def _generate_incremental_export(incremental_export):
export_instance = incremental_export.export_instance
export_instance.export_format = Format.UNZIPPED_CSV # force to unzipped CSV
checkpoint = incremental_export.last_valid_checkpoint
# Remove the date period from the ExportInstance, since this is added automatically by Daily Saved exports
export_instance.filters.date_period = None
filters = export_instance.get_filters()
if checkpoint:
filters.append(
ServerModifiedOnRangeFilter(gt=checkpoint.last_doc_date))
class LastDocTracker:
def __init__(self, doc_iterator):
self.doc_iterator = doc_iterator
self.last_doc = None
self.doc_count = 0
def __iter__(self):
for doc in self.doc_iterator:
self.last_doc = doc
self.doc_count += 1
yield doc
with TransientTempfile() as temp_path, metrics_track_errors(
'generate_incremental_exports'):
writer = get_export_writer([export_instance],
temp_path,
allow_pagination=False)
with writer.open([export_instance]):
query = _get_export_query(export_instance, filters)
query = query.sort('server_modified_on'
) # reset sort to this instead of opened_on
docs = LastDocTracker(query.run().hits)
write_export_instance(writer, export_instance, docs)
export_file = ExportFile(writer.path, writer.format)
if docs.doc_count <= 0:
return
new_checkpoint = incremental_export.checkpoint(
docs.doc_count, docs.last_doc.get('server_modified_on'))
with export_file as file_:
db = get_blob_db()
db.put(file_,
domain=incremental_export.domain,
parent_id=new_checkpoint.blob_parent_id,
type_code=CODES.data_export,
key=str(new_checkpoint.blob_key),
timeout=24 * 60)
return new_checkpoint
def process_bounced_emails():
if settings.RETURN_PATH_EMAIL and settings.RETURN_PATH_EMAIL_PASSWORD:
try:
with BouncedEmailManager(
delete_processed_messages=True
) as bounced_manager, metrics_track_errors(
'process_bounced_emails_task'):
bounced_manager.process_daemon_messages()
except Exception as e:
notify_exception(
None,
message="Encountered error while processing bounced emails",
details={
'error': e,
})
def populate_export_download_task(domain,
export_ids,
exports_type,
username,
es_filters,
download_id,
owner_id,
filename=None,
expiry=10 * 60):
"""
:param expiry: Time period for the export to be available for download in minutes
"""
email_requests = EmailExportWhenDoneRequest.objects.filter(
domain=domain, download_id=download_id)
if settings.STALE_EXPORT_THRESHOLD is not None and not email_requests.count(
):
delay = get_task_time_to_start(
populate_export_download_task.request.id)
if delay.total_seconds() > settings.STALE_EXPORT_THRESHOLD:
metrics_counter('commcare.exports.rejected_unfresh_export')
raise RejectedStaleExport()
export_instances = [
get_export(exports_type, domain, export_id, username)
for export_id in export_ids
]
with TransientTempfile() as temp_path, metrics_track_errors(
'populate_export_download_task'):
export_file = get_export_file(
export_instances,
es_filters,
temp_path,
# We don't have a great way to calculate progress if it's a bulk download,
# so only track the progress for single instance exports.
progress_tracker=populate_export_download_task
if len(export_instances) == 1 else None)
file_format = Format.from_format(export_file.format)
filename = filename or export_instances[0].name
with export_file as file_:
db = get_blob_db()
db.put(
file_,
domain=domain,
parent_id=domain,
type_code=CODES.data_export,
key=download_id,
timeout=expiry,
)
expose_blob_download(
download_id,
expiry=expiry * 60,
mimetype=file_format.mimetype,
content_disposition=safe_filename_header(
filename, file_format.extension),
download_id=download_id,
owner_ids=[owner_id],
)
for email_request in email_requests:
try:
couch_user = CouchUser.get_by_user_id(email_request.user_id,
domain=domain)
except CouchUser.AccountTypeError:
pass
else:
if couch_user is not None:
process_email_request(domain, download_id,
couch_user.get_email())
email_requests.delete()
