def sendDupMessage(newE, dups, commenter, db, byCrc=True):
# newE = (newE[u'datetime'], "", newE[u'id'], newE[u'title'])
newE = (newE[u'datetime'], "", newE[u'account_id'], newE[u'id'], newE[u'title'])
#datetime, userurl, userid, link, title = dups[-1]
# trip = False
# if lastSeen[2] == firstSeen[2]:
# newE, lastSeen = sorted([newE, lastSeen], reverse=True)
# else:
# newE, lastSeen, firstSeen = sorted([newE, lastSeen, firstSeen], reverse=True)
# trip = True
if db.commentWritten(newE[2]):
return
message = list()
#commenter.appendComment(galleryId, message, retries, parentCommentId=-1, childComment=None):
# shouldn'T be needed as the db sort it by date, but sometimes there is a bug ?! TODO:
dups.sort()
lastSeen = dups[-1]
firstSeen = dups[0]
seen = len(dups)
trip = firstSeen[3] != lastSeen[3]
# last seen
to, tn = (lastSeen[4].lower().strip(), newE[4].lower().strip())
titleInf = u"Title similarity: " + genUnicodeSuccessBar(5, 1-levenshtein_n(to, tn), 1, [u'★'], u'☆', u'', u'')
if byCrc:
#commentId = commenter.writeComment(newE[2], str_duplicate % (crudeTimeFormat(newE[0]-lastSeen[0]),lastSeen[2], titleInf))
message = [commenter.TYPE_COMMENT_OR_REPLY, newE[3], str_duplicate % (crudeTimeFormat(newE[0]-lastSeen[0]),lastSeen[3], titleInf), 5, -1]
else:
#commentId = commenter.writeComment(newE[2], str_similar % (crudeTimeFormat(newE[0]-lastSeen[0]), lastSeen[2], titleInf))
message = [commenter.TYPE_COMMENT_OR_REPLY, newE[3], str_similar % (crudeTimeFormat(newE[0]-lastSeen[0]), lastSeen[3], titleInf), 5, -1]
# first seen
if trip:
to, tn = (firstSeen[4].lower().strip(), newE[4].lower().strip())
titleInf = u"Title similarity: " + genUnicodeSuccessBar(5, 1-levenshtein_n(to, tn), 1, [u'★'], u'☆', u'', u'')
if byCrc:
if seen > 2:
#commentId = commenter.writeReply(newE[2], commentId, str_duplicate_firstTimes % (seen-1, crudeTimeFormat(newE[0]-firstSeen[0]), firstSeen[2], titleInf))
message.append( [commenter.TYPE_COMMENT_OR_REPLY, newE[3], str_duplicate_firstTimes % (seen-1, crudeTimeFormat(newE[0]-firstSeen[0]), firstSeen[3], titleInf), 7, -1] )
else:
#commentId = commenter.writeReply(newE[2], commentId, str_duplicate_first % (crudeTimeFormat(newE[0]-firstSeen[0]), firstSeen[2], titleInf))
message.append( [commenter.TYPE_COMMENT_OR_REPLY, newE[3], str_duplicate_first % (crudeTimeFormat(newE[0]-firstSeen[0]), firstSeen[3], titleInf), 7, -1] )
else:
if seen > 2:
#commentId = commenter.writeReply(newE[2], commentId, str_similar_firstTimes % (seen-1, crudeTimeFormat(newE[0]-firstSeen[0]), firstSeen[2], titleInf))
message.append( [commenter.TYPE_COMMENT_OR_REPLY, newE[3], str_similar_firstTimes % (seen-1, crudeTimeFormat(newE[0]-firstSeen[0]), firstSeen[3], titleInf), 7, -1] )
else:
#commentId = commenter.writeReply(newE[2], commentId, str_similar_first % (crudeTimeFormat(newE[0]-firstSeen[0]), firstSeen[2], titleInf))
message.append( [commenter.TYPE_COMMENT_OR_REPLY, newE[3], str_similar_first % (crudeTimeFormat(newE[0]-firstSeen[0]), firstSeen[3], titleInf), 7, -1] )
#sleep(28)
commenter.queue.append(message)
db.logComment(newE[3], -1)
def shouldIPost(con, elem, dups, maxPostTime):
#return True
#lastDup = dups[-1]
datetime, userurl, userid, link, title = dups[-1]
to, tn = (elem[u'title'].lower().strip(), title.lower().strip())
levn = levenshtein_n(to, tn)
# TODO: make this readable
return (
time() < elem[u'datetime'] + maxPostTime
and (
not (
# TODO: use userblacklist by id plus names till all entries have ids
( elem[u'account_url'] is not None and con.is_user_blocked(int(elem[u'account_id'])))
#or ( userurl is not None and lastDup[1].lower() in userBlackList )
)
)
and (
levn < 0.4 or (
abs(elem[u'datetime']-datetime) < 60*60*24*60 # time diff < 60 days
and u'mrw' not in elem[u'title'].lower().split(u' ') # not mrw
and u'mfw' not in elem[u'title'].lower().split(u' ') # not mfw
and not elem[u'title'].lower().startswith(u'when') # not start with when
)
)
# TODO: drop this when all user have an userid
and (elem[u'account_url'] is None or elem[u'account_url'] != userurl)
and (elem['account_id'] != 0 or elem['account_id'] != userid)
)
