def test_create(self):
accessor_save_kw = self.__keywords()
source_name = accessor_save_kw['source']
content_type = accessor_save_kw['content_type']
body = accessor_save_kw['body']
self.moxer.StubOutWithMock(ArtifactInfo, "all", use_mock_anything=True)
self.moxer.StubOutWithMock(ArtifactSource, "get_or_create", use_mock_anything=True)
self.moxer.StubOutWithMock(Counters, "source_counter", use_mock_anything=True)
self.moxer.StubOutWithMock(ArtifactInfo, "create", use_mock_anything=True)
self.moxer.StubOutWithMock(ArtifactContent, "create", use_mock_anything=True)
source = MockEntity(key_name=source_name)
ArtifactInfo.all(keys_only=True).AndReturn(MockQuery(None, keys_only=True))
ArtifactSource.get_or_create(source_name).AndReturn(source)
counter = self.moxer.CreateMockAnything()
Counters.source_counter(source_name).AndReturn(counter)
counter.increment()
# TODO: I wish I could ignore keywords
md5 = ArtifactAccessor._content_md5(source_name, content_type, body)
info_save_kw = dict(source=source, source_name=source_name, content_type=content_type, content_md5=md5)
info_key = MockKey(name=self.test_id)
ArtifactInfo.create(**info_save_kw).AndReturn(info_key)
content_save_kw = dict(source=source, source_name=source_name, info=info_key, body=body)
ArtifactContent.create(info_key.name(), **content_save_kw).AndReturn(MockKey(name=self.test_id))
self.moxer.ReplayAll()
info, content, source = ArtifactAccessor.create(**accessor_save_kw)
print 'info:%s, content:%s, source:%s' % (info, content, source)
self.moxer.VerifyAll()
def _test_delete(self):
self.moxer.StubOutWithMock(ArtifactInfo, "get_by_guid", use_mock_anything=True)
self.moxer.StubOutWithMock(ArtifactContent, "get_by_guid", use_mock_anything=True)
guid = 'blah'
ArtifactInfo.get_by_guid(guid).AndReturn(MockEntity(MockKey(name=guid)))
ArtifactContent.get_by_guid(guid).AndReturn(MockEntity(MockKey(name=guid)))
self.moxer.ReplayAll()
ArtifactAccessor.delete(guid)
self.moxer.VerifyAll()
def test_create_duplicate(self):
self.moxer.StubOutWithMock(ArtifactInfo, "all", use_mock_anything=True)
ArtifactInfo.all(keys_only=True).AndReturn(MockQuery(range(1), keys_only=True))
self.moxer.ReplayAll()
try:
ArtifactAccessor.create(**self.__keywords())
self.fail("exception expected")
except DuplicateDataException, ex:
pass
def test_create_duplicate(self):
self.moxer.StubOutWithMock(ArtifactInfo, "all", use_mock_anything=True)
ArtifactInfo.all(keys_only=True).AndReturn(
MockQuery(range(1), keys_only=True))
self.moxer.ReplayAll()
try:
ArtifactAccessor.create(**self.__keywords())
self.fail("exception expected")
except DuplicateDataException, ex:
pass
def test_delete_nonexistent(self):
self.moxer.StubOutWithMock(ArtifactInfo, "get_by_guid", use_mock_anything=True)
self.moxer.StubOutWithMock(ArtifactContent, "get_by_guid", use_mock_anything=True)
guid = 'blah'
ArtifactInfo.get_by_guid(guid)
ArtifactContent.get_by_guid(guid)
self.moxer.ReplayAll()
try:
ArtifactAccessor.delete(guid)
self.fail("exception expected")
except NotFoundException, ex:
pass
def test_create(self):
accessor_save_kw = self.__keywords()
source_name = accessor_save_kw['source']
content_type = accessor_save_kw['content_type']
body = accessor_save_kw['body']
self.moxer.StubOutWithMock(ArtifactInfo, "all", use_mock_anything=True)
self.moxer.StubOutWithMock(ArtifactSource,
"get_or_create",
use_mock_anything=True)
self.moxer.StubOutWithMock(Counters,
"source_counter",
use_mock_anything=True)
self.moxer.StubOutWithMock(ArtifactInfo,
"create",
use_mock_anything=True)
self.moxer.StubOutWithMock(ArtifactContent,
"create",
use_mock_anything=True)
source = MockEntity(key_name=source_name)
ArtifactInfo.all(keys_only=True).AndReturn(
MockQuery(None, keys_only=True))
ArtifactSource.get_or_create(source_name).AndReturn(source)
counter = self.moxer.CreateMockAnything()
Counters.source_counter(source_name).AndReturn(counter)
counter.increment()
# TODO: I wish I could ignore keywords
md5 = ArtifactAccessor._content_md5(source_name, content_type, body)
info_save_kw = dict(source=source,
source_name=source_name,
content_type=content_type,
content_md5=md5)
info_key = MockKey(name=self.test_id)
ArtifactInfo.create(**info_save_kw).AndReturn(info_key)
content_save_kw = dict(source=source,
source_name=source_name,
info=info_key,
body=body)
ArtifactContent.create(info_key.name(), **content_save_kw).AndReturn(
MockKey(name=self.test_id))
self.moxer.ReplayAll()
info, content, source = ArtifactAccessor.create(**accessor_save_kw)
print 'info:%s, content:%s, source:%s' % (info, content, source)
self.moxer.VerifyAll()
def put(cls, rhandler, guid, **kw):
helper = RequestHelper(rhandler)
artifact = ArtifactInfo.get_by_guid(guid)
if not artifact:
helper.error(404)
return
# removes existing properties
props = ArtifactInfo.properties().keys()
for prop in props:
delattr(artifact, prop)
# save artifact
ArtifactInfo.save(artifact)
def delete_by_name(cls, source_name):
source = ArtifactSource.get_by_name(source_name)
logging.debug("delete_by_name source: %s" % source)
if not source:
raise NotFoundException("ArtifactSource %s" % source_name)
# checks for feeds linked to source
feed = FeedAccessor.get_by_source_name(source_name, return_none=True)
if feed:
raise ConflictingDataException("ArtifactSource '%s' is referenced by Feed '%s'" % (source_name, feed.url))
# finds and deletes artifacts for source
info_keys = ArtifactInfo.find_by_source(source, keys_only=True)
content_keys = ArtifactContent.find_by_source(source)
# zips keys to delete info/content pairs back-to-back
for artifact_keys in zip(info_keys, content_keys):
db.delete(artifact_keys)
# deletes extras if info/content sizes don't match
# (this would be a data bug somewhere)
content_len = content_keys.count()
info_len = info_keys.count()
if content_len < info_len:
db.delete(content_keys[info_len:])
elif info_len > content_len:
db.delete(info_keys[content_len:])
# deletes source
db.delete(source)
def create(cls, **kw):
"""
keywords:
source
content_type
body
returns:
tuple: (ArtifactInfo key, ArtifactContent key, ArtifactSource key)
raises:
DuplicateDataException - if artifact already exists
"""
if not kw:
raise IllegalArgumentException("keywords must be provided")
source_name = kw.pop("source", None)
content_type = kw.get("content_type")
if not source_name:
raise IllegalArgumentException("source keyword must be provided.")
elif not content_type:
raise IllegalArgumentException("content_type keyword must be provided.")
# I pop "body" since I can't include it as a keyword for ArtifactInfo.create()
body = kw.pop("body", None)
# hashes content to avoid saving a duplicate
content_md5 = cls._content_md5(source_name, content_type, body)
found_artifact_key = ArtifactInfo.find_by_content_md5(content_md5, keys_only=True).get()
if found_artifact_key:
raise DuplicateDataException("artifact %s" % (found_artifact_key.name()))
return cls._create(source_name, body, content_md5, **kw)
def post(self, **kw):
helper = RequestHelper(self)
results = {}
source_q = ArtifactSource.all()
for s in source_q:
artifact_q = ArtifactInfo.find_by_source(s)
count = len([a for a in artifact_q])
counter = Counters.source_counter(s.name)
old_count = counter.count()
counter.set(count)
source_result = { 'old': old_count }
# if source is linked to a feed, I can't delete it
feed = Feed.get_by_source(s, return_none=True)
if feed:
source_result['feed'] = feed.url
if not count and not feed:
s.delete()
source_result['deleted'] = True
if count:
source_result['new'] = count
results[s.name] = source_result
helper.write_json(results)
def post(self, **kw):
helper = RequestHelper(self)
results = {}
source_q = ArtifactSource.all()
for s in source_q:
artifact_q = ArtifactInfo.find_by_source(s)
count = len([a for a in artifact_q])
counter = Counters.source_counter(s.name)
old_count = counter.count()
counter.set(count)
source_result = {'old': old_count}
# if source is linked to a feed, I can't delete it
feed = Feed.get_by_source(s, return_none=True)
if feed:
source_result['feed'] = feed.url
if not count and not feed:
s.delete()
source_result['deleted'] = True
if count:
source_result['new'] = count
results[s.name] = source_result
helper.write_json(results)
def find_or_create(cls, **kw):
"""
returns:
tuple: (ArtifactInfo key, ArtifactContent key, ArtifactSource key, created)
"""
if not kw:
raise IllegalArgumentException("keywords must be provided")
source_name = kw.pop("source", None)
content_type = kw.get("content_type")
if not source_name:
raise IllegalArgumentException("source keyword must be provided.")
elif not content_type:
raise IllegalArgumentException(
"content_type keyword must be provided.")
# I pop "body" since I can't include it as a keyword for ArtifactInfo.create()
body = kw.pop("body", None)
# hashes content to avoid saving a duplicate
content_md5 = cls._content_md5(source_name, content_type, body)
found_artifact = ArtifactInfo.find_by_content_md5(content_md5).get()
if found_artifact:
info_key = found_artifact.key()
content_key = ArtifactContent.get_by_guid(
found_artifact.guid).key()
source_key = found_artifact.source.key()
created = False
else:
info_key, content_key, source_key = cls._create(
source_name, body, content_md5, **kw)
created = True
return (info_key, content_key, source_key, created)
def delete_by_name(cls, source_name):
source = ArtifactSource.get_by_name(source_name)
logging.debug("delete_by_name source: %s" % source)
if not source:
raise NotFoundException('ArtifactSource %s' % source_name)
# checks for feeds linked to source
feed = FeedAccessor.get_by_source_name(source_name, return_none=True)
if feed:
raise ConflictingDataException(
"ArtifactSource '%s' is referenced by Feed '%s'" %
(source_name, feed.url))
# finds and deletes artifacts for source
info_keys = ArtifactInfo.find_by_source(source, keys_only=True)
content_keys = ArtifactContent.find_by_source(source)
# zips keys to delete info/content pairs back-to-back
for artifact_keys in zip(info_keys, content_keys):
db.delete(artifact_keys)
# deletes extras if info/content sizes don't match
# (this would be a data bug somewhere)
content_len = content_keys.count()
info_len = info_keys.count()
if content_len < info_len:
db.delete(content_keys[info_len:])
elif info_len > content_len:
db.delete(info_keys[content_len:])
# deletes source
db.delete(source)
def find_or_create(cls, **kw):
"""
returns:
tuple: (ArtifactInfo key, ArtifactContent key, ArtifactSource key, created)
"""
if not kw:
raise IllegalArgumentException("keywords must be provided")
source_name = kw.pop("source", None)
content_type = kw.get("content_type")
if not source_name:
raise IllegalArgumentException("source keyword must be provided.")
elif not content_type:
raise IllegalArgumentException("content_type keyword must be provided.")
# I pop "body" since I can't include it as a keyword for ArtifactInfo.create()
body = kw.pop("body", None)
# hashes content to avoid saving a duplicate
content_md5 = cls._content_md5(source_name, content_type, body)
found_artifact = ArtifactInfo.find_by_content_md5(content_md5).get()
if found_artifact:
info_key = found_artifact.key()
content_key = ArtifactContent.get_by_guid(found_artifact.guid).key()
source_key = found_artifact.source.key()
created = False
else:
info_key, content_key, source_key = cls._create(source_name, body, content_md5, **kw)
created = True
return (info_key, content_key, source_key, created)
def get(self, **kw):
helper = RequestHelper(self)
q = self.request.get("q", None)
output = self.request.get("o", None)
max_results = int(self.request.get("max", -1))
if not q:
helper.error(400, "q not provided.")
return
q_results = ArtifactContent.all().search(q)
json_results = None
if output == "short":
json_results = {}
json_results["count"] = q_results.count()
elif output == "id":
json_results = {}
count = q_results.count()
if max_results > 0 and max_results < q_results.count():
count = max_results
json_results["count"] = count
ids = []
json_results["ids"] = ids
results = q_results.fetch(1000) if max_results == -1 else q_results.fetch(max_results)
for c in results:
ids.append(c.guid)
else:
json_results = []
if q_results.count():
for content in q_results.fetch(10):
info = ArtifactInfo.get_by_guid(content.guid)
json_results.append(ArtifactsHelper.artifact_to_hash(info, content))
helper.write_json(json_results)
def _test_delete(self):
self.moxer.StubOutWithMock(ArtifactInfo,
"get_by_guid",
use_mock_anything=True)
self.moxer.StubOutWithMock(ArtifactContent,
"get_by_guid",
use_mock_anything=True)
guid = 'blah'
ArtifactInfo.get_by_guid(guid).AndReturn(MockEntity(
MockKey(name=guid)))
ArtifactContent.get_by_guid(guid).AndReturn(
MockEntity(MockKey(name=guid)))
self.moxer.ReplayAll()
ArtifactAccessor.delete(guid)
self.moxer.VerifyAll()
def test_delete_nonexistent(self):
self.moxer.StubOutWithMock(ArtifactInfo,
"get_by_guid",
use_mock_anything=True)
self.moxer.StubOutWithMock(ArtifactContent,
"get_by_guid",
use_mock_anything=True)
guid = 'blah'
ArtifactInfo.get_by_guid(guid)
ArtifactContent.get_by_guid(guid)
self.moxer.ReplayAll()
try:
ArtifactAccessor.delete(guid)
self.fail("exception expected")
except NotFoundException, ex:
pass
def get(cls, rhandler, guid, **kw):
helper = RequestHelper(rhandler)
artifact_info = ArtifactInfo.get_by_guid(guid)
artifact_content = ArtifactContent.get_by_guid(guid)
if artifact_info and artifact_content:
artifact_hash = ArtifactsHelper.artifact_to_hash(artifact_info, artifact_content)
helper.write_json(artifact_hash)
else:
helper.error(404)
def test_delete_by_name_deletes_source_with_no_referencing_feed(self):
self.m.StubOutWithMock(ArtifactSource, "get_by_name")
self.m.StubOutWithMock(FeedAccessor, "get_by_source_name")
self.m.StubOutWithMock(ArtifactInfo, "find_by_source")
self.m.StubOutWithMock(ArtifactContent, "find_by_source")
self.m.StubOutWithMock(db, "delete")
self.m.StubOutWithMock(memcache, "delete")
name = "mhawthorne"
source = MockEntity(key_name=name)
ArtifactSource.get_by_name(name).AndReturn(source)
FeedAccessor.get_by_source_name(name, return_none=True)
ArtifactInfo.find_by_source(source, keys_only=True).AndReturn(MockQuery(range(0,0)))
ArtifactContent.find_by_source(source).AndReturn(MockQuery(range(0,0)))
db.delete(source)
memcache.delete(IsA(str)).AndReturn(1)
self.m.ReplayAll()
ArtifactSourceAccessor.delete_by_name(name)
self.m.VerifyAll()
def get(self, **kw):
helper = RequestHelper(self)
start = int(self.request.get("start", 0))
count = int(self.request.get("count", 10))
q = ArtifactInfo.all().order("-modified")
json_results = []
if q.count():
for a_info in q.fetch(count, start):
a_content = ArtifactAccessor.get_content_by_guid(a_info.guid)
json_results.append(ArtifactsHelper.artifact_to_hash(a_info, a_content))
helper.write_json(json_results)
def test_delete_by_name_deletes_source_with_no_referencing_feed(self):
self.m.StubOutWithMock(ArtifactSource, "get_by_name")
self.m.StubOutWithMock(FeedAccessor, "get_by_source_name")
self.m.StubOutWithMock(ArtifactInfo, "find_by_source")
self.m.StubOutWithMock(ArtifactContent, "find_by_source")
self.m.StubOutWithMock(db, "delete")
self.m.StubOutWithMock(memcache, "delete")
name = "mhawthorne"
source = MockEntity(key_name=name)
ArtifactSource.get_by_name(name).AndReturn(source)
FeedAccessor.get_by_source_name(name, return_none=True)
ArtifactInfo.find_by_source(source, keys_only=True).AndReturn(
MockQuery(range(0, 0)))
ArtifactContent.find_by_source(source).AndReturn(MockQuery(range(0,
0)))
db.delete(source)
memcache.delete(IsA(str)).AndReturn(1)
self.m.ReplayAll()
ArtifactSourceAccessor.delete_by_name(name)
self.m.VerifyAll()
def _create(cls, source_name, body, content_md5, **kw):
# saves source, if unique
source_key = ArtifactSource.get_or_create(source_name)
# saves ArtifactInfo
a_info_key = ArtifactInfo.create(content_md5=content_md5, source=source_key, source_name=source_name, **kw)
# saves ArtifactContent
guid = a_info_key.name()
a_content_key = ArtifactContent.create(
guid, body=body, source=source_key, source_name=source_name, info=a_info_key
)
# bump source counter
# it's important to do this AFTER the artifacts are saved
Counters.source_counter(source_name).increment()
return a_info_key, a_content_key, source_key
def delete(cls, guid):
logger = LoggerFactory.logger(cls.__name__)
a_info = ArtifactInfo.get_by_guid(guid)
a_content_key = ArtifactContent.get_by_guid(guid)
if not (a_info or a_content_key):
# neither record found
raise NotFoundException("artifact %s" % guid)
elif not (a_info and a_content_key):
# one record found; one missing
logger.warn("artifact %s; missing data; info=%s; content=%s" %
(guid, a_info.key().name(), a_content_key))
# I delete what I can
keys = []
if a_info: keys.append(a_info)
if a_content_key: keys.append(a_content_key)
db.delete(keys)
# decrease source counter
Counters.source_counter(a_info.source.name).decrement()
class ArtifactBulkDeleteHandler(webapp.RequestHandler):
""" I wanted this to be in the DELETE method for ArtifactsHandler, but a client limitation currently prevents it """
def post(self):
helper = RequestHelper(self)
json_body = self.request.body
if not json_body:
helper.error(400, "body required")
return
decoded_body = urllib.unquote(json_body)
try:
body_hash = json.loads(decoded_body)
except json.JSONDecodeError, e:
msg = "malformed json: %s" % decoded_body
helper.error(400, msg)
logging.info(msg)
return
ids = body_hash.get("ids", None)
if not ids:
msg = "no 'ids' field provided in JSON"
helper.error(400, msg)
logging.info(msg)
return
logging.info("deleting %s artifact(s)" % len(ids))
contents = [c for c in ArtifactContent.get_by_key_name(ids) if c is not None]
logging.info("deleting %s ArtifactContent instances" % len(contents))
db.delete(contents)
infos = [i for i in ArtifactInfo.get_by_key_name(ids) if i is not None]
logging.info("deleting %s ArtifactInfo instances" % len(infos))
db.delete(infos)
helper.set_status(204)
def delete(cls, guid):
logger = LoggerFactory.logger(cls.__name__)
a_info = ArtifactInfo.get_by_guid(guid)
a_content_key = ArtifactContent.get_by_guid(guid)
if not (a_info or a_content_key):
# neither record found
raise NotFoundException("artifact %s" % guid)
elif not (a_info and a_content_key):
# one record found; one missing
logger.warn("artifact %s; missing data; info=%s; content=%s" % (guid, a_info.key().name(), a_content_key))
# I delete what I can
keys = []
if a_info:
keys.append(a_info)
if a_content_key:
keys.append(a_content_key)
db.delete(keys)
# decrease source counter
Counters.source_counter(a_info.source.name).decrement()
def _create(cls, source_name, body, content_md5, **kw):
# saves source, if unique
source_key = ArtifactSource.get_or_create(source_name)
# saves ArtifactInfo
a_info_key = ArtifactInfo.create(content_md5=content_md5,
source=source_key,
source_name=source_name,
**kw)
# saves ArtifactContent
guid = a_info_key.name()
a_content_key = ArtifactContent.create(guid,
body=body,
source=source_key,
source_name=source_name,
info=a_info_key)
# bump source counter
# it's important to do this AFTER the artifacts are saved
Counters.source_counter(source_name).increment()
return a_info_key, a_content_key, source_key
def create(cls, **kw):
"""
keywords:
source
content_type
body
returns:
tuple: (ArtifactInfo key, ArtifactContent key, ArtifactSource key)
raises:
DuplicateDataException - if artifact already exists
"""
if not kw:
raise IllegalArgumentException("keywords must be provided")
source_name = kw.pop("source", None)
content_type = kw.get("content_type")
if not source_name:
raise IllegalArgumentException("source keyword must be provided.")
elif not content_type:
raise IllegalArgumentException(
"content_type keyword must be provided.")
# I pop "body" since I can't include it as a keyword for ArtifactInfo.create()
body = kw.pop("body", None)
# hashes content to avoid saving a duplicate
content_md5 = cls._content_md5(source_name, content_type, body)
found_artifact_key = ArtifactInfo.find_by_content_md5(
content_md5, keys_only=True).get()
if found_artifact_key:
raise DuplicateDataException("artifact %s" %
(found_artifact_key.name()))
return cls._create(source_name, body, content_md5, **kw)
def count_infos(cls, source):
return ArtifactInfo.find_by_source(source, keys_only=True).count()
def find_newer(cls, timestamp, **kw):
return ArtifactInfo.find_newer(timestamp, **kw)
def find_artifact_counts_newer(cls, datetime, **kw):
counts = KeyCounter()
for art in ArtifactInfo.find_newer(datetime, **kw):
counts.increment(art.source_name)
return counts.to_hash()
def count_infos(cls, source):
return ArtifactInfo.find_by_source(source, keys_only=True).count()
def find_newer(cls, timestamp, **kw):
return ArtifactInfo.find_newer(timestamp, **kw)
def find_artifact_counts_newer(cls, datetime, **kw):
counts = KeyCounter()
for art in ArtifactInfo.find_newer(datetime, **kw):
counts.increment(art.source_name)
return counts.to_hash()