Home / ubuntu / Questions / 1067045
Accepted
Yura Shinkarev
Asked: 2018-08-20 22:56:16 +0800 CST

Wrong sed with regexp

  • 772

I have log file. I want extract email list from specific lines. Sample log file:

05:06:48,311  INFO http-nio-8080-exec-81 controller.IndexController:221 - Attempt to login: [[email protected]]. Headers: ["content-length": "52", ...]
15:06:48,311  INFO http-nio-8080-exec-81 controller.IndexController:221 - Attempt to login: [[email protected]]. Headers: ["content-length": "52", ...]
09:40:21,187  INFO http-nio-8080-exec-31 security.AuthAuthenticationProvider:81 - User: [16167]. Wrong code. Telegram: [0]. Google: [0]

My script:

cat file.log | sed -r 's/.*Attempt to login: \[(.*)\]\..*/\1/' | sort | uniq > file.log.filtered.txt

I get file.log.filtered.txt:

09:40:21,187  INFO http-nio-8080-exec-31 security.AuthAuthenticationProvider:81 - User: [16167]. Wrong code. Telegram: [0]. Google: [0]
[email protected]

Why I have line with "AuthAuthenticationProvider" in filtered file?

command-line sed text-processing

1 Answers

  1. Best Answer

    Because you are running your sed command only on lines which contains : Attempt to login so other lines will stay untouched.

    You can change your code like this:

    grep 'Attempt' file.log | sed -r 's/.*Attempt to login: \[(.*)\]\..*/\1/' | sort | uniq > file.log.filtered.txt

    so it exclude the other lines first, then extracts email addresses.

    or this one:

    sed -re 's/.*Attempt to login: \[(.*@.*)\]\..*/\1/;t;d' file.log | sort | uniq

    or even this one:

    sed -nr 's/.*Attempt to login: \[(.*@.*)\]\..*/\1/p' file.log | sort | uniq

    Or just use grep:

    grep 'Attempt' file.log | grep -o '[[:alnum:]+\.\_\-]*@[[:alnum:]+\.\_\-]*' | sort | uniq

    Regex for last command form here

    • 5