sed, grep, awk, regex -- extracting a matched substring from a file/string

Shell Programming and Scripting
1

Ok, I'm stumped and can't seem to find relevant info.
(I'm not even sure, I might have asked something similar before.):

I'm trying to use shell scripting/UNIX commands to extract URLs from a fairly large web page, with a view to ultimately wrapping this in PHP with exec() and including the URLs in a webpage that I'm trying to then generate for myself.

Here's what I have so far:

I'm catching the page with cURL:

# curl -s http://archive.wbai.org/

I'm then grepping all the lines that include the (case-insensitive) string "talkback":

# curl -s http://archive.wbai.org/ | grep -i talkback

NB: I wanted to grep for either "talk back" or "talkback", and the docs I found said that I could use "talk*back", wherein the asterisk would signify zero or more characters, but I can't seem to get this to work.

Having gotten this far, I have a number of strings like this one (which is supposed to be one single line):

<td width="5%" valign="top" bgcolor="#767676" align="center"> \
<span class=headline3>1 \
<td width="10%" valign="top" bgcolor="#EFEFEF"><span class=archivelink> \
<a href="pls.php?mp3fil=4541"><u>Play</u></a></span>� \
<span class=archivelink> \
<a href="http://archive.wbai.org/files/mp3/060222_150002talkback.MP3"> \
<u>Download</u></a>

I now want to extract just the second URL, the one with the .mp3 file.
I have tried to match

href*\.MP3

and then to somehow only get the URLs printed, but it just doesn't seem to work.

NB: I am aware that if I got awk to work right, I might no longer need to grep. For now, I'm still grepping though. Also, I heard that first grepping and then awking might hypothetically be a tiny bit quicker as allegedly, reportedly grep is (reputed to be) somewhat faster than awk. (Comments welcome.)

I am also aware that it's probably perfectly possible (and conceivably even quicker) to do all of this in PHP. I haven't tried that because (a) my PHP skills blow even harder than my scripting skills and (b) I'd really like to know how to do this kind of manipulation using "standard" UNIX shell commands. (Yes, yes, I know, a lot of people consider PHP "standard" as well, and one can even write php shell scripts, etc. etc... but PLEASE, have mercy on my soul. ;))

I'd be extremely grateful for any help on this. :slight_smile:

Edit:
PS: I should add that I want to make as little assumptions about the source page as possible, so I don't want to just extract the nth $something (say, e.g. $9) with awk, because I don't want to assume that the talkback .mp3 URL always stays in the same place.

2

How about

sed -n -e "s_.*a href=.\([^\"]*[Tt][Aa][Ll][Kk][.]*[Bb][Aa][Cc][Kk].[Mm][Pp]3\).*_\1_p"

It says, capture everything within the quotes until you encounter .mp3 or .MP3 or anything similiar. It should handle "talk back", "talkback" and is case insensitive.

3

Thanks a bunch vino. Your kung fu is strong.
As for myself, I will actually continue to read and search for a while, until I am darn sure I fully understand everything. But now I definitely know which way to go! :slight_smile: Thanks again!