perl regex multi-line substitution weirdness

[acutchin]

I am trying to do a simple multi-line substitution in perl, but encountering intractable weirdness. How in the world do I replace BOTH of the lines with "FOOBAR"? The file does not appear to have any windows/dos-style line breaks or CR characters.

# ls -l fb
-rw------- 1 aaronc devel 8 Nov 3 11:28 fb
# cat fb
foo
bar
# cat fb | perl -nle '$_ =~ s/foo/FOO/; print $_'
FOO
bar
# cat fb | perl -nle '$_ =~ s/foo\nbar/FOOBAR/m; print $_'
foo
bar
# cat fb | perl -nle '$_ =~ s/foo$bar/FOOBAR/m; print $_'
FOOBAR
bar
# cat fb | perl -nle '$_ =~ s/^foo$bar$/FOOBAR/m; print $_'
FOOBAR
bar
# cat fb | perl -nle '$_ =~ s/^foo$^bar$/FOOBAR/m; print $_'
foo
bar
# cat fb | perl -nle '$_ =~ s/^foo\nbar$/FOOBAR/m; print $_'
foo
bar
# cat fb | perl -nle '$_ =~ s/^foo\nbar$/FOOBAR/mg; print $_'
foo
bar
# cat fb | perl -nle '$_ =~ s/foo\nbar/FOOBAR/mg; print $_'
foo
bar
# cat fb | perl -nle '$_ =~ s/^foo$\n^bar$/FOOBAR/m; print $_'
foo
bar
# cat fb | perl -nle '$_ =~ s/^foo$\n^bar$/FOOBAR/mg; print $_'
foo
bar

-- Aaron

 

[MillerH]

I suggest that you ask specifically what you're trying to do. Here is a regex that will do what you ask, but it's probably catching a lot more than what you want. Ask a specific problem, or just continue reading up on regex's. Good luck.

cat fb | perl -nle 's/^\w+/FOOBAR/g; print'

 

[acutchin]

Thank you MillerH. Maybe this is more clear:

# cat file
one
two
three
four

after parsing, i want the file to read:

# cat file
one
FOOBAR
four

how do i do that?

-- Aaron

 

[MillerH]

Your problem arises in the use of the -n parameter. This makes your code synonymous with the following:

while (<>) {
    # Substitution code here.
    s/^\w+/FOOBAR/g;
    # Print code here
    print;
}

That code process through each line of the file one at a time, instead of as one big chuck. Because of this, the regular expression never sees beyond each individual line.

I never run perl from the command line like you're doing, so I don't yet know the proper way to get this to work. Try this example to confirm what I'm telling you though.

cat fb | perl -nle "s/./FOOBAR/; print"

I originally believed that the regex would only run once, and would replace the first character of the file with FOOBAR. Instead it ran 4 times and replaced the first character on every line.

FOOBARne
FOOBARwo
FOOBARhree
FOOBARour

There should be a way to get the regex to parse the entire file in one pass, but I don't know what it is yet. I'll look into it more this afternoon, but maybe someone else already has experience doing this type of operation.