nlp

bpe.py

###byte pair encoding
###Neural Machine Translation of Rare Words with Subword Units
###from https://plmsmile.github.io/2017/10/19/subword-units/
import re
def process_raw_words(words, endtag='-'):
    '''把单词分割成最小的符号，并且加上结尾符号'''
    vocabs = {}
    for word, count in words.items():
        # 加上空格
        word = re.sub(r'([a-zA-Z])', r' \1', word)
        word += ' ' + endtag
        vocabs[word] = count
    return vocabs

def get_symbol_pairs(vocabs):
    ''' 获得词汇中所有的字符pair，连续长度为2，并统计出现次数
    Args:
        vocabs: 单词dict，(word, count)单词的出现次数。单词已经分割为最小的字符
    Returns:
        pairs: ((符号1, 符号2), count)
    '''
    #pairs = collections.defaultdict(int)
    pairs = dict()
    for word, freq in vocabs.items():
        # 单词里的符号
        symbols = word.split()
        for i in range(len(symbols) - 1):
            p = (symbols[i], symbols[i + 1])
            pairs[p] = pairs.get(p, 0) + freq
    return pairs

def merge_symbols(symbol_pair, vocabs):
    '''把vocabs中的所有单词中的'a b'字符串用'ab'替换
    Args:
        symbol_pair: (a, b) 两个符号
        vocabs: 用subword(symbol)表示的单词，(word, count)。其中word使用subword空格分割
    Returns:
        vocabs_new: 替换'a b'为'ab'的新词汇表
    '''
    vocabs_new = {}
    raw = ' '.join(symbol_pair)
    merged = ''.join(symbol_pair)
    # 非字母和数字字符做转义
    bigram =  re.escape(raw)
    p = re.compile(r'(?<!\S)' + bigram + r'(?!\S)')
    for word, count in vocabs.items():
        word_new = p.sub(merged, word)
        vocabs_new[word_new] = count
    return vocabs_new

raw_words = {"low":5, "lower":2, "newest":6, "widest":3}
vocabs = process_raw_words(raw_words)
print(vocabs)

num_merges = 10
for i in range(num_merges):
    pairs = get_symbol_pairs(vocabs)
    # 选择出现频率最高的pair
    symbol_pair = max(pairs, key=pairs.get)
    vocabs = merge_symbols(symbol_pair, vocabs)
print (vocabs)

chinese_converter.py

#!/usr/bin/env python
# -*- coding:GBK -*- 

"""汉字处理的工具:
判断unicode是否是汉字，数字，英文，或者其他字符。
全角符号转半角符号。"""
def is_chinese(uchar):
        """判断一个unicode是否是汉字"""
        if uchar >= u'u4e00' and uchar<=u'u9fa5':
                return True
        else:
                return False

def is_number(uchar):
        """判断一个unicode是否是数字"""
        if uchar >= u'u0030' and uchar<=u'u0039':
                return True
        else:
                return False

def is_alphabet(uchar):
        """判断一个unicode是否是英文字母"""
        if (uchar >= u'u0041' and uchar<=u'u005a') or (uchar >= u'u0061' and uchar<=u'u007a'):
                return True
        else:
                return False

def is_other(uchar):
        """判断是否非汉字，数字和英文字符"""
        if not (is_chinese(uchar) or is_number(uchar) or is_alphabet(uchar)):
                return True
        else:
                return False

def B2Q(uchar):
        """半角转全角"""
        inside_code=ord(uchar)
        if inside_code<0x0020 or inside_code>0x7e:      #不是半角字符就返回原来的字符
                return uchar
        if inside_code==0x0020: #除了空格其他的全角半角的公式为:半角=全角-0xfee0
                inside_code=0x3000
        else:
                inside_code+=0xfee0
        return unichr(inside_code)

def Q2B(uchar):
        """全角转半角"""
        inside_code=ord(uchar)
        if inside_code==0x3000:
                inside_code=0x0020
        else:
                inside_code-=0xfee0
        if inside_code<0x0020 or inside_code>0x7e:      #转完之后不是半角字符返回原来的字符
                return uchar
        return unichr(inside_code)

def stringQ2B(ustring):
        """把字符串全角转半角"""
        return "".join([Q2B(uchar) for uchar in ustring])

def uniform(ustring):
        """格式化字符串，完成全角转半角，大写转小写的工作"""
        return stringQ2B(ustring).lower()

def string2List(ustring):
        """将ustring按照中文，字母，数字分开"""
        retList=[]
        utmp=[]
        for uchar in ustring:
                if is_other(uchar):
                        if len(utmp)==0:
                                continue
                        else:
                                retList.append("".join(utmp))
                                utmp=[]
                else:
                        utmp.append(uchar)
        if len(utmp)!=0:
                retList.append("".join(utmp))
        return retList

if __name__=="__main__":
        #test Q2B and B2Q
        for i in range(0x0020,0x007F):
                print Q2B(B2Q(unichr(i))),B2Q(unichr(i))

        #test uniform
        ustring=u'中国 人名ａ高频Ａ'
        ustring=uniform(ustring)
        ret=string2List(ustring)
        print ret

utils.py

import unicodedata
import string

all_letters = string.ascii_letters + " .,;'"
n_letters = len(all_letters)

# Turn a Unicode string to plain ASCII, thanks to http://stackoverflow.com/a/518232/2809427
def unicodeToAscii(s):
    return ''.join(
        c for c in unicodedata.normalize('NFD', s)
        if unicodedata.category(c) != 'Mn'
        and c in all_letters
    )

print(unicodeToAscii('Ślusàrski'))

#python2 and python3 print unicode
import sys, codecs
print(sys.stdout.encoding)
if sys.stdout.encoding is None or sys.stdout.encoding == 'ANSI_X3.4-1968':
    utf8_writer = codecs.getwriter('UTF-8')
    if sys.version_info.major < 3:
        sys.stdout = utf8_writer(sys.stdout, errors='replace')
    else:
        sys.stdout = utf8_writer(sys.stdout.buffer, errors='replace')