A simple, linear, non-recursive data-interchange format designed to be easily emitted, and parsed by any language.
The primary purpose of the fable format is to provide an easy way to store and transport multiple variables, and tables (in a more hygienic, CSV-like format) with required type declarations, and optional header name declarations. The goal was to create a format that could be emitted from any language as simply as possible (originally designed to be emitted from a Fortran95 codebase).
The current specification version is 0.2.0. (The %% directive is provided for forward compatibility with future versions of the specification/parser.)
example.fable
%% 0.2.0
# ^specification version - can be used by parser for
# backward/forward compatability
# comments work, empty lines are ignored, extra space between
# variables and delimiters is also ignored
# key-value declaration and supported types
integer my_int 10
float my_float 3.14
boolean my_bool true
boolean my_other_bool false
string my_string "hello world"
# there are also "nullable" types as modifiers of the
# "primitive" types that can accept `null`
integer? age null
boolean? is_present null
string? name null
# these cause a parsing error, for example
# integer years null # requires integer? declaration
# float salary null # requires float? declaration
# float types support scientific notation, infinity, and "not a number" values
float a_num 1.4151426
float b_num +1.4151426
float c_num -1.4151426
# scientific notation
float d_num 5.61e+13
float e_num 1e06
float f_num -8.71E-4
# separators work too
float g_num 218_411.897_871
# infinity
float h_num inf
float i_num +inf
float j_num -inf
# not a number
float k_num nan
float l_num +nan
float m_num -nan
# note, this is different than nullability!
# nan and null are two different concepts
float ok_float_1 nan
float ok_float_2 14.51
# float bad_float null # error, need to use: float?
float? ok_float_3 null # this is the correct way
float? ok_float_4 nan # also ok
# table declarations are similar but entries
# span over multiple rows and types are *required*
# each header name is delimited by a comma and requires quotes
# each column type follows the same rules as variables above
# table without a header
table example_table_no_header
integer,integer,integer,float,float,integer
0,1,83,0.01,7136,36
0,2,96,0.01,7579,39
0,3,95,0.02,7469,34
0,4,98,0.02,6227,38
0,5,96,0.04,7769,40
1,1,99,0.05,6219,36
1,2,89,0.01,6891,38
1,3,96,0.05,6926,39
1,4,80,0.05,7082,35
1,5,98,0.03,6200,39
# table with a header row
table+ example_table_with_header
integer,integer,integer,float,float,integer
"time","pipe_section","velocity","mass","pressure","temperature"
0,1,83,0.01,7136,36
0,2,96,0.01,7579,39
0,3,95,0.02,7469,34
0,4,98,0.02,6227,38
0,5,96,0.04,7769,40
1,1,99,0.05,6219,36
1,2,89,0.01,6891,38
1,3,96,0.05,6926,39
1,4,80,0.05,7082,35
1,5,98,0.03,6200,39
# table with some nullable columns and header
table+ students
integer,string,integer,float?
"id","name","grade","test_score"
0,"Jane Doe",2,98.1
1,"John Doe",3,null
2,"Other Doe",1,nan
The fable format is designed to be language agnostic and easily parsed by any language.
The following parser is built in Python and provides a familiar interface and almost identical usage to the builtin json library.
import fable
# load and parse the example file shown in the specification
with open('example.fable', 'r') as f:
document = fable.load(f)
print(document){'a_num': 1.4151426,
'age': None,
'b_num': 1.4151426,
'c_num': -1.4151426,
'd_num': 56100000000000.0,
'e_num': 1000000.0,
'example_table_no_header': {'header': None,
'values': [[0, 1, 83, 0.01, 7136.0, 36],
[0, 2, 96, 0.01, 7579.0, 39],
[0, 3, 95, 0.02, 7469.0, 34],
[0, 4, 98, 0.02, 6227.0, 38],
[0, 5, 96, 0.04, 7769.0, 40],
[1, 1, 99, 0.05, 6219.0, 36],
[1, 2, 89, 0.01, 6891.0, 38],
[1, 3, 96, 0.05, 6926.0, 39],
[1, 4, 80, 0.05, 7082.0, 35],
[1, 5, 98, 0.03, 6200.0, 39]]},
'example_table_with_header': {'header': ['time',
'pipe_section',
'velocity',
'mass',
'pressure',
'temperature'],
'values': [[0, 1, 83, 0.01, 7136.0, 36],
[0, 2, 96, 0.01, 7579.0, 39],
[0, 3, 95, 0.02, 7469.0, 34],
[0, 4, 98, 0.02, 6227.0, 38],
[0, 5, 96, 0.04, 7769.0, 40],
[1, 1, 99, 0.05, 6219.0, 36],
[1, 2, 89, 0.01, 6891.0, 38],
[1, 3, 96, 0.05, 6926.0, 39],
[1, 4, 80, 0.05, 7082.0, 35],
[1, 5, 98, 0.03, 6200.0, 39]]},
'f_num': -0.000871,
'g_num': 218411.897871,
'h_num': inf,
'i_num': inf,
'is_present': None,
'j_num': inf,
'k_num': nan,
'l_num': nan,
'm_num': nan,
'my_bool': True,
'my_float': 3.14,
'my_int': 10,
'my_other_bool': False,
'my_string': 'hello world',
'name': None,
'ok_float_1': nan,
'ok_float_2': 14.51,
'ok_float_3': None,
'ok_float_4': nan,
'students': {'header': ['id', 'name', 'grade', 'test_score'],
'values': [[0, 'Jane Doe', 2, 98.1],
[1, 'John Doe', 3, None],
[2, 'Other Doe', 1, nan]]}}