bash scripting: batch file renamer

Bit by bit, I have replaced repetitive tasks completed using downloaded programs with scripts that do the same thing more efficiently, without the residual registry and file junk many programs leave, and with the flexibility to allow me to modify the script to suite specific needs in the future. This was the case with batch renaming files. I often had data from experiments or other projects that was in a regular, but undesirable, naming formats (thanks in part to industrial software designers and their penchant for meaningless output filenames). Changing either involved manually hitting F2 and changing the name (yuck!), downloading a program (lets avoid doing that...), or writing a cheap sed function.

I have finally decided to extend the use of sed and mv to a fully functional script for batch renaming files in a directory. I'll briefly comment on the code. A batch version (for Windows) may come in the future. Else, just download cygwin to use the bash shell environment in Windows, you'll be glad for it.

For those who want to dive right in, the script can be downloaded at:

batch renamer

The script is fairly basic, but I wanted it to be slightly flexible without getting bogged down in needless exception handling (for another time). To start, the script asks for the directory and changes to that if needed. A function to ask for user input is defined for use later. The internal field separator is changed to allow parsing of files and directories with spaces.

Then it ask for the number of different regular expressions to run on the files in the directory. This is useful for two reasons. It allows one to change multiple different batches of files at once (assuming they have some difference in how they are named) or to do multiple different search and replace operations on the same file. Further, the method I use allows several steps of replacement. For example, say you want to remove all spaces, normally the end sed command (inside some file-replacing loop for function) would be sed 's/ /_/g', but I just ask for you to hit spacebar and underscore and you're on your way. In addition, the script confirms the regexp pattern is the one you want by displaying an example use on one file in the directory.

Further, the regular expressions can be chained, e.g. sed 's/ /_/g;s/foo/bar/g'. The semicolon allows multiple regular expressions to be used with sed without needing to pipe the commands. I accomplish this automatically from the multiple regular expressions and replacements using a simple loop.

The rest of the script simply shows the user modified filenames and then renames the them after confirming those changes are desired. To save a bit of time and unnecessary warnings, I check that the old and new filename are actually different before proceeding.

There are a couple things I would like to improve, such as ability to select specific files or choose the type (by extension) of file to change. The later will just involve integration with grep or find, but the former might be a little trickier if I want to stay within the confines of bash. Might play around with select, dialog, or just a loop asking for files from a list. Probably better if files are just organized and segregated properly...

Alright, thats all. The code was written in a dash and the post should be quick as well.