def test_parse_compressed_file(self):
filename = path.join(
path.dirname(__file__), '../testdata/test_file_parser.tsv.gz')
parser = dsrf_file_parser.DSRFFileParser(self.logger, None, None, filename)
parser.row_validators_list = self.row_validators_list
for expected, actual in zip(
self.expected_blocks, parser.parse_file(1)):
self.assertMultiLineEqual(str(expected), str(actual))
self.assertEqual(self.logger._counts['error'], 0)
self.assertEqual(self.logger._counts['warn'], 0)
self.assertGreater(self.logger._counts['info'], 0)
def parse_report(self,
files_list,
dsrf_xsd_file,
avs_xsd_file,
human_readable=False,
write_head=True):
"""Parses a dsrf report to block objects.
The blocks are transferred to the queue.
Args:
files_list: A list of files in the report to parse.
dsrf_xsd_file: Optional user-provided path to custom XSD.
avs_xsd_file: Optional user-provided path to custom AVS XSD.
human_readable: If True, write the block to the queue in a human readable
form. Otherwise, write the block as a raw bytes.
write_head: If set to False, the header will not be written to the queue.
Returns:
dsrf_logger.DSRFLogger object.
"""
file_path_to_name_map = {
file_path: path.basename(file_path)
for file_path in files_list
}
expected_components = constants.FILE_NAME_COMPONENTS
self.logger.info('Validating the report file names.')
report_validator = report_files_validators.ReportFilesValidator(
file_name_validators.FileNameValidator(expected_components),
self.logger)
report_validator.validate_file_names(
list(file_path_to_name_map.values()))
blocks = defaultdict(set)
for file_path, file_name in six.iteritems(file_path_to_name_map):
file_parser = dsrf_file_parser.DSRFFileParser(
self.logger, dsrf_xsd_file, avs_xsd_file, file_path)
file_name_dict = file_name_validators.FileNameValidator.split_file_name(
file_name, expected_components)
file_number = file_name_dict['x']
self.logger.info('Start parsing file number %s.', file_number)
for block in file_parser.parse_file(int(file_number)):
if block.type == block_pb2.BODY:
for compared_file_number, file_blocks in six.iteritems(
blocks):
if block.number in file_blocks:
raise error.ReportValidationFailure(
'The block number %s is not unique. It appears in files '
'number: %s and %s.' %
(block.number,
min(file_number, compared_file_number),
max(file_number, compared_file_number)))
blocks[file_number].add(block.number)
elif block.type == block_pb2.HEAD:
try:
self.validate_head_block(block, file_name,
file_name_dict)
except error.FileNameValidationFailure as e:
self.logger.error(e)
if not write_head:
# Skip writing the header to the queue, if requested.
continue
else:
# FOOT
continue
self.write_to_queue(block, self.logger, human_readable)
try:
self.logger.raise_if_fatal_errors_found()
except error.ReportValidationFailure as e:
sys.stderr.write(constants.COLOR_RED + constants.BOLD +
'\n[Cell validation] ' + str(e) + constants.ENDC)
return self.logger
def _get_file_parser(self, row_validators=None):
parser = dsrf_file_parser.DSRFFileParser(
self.logger, None, None, 'filename')
parser.row_validators_list = row_validators or self.row_validators_list
return parser
