Bash script to replace all occurrences of string in files, including filenames

Question

I want to be able to replace the occurrences of one string with another in the file contents and filenames of all files within a directory minus the ".git" directory. I have the following code, which is good for changing the filenames but doesn't replace strings inside the files. What am I doing wrong?

#!/bin/bash

# Replace occurances in files, lowercase
lower1=$(echo $1 | tr '[:upper:]' '[:lower:]')
lower2=$(echo $2 | tr '[:upper:]' '[:lower:]')
echo "Replacing $lower1 with $lower2..."
find . -type f \! -iregex '.\|./.git' -exec perl -i -pe 's/$lower1/$lower2/g' {} \;

# Replace all other occurances in files to capitalised
upper2=$(echo ${2:0:1} | tr '[:lower:]' '[:upper:]')${2:1}
echo "Replacing all $1 with $upper2..."
find . -type f \! -iregex '.\|./.git' -exec perl -i -pe 's/$1/$upper2/gi' {} \;

# Replace filenames. Use "bash -c" to pass files as arguments to mv command.
# This worked
find . -name "*$lower1*" -exec bash -c 'mv "$1" "${1/$2/$3}"' -- {} $lower1 $lower2 \;

I did try sed, but that wasn't allowing case insensitivity.

I am testing this on OSX Mountain Lion if that makes a difference, though I do want this to work on Linux as well. I'm away from my Linux machine at the moment.

Answer 1

Just use File::Find

The following code will edit an entire directory tree, so I'd be careful how I used it. Perhaps some testing first? :)

This script accepts two strings as arguments, and then replaces all occurances of $string1 with $string2. For perfect lowercase matches, it will replace with a lowercase $string2, but for all other matches it will replace with uppercase $string2.

Currently, it will just do lowercase for file renames.

use strict;
use warnings;
use autodie;

use File::Find;

die "Usage: $0 From_string To_string\n" if @ARGV != 2;

my ($from, $to) = map lc, @ARGV;

finddepth(sub {
    return if $File::Find::dir =~ /.git\b/;

    # Inplace Edit of files
    if (-f) {
        local @ARGV = $_;
        local $^I = '.bak';
        while (<>) {
            s/\Q$from/$to/g;
            s/\Q$from/\U$to/ig;
            print; 
        }
        #unlink "$_$^I";  # Uncomment if you want to delete backups
    }

    # Rename File
    my $newfile = $_;
    if ($newfile =~ s/\Q$from/$to/ig) {
        rename $_, $newfile;
    }
}, '.');

Answer 2

find . -name .git -prune -o type f -exec perl -i -pe "s/\Q$1\E/\L$2\E/i" {} +
find . -name .git -prune -o type f -name "*$lower1*" -exec mmv "*$lower1*" "#1$lower2#2" {} +

If you are doing a lot of file renames, you should check for accidental removal of intermediate results. mmv does this admirably.

+ is a useful performance boost for find.

Answer 3

Here is the solution I managed to come up with. Thanks to all that helped me come to it:

#!/bin/bash

# Replace occurances in files, lowercase
lower1=$(echo $1 | tr '[:upper:]' '[:lower:]')
lower2=$(echo $2 | tr '[:upper:]' '[:lower:]')
echo "Replacing $lower1 with $lower2..."
find . -type f \! -iregex '.\|./.git' -exec perl -i'' -pe "s/$lower1/$lower2/g" {} +

# Replace all other occurances in files to capitalised
upper2=$(echo ${2:0:1} | tr '[:lower:]' '[:upper:]')${2:1}
echo "Replacing all $1 with $upper2..."
find . -type f \! -iregex '.\|./.git' -exec perl -i'' -pe "s/$1/$upper2/gi" {} +

# Replace filenames.
mmv -r ";*$lower1*" "#2$lower2#3"