CODEWITHSUNDEEP
shellawksedgrep
Francesco Marchioni
Francesco Marchioni

Reputation: 4328

Bash: Extract variable value from string

I'd need to extract the value of a variable "error" from a log file. Here's a sample line:

WARN (Periodic Recovery) IJ000906: error=15 check server.log

I'd need to capture the value of "error". Looking on similar answers, I've come up with:

echo "WARN (Periodic Recovery) IJ000906: error=15 check server.log" |  grep -P '\d+ (error=?)' -o

However it does not produce any value. Can you recommend a working solution for this case?

Upvotes: 3

Views: 879

Answers (6)

RavinderSingh13
RavinderSingh13

Reputation: 133428

1st solution: With your shown samples, please try following awk code.

awk -F'error=| check' '{print $2}' Input_file

Explanation: Simple explanation would be, setting field separators as error= OR check for all the lines. Then printing 2nd field of line, which will print value after error= and before check as per shown samples.

2nd solution: Using match function of awk here.

awk 'match($0,/error=[^[:space:]]+/){print substr($0,RSTART+6,RLENGTH-6)}' Input_file

Upvotes: 1

anubhava
anubhava

Reputation: 784918

You may use this grep:

s='WARN (Periodic Recovery) IJ000906: error=15 check server.log'
grep -oP '\berror=\K\d+' <<< "$s"

15

RegEx Details:

  • \b: Match word boundary
  • error=: Match error= text
  • \K: Reset matched info
  • \d+: Match 1+ digits and print it

Upvotes: 2

Cyrus
Cyrus

Reputation: 88553

With bash >= 3.0.

v='WARN (Periodic Recovery) IJ000906: error=15 check server.log'

[[ $v =~ error=([0-9]+) ]] && echo "${BASH_REMATCH[1]}"

Output:

15

Upvotes: 1

Daweo
Daweo

Reputation: 36360

I would use GNU AWK following way, let file.txt content be

WARN (Periodic Recovery) IJ000906: error=15 check server.log

then

awk 'BEGIN{FPAT="error=[0-9]+"}{print substr($1,7)}' file.txt

output

15

Explanation: I inform GNU AWK that column is error= followed by 1 or more digits using field pattern (FPAT), for every line print first field starting from 7th charater, using substr string function. 7 as error= has 6 characters. Note: this solution will print first occurence of error=value for each line.

(tested in gawk 4.2.1)

Upvotes: 2

glenn jackman
glenn jackman

Reputation: 246744

For a perl-compatible regular expression, you're looking for a "lookbehind" assertion.

To find digits that are preceded by the string "error=", you want:

echo "$line" | grep -o -P '(?<=error=)\d+'    # => 15

See the pcresyntax(3) man page

Upvotes: 4

sseLtaH
sseLtaH

Reputation: 11207

Using sed

$ echo "WARN (Periodic Recovery) IJ000906: error=15 check server.log" | sed 's/.*error=\([^ ]*\).*/\1/'
15

Upvotes: 4

Related Questions