Shell Programming and Scripting

I desperately need to write a script that go into text documents that list the location of files in small caps and have it convert the directories and subdirectories that lead to the file as Title Caps while leaving the file itself in small caps... to illustrate what I mean:

I need to turn txt file upon txt files containing:
-----------------------------------------------------------------
\music\artist\album\title-in-small-caps.mp3
......
\music\arc-en-ciel\jam sessions\12 alec alec boum!.mp3
\music\babel soundtrack\babel original soundtrack\10 bibo no aozora_endless flight_bab.mp3
\music\benjamin zephaniah\belly of de beast\wake up.mp3
-----------------------------------------------

into:
------------------------------------------------
\Music\Artist\Album\title-in-small-caps.mp3
...........
\Music\Arc-En-Ciel\Jam Sessions\12 alec alec boum!.mp3
\Music\Babel Soundtrack\Babel Original Soundtrack\10 bibo no aozora_endless flight_bab.mp3
\Music\Benjamin Zephaniah\Belly Of De Beast\wake up.mp3
---------------------------

I can think up the algorithm easily but I have zero knowledge of the synthax need to pull this off:

Algoritm:
for each line, count the number of "/" characters there are, if a string is between two "/" characters, TitleCase it. If the string is between a "/" character and a ".mp3" string, do not change it and move on to next line.

how would I go about doing this in a script or pearl?

This thread discusses capitalization but it will capitalize everything... [see trackback link]


Does anyone have any ideas?

What Operating System and version are you running?
What Shell you you use (e.g. bash, ksh, real Bourne, whatever)?
Do you use a UK or American character set, or a character set from some other Country? It matters because in alternative character sets the range "a-z" can either include characters we don't want to change or miss out characters which we do want to change.

How many directories? This is a question about whether efficiency matters.
To count the directories:

Code:

find . -type d -print | wc -l

Do you have any existing directories which are in the desired format? This a very important question because we do not want to overwrite existing directories.


Footnote. The directory format in your examples uses the character "\" to delimit directories. This is a feature of M$ Operating Systems such as Windows, and not a feature of unix Operating Systems which use the character "/". What is the exact format of your directoy names on your disc?

@methyl brings up some very good points, which may affect how this PERL script:

Code:

$, = '\\';

while (<>) {
    @F = split /\\/;
    $f = pop @F;
    print(( map { ucfirst($_) } @F ), $f);
}

works on your system with your data.

Thank you for your help friends.

I am dealing here wih Directories of my music library. All the directories are already named in the correct format (they are Title Case). What I am acutally trying to modify is not the directories themselves but multiple m3u playlists written by a windows program where case is irrelivant in address where the directories are listed in lower case.

Usually modifying the case of text can be accomplished in any word processor but the tricky part is that although directories are in Title Case on the hard drive, the actual mp3 files they contain are in lower case. So when altering my playlists, the directories need to be listed in Title Case whereas the mp3s need to remain unchanged (in lower case). I have thousands of lines to modify so, I'm trying to see if there is a script that can do this without having me go at it one by one.

I use an american charset on my Linux. I am on Kubuntu 10.04 running KDE. I'm using Bash (the console program itself is called Konsole). I altering an m3u text file writen by an american windows based program called DoubleTwist and this is part of my effort to leave this program and migrate all my music data to SongBird. I hope to make a tutorial for other people wanting to leave this program as well.

Try something like...

Code:

$ cat a1.awk
BEGIN{FS=OFS="\\"}
{
   for(i=1;i<NF;i++)
      $i=titlecase($i)
   print $0
}

$ cat file1.txt
\music\artist\album\title-in-small-caps.mp3
......
\music\arc-en-ciel\jam sessions\12 alec alec boum!.mp3
\music\babel soundtrack\babel original soundtrack\10 bibo no aozora_endless flight_bab.mp3
\music\benjamin zephaniah\belly of de beast\wake up.mp3

$ gawk -f titlecase.awk -f a1.awk file1.txt > file2.txt

$ cat file2.txt
\Music\Artist\Album\title-in-small-caps.mp3
......
\Music\Arc-En-Ciel\Jam Sessions\12 alec alec boum!.mp3
\Music\Babel Soundtrack\Babel Original Soundtrack\10 bibo no aozora_endless flight_bab.mp3
\Music\Benjamin Zephaniah\Belly of De Beast\wake up.mp3

$

You would have to download the titlecase function from http://www.pement.org/awk/titlecase.awk.txt

Code:

$ ruby -F'\\'  -ane '$F[1..-2].map{|x| x.gsub!(/\w+/) {|y| y.capitalize} };puts $F.join("\\")' file 
\Music\Artist\Album\title-in-small-caps.mp3
\Music\Arc-En-Ciel\Jam Sessions\12 alec alec boum!.mp3
\Music\Babel Soundtrack\Babel Original Soundtrack\10 bibo no aozora_endless flight_bab.mp3
\Music\Benjamin Zephaniah\Belly Of De Beast\wake up.mp3

Thanks I'll give it a whirl...