你如何 pipe 并从尾部过滤文本作为脚本中变量的输入？

Question

Backstory

I am trying to create a script that updates a "device" through the devices cli, but it doesn't accept any form of command following the establishment of an ssh connection. for this reason i have started using screen to logging the output from the device and then attempting to filter the log for relevant info so i can pass commands back to the remote device by stuffing it into screens buffer.(kind of a ramshackled way of doing it but its all i can think of.

Issue I need to use some combo of grep and sed or awk to filter out one of two outputs i'm looking for respectively "SN12345678" '\w[a-zA-Z]\d{6-10}' and "finished" inside screenlog.2 I've got regex patterns for both of these but i cannot seem to get the right output and assign it to a variable

.screenrc (relevant excerpt)

screen -t script 0 ./script
screen -t local 1 bash
screen -t remote 2 bash
screen -t Shell 3 bash

./script

screen -p 2 -X log on #turns logging on window 2
screen -p 3 -X stuff 'tail-Fn 0 screenlog.2 | #SOMESED Function that i cant figure out'
screen -p 2 -X stuff 'ssh -o "UserKnownHostsFile /dev/null" -o "StrictHostKeyChecking=no" admin@192.168.0.1^M' && echo "Stuffed ssh login -> window 2"
sleep 2 # wait for ssh connection
screen -p 2 -X stuff admin^M && echo "stuffed pw"
sleep 4 # wait for auth
screen -p 2 -X stuff "copy sw ftp://ftpuser:admin@192.168.0.2/dev_uimage-4_4_5-26222^M" && echo "initiated flash"
screen -p 2 -X stuff "copy license ftp://ftpuser:admin@192.168.0.2/$(result of sed from screenlog.2).lic^M" && echo "uploading license"

sorry if this is a bit long winded i've been wracking my brain for the last few days trying to get this to work. Thank you for your time!

Answer 1

Answer

Regular Expression

Looking at the example regex you provided, I'm going to assume SN can't just be hardcoded and that it could be uppercase,lowercase,digit for first character and uppercase,lowercase for the second digit, so I think you are looking for:

grep -Eo '[[:alnum:]][[:alpha:]][[:digit:]]{6,10}' # Works regardless of computer's local settings
# OR
egrep -o '[[:alnum:]][[:alpha:]][[:digit:]]{6,10}' # Works regardless of computer's local settings

# OR

grep -Eo '[0-9A-Za-z][A-Za-z][0-9]{6,10}'
# OR
egrep -o '[0-9A-Za-z][A-Za-z][0-9]{6,10}'

These are exact conversions of your regular expression (includes the _ as. a possibility of the first character):

grep -Eo '[[:alnum:]_][[:alpha:]][[:digit:]]{6,10}' # Works regardless of computer's local settings
# OR
grep -Eo '[0-9A-Za-z_][A-Za-z][0-9]{6,10}'
# OR (non-extended regular expressions)
grep -o '[[:alnum:]_][[:alpha:]][[:digit:]]\{6,10\}'
grep -o '[0-9A-Za-z_][A-Za-z][0-9]\{6,10\}'

Reuse the Match

I don't know how you would assign the output to a variable, but I would just write it to a file and delete the file afterwards (assuming the "script" and "Shell" windows have the same pwd [present working directory]):

. . .
screen -p 3 -X stuff 'tail -Fn1 screenlog.2 | grep -Eo "[[:alnum:]][[:alpha:]][[:digit:]]{6,10}" >> SerialNumberOrID^M'
. . .
screen -p 2 -X stuff "copy license ftp://ftpuser:admin@192.168.0.2/$(cat SerialNumberOrID).lic^M" && echo "uploading license"
rm -f SerialNumberOrID

Explanation

Regular Expression

I'm fairly confident that grep , sed , and awk (and most POSIX compliant utilities) don't support \w and \d . Those are Perl-like flags . You can pass -E to grep and sed to make them use extended regular expressions (will save you from having to do as much escaping).

Command Changes

Writing the match to a file seemed like the best way to reuse it. Using >> ensures that we append to the file, so that grep will only write the matching expression to the file and won't overwrite it with an empty file. This is why it's necessary to delete the file at the end of your script (so that it won't mess up next run and also so you don't have unnecessary files laying around). In the license upload command, we use cat to output the contents of the file in-line. I also changed the tail command to tail -Fn1 because I'm pretty sure you need to at least have 1 for it to feed a line into grep.

Resources

• https://en.wikibooks.org/wiki/Regular_Expressions/POSIX_Basic_Regular_Expressions
• https://en.wikibooks.org/wiki/Regular_Expressions/POSIX-Extended_Regular_Expressions
• grep, sed, and awk man pages