simple string manipulation

MiCr0

Hi All,
my brain is asleep, and i need to do a simple piece of string manipulation
i've got a file containing about 100k records like this

&SEQUENCENUMBER=43|TIMESTAMP=20071023113913|TRIGGER=interim|CONTEXTID=node1;167773127+471dd383@1|MSISDN=1234567890|IPADDRESS=10.11.12.13|CLIENTPROTOCOL=rtsp|NETWORKELEMENTID=node1|SERVICETYPE=streaming|APPLICATIONDESCRIPTION=Streaming/downloading service|CONTENTURL=rtsp://100.10.12.13/vod/beer.3gp|TRANSFERREDVOLUME=0|TIMEDURATION=0|CHARGINGSTATUS=notDebited|CONTENTRATE=2000|PRICETYPE=pricePerMinute|CURRENCYCODE=978|CURRENCYEXPONENT=-2|&

i need to extract these values
167773127+471dd38

Using a command like awk like this it works
awk -F'|' '{print $4}' input.ASCII | awk -F';' '{print $2}'

but its not very efficient.
any ideas how to avoid the second awk?

brianmc

Hi, nothing terribly simple spings to mind for basic AWK unless there is something more uniform about the structure of that 4th field in each record that could be used.

If the length of the field and the position within the field of the vale you want is standard in each record you could use the substr function.

If not, then nawk or gawk might solve your problem with a sub comand to substitute the first part of the fields value up to the "=" with nothing.

If none of this works out and you have to use traditional AWK (not nawk or gawk) then you could write a loop that uses substr to look at each character in the value one at a time up to the "=" and then extract the remaining characters.

MiCr0

thanks for that

Any one know is there a quicker way to do it that doesn't involve awk?

djmarkus

For what your doing it looks like you could just use cut

brianmc

djmarkus wrote: »

For what your doing it looks like you could just use cut

Yep, "cut" into another "cut" or into "sed". Or "awk" into "sed". I don't know if any of it is much more efficient though. What shell are you using? The following should work in ksh...

FIELD=$(awk -F"|" '{print $4}')
FIELD=${FIELD##*;}

Probably works like that in Bash too but I've not tried it.

I was talking about "=" symbols in my last post where I meant to say ";" symbols btw. Also the ";" in the second command line above may need to be \ escaped. Haven't tested.

MiCr0

awk -F'|' '{print substr($4,17, length($0)-1)}' file.name

brianmc

MiCr0 wrote: »

awk -F'|' '{print substr($4,17, length($0)-1)}' file.name

Curious about this bit...

length($0) - 1

That's the length of the entire record which is obviously longer than the length of the fourth field but otherwise an irrelevant number. Could you just replace that with, say 999. I.E. a large number.

carveone

MiCr0 wrote: »

i need to extract these values
167773127+471dd38

Using a command like awk like this it works
awk -F'|' '{print $4}' input.ASCII | awk -F';' '{print $2}'

Bit late but:

awk -F '[|;]' '{ print $5 }'

or

awk -F '|' '{ str = $4; sub(/^.*;/, "", str); print str; }'

or

awk '{ str = $0; sub(/^.*;/, "", str); sub(/\|.*$/, "", str); print str }'

or many others...

As for substr, just leave off the last parameter to specify the remaining length. substr("washington", 5) returns "ington".