def test_crosswalk(input_format, output_format=constants.FORMAT_METAJSON, all_in_one_file=True):
logging.info("*** Test crosswalk : {0}".format(input_format))
base_dir = os.path.join(os.getcwd(), "data")
input_dir = os.path.join(base_dir, input_format)
output_file_extension = io_service.guess_file_extension_from_format(output_format)
input_file_list = io_service.get_relevant_file_list_by_format(input_dir, input_format)
if input_file_list:
results = crosswalks_service.parse_and_convert_file_list(input_file_list, input_format, output_format, "test", "", False, all_in_one_file)
col_id = "".join(["test_", input_format, "_to_", output_format])
col_title = "".join(["Test ", input_format, " to ", output_format])
if all_in_one_file:
output_file_path = os.path.join(base_dir, "result", "result_" + input_format + "_" + output_format + "." + output_file_extension)
io_service.write_items_in_one_file(col_id, col_title, results, output_file_path, output_format)
else:
#for idx, result in enumerate(results):
output_dir_path = os.path.join(base_dir, "result", "result_" + input_format + "_" + output_format)
io_service.write_items(col_id, col_title, results, output_dir_path, output_format)
input_format = constants.FORMAT_UNIMARC
output_dir_path = os.path.join("data", "num", "output")
if not os.path.exists(output_dir_path):
os.mkdir(output_dir_path)
error_file_name = "".join(["validation-", corpus, ".txt"])
error_file_path = os.path.join(output_dir_path, error_file_name)
#logging.debug("error_file_path: {}".format(error_file_path))
# conf corpus
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()
input_format = constants.FORMAT_UNIMARC
output_dir_path = os.path.join("data", "num", "output")
if not os.path.exists(output_dir_path):
os.mkdir(output_dir_path)
error_file_name = "".join(["validation-", corpus, ".txt"])
error_file_path = os.path.join(output_dir_path, error_file_name)
#logging.debug("error_file_path: {}".format(error_file_path))
# conf corpus
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
