File size: 5,684 Bytes
ee2270f |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 |
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
import re
import regex
from itertools import chain
class MosesPunctNormalizer:
"""
This is a Python port of the Moses punctuation normalizer from
https://github.com/moses-smt/mosesdecoder/blob/master/scripts/tokenizer/normalize-punctuation.perl
"""
EXTRA_WHITESPACE = [ # lines 21 - 30
(r"\r", r""),
(r"\(", r" ("),
(r"\)", r") "),
(r" +", r" "),
(r"\) ([.!:?;,])", r")\g<1>"),
(r"\( ", r"("),
(r" \)", r")"),
(r"(\d) %", r"\g<1>%"),
(r" :", r":"),
(r" ;", r";"),
]
NORMALIZE_UNICODE_IF_NOT_PENN = [(r"`", r"'"), (r"''", r' " ')] # lines 33 - 34
NORMALIZE_UNICODE = [ # lines 37 - 50
("„", r'"'),
("“", r'"'),
("”", r'"'),
("–", r"-"),
("—", r" - "),
(r" +", r" "),
("´", r"'"),
("([a-zA-Z])‘([a-zA-Z])", r"\g<1>'\g<2>"),
("([a-zA-Z])’([a-zA-Z])", r"\g<1>'\g<2>"),
("‘", r"'"),
("‚", r"'"),
("’", r"'"),
(r"''", r'"'),
("´´", r'"'),
("…", r"..."),
]
FRENCH_QUOTES = [ # lines 52 - 57
("\u00A0«\u00A0", r'"'),
("«\u00A0", r'"'),
("«", r'"'),
("\u00A0»\u00A0", r'"'),
("\u00A0»", r'"'),
("»", r'"'),
]
HANDLE_PSEUDO_SPACES = [ # lines 59 - 67
("\u00A0%", r"%"),
("nº\u00A0", "nº "),
("\u00A0:", r":"),
("\u00A0ºC", " ºC"),
("\u00A0cm", r" cm"),
("\u00A0\\?", "?"),
("\u00A0\\!", "!"),
("\u00A0;", r";"),
(",\u00A0", r", "),
(r" +", r" "),
]
EN_QUOTATION_FOLLOWED_BY_COMMA = [(r'"([,.]+)', r'\g<1>"')]
DE_ES_FR_QUOTATION_FOLLOWED_BY_COMMA = [
(r',"', r'",'),
(r'(\.+)"(\s*[^<])', r'"\g<1>\g<2>'), # don't fix period at end of sentence
]
DE_ES_CZ_CS_FR = [
("(\\d)\u00A0(\\d)", r"\g<1>,\g<2>"),
]
OTHER = [
("(\\d)\u00A0(\\d)", r"\g<1>.\g<2>"),
]
# Regex substitutions from replace-unicode-punctuation.perl
# https://github.com/moses-smt/mosesdecoder/blob/master/scripts/tokenizer/replace-unicode-punctuation.perl
REPLACE_UNICODE_PUNCTUATION = [
(",", ","),
(r"。\s*", ". "),
("、", ","),
("”", '"'),
("“", '"'),
("∶", ":"),
(":", ":"),
("?", "?"),
("《", '"'),
("》", '"'),
(")", ")"),
("!", "!"),
("(", "("),
(";", ";"),
("」", '"'),
("「", '"'),
("0", "0"),
("1", "1"),
("2", "2"),
("3", "3"),
("4", "4"),
("5", "5"),
("6", "6"),
("7", "7"),
("8", "8"),
("9", "9"),
(r".\s*", ". "),
("~", "~"),
("’", "'"),
("…", "..."),
("━", "-"),
("〈", "<"),
("〉", ">"),
("【", "["),
("】", "]"),
("%", "%"),
]
def __init__(
self,
lang="en",
penn=True,
norm_quote_commas=True,
norm_numbers=True,
pre_replace_unicode_punct=False,
post_remove_control_chars=False,
):
"""
:param language: The two-letter language code.
:type lang: str
:param penn: Normalize Penn Treebank style quotations.
:type penn: bool
:param norm_quote_commas: Normalize quotations and commas
:type norm_quote_commas: bool
:param norm_numbers: Normalize numbers
:type norm_numbers: bool
"""
self.substitutions = [
self.EXTRA_WHITESPACE,
self.NORMALIZE_UNICODE,
self.FRENCH_QUOTES,
self.HANDLE_PSEUDO_SPACES,
]
if penn: # Adds the penn substitutions after extra_whitespace regexes.
self.substitutions.insert(1, self.NORMALIZE_UNICODE_IF_NOT_PENN)
if norm_quote_commas:
if lang == "en":
self.substitutions.append(self.EN_QUOTATION_FOLLOWED_BY_COMMA)
elif lang in ["de", "es", "fr"]:
self.substitutions.append(self.DE_ES_FR_QUOTATION_FOLLOWED_BY_COMMA)
if norm_numbers:
if lang in ["de", "es", "cz", "cs", "fr"]:
self.substitutions.append(self.DE_ES_CZ_CS_FR)
else:
self.substitutions.append(self.OTHER)
self.substitutions = list(chain(*self.substitutions))
self.pre_replace_unicode_punct = pre_replace_unicode_punct
self.post_remove_control_chars = post_remove_control_chars
def normalize(self, text):
"""
Returns a string with normalized punctuation.
"""
# Optionally, replace unicode puncts BEFORE normalization.
if self.pre_replace_unicode_punct:
text = self.replace_unicode_punct(text)
# Actual normalization.
for regexp, substitution in self.substitutions:
# print(regexp, substitution)
text = re.sub(regexp, substitution, str(text))
# print(text)
# Optionally, replace unicode puncts BEFORE normalization.
if self.post_remove_control_chars:
text = self.remove_control_chars(text)
return text.strip()
def replace_unicode_punct(self, text):
for regexp, substitution in self.REPLACE_UNICODE_PUNCTUATION:
text = re.sub(regexp, substitution, str(text))
return text
def remove_control_chars(self, text):
return regex.sub(r"\p{C}", "", text)
|