getting selected lines in a file

viveksnv · 02-27-2008, 03:26 AM

Hello all,

my file has
KEY 9D1D479E5AEE1C4219B3B56735266609
########GET http://top100-images.rambler.ru/top100/w1.gif
########STORE_OK IN_MEMORY SWAPOUT_NONE PING_DONE
########CACHABLE,DISPATCHED,NEGCACHED,VALIDATED
########04087859 LU:1204088178 LM:-1 EX:1204088478
########0 locks, 0 clients, 1 refs
########Swap Dir -1, File 0XFFFFFFFF
KEY D57DBE38D0D3F84960FE6526A289CC6D
########GET http://wiki.squid-cache.org/wiki/squ...css/common.css
########STORE_OK IN_MEMORY SWAPOUT_DONE PING_DONE
########CACHABLE,DISPATCHED,VALIDATED
########LV:1203939464 LU:1204085787 LM:1117795312 EX:-1

IMPORTANT:
(######## -> space. Not char

I try print only the
http://top100-images.rambler.ru/top100/w1.gif
http://wiki.squid-cache.org/wiki/squ...css/common.css

Please help me.

ghostdog74 · 02-27-2008, 03:29 AM

what is the code you have?

livetoday · 02-27-2008, 04:05 AM

while playing with awk command to answer the query ...I found that the following command is giving the required output..

Code:

cat file |  awk '$1 = /http/ {print $2}'
http://top100-images.rambler.ru/top100/w1.gif
http://wiki.squid-cache.org/wiki/squ...css/common.css

But I don't know how ??

I am using GNU awk. But I don't know how awk is working in this scenario? Can anyone explain ?

ghostdog74 · 02-27-2008, 05:09 AM

Quote:

Originally Posted by livetoday

while playing with awk command to answer the query ...I found that the following command is giving the required output..

Code:

cat file |  awk '$1 = /http/ {print $2}'
http://top100-images.rambler.ru/top100/w1.gif
http://wiki.squid-cache.org/wiki/squ...css/common.css

But I don't know how ??

I am using GNU awk. But I don't know how awk is working in this scenario? Can anyone explain ?

first you can do away with the cat. Its useless in this scenario. Your usage of awk is a bit "unconventional". however it will still work in this case.

Quote:

Code:

$1 = /http/

the above means , look for pattern http, then if found, assign its value to $1. In awk, if a pattern is found, the return value would be 1. So $1=/http/ means assign 1 to $1. you can try this

Code:

# awk '$1=/KEY/{print $1,$2}' file
1 9D1D479E5AEE1C4219B3B56735266609
1 D57DBE38D0D3F84960FE6526A289CC6D

the "usual" way is this ...

Code:

# awk '/http/{print $2}' file
http://top100-images.rambler.ru/top100/w1.gif
http://wiki.squid-cache.org/wiki/squ...css/common.css

livetoday · 02-28-2008, 06:52 AM

Thanks a lot for this detailed explanation. This certainly clears my doubts however while reading through your answer a new query pops up.

Is is possible to print only that field in which awk has found the pattern.

Quote:

Originally Posted by ghostdog74

first you can do away with the cat. Its useless in this scenario. Your usage of awk is a bit "unconventional". however it will still work in this case.

the above means , look for pattern http, then if found, assign its value to $1. In awk, if a pattern is found, the return value would be 1. So $1=/http/ means assign 1 to $1. you can try this

Code:

# awk '$1=/KEY/{print $1,$2}' file
1 9D1D479E5AEE1C4219B3B56735266609
1 D57DBE38D0D3F84960FE6526A289CC6D

the "usual" way is this ...

Code:

# awk '/http/{print $2}' file
http://top100-images.rambler.ru/top100/w1.gif
http://wiki.squid-cache.org/wiki/squ...css/common.css

pixellany · 02-28-2008, 07:06 AM

sed -n '/http:/p' filename

Quote:

Originally Posted by ghostdog74

what is the code you have?

Why didn't we stop here??...

ghostdog74 · 02-28-2008, 07:14 AM

you can use a for loop to go through the fields

Code:

awk '
{
  for(i=1;i<=NF;i++){
    if ($i ~ /pattern/) print $i
  }
}
'

you can also use match() with RSTART and RLENGTH as the variables set when the pattern is found. However they just show you where the pattern starts and end, not which field though.

ghostdog74 · 02-28-2008, 07:16 AM

Quote:

Originally Posted by pixellany

sed -n '/http:/p' filename

Why didn't we stop here??...

hmm...beats me

livetoday · 02-28-2008, 09:55 PM

Thanks again....

Yeah ! I also feel sed is easiser to use to get the desired output in this case.

Code:

sed -n 's/.*\(http:*\)/\1/p' filename

http://top100-images.rambler.ru/top100/w1.gif
http://wiki.squid-cache.org/wiki/squ...css/common.css

Quote:

Originally Posted by ghostdog74

you can use a for loop to go through the fields

Code:

awk '
{
  for(i=1;i<=NF;i++){
    if ($i ~ /pattern/) print $i
  }
}
'

you can also use match() with RSTART and RLENGTH as the variables set when the pattern is found. However they just show you where the pattern starts and end, not which field though.

ghostdog74 · 02-28-2008, 10:27 PM

Quote:

Originally Posted by livetoday

Thanks again....

Yeah ! I also feel sed is easiser to use to get the desired output in this case.

for this case, sure.

Quote:

Code:

sed -n 's/.*\(http:*\)/\1/p' filename

http://top100-images.rambler.ru/top100/w1.gif
http://wiki.squid-cache.org/wiki/squ...css/common.css

you might want to add an extra space after the http pattern so that you won't grab more than what you want.