def purge_orphaned_gridfs_files(self, cutoff_hours=24):
cutoff = dateutil.utcnow() - datetime.timedelta(hours=cutoff_hours)
gridfs = self.get_gridfs()
old_ids = [item['_id'] for item in gridfs._GridFS__files.find(
{'parents':[], 'lastmodDate':{'$lt':cutoff}}, fields=[])]
for id in old_ids:
gridfs.delete(id)
return len(old_ids)
def test_dateutil(self):
import datetime, pytz
from audrey import dateutil
utc_dt_aware = datetime.datetime(2012, 7, 4, 16, 20, tzinfo=pytz.utc)
utc_dt_naive = datetime.datetime(2012, 7, 4, 16, 20)
est_dt_aware = datetime.datetime(2012, 7, 4, 11, 20, tzinfo=pytz.timezone('US/Eastern'))
est_dt_naive = datetime.datetime(2012, 7, 4, 11, 20)
self.assertEqual(dateutil.convert_aware_datetime(utc_dt_aware, 'US/Eastern'), est_dt_aware)
self.assertEqual(dateutil.convert_aware_datetime(est_dt_aware, 'utc'), utc_dt_aware)
self.assertEqual(dateutil.convert_naive_datetime(utc_dt_naive, 'utc', 'US/Eastern'), est_dt_aware)
self.assertEqual(dateutil.convert_naive_datetime(est_dt_naive, 'US/Eastern', 'utc'), utc_dt_aware)
self.assertEqual(dateutil.make_naive(est_dt_aware), est_dt_naive)
now_zero = dateutil.utcnow(zero_seconds=True)
now_later = dateutil.utcnow(zero_seconds=False)
self.assertEqual(now_zero.second, 0)
self.assertTrue(now_later > now_zero)
self.assertEqual(now_zero.tzinfo, pytz.utc)
self.assertEqual(now_later.tzinfo, pytz.utc)
def save(self, validate_schema=True, index=True, set_modified=True, set_etag=True):
"""
Save this object in MongoDB (and optionally ElasticSearch).
:param validate_schema: Should the colander schema be validated first? If ``True``, may raise :class:`colander.Invalid`.
:type validate_schema: boolean
:param index: Should the object be (re-)indexed in ElasticSearch?
:type index: boolean
:param set_modified: Should the object's last modified timestamp (``_modified``) be updated?
:type set_modified: boolean
:param set_etag: Should the object's Etag be updated?
:type set_etag: boolean
"""
if validate_schema:
self.validate_schema() # May raise a colander.Invalid exception
if set_modified:
self._modified = dateutil.utcnow()
if not getattr(self, '_created', None): self._created = self._modified
if set_etag:
self._etag = self.generate_etag()
# Determine all the GridFS file ids that this object
# now refers to and used to refer to.
root = find_root(self)
fs_files_coll = root.get_gridfs()._GridFS__files
new_file_ids = set([x._id for x in self.get_all_files()])
old_file_ids = set()
if self._id:
dbref = self.get_dbref()
for item in fs_files_coll.find({'parents':dbref}, fields=[]):
old_file_ids.add(item['_id'])
# Persist the object in Mongo.
doc = self.get_mongo_save_doc()
id = self.get_mongo_collection().save(doc, safe=True)
if not self._id:
self._id = id
dbref = self.get_dbref()
# Update GridFS file "parents".
ids_to_remove = old_file_ids - new_file_ids
ids_to_add = new_file_ids - old_file_ids
if ids_to_remove:
fs_files_coll.update({'_id':{'$in':list(ids_to_remove)}}, {"$pull":{"parents":dbref}, "$set":{"lastmodDate": dateutil.utcnow()}}, multi=True)
if ids_to_add:
fs_files_coll.update({'_id':{'$in':list(ids_to_add)}}, {"$addToSet":{"parents":dbref}, "$set":{"lastmodDate": dateutil.utcnow()}}, multi=True)
if index: self.index()
def create_gridfs_file(self, file, filename, mimetype, parents=None):
""" Create a new GridFS file.
:param file: file content/data
:type file: a file-like object (providing a read() method) or a string
:param filename: a filename
:type filename: string
:param mimetype: a mime-type
:type mimetype: string
:param parents: list of references to the Objects that "own" the file
:type parents: list of :class:`bson.dbref.DBRef` instances, or ``None``
:rtype: :class:`audrey.resources.file.File`
"""
if parents is None: parents = []
# FIXME: if image, get dimensions (via PIL?) and store as two custom attributes (width and height)
return File(self.get_gridfs().put(file, filename=filename, contentType=mimetype, parents=parents, lastmodDate=dateutil.utcnow()))
def _pre_delete(self):
# Remove from ElasticSearch
self.unindex()
# Update parents attribute of related GridFS files
dbref = self.get_dbref()
root = find_root(self)
fs_files_coll = root.get_gridfs()._GridFS__files
fs_files_coll.update({'parents':dbref}, {"$pull":{"parents":dbref}, "$set":{"lastmodDate": dateutil.utcnow()}}, multi=True)