Scripting Question - Strip filename from URL leaving Domain and Path

adam1mc · 03-08-2010, 11:37 AM

I'm new to linux and scipting in general. I have a text file with several thousand URL's - 1 URL per line. I need to come up with a few simple commands that can parse the domain information and path from a URL while stripping the filename and extension.

For example, if I have http://blah.com/123/abc.exe I want to strip the abc.exe and just leave http://blah.com/123/

Can anyone please help me out with this?

Thanks

jamescondron · 03-08-2010, 11:50 AM

Now personally I'd do this in python, because it is easier to follow in this example without having to learn about regular expressions (http://en.wikipedia.org/wiki/Regular_expressions).
You should also have python installed.

Code:

#!/usr/bin/env python
#-*-coding:utf-8 -*-
#
file = open( "/path/to/file", 'r' ).readlines()
for url in file:
  url_tmp = ""
  for part in url.split("/"):
    url_tmp += "%s/" % part
  # Do something here with 'url_tmp'

Simply put, we open in file and store it in an array, one line in the file per element.
Then we use split() on each url; which as it suggests splits further by the forward slash into another array (We're defining it on the fly, no need to name it) and then steps into every element in there, adds it to the string 'url_tmp' and adds a forward slash back in.

There are better ways, but this is a small and powerful and will be more than good enough

adam1mc · 03-08-2010, 12:03 PM

That could work too. I'm just trying to learn the basics without getting too complex. :-)

Here is what I ended up using that worked well:

rev urllist | cut -d/ -f2- | rev > newurllist

pixellany · 03-08-2010, 12:14 PM

Another perspective: Learn Regular Expressions

In your example, you have more than the domain name---that would be "http://blah.com"

Here are 2 example Regexes that match common URLs (Not tested--may need extended regex flags. Note also that the "/" needs to be escaped in some situations):

Code:

1.  http://([^\.]/.)*(com|net|org|edu|mil)

2.  http://[^/]*

##In a way, this might be more robust, but it also matches all manner of invalid URLs. In your case, you might want to match everything that was intended to be a URL in the context.)

Using the second example, you can do the editing with SED:

Code:

sed 's%\(http://[^/]*\)[^ ]*%\1%g' filename > newfilename

This uses a backreference --- \(...\) ---to capture everything up to the first "/" after the http:// and then re-insert that in place of the larger expressiong (which goes up the the first space)

Again---none of this tested. For a good reference on SED an other things, go here:
http://www.grymoire.com/Unix/

pixellany · 03-08-2010, 12:16 PM

Hmmm---perhaps OP has a better answer---and in less time. Funny how that happens.....

catkin · 03-08-2010, 12:29 PM

Code:

while read
do
    echo ${REPLY%/*}
done < input.txt

rweaver · 03-08-2010, 12:32 PM

My suggestion would have been along the lines of--

Code:

for i in $(cat filename.txt); do echo $i | cut -f1-$(echo $i | awk -F\/ '{print NF-1}') -d\/; done

Which isn't an elegant solution. However, I forgot about rev and it would enable you to simplify it to a single cut as the op did. Good call!

I don't often use rev and often lament that cut doesn't support working backwards on a line... /facepalm.

Good solution OP

custangro · 03-08-2010, 12:36 PM

Quote:

Originally Posted by rweaver

My suggestion would have been along the lines of--

Code:

for i in $(cat filename.txt); do echo $i | cut -f1-$(echo $i | awk -F\/ '{print NF-1}') -d\/; done

Which isn't an elegant solution. However, I forgot about rev and it would enable you to simplify it to a single cut as the op did. Good call!

I don't often use rev and often lament that cut doesn't support working backwards on a line... /facepalm.

Good solution OP

Sometimes you learn more from doing it yourself. I agree good job OP

penguiniator · 03-08-2010, 03:11 PM

Code:

#!/bin/bash
while read url; do
    echo ${url%/*}/
done < file

chrism01 · 03-09-2010, 12:16 AM

Alternatively

Code:

basename http://blah.com/123/abc.exe 
abc.exe

Sorry; you wanted the opposite, didn't you

Code:

dirname http://blah.com/123/abc.exe
http://blah.com/123

rweaver · 03-09-2010, 12:52 PM

Lots of interesting and good answers in this thread. Make sure you tag your posts up to make them easier for others to find in the future!