Prepend string to each key in JSON file

You can use a JSON parser such as jq to parse your JSON file. The advantage of this is that valid JSON can be handled regardless of its shape:

jq 'with_entries(.key |= "x_" + .)' <file.json

Input

{"key_1": 12, "key_2": 34, "key_3": 56}

Output

{
  "x_key_1": 12,
  "x_key_2": 34,
  "x_key_3": 56
}

You can keep the JSON on one line with jq -c …:

{"x_key_1":12,"x_key_2":34,"x_key_3":56}

This solution is shown as an example in the documentation for jq - see man jq and search for with_entries - but as it's a reference document it can be hard to find things you don't know how to describe.

Taking the question in the title literally, the following appends some user-provided string to all keys in the input document recursively:

jq --arg prefix x_ '
    walk(
        if type == "object" then 
            with_entries(.key = $prefix + .key)
        else . end
    )' file

The key is updated similarly to what's done in roaima's answer, but without using the update operator. The prefix string is also given on the command line instead of being a hard-coded string.

Test run:

$ cat file
{
  "key_1": 12,
  "key_2": 34,
  "key_3": [
    {
      "key_1": "{\"key_A\":123}",
      "key_2": 34,
      "key_3": 56
    },
    {
      "key_1": "{\"key_A\":123}",
      "key_2": 34,
      "key_3": 56
    }
  ]
}

$ jq --arg prefix x_ 'walk(if type == "object" then with_entries(.key = $prefix + .key) else . end)' file
{
  "x_key_1": 12,
  "x_key_2": 34,
  "x_key_3": [
    {
      "x_key_1": "{\"key_A\":123}",
      "x_key_2": 34,
      "x_key_3": 56
    },
    {
      "x_key_1": "{\"key_A\":123}",
      "x_key_2": 34,
      "x_key_3": 56
    }
  ]
}

If your json format is in one-line then the following should work:

sed -i 's/"\([^"]*\)":/"x_\1":/g' json_file.json

If your json format looks like this:

{
   "key_1": 1, 
   "key_2": "v2", 
   "key_3": 3
}

You could use the sed command above but you also can use the following sed which is a little easier:

sed -i 's/"\(.*\)":/"x_\1":/'  json_file.json

Note: The solution above works well if you just have numbers or strings that don't have the following pattern:

{ "key_1": "\"Sale\": foobar" }
// Or
{ "\"key_1\"": "\"Sale\": foobar" } // I think  this case is less common

So the best way to edit the keys correctly is by using the roaima's answer. I tried another possible solution with sed (but I'm not sure if there are other cases where it might fail). If your json file looks like this:

{"key_1": "\"Sale\": foobar", "key_2": "\"3\":", "key_3": 56}
// Or maybe
{"key_1": "\"Sale:sale\": foobar", "key_2": "\"3\":", "key_3": 56}

you can try using:

sed -i 's/"\([^"\][^"\]*\)":/"x_\1":/g' json_file.json

However the command above will fail if the json file looks like this:

{"\"key_1\"": "\"Sale:\": foobar", "key_2": "\"3\":", "key_3": 56}
// If the json keys  have \"somekey\" as their keys then the sed command will fail. 
// So using sed for this task might be a little tricky for avoiding these problems.