Print fields between two delimiters 1

[AnotherAlan]

Hi All,

Just a quick puzzle really as I have found a suitable workaround, but I was attempting to print the fields from the below example loglines from the word AFTER to the numerals after the word took;

i.e. AFTER Initial lock of quote took 13
AFTER getData for 46681 took 0

2011-05-17 [[ACTIVE] ExecuteThread: '186' for queue: 'weblogic.kernel.Default (self-tuning)'] INFO QuoteServiceImpl - AFTER Initial lock of quote took 13 ms for quote 4
2011-05-17 [[ACTIVE] ExecuteThread: '186' for queue: 'weblogic.kernel.Default (self-tuning)'] INFO QuoteServiceImpl - AFTER getData for 46681 took 0 ms for quote 4

Although I could get the field number for AFTER using a for loop I couldn't work out how to put an end delimiter to it.

Anyway, I got what I needed by using awk and sed;
nawk -F- '/AFTER/ {print $5}' | sed s'/ms.*//g'
but was interested if there was an Awk solution.

Thanks

 

[feherke]

Hi

nawk -F- '/AFTER/ {sub(/ms.*/,"",$5); print $5}'

Tested with [tt]gawk[/tt] and [tt]mawk[/tt].

( I heard that some old AWK implementations were unable to use [tt]sub()[/tt] directly on the fields. For those you ha(d|ve) to put [tt]$5[/tt] in a variable then [tt]sub()[/tt] that. )

Feherke.

http://free.rootshell.be/~feherke/

 

[AnotherAlan]

Hi Feherke, thanks for the speedy reply.

May I ask another, just for my learning benefit. Feel free to ignore it of course, having already provided an answer ;-)

Using the same examples, if I wanted to print just fields between "AFTER" and "took" and then the numerals after "took"; bearing in mind that the field placings can be completely variable

i.e. Initial lock of quote 13
getData for 46681 0

Can this be easily achieved with Awk?

Many Thanks

 

[feherke]

Hi

I am sure you can [tt]sub()[/tt] out the undesired words yourself, so I would bring another approach into your attention - use the undesired words as field separators :

awk -F ' (AFTER|took|ms) ' 'NF==4{print$2,$3}'

[gray]# or[/gray]

awk -F ' AFTER | took | ms ' 'NF==4{print$2,$3}'

Tested with [tt]gawk[/tt] and [tt]mawk[/tt].

Note that using more complex regular expression as field separator can slow down the processing.


Feherke.

http://free.rootshell.be/~feherke/

 

[AnotherAlan]

That is cunning and very clever, thanks Feherke. If I could award another star I would.
A lesson in thinking around the problem rather than just bulldozing my way in with complexity.

Appreciate the guidance as always.
Alan

 

[feherke]

Hi

Partially to my relaxation, two more alternative approaches, both [tt]gawk[/tt]-specific :

awk 'match($0,/AFTER (.+) took (.+) ms/,a){print a[1],a[2]}'

[gray]# or[/gray]

awk 'match($0,/AFTER (.+) took ([[:digit:]]+)/,a){print a[1],a[2]}'

awk '/AFTER/{print gensub(/.* AFTER (.+) took (.+) ms .*/,"[i]1 [/i]2","")}'

[gray]# or[/gray]

awk '/AFTER/{print gensub(/.* AFTER (.+) took ([[:digit:]]+) .*/,"[i]1 [/i]2","")}'

Tested with [tt]gawk[/tt].

[tt]gensub()[/tt] is [tt]gawk[/tt] extension. But [tt]match()[/tt] is POSIX, only its 3[sup]rd[/sup] parameter is not standard, it may be present in [tt]nawk[/tt] too.


Feherke.

http://free.rootshell.be/~feherke/

 

[AnotherAlan]

Cool, more tricks to add to the awk armoury.
Thanks