如果使用方括号作为定界符，则Perl regex不起作用。 为什么？

Question

I'm very experienced with Perl and regex. However, this is making me crazy, I just can't find an answer to it and I cannot see a reason for it either. Please look at the following code:

my $str = 'Hello[world]';

say $str =~ m/\w+\[.*?\]/ ? 'Yes' : 'No';
say $str =~ m[\w+\[.*?\]] ? 'Yes' : 'No';
say $str =~ m(\w+\[.*?\]) ? 'Yes' : 'No';

The output of this is:

Yes
No
Yes

As you can see, the only thing I'm changing is the regex delimiter, and the expression is not working as I would expect when the delimiter is square brackets.

Can someone please explain why the second one is not matching?

Thanks in advance,

Francisco

Answer 1

The B::Deparse module comes to your rescue:

$ perl -MO=Deparse foo.pl
my $str = 'Hello[world]';
say $str =~ /\w+\[.*?\]/u ? 'Yes' : 'No';
say $str =~ /\w+[.*?]/u ? 'Yes' : 'No';
say $str =~ /\w+\[.*?\]/u ? 'Yes' : 'No';
foo.pl syntax OK

As you can see, the escaping of [ ] in your regex meant that perl now interpreted them as meta characters, and not delimiters. You need two levels of escape. Which I am not sure is even possible to do, since \\\\ will be interpreted as literal backslash.

To be extra clear: In a normal regex, the brackets [] have a meta character status. So in order to match them literally, they need to be escaped. When using them as delimiters, you add another meta character status to them: They are also delimiters. So both meta character statuses need to be escaped.

This will work as intended:

say $str =~ m[\w+\Q\[\E.*?\Q\]\E] ? 'Yes' : 'No';

Of course, the lesson here is to choose your delimiters wisely.

Answer 2

在执行匹配之前，请尝试用>替换[和<和] > （或其他类似的替换符号）。