def _raw_preprocess(self, number_per_slice):
"""Splits raw data into slices. keep start time of each slice in a json file.
Args:
number_per_slice: An int of records to keep for each slice.
Returns:
Error string if an error occurs, None if complete.
"""
raw_slice_metadata = Metadata(
self._preprocess_dir, strategy=None, level=RAW_LEVEL_DIR,
bucket=self._preprocess_bucket)
raw_data = RawDataProcessor(
self._metadata['raw_file'], number_per_slice, self._raw_bucket)
slice_index = 0
raw_start_times = list()
record_count = 0
timespan_start = timespan_end = -1
while raw_data.readable():
slice_name = utils.get_slice_path(
self._preprocess_dir, RAW_LEVEL_DIR, utils.get_slice_name(slice_index))
print("Slice name: " + slice_name)
level_slice = LevelSlice(
slice_name, bucket=self._preprocess_bucket)
raw_slice = raw_data.read_next_slice()
print(raw_slice)
if isinstance(raw_slice, str):
return raw_slice
level_slice.save(raw_slice)
raw_start_times.append(raw_slice[0][0])
slice_index += 1
record_count += len(raw_slice)
if timespan_start == -1:
timespan_start = raw_slice[0][0]
timespan_end = raw_slice[-1][0]
self._metadata['raw_number'] = record_count
self._metadata['start'] = timespan_start
self._metadata['end'] = timespan_end
levels, level_names = self._get_levels_metadata(
record_count, timespan_end-timespan_start)
self._metadata['levels']['names'] = level_names
for name, level in zip(level_names, levels):
self._metadata["levels"][name] = level
for index, raw_slice_start in enumerate(raw_start_times):
raw_slice_metadata[self._metadata['levels']
[RAW_LEVEL_DIR]['names'][index]] = raw_slice_start
raw_slice_metadata.save()
return None
def test_save_parameter(self, test_records1):
"""Tests if parameter records saved."""
tmpfile = NamedTemporaryFile()
test_save_slice = LevelSlice(tmpfile.name)
formatted_test_records = {test_records1[0][2]: test_records1}
test_save_slice.save(test_records1)
test_read_slice = LevelSlice(tmpfile.name)
test_read_slice.read()
assert test_read_slice._records == formatted_test_records
tmpfile.close()
def _single_level_downsample(self, strategy, prev_level, curr_level,
level_metadata):
"""Downsamples for one single level.
Args:
strategy: A string representing a downsampling strategy.
prev_level: A string of the name of the current level.
curr_level: A string of the name of the previous level.
level_metadata: A metadata object for this level.
Returns:
A dict of metadata for the current level.
"""
curr_slice_names = self._metadata['levels'][curr_level]['names']
prev_slice_names = self._metadata['levels'][prev_level]['names']
slice_index = 0
curr_slice_path = utils.get_slice_path(
self._preprocess_dir, curr_level,
utils.get_slice_name(slice_index), strategy)
curr_level_slice = LevelSlice(curr_slice_path,
bucket=self._preprocess_bucket)
for prev_slice_name in prev_slice_names:
prev_slice_path = utils.get_slice_path(self._preprocess_dir,
prev_level, prev_slice_name,
strategy)
prev_level_slice = LevelSlice(prev_slice_path,
bucket=self._preprocess_bucket)
prev_level_slice.read()
prev_level_downsample = prev_level_slice.downsample(
strategy, self._downsample_level_factor)
curr_level_slice.add_records(prev_level_downsample)
if curr_level_slice.get_records_count() >= self._number_per_slice:
curr_level_slice.save()
level_metadata[curr_slice_names[
slice_index]] = curr_level_slice.get_first_timestamp()
slice_index += 1
curr_slice_path = utils.get_slice_path(
self._preprocess_dir, curr_level,
utils.get_slice_name(slice_index), strategy)
curr_level_slice = LevelSlice(curr_slice_path,
bucket=self._preprocess_bucket)
curr_level_slice.save()
level_metadata[curr_slice_names[
slice_index]] = curr_level_slice.get_first_timestamp()
return level_metadata