Reg Expression Question

windisch · 10-03-2005, 02:04 PM

I am trying to modify a reg expression built by a poster here.

Code:

msg=$(echo $line|sed 's/.*\] \(.*\)/\1/')

I am having trouble getting this to recognize files that have parentheses in the file name. Could someone help me out with that?

Dark_Helmet · 10-03-2005, 02:18 PM

It would be helpful if you could provide some examples of the text being processed and what you'd like the result to be.

As far as I can tell, the expression you give above doesn't look for a parenthesis at all. It looks for the last occurrence of a literal close square bracket ( ] ), ignores the space immediately after the bracket, and returns everything else in the line.

EDIT:
BTW, happy birthday!

windisch · 10-03-2005, 04:48 PM

Thanks for noticing my birthday.

Anyway here is an example of the text

Here is what the script is monitoring:

Code:

<10/03/05@17:38:04> [DECODE] Opened Majora's Mask - 42 - Magic Hag's House.mp3
<10/03/05@17:38:26> [MAIN] Title Updated
<10/03/05@17:39:34> [DECODE] Opened 32 - Daughter of Madain Sari.mp3
<10/03/05@17:39:57> [MAIN] Title Updated
<10/03/05@16:57:34> [DECODE] Opened 09 - The Sons (Feat. Brett Reilly) - Too Much Of A Good Thing.mp3
<10/03/05@16:57:56> [MAIN] Title Updated
<10/03/05@17:02:46> [DECODE] Opened 08 - Liberi Fatali (FF VIII).mp3

Here is an example of file names:

Code:

/video3/MythMusic/U2/U2 - The Best of 1980-1990/210 - Luminous Times (Hold on to Love).mp3
/video3/MythMusic/U2/U2 - The Best of 1980-1990/211 - Hallelujah Here She Comes.mp3
/video3/MythMusic/U2/U2 - The Best of 1980-1990/212 - Silver And Gold.mp3
/video3/MythMusic/Nickelback/Silver Side Up/01 Never Again.mp3
/video3/MythMusic/Nickelback/Silver Side Up/02 How You Remind Me.mp3

Most of the files do not have () or [] in them, but there are some that do, and throw off my reg expression.

Dark_Helmet · 10-03-2005, 11:20 PM

Ok, I may not have a complete grasp on the problem, because I'm not encountering a problem with filenames having parentheses in them. So double check what I'm doing here. Here we go...

The expression, as you have it above, modifies the monitored data you have to this:

Code:

Opened Majora's Mask - 42 - Magic Hag's House.mp3
Title Updated
Opened 32 - Daughter of Madain Sari.mp3
Title Updated
Opened 09 - The Sons (Feat. Brett Reilly) - Too Much Of A Good Thing.mp3
Title Updated
Opened 08 - Liberi Fatali (FF VIII).mp3

That's the result I get from running the command on my system. So that's what I'm going to work with. If that's not the output you want, then we'll need to tinker with the expression some more.

I noticed the data alternates between "[DECODE]" and "[MAIN]" and those can serve as markers to prevent the "greedy" nature of an RE from getting too much. The modified version is:

Code:

msg=$( echo $line | sed 's%.*\[\(\(DECODE\)\|\(MAIN\)\)\] \(.*\)%\4%'

A couple things, I changed the format from "s///" to "s%%%". It's a little dizzying (to me) to see forward slashes and backslashes mingling. I'm not goinf to go so far as to say the new expression is clearer because of it, but oh well. Also, I added spaces to to suit my style.

So what does this beast of an expression do? It ignores anything until it comes to an open-close set of square brackets that contain either "DECODE" or "MAIN". It ignores the space immediately after the close square bracket, and returns everything following that space to the end of the line. So, on a fictitious data set of:

Code:

<10/03/05@17:38:04> [DECODE] Opened Majora's Mask - 42 - Magic Hag's House.mp3
<10/03/05@17:38:26> [MAIN] Title Updated
<10/03/05@17:39:34> [DECODE] Opened 32 [disc 1]- Daughter of Madain Sari.mp3
<10/03/05@17:39:57> [MAIN] Title Updated
<10/03/05@16:57:34> [DECODE] Opened 09 [disc 20] - The Sons (Feat. Brett Reilly) - Too Much Of A Good Thing.mp3
<10/03/05@16:57:56> [MAIN] Title Updated
<10/03/05@17:02:46> [DECODE] Opened 08 [disco [pt2]] - Liberi Fatali (FF VIII).mp3

The result would be:

Code:

Opened Majora's Mask - 42 - Magic Hag's House.mp3
Title Updated
Opened 32 [disc 1]- Daughter of Madain Sari.mp3
Title Updated
Opened 09 [disc 20] - The Sons (Feat. Brett Reilly) - Too Much Of A Good Thing.mp3
Title Updated
Opened 08 [disco [pt2]] - Liberi Fatali (FF VIII).mp3

So we're back at the original output except the expression doesn't get hung up on square brackets later in the filename. Like I said, if that's not exactly what you're looking for, we can refine the expression. It'll take some desired intput->output examples to flesh out the expression.

windisch · 10-04-2005, 06:37 AM

You're right, I changed my script to look and see what the reg expression, and it doesn't have a problem with (), it does with []. I assumed that it was a problem with parenthese because when I request a song, it loops on songs that have parenthese. My problem must be with the awk command I use to remove the song from my playlist. I would appreciate your help to get this to recognize brackets. The only thing I need to change in your code, is that I don't need to grab the word Opened. My goal is to grab the entire file name after Opened.

Dark_Helmet · 10-04-2005, 10:56 AM

Ok, so you don't even want the "Title updated" business either, right? Ok, I was thinking this script was pulling off the date & time info and handng the stripped text to some other utility. Actually, if it's going to awk, then it is, so I was right

It's just not ripping off enough... Easily remedied...

Try this on for size:

Code:

msg=$( echo $line | sed 's%[^]]*\] \(\(Title Updated\)\|\(Opened \)\)\(.*\)%\4%'

Since it's chenged a bit from the previous version (same basic premise though), I'll give a quick run-down:

Match any non-] character until the first literal ] is reached => [^]]*\]
Ignore the space after ]
Match explicit text: "Title Updated" or "Opened " (notice the space with Opened) => \(\(Title Updated\)\|\(Opened \)\)
Match, and save, any text following the last either-or match => \(.*\)
Replace everything with the saved text

So this RE should just give you the filename, or a blank empty line (in the case of a "Title Updated" entry). I gave it a shot with two different input lines (i.e. not thorough testing), and it worked as expected. So give it a test drive, and let me know if it needs anymore tweaking.

windisch · 10-04-2005, 11:08 AM

Thanks, I'll see how it works.