Compatibility with JSON

The Jonquil project provides a flavor of TOML Fortran which can handle JSON. It is available from the `toml-f/jonquil <https://github.com/toml-f/jonquil`__ repository and can be included using the same steps as shown for TOML Fortran in Using TOML Fortran.

Tip

All types and procedures in TOML Fortran and Jonquil are fully compatible. Everything that is possible with TOML Fortran can be done with Jonquil, they can be mixed and matched without restriction. In short, Jonquil is TOML Fortran.

Using Jonquil

You can follow all recipes and tutorials in this documentation using Jonquil by just replacing the tomlf module with the jonquil module. All derived types, procedure names and interfaces use the json_ prefix instead of toml_, with the exception of the toml_table type which becomes a json_object.

The following example program shows how to load JSON data from a string and access the value using the build interface.

app/demo.f90

program demo
  use jonquil, only : json_loads, json_value, json_object, json_error, cast_to_object, &
    & get_value
  implicit none

  class(json_value), allocatable :: val
  type(json_object), pointer :: object
  type(json_error), allocatable :: error
  integer :: ival

  call json_loads(val, '{"a":1,"b":2}', error=error)
  if (allocated(error)) then
    print '(a)', error%message
    stop
  end if

  object => cast_to_object(val)
  if (associated(object)) then
    call get_value(object, "a", ival)
    print '(a,1x,i0)', "a is", ival

    call get_value(object, "b", ival)
    print '(a,1x,i0)', "b is", ival
  end if

In contrast to TOML Fortran the loaded data structure is always returned as a polymorphic value, which can be dispatched by using a select type construct or by obtaining a pointer using cast_to_object / cast_to_array as appropriate.

Jonquil promises seamless compatibility with TOML Fortran, the data structure we just loaded can be manipulated with any procedure from the tomlf module and vice versa.

app/demo.f90

  block
    use tomlf, only : toml_table, toml_array, add_array, set_value, toml_serialize

    integer :: it
    type(toml_table), pointer :: table
    type(toml_array), pointer :: array

    ! Add an array to the object
    call add_array(object, "c", array)
    call set_value(array, [-1, 0, 1])

    ! Create a new subobject / subtable
    call get_value(object, "d", table, requested=.true.)
    call set_value(table, "sub", "table")

    print '(a)', "# representation in TOML land"
    print '(a)', toml_serialize(object)
  end block

Since the json_object is just a flavor of the toml_table type, it can be used in the same way as any other table.