Micro Object Notation
by Douglas Lau
MuON is a text format for data serialization. It is suitable for configuration files and data interchange — as expressive as other formats, but much simpler.
# MuON example
movie: Alien
director: Ridley Scott
cast:=Sigourney Weaver
:=Tom Skerritt
:=John Hurt
release: 1979-06-22
region: USA
release: 1979-09-06
region: UK
gross: 203_630_630
emoji: 👽 👾
MuON is Unicode text encoded in UTF-8, with no byte-order mark. MuON is case sensitive and line based.
A line is a definition, comment or blank, and must end with a single line
feed character (U+000A). Comments begin with a number sign #, which may
be preceded by spaces. Blank lines contain no characters.
# Example comment
A definition maps a key to a value, with a colon and space between.
key: value
If the value is empty, the space is not required.
Some definitions create branches. Starting from a root record, all branches form a tree. With no indents, definitions are contained in the root. After a branch, subsequent definitions with one more indent are contained in it.
key_in_root: value in root
branch:
key_in_branch: value in branch
Definition indents are exactly 2, 3 or 4 spaces (U+0020). Nested branches use multiple indents. The number of spaces must be the same for all indents in a file.
family: Ursidae
# One indent; 3 spaces
genus: Ailuropoda
# Two indents; 6 spaces
species: A. melanoleuca 🐼
A key is a sequence of one or more characters. It must be "quoted"
if it contains a colon or begins with a space, quote mark or number sign. In
this case, all quote marks in the key must be doubled.
# "skeleton" key begins with a quote mark
"""skeleton"" key": value
Also, a key should be quoted if it begins with whitespace, contains control characters or homoglyphs of colon.
A value is a sequence of characters. With the exception of line feed, any Unicode character is allowed.
A schema is a template with types as values. It can be separate or prepended to a MuON file. In either case, it begins and ends with a line of three colons.
:::
# Example MuON schema
movie: list record
title: text
director: text Alan Smithee
cast: list text
release: list record
release_date: date >=1878-01-01
region: text
gross: int
emoji: optional text
:::
There are eleven available types: text, bool, int, number,
datetime, date, time, record, choice, dictionary and any. They
are used to parse objects from values.
An optional or list modifier may precede the type, followed by a space.
One or two constraints may follow the type, with a space between. This is
one of four specifiers >, >=, < or <=, followed by a value. For int,
number, datetime, date and time types, it defines a subrange of valid
values. For text, it restricts the count of characters.
A default value may be included after the type and any constraints, also
separated with a space. It is a value for the type, used when a definition is
not present. Allowed types are text, bool, int, number, datetime,
date and time. Defaults are not allowed with optional or list
modifiers.
Text is a sequence of characters.
:::
greeting: text Hello!
farewell: text Goodbye!
:::
# greeting is Hello!
farewell: Be seeing you.
Because values cannot contain line feeds, they can only be represented using
multiple definitions. The text must be split into values between each line
feed. For each definition after the first, use a text append separator,
which is :> instead of the usual : before the value.
When appending, use a blank key — a sequence of spaces with the same number of characters as the key.
lyric: Out in the garden
:>There's half of a heaven
Bool is a boolean: either true or false.
earth_is_flat: false
Int is an integer (whole number) in one of three forms:
- Decimal: sequence of digits
0-9. May have a+or-sign prefix - Binary:
bfollowed by sequence of digits0or1 - Hexadecimal:
xfollowed by sequence of digits0-9,A-Fora-f
An underscore may be inserted between digits to improve readability.
locke: 4
reyes: b1000
ford: x0F
jarrah: +16
shephard: b01_0111
kwon: x2a
If no constraints are given, an integer has no bounds (BigInt).
:::
uint8: int >=0 <=255
rank: int >0 <6
:::
uint8: 49
rank: 3
A number is a 64-bit floating point number, made up of these parts:
- Whole number part (same as decimal int)
- Fractional part (decimal point followed by sequence of digits
0-9) - Exponent part (
efollowed by decimal int)
One or both of the whole or fractional parts must be present, but the exponent part is not required. As with ints, underscores may be included.
The values inf and NaN stand for infinity and not a number,
respectively. Either can be prefixed with a + or - sign.
prime: 37
log_e_2: .6931471805599453
mercury: -38.83440
planck: 6.626_070_15e-34
buzz: +inf
avogadro: 6.022_140_76e23
Datetime is date, time and offset, as specified by date-time
from RFC 3339. The date and
time are separated by an uppercase T only. If the offset is represented by
Z, it must also be uppercase.
moonwalk: 1969-07-21T02:56:00Z
Date is year, month and day, as specified by full-date from
RFC 3339.
birthday: 2019-08-01
Time is hour, minute and second, as specified by partial-time from
RFC 3339.
start: 08:00:00
end: 15:58:14.593849001
A record is a branch containing fields as subsequent definitions. A record represents all of its fields.
:::
book: record
title: text
author: text
year: int
:::
book:
title: If on a winter's night a traveler
author: Italo Calvino
year: 1979
Field keys are often used in programming languages as identifiers. For compatibility, they should contain only ASCII alphanumeric or underscore characters.
Since records do not use their values, they can substitute for the first
field, which must then be left out. Like defaults, substitution is only
allowed for text, bool, int, number, datetime, date and time types
without optional or list modifiers.
book: The Left Hand of Darkness
author: Ursula K. Le Guin
year: 1969
In the schema, a record id can follow record after a space. It is used
if a record exists in more than one place. After the first definition, the
fields do not need to be included.
:::
player: record Character
name: text
health: int
nemesis: record Character
:::
player: Arthur
health: 50
nemesis: Mordred
health: 60
A choice is a branch containing variants as subsequent definitions. A choice represents exactly one of the variants.
A variant containing no data is declared as a definition with no type. These variants can substitute for the choice value.
:::
pill: choice
red
blue
:::
pill: red
Variants can also contain arbitrary data. These must be declared with a subsequent definition.
:::
strategy: choice
attack: int
retreat
surrender: text
:::
strategy:
attack: 50
In the schema, a variant id can follow choice after a space. This works
in the same way as a record id.
:::
face_a: choice direction
North
South
East
West
face_b: choice direction
:::
face_a: North
face_b: East
A dictionary is a branch for associative arrays — useful if keys are not
known in advance. The schema must contain a single definition with types for
both key and value. The key type is restricted to text, bool, int,
number, datetime, date or time.
:::
num_word: dictionary
text: int
:::
num_word:
fifty: 50
one: 1
thirteen: 13
Any is a branch containing data of any type. It should be used for data which does not fit into a rigid schema.
:::
product: list record
name: text
price: number
details: any
:::
product: duct tape
price: 4.99
details:
color: silver
width: 8 cm
product: machete
price: 29.99
details:
length: 50 cm
weight: 0.5 kg
Optional types are not required — the absence of a definition represents
a None or null value.
:::
name: text
occupation: optional text
:::
name: Surfer Joe
# no occupation
A list is parsed as a sequence of objects, separated by spaces. If a list is empty, omit its definition.
:::
show_times: list time
healthy_snacks: list text
:::
show_times: 15:40:00 18:00:00 20:20:00
# no healthy_snacks
Like text, lists can be appended. All objects are added to the end.
fibonacci: 0 1 1 2 3
: 5 8 13 21 34
# same as fibonacci: 0 1 1 2 3 5 8 13 21 34
When appending to list record or list dictionary, the key cannot be blank, since the definitions are not consecutive.
person: George Washington
birthday: 1732-02-22
person: Abraham Lincoln
birthday: 1809-02-12
For list text, objects are separated by spaces, just like other lists. If
a text object contains spaces, use the text value separator := to treat
an entire value as a single object. The text append separator :> will
also append an entire value to the previous text object.
shopping: avocado banana
:=cream cheese
: cucumber
:=ice cream
: raw
:>burger! (mmmm)
Any feedback, bug reports, spelling fixes, or text clarity improvements are welcome! Please create an issue.