def export_corpus(corpus, output_file_path, output_format, all_in_one_file, one_record_per_copy=False): if corpus and output_file_path: # fetch metajson_list = repository_service.get_documents(corpus) # one record per physical resource if one_record_per_copy: metajson_tmp = export_one_record_per_copy(metajson_list) else: metajson_tmp = metajson_list # convert results = crosswalks_service.convert_metajson_list( metajson_tmp, output_format, all_in_one_file) # export if all_in_one_file: io_service.write_items_in_one_file(corpus, corpus, results, output_file_path, output_format) else: io_service.write_items(corpus, corpus, results, output_file_path, output_format)
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)
def export_corpus(corpus, output_file_path, output_format, all_in_one_file, one_record_per_copy=False): if corpus and output_file_path: # fetch metajson_list = repository_service.get_documents(corpus) # one record per physical resource if one_record_per_copy: metajson_tmp = export_one_record_per_copy(metajson_list) else: metajson_tmp = metajson_list # convert results = crosswalks_service.convert_metajson_list(metajson_tmp, output_format, all_in_one_file) # export if all_in_one_file: io_service.write_items_in_one_file(corpus, corpus, results, output_file_path, output_format) else: io_service.write_items(corpus, corpus, results, output_file_path, output_format)