Parse value from different JSON strings (No jq)

Question

I'm using bash script that needs to read the JSON output and parse a value from different JSON variables or strings. Here's the sample output. It needs to read the value next to the Content or from any other variable. Such as,

Lookup Content and be able to print Value1.

Lookup DeviceType and be able to print Value4

Sample Ouput: {"Content":"Value1","CreationMethod":"Value2","database":"Value3","DeviceType":"Value4"}

I tried the combination of sed and awk sed 's/["]/ /g' | awk '{print $4}', but only if the position of Content remains the same in the output. otherwise in the different JSON output, the positioning of Content changes that puts the value out of scope thus awk '{print $4}' picks up the wrong value.

This would be trivial with jq, or any other JSON parser. With sed and or awk, you would not only have to rely on the document always being formatted exactly the same way, but also do your own string decoding in case the data is JSON encoded. — Kusalananda, Nov 08 '19 at 15:56
You want to get the values only? Or would you want a specific output text? How's the lookup performed, with grep? — schrodingerscatcuriosity, Nov 08 '19 at 16:10
From the output, if it can grep "Content": and then only print the Value1 — Riz, Nov 08 '19 at 16:22
Please looks here: https://unix.stackexchange.com/questions/24140/return-only-the-portion-of-a-line-after-a-matching-pattern — mariaczi, Nov 08 '19 at 16:23
How about Python? Any scripting language? What is the reason for ruling out the most suitable tools for the job? — JigglyNaga, Nov 08 '19 at 17:12

score 0 · Answer 1 · answered Nov 08 '19 at 16:41

0

Well, if I understand you correctly, with GNU grep and tr:

tr ',' '\n' < foo | grep -Po "(?<=Content\":\").*(?=\")"

Output:

Value1

With other keyword:

tr ',' '\n' < foo | grep -Po "(?<=database\":\").*(?=\")"

Value 3

Replace the commas for a new line:

tr ',' '\n'

Grep -o only -P perl like regex

grep -Po

Catch the pattern ("<keyword>:"")<value>(") and print only <value>.

answered Nov 08 '19 at 16:41

schrodingerscatcuriosity

12,396

1

Or s="Content"; grep -Po "(?<=$s\":\")[^\"]*" file – bu5hman Nov 08 '19 at 22:02

bu5hman · Answer 2 · 2019-11-08T22:30:13.813

0

jqless and perlless sledgehammer using OP's intended sed/awk

s="Content"; sed 's/[{}"]//g' file | awk -v s=$s -F ":" -v RS=',' '($1 == s) {print $2}'

edited Nov 08 '19 at 22:30

answered Nov 08 '19 at 22:22

bu5hman

4,756

score 0 · Answer 3 · answered Nov 09 '19 at 12:49

0

OK, I admit that this is awful...

$ awk 'BEGIN {FPAT="\"[^\"]+\""; RS=","; pat=ARGV[1]; delete ARGV[1]} 
     $1 ~ pat {print $2} '       Content ex.json
"Value1"

answered Nov 09 '19 at 12:49

JJoao

12,170
1
23
45

Kusalananda · Answer 4 · 2021-05-15T21:00:31.890

0

$ jq -r '.Content' file
Value1
$ jq -r '.DeviceType' file
Value4

There is no reason to not install jq to parse the JSON properly, safely, and without having to care about decoding JSON-encoded strings manually etc.

The jq utility is available for download in a form that does not require root permissions for installation in your home directory (no external runtime dependencies).

edited May 15 '21 at 21:00

answered May 15 '21 at 20:43

Kusalananda

333,661

Parse value from different JSON strings (No jq)

4 Answers4

Linked