metajson_spire.json")
mods_file_path = os.path.join(data_result_dir, "result_didl_mods_spire.json")
repec_file_path = os.path.join(data_result_dir, "result_repec.txt")
date_path = datetime.datetime.now()
# validate
corpus_service.validate_corpus(corpus, error_file_path)
date_validate = datetime.datetime.now()
chrono.chrono_trace("Validate corpus", date_path, date_validate, None)
# export MetaJSON
corpus_service.export_corpus(corpus, metajson_file_path, constants.FORMAT_METAJSON, True)
date_export_metajson = datetime.datetime.now()
chrono.chrono_trace("Export corpus as MetaJSON", date_validate, date_export_metajson, None)
# export MODS
corpus_service.export_corpus(corpus, mods_file_path, constants.FORMAT_MODS, True)
date_export_mods = datetime.datetime.now()
chrono.chrono_trace("Export corpus as MODS", date_export_metajson, date_export_mods, None)
# export RePEc
corpus_service.export_corpus(corpus, repec_file_path, constants.FORMAT_REPEC, True)
corpus_service.clean_corpus(corpus)
corpus_service.conf_corpus(corpus, "aime")
date_clean = datetime.datetime.now()
chrono.chrono_trace("Clean and conf corpus", date_begin, date_clean, None)
# import
input_file_paths = io_service.get_relevant_file_list_by_format(
input_dir_path, input_format)
results = corpus_service.import_metadata_files(corpus, input_file_paths,
input_format, source,
rec_id_prefix, True, None)
date_import = datetime.datetime.now()
chrono.chrono_trace("Import corpus", date_clean, date_import, None)
# Validate
corpus_service.validate_corpus(corpus, error_file_path)
date_validate = datetime.datetime.now()
chrono.chrono_trace("Validate corpus", date_import, date_validate, None)
# Export mods
corpus_service.export_corpus(corpus, output_dir_path,
constants.FORMAT_MODS, False, True)
date_export_mods = datetime.datetime.now()
chrono.chrono_trace("Export corpus mods", date_validate, date_export_mods,
None)
# Export oai_dc
corpus_service.export_corpus(corpus, output_dir_path,
constants.FORMAT_OAI_DC, False, True)
date_export_oai_dc = datetime.datetime.now()
chrono.chrono_trace("Export corpus oai_dc", date_export_mods,