Bash to sort ogg files by internal artist, album info

Shay · 05-08-2007, 08:29 PM

I've got a pile of ogg files in a folder. I would like to write a bash script to separate the files into artist and album folders. Each file contains strings that look like:
artist=some artist^
album=some album^

I thought there might be a way to use bash and perl to discover the album names, create folders, and move the files into those folders.

I think if I get past the first step, I can get the rest.
grep 'album=' *ogg doesn't print the line containing album=some album^
Instead, I get this:
Binary file song1.ogg matches
Binary file song2.ogg matches
Binary file song3.ogg matches

That would be useful enough to me if I could just generate a list of all of the album names before I start.

Any help appreciated.

Sonneteer · 05-09-2007, 12:24 AM

Instead of greping the files directly, use ogginfo on the files, and pipe the output of that to grep.

Code:

ogginfo *.ogg | grep album

Something like this would be able to give you a list of all the albums:

Code:

ogginfo *.ogg | grep album | uniq | cut -d = -f 2

wjevans_7d1@yahoo.co · 05-09-2007, 12:30 AM

Go with soneteer's solution. But for further background, the following will apply to other files which have a mixture of ASCII characters and binary data. (I didn't know about ogginfo, and typed the following before I saw his response.)

ogg files contain not only printable ASCII characters, but also binary data. grep usually works with printable ASCII data, line by line. With binary data, there is no such thing as a "line" in the ASCII sense, and what would be the line termination character (LF) in ASCII data can appear anywhere in plain binary data, and that byte means whatever the binary usage would dictate.

So grep goes out of its way to detect a binary file, and just let you know whether the specified string is within that file.

You have two ways to fix this: the wrong way, and the right way.

The wrong way, with which you may wish to experiment if you have nothing better to do, is this:

Code:

grep -a 'album=' *ogg

That will ignore the possibility that the file might contain binary data (which is the case with ogg files). What you really want to do, though, is consider just the ASCII parts of each file. You can pipe each file through the strings command, and send the output through grep. Try this:

Code:

for xxx in *ogg
do
  strings $xxx | grep 'album=' | sed -e "s/^/$xxx:/"
done

Note that the first occurrence of xxx is not preceded by a dollar sign. Also notice the double quotes around the sed parameter; using single quotes would prevent the $xxx from being interpreted properly. With double quotes, the $xxx will get changed into the name of the individual ogg file; with single quotes, it will remain $xxx, and that's what you'll see at the beginning of each line of output.

For more information:

Code:

man grep
man strings
man sed

Hope this helps.