我想在一个较大的字符串中找到子字符串的第一个索引。我只希望它匹配整个单词，我希望它不区分大小写，但我希望它把 CamelCase 视为单独的单词。
下面的代码可以解决问题，但速度很慢。我想加快速度。有什么建议么？我正在尝试一些正则表达式的东西，但找不到可以处理所有边缘情况的东西。
def word_start_index(text, seek_word):
start_index = 0
curr_word = ""
def case_change():
return curr_word and ch.isupper() and curr_word[-1].islower()
def is_match():
return curr_word.lower() == seek_word.lower()
for i, ch in enumerate(text):
if case_change() or not ch.isalnum():
if is_match():
return start_index
curr_word = ""
start_index = None
if ch.isalnum():
if start_index is None:
start_index = i
curr_word += ch
if is_match():
return start_index
if __name__ == "__main__":
# 01234567890123456789012345
test_text = "a_foobar_FooBar baz golf_CART"
test_words = ["a", "foo", "bar", "baz", "golf", "cart", "fred"]
for word in test_words:
match_start = word_start_index(test_text, word)
print match_start, word
输出:
0 a
9 foo
12 bar
16 baz
20 golf
25 cart
None fred
最佳答案
word_emitter(如下)接受一个文本字符串并在找到时产生小写的“单词”，一次一个(连同它们的位置)。
它将所有下划线替换为空格。然后它将文本拆分为一个列表。例如，
"a_foobar_FooBar baz golf_CART Foo"
成为
['a', 'foobar', 'FooBar', 'baz', 'golf', 'CART', 'Foo']
当然，您还希望将驼峰命名法单词视为单独的单词。
因此，对于上面列表中的每一部分，我们使用正则表达式模式 '(.*[a-z])(?=[A-Z])'
拆分驼峰词。此正则表达式使用 re 模块的前瞻运算符 (?=...)。
也许这是整个事情中最棘手的部分。
word_emitter 然后一次生成一个单词及其相关位置。
一旦有了将文本拆分为“单词”的函数，剩下的就很简单了。
我还切换了循环的顺序，因此您只循环一次 test_text。如果 test_text 与 test_words 相比很长，这将加快速度。
import re
import string
import itertools
nonspace=re.compile('(\S+)')
table = string.maketrans(
'_.,!?;:"(){}@#$%^&*-+='+"'",
' ',
)
def piece_emitter(text):
# This generator splits text into 2-tuples of (positions,pieces).
# Given "a_foobar_FooBar" it returns
# ((0,'a'),
# (2,'foobar'),
# (9,'FooBar'),
# )
pos=0
it=itertools.groupby(text,lambda w: w.isspace())
for k,g in it:
w=''.join(g)
w=w.translate(table)
it2=itertools.groupby(w,lambda w: w.isspace())
for isspace,g2 in it2:
word=''.join(g2)
if not isspace:
yield pos,word
pos+=len(word)
def camel_splitter(word):
# Given a word like 'FooBar', this generator yields
# 'Foo', then 'Bar'.
it=itertools.groupby(word,lambda w: w.isupper())
for k,g in it:
w=''.join(g)
if len(w)==1:
try:
k1,g1=next(it)
w+=''.join(g1)
except StopIteration:
pass
yield w
def word_emitter(piece):
# Given 'getFooBar', this generator yields in turn the elements of the sequence
# ((0,'get'),
# (0,'getFoo'),
# (0,'getFooBar'),
# (3,'Foo'),
# (3,'FooBar'),
# (6,'Bar'),
# )
# In each 2-tuple, the number is the starting position of the string,
# followed by the fragment of camelCase word generated by camel_splitter.
words=list(camel_splitter(piece))
num_words=len(words)
for i in range(0,num_words+1):
prefix=''.join(words[:i])
for step in range(1,num_words-i+1):
word=''.join(words[i:i+step])
yield len(prefix),word
def camel_search(text,words):
words=dict.fromkeys(words,False)
for pos,piece in piece_emitter(text):
if not all(words[test_word] for test_word in words):
for subpos,word in word_emitter(piece):
for test_word in words:
if not words[test_word] and word.lower() == test_word.lower():
yield pos+subpos,word
words[test_word]=True
break
else:
break
for word in words:
if not words[word]:
yield None,word
if __name__ == "__main__":
# 01234567890123456789012345
test_text = "a_foobar_FooBar baz golf_CART"
test_words = ["a", "foo", "bar", "baz", "golf", "cart", "fred"]
for pos,word in camel_search(test_text,test_words):
print pos,word.lower()
以下是我用来检查程序的单元测试:
import unittest
import sys
import camel
import itertools
class Test(unittest.TestCase):
def check(self,result,answer):
for r,a in itertools.izip_longest(result,answer):
if r!=a:
print('%s != %s'%(r,a))
self.assertTrue(r==a)
def test_piece_emitter(self):
tests=(("a_foobar_FooBar baz? golf_CART Foo 'food' getFooBaz",
((0,'a'),
(2,'foobar'),
(9,'FooBar'),
(16,'baz'),
(21,'golf'),
(26,'CART'),
(31,'Foo'),
(36,'food'),
(42,'getFooBaz'),
)
),
)
for text,answer in tests:
result=list(camel.piece_emitter(text))
print(result)
self.check(result,answer)
def test_camel_splitter(self):
tests=(('getFooBar',('get','Foo','Bar')),
('getFOObar',('get','FOO','bar')),
('Foo',('Foo',)),
('getFoo',('get','Foo')),
('foobar',('foobar',)),
('fooBar',('foo','Bar')),
('FooBar',('Foo','Bar')),
('a',('a',)),
('fooB',('foo','B')),
('FooB',('Foo','B')),
('FOOb',('FOO','b')),
)
for word,answer in tests:
result=camel.camel_splitter(word)
self.check(result,answer)
def test_word_emitter(self):
tests=(("a",
((0,'a'),) ),
('getFooBar',
((0,'get'),
(0,'getFoo'),
(0,'getFooBar'),
(3,'Foo'),
(3,'FooBar'),
(6,'Bar'),
)
)
)
for text,answer in tests:
result=list(camel.word_emitter(text))
print(result)
self.check(result,answer)
def test_camel_search(self):
tests=(("a_foobar_FooBar baz? golf_CART Foo 'food' getFooBaz",
("a", "foo", "bar", "baz", "golf", "cart", "fred", "food",
'FooBaz'),
((0,'a'),
(9,'Foo'),
(12,'Bar'),
(16,'baz'),
(21,'golf'),
(26,'CART'),
(36,'food'),
(45,'FooBaz'),
(None,'fred')
)
),
("\"Foo\"",('Foo',),((1,'Foo'),)),
("getFooBar",('FooBar',),((3,'FooBar'),)),
)
for text,search_words,answer in tests:
result=list(camel.camel_search(text,search_words))
print(result)
self.check(result,answer)
if __name__ == '__main__':
unittest.main(argv = unittest.sys.argv + ['--verbose'])
关于python - 棘手的字符串匹配，我们在Stack Overflow上找到一个类似的问题： https://stackoverflow.com/questions/2127188/