def _send_modified_event(self, **kw):
kwa = dict(kw)
kwa['new_subscriptions'] = list(self._added_subs)
kwa['removed_subscriptions'] = list(self._removed_subs)
self._added_subs = set()
self._removed_subs = set()
NewsBucket._send_modified_event(self, **kwa)
def save(self):
try:
NewsBucket.save(self)
except ResourceConflict:
raise
finally:
# just best effort here, we assume conflicts indicate better
# information arrived...
self._context.db.update(self._updated_news_items.values())
self._updated_news_items = {}
def test_composite_subs_by_title(ctx):
from melkman.db.bucket import NewsBucket
from melkman.db.composite import Composite, view_composite_subscriptions_by_title
from random import shuffle
cc = Composite.create(ctx)
buckets = []
for i in range(10):
bucket = NewsBucket.create(ctx)
bucket.title = 'bucket %d' % i
bucket.save()
buckets.append(bucket)
shuffled_buckets = list(buckets)
shuffle(shuffled_buckets)
for bucket in shuffled_buckets:
cc.subscribe(bucket)
cc.save()
# should come out in alphabetical order
for i, row in enumerate(view_composite_subscriptions_by_title(ctx.db, startkey=[cc.id, None], endkey=[cc.id, {}])):
assert row.value['bucket_id'] == buckets[i].id
assert i + 1 == len(buckets)
def get_initial_batch(id):
bucket = NewsBucket.get(id, ctx)
if bucket is None:
abort(404)
batch_args = _get_batch_args()
entries, next = _bucket_latest_entries_batch(bucket, **batch_args)
return Dibject(entries=entries, next=next)
def test_init_subscription(ctx):
from eventlet import sleep, spawn
from melkman.aggregator.worker import run_aggregator
from melkman.db.bucket import NewsBucket
from melkman.db.composite import Composite
agg = spawn(run_aggregator, ctx)
c = Composite.create(ctx)
c.save()
entries = []
bucket = NewsBucket.create(ctx)
for i in range(5):
eid = random_id()
entries.append(eid)
bucket.add_news_item(eid)
bucket.save()
sleep(.5)
c.subscribe(bucket)
c.save()
sleep(.5)
c.reload()
for eid in entries:
assert c.has_news_item(eid)
agg.kill()
agg.wait()
def get_rejected(self, db):
if self._rejected is not None and self._rejected.id == self.rejected_ref:
return self._rejected
elif self.rejected_ref:
self._rejected = NewsBucket.load(db, self.rejected_ref)
return self._rejected
else:
return None
def test_composites_by_sub(ctx):
from melkman.db.bucket import NewsBucket
from melkman.db.composite import Composite, view_composites_by_subscription
c1 = Composite.create(ctx)
c2 = Composite.create(ctx)
bucket1 = NewsBucket.create(ctx)
bucket1.save()
bucket2 = NewsBucket.create(ctx)
bucket2.save()
bucket3 = NewsBucket.create(ctx)
bucket3.save()
c1.subscribe(bucket1)
c1.save()
c2.subscribe(bucket1)
c2.subscribe(bucket2)
c2.save()
count = 0
seen = set()
for r in view_composites_by_subscription(ctx.db, include_docs=True, startkey=bucket1.id, endkey=bucket1.id):
comp = Composite.wrap(r.doc)
seen.add(comp.id)
count += 1
assert count == 2
assert c1.id in seen
assert c2.id in seen
count = 0
seen = set()
for r in view_composites_by_subscription(ctx.db, include_docs=True, startkey=bucket2.id, endkey=bucket2.id):
comp = Composite.from_doc(r.doc, ctx)
seen.add(comp.id)
count += 1
assert count == 1
assert c2.id in seen
for r in view_composites_by_subscription(ctx.db, include_docs=True, startkey=bucket3.id, endkey=bucket3.id):
assert False, 'unexpected subscription'
def get_feed_info(id):
bucket = NewsBucket.get(id, ctx)
if bucket is None:
abort(404)
feed = Dibject()
feed.id = id
feed.title = bucket.title
feed.timestamp = bucket.last_modification_date or datetime.utcnow()
try:
limit = int(request.params.get('limit', DEFAULT_FEED_SIZE))
except:
limit = DEFAULT_FEED_SIZE
limit = min(limit, MAX_FEED_SIZE)
feed.entries = _bucket_latest_entries_batch(bucket, limit=limit)
return feed
def get_batch(self, id):
bucket = NewsBucket.get(id, ctx)
if bucket is None:
abort(404)
batch_args = _get_batch_args()
entries, next = _bucket_latest_entries_batch(bucket, **batch_args)
batch = Dibject(next=next)
if request.is_xhr:
# this html section can be optionally omitted by specifying
# the query argument no_html=True
if not asbool(request.params.get('no_html', False)):
batch.html = render_entries_html(entries)
batch.entries = [i.item_id for i in entries]
return json_response(batch)
else:
batch.entries = entries
return self._show_batch(id, batch)
def reload(self):
NewsBucket.reload(self)
self._updated_news_items = {}
def __init__(self, *args, **kw):
NewsBucket.__init__(self, *args, **kw)
self._rejected = None
self._added_subs = set()
self._removed_subs = set()
def delete(self):
NewsBucket.delete(self)
if self._rejected is not None:
self._rejected.delete()
def test_modified_updates_composite(ctx):
from eventlet import sleep, spawn
from melkman.aggregator.worker import run_aggregator
from melkman.db.bucket import NewsBucket
from melkman.db.composite import Composite
agg = spawn(run_aggregator, ctx)
b = []
c = []
# make some buckets and composites.
for i in range(3):
bucket = NewsBucket.create(ctx)
bucket.save()
b.append(bucket)
comp = Composite.create(ctx)
comp.save()
c.append(comp)
# set up some subscriptions
c[0].subscribe(b[0])
c[1].subscribe(b[1])
c[2].subscribe(c[0])
c[2].subscribe(c[1])
for i in range(3):
c[i].save()
id1 = random_id()
b[0].add_news_item(id1)
log.debug("updating bucket 0 (%s) with item %s..." % (b[0].id, id1))
b[0].save()
sleep(1)
# refresh them from the db...
for i in range(3):
c[i].reload()
assert c[0].has_news_item(id1)
assert not c[1].has_news_item(id1)
assert c[2].has_news_item(id1)
id2 = random_id()
b[1].add_news_item(id2)
log.debug("updating bucket 1 (%s) with item %s..." % (b[1].id, id2))
b[1].save()
sleep(1)
# refresh them from the db...
for i in range(3):
c[i].reload()
assert not c[0].has_news_item(id2)
assert c[1].has_news_item(id2)
assert c[2].has_news_item(id2)
id3 = random_id()
b[2].add_news_item(id3)
log.debug("updating bucket 2 (%s) with item %s..." % (b[2].id, id3))
b[2].save()
sleep(1)
# refresh them from the db...
for i in range(3):
c[i].reload()
assert not c[0].has_news_item(id3)
assert not c[1].has_news_item(id3)
assert not c[2].has_news_item(id3)
agg.kill()
agg.wait()
def save(self):
NewsBucket.save(self)
if self._rejected is not None:
self._rejected.save(self)
def __init__(self, *args, **kw):
NewsBucket.__init__(self, *args, **kw)
self._updated_news_items = {}
def _handle_new_subscriptions(message_data, message, context):
"""
helper handler called when new subscriptions are added to
a composite.
"""
try:
new_subscriptions = message_data.get('new_subscriptions', [])
if len(new_subscriptions) == 0:
log.warn("Ignoring init_subscription with no new subscriptions...")
return
cid = message_data.get('bucket_id', None)
if cid is None:
log.error("Ignoring init_subscription with no bucket_id: %s" % message_data)
return
composite = Composite.get(cid, context)
if composite is None or not 'Composite' in composite.document_types:
log.error("Ignoring subscription update for non-existent composite %s" % cid)
return
new_feeds = []
updates = 0
for sub in new_subscriptions:
if not sub in composite.subscriptions:
log.warn("ignoring subscription %s -> %s, not in composite" % (sub, cid))
continue
bucket = NewsBucket.get(sub, context)
if bucket is None:
log.warn("Ignoring init subscription to unknown object (%s)" % composite.subscriptions[sub])
continue
# try 'casting' to a RemoteFeed
if 'RemoteFeed' in bucket.document_types:
rf = RemoteFeed.from_doc(bucket.unwrap(), context)
# mark as needing immediate fetch if
# there is no history for this feed.
if len(rf.update_history) == 0:
new_feeds.append(rf.url)
continue
try:
log.debug("init subscription %s -> %s" % (sub, cid))
updates += composite.init_subscription(sub)
sleep(0) # yield control
except:
log.error("Error initializing subscription %s -> %s: %s" % (sub, cid, traceback.format_exc()))
if updates > 0:
try:
composite.save()
except ResourceConflict:
# not a big deal in this case. This basically means
# our timestamp did not become the latest -- we
# have made no alterations other than adding items.
# Our additions succeed/fail independently of this as they
# are separate documents.
pass
# request that we start indexing anything new...
for url in new_feeds:
request_feed_index(url, context)
except:
log.error("Error handling init_subscrition %s: %s" % (message_data, traceback.format_exc()))
raise
def reload(self):
NewsBucket.reload(self)
self._rejected = None
self._added_subs = set()
self._removed_subs = set()