def delete(self, *args, **kwargs):
importer_locations = []
upload_files = [_file.file for _file in UploadFile.objects.filter(upload=self)]
super(Upload, self).delete(*args, **kwargs)
try:
session = gs_uploader.get_session(self.import_id)
except (NotFound, Exception):
session = None
if session:
for task in session.tasks:
if getattr(task, 'data'):
importer_locations.append(
getattr(task.data, 'location'))
try:
session.delete()
except Exception:
logging.warning('error deleting upload session')
for _file in upload_files:
try:
if os.path.isfile(_file.path):
os.remove(_file.path)
except Exception as e:
logger.warning(e)
for _location in importer_locations:
try:
shutil.rmtree(_location)
except Exception as e:
logger.warning(e)
if self.upload_dir and os.path.exists(self.upload_dir):
try:
shutil.rmtree(self.upload_dir)
except Exception as e:
logger.warning(e)
def delete(self, *args, **kwargs):
importer_locations = []
super().delete(*args, **kwargs)
try:
session = gs_uploader.get_session(self.import_id)
except (NotFound, Exception):
session = None
if session:
for task in session.tasks:
if getattr(task, 'data'):
importer_locations.append(getattr(task.data, 'location'))
try:
session.delete()
except Exception:
logging.warning('error deleting upload session')
for _location in importer_locations:
try:
shutil.rmtree(_location)
except Exception as e:
logger.warning(e)
# here we are deleting the local that soon will be removed
if self.upload_dir and os.path.exists(self.upload_dir):
try:
shutil.rmtree(self.upload_dir)
except Exception as e:
logger.warning(e)
def _update_upload_session_state(self, upload_session_id: int):
"""Task invoked by 'upload_workflow.chord' in order to process all the 'PENDING' Upload tasks."""
lock_id = f'{self.request.id}'
with AcquireLock(lock_id) as lock:
if lock.acquire() is True:
_upload = Upload.objects.get(id=upload_session_id)
session = _upload.get_session.import_session
if not session or session.state != enumerations.STATE_COMPLETE:
session = gs_uploader.get_session(_upload.import_id)
if session:
try:
content = next_step_response(None, _upload.get_session).content
if isinstance(content, bytes):
content = content.decode('UTF-8')
response_json = json.loads(content)
_success = response_json.get('success', False)
_redirect_to = response_json.get('redirect_to', '')
if _success:
if 'upload/final' not in _redirect_to and 'upload/check' not in _redirect_to:
_upload.set_resume_url(_redirect_to)
_upload.set_processing_state(enumerations.STATE_WAITING)
else:
if session.state == enumerations.STATE_COMPLETE and _upload.state == enumerations.STATE_PENDING:
if not _upload.resource or not _upload.resource.processed:
final_step_view(None, _upload.get_session)
_upload.set_processing_state(enumerations.STATE_RUNNING)
except (NotFound, Exception) as e:
logger.exception(e)
if _upload.state not in (enumerations.STATE_COMPLETE, enumerations.STATE_PROCESSED):
_upload.set_processing_state(enumerations.STATE_INVALID)
if _upload.resource:
resource_manager.delete(_upload.resource.uuid)
def run_import(upload_session, async):
"""Run the import, possibly asynchronously.
Returns the target datastore.
"""
import_session = upload_session.import_session
import_session = gs_uploader.get_session(import_session.id)
task = import_session.tasks[0]
if import_session.state == 'INCOMPLETE':
if task.state != 'ERROR':
raise Exception('unknown item state: %s' % task.state)
elif import_session.state == 'PENDING' and task.target.store_type == 'coverageStore':
if task.state == 'READY':
import_session.commit(async)
elif import_session.state == 'PENDING' and task.target.store_type == 'coverageStore':
if task.state == 'READY':
import_session.commit(async)
# if a target datastore is configured, ensure the datastore exists
# in geoserver and set the uploader target appropriately
if ogc_server_settings.GEOGIG_ENABLED and upload_session.geogig is True \
and task.target.store_type != 'coverageStore':
target = create_geoserver_db_featurestore(
store_type='geogig',
store_name=upload_session.geogig_store,
author_name=upload_session.user.username,
author_email=upload_session.user.email)
_log(
'setting target datastore %s %s',
target.name,
target.workspace.name)
task.set_target(target.name, target.workspace.name)
elif ogc_server_settings.datastore_db and task.target.store_type != 'coverageStore':
target = create_geoserver_db_featurestore()
_log(
'setting target datastore %s %s',
target.name,
target.workspace.name)
task.set_target(target.name, target.workspace.name)
else:
target = task.target
if upload_session.update_mode:
_log('setting updateMode to %s', upload_session.update_mode)
task.set_update_mode(upload_session.update_mode)
_log('running import session')
# run async if using a database
import_session.commit(async)
# @todo check status of import session - it may fail, but due to protocol,
# this will not be reported during the commit
return target
def run_import(upload_session, async_upload=_ASYNC_UPLOAD):
"""Run the import, possibly asynchronously.
Returns the target datastore.
"""
# run_import can raise an exception which callers should handle
import_session = upload_session.import_session
import_session = gs_uploader.get_session(import_session.id)
task = import_session.tasks[0]
import_execution_requested = False
if import_session.state == 'INCOMPLETE':
if task.state != 'ERROR':
raise Exception(_('unknown item state: %s' % task.state))
elif import_session.state == 'PENDING' and task.target.store_type == 'coverageStore':
if task.state == 'READY':
import_session.commit(async_upload)
import_execution_requested = True
if task.state == 'ERROR':
progress = task.get_progress()
raise Exception(
_('error during import: %s' % progress.get('message')))
# if a target datastore is configured, ensure the datastore exists
# in geoserver and set the uploader target appropriately
if ogc_server_settings.GEOGIG_ENABLED and upload_session.geogig is True \
and task.target.store_type != 'coverageStore':
target = create_geoserver_db_featurestore(
store_type='geogig',
store_name=upload_session.geogig_store,
author_name=upload_session.user.username,
author_email=upload_session.user.email)
_log('setting target datastore %s %s', target.name,
target.workspace.name)
task.set_target(target.name, target.workspace.name)
elif ogc_server_settings.datastore_db and task.target.store_type != 'coverageStore':
target = create_geoserver_db_featurestore(
# store_name=ogc_server_settings.DATASTORE,
store_name=ogc_server_settings.datastore_db['NAME'])
_log('setting target datastore %s %s', target.name,
target.workspace.name)
task.set_target(target.name, target.workspace.name)
else:
target = task.target
if upload_session.update_mode:
_log('setting updateMode to %s', upload_session.update_mode)
task.set_update_mode(upload_session.update_mode)
_log('running import session')
# run async if using a database
if not import_execution_requested:
import_session.commit(async_upload)
# @todo check status of import session - it may fail, but due to protocol,
# this will not be reported during the commit
return target
def run_import(upload_session, async):
"""Run the import, possibly asynchronously.
Returns the target datastore.
"""
import_session = upload_session.import_session
import_session = gs_uploader.get_session(import_session.id)
task = import_session.tasks[0]
if import_session.state == 'INCOMPLETE':
if task.state != 'ERROR':
raise Exception('unknown item state: %s' % task.state)
elif import_session.state == 'PENDING' and task.target.store_type == 'coverageStore':
if task.state == 'READY':
import_session.commit(async)
elif import_session.state == 'PENDING' and task.target.store_type == 'coverageStore':
if task.state == 'READY':
import_session.commit(async)
# if a target datastore is configured, ensure the datastore exists
# in geoserver and set the uploader target appropriately
if ogc_server_settings.GEOGIG_ENABLED and upload_session.geogig is True \
and task.target.store_type != 'coverageStore':
target = create_geoserver_db_featurestore(
store_type='geogig',
store_name=upload_session.geogig_store)
_log(
'setting target datastore %s %s',
target.name,
target.workspace.name)
task.set_target(target.name, target.workspace.name)
elif ogc_server_settings.datastore_db and task.target.store_type != 'coverageStore':
target = create_geoserver_db_featurestore()
_log(
'setting target datastore %s %s',
target.name,
target.workspace.name)
task.set_target(target.name, target.workspace.name)
else:
target = task.target
if upload_session.update_mode:
_log('setting updateMode to %s', upload_session.update_mode)
task.set_update_mode(upload_session.update_mode)
_log('running import session')
# run async if using a database
import_session.commit(async)
# @todo check status of import session - it may fail, but due to protocol,
# this will not be reported during the commit
return target
def get_resume_url(self):
if self.state == Upload.STATE_WAITING and self.import_id:
return f"{reverse('data_upload')}?id={self.import_id}"
else:
session = None
try:
if not self.import_id:
raise NotFound
session = self.get_session.import_session
if not session or session.state != Upload.STATE_COMPLETE:
session = gs_uploader.get_session(self.import_id)
except (NotFound, Exception):
if self.state not in (Upload.STATE_COMPLETE,
Upload.STATE_PROCESSED):
self.state = Upload.STATE_INVALID
Upload.objects.filter(id=self.id).update(
state=Upload.STATE_INVALID)
if session:
lock_id = f'{self.import_id}'
with AcquireLock(lock_id) as lock:
if lock.acquire() is True:
try:
content = next_step_response(
None, self.get_session).content
if isinstance(content, bytes):
content = content.decode('UTF-8')
response_json = json.loads(content)
if response_json[
'success'] and 'redirect_to' in response_json:
if 'upload/final' not in response_json[
'redirect_to'] and 'upload/check' not in response_json[
'redirect_to']:
self.state = Upload.STATE_WAITING
Upload.objects.filter(id=self.id).update(
state=Upload.STATE_WAITING)
return f"{reverse('data_upload')}?id={self.import_id}"
else:
next = get_next_step(self.get_session)
if next == 'final' and session.state == Upload.STATE_COMPLETE and self.state == Upload.STATE_PENDING:
if not self.layer or not self.layer.processed:
from .views import final_step_view
final_step_view(
None, self.get_session)
self.state = Upload.STATE_RUNNING
Upload.objects.filter(
id=self.id).update(
state=Upload.STATE_RUNNING)
except (NotFound, Exception) as e:
logger.exception(e)
if self.state not in (Upload.STATE_COMPLETE,
Upload.STATE_PROCESSED):
self.state = Upload.STATE_INVALID
Upload.objects.filter(id=self.id).update(
state=Upload.STATE_INVALID)
return None
def _update_upload_session_state(self, upload_session_id: int):
"""Task invoked by 'upload_workflow.chord' in order to process all the 'PENDING' Upload tasks."""
_upload = Upload.objects.get(id=upload_session_id)
session = _upload.get_session.import_session
if not session or session.state != enumerations.STATE_COMPLETE:
session = gs_uploader.get_session(_upload.import_id)
if session:
try:
_response = next_step_response(None, _upload.get_session)
_upload.refresh_from_db()
_content = _response.content
if isinstance(_content, bytes):
_content = _content.decode('UTF-8')
_response_json = json.loads(_content)
_success = _response_json.get('success', False)
_redirect_to = _response_json.get('redirect_to', '')
if _success:
if 'upload/final' not in _redirect_to and 'upload/check' not in _redirect_to:
_upload.set_resume_url(_redirect_to)
_upload.set_processing_state(enumerations.STATE_WAITING)
else:
if session.state == enumerations.STATE_COMPLETE and _upload.state == enumerations.STATE_PENDING:
if not _upload.resource or not _upload.resource.processed:
_response = final_step_view(None, _upload.get_session)
_upload.refresh_from_db()
_content = _response.content
if isinstance(_content, bytes):
_content = _content.decode('UTF-8')
_response_json = json.loads(_content)
_success = _response_json.get('success', False)
_status = _response_json.get('status', 'error')
if _status == 'error':
# GeoNode Layer creation errored!
_upload.set_processing_state(enumerations.STATE_INVALID)
elif _status == 'pending':
# GeoNode Layer not ready yet...
_upload.set_processing_state(enumerations.STATE_PENDING)
elif _upload.state != enumerations.STATE_PROCESSED:
if _upload.resource and _upload.resource.processed:
# GeoNode Layer successfully processed...
_upload.set_processing_state(enumerations.STATE_PROCESSED)
else:
# GeoNode Layer updating...
_upload.set_processing_state(enumerations.STATE_RUNNING)
logger.debug(f"Upload {upload_session_id} updated with state {_upload.state}.")
except (NotFound, Exception) as e:
logger.exception(e)
if _upload.state not in (enumerations.STATE_COMPLETE, enumerations.STATE_PROCESSED):
_upload.set_processing_state(enumerations.STATE_INVALID)
logger.error(f"Upload {upload_session_id} deleted with state {_upload.state}.")
def delete(self, cascade=True):
models.Model.delete(self)
if cascade:
try:
session = gs_uploader.get_session(self.import_id)
except NotFound:
session = None
if session:
try:
session.delete()
except:
logging.exception('error deleting upload session')
if self.upload_dir and path.exists(self.upload_dir):
shutil.rmtree(self.upload_dir)
def delete(self, cascade=True):
models.Model.delete(self)
if cascade:
try:
session = gs_uploader.get_session(self.import_id)
except NotFound:
session = None
if session:
try:
session.delete()
except BaseException:
logging.exception('error deleting upload session')
if self.upload_dir and path.exists(self.upload_dir):
shutil.rmtree(self.upload_dir)
def run_import(upload_session, async_upload=_ASYNC_UPLOAD):
"""Run the import, possibly asynchronously.
Returns the target datastore.
"""
# run_import can raise an exception which callers should handle
import_session = upload_session.import_session
import_session = gs_uploader.get_session(import_session.id)
task = import_session.tasks[0]
import_execution_requested = False
if import_session.state == 'INCOMPLETE':
if task.state != 'ERROR':
raise Exception(_('unknown item state: %s' % task.state))
elif import_session.state == 'PENDING' and task.target.store_type == 'coverageStore':
if task.state == 'READY':
import_session.commit(async_upload)
import_execution_requested = True
if task.state == 'ERROR':
progress = task.get_progress()
raise Exception(_(
'error during import: %s' %
progress.get('message')))
# if a target datastore is configured, ensure the datastore exists
# in geoserver and set the uploader target appropriately
if ogc_server_settings.datastore_db and task.target.store_type != 'coverageStore':
target = create_geoserver_db_featurestore(
# store_name=ogc_server_settings.DATASTORE,
store_name=ogc_server_settings.datastore_db['NAME']
)
_log(
'setting target datastore %s %s',
target.name,
target.workspace.name)
task.set_target(target.name, target.workspace.name)
else:
target = task.target
if upload_session.update_mode:
_log('setting updateMode to %s', upload_session.update_mode)
task.set_update_mode(upload_session.update_mode)
_log('running import session')
# run async if using a database
if not import_execution_requested:
import_session.commit(async_upload)
# @todo check status of import session - it may fail, but due to protocol,
# this will not be reported during the commit
return target
def get_import_url(self):
session = None
try:
if not self.import_id:
raise NotFound
session = self.get_session.import_session
if not session or session.state != enumerations.STATE_COMPLETE:
session = gs_uploader.get_session(self.import_id)
except (NotFound, Exception):
if self.state not in (enumerations.STATE_COMPLETE, enumerations.STATE_PROCESSED):
self.set_processing_state(enumerations.STATE_INVALID)
if session and self.state != enumerations.STATE_INVALID:
return f"{ogc_server_settings.LOCATION}rest/imports/{session.id}"
else:
return None
def get_import_url(self):
session = None
try:
if not self.import_id:
raise NotFound
session = self.get_session.import_session
if not session or session.state != Upload.STATE_COMPLETE:
session = gs_uploader.get_session(self.import_id)
except (NotFound, Exception):
if self.state not in (Upload.STATE_COMPLETE, Upload.STATE_PROCESSED):
self.state = Upload.STATE_INVALID
Upload.objects.filter(id=self.id).update(state=Upload.STATE_INVALID)
if session and self.state != Upload.STATE_INVALID:
return f"{ogc_server_settings.LOCATION}rest/imports/{session.id}"
else:
return None
def delete(self, *args, **kwargs):
importer_locations = []
super().delete(*args, **kwargs)
try:
session = gs_uploader.get_session(self.import_id)
except (NotFound, Exception):
session = None
if session:
for task in session.tasks:
if getattr(task, 'data'):
importer_locations.append(
getattr(task.data, 'location'))
try:
session.delete()
except Exception:
logging.warning('error deleting upload session')
# we delete directly the folder with the files of the resource
if self.resource:
for _file in self.resource.files:
try:
if storage_manager.exists(_file):
storage_manager.delete(_file)
except Exception as e:
logger.warning(e)
# Do we want to delete the files also from the resource?
ResourceBase.objects.filter(id=self.resource.id).update(files={})
for _location in importer_locations:
try:
shutil.rmtree(_location)
except Exception as e:
logger.warning(e)
# here we are deleting the local that soon will be removed
if self.upload_dir and os.path.exists(self.upload_dir):
try:
shutil.rmtree(self.upload_dir)
except Exception as e:
logger.warning(e)
layer_values = values or _get_layer_values(layer, upload_session, expand=0)
att_list = _get_time_dimensions(layer, upload_session)
_is_eligible = att_list or False
if upload_session and _is_eligible:
upload_session.time = True
return (_is_eligible, layer_values)
def run_import(upload_session, async=_ASYNC_UPLOAD):
"""Run the import, possibly asynchronously.
Returns the target datastore.
"""
# run_import can raise an exception which callers should handle
import_session = upload_session.import_session
import_session = gs_uploader.get_session(import_session.id)
task = import_session.tasks[0]
import_execution_requested = False
if import_session.state == 'INCOMPLETE':
if task.state != 'ERROR':
raise Exception(_('unknown item state: %s' % task.state))
elif import_session.state == 'PENDING' and task.target.store_type == 'coverageStore':
if task.state == 'READY':
import_session.commit(async)
import_execution_requested = True
if task.state == 'ERROR':
progress = task.get_progress()
raise Exception(_(
'error during import: %s' %
progress.get('message')))
layer_values = values or _get_layer_values(layer, upload_session, expand=0)
att_list = _get_time_dimensions(layer, upload_session)
_is_eligible = att_list or False
if upload_session and _is_eligible:
upload_session.time = True
return (_is_eligible, layer_values)
def run_import(upload_session, async=_ASYNC_UPLOAD):
"""Run the import, possibly asynchronously.
Returns the target datastore.
"""
# run_import can raise an exception which callers should handle
import_session = upload_session.import_session
import_session = gs_uploader.get_session(import_session.id)
task = import_session.tasks[0]
import_execution_requested = False
if import_session.state == 'INCOMPLETE':
if task.state != 'ERROR':
raise Exception('unknown item state: %s' % task.state)
elif import_session.state == 'PENDING' and task.target.store_type == 'coverageStore':
if task.state == 'READY':
import_session.commit(async)
import_execution_requested = True
if task.state == 'ERROR':
progress = task.get_progress()
raise Exception('error during import: %s' %
progress.get('message'))
# if a target datastore is configured, ensure the datastore exists
def finalize_incomplete_session_uploads(self, *args, **kwargs):
"""The task periodically checks for pending and stale Upload sessions.
It runs every 600 seconds (see the PeriodTask on geonode.upload._init_),
checks first for expired stale Upload sessions and schedule them for cleanup.
We have to make sure To NOT Delete those Unprocessed Ones,
which are in live sessions.
After removing the stale ones, it collects all the unprocessed and runs them
in parallel."""
lock_id = f'{self.request.id}'
with AcquireLock(lock_id) as lock:
if lock.acquire() is True:
_upload_ids = []
_upload_tasks = []
# Check first if we need to delete stale sessions
expiry_time = now() - timedelta(hours=UPLOAD_SESSION_EXPIRY_HOURS)
for _upload in Upload.objects.exclude(
state=enumerations.STATE_PROCESSED).exclude(
date__gt=expiry_time):
_upload.set_processing_state(enumerations.STATE_INVALID)
_upload_ids.append(_upload.id)
_upload_tasks.append(
_upload_session_cleanup.signature(args=(_upload.id, )))
upload_workflow_finalizer = _upload_workflow_finalizer.signature(
args=(
'_upload_session_cleanup',
_upload_ids,
),
immutable=True).on_error(
_upload_workflow_error.signature(args=(
'_upload_session_cleanup',
_upload_ids,
),
immutable=True))
upload_workflow = chord(_upload_tasks,
body=upload_workflow_finalizer)
upload_workflow.apply_async()
# Let's finish the valid ones
_processing_states = (enumerations.STATE_RUNNING,
enumerations.STATE_INVALID,
enumerations.STATE_PROCESSED)
for _upload in Upload.objects.exclude(
state__in=_processing_states):
session = None
try:
if not _upload.import_id:
raise NotFound
session = _upload.get_session.import_session
if not session or session.state != enumerations.STATE_COMPLETE:
session = gs_uploader.get_session(_upload.import_id)
except (NotFound, Exception) as e:
logger.exception(e)
session = None
if _upload.state not in (enumerations.STATE_COMPLETE,
enumerations.STATE_PROCESSED):
_upload.set_processing_state(
enumerations.STATE_INVALID)
if _upload.resource:
resource_manager.delete(_upload.resource.uuid)
if session:
_upload_ids.append(_upload.id)
_upload_tasks.append(
_update_upload_session_state.signature(
args=(_upload.id, )))
upload_workflow_finalizer = _upload_workflow_finalizer.signature(
args=(
'_update_upload_session_state',
_upload_ids,
),
immutable=True).on_error(
_upload_workflow_error.signature(args=(
'_update_upload_session_state',
_upload_ids,
),
immutable=True))
upload_workflow = chord(_upload_tasks,
body=upload_workflow_finalizer)
upload_workflow.apply_async()
