[SOLVED] awk add value to column

[RMC]Pip · 04-23-2012, 10:00 AM

Hi,

I want to add a value to a specific column of a file if that line starts with a pattern.

Samle file:

Quote:

ExecutionTime = 28.26 s ClockTime = 31 s
blabla
ExecutionTime = 53.7 s ClockTime = 59 s
bla
ExecutionTime = 77.28 s ClockTime = 86 s
yeah
ExecutionTime = 101.14 s ClockTime = 114 s
anyway
ExecutionTime = 123.02 s ClockTime = 140 s

I want to modify the two numerical values.

My utility function for this task:

Code:

timeCorrectLog()
{
	local execTime=$2
	local clockTime=$3

	awk -F '[ \t"#]*' '/^ExecutionTime/{sub($3,$3+'$execTime')}1' $1 > tmpAWK
	awk -F '[ \t"#]*' '/^ExecutionTime/{sub($7,$7+'$clockTime')}1' tmpAWK > timeCorrect.$1
	rm -f tmpAWK
}

The command:

Code:

timeCorrectLog file 86728.8 95351

The output:

Quote:

ExecutionTime = 86757.1 s ClockTime = 95382 s
blabla
ExecutionTime = 86782.5 s ClockTime = 95410 s
bla
ExecutionTime = 95437806.1 s ClockTime = 86 s
yeah
ExecutionTime = 86829.9 s ClockTime = 95465 s
anyway
ExecutionTime = 86851.8 s ClockTime = 95491 s

Clearly, something went very wrong in line 5. I used the script for a huge log file, and it messed up only that line, have no clue what is happening!

In short, this is the command that causes the trouble:

Code:

echo "ExecutionTime = 86806.1 s  ClockTime = 86 s" | awk -F '[ \t"#]*' '/^ExecutionTime/{sub($7,$7+95351)}1'

Thank you!

colucix · 04-23-2012, 10:23 AM

Using the sub function it substitutes the first occurrence of $7 (that is the string "86") and not $7 itself. In that line there is a "86" at the beginning of the third field:

Code:

ExecutionTime = 86806.1 s  ClockTime = 86 s

You should specify the substitution has to be made in $7:

Code:

sub($7,$7+95351,$7)

but the correct way is simply:

Code:

$7 += 95351

that is the same as

Code:

$7 = $7 + 95351

[RMC]Pip · 04-23-2012, 10:30 AM

Thanks a lot colucix!

colucix · 04-23-2012, 10:40 AM

You're welcome! An aside note: you can do it using a single awk call:

Code:

#!/bin/bash
#
timeCorrectLog()
{
	local execTime=$2
	local clockTime=$3

	awk -F '[ \t"#]*' '/^ExecutionTime/{$3 += '$execTime'; $7 += '$clockTime'}1' $1 > timeCorrect.$1
}

This should speed-up the process, avoiding the need for the temporary file.

grail · 04-23-2012, 12:28 PM

You may even wish to look at the -v option for awk to set variables to be used within the command.

[RMC]Pip · 04-24-2012, 02:28 AM

Thanks, really great help!

grail · 04-24-2012, 03:42 AM

Please mark as SOLVED once you have a solution

[RMC]Pip · 04-24-2012, 05:27 AM

Yepp, just waited a bit in case of someone has some more interesting stuff!

Wait is over, thank you all!