[SOLVED] Reading the content of line number $ (Shell Script)

WhisperiN · 11-11-2009, 12:59 PM

Hello Geeks and Fellows,

I'm writing a shell script.. and in some part of it I need that script to read the content of a line by giving that line number.

Well, I found a work-around to achieve that, but I'm looking for something more efficient to enhance the script performance.

Here is the work-around I got to:

Code:

#!/bin/bash

LineNumber=5
FileLocation='/some/locations/file'

# Reading the content of line number $LineNmber stored in FileLocation

Content=$(head $FileLocation -n$LineNumber | tail -n1)

# End

So, is there any thing simpler that that?

Regards

sycamorex · 11-11-2009, 01:05 PM

You can use sed:

Code:

num=10
file=my_file
sed -n "$num p" $file

WhisperiN · 11-11-2009, 05:23 PM

Thanks sycamorex..

That was good..

I hope to see some more choices..

Regards

chrism01 · 11-12-2009, 12:18 AM

Perl snippet

Code:

# '$.' is current line num
open(FH,'foo.txt') or die "$!"; 
while (<FH>) 
{ 
  print "$_" if( $. == 5 ); 
} 
close(FH);

ghostdog74 · 11-12-2009, 12:30 AM

read a line using line number in gawk

Code:

#!/bin/bash
num=10
gawk -v ln="$num" 'NR==ln' file

bigearsbilly · 11-12-2009, 03:35 AM

shorter perl

Code:

 perl -ne 'print if $. == 3' file

WhisperiN · 11-15-2009, 07:44 PM

Thanks all..

That's fair enough

Much Regards..

stevenworr · 11-17-2009, 11:53 AM

Why don't you just do it in bash?

Code:

#!/bin/bash

typeset -i ln
typeset -i ii
ln=$1
fn=$2
for ((ii=1; ii < ln + 1; ii=ii+1 ))
do
    read line
    (( ii == ln )) && echo "$line"
done < $fn

Now you're not starting any subprocesses.

ghostdog74 · 11-17-2009, 02:10 PM

Quote:

Originally Posted by stevenworr

Why don't you just do it in bash?

[/CODE]

Now you're not starting any subprocesses.

one of the advantage is that if he is reading a very big file, using a file reading tool such as awk is the way to go. bash's loop+read is simply too slow.

gnashley · 11-17-2009, 03:27 PM

But, if the files are small and he is doing lots of reads on multiple files, the bash solution will be much faster, than calling any external program. I suspect that sed would be faster than awk, which would be faster than perl -because of latency in forking the external program.
Here's another example using bash:

Code:

#!/bin/bash
LineNumber=5
FileLocation='/some/locations/file'
# or: LineNumber=$1 FileLocation=$2

COUNT=1
while read LINE ; do
  # break once the line is found to avoid looping through the whole file
  [[ $COUNT -eq $LineNumber ]] && echo $LINE && break
done < $FileLocation

ghostdog74 · 11-17-2009, 06:20 PM

Quote:

Originally Posted by gnashley

But, if the files are small and he is doing lots of reads on multiple files, the bash solution will be much faster, than calling any external program.

how about a small test on file with 20000 lines ( a small file) and getting the 1000th line

Code:

$ head -5 file
     1  this is a line
     2  this is a line
     3  this is a line
     4  this is a line
     5  this is a line
$ tail -5 file
 19996  this is a line
 19997  this is a line
 19998  this is a line
 19999  this is a line
 20000  this is a line

$ more test.sh
#!/bin/bash
LineNumber=$1
FileLocation="file1"
COUNT=1
while read LINE ; do
  [[ $COUNT -eq $LineNumber ]] && echo $LINE && break
  let COUNT=COUNT+1
done < $FileLocation

$ time ./test.sh 1000
1000 this is a line

real    0m0.090s
user    0m0.072s
sys     0m0.012s

$ time ./test.sh 1000
1000 this is a line

real    0m0.096s
user    0m0.073s
sys     0m0.012s

$ time awk 'NR==1000{print;exit}' file
  1000  this is a line

real    0m0.003s
user    0m0.001s
sys     0m0.001s

$ time awk 'NR==1000{print;exit}' file
  1000  this is a line

real    0m0.004s
user    0m0.000s
sys     0m0.003s

$ time sed -n '1000{p;q}' file
  1000  this is a line

real    0m0.003s
user    0m0.000s
sys     0m0.003s

$ time sed -n '1000{p;q}' file
  1000  this is a line

real    0m0.003s
user    0m0.001s
sys     0m0.002s

calling sed or awk is much faster. not to mention the amount of bash code to cook up as well.