RE question; how to insert a digit between const[a-z]*3 && inc[0-9]*3?

[taint]

Greetings, I've been maintaining some 9k html records for a few years now. The names of the files begin with the same 3 alpha characters, followed by 4, or less digits (0-9). I've been maintaining them via quick one-off sed scripts. But recently I decided I needed to re-numerate them, as it would be ultimately be more efficient. I also decided (given I /really/ love perl) that it would make more sense to maintain a set of Perl scripts to better maintain these documents, then it hit me -- I have no idea how to take on the task of re-numerating the references within these documents.
PROBLEM:

changing all references to:
abc[0-9][0-9][0-9].html
into
abc0[0-9][0-9][0-9].html

the same goes for abc[0-9][0-9].html
in short; I need all of the files to be 4 digits long -- not 1,2, & 3, as they are now.
I'm fair with search & replace, add, and delete. But I /suck/ at hold patterns -- which I think I need here. I tried to match against:

 /abc1[0-9][0-9].html

with the intent of matching within the 100-199.html documents. But all my efforts were an epic fail.
Pointers, solutions would be greatly appreciated.

Thank you for all your time, and consideration.

[tye]

s{(?<=[a-z])([0-9]{1,3})(?=[.])}{
    sprintf "%04d", $1
}ge;

...with whatever additions are needed near the start and end of that regex in order to only match the parts that you want to.

[hdb]

Can it not be as simple as

s{abc(\d+)\.html}{sprintf("abc%04d.html",$1)}ge;

if all the filenames are of the form abc\d+.html or is there a catch?

[tye]

I doubt "abc" was as literal as you interpreted it to be.

[taint]

Greetings tye.
Thanks for taking the time to respond.


s{(?<=[a-z])([0-9]{1,3})(?=[.])}{
    sprintf "%04d", $1
}ge;

Looks brilliant in it's simplicity. But my head is still in sed(1) scripting mode. So I fear I am not yet following it. :(

I'm attempting to sort it out now. :)

Thanks again!

[kcott]

G'day taint,

Here's another way to do it:

$ perl -Mstrict -Mwarnings -E '
    my @files = qw{abc1.html abc12.html abc123.html abc1234.html};
    say for map {
        s/^(abc)(\d{1,3})(\.html)$/$1 . "0" x (4 - length $2) . $2 . $
+3/e; $_
    } @files;
'
abc0001.html
abc0012.html
abc0123.html
abc1234.html

Update: "... the files begin with the same 3 alpha characters ..." so \w{3} is far too generic: s/\w{3}/abc/

[taint]

Greetings Ken, and thanks for the reply.
This looks like a complete solution!
Given that all of the files end in .html, and begin with abc
I should be able to simply read in the entire directory.

opendir(DIR, $dirname) or die "can't opendir $dirname: $!";
while (defined($file = readdir(DIR))) {
# use your suggestion on "$dirname/$file"
}
closedir(DIR);

I'm still sorting it all out. But wanted to take the time to thank you
for your help.

Thanks again!

--chris

[kcott]

Given that context, you might want to try something like this technique (untested):

while (defined($file = readdir(DIR))) {
    next unless $file =~ /^(abc)(\d{1,3})(\.html)$/;
    rename "$dirname/$file" "$dirname/" . $1 . "0" x (4 - length $2) .
+ $2 . $3;
}

I'd also recommend that you try this with a print to ensure you're in the right directory and targetting the correct files before making thousands of changes; then change to rename. Some error checking/handling might also be useful.

[MidLifeXis]

Depending on your server software, you could also configure your server to generate a 301 response for the old document names. I have done this in the past on apache with mod_rewrite. It could also be done with a mod_perl handler or even a dispatcher if you drive everything through a dispatch script.