Sed help please

Tinkster · 04-20-2005, 12:28 AM

My thing is non-destructive

As you see in
awk -f clean-up.awk phonelist > phonelist.new
it creates a phonelist.new file and leaves it up
to you to replace the old one if it's correct.

As for the learning ... there's a pretty good book
out there by Ellie Quigley, "Unix shells by example",
that covers grep, sed and awk (not exhaustively)
but with good explanations.

Cheers,
Tink

ahh · 04-20-2005, 05:51 AM

Quote:

Originally posted by nazs
Thank you to everyone who has responded . You all are awesom. I will try it out again when i go to work tomorrow. Does anyone have any recomendations on how to learn to write scripts like this. (Ex. books or web pages, or cdroms) I would love to learn this stuff. Please let me know. I will get back to you all and let you know how things work and i definitely will try it out on a test_copy first.

Nazs

A place to start would be to type

Code:

info

in a terminal. This would show you the utilities available, and a description of how to use them.

Awk is an awesome tool, but sometimes its overkill

Run this script

Code:

#!/bin/bash

FULL_LIST=/path/to/list
DEL_LIST=/path/to/names

for i in $(cat $DEL_LIST); do
  x=$(grep -n $i $FULL_LIST | cut -d ":" -f 1)
  while [ $(sed -n $x'p' $FULL_LIST) ]; do
    sed -ie $x'd' $FULL_LIST
  done
done

on your copy.

Use the man pages and/or info pages for grep, cut and sed to see how these tools are being used. There are plenty of bash tutorials on line that will show you how to write a script using them.

http://www.google.co.uk/search?hl=en&q=bash+tutorial

If you want a closer look at what is going on when the script runs, change the first line to

Code:

#!/bin/bash -x

and it will print out what it is doing.

homey · 04-20-2005, 07:49 AM

That is a beauty ahh and I have been severily outclassed!

I looked at the option p for sed but didn't quite catch the fact that this is what I needed to get to the next blank line. That's why I was having to manually put in the start and stop lines.

Tinkster · 04-20-2005, 01:32 PM

Quote:

Originally posted by ahh
Awk is an awesome tool, but sometimes its overkill :)

Your shell version leaves tons of blank lines
behind, and it takes longer to run ...

Code:

[tink@diggn:~/tmp4]$ time test.awk testing.ori > testing.2

real    0m0.004s
user    0m0.000s
sys     0m0.000s
[tink@diggn:~/tmp4]$ time ./rem.sh 

real    0m0.033s
user    0m0.000s
sys     0m0.030s

I still like the awk version better ;)

Cheers,
Tink

ahh · 04-20-2005, 04:59 PM

Quote:

Originally posted by Tinkster
Your shell version leaves tons of blank lines
behind, and it takes longer to run ...
...
I still like the awk version better

Yeah, I know its slower and leaves blank lines, but its shorter and easier, and the blank lines are easily removed with 'cat -s'.

But I agree, yours is technically superior.

nazs · 04-20-2005, 06:38 PM

Tink i have a question on your revised script. iIf i have one file with the list of names to be deleted and also a file with all the names and nubers it needs to go through to make the deletions from. Where do i place the path to the file with the names to be deleted and also where do i put the path to the list with all the names and numbers? i know i am suppose to put paths where it says /path/to/killem just not sure where what goes where.

Thanks Nazs

Tinkster · 04-20-2005, 06:59 PM

Quote:

Originally posted by nazs
Tink i have a question on your revised script. iIf i have one file with the list of names to be deleted and also a file with all the names and nubers it needs to go through to make the deletions from. Where do i place the path to the file with the names to be deleted and also where do i put the path to the list with all the names and numbers? i know i am suppose to put paths where it says /path/to/killem just not sure where what goes where.

Thanks Nazs

The name of the file with the names to be deleted is the killem
in my /path/to/killem ...

The file you want to EDIT goes as a parameter to the awk-script.
If you change the awk to look like that:

Code:

#!/usr/bin/awk -f
BEGIN {
  count = 0
  while (( getline < "killem") > 0 ){
    count += 1
    name[count]=$1
  }
  close( "killem" )
  RS=""; FS="\n";OFS="\n\n"
}
{
  flag = 1
  for ( i=1; i<=count;i++){
    if($1 ~ name[i]) flag = 0
  }
  if( flag == 1  ) {print $0; printf "\n"}
}

and make it executable (chmod u+x FILE) you can actually
run it like you would have done the shell-script, too.
./FILE phonelist > newlist

Cheers,
Tink