def getWholePost(db, postId):
""" get a (title, post and all its answers) """
title = None
wholePost = None
for question in util.iterateQuestions(db, postList=[postId]):
answers = "\n\n".join([answer.body for answer in util.iterateAnswers(db, postId)])
title = question.title
wholePost = "\n\n".join([question.title, question.body, answers, question.tags])
return (title, wholePost)
def getWholePost(db, postId):
""" get a (title, post and all its answers) """
title = None
wholePost = None
for question in util.iterateQuestions(db, postList=[postId]):
answers = "\n\n".join(
[answer.body for answer in util.iterateAnswers(db, postId)])
title = question.title
wholePost = "\n\n".join(
[question.title, question.body, answers, question.tags])
return (title, wholePost)
def __iter__(self):
for question in util.iterateQuestions(self.db, self.topic, self.postList):
answers = [answer for answer in util.iterateAnswers(self.db, [question.id])]
tokens = tokenizePost(question.title, question.body, [answer.body for answer in answers], question.tags)
if Config.debug and self.ctr > 0 and (self.ctr % 5000)==0:
now = time.time()
print >>sys.stderr, "Posts imported:", self.ctr, "(in %0.1fs, %0.2fpost/s)" % (
(now-self.t0), self.ctr/(now-self.tbegin)
)
self.t0 = now
self.corpusToPost[self.ctr] = question.id
self.ctr += 1
yield self.dictionary.doc2bow([utoken for utoken in self.unicodifyTokens(tokens)], allow_update=True)
def __iter__(self):
for question in util.iterateQuestions(self.db, self.topic,
self.postList):
answers = [
answer
for answer in util.iterateAnswers(self.db, [question.id])
]
tokens = tokenizePost(question.title, question.body,
[answer.body for answer in answers],
question.tags)
if Config.debug and self.ctr > 0 and (self.ctr % 5000) == 0:
now = time.time()
print >> sys.stderr, "Posts imported:", self.ctr, "(in %0.1fs, %0.2fpost/s)" % (
(now - self.t0), self.ctr / (now - self.tbegin))
self.t0 = now
self.corpusToPost[self.ctr] = question.id
self.ctr += 1
yield self.dictionary.doc2bow(
[utoken for utoken in self.unicodifyTokens(tokens)],
allow_update=True)
