def get(self):
queue = getQueue()
stat = getStat()
now = datetime.now()
while len(queue) > 0 and now - queue[0][1] > timedelta(hours=1):
if queue[0][0] in stat:
stat[queue[0][0]] = stat[queue[0][0]] - 1
if stat[queue[0][0]] == 0:
del (stat[queue[0][0]])
else:
logging.error(
"Fatal error with trending stream data. Data in queue but does not in stat."
)
queue.popleft()
trending = [(x.key.id(), stat[x.key.id()] if x.key.id() in stat else 0)
for x in Stream.query(
ancestor=getStreamKey()).order(-Stream.viewCount)]
trending.sort(key=lambda x: -x[1])
updateQueue(queue)
updateStat(stat)
updateTrendingStreams(trending)
n = 0
trendingStreams = []
for stat in trending:
stream = Stream.get_by_id(stat[0], getStreamKey())
if stream:
trendingStreams.append((stream, stat[1]))
n = n + 1
if n == Trending_Stream_Handler.digest_num:
break
for frequency in Trending_Stream_Handler.email_digest_counter.keys():
Trending_Stream_Handler.email_digest_counter[
frequency] = Trending_Stream_Handler.email_digest_counter[
frequency] + 1
if Trending_Stream_Handler.email_digest_counter[
frequency] == Trending_Stream_Handler.email_digest_bound[
frequency]:
Trending_Stream_Handler.email_digest_counter[frequency] = 0
for userSetting in UserEmailSetting.query(
UserEmailSetting.email_update_rate == frequency):
emails.sendTrendingStreamEmails(userSetting.user,
userSetting.name,
trendingStreams)
self.response.write(str(trending))
def getTrendingStreams():
trending = memcache.get('TrendingStat')
if trending:
return trending
trending = [
(x.key.id(), 0)
for x in Stream.query(ancestor=getStreamKey()).order(Stream.viewCount)
]
updateTrendingStreams(trending)
return trending
def getTopStreams(max_num):
trending = getTrendingStreams()
n = 0
trendingStreams = []
for stat in trending:
stream = Stream.get_by_id(stat[0], getStreamKey())
if stream:
trendingStreams.append((stream, stat[1]))
n = n + 1
if n == max_num:
break
return trendingStreams
def getTrie():
trie = memcache.get('search')
if trie:
return trie
trie = Trie()
for stream in Stream.query(ancestor=getStreamKey()):
sid = stream.key.id()
for str in formatSearchContent(stream.name):
trie.add(str, sid)
for str in formatSearchContent(stream.tag):
trie.add(str, sid)
for str in formatSearchContent("".join(
re.findall(r'(.+)@', stream.user))):
trie.add(str, sid)
updateTrie(trie)
return trie
def search_streams(query_words):
query_words = formatSearchContent(query_words)
sids = []
relevence = {}
trie = getTrie()
for word in query_words:
sids_of_word = set()
for string in trie.searchSubstring(word):
sids_of_string = trie.get(string)
for sid in sids_of_string:
if not (sid in relevence):
relevence[sid] = 0.0
relevence[sid] = relevence[sid] + float(
len(word)) / len(string)
sids_of_word = sids_of_word.union(sids_of_string)
sids.append(sids_of_word)
ret = [
Stream.get_by_id(sid, getStreamKey())
for sid in reduce(lambda x, y: x.intersection(y), sids)
]
return sorted(ret, key=lambda x: -relevence[x.key.id()])
