def tokenizer_pku():
f = open('dict.json', 'r', encoding='utf-8')
dict = json.load(f)
f.close()
trie = Trie()
trie.update(dict)
print(type(trie))
text = 'NLP统计模型没有加规则，聪明人知道自己加。英文、数字、自定义词典统统都是规则。'
print(split_sents(text, trie))
tokenizer = hanlp.load('PKU_NAME_MERGED_SIX_MONTHS_CONVSEG')
tokenizer = hanlp.pipeline() \
.append(split_sents, output_key=('parts', 'offsets', 'words'), trie=trie) \
.append(tokenizer, input_key='parts', output_key='tokens') \
.append(merge_parts, input_key=('tokens', 'offsets', 'words'), output_key='merged')
print(tokenizer(text))
def create_trie():
trie = Trie()
trie['自然'] = 'nature'
trie['自然人'] = 'human'
trie['自然语言'] = 'language'
trie['自语'] = 'talk to oneself'
trie['入门'] = 'introduction'
return trie
def __init__(self,
hanlp_tokenizer,
hanlp_tagger,
user_dict_path,
stop_words_path,
consider_tags_path,
ignore_tag='-'):
self.hanlp_tokenizer = hanlp_tokenizer
self.tagger = hanlp_tagger
self.ignore_tag = ignore_tag
self.stop_words = self.load_stop_words(stop_words_path)
self.considered_tags = self.load_consider_tags(consider_tags_path)
self.user_dict = self.load_user_dict(user_dict_path)
self.trie = Trie()
self.trie.update(self.user_dict)
self.tokenizer = hanlp.pipeline() \
.append(self.split_sentences, output_key=('parts', 'offsets', 'words')) \
.append(self.hanlp_tokenizer, input_key='parts', output_key='tokens') \
.append(self.merge_parts, input_key=('tokens', 'offsets', 'words'), output_key='merged')
def split_sents(text: str, trie: Trie):
words = trie.parse_longest(text)
sents = []
pre_start = 0
offsets = []
for word, value, start, end in words:
if pre_start != start:
sents.append(text[pre_start:start])
offsets.append(pre_start)
pre_start = end
if pre_start != len(text):
sents.append(text[pre_start:])
offsets.append(pre_start)
return sents, offsets, words
def split_sents(text: str, trie: Trie):
words = trie.parse_longest(text)
# https://github.com/hankcs/HanLP/blob/master/tests/demo/zh/demo_cws_trie.py
# 对官方提供的自定义字典的改造，以处理以下问题：
# “千米/时”应划分为完整的，而不是再划分出“米/时”
keys_rm_list = []
for i, key_i in enumerate(words):
for j, key_j in enumerate(words):
if (i != j) and (key_i[3]
== key_j[3]) and (key_i[2] < key_j[2]):
keys_rm_list.append((key_j[2], key_j[3]))
elif (i != j) and (key_i[3]
== key_j[3]) and (key_i[2] > key_j[2]):
keys_rm_list.append((key_i[2], key_i[3]))
elif (i != j) and (key_i[2]
== key_j[2]) and (key_i[3] < key_j[3]):
keys_rm_list.append((key_i[2], key_i[3]))
elif (i != j) and (key_i[2]
== key_j[2]) and (key_i[3] > key_j[3]):
keys_rm_list.append((key_j[2], key_j[3]))
words = list(filter(lambda x: (x[2], x[3]) not in keys_rm_list, words))
sents = []
pre_start = 0
offsets = []
for word, value, start, end in words:
if pre_start != start:
sents.append(text[pre_start:start])
offsets.append(pre_start)
pre_start = end
if pre_start != len(text):
sents.append(text[pre_start:])
offsets.append(pre_start)
return sents, offsets, words
class hanLPTokenizer:
def __init__(self,
hanlp_tokenizer,
hanlp_tagger,
user_dict_path,
stop_words_path,
consider_tags_path,
ignore_tag='-'):
self.hanlp_tokenizer = hanlp_tokenizer
self.tagger = hanlp_tagger
self.ignore_tag = ignore_tag
self.stop_words = self.load_stop_words(stop_words_path)
self.considered_tags = self.load_consider_tags(consider_tags_path)
self.user_dict = self.load_user_dict(user_dict_path)
self.trie = Trie()
self.trie.update(self.user_dict)
self.tokenizer = hanlp.pipeline() \
.append(self.split_sentences, output_key=('parts', 'offsets', 'words')) \
.append(self.hanlp_tokenizer, input_key='parts', output_key='tokens') \
.append(self.merge_parts, input_key=('tokens', 'offsets', 'words'), output_key='merged')
def split_sentences(self, text: str):
words = self.trie.parse_longest(text)
sentences = []
pre_start = 0
offsets = []
for word, value, start, end in words:
if pre_start != start:
sentences.append(text[pre_start:start])
offsets.append(pre_start)
pre_start = end
if pre_start != len(text):
sentences.append(text[pre_start:])
offsets.append(pre_start)
return sentences, offsets, words
@staticmethod
def merge_parts(parts, offsets, words):
items = [(i, p) for (i, p) in zip(offsets, parts)]
items += [(start, [word]) for (word, value, start, end) in words]
# In case you need the tag, use the following line instead
# items += [(start, [(word, value)]) for (word, value, start, end) in words]
return [each for x in sorted(items) for each in x[1]]
def tokenize(self, text):
"""
:param text: str
:return:
"""
return self.tokenizer(text)['merged']
def tag(self, tokens):
"""
:param tokens: list
:return:
"""
return self.tagger(tokens)
def tag_stop_words(self, tokens, tags):
new_tags = []
for i in range(len(tokens)):
if tokens[i] in self.stop_words:
new_tags.append(self.ignore_tag)
else:
new_tags.append(tags[i])
return new_tags
def tag_unconsidered_tags(self, tags):
new_tags = []
for tag in tags:
if tag.lower() in self.considered_tags:
new_tags.append(tag)
else:
new_tags.append(self.ignore_tag)
return new_tags
def tokenize_filter(self, text):
tokens = self.tokenize(text)
tags = self.tag(tokens)
tags = self.tag_stop_words(tokens, tags) # remove stop words
tags = self.tag_unconsidered_tags(tags) # tag filter
tagged_tokens = []
for i in range(len(tags)):
tagged_tokens.append((tokens[i], tags[i]))
return tagged_tokens
@staticmethod
def load_txt_data(path, mode='utf-8-sig', origin=False):
"""
This func is used to reading txt file
:param origin:
:param path: path where file stored
:param mode:
:type path: str
:return: string lines in file in a list
:rtype: list
"""
if type(path) != str:
raise TypeError
res = []
file = open(path, 'rb')
lines = file.read().decode(mode, errors='ignore')
for line in lines.split('\n'):
line = line.strip()
if origin:
res.append(line)
else:
if line:
res.append(line)
file.close()
return res
def load_user_dict(self, path):
raw = self.load_txt_data(path)
user_word_dict = {}
for i in range(len(raw)):
word = raw[i].split('\t')[0]
if word not in user_word_dict:
user_word_dict[word] = ' '
return user_word_dict
def load_stop_words(self, path):
return set(self.load_txt_data(path) + stopwords.words('english'))
def load_consider_tags(self, path):
return set(
[x.split('\t')[0].lower() for x in self.load_txt_data(path)])
# -*- coding:utf-8 -*-
# Author: hankcs
# Date: 2019-12-28 21:25
from hanlp.common.trie import Trie
import hanlp
tokenizer = hanlp.load('LARGE_ALBERT_BASE')
text = 'NLP统计模型没有加规则，聪明人知道自己加。英文、数字、自定义词典统统都是规则。'
print(tokenizer(text))
trie = Trie()
trie.update({'自定义词典': 'custom_dict', '聪明人': 'smart'})
def split_sents(text: str, trie: Trie):
words = trie.parse_longest(text)
sents = []
pre_start = 0
offsets = []
for word, value, start, end in words:
if pre_start != start:
sents.append(text[pre_start:start])
offsets.append(pre_start)
pre_start = end
if pre_start != len(text):
sents.append(text[pre_start:])
offsets.append(pre_start)
return sents, offsets, words
# -*- coding:utf-8 -*-
# Author: hankcs
# Date: 2020-11-11 11:08
from hanlp.common.trie import Trie
trie = Trie({'密码', '码'})
print(trie.parse_longest('密码设置'))
# -*- coding:utf-8 -*-
# Author: hankcs
# Date: 2019-12-28 21:25
from hanlp.common.trie import Trie
import hanlp
tokenizer = hanlp.load('LARGE_ALBERT_BASE')
text = 'NLP统计模型没有加规则，聪明人知道自己加。英文、数字、自定义词典统统都是规则。'
print(tokenizer(text))
trie = Trie()
trie.update({'自定义': 'custom', '词典': 'dict', '聪明人': 'smart'})
def split_sents(text: str, trie: Trie):
words = trie.parse_longest(text)
sents = []
pre_start = 0
offsets = []
for word, value, start, end in words:
if pre_start != start:
sents.append(text[pre_start: start])
offsets.append(pre_start)
pre_start = end
if pre_start != len(text):
sents.append(text[pre_start:])
offsets.append(pre_start)
return sents, offsets, words
# pipeline = hanlp.pipeline() \
# .append(hanlp.utils.rules.split_sentence, output_key='sentences') \
# .append(tokenizer, output_key='tokens') \
# .append(tagger, output_key='part_of_speech_tags') \
# .append(syntactic_parser, input_key=('tokens', 'part_of_speech_tags'), output_key='syntactic_dependencies') \
# .append(semantic_parser, input_key=('tokens', 'part_of_speech_tags'), output_key='semantic_dependencies')
# 添加自定义字典，也就是上面的num_keys_list + units_list
custrom_dict = {}
keys_list = num_keys_list + units_list
keys_list.sort(key=lambda x: len(x), reverse=True)
for key in keys_list:
custrom_dict[key] = key
trie = Trie()
# trie.update({'自定义词典': 'custom_dict', '聪明人': 'smart'})
trie.update(custrom_dict)
def split_sents(text: str, trie: Trie):
words = trie.parse_longest(text)
# https://github.com/hankcs/HanLP/blob/master/tests/demo/zh/demo_cws_trie.py
# 对官方提供的自定义字典的改造，以处理以下问题：
# “千米/时”应划分为完整的，而不是再划分出“米/时”
keys_rm_list = []
for i, key_i in enumerate(words):
for j, key_j in enumerate(words):
if (i != j) and (key_i[3]
== key_j[3]) and (key_i[2] < key_j[2]):
keys_rm_list.append((key_j[2], key_j[3]))