因此，我被告知，连接字符串是一个非常昂贵的过程，我们最好将其附加到列表中然后再将它们加入，而这整个都是在Internet上进行的。毕竟，不创建临时字符串总比这样做更好。
但是，当我在Python 3.7上进行测试时，似乎相反。
我在这里有两个功能。 editText1应该比def editText0(text, slices):
# text is the string to be edited.
# slices should be in the form of {(start0, end0): chars0, (start1, end1): chars1 ...}.
# text is going to be split into lices by '\n'
# slicing is going to be applied to every line individually
# the new text is going to be in the form of text[n][start0:end0] + chars0 + text[n + 1][start1:end1] + chars1 + ...
text = text.split('\n')[:-1] # get lines. We don't need the last item in the list since it is empty, hence [:-1]
finalText = '' # fully processed text
for line in text:
# for every line
finalLine = '' # final state of the current line
for ((start, end), chars) in slices.items():
finalLine += line[start : end] + chars
finalText += finalLine
finalText += '\n'
return finalText
def editText1(text, slices):
text = text.split('\n')[:-1] # get lines. We don't need the last item in the list since it is empty, hence [:-1]
finalText = "\n".join(''.join(line[start : end] + chars for ((start, end), chars) in slices.items()) for line in text)
return finalText
text = 'abcdefghijklmnopqrstuvwxyz\n' * 4 # 4 lines, each 26 characters and a '\n'
slices = {(0, 11): "-", (0, 26): "~", (20, 24): ""}
print(timeit.timeit("editText0(text, slices)", "from __main__ import editText0, text, slices"))
print(timeit.timeit("editText1(text, slices)", "from __main__ import editText1, text, slices"))
我给他们计时的时候：
{{1}}
输出为5.162253748000012和7.014296320999847。基于串联的功能几乎总是快2整秒。
这是怎么回事？互联网上有关此问题的大多数讨论似乎是从很久以前开始的，是否是因为Python本身进行了某种优化？在2018年串联字符串的最佳方法是什么？