Unable to assign output of sed or awk command to variables? Can only print them

问题

I use this command to extract 2 lines from a text file:

cat file1 | grep -A7 SECTIONA  | grep -E 'Address|BackupAddress'

This produces the two lines below:

Address host1  port_1  Address
BackupAddress host2 port_2  BackupAddress

I need to assign (not print) the host and port columns to distinct global variables to use later in the script.

hosta="host1"
porta="port_1"
hostb="host2"
portb="port_2"

A member suggested I use this and I get the desired output, however I cannot use the variables in the current shell? When I try to print or use them they come out blank. I also have an awk solution to get the desired output, but both sed and awk commands just print the above and do not create the variables?

sed -n 's/^Address *\([^ ]\+\) *\([^ ]\+\).*/hosta="\1"\nporta="\2"/p;
s/^BackupAddress *\([^ ]\+\) *\([^ ]\+\).*/hostb="\1"\nportb="\2"/p' file1

I can get around all this by breaking the sed and awk command into 4 sections and run the same query 4 times but there must be a better way. for example I can do:

hosta=`cat file | sed/or/awk command` 
porta=`cat file | sed/or/awk command` 
hostb=`cat file | sed/or/awk command` 
portb=`cat file | sed/or/awk command`

When I do the query using awk or sed command to get the desired output they get assigned to variables and I can print or use the variables as I like; but I do not want to run the query 4 times.

Can anyone help please

回答1:

Checking that input doesn't contain invalid character, the output can be sourced same as "eval"ed. in that case negative character set [^ ] should be replaced by a positive character set to match expected characters

output=$(sed ...)
eval "$output"

eval "$(sed ...)"

source <(sed ...)

回答2:

this is the case all the time. The host address and port lines are generic across all the files...

...So there are two lines that I grep out of the files and they will always contain 4 variables.

Given the specificity of the application, you could:

Scan the file once
Save the results into a variable
Parse only the variable to assign the host and port variables

Given the input file input.txt, which contains:

Address host1  port_1  Address
BackupAddress host2 port_2  BackupAddress

You could use this script:

#!/bin/bash

lines="$(grep -P "Address|BackupAddress" <input.txt)"             #Get the two lines needed

hosta="$(awk '/^Address/ { print $2 }' <<<"${lines[@]}")"         #Assign $hosta
porta="$(awk '/^Address/ { print $3 }' <<<"${lines[@]}")"         #Assign $porta
hostb="$(awk '/^BackupAddress/ { print $2 }' <<<"${lines[@]}")"   #Assign $hostb
portb="$(awk '/^BackupAddress/ { print $3 }' <<<"${lines[@]}")"   #Assign $portb

#Use the results how you like.  Here we just echo them back:

echo "$hosta"
echo "$porta"
echo "$hostb"
echo "$portb"

来源：https://stackoverflow.com/questions/44545583/unable-to-assign-output-of-sed-or-awk-command-to-variables-can-only-print-them

标签

bash

awk

sed

global-variables