def export_skeleton_as_nrrd_async(skeleton_id,
source_ref,
target_ref,
user_id,
mirror=True,
create_message=True):
result = export_skeleton_as_nrrd(skeleton_id, source_ref, target_ref,
user_id, mirror)
if create_message:
msg = Message()
msg.user = User.objects.get(pk=int(user_id))
msg.read = False
if result['errors']:
msg.title = "No NRRD file could be creaed for skeleton {}".format(
skeleton_id)
msg.text = "There was at least one error during the NRRD export: {}".format(
'\n'.join(result['errors']))
msg.action = ""
else:
url = urljoin(
urljoin(settings.MEDIA_URL,
settings.MEDIA_EXPORT_SUBDIRECTORY),
result['nrrd_name'])
msg.title = "Exported skeleton {} as NRRD file".format(skeleton_id)
msg.text = "The requested skeleton was exported as NRRD file. You " \
"can download it from this location: <a href='{}'>{}</a>".format(url, url)
msg.action = url
msg.save()
return "Errors: {}".format('\n'.join(
result['errors'])) if result['errors'] else result['nrrd_path']
def process_crop_job(job, create_message=True):
""" This method does the actual cropping. It controls the data extraction
and the creation of the sub-stack. It can be executed as Celery task.
"""
try:
# Create the sub-stack
cropped_stack = extract_substack(job)
# Create tho output image
outputImage = ImageList()
for img in cropped_stack:
outputImage.append(img)
# Save the resulting micro_stack to a temporary location
no_error_occured = True
error_message = ""
# Only produce an image if parts of stacks are within the output
if len(cropped_stack) > 0:
outputImage.writeImages(job.output_path.encode('ascii', 'ignore'))
# Add some meta data to the image
addMetaData(job.output_path, job, cropped_stack)
else:
no_error_occured = False
error_message = "A region outside the stack has been selected. " \
"Therefore, no image was produced."
except (IOError, OSError, ValueError) as e:
no_error_occured = False
error_message = str(e)
# Delete the file if parts of it have been written already
if os.path.exists(job.output_path):
os.remove(job.output_path)
if create_message:
# Create a notification message
bb_text = "( %s, %s, %s ) -> ( %s, %s, %s )" % (job.x_min, job.y_min, \
job.z_min, job.x_max, job.y_max, job.z_max)
msg = Message()
msg.user = User.objects.get(pk=int(job.user.id))
msg.read = False
if no_error_occured:
file_name = os.path.basename(job.output_path)
url = os.path.join(settings.CATMAID_URL,
"crop/download/" + file_name + "/")
msg.title = "Microstack finished"
msg.text = "The requested microstack %s is finished. You can " \
"download it from this location: <a href='%s'>%s</a>" % \
(bb_text, url, url)
msg.action = url
else:
msg.title = "Microstack could not be created"
msg.text = "The requested microstack %s could not be created due " \
"to an error while saving the result (%s)." % \
(bb_text, error_message)
msg.action = ""
msg.save()
return job.output_path if no_error_occured else error_message
def process_crop_job(job, create_message=True):
""" This method does the actual cropping. It controls the data extraction
and the creation of the sub-stack. It can be executed as Celery task.
"""
try:
# Create the sub-stack
cropped_stack = extract_substack(job)
# Create tho output image
outputImage = ImageList()
for img in cropped_stack:
outputImage.append( img )
# Save the resulting micro_stack to a temporary location
no_error_occured = True
error_message = ""
# Only produce an image if parts of stacks are within the output
if len( cropped_stack ) > 0:
outputImage.writeImages( job.output_path.encode('ascii', 'ignore') )
# Add some meta data to the image
addMetaData( job.output_path, job, cropped_stack )
else:
no_error_occured = False
error_message = "A region outside the stack has been selected. " \
"Therefore, no image was produced."
except (IOError, OSError, ValueError) as e:
no_error_occured = False
error_message = str(e)
# Delete the file if parts of it have been written already
if os.path.exists( job.output_path ):
os.remove( job.output_path )
if create_message:
# Create a notification message
bb_text = "( %s, %s, %s ) -> ( %s, %s, %s )" % (job.x_min, job.y_min, \
job.z_min, job.x_max, job.y_max, job.z_max)
msg = Message()
msg.user = User.objects.get(pk=int(job.user.id))
msg.read = False
if no_error_occured:
file_name = os.path.basename( job.output_path )
url = os.path.join( settings.CATMAID_URL, "crop/download/" + file_name + "/")
msg.title = "Microstack finished"
msg.text = "The requested microstack %s is finished. You can " \
"download it from this location: <a href='%s'>%s</a>" % \
(bb_text, url, url)
msg.action = url
else:
msg.title = "Microstack could not be created"
msg.text = "The requested microstack %s could not be created due " \
"to an error while saving the result (%s)." % \
(bb_text, error_message)
msg.action = ""
msg.save()
return job.output_path if no_error_occured else error_message
def process_crop_job(job: CropJob, create_message=True) -> str:
""" This method does the actual cropping. It controls the data extraction
and the creation of the sub-stack. It can be executed as Celery task.
"""
try:
# Create the sub-stack
cropped_stack = extract_substack(job)
# Save the resulting micro_stack to a temporary location
no_error_occured = True
error_message = ""
# Only produce an image if parts of stacks are within the output
if len(cropped_stack) > 0:
metadata = job.create_tiff_metadata(len(cropped_stack))
cropped_stack[0].save(job.output_path, compression="raw", save_all=True,
append_images=cropped_stack[1:], tiffinfo=metadata)
else:
no_error_occured = False
error_message = "A region outside the stack has been selected. " \
"Therefore, no image was produced."
except (IOError, OSError, ValueError) as e:
no_error_occured = False
error_message = str(e)
# Delete the file if parts of it have been written already
if os.path.exists( job.output_path ):
os.remove( job.output_path )
if create_message:
# Create a notification message
bb_text = "( %s, %s, %s ) -> ( %s, %s, %s )" % (job.x_min, job.y_min, \
job.z_min, job.x_max, job.y_max, job.z_max)
user = User.objects.get(pk=int(job.user.id))
msg = Message()
msg.user = user
msg.read = False
if no_error_occured:
file_name = os.path.basename( job.output_path )
url = os.path.join( settings.CATMAID_URL, "crop/download/" + file_name + "/")
msg.title = "Microstack finished"
msg.text = "The requested microstack %s is finished. You can " \
"download it from this location: <a href='%s'>%s</a>" % \
(bb_text, url, url)
msg.action = url
else:
msg.title = "Microstack could not be created"
msg.text = "The requested microstack %s could not be created due " \
"to an error while saving the result (%s)." % \
(bb_text, error_message)
msg.action = ""
msg.save()
notify_user(user.id, msg.id, msg.title)
return job.output_path if no_error_occured else error_message
def create_message(self, title, message, url):
msg = Message()
msg.user = User.objects.get(pk=int(self.job.user.id))
msg.read = False
msg.title = title
msg.text = message
msg.action = url
msg.save()
def create_message(self, title, message, url):
msg = Message()
msg.user = User.objects.get(pk=int(self.job.user.id))
msg.read = False
msg.title = title
msg.text = message
msg.action = url
msg.save()
def export_skeleton_as_nrrd_async(skeleton_id, source_ref, target_ref, user_id,
mirror=True, create_message=True):
result = export_skeleton_as_nrrd(skeleton_id, source_ref, target_ref,
user_id, mirror)
if create_message:
msg = Message()
msg.user = User.objects.get(pk=int(user_id))
msg.read = False
if result['errors']:
msg.title = "No NRRD file could be creaed for skeleton {}".format(skeleton_id)
msg.text = "There was at least one error during the NRRD export: {}".format('\n'.join(result['errors']))
msg.action = ""
else:
url = urljoin(urljoin(settings.MEDIA_URL, settings.MEDIA_EXPORT_SUBDIRECTORY), result['nrrd_name'])
msg.title = "Exported skeleton {} as NRRD file".format(skeleton_id)
msg.text = "The requested skeleton was exported as NRRD file. You " \
"can download it from this location: <a href='{}'>{}</a>".format(url, url)
msg.action = url
msg.save()
return "Errors: {}".format('\n'.join(result['errors'])) if result['errors'] else result['nrrd_path']
if create_message:
# Create a notification message
bb_text = "( %s, %s, %s ) -> ( %s, %s, %s )" % (job.x_min, job.y_min, \
job.z_min, job.x_max, job.y_max, job.z_max)
msg = Message()
msg.user = User.objects.get(pk=int(job.user.id))
msg.read = False
if no_error_occured:
file_name = os.path.basename( job.output_path )
url = os.path.join( settings.CATMAID_URL, "crop/download/" + file_name + "/")
msg.title = "Microstack finished"
msg.text = "The requested microstack %s is finished. You can " \
"download it from this location: <a href='%s'>%s</a>" % \
(bb_text, url, url)
msg.action = url
else:
msg.title = "Microstack could not be created"
msg.text = "The requested microstack %s could not be created due " \
"to an error while saving the result (%s)." % \
(bb_text, error_message)
msg.action = ""
msg.save()
return None if no_error_occured else error_message
def start_asynch_process( job ):
""" It launches the data extraction and sub-stack building as a seperate process.
This process uses the addmessage command with manage.py to write a message for the
user into the data base once the process is done.
"""
def query_segmentation_async(
result,
project_id,
user_id,
ssh_key,
ssh_user,
local_temp_dir,
segmentations_dir,
server,
job_name,
job_type,
):
result.status = "computing"
result.save()
msg_user(user_id, "autoproofreader-result-update", {"status": "computing"})
# copy temp files from django local temp media storage to server temp storage
setup = (
"scp -i {ssh_key} -pr {local_dir} "
+ "{ssh_user}@{server_address}:{server_results_dir}/{job_dir}"
).format(
**{
"local_dir": local_temp_dir,
"server_address": server["address"],
"server_results_dir": server["results_dir"],
"job_dir": job_name,
"ssh_key": ssh_key,
"ssh_user": ssh_user,
}
)
files = {}
for f in local_temp_dir.iterdir():
files[f.name.split(".")[0]] = Path(
"~/", server["results_dir"], job_name, f.name
)
if job_type == "diluvian":
extra_parameters = (
"--model-weights-file {model_file} "
+ "--model-training-config {model_config_file} "
+ "--model-job-config {job_config_file} "
+ "--volume-file {volume_file} "
).format(
**{
"model_file": server["model_file"],
"model_config_file": files["model_config"],
"job_config_file": files["diluvian_config"],
"volume_file": files["volume"],
}
)
elif job_type == "cached_lsd":
extra_parameters = "--cached-lsd-config {} ".format(files["cached_lsd_config"])
else:
extra_parameters = ""
# connect to the server and run the autoproofreader algorithm on the provided skeleton
query_seg = (
"ssh -i {ssh_key} {ssh_user}@{server}\n"
+ "source {server_ff_env_path}\n"
+ "sarbor-error-detector "
+ "--skeleton-csv {skeleton_file} "
+ "--sarbor-config {sarbor_config} "
+ "--output-file {output_file} "
+ "{segmentation_type} "
+ "{type_parameters}"
).format(
**{
"ssh_key": ssh_key,
"ssh_user": ssh_user,
"server": server["address"],
"server_ff_env_path": server["env_source"],
"skeleton_file": files["skeleton"],
"sarbor_config": files["sarbor_config"],
"output_file": Path(server["results_dir"], job_name, "outputs"),
"segmentation_type": job_type.replace("_", "-"),
"type_parameters": extra_parameters,
}
)
# Copy the numpy file containing the volume mesh and the csv containing the node connections
# predicted by the autoproofreader run.
fetch_files = (
"scp -i {ssh_key} -r {ssh_user}@{server}:"
+ "{server_results_dir}/{server_job_dir}/* {local_temp_dir}\n"
).format(
**{
"ssh_key": ssh_key,
"ssh_user": ssh_user,
"server": server["address"],
"server_results_dir": server["results_dir"],
"server_job_dir": job_name,
"local_temp_dir": local_temp_dir,
}
)
process = subprocess.Popen(
"/bin/bash", stdin=subprocess.PIPE, stdout=subprocess.PIPE, encoding="utf8"
)
out, err = process.communicate(setup)
logging.info(out)
process = subprocess.Popen(
"/bin/bash", stdin=subprocess.PIPE, stdout=subprocess.PIPE, encoding="utf8"
)
out, err = process.communicate(query_seg)
logging.info(out)
process = subprocess.Popen(
"/bin/bash", stdin=subprocess.PIPE, stdout=subprocess.PIPE, encoding="utf8"
)
out, err = process.communicate(fetch_files)
logging.info(out)
nodes_path = Path(local_temp_dir, "outputs", "nodes.obj")
Node = namedtuple("Node", ["node_id", "parent_id", "x", "y", "z"])
# b is the branch score. c is the connectivity score
Ranking = namedtuple(
"Ranking", ["node_id", "parent_id", "c", "b", "b_dx", "b_dy", "b_dz"]
)
Combined = namedtuple(
"Combined",
["node_id", "parent_id", "x", "y", "z", "b", "c", "b_dx", "b_dy", "b_dz"],
)
# Nodes are mandatory
if nodes_path.exists():
nodes = {row[0]: Node(*row) for row in pickle.load(nodes_path.open("rb"))}
else:
result.status = "failed"
result.save()
return "failed"
rankings_path = Path(local_temp_dir, "outputs", "rankings.obj")
# Rankings are mandatory
if rankings_path.exists():
with rankings_path.open("r") as f:
rankings = {
row[0]: Ranking(*row) for row in pickle.load(rankings_path.open("rb"))
}
node_data = [
Combined(**{**nodes[nid]._asdict(), **rankings[nid]._asdict()})
for nid in nodes.keys()
]
proofread_nodes = [
ProofreadTreeNodes(
node_id=row.node_id,
parent_id=row.parent_id,
x=row.x,
y=row.y,
z=row.z,
connectivity_score=row.c,
branch_score=row.b,
branch_dx=row.b_dx,
branch_dy=row.b_dy,
branch_dz=row.b_dz,
reviewed=False,
result=result,
user_id=user_id,
project_id=project_id,
editor_id=user_id,
)
for row in node_data
]
ProofreadTreeNodes.objects.bulk_create(proofread_nodes)
else:
result.status = "failed"
result.save()
return "failed"
mesh_path = Path(local_temp_dir, "outputs", "mesh.stl")
# Mesh is optional
if mesh_path.exists():
with mesh_path.open("r") as f:
stl_str = f.read()
try:
vertices, triangles = _stl_ascii_to_indexed_triangles(stl_str)
except InvalidSTLError as e:
raise ValueError("Invalid STL file ({})".format(str(e)))
mesh = TriangleMeshVolume(
project_id,
user_id,
{"type": "trimesh", "title": job_name, "mesh": [vertices, triangles]},
)
mesh_volume = mesh.save()
result.volume = Volume.objects.get(id=mesh_volume)
segmentation_path = Path(local_temp_dir, "outputs", "segmentations.n5")
segmentation_dir = Path(segmentations_dir)
if segmentation_path.exists():
segmentation_dir.mkdir(parents=True, exist_ok=True)
segmentation_path.rename(segmentation_dir / "segmentations.n5")
cleanup = (
"rm -r {local_temp_dir}\n"
+ "ssh -i {ssh_key} {ssh_user}@{server}\n"
+ "rm -r {server_results_dir}/{server_job_dir}"
).format(
**{
"ssh_key": ssh_key,
"ssh_user": ssh_user,
"server": server["address"],
"server_results_dir": server["results_dir"],
"server_job_dir": job_name,
"local_temp_dir": local_temp_dir,
}
)
process = subprocess.Popen(
"/bin/bash", stdin=subprocess.PIPE, stdout=subprocess.PIPE, encoding="utf8"
)
out, err = process.communicate(cleanup)
logging.info(out)
msg = Message()
msg.user = User.objects.get(pk=int(user_id))
msg.read = False
msg.title = "Job {} complete!"
msg.text = "IM DOING SOME STUFF, CHECK IT OUT"
msg.action = "localhost:8000"
notify_user(user_id, msg.id, msg.title)
result.completion_time = datetime.datetime.now(pytz.utc)
result.status = "complete"
result.save()
msg_user(user_id, "autoproofreader-result-update", {"status": "completed"})
return "complete"
# Create a notification message
bb_text = "( %s, %s, %s ) -> ( %s, %s, %s )" % (job.x_min, job.y_min, \
job.z_min, job.x_max, job.y_max, job.z_max)
msg = Message()
msg.user = User.objects.get(pk=int(job.user.id))
msg.read = False
if no_error_occured:
file_name = os.path.basename(job.output_path)
url = os.path.join(settings.CATMAID_URL,
"crop/download/" + file_name + "/")
msg.title = "Microstack finished"
msg.text = "The requested microstack %s is finished. You can " \
"download it from this location: <a href='%s'>%s</a>" % \
(bb_text, url, url)
msg.action = url
else:
msg.title = "Microstack could not be created"
msg.text = "The requested microstack %s could not be created due " \
"to an error while saving the result (%s)." % \
(bb_text, error_message)
msg.action = ""
msg.save()
return None if no_error_occured else error_message
def start_asynch_process(job):
""" It launches the data extraction and sub-stack building as a seperate process.
This process uses the addmessage command with manage.py to write a message for the
user into the data base once the process is done.