grep

blah1234 · 12-18-2018, 10:25 AM

Hello,

I have a file I'm trying to figure out a way to output certain information from a log file. I'm looking for the regexp number and the network with the cidr (the /30 and /24).

Example of the log file:

input.txt
Router# show ip bgp regexp 65000
Network Next Hop Metric LocPrf Weight Path
* 10.1.0.0/30 172.16.0.6 0 300 100 ?
*> 172.16.0.1 0 0 100 ?
* 172.16.0.0/24 172.16.0.6 0 300 100 ?
* 172.16.0.1 0 0 100 ?

Router# show ip bgp regexp 65100
Network Next Hop Metric LocPrf Weight Path
* 10.2.0.0/30 172.16.0.6 0 300 100 ?
*> 172.16.0.1 0 0 100 ?
* 10.3.0.0/30 172.16.0.1 0 100 300 ?
* 172.17.0.0/30 172.16.0.6 0 300 100 ?

output.txt

65000
10.1.0.0/30
172.16.0.0/24

65100
10.2.0.0/30
10.3.0.0/30
172.17.0.0/30

I have: cat input.txt | egrep "/|show ip bgp regexp" | cut -d " " -f4 > output.txt but I'm getting:

regexp
10.1.0.0/30
172.16.0.0/30
regexp
10.2.0.0/30
10.3.0.0/30
172.17.0.0/30

Any suggestions on using cut with multiple delimiters? I'm open to another method of doing this. Thanks in advanced!

pan64 · 12-18-2018, 10:36 AM

this is a job for awk:

Code:

awk '/pattern/ { print something }
     /pattern/ { print something }
 .... '

l0f4r0 · 12-18-2018, 10:54 AM

Code:

awk '/Router# show ip bgp regexp/{print $6} $2 ~ /\//{print $2}' input.txt

teckk · 12-18-2018, 12:57 PM

Any one of these examples, and more.

Code:

cut -d ' ' -f6 <<< $(grep 'Router' input.txt)

grep 'Router' input.txt | cut -d ' ' -f6

awk '/Router/' input.txt | cut -d ' '  -f6

awk '/Router/{print $6}' input.txt

head -n1 input.txt | cut -d ' '  -f6

grep -o '[0-9][0-9][0-9][0-9][0-9]' input.txt

grep  -o '[0-9]\{5\}' input.txt

l0f4r0 · 12-18-2018, 02:47 PM

^ Global remark: you've missed the CIDR requirement part

Code:

awk '/Router/' input.txt | cut -d ' '  -f6

Really not optimized

Code:

grep -o '[0-9][0-9][0-9][0-9][0-9]' input.txt
grep  -o '[0-9]\{5\}' input.txt

Very hazardous as OP didn't say numbers after "regexp" are always 5 digits long

pan64 · 12-19-2018, 03:04 AM

Quote:

Originally Posted by teckk

Any one of these examples, and more.

Code:

cut -d ' ' -f6 <<< $(grep 'Router' input.txt)

grep 'Router' input.txt | cut -d ' ' -f6

awk '/Router/' input.txt | cut -d ' '  -f6

syg00 · 12-19-2018, 03:32 AM

KISS - for field based data I would usually also use awk, but if we take some assumptions based on request/limited data, grep can handle this in a readable manner.

Code:

grep -oE "([[:digit:]]+$|[[:digit:].]+/[[:digit:]]{1,2})" input.file

danielbmartin · 12-19-2018, 06:37 AM

Quote:

Originally Posted by syg00

... grep can handle this in a readable manner.

Code:

grep -oE "([[:digit:]]+$|[[:digit:].]+/[[:digit:]]{1,2})" input.file

An impressive solution! Impressive, but not readable for those with lesser skills (such as me). Please explain!

Daniel B. Martin

.

pan64 · 12-19-2018, 07:43 AM

it is quite easy, if you find a way to split it:

Code:

(                # opening 
[                # character set beginning (bracket expression in man page)
[:digit:]        # character class, digits: 0123456789
]                # end of character set
+                # multiplier (one ore more)
$                # end of line
|                # this is an OR
[                # again a char set
[:digit:]        # including digits
.                # and a dot
]                # end of set
+                # multiplier
/                # just a /
[[:digit:]]      # this could be [0-9] too
{1,2}            # 1 or 2 instances allowed
)                # closing

see man grep for additional details

danielbmartin · 12-19-2018, 10:53 AM

Thank you syg00 for the excellent one-line grep solution.

Thank you pan64 for the explanatory breakdown. Yes, man grep provides additional details but (for me) the complexity lies in the RegEx. I'm still digesting your post. One question: this ...

Code:

{1,2}            # 1 or 2 instances allowed

... allows for 1 or 2 instances but the second "paragraph" in the InFile (headed 65100) has 3 instances of a slash, and yet the code handles them properly. (???)

Daniel B. Martin

.

pan64 · 12-19-2018, 11:03 AM

Quote:

Originally Posted by danielbmartin

I'm still digesting your post. One question: this ...

Code:

{1,2}            # 1 or 2 instances allowed

... allows for 1 or 2 instances but the second "paragraph" in the InFile (headed 65100) has 3 instances of a slash, and yet the code handles them properly. (???)

{1,2} - repeater if you want to say - will work on the previous item only, which is in this case a character set or bracket expression [ ] (actually containing digits).
In short [[:digits:]]{1,2} means one or two digits

l0f4r0 · 12-19-2018, 11:06 AM

Quote:

Originally Posted by danielbmartin

One question: this ...

Code:

{1,2}            # 1 or 2 instances allowed

... allows for 1 or 2 instances but the second "paragraph" in the InFile (headed 65100) has 3 instances of a slash, and yet the code handles them properly. (???)

You are mixing CIDR network digit instances with CIDR instances.
/[[:digit:]]{1,2} means that CIDR network value can be 1 or 2 digit-long (in our case, in the 2nd paragraph, the value is equal to "30" so 2 digit-long).
It's not related to the number of CIDR instances (3 instances which are: 10.2.0.0/30, 10.3.0.0/30 and 172.17.0.0/30 in the 2nd paragraph).

syg00 · 12-19-2018, 06:20 PM

Thanks from me too to pan64 for the excellent breakdown. One advantage of world-wide subscibers; some-one else can answer while I sleep ....