RegEx是否可以匹配除标点符号之外的所有非单词？

Question

For sentences like:

sent = "This i$s a s[[]ample sentence.\nAnd another <<one>>.
        \nMoreover, it is 'filtered'!"

I would like to get:

"This is a sample sentence. And another one. Moreover, it is filtered."

Thus, I thought using re.sub should be the way to go. However, RegEx doesn't work as expected (like it pretty much always does^^).

My idea was to use \\W to match every non-word and then exclude [.,;!?] to keep the punctuation. The last RegEx I've tried was:

re.sub(r"(\W[^\.\,\;\?\!])", "", sent)

Unfortunately, [^\\.\\,\\;\\?\\!] does match for anything that does not contain an entry of [.,;!?] , instead of simply saying 'do not match these characters literally'.

How can I exclude these characters from match?

Answer 1

The \\W needs to be integrated into the negated character class. \\W is the same as [^\\w] , so you'll end up with [^\\w.,;!?] . You should repeat this character class, to match contiguous occurences in a single step - [^\\w.,;!?]+ .

It seems you also want to keep spaces, so you should add them to your character class.

Reeding deeper into your question, you also want to replace newlines with a space and ! with . . This makes it a multiple step solution. First filter out anything unwanted [^\\w.,;!? \\n]+ [^\\w.,;!? \\n]+ , in a next step replace \\n with and ! with . .