I have a little script that kind of publishes my dev php files to live versions. All my dev files look like "index.dev.php" & the live version looks like "index.php". The script strips out the .dev from references inside the file (includes, hrefs, forms) when it publishes & does this with a sed.

The problem is, it's replacing a couple extra ".dev"'s in the file that I'd rather it not, so I'm looking for a regex that replaces:

[a-z].dev.php with [a-z].php

...meaning that it only strips .dev if it is preceded by a letter. The problem I have is that it's also replacing that preceding letter. How do you carry over a character from the find to the replace? My current regex looks like:

sed 's/[a-z]\.dev\.php"/\.php"/g'

but that makes the line:

include "foo.dev.php";

become

include "fo.php";

when I need it to be

include "foo.php";

ideas?

I'm not 100% positive, but try the following:
Not sure if you want the "'s either.

You need to use a placeholder:ta0kira

PS I find sed a lot easier to use with -r. That lets you use () without \, plus a few other things.

fantastic. Thanks!