从Moses Machine Translation Toolkit:
给出配置文件#########################
### MOSES CONFIG FILE ###
#########################
# input factors
[input-factors]
0
# mapping steps
[mapping]
0 T 0
[distortion-limit]
6
# feature functions
[feature]
UnknownWordPenalty
WordPenalty
PhrasePenalty
PhraseDictionaryMemory name=TranslationModel0 num-features=4 path=/home/gillin/jojomert/phrase-jojo/work.src-ref/training/model/phrase-table.gz input-factor=0 output-factor=0
LexicalReordering name=LexicalReordering0 num-features=6 type=wbe-msd-bidirectional-fe-allff input-factor=0 output-factor=0 path=/home/gillin/jojomert/phrase-jojo/work.src-ref/training/model/reordering-table.wbe-msd-bidirectional-fe.gz
Distortion
KENLM lazyken=0 name=LM0 factor=0 path=/home/gillin/jojomert/ru.kenlm order=5
# dense weights for feature functions
[weight]
UnknownWordPenalty0= 1
WordPenalty0= -1
PhrasePenalty0= 0.2
TranslationModel0= 0.2 0.2 0.2 0.2
LexicalReordering0= 0.3 0.3 0.3 0.3 0.3 0.3
Distortion0= 0.3
LM0= 0.5
我需要阅读[weights]
部分的参数:
UnknownWordPenalty0= 1
WordPenalty0= -1
PhrasePenalty0= 0.2
TranslationModel0= 0.2 0.2 0.2 0.2
LexicalReordering0= 0.3 0.3 0.3 0.3 0.3 0.3
Distortion0= 0.3
LM0= 0.5
我一直这样做:
def read_params_from_moses_ini(mosesinifile):
parameters_string = ""
for line in reversed(open(mosesinifile, 'r').readlines()):
if line.startswith('[weight]'):
return parameters_string
else:
parameters_string+=line.strip() + ' '
获得此输出:
LM0= 0.5 Distortion0= 0.3 LexicalReordering0= 0.3 0.3 0.3 0.3 0.3 0.3 TranslationModel0= 0.2 0.2 0.2 0.2 PhrasePenalty0= 0.2 WordPenalty0= -1 UnknownWordPenalty0= 1
然后使用
解析输出moses_param_pattern = re.compile(r'''([^\s=]+)=\s*((?:[^\s=]+(?:\s|$))*)''')
def parse_parameters(parameters_string):
return dict((k, list(map(float, v.split())))
for k, v in moses_param_pattern.findall(parameters_string))
mosesinifile = 'mertfiles/moses.ini'
print (parse_parameters(read_params_from_moses_ini(mosesinifile)))
得到:
{'UnknownWordPenalty0': [1.0], 'PhrasePenalty0': [0.2], 'WordPenalty0': [-1.0], 'Distortion0': [0.3], 'LexicalReordering0': [0.3, 0.3, 0.3, 0.3, 0.3, 0.3], 'TranslationModel0': [0.2, 0.2, 0.2, 0.2], 'LM0': [0.5]}
当前的解决方案涉及从配置文件中读取一些疯狂的反转线,然后是相当复杂的正则表达式读取以获取参数。
是否有更简单或更少的hacky / verbose方式来读取文件并实现所需的参数字典输出?
是否可以更改configparser以便它读取moses配置文件?这很难,因为它有一些错误的部分,实际上是参数,例如[distortion-limit]
并且6
没有关键值。在经过验证的configparse-able文件中,它可能是distortion-limit = 6
。
注意:本机python configparser
无法处理moses.ini
配置文件。来自How to read and write INI file with Python3?的答案无效。
答案 0 :(得分:1)
你可以这样做。
x="""#########################
### MOSES CONFIG FILE ###
#########################
# input factors
[input-factors]
0
# mapping steps
[mapping]
0 T 0
[distortion-limit]
6
# feature functions
[feature]
UnknownWordPenalty
WordPenalty
PhrasePenalty
PhraseDictionaryMemory name=TranslationModel0 num-features=4 path=/home /gillin/jojomert/phrase-jojo/work.src-ref/training/model/phrase-table.gz input-factor=0 output-factor=0
LexicalReordering name=LexicalReordering0 num-features=6 type=wbe-msd-bidirectional-fe-allff input-factor=0 output-factor=0 path=/home/gillin/jojomert/phrase-jojo/work.src-ref/training/model/reordering-table.wbe-msd-bidirectional-fe.gz
Distortion
KENLM lazyken=0 name=LM0 factor=0 path=/home/gillin/jojomert/ru.kenlm order=5
# dense weights for feature functions
[weight]
UnknownWordPenalty0= 1
WordPenalty0= -1
PhrasePenalty0= 0.2
TranslationModel0= 0.2 0.2 0.2 0.2
LexicalReordering0= 0.3 0.3 0.3 0.3 0.3 0.3
Distortion0= 0.3
LM0= 0.5"""
print [(i,j.split()) for i,j in re.findall(r"([^\s=]+)=\s*([\d.\s]+(?<!\s))",re.findall(r"\[weight\]([\s\S]*?)(?:\n\[[^\]]*\]|$)",x)[0])]
输出:[('UnknownWordPenalty0', ['1']), ('PhrasePenalty0', ['0.2']), ('TranslationModel0', ['0.2', '0.2', '0.2', '0.2']), ('LexicalReordering0', ['0.3', '0.3', '0.3', '0.3', '0.3', '0.3']), ('Distortion0', ['0.3']), ('LM0', ['0.5'])]
`
答案 1 :(得分:1)
这是另一个基于正则表达式的简短解决方案,它返回与输出类似的值的字典:
import re
from collections import defaultdict
dct = {}
str="MOSES_INI_FILE_CONTENTS"
#get [weight] section
match_weight = re.search(r"\[weight][^\n]*(?:\n(?!$|\n)[^\n]*)*", str) # Regex is identical to "(?s)\[weight].*?(?:$|\n\n)"
if match_weight:
weight = match_weight.group() # get the [weight] text
dct = dict([(x[0], [float(x) for x in x[1].split(" ")]) for x in re.findall(r"(\w+)\s*=\s*(.*)\s*", weight)])
print dct
请参阅IDEONE demo
结果字典内容:
{'UnknownWordPenalty0': [1.0], 'LexicalReordering0': [0.3, 0.3, 0.3, 0.3, 0.3, 0.3], 'LM0': [0.5], 'PhrasePenalty0': [0.2], 'TranslationModel0': [0.2, 0.2, 0.2, 0.2], 'Distortion0': [0.3], 'WordPenalty0': [-1.0]}
逻辑:
[weight]
块。它可以使用r"\[weight][^\n]*(?:\n(?!$|\n)[^\n]*)*"
字面上匹配[weight]
的正则表达式完成,然后它会匹配每个字符任意次,直到双\n
符号(正则表达式使用展开循环技术和跨越几行的较长文本是好的。相同的基于懒惰点的正则表达式是[r"(?s)\[weight].*?(?:$|\n\n)"
],但效率不高(第一个正则表达式为62步,第二个正则表达式为528,以便在当前MOSES.ini文件中找到匹配项),但绝对是更具可读性。re.findall(r"(\w+)\s*=\s*(.*)\s*", weight)
方法以收集所有键值对。使用的正则表达式是一个简单的(\w+)\s*=\s*(.*)\s*
匹配,并在组1中捕获一个或多个字母数字符号((\w+)
),后跟任意数量的空格=
,再次任意数量的空格({{ 1}}),然后匹配并捕获到第2组中的任何符号,但直到字符串末尾的换行符。随后的sapces尾随换行符将使用最终\s*=\s*
。答案 2 :(得分:1)
如果没有正则表达式,你可以这样做:
flag = False
result = dict()
with open('moses.ini', 'rb') as fh:
for line in fh:
if flag:
parts = line.rstrip().split('= ')
if len(parts) == 2:
result[parts[0]] = [float(x) for x in parts[1].split()]
else:
break
elif line.startswith('[weight]'):
flag = True
print(result)
在循环中逐行读取文件,当到达[weight]
时,标志设置为True
并且为所有下一行提取键/值(s),直到空白行或文件的结尾。
这样,只有当前行被加载到内存中,一旦到达[weight]
块的末尾,程序就会停止读取文件。
使用itertools
的其他方式:
from itertools import *
result = dict()
with open('moses.ini', 'rb') as fh:
a = dropwhile(lambda x: not(x.startswith('[weight]')), fh)
a.next()
for k,v in takewhile(lambda x: len(x)==2, [y.rstrip().split('= ') for y in a]):
result[k] = [float(x) for x in v.split()]
print(result)