Bash, pattern substitution for json-like multiline string-CodePudding

I wrote a bash script that essentially turns df -a command into multiple, new-line separated, Json formatted objects.

PATTERN='!/loop/ && /^\// {printf \
"\"disk.device\":\"%s\","\
"\"disk.mount\":\"%s\","\
"\"disk.total\":%d,"\
"\"disk.used\":%d,"\
"\"disk.free\":%d,"\
"\"disk.percentage\":%.2f\n", $1, $6, $2, $2-$4, $4, ($2-$4)/($2 1)*100}'

MODE=""
NAME="disk.device"
PAYLOAD="$(df -a | awk "$PATTERN")"

jsonify.sh $MODE "$PAYLOAD"

The script works in two steps:

creating a "proto-json" payload
converting it into actual json

The $PAYLOAD variable is constructed via awk and then passed to the jsonify function.

Here is an example output:

"disk.device":"/dev/nvme0n1p2","disk.mount":"/","disk.total":47745772,"disk.used":29455764,"disk.free":18290008,"disk.percentage":61,69
"disk.device":"/dev/nvme0n1p1","disk.mount":"/boot/efi","disk.total":583544,"disk.used":5356,"disk.free":578188,"disk.percentage":0,92
"disk.device":"/dev/nvme0n1p4","disk.mount":"/home","disk.total":427115864,"disk.used":46328340,"disk.free":380787524,"disk.percentage":10,85
"disk.device":"/dev/fuse","disk.mount":"/run/user/1000/doc","disk.total":0,"disk.used":0,"disk.free":0,"disk.percentage":0,00

Unfortunately a new requirement emerged, and now the value of the "disk.device" needs to be substituted from the actual name of the disk to disk_n, where n is an integer.

In the above example the result would look like:

"disk.device":"disk_0","disk.mount":"/","disk.total":47745772,"disk.used":29455764,"disk.free":18290008,"disk.percentage":61,69
"disk.device":"disk_1","disk.mount":"/boot/efi","disk.total":583544,"disk.used":5356,"disk.free":578188,"disk.percentage":0,92
"disk.device":"disk_2","disk.mount":"/home","disk.total":427115864,"disk.used":46328340,"disk.free":380787524,"disk.percentage":10,85
"disk.device":"disk_3","disk.mount":"/run/user/1000/doc","disk.total":0,"disk.used":0,"disk.free":0,"disk.percentage":0,00

Is it possible to work out a solution using a single command (maybe awk?) or do i need an explicit loop that goes trough all the lines of the payload? In case of the latter how would you proceed ?

CodePudding user response：

echo "$output" | awk -F, -v OFS=, '{
    disk_n = "disk_" (NR - 1)
    sub(/"[^"] "$/, "\"" disk_n "\"", $1)
    print
}'

CodePudding user response：

In GNU AWK one might use postincrease to get running number starting at zero, consider following example, let file.txt content be

and you want to output ordinal followed by numer for values >= 500, then you might do

awk '$1>=500{printf "%d %s\n", i  , $1}' file.txt

which gives output

0 1000
1 10000
2 100000

(tested in gawk 4.2.1)