def parseString(inString, silence=False):
if sys.version_info.major == 2:
from StringIO import StringIO as IOBuffer
else:
from io import BytesIO as IOBuffer
parser = None
doc = parsexml_(IOBuffer(inString), parser)
rootNode = doc.getroot()
rootTag, rootClass = get_root_tag(rootNode)
if rootClass is None:
rootTag = 'stringsType'
rootClass = stringsType
rootObj = rootClass.factory()
rootObj.build(rootNode)
# Enable Python to collect the space used by the DOM.
doc = None
if not silence:
sys.stdout.write('<?xml version="1.0" ?>\n')
rootObj.export(
sys.stdout,
0,
name_=rootTag,
namespacedef_=
'xmlns:tns="https://leoschwarz.github.io/opensim-protocol/schema/StringsSchema.xsd"'
)
return rootObj
def decompress_case(archive_path, debug=False):
"""Decompress a archive in memory.
Args:
archive_path (str): The filepath of the archive
debug (bool): To print debug information
Returns:
slices (list): The list of decompressed files
"""
mode = ''
if archive_path.lower().endswith('.gz') or archive_path.lower().endswith(
'.tgz'):
mode = 'r:gz'
elif archive_path.lower().endswith(
'.bz2') or archive_path.lower().endswith('.tbz'):
mode = 'r:bz2'
tar_archive = tarfile.open(archive_path, mode)
slices = []
if debug:
print("Untar...")
start = time.time()
while True:
tarinfo = tar_archive.next()
if tarinfo is None:
break
if tarinfo.isfile():
file_object = tar_archive.extractfile(tarinfo)
file_like_object = IOBuffer(file_object.read())
file_object.close()
file_like_object.seek(0)
try:
dcm = pydicom.read_file(file_like_object)
slices.append(dcm)
except:
continue
end_untar = time.time()
if debug:
print("Untar time: {}".format(end_untar - start))
return slices
def parseString(inString, silence=False):
if sys.version_info.major == 2:
from StringIO import StringIO as IOBuffer
else:
from io import BytesIO as IOBuffer
parser = None
doc = parsexml_(IOBuffer(inString), parser)
rootNode = doc.getroot()
rootTag, rootClass = get_root_tag(rootNode)
if rootClass is None:
rootTag = 'image_descriptor'
rootClass = image_descriptor
rootObj = rootClass.factory()
rootObj.build(rootNode)
# Enable Python to collect the space used by the DOM.
doc = None
if not silence:
sys.stdout.write('<?xml version="1.0" ?>\n')
rootObj.export(sys.stdout, 0, name_=rootTag, namespacedef_='')
return rootObj
def peek_compressed_study(archive_path):
"""Peeks the `StudyInstanceUID` of a compressed DICOM archive, that is,
reads the first DICOM file in the archive and returns the associated ID.
Args:
archive_path (str): The filepath of the archive.
Returns:
str or None: The StudyInstanceUID if possible, None otherwise.
"""
mode = ''
if archive_path.lower().endswith('.gz') or archive_path.lower().endswith(
'.tgz'):
mode = 'r:gz'
elif archive_path.lower().endswith(
'.bz2') or archive_path.lower().endswith('.tbz'):
mode = 'r:bz2'
try:
tar_archive = tarfile.open(archive_path, mode)
except:
return None
dcm = None
while True:
tarinfo = tar_archive.next()
if tarinfo is None:
break
if tarinfo.isfile():
file_object = tar_archive.extractfile(tarinfo)
file_like_object = IOBuffer(file_object.read())
file_object.close()
file_like_object.seek(0)
try:
dcm = pydicom.read_file(file_like_object)
except:
continue
break
if dcm is not None:
return dcm.StudyInstanceUID
return None
def parseString(inString, silence=False):
if sys.version_info.major == 2:
from StringIO import StringIO as IOBuffer
else:
from io import BytesIO as IOBuffer
parser = None
doc = parsexml_(IOBuffer(inString), parser)
rootNode = doc.getroot()
rootTag, rootClass = get_root_tag(rootNode)
if rootClass is None:
rootTag = 'carrierType'
rootClass = carrierType
rootObj = rootClass.factory()
rootObj.build(rootNode)
# Enable Python to collect the space used by the DOM.
doc = None
## if not silence:
## sys.stdout.write('<?xml version="1.0" ?>\n')
## rootObj.export(
## sys.stdout, 0, name_=rootTag,
## namespacedef_='xmlns:target="http://cars.example.com/schema"')
return rootObj
def write_reordered_traj(temp_inds, byte_inds, outtemps, temps,
frametuple_dict, nprod, writefreq, outtrajfns,
infobjs):
"""
Reorders trajectories by temp. and writes them to disk
:param temp_inds: list index of temps (in the list of all temps) for which
reordered trajs will be produced on this proc.
:param byte_inds: dict containing the (previously stored) byte indices
for each replica file (key = replica number)
:param outtemps: list of all temps for which to produce reordered trajs.
:param temps: list of all temps used in the REMD simulation.
:param outtrajfns: list of filenames for output (ordered) trajs.
:param frametuple_dict: dict containing a tuple (replica #, frame #)
for each temp.
:param nprod: number of production timesteps.
Last (nprod / writefreq) frames
from the end will be written to disk.
:param writefreq: traj dump frequency in LAMMPS
:param infobjs: list of file pointers to input (unordered) trajs.
"""
nframes = int(nprod / writefreq)
for n in temp_inds:
# open string-buffer and file
buf = IOBuffer()
of = readwrite(outtrajfns[n], "wb")
# get frames
abs_temp_ind = np.argmin(abs(temps - outtemps[n]))
frametuple = frametuple_dict[abs_temp_ind][-nframes:]
# write frames to buffer
if me == ROOT:
pb = tqdm(frametuple,
desc=("Buffering trajectories for writing"),
leave=True,
position=ROOT + 2 * me,
unit='frame/replica',
unit_scale=True)
iterable = pb
else:
iterable = frametuple
for i, (rep, frame) in enumerate(iterable):
infobj = infobjs[rep]
start_ptr = int(byte_inds[rep][frame, 1])
stop_ptr = int(byte_inds[rep][frame + 1, 1])
byte_len = stop_ptr - start_ptr
infobj.seek(start_ptr)
buf.write(infobj.read(byte_len))
if me == ROOT: pb.close()
# write buffer to disk
if me == ROOT: print("Writing buffer to file")
of.write(buf.getvalue())
of.close()
buf.close()
for i in infobjs:
i.close()
return
def process_archive(archive_filepath, prefix=".", img_format="jpg"):
logger = IOBuffer()
archive_path = str(archive_filepath)
case = dcmtools.decompress_case(archive_path)
studyInstanceUID, study = dcmtools.load_study(case)
if studyInstanceUID is None:
logger.write(
'No StudyInstanceUID found in filename {}'.format(archive_path))
return
folder = Path(prefix) / studyInstanceUID
folder.mkdir(exist_ok=True)
logger.write("Saving meta info: {}".format(folder))
studies_meta = []
for series in study:
img = series[0]
meta_info = {
'Filename': str(archive_path),
'StudyInstanceUID': str(img.StudyInstanceUID),
'SeriesInstanceUID': str(img.SeriesInstanceUID),
'PatientSex': str(img.PatientSex),
'AcquisitionDate': int(img.AcquisitionDate),
'SliceThicknessX': float(img.SliceThicknessX),
'SliceThicknessY': float(img.SliceThicknessY),
'SliceThicknessZ': float(img.SliceThicknessZ),
'PatientAge': int(img.PatientAge[:-1]),
'StudyDescription': str(img.StudyDescription),
'SeriesDescription': str(img.SeriesDescription)
}
studies_meta.append(meta_info)
logger.write(str(meta_info))
voxels, spacing = dcmtools.clip_voxel_values(series)
mid_slice_X = voxels[int(voxels.shape[0] // 2), :, :]
mid_slice_Y = np.flipud(voxels[:, int(voxels.shape[1] // 2), :])
mid_slice_Z = np.flipud(voxels[:, :, int(voxels.shape[2] // 2)])
scipy.misc.imsave(
folder / "{}-{}.{}".format(img.SeriesInstanceUID, "X", img_format),
mid_slice_X)
scipy.misc.imsave(
folder / "{}-{}.{}".format(img.SeriesInstanceUID, "Y", img_format),
mid_slice_Y)
scipy.misc.imsave(
folder / "{}-{}.{}".format(img.SeriesInstanceUID, "Z", img_format),
mid_slice_Z)
df_meta_info = pd.DataFrame(studies_meta)
df_meta_info.to_csv(folder / "{}.{}".format(studyInstanceUID, "csv"))
# rewind the tape
logger.seek(0)
return logger
# a list of all files to be processed
archives_todo = []
# traverse directory and find archives
if path.is_dir():
# grab the first DICOM image in the directory
for (dirpath, dirnames, filenames) in os.walk(str(path)):
for fname in filenames:
fp = Path(dirpath) / fname
if fp.suffix[1:] in archive_suffixes:
archives_todo.append(fp)
elif path.is_file():
# only one file given as input
if path.suffix[1:] in archive_suffixes:
archives_todo.append(path)
logger = IOBuffer()
# start worker processes
with tqdm(desc="CTInfo", total=len(archives_todo), unit="file") as pbar:
with Pool(processes=multiprocessing.cpu_count()) as pool:
for log in tqdm(
pool.imap_unordered(
partial(process_archive, prefix=args.prefix),
archives_todo)):
pbar.update(1)
logger.write(log.getvalue())
log_filep = Path(args.prefix) / "{}.log".format("MetaInfo")
with open(log_filep, "w") as fp:
# rewind the tape
logger.seek(0)
shutil.copyfileobj(logger, fp)
