"""

if len(target_list) == 0:

return None

s_set = set([item for item in source])

contain_score = []

un_contain_score = [] # target当中相比于source多出来的部分

for target in target_list:

t_set = set([t for t in target])

contain_score.append(len(s_set & t_set))

un_contain_score.append(0)

char_match_score = [contain_score[idx] for idx in range(len(target_list))]

# 如果最高匹配分数为0,说明一个匹配的都没有，那么返回None

if max(char_match_score) == 0: return None

# 下面计算编辑距离分数

e_d_score = [ len(source) - editdistance.eval(source, t) for t in target_list]

final_score = [char_match_score[idx] + e_d_score[idx] for idx in range(len(target_list))]

"""

if len(target_list) == 0:

return None

s_set = set([item for item in source])

contain_score = []

un_contain_score = []# target当中相比于source多出来的部分

for target in target_list:

t_set = set([t for t in target])

contain_score.append(len(s_set & t_set))

# un_contain_score.append(len(t_set.difference(s_set))) #

un_contain_score.append(0)

char_match_score = [contain_score[idx] * 4 for idx in range(len(target_list))]

# 如果最高匹配分数为0,说明一个匹配的都没有，返回None

if max(char_match_score) == 0: return None

# 如果匹配分数为0，直接给最小分

for idx in range(len(target_list)):

if char_match_score[idx] == 0:

char_match_score[idx] = -65530

# 下面计算编辑距离分数

e_d_score = [ len(source) - editdistance.eval(source, t) for t in target_list]

final_score = [char_match_score[idx] + e_d_score[idx] for idx in range(len(target_list))]

if re.findall('[^\d\-\.\%]', source):

return most_similar_out(source, target_list)

else:

new_target_list = [target for target in target_list if check_num_exactly_match(source, target)[0] >= 1]

if len(new_target_list) == 0: return None

"""处理question中的数字映射问题"""

try:

mat_re = re.compile(re_expre)

q_mat = mat_re.findall(question)

alab_num = upper_to_num_dict.get(q_mat[0][0])

question = question.replace(q_mat[0][0] + q_mat[0][1], str(alab_num) + q_mat[0][1])

return question

except:

# @param S, a list of integer

# @return a list of lists of integer

def __init__(self, max_len=5):

self.max_len = max_len

def dfs(self, start, S, result, father_subsets):

if len(father_subsets) > self.max_len: return

result.append(father_subsets)

for i in range(start, len(S)):

self.dfs(i+1, S, result, father_subsets+[S[i]])

def subsets(self, S):

# none case

if S is None:

return []

# deep first search

result = []

self.dfs(0, sorted(S), result, [])

"""从句子s中找与w最相近的片段，

sw = jieba.lcut(s)

sl = [x for x in list(sw)]

if mode == 'output': # 这里只需要组合最大长度为５吧　

s_subset = Subsets().subsets(sw)

sl = [''.join(item) for item in s_subset if len(item) > 0]

elif mode == 'input': # 继续采用之前的匹配方式

sl.extend([char for char in s])

sl.extend([''.join(i) for i in zip(sw, sw[1:])]) # 2-gram

sl.extend([''.join(i) for i in zip(sw, sw[1:], sw[2:])]) # 3-gram

sl.extend([''.join(i) for i in zip(sw, sw[1:], sw[2:], sw[3:])]) # 4-grarm

sl.extend([''.join(i) for i in zip(sw, sw[1:], sw[2:], sw[3:], sw[4:])]) # 5-gram

sl.extend([''.join(i) for i in zip(sw, sw[1:], sw[2:], sw[3:], sw[4:], sw[5:])]) # 6-gram

sl.extend([''.join(i) for i in zip(sw, sw[1:], sw[2:], sw[3:], sw[4:], sw[5:], sw[6:])]) # 7-gram

else:

raise ValueError('Unsupported mode! ')

return most_similar(w, sl) # delte in 08-19

"""

# col_name中的括号去掉

col_name = re.sub(re_brack, "", col_name)

question = trans_question_acc(question)

col_in_question = most_similar_2(col_name, question)

start_idx = 0

if col_in_question is not None and col_in_question in question:

start_idx = question.index(col_in_question) + len(col_in_question)

elif col_in_question is not None:# 标题不在question里面,分散在里面

pass

else:

return None, None, None

pattern = re.compile(r'-\d+\.\d+|-\d+|\d+\.\d+|\d+')

num_find = pattern.findall(question, start_idx) # 匹配到标题后的第一个数字

if num_find is None:

return

try:

start_idx = question.find(num_find[0], start_idx) # 第一个数字是我们想要标注的位置,找到他的起始位置

end_idx = start_idx + len(num_find[0])

# print(start_idx, '\t', end_idx, '\t', question[start_idx:end_idx])

return start_idx, end_idx, question[start_idx:end_idx]

except:

print('except found')

question = '收入为-10.5的单位有哪些'

col_name = '收入'

word = '-10.5'

assert alap_an_cn_mark(question, col_name, word) == (3, 8, '-10.5')

question = '收入为-100的单位有哪些'

col_name = '收入'

word = '-100'

assert alap_an_cn_mark(question, col_name, word) == (3, 7, '-100')

question = '你好啊，我想知道出现次数大于8万，频率还高于0.1的都是什么词来着'

col_name = '频率'

val = '0.15666'

assert alap_an_cn_mark(question, col_name, val) == (22, 25, '0.1')

question = '你好啊，我想知道出现次数大于8万，频率还高于4325的都是什么词来着'

col_name = '频率'

val = '133'

assert alap_an_cn_mark(question, col_name, val) == (22, 26, '4325')

question = '你好啊，我想知道出现次数大于8万，频率还高于4万的都是什么词来着'

col_name = '频率'

val = '133'

assert alap_an_cn_mark(question, col_name, val) == (22, 23, '4')

question = '2020年通车高铁线路中长超过100公里，投资高于100亿的线路叫啥名呀'

col_name = '线路长度（公里）'

#"线路名称", "沿线地区", "线路长度（公里）", "投资金额（亿元）"

val = '100'

assert alap_an_cn_mark(question, col_name, val) == (15, 18, '100')

question = '哪些股票是周涨跌幅小于0或年涨跌幅大于0的？'

col_name = '投资金额（亿元）'

#"线路名称", "沿线地区", "线路长度（公里）", "投资金额（亿元）"

val = '100'

assert alap_an_cn_mark(question, col_name, val) == (None, None, None)

question = '请问在什么时候贷款利率调整前大于6%并且贷款利率调整后大于6%？'

col_name = '贷款利率调整前'

#"贷款利率调整前", "贷款利率调整后"

val = '6'

assert alap_an_cn_mark(question, col_name, val) == (16, 17, '6')

question = '请问在什么时候贷款利率调整前大于6%并且贷款利率调整后大于6%？'

col_name = '贷款利率调整后'

#"贷款利率调整前", "贷款利率调整后"

val = '6'

assert alap_an_cn_mark(question, col_name, val) == (29, 30, '6')

question = '我想知道上周5综艺收视率超过3%的，在湖南台播的都有几个啊'

col_name = '收视率'# 百分号也干掉

word = '0.3'

assert alap_an_cn_mark(question, col_name, word) == (14, 15, '3')

question = '哪些城市的成交面积在本周是低于2的'

col_name = '本周成交面积'

word = '1.67'

assert alap_an_cn_mark(question, col_name, word) == (15, 16, '2')

question = '有没有最新股价超过5块一股而且持股数量超过五百万股的模拟组合啊'

col_name = '持股数量'

word = '500'

assert alap_an_cn_mark(question, col_name, word) == (21, 28, '5000000')

question = '这周票房达到八千万以上的影片共有几部呀'

col_name = '本周票房'

word = '8000'

assert alap_an_cn_mark(question, col_name, word) == (6, 14, '80000000')

question = '电视剧收视率排名前3的都是什么剧啊，是在哪个台播的呀'

col_name = '排名'

word = '4'

assert alap_an_cn_mark(question, col_name, word) == (9, 10, '3')

question = '新房成交环比上周大于20%而且累计同比也大于20%的上周平均成交量为多少'

col_name = '累计同比'

word = '10'

assert alap_an_cn_mark(question, col_name, word) == (22, 24, '20')

question = '场均人次小于10而且上映了10天以上的电影最大累计票房为多少万'

col_name = '上映天数'

word = '10'