13

I was hoping to curl data into a shell script variable, then use cat or awk to operate on that variable as if it was a file.

Is this possible, or is there a workaround? Do I have to save the curl command into a file?

Do I have to run curl several times, for each bit of string formatting I want to do?

I am curling ipinfo.io, and want to output the matching lines for "city" and "region". I know how to use awk and sed to format the output as desired.

Try for example, curl ipinfo.io/"8.8.8.8" for the result for a Google DNS IP address.

I'd like to print just "Mountain View" and "California," preferably on the same line, with some common formatting, like "Mountain View, California".

5
  • 1
    Whats your input and desired output?
    – heemayl
    Dec 29, 2015 at 6:04
  • You can, but there's no point in storing curl's output (unless you want to use it multiple times), you can directly feed it to cat or awk. As heemayl said please explain exactly what you want to do.
    – kos
    Dec 29, 2015 at 6:06
  • i am curling ipinfo.io, and want to echo the matching lines for city and region. I know how to use awk and sed to format the output as desired. Try for example, curl ipinfo.io/"8.8.8.8" for the result for a google dns ip address
    – cfye14
    Dec 29, 2015 at 6:11
  • What do you want to print exactly? The whole line as in the output or only "Mountain View" and "California"?
    – kos
    Dec 29, 2015 at 6:23
  • 1
    just "Mountain View" and "California," preferably on the same line, with some common formatting, like you know, "Mountain View, California"
    – cfye14
    Dec 29, 2015 at 6:26

4 Answers 4

15

In general there are more appropriate ways of parsing JSON objects, but since in this case the JSON object is very simple you may store curl's output in a variable (which is possible) and just use AWK:

var="$(curl ipinfo.io/"8.8.8.8" 2>/dev/null)"
<<<"$var" awk -F'"' '$2=="city"{printf("%s, ", $4)}$2=="region"{print $4}'
% var="$(curl ipinfo.io/"8.8.8.8" 2>/dev/null)"
% <<<"$var" awk -F'"' '$2=="city"{printf("%s, ", $4)}$2=="region"{print $4}'
Mountain View, California

However unless you want to use curl's output multiple times you may just use a pipe:

curl ipinfo.io/"8.8.8.8" 2>/dev/null | awk -F'"' '$2=="city"{printf("%s, ", $4)}$2=="region"{print $4}'
curl ipinfo.io/"8.8.8.8" 2>/dev/null | awk -F'"' '$2=="city"{printf("%s, ", $4)}$2=="region"{print $4}'
Mountain View, California

<<< is a form of input redirection called "here string"; it redirects the STDIN of a command from the terminal to a string.

What happens here is that $var is expanded between the double quotes; the STDIN of the AWK command is redirected from the terminal to the expanded string and AWK consequently reads the string as its input file.

2
  • can you explain the <<< part a bit more? edit: this I take it?
    – cfye14
    Dec 29, 2015 at 6:56
  • 1
    @cfye14 I added an explanation for the <<< token: one thing, note that I asked all those informations to be extremely sure of the scope of your question (to avoid anyone giving potentially unuseful answers); I understand that the processing itself was probably not necessary. janos' answer below is just as much as valid.
    – kos
    Dec 29, 2015 at 7:12
6

I was hoping to curl data into a shell script variable, then use cat or awk to operate on that variable as if it were a file. Is this possible, or is there a workaround?

Sure you can. If the content is not too large, and if you want to run multiple commands to parse the content, then it's a good thing to cache it in memory rather than re-download every time.

To store the result of curl in a variable:

ipinfo=$(curl ipinfo.io/8.8.8.8)

To run commands on it:

<<< "$ipinfo" awk ...
<<< "$ipinfo" sed ...

The double-quotes no `"$ipinfo" are important to preserve all the whitespace characters.

A "workaround" to not saving the content in a variable is to figure out a way to process the content in a single pipeline, like @kos did.

0
2

Another easy and more readable way to do this, you can use python, which exists in all linux based systems, to parse the json for you

Simply, Import json and sys packages to load the json object, then print from the result object any property you need:

curl ipinfo.io/"8.8.8.8" 2>/dev/null | 
  python -c 'import  json,sys; 
             result=json.load(sys.stdin); 
             print(result["'city'"] + ", " + result["'region'"])';

I added the script in a more readable way however you should put in one line or add \ by the end of each line if you need to write it in multiple lines.

1

Since the output of curl will be JSON the appropriate/simpler tool seems to be jq (a lightweight and flexible command-line JSON processor):

$ curl ipinfo.io/"1.1.1.1" 2>/dev/null | jq --raw-output '"\(.city), \(.region)"'
White Hall, Alabama

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .