def addResult(arg, q):
amod = [
qm if i in q['pos'] else '<a pos=%d href="#">' % i + qm + '</a>'
for i, qm in enumerate(q['mod'])
]
title = ' '.join(q['mod'])
if arg['type'] == 0:
(retId, flag) = addCluster(
arg['senmap'], arg['senlist'], title, {
'display': ' '.join(amod),
'pos': q['pos'],
'other': '_other_' in q['mod']
})
if flag:
markSent = cleaned_sentence(arg['sent'], arg['keypos'])
arg['strlist'].append({'sentence': markSent, 'sen': retId})
elif arg['type'] == 1:
if title == arg['title']:
title = ' '.join(q['word'])
(retId, flag) = addCluster(arg['senmap'], arg['senlist'], title, {
'display': title,
'pos': [],
'other': '_other_' in q['word']
})
if flag:
markSent = cleaned_sentence(arg['sent'], arg['keypos'])
arg['strlist'].append({'sentence': markSent, 'sen': retId})
def addResult(arg, q):
amod = [qm if i in q['pos'] else '<a pos=%d href="#">' % i + qm + '</a>'
for i, qm in enumerate(q['mod'])]
title = ' '.join(q['mod'])
if arg['type'] == 0:
(retId, flag) = addCluster(arg['senmap'], arg['senlist'], title,
{'display': ' '.join(amod), 'pos': q['pos'], 'other': '_other_' in q['mod']})
if flag:
markSent = cleaned_sentence(arg['sent'], arg['keypos'])
arg['strlist'].append({'sentence': markSent, 'sen': retId})
elif arg['type'] == 1:
if title == arg['title']:
title = ' '.join(q['word'])
(retId, flag) = addCluster(arg['senmap'], arg['senlist'], title,
{'display': title, 'pos': [], 'other': '_other_' in q['word']})
if flag:
markSent = cleaned_sentence(arg['sent'], arg['keypos'])
arg['strlist'].append({'sentence': markSent, 'sen': retId})
def addResult(arg, res, pos):
qpos = [i for i, qm in enumerate(res) if not qm.isupper()]
amod = [qm if i in qpos else '<a pos=%d href="#">' % i + qm + '</a>'
for i, qm in enumerate(res)]
title = ' '.join(res)
(retId, flag) = addCluster(arg['senmap'], arg['senlist'], title,
{'display': ' '.join(amod), 'pos': qpos, 'other': '_other_' in res})
if flag:
pos = [i for i in pos if 0 <= i and i < len(arg['sent'])]
markSent = cleaned_sentence(arg['sent'], pos)
arg['strlist'].append({'sentence': markSent, 'sen': retId})
def get_qtree_db(tree, tokens, key, ctype):
nkey = []
for k in key:
if not is_upper(tokens[k]):
nkey.append(k)
if ctype == 2:
# keys = [stemmer_value(tokens[k]) for k in nkey]
keys = [tokens[k]['lemma'] for k in nkey]
keys.sort(key=lambda word: -len(word))
# rs = cl.find({'tokens.s': {'$all': keys}})
rs = cl.find({'tokens.l': {'$all': keys}})
else:
# keys = [{'$elemMatch': {'s': stemmer_value(tokens[k]), 'q': getq(tokens[k]['pos'])}} for k in nkey]
# keys.sort(key=lambda word: -len(word['$elemMatch']['s']))
# rs = cl.find({'tokens': {'$all': keys}})
keys = [{
'$elemMatch': {
'l': tokens[k]['lemma'],
'q': getq(tokens[k]['pos'])
}
} for k in nkey]
keys.sort(key=lambda word: -len(word['$elemMatch']['l']))
rs = cl.find({'tokens': {'$all': keys}})
# retJson = {'result': [], 'desc': {'sen': []}}
# senmap = {}
# senlist = retJson['desc']['sen']
retJson = {'result': [], 'desc': {'sen': []}}
senmap2 = {}
senlist2 = retJson['desc']['sen']
strlist = retJson['result']
for sen in rs:
sent = sen['tree0']
tk = sen['tokens']
qtree = transfer_Node_i(sent)
tp = check_find(tree, key, tokens, qtree, tk, ctype)
if tp:
# senId = addCluster(senmap, senlist, tp.resultSent)
(senId2, flag1) = addCluster(senmap2, senlist2, tp.resultSent2,
{'len': tp.cost})
if flag1:
markSent = cleaned_sentence([w['t'] for w in tk], tp.qkey)
resultDict = {'sentence': markSent, 'sen': senId2}
strlist.append(resultDict)
senlist2.sort(key=lambda word: -word['count'] * 100 + word['len']
if word['title'] != '_others_' else 0)
result_part(retJson)
return retJson
def get_query_db2(tree, msg, tokens, keys):
keys.sort(key=lambda word: -len(word['$elemMatch']['l']))
rs = cl.find({'tokens': {'$all': keys}})
retJson = {'result': [], 'desc': {}}
strlist = retJson['result']
cnt = 0
for sen in rs:
sent = sen['tree0']
tk = sen['tokens']
tp = check_find(msg, tokens, sent, tk)
if tp != -1:
tplist = get_pos_list(msg, sent, tp)
markSent = cleaned_sentence([w['t'] for w in tk], tplist)
strlist.append({'sentence': markSent})
#print sen['sentence'], tplist
cnt = cnt + 1
return retJson
def get_qtree_db(tree, tokens, key, ctype):
nkey = []
for k in key:
if not is_upper(tokens[k]):
nkey.append(k)
if ctype == 2:
# keys = [stemmer_value(tokens[k]) for k in nkey]
keys = [tokens[k]['lemma'] for k in nkey]
keys.sort(key=lambda word: -len(word))
# rs = cl.find({'tokens.s': {'$all': keys}})
rs = cl.find({'tokens.l': {'$all': keys}})
else:
# keys = [{'$elemMatch': {'s': stemmer_value(tokens[k]), 'q': getq(tokens[k]['pos'])}} for k in nkey]
# keys.sort(key=lambda word: -len(word['$elemMatch']['s']))
# rs = cl.find({'tokens': {'$all': keys}})
keys = [{'$elemMatch': {'l': tokens[k]['lemma'], 'q': getq(tokens[k]['pos'])}} for k in nkey]
keys.sort(key=lambda word: -len(word['$elemMatch']['l']))
rs = cl.find({'tokens': {'$all': keys}})
# retJson = {'result': [], 'desc': {'sen': []}}
# senmap = {}
# senlist = retJson['desc']['sen']
retJson = {'result': [], 'desc': {'sen': []}}
senmap2 = {}
senlist2 = retJson['desc']['sen']
strlist = retJson['result']
for sen in rs:
sent = sen['tree0']
tk = sen['tokens']
qtree = transfer_Node_i(sent)
tp = check_find(tree, key, tokens, qtree, tk, ctype)
if tp:
# senId = addCluster(senmap, senlist, tp.resultSent)
(senId2, flag1) = addCluster(senmap2, senlist2, tp.resultSent2, {'len': tp.cost})
if flag1:
markSent = cleaned_sentence([w['t'] for w in tk], tp.qkey)
resultDict = {'sentence': markSent, 'sen': senId2}
strlist.append(resultDict)
senlist2.sort(key=lambda word: -word['count'] * 100 + word['len'] if word['title'] != '_others_' else 0)
result_part(retJson)
return retJson
def addResult(arg, title):
(retId, flag) = addCluster(arg['senmap'], arg['senlist'], title)
# print arg['senmap'], title
if flag:
markSent = cleaned_sentence(arg['sent'], [arg['keypos']])
arg['strlist'].append({'sentence': markSent, 'sen': retId})
def addResult(arg, title):
(retId, flag) = addCluster(arg['senmap'], arg['senlist'], title)
# print arg['senmap'], title
if flag:
markSent = cleaned_sentence(arg['sent'], [arg['keypos']])
arg['strlist'].append({'sentence': markSent, 'sen': retId})