def _process_chunk_not_containing_data_record_or_header_part(self, chunk):
"""
This method processes a chunk that does not contain a data record or header. This case is
not applicable to "non_data". For cspp file streams, we expect some lines in the file that
we do not care about, and we will not consider them "non_data".
@param chunk A regular expression match object for a cspp header row
"""
# Check for the expected timestamp line we will ignore
timestamp_line_match = TIMESTAMP_LINE_MATCHER.match(chunk)
# Check for other status messages we can ignore
ignore_match = IGNORE_MATCHER.match(chunk)
if timestamp_line_match is not None or ignore_match is not None:
# Ignore
pass
else:
# OK. We got unexpected data
log.warn('got unrecognized row %s at position %s', chunk,
self._read_state[StateKey.POSITION])
self._exception_callback(
RecoverableSampleException("Found an invalid chunk: %s" %
chunk))
def parse_file(self):
for line in self._stream_handle:
battery_match = BATTERY_DATA_MATCHER.match(line)
gps_match = GPS_DATA_MATCHER.match(line)
# If we found a data match, let's process it
if battery_match is not None:
self._process_data_match(self._battery_status_class, battery_match)
elif gps_match is not None:
self._process_data_match(self._gps_adjustment_class, gps_match)
else:
# Check for head part match
header_part_match = HEADER_PART_MATCHER.match(line)
if header_part_match is not None:
self._process_header_part_match(header_part_match)
elif HEX_ASCII_LINE_MATCHER.match(line):
self._process_line_not_containing_data_record_or_header_part(line)
elif not TIMESTAMP_LINE_MATCHER.match(line) and not \
(IGNORE_MATCHER is not None and IGNORE_MATCHER.match(line)):
log.warn("non_data: %s", line)
self._exception_callback(RecoverableSampleException("Found d bytes"))
def parse_file(self):
"""
Parse NUTNR J CSPP text file.
"""
# loop over all lines in the data file and parse the data to generate Winch CSPP particles
for line in self._stream_handle:
data_match = DATA_MATER.match(line)
# If we found a data match, let's process it
if data_match is not None:
self._process_data_match(data_match)
else:
# Check for head part match
header_part_match = HEADER_PART_MATCHER.match(line)
if header_part_match is not None:
header_part_key = header_part_match.group(
HeaderPartMatchesGroupNumber.HEADER_PART_MATCH_GROUP_KEY)
header_part_value = header_part_match.group(
HeaderPartMatchesGroupNumber.HEADER_PART_MATCH_GROUP_VALUE)
if header_part_key in self._header_state.keys():
self._header_state[header_part_key] = string.rstrip(header_part_value)
else:
if HEX_ASCII_LINE_MATCHER.match(line):
# we found a line starting with the timestamp, depth, and
# suspect timestamp, followed by all hex ascii chars
log.warn('got hex ascii corrupted data %s ', line)
self._exception_callback(RecoverableSampleException(
"Found hex ascii corrupted data: %s" % line))
# ignore the expected timestamp line and any lines matching the ignore regex,
# otherwise data is unexpected
elif not TIMESTAMP_LINE_MATCHER.match(line) and not \
(IGNORE_MATCHER is not None and IGNORE_MATCHER.match(line)):
# Unexpected data was found
log.warn('got unrecognized row %s', line)
self._exception_callback(RecoverableSampleException("Found an invalid chunk: %s" % line))
def _process_chunk_not_containing_data_record_or_header_part(self, chunk):
"""
This method processes a chunk that does not contain a data record or header. This case is
not applicable to "non_data". For cspp file streams, we expect some lines in the file that
we do not care about, and we will not consider them "non_data".
@param chunk A regular expression match object for a cspp header row
"""
# Check for the expected timestamp line we will ignore
timestamp_line_match = TIMESTAMP_LINE_MATCHER.match(chunk)
# Check for other status messages we can ignore
ignore_match = IGNORE_MATCHER.match(chunk)
if timestamp_line_match is not None or ignore_match is not None:
# Ignore
pass
else:
# OK. We got unexpected data
log.warn('got unrecognized row %s at position %s', chunk, self._read_state[StateKey.POSITION])
self._exception_callback(RecoverableSampleException("Found an invalid chunk: %s" % chunk))