Replace third word in double quotes from file

Question

This is the json file that we want to edit

Example 1

{
  "topics": [{"topic": "hfgt_kejd_tdsfd”}],
  "version": 1
}

Example 2

{
  "topics": [{"topic": "hfgt_kj_fsrfdfd”}],
  "version": 1
}

We want to replace the third word in the line topics with other word ( by sed or perl one liner )

regarding to example 1 , Expected results when we want to replace hfgt_kejd_tdsfd with test_1

{
  "topics": [{"topic": "test_1”}],
  "version": 1
}

example 3

more /tmp/file.json

{
  "topics": [{"topic": "TOPIC_GGGG”}],
  "version": 1
}


# sed  's/\(topic\": "\)[a-z_]*/\1test_1/'  /tmp/file.json

{
  "topics": [{"topic": "test_1TOPIC_GGGG”}],
  "version": 1
}

Answer 1

Using jq:

$ jq '.topics[0].topic |= "test_1"' file.json
{
  "topics": [
    {
      "topic": "test_1"
    }
  ],
  "version": 1
}

This reads the JSON document and modifies the value of the topic entry of the first element of the array topics to the string test_1.

If you have your value in a variable (encoded in UTF-8):

$ val='Some value with "double quotes"'
$ jq --arg string "$val"  '.topics[0].topic |= $string' file.json
{
  "topics": [
    {
      "topic": "Some value with \"double quotes\""
    }
  ],
  "version": 1
}

---

Using Perl:

$ perl -MJSON -0777 -e '$h=decode_json(<>); $h->{topics}[0]{topic}="test_1"; print encode_json($h), "\n"' file.json
{"topics":[{"topic":"test_1"}],"version":1}

With a variable:

$ val='Some value with "double quotes"'
$ STRING=$val perl -MJSON -0777 -e '$string = $ENV{STRING}; utf8::decode $string; $h=decode_json(<>); $h->{topics}[0]{topic}=$string; print encode_json($h), "\n"' file.json
{"topics":[{"topic":"Some value with \"double quotes\""}],"version":1}

Both of these uses the Perl JSON module to decode the JSON document, change the value that needs changing, and then output the re-encoded data structure. The error handling is left as an exercise.

For the second piece of code, the value to be inserted is passed as an environment variable, STRING, into the Perl code. This is due to reading the JSON document from file in "slurp" mode with -0777.

Answer 2

Instead of matching [A-Za-z_] (which by the way matches a lot more than the English letters and underscore in most locales), use [^"] for any character other than " (by the way, the sample in your question has a ” (U+201D, RIGHT DOUBLE QUOTATION MARK) instead of a json quote (", U+0022, QUOTATION MARK) as the closing quote after the topic name, I'll assume that's not the case in your original files as that would make it invalid json):

sed 's/\("topic": "\)[^"]*/\1test_1/' < file.json

(note that it's your responsibility to make sure the topic name replacement is properly JSON-encoded: UTF-8 charset, control characters encoded (like \n for newline), and " encoded as \").

Answer 3

sed -i 's/\(topic\": "\)[A-Za-z_]*/\1test_1/' file.json