---

5.4.3 JSON data encoding/decoding ¶

JavaScript Object Notation, in short JSON, is a very common human readable and structured data format. GNU Octave supports encoding and decoding this format with the following two functions.

 

: JSON_txt = jsonencode (object) ¶

: JSON_txt = jsonencode (…, "ConvertInfAndNaN", TF) ¶

: JSON_txt = jsonencode (…, "PrettyPrint", TF) ¶

Encode Octave data types into JSON text.

The input object is an Octave variable to encode.

The output JSON_txt is the JSON text that contains the result of encoding object.

If the value of the option "ConvertInfAndNaN" is true then NaN, NA, -Inf, and Inf values will be converted to "null" in the output. If it is false then they will remain as their original values. The default value for this option is true.

If the value of the option "PrettyPrint" is true, the output text will have indentations and line feeds. If it is false, the output will be condensed and written without whitespace. The default value for this option is false.

Programming Notes:

• Complex numbers are not supported.
• classdef objects are first converted to structs and then encoded.
• To preserve escape characters (e.g., "\n"), use single-quoted strings.
• Every character after the null character ("\0") in a double-quoted string will be dropped during encoding.
• Encoding and decoding an array is not guaranteed to preserve the dimensions of the array. In particular, row vectors will be reshaped to column vectors.
• Encoding and decoding is not guaranteed to preserve the Octave data type because JSON supports fewer data types than Octave. For example, if you encode an int8 and then decode it, you will get a double.

This table shows the conversions from Octave data types to JSON data types:

Octave data type	JSON data type
logical scalar	Boolean
logical vector	Array of Boolean, reshaped to row vector
logical array	nested Array of Boolean
numeric scalar	Number
numeric vector	Array of Number, reshaped to row vector
numeric array	nested Array of Number
NaN, NA, Inf, -Inf when "ConvertInfAndNaN" = true	"null"
NaN, NA, Inf, -Inf when "ConvertInfAndNaN" = false	"NaN", "NaN", "Infinity", "-Infinity"
empty array	"[]"
character vector	String
character array	Array of String
empty character array	""
cell scalar	Array
cell vector	Array, reshaped to row vector
cell array	Array, flattened to row vector
struct scalar	Object
struct vector	Array of Object, reshaped to row vector
struct array	nested Array of Object
classdef object	Object

Examples:

jsonencode ([1, NaN; 3, 4])
⇒ [[1,null],[3,4]]

jsonencode ([1, NaN; 3, 4], "ConvertInfAndNaN", false)
⇒ [[1,NaN],[3,4]]

## Escape characters inside a single-quoted string
jsonencode ('\0\a\b\t\n\v\f\r')
⇒ "\\0\\a\\b\\t\\n\\v\\f\\r"

## Escape characters inside a double-quoted string
jsonencode ("\a\b\t\n\v\f\r")
⇒ "\u0007\b\t\n\u000B\f\r"

jsonencode ([true; false], "PrettyPrint", true)
⇒ ans = [
     true,
     false
   ]

jsonencode (['foo', 'bar'; 'foo', 'bar'])
⇒ ["foobar","foobar"]

jsonencode (struct ('a', Inf, 'b', [], 'c', struct ()))
⇒ {"a":null,"b":[],"c":{}}

jsonencode (struct ('structarray', struct ('a', {1; 3}, 'b', {2; 4})))
⇒ {"structarray":[{"a":1,"b":2},{"a":3,"b":4}]}

jsonencode ({'foo'; 'bar'; {'foo'; 'bar'}})
⇒ ["foo","bar",["foo","bar"]]

jsonencode (containers.Map({'foo'; 'bar'; 'baz'}, [1, 2, 3]))
⇒ {"bar":2,"baz":3,"foo":1}

See also: jsondecode.

 

: object = jsondecode (JSON_txt) ¶

: object = jsondecode (…, "ReplacementStyle", rs) ¶

: object = jsondecode (…, "Prefix", pfx) ¶

: object = jsondecode (…, "makeValidName", TF) ¶

Decode text that is formatted in JSON.

The input JSON_txt is a string that contains JSON text.

The output object is an Octave object that contains the result of decoding JSON_txt.

For more information about the options "ReplacementStyle" and "Prefix", see matlab.lang.makeValidName.

If the value of the option "makeValidName" is false then names will not be changed by matlab.lang.makeValidName and the "ReplacementStyle" and "Prefix" options will be ignored.

NOTE: Decoding and encoding JSON text is not guaranteed to reproduce the original text as some names may be changed by matlab.lang.makeValidName.

This table shows the conversions from JSON data types to Octave data types:

JSON data type	Octave data type
Boolean	scalar logical
Number	scalar double
String	vector of characters
Object	scalar struct (field names of the struct may be different from the keys of the JSON object due to matlab_lang_makeValidName
null, inside a numeric array	NaN
null, inside a non-numeric array	empty double array []
Array, of different data types	cell array
Array, of Booleans	logical array
Array, of Numbers	double array
Array, of Strings	cell array of character vectors (cellstr)
Array of Objects, same field names	struct array
Array of Objects, different field names	cell array of scalar structs

Examples:

jsondecode ('[1, 2, null, 3]')
    ⇒ ans =

      1
      2
    NaN
      3

jsondecode ('["foo", "bar", ["foo", "bar"]]')
    ⇒ ans =
       {
         [1,1] = foo
         [2,1] = bar
         [3,1] =
         {
           [1,1] = foo
           [2,1] = bar
         }

       }

jsondecode ('{"nu#m#ber": 7, "s#tr#ing": "hi"}', ...
            'ReplacementStyle', 'delete')
    ⇒ scalar structure containing the fields:

         number = 7
         string = hi

jsondecode ('{"nu#m#ber": 7, "s#tr#ing": "hi"}', ...
            'makeValidName', false)
    ⇒ scalar structure containing the fields:

         nu#m#ber = 7
         s#tr#ing = hi

jsondecode ('{"1": "one", "2": "two"}', 'Prefix', 'm_')
    ⇒ scalar structure containing the fields:

         m_1 = one
         m_2 = two

See also: jsonencode, matlab.lang.makeValidName.