347 lines
13 KiB
Python
347 lines
13 KiB
Python
import getopt
|
||
import json
|
||
import math
|
||
import os
|
||
import re
|
||
import sys
|
||
import traceback
|
||
from numpy import NaN
|
||
import requests, time
|
||
|
||
import pandas as pd
|
||
from flask import Flask, jsonify, render_template, request
|
||
from flask_cors import CORS
|
||
from pypinyin import Style, lazy_pinyin, style
|
||
|
||
app = Flask(__name__)
|
||
app.config['JSON_AS_ASCII'] = False
|
||
CORS(app)
|
||
|
||
TONE_TABLE = [
|
||
set(['ā', 'ē', 'ī', 'ō', 'ū', 'ǖ']),
|
||
set(['á', 'é', 'í', 'ó', 'ú', 'ǘ']),
|
||
set(['ǎ', 'ě', 'ǐ', 'ǒ', 'ǔ', 'ǚ']),
|
||
set(['à', 'è', 'ì', 'ò', 'ù', 'ǜ']),
|
||
set(['a', 'e', 'i', 'o', 'u', 'ü'])
|
||
]
|
||
|
||
TONE_MAP = {
|
||
'ā': 'a', 'ē': 'e', 'ī': 'i', 'ō': 'o', 'ū': 'u', 'ǖ': 'v',
|
||
'á': 'a', 'é': 'e', 'í': 'i', 'ó': 'o', 'ú': 'u', 'ǘ': 'v',
|
||
'ǎ': 'a', 'ě': 'e', 'ǐ': 'i', 'ǒ': 'o', 'ǔ': 'u', 'ǚ': 'v',
|
||
'à': 'a', 'è': 'e', 'ì': 'i', 'ò': 'o', 'ù': 'u', 'ǜ': 'v',
|
||
'a': 'a', 'e': 'e', 'i': 'i', 'o': 'o', 'u': 'u', 'ü': 'v'
|
||
}
|
||
|
||
INITIAL_WEIGHT = {"y":12789,"j":9523,"sh":9043,"zh":7735,"x":7656,"b":7649,"l":7240,"d":6712,"h":5992,"q":5711,"g":5614,"w":5475,"m":5111,"f":5059,"ch":4965,"t":4680,"r":3906,"z":3186,"s":3149,"c":2236,"k":2210,"":2142,"n":2079,"p":1866,"，":549,"、":1}
|
||
FINAL_WEIGHT = {"i":21826,"u":18074,"an":7550,"ing":5519,"ian":5083,"ou":4657,"en":4652,"eng":4648,"ong":4567,"ang":4460,"ao":4450,"e":4446,"in":3827,"ai":3721,"uan":3337,"uo":3279,"ei":3205,"a":3141,"ui":2936,"un":2366,"iao":2299,"iu":1937,"iang":1845,"ie":1786,"ue":1367,"o":1356,"ia":1231,"uang":1061,"er":1008,"ua":864,"，":549,"v":417,"iong":381,"uai":373,"ve":59,"、":1}
|
||
|
||
def tone_only(pinyin, **kwargs):
|
||
for i in range(5):
|
||
if len(set(pinyin).intersection(TONE_TABLE[i])) > 0:
|
||
return str(i + 1)
|
||
return pinyin
|
||
|
||
def pinyin_to_tone(pinyin, **kwargs):
|
||
pinyin = ''.join(filter(str.isdigit, pinyin))
|
||
if(len(pinyin) <= 0):
|
||
return '5'
|
||
if(pinyin[0] == '1'):
|
||
return '1'
|
||
elif (pinyin[0] == '2'):
|
||
return '2'
|
||
elif (pinyin[0] == '3'):
|
||
return '3'
|
||
elif (pinyin[0] == '4'):
|
||
return '4'
|
||
return '5'
|
||
|
||
|
||
def pinyin_only(pinyin, **kwargs):
|
||
for i in range(5):
|
||
if len(set(pinyin).intersection(TONE_TABLE[i])) > 0:
|
||
for s in TONE_TABLE[i]:
|
||
pinyin = pinyin.replace(s, TONE_MAP[s])
|
||
return pinyin
|
||
|
||
def merge_pinyin_tone(pinyin, tone):
|
||
plist = pinyin.split()
|
||
for i, t in enumerate(tone):
|
||
if(tone != '5'):
|
||
plist[i] = plist[i] + t
|
||
return ' '.join(plist)
|
||
|
||
style.register('TONE_ONLY', tone_only)
|
||
|
||
def trim_space(s):
|
||
return s.replace(' ', '').replace(',', '').replace('，', '')
|
||
|
||
def remove_at(s, i):
|
||
return s[0:i] + s[i+1:]
|
||
|
||
def compute_pinyin(s, style=None):
|
||
if style is None:
|
||
return ' '.join(lazy_pinyin(s))
|
||
else:
|
||
return ' '.join(lazy_pinyin(s, style=style))
|
||
|
||
all_idiom = None
|
||
|
||
def query_word(word):
|
||
# {"tipsList":[{"fpindex":"andmkbblbeoojhcp","fpinyin":"bā miàn zhāng luo","fname":"八面张罗"}],"found":"true"}
|
||
ret = requests.get("https://cy.hwxnet.com/searchtips.do?wd=%s×tamp=%d&qt=1" % (word,int(time.time() * 1000)))
|
||
time.sleep(0.1)
|
||
data = ret.json()
|
||
if len(data["tipsList"]) > 0:
|
||
return data["tipsList"][0]['fpinyin']
|
||
return ''
|
||
|
||
|
||
def init():
|
||
global all_idiom
|
||
if os.path.exists("all_idiom.csv"):
|
||
all_idiom = pd.read_csv('all_idiom.csv')
|
||
all_idiom["pinyin_initials"].str.split(",").explode("splited").value_counts().to_json()
|
||
all_idiom["pinyin_finals"].str.split(",").explode("splited").value_counts().to_json()
|
||
else:
|
||
all_idiom = pd.read_json('idiom.json')
|
||
idiom_frequency = pd.read_csv('idiom_frequency.csv')
|
||
all_idiom = all_idiom.merge(idiom_frequency, how='outer', on='word')
|
||
all_idiom['frequency'] = all_idiom['frequency'].fillna(1).astype(int)
|
||
all_idiom['pinyin'] = all_idiom.apply(lambda x: compute_pinyin(x['word'], style=Style.TONE) if x['pinyin'] != NaN else x['pinyin'], axis=1)
|
||
all_idiom['pinyin_r'] = all_idiom.apply(lambda x: pinyin_only(x['pinyin']) if x['pinyin_r'] != NaN else x['pinyin_r'], axis=1)
|
||
all_idiom['pinyin_rt'] = all_idiom.apply(lambda x: int(''.join(map(lambda y: str(len(y)), re.split('[ ,，]',x['pinyin_r'])))), axis=1)
|
||
all_idiom['pinyin_tone'] = all_idiom.apply(lambda x: ''.join([tone_only(y) for y in x['pinyin'].split()]), axis=1)
|
||
all_idiom['pinyin_initials' ] = all_idiom.apply(lambda x: ','.join(list(lazy_pinyin(x['word'], style=Style.INITIALS, strict=False))), axis=1)
|
||
all_idiom['pinyin_finals'] = all_idiom.apply(lambda x: ','.join(list(lazy_pinyin(x['word'], style=Style.FINALS, strict=False))), axis=1)
|
||
all_idiom.to_csv("all_idiom.csv")
|
||
|
||
def fetch_next_input(parameter):
|
||
parameter_rst = parameter.split(';', 1)
|
||
if len(parameter_rst) > 1:
|
||
parameter_rst = parameter_rst[1]
|
||
else:
|
||
parameter_rst = ''
|
||
parameter = parameter.split(';')[0]
|
||
return parameter,parameter_rst
|
||
|
||
def filter_group_mode1(parameter, group, hits):
|
||
for i in range(len(parameter)):
|
||
key0 = 'pinyin0_%d' % i
|
||
key1 = 'pinyin1_%d' % i
|
||
key2 = 'pinyin2_%d' % i
|
||
group[key0] = group.apply(lambda x: ''.join(set(trim_space(x['pinyin_r']))), axis=1)
|
||
group[key1] = group.apply(lambda x: ''.join(set(remove_at(trim_space(x['pinyin_r']), i))), axis=1)
|
||
group[key2] = group.apply(lambda x: trim_space(x['pinyin_r'])[i], axis=1)
|
||
includes = set()
|
||
for i, hit in enumerate(list(trim_space(hits))):
|
||
if hit == '2':
|
||
includes.add(parameter[i])
|
||
for i, hit in enumerate(list(trim_space(hits))):
|
||
key0 = 'pinyin0_%d' % i
|
||
key1 = 'pinyin1_%d' % i
|
||
key2 = 'pinyin2_%d' % i
|
||
target = parameter[i]
|
||
if hit == '0' and target not in includes:
|
||
group = group[~group[key0].str.contains(target, regex=False, case=True, na=False)]
|
||
elif hit == '1':
|
||
group = group[(group[key2] != target) & (group[key1].str.contains(target, regex=False, case=True, na=False))]
|
||
includes.add(target)
|
||
elif hit == '2':
|
||
group = group[group[key2] == target]
|
||
if len(group) <= 1:
|
||
break
|
||
return group
|
||
|
||
def filter_with_target_field(group, field_name, values, value_hits):
|
||
includes_value = set()
|
||
for i in range(4):
|
||
if value_hits[i] == '2':
|
||
includes_value.add(values[i])
|
||
for i in range(4):
|
||
if value_hits[i] == '0' and values[i] not in includes_value:
|
||
group = group[~group[field_name].str.contains(values[i], regex=False, case=True, na=False)]
|
||
elif value_hits[i] == '1':
|
||
group = group[(group[field_name].str[i] != values[i]) & group[field_name].str.contains(values[i], regex=False, case=True, na=False)]
|
||
includes_value.add(values[i])
|
||
elif value_hits[i] == '2':
|
||
group = group[group[field_name].str[i] == values[i]]
|
||
if len(group) <= 1:
|
||
break
|
||
return group
|
||
|
||
def filter_group_model2(parameter, group, hits, tones, tone_hits, word_hits):
|
||
group = filter_with_target_field(group, 'pinyin_tone', tones, tone_hits)
|
||
if len(group) <= 1:
|
||
return group
|
||
group = filter_with_target_field(group, 'word', parameter, word_hits)
|
||
if len(group) <= 1:
|
||
return group
|
||
hits = hits.split()
|
||
for i in range(4):
|
||
target = parameter[i]
|
||
targets = list()
|
||
targets.append(lazy_pinyin(target, style=Style.INITIALS, strict=False)[0])
|
||
targets.append(lazy_pinyin(target, style=Style.FINALS, strict=False)[0])
|
||
pinyin_hit = hits[i]
|
||
for j,name in enumerate(['initials', 'finals']):
|
||
key = 'pinyin_%s' % name
|
||
if pinyin_hit[j] == '0':
|
||
group = group[group[key].str.count('(^|[,])%s([,]|$)' + targets[j]) == 0]
|
||
elif pinyin_hit[j] == '1':
|
||
group = group[(group[key].str.count('(^|[,])%s([,]|$)' % targets[j]) > 0) & (group[key].str.count(('^(\w*,){%d}%s([,]|$)' % (i, targets[j]))) == 0)]
|
||
elif pinyin_hit[j] == '2':
|
||
group = group[group[key].str.count(('^(\w*,){%d}%s([,]|$)' % (i, targets[j]))) > 0]
|
||
if len(group) <= 1:
|
||
return group
|
||
return group
|
||
|
||
def filter_logic(mode, parameter):
|
||
global all_idiom
|
||
if mode == '0':
|
||
groups = all_idiom.groupby(by='pinyin_rt')
|
||
group = groups.get_group(int(parameter)).copy()
|
||
return all_idiom, group
|
||
elif mode == '1':
|
||
parameter, parameter_rst = fetch_next_input(parameter)
|
||
|
||
hits = parameter.split(',')[1]
|
||
parameter = trim_space(parameter.split(',')[0])
|
||
|
||
count = ''.join([str(len(x)) for x in hits.split()])
|
||
groups = all_idiom.groupby(by='pinyin_rt')
|
||
group = groups.get_group(int(count)).copy()
|
||
|
||
while(True):
|
||
group = filter_group_mode1(parameter, group, hits)
|
||
if(len(group) > 1 and len(parameter_rst) > 0):
|
||
parameter, parameter_rst = fetch_next_input(parameter_rst)
|
||
|
||
hits = parameter.split(',')[1]
|
||
parameter = trim_space(parameter.split(',')[0])
|
||
elif len(group) <= 0:
|
||
break
|
||
else:
|
||
break
|
||
return all_idiom, group
|
||
elif mode == '2':
|
||
four_idiom = all_idiom[all_idiom['word'].str.len() == 4]
|
||
group = four_idiom[four_idiom['pinyin_tone'].str.startswith(parameter)].copy()
|
||
return four_idiom, group
|
||
elif mode == '3':
|
||
four_idiom = all_idiom[all_idiom['word'].str.len() == 4]
|
||
|
||
parameter, parameter_rst = fetch_next_input(parameter)
|
||
|
||
hits = parameter.split(',')[1]
|
||
parameter = parameter.split(',')[0]
|
||
tones = ''.join(lazy_pinyin(parameter, style="TONE_ONLY"))
|
||
tone_hits=hits[-4:]
|
||
word_hits=hits[-9:-5]
|
||
hits=hits[:-10]
|
||
|
||
group = four_idiom.copy()
|
||
while(True):
|
||
group = filter_group_model2(parameter, group, hits, tones, tone_hits, word_hits)
|
||
if(len(group) > 1 and len(parameter_rst) > 0):
|
||
parameter, parameter_rst = fetch_next_input(parameter_rst)
|
||
|
||
hits = parameter.split(',')[1]
|
||
parameter = parameter.split(',')[0]
|
||
tones = ''.join(lazy_pinyin(parameter, style="TONE_ONLY"))
|
||
tone_hits=hits[-4:]
|
||
word_hits=hits[-9:-5]
|
||
hits=hits[:-10]
|
||
elif len(group) <= 0:
|
||
break
|
||
else:
|
||
break
|
||
return four_idiom, group
|
||
elif mode == '4':
|
||
four_idiom = all_idiom[all_idiom['word'].str.len() == 4]
|
||
group = four_idiom[four_idiom['word'].str.count(parameter) > 0].copy()
|
||
return four_idiom, group
|
||
elif mode == '5':
|
||
four_idiom = all_idiom[all_idiom['word'].str.len() == 4]
|
||
group = four_idiom[four_idiom['pinyin_r'].str.count(parameter) > 0].copy()
|
||
return four_idiom, group
|
||
return None, None
|
||
|
||
def print_max_group(all_idiom, group, num):
|
||
for item in get_max_group(all_idiom, group, num):
|
||
print(item)
|
||
|
||
def get_pinyin_weight(initials, finals):
|
||
initials = set(initials.split(','))
|
||
weight = sum([INITIAL_WEIGHT[w] for w in initials])
|
||
finals = set(finals.split(','))
|
||
weight += sum([FINAL_WEIGHT[w] for w in finals])
|
||
return weight
|
||
|
||
def get_max_group(all_idiom, group, num):
|
||
result = list()
|
||
if(len(group) == 0):
|
||
return result
|
||
group['pinyin_c'] = group.apply(lambda x: (math.log(x['frequency'], 2)/16 + 1) * get_pinyin_weight(x['pinyin_initials'], x['pinyin_finals']), axis=1)
|
||
ret_list = group.nlargest(num, ['pinyin_c', 'frequency']).index.tolist()
|
||
for i in ret_list:
|
||
result.append(json.loads(all_idiom.loc[i].to_json(orient = 'index',force_ascii=False)))
|
||
return result
|
||
|
||
@app.route('/predict', methods=['GET'])
|
||
def predict():
|
||
try:
|
||
mode = request.args.get('mode')
|
||
parameter = request.args.get('parameter')
|
||
if len(trim_space(parameter)) == 0:
|
||
return jsonify({'status': 1, 'message': '请输入参数', 'result': []})
|
||
all_idiom, group = filter_logic(mode, parameter)
|
||
result = get_max_group(all_idiom, group, 6)
|
||
return jsonify({'status': 0, 'message': 'success', 'result': result})
|
||
except:
|
||
traceback.print_exc()
|
||
return jsonify({'status': 2, 'message': '执行出错', 'result': []})
|
||
|
||
|
||
@app.route('/', methods=['GET'])
|
||
def index():
|
||
return render_template('index.html')
|
||
|
||
def main(argv):
|
||
# mode = '0'
|
||
# parameter = '3323'
|
||
# mode = '1'
|
||
# parameter = 'bai vvv vv vvv,012 000 00 000;bai tou er xin,012 010 10 001'
|
||
mode = '2'
|
||
parameter = '2134'
|
||
# mode = '3'
|
||
# parameter = '行之有效,00 11 10 00 0100 1101;各执一词,00 11 12 22 0010 1022'
|
||
num = 6
|
||
try:
|
||
opts, args = getopt.getopt(argv, "hm:p:n:", ["mode=", "parameter=", "num="])
|
||
except getopt.GetoptError:
|
||
print('test.py -m <mode> -p <parameter> -n <num>')
|
||
sys.exit(2)
|
||
for opt, arg in opts:
|
||
if opt == '-h':
|
||
print('test.py -i <mode> -o <parameter>')
|
||
sys.exit()
|
||
elif opt in ("-m", "--mode"):
|
||
mode = arg
|
||
elif opt in ("-p", "--parameter"):
|
||
parameter = arg
|
||
elif opt in ("-n", "--num"):
|
||
num = int(arg)
|
||
all_idiom, group = filter_logic(mode, parameter)
|
||
if all_idiom is not None and len(group) > 0:
|
||
print_max_group(all_idiom, group, num)
|
||
else:
|
||
print('未找到匹配项')
|
||
|
||
if __name__ == '__main__':
|
||
current_work_dir = os.path.dirname(__file__)
|
||
os.chdir(current_work_dir)
|
||
init()
|
||
# main(sys.argv[1:])
|
||
app.run(debug=True, host="0.0.0.0")
|