def __init__(self, imzml_path: pathlib.Path):
try:
imzml_parser = ImzMLParser(imzml_path, parse_lib="ElementTree")
self.spectrum_reader = imzml_parser.portable_spectrum_reader()
del imzml_parser
except Exception as e:
raise ImzMLError(format_exc()) from e
self._stream = None
def __init__(self, storage: Storage, imzml_cobject: CloudObject,
ibd_cobject: CloudObject):
imzml_parser = ImzMLParser(
storage.get_cloudobject(imzml_cobject, stream=True),
ibd_file=None,
parse_lib='ElementTree',
include_spectra_metadata=METADATA_FIELDS,
)
self._ibd_cobject = ibd_cobject
self.imzml_reader = imzml_parser.portable_spectrum_reader()
super().__init__(imzml_parser)
def test_portable_get_spectrum(imzml_path, ibd_path, parse_lib):
# get normal parser
parser = ImzMLParser(imzml_path, parse_lib=parse_lib)
# get detached parser and get handle of the portable reader
detached_parser = ImzMLParser(imzml_path, parse_lib=parse_lib)
portable_reader = detached_parser.portable_spectrum_reader()
# pickle and unpickle to ensure it survives for its intended use case
portable_reader = pickle.loads(pickle.dumps(portable_reader))
for idx in range(parser.n_pixels):
mz_x, mz_y = parser.get_spectrum(idx)
_mz_x2, _mz_y2 = portable_reader.get_spectrum(idx)
assert np.all(mz_x == _mz_x2)
assert np.all(mz_y == _mz_y2)
def load_and_split_ds_vm(storage, imzml_cobject, ibd_cobject, ds_segm_size_mb,
sort_memory):
stats = []
with TemporaryDirectory() as tmp_dir:
logger.info("Temp dir is {}".format(tmp_dir))
imzml_dir = Path(tmp_dir) / 'imzml'
res = imzml_dir.mkdir()
logger.info("Create {} result {}".format(imzml_dir, res))
segments_dir = Path(tmp_dir) / 'segments'
res = segments_dir.mkdir()
logger.info("Create {} result {}".format(segments_dir, res))
logger.info('Downloading dataset...')
t = time()
imzml_path, ibd_path = download_dataset(imzml_cobject, ibd_cobject,
imzml_dir, storage)
stats.append(('download_dataset', time() - t))
logger.info('Loading parser...')
t = time()
imzml_parser = ImzMLParser(str(imzml_path))
imzml_reader = imzml_parser.portable_spectrum_reader()
stats.append(('load_parser', time() - t))
logger.info('Defining segments bounds...')
t = time()
ds_segments_bounds = define_ds_segments(
imzml_parser, ds_segm_size_mb=ds_segm_size_mb)
segments_n = len(ds_segments_bounds)
stats.append(('define_segments', time() - t))
logger.info('Segmenting...')
t = time()
chunks_n, ds_segms_len = make_segments(imzml_reader, ibd_path,
ds_segments_bounds,
segments_dir, sort_memory)
stats.append(('dataset_segmentation', time() - t))
logger.info('Uploading segments...')
t = time()
ds_segms_cobjects = upload_segments(storage, segments_dir, chunks_n,
segments_n)
stats.append(('upload_segments', time() - t))
return imzml_reader, ds_segments_bounds, ds_segms_cobjects, ds_segms_len, stats
def get_portable_imzml_reader(storage):
imzml_stream = requests.get(imzml_path, stream=True).raw
parser = ImzMLParser(imzml_stream, ibd_file=None)
imzml_reader = parser.portable_spectrum_reader()
imzml_cobject = storage.put_cobject(pickle.dumps(imzml_reader))
return imzml_reader, imzml_cobject
def get_portable_imzml_reader(storage):
imzml_stream = storage.get_cloudobject(imzml_cobject, stream=True)
parser = ImzMLParser(imzml_stream, ibd_file=None)
imzml_reader = parser.portable_spectrum_reader()
imzml_reader_cobject = storage.put_cloudobject(serialise(imzml_reader))
return imzml_reader, imzml_reader_cobject