[SOLVED] Bash script help needed

anotherunixnoob · 11-03-2010, 02:27 PM

I have a script that starts to read through a txt file and finds a key word, as it is doing this it counts each line until it gets to its desired line(line y). Then it starts to re-read the file from the beginning until it gets line y+x where it performs some operations on that line to get the data I need and from there I can export the variable when needed.

The script does this for every piece of data that I need and I am looking for a way to do this and use considerably less code. What I need is a point in the right direction on how to do it so I can implement it in the rest of my script.

My code and a sample of the data I am reading are below.

Thanks in advance.

Code:

Code:

for i in `cat hardware`
do
		let "getnum++"
		if [[ $i == \<Processors\> ]]
			then
				break				
		fi
done

let "getnum = $getnum + 5"

for i in `cat hardware`
do
		let "counter++"
		if [[ $counter == $getnum ]]
			then
				cpuspeed=${i:7:7}
		fi
done

Data:

Code:

(line y)<Processors>
        <Processor is64bitCapable="yes">
          <Manufacturer>Intel</Manufacturer>
          <Version/>
(line y+x)<Speed>3000MHz</Speed>

crts · 11-03-2010, 03:11 PM

Hi,

sounds like homework. Anyway, you could start by doing this all with just one loop.

catkin · 11-03-2010, 03:15 PM

How about something like (not tested)

Code:

while read line
do
    if [[ [[ $line == \<Processors\> ]]; then
        for (( i=0; i<5; i++ ))
        do
             read line
        done
        <do what you want with the line 5 after Processors>
    fi 
done < input.txt

ntubski · 11-03-2010, 03:23 PM

How about using something that understands xml:

Code:

cpuspeed=$(xmlstarlet sel -t -v //Speed data.xml)

anotherunixnoob · 11-03-2010, 04:27 PM

Quote:

Originally Posted by ntubski

How about using something that understands xml:

Code:

cpuspeed=$(xmlstarlet sel -t -v //Speed data.xml)

I put 'bash script' in the title for a reason. Thanks for the suggestion anyway though.

And its not homework for school its homework for work.

theNbomr · 11-03-2010, 06:28 PM

Quote:

Originally Posted by anotherunixnoob

I put 'bash script' in the title for a reason. Thanks for the suggestion anyway though.

If someone asked you how to build a house with a toothbrush, you would probably mention that it isn't the right tool for the job (even if you did know how to accomplish such a task). Your problem is clearly looking for a solution using awk or Perl. Here's a shell script that uses Perl.

Code:

#! /bin/sh 
perl -e '@data=<>;foreach $i (1..@data){ if($data[$i]=~m/<Processors>/){print $data[$i+5];}}' YourInputData.File

--- rod.

ntubski · 11-03-2010, 11:07 PM

Quote:

Originally Posted by anotherunixnoob

I put 'bash script' in the title for a reason. Thanks for the suggestion anyway though.

What makes it not a bash script? Is this better?

Code:

x=$(grep -n '<Processors>' hardware | grep -o '^[0-9]*')
let 'y = x + 4'
cpuspeed=$(sed -n "${y}s/<[^>]*>\| *//gp" hardware)

anotherunixnoob · 11-04-2010, 09:29 AM

I am using a tooth brush to build my house because there are only tooth brushes on site. I'm writing this script to be used on a system that doesn't have perl installed. In fact it has very little installed because its a shell that we use during the boot process to update firmware on new machines that are being put into production. Could I install perl and then rebuild the boot image? I sure could, I just find writing a simple bash script to be considerably easier.

catkin's code revised

Code:

while read line
do
    if [[ $line == \<Processors\> ]]; then
        for (( i=0; i<4; i++ ))
        do
             read line
        done
        echo $line
       # <do what you want with the line 5 after Processors>
    fi
done < input.txt

When I was searching around on the tubes(before I posted) this is actually what I was looking for. I was looking for a way to find text that would be unique, and then be able to advance the line reader a certain amount of times to the text that I needed. All I could think of was how to do this in other languages like c++, java or perl. Thanks for the help catkin.

Quote:

Originally Posted by ntubski

What makes it not a bash script? Is this better?

Code:

x=$(grep -n '<Processors>' hardware | grep -o '^[0-9]*')
let 'y = x + 4'
cpuspeed=$(sed -n "${y}s/<[^>]*>\| *//gp" hardware)

I jumped to quickly to a conclusion that your original code wasn't a bash script seeing 'xmlstarlet'. What I should have said was something about not having that particular utility. I look at your new code and the line that includes 'sed' confuses and infuriates me haha. I guess I'll have to read the man pages and some tutorials on sed. Thanks for the help.

crts · 11-04-2010, 10:32 AM

Quote:

Originally Posted by anotherunixnoob

I guess I'll have to read the man pages and some tutorials on sed. Thanks for the help.

So, you can use sed for the task?

Code:

cpuspeed=$(sed -rn '/other/,+5 p' inputfile | sed -rn '$ s/[^>]*>([^<]*).*/\1/')
# or without the pipe
cpuspeed=$(sed -rn '/<Processors>/,/<Speed>.*<\/Speed>/ {/<Speed>.*<\/Speed>/ {s@.*<Speed>([^<]*)</Speed>.*@\1@;p}}' file)

The first one depends on "advancing" 5 lines (from your sample data it should be 4 though) to get the info, the second alternative searches for the pattern '<Speed>' after the pattern '<Processors>' is encountered.

Yes, sed replaces the both of your loops.

anotherunixnoob · 11-04-2010, 04:41 PM

Awesome thanks for the help. Right now my code looks like a mess but I'll keep working on it and as I get more familiar with the slightly more advanced stuff I'll be able to clean it up.