The Koto Language Guide

As you're reading this guide, you're encouraged to play around with the examples to get a feel for the language.

When you see a icon below an example, clicking it will open the example in the Koto Playground, where you can run the code and see what happens as you make changes.

Language Basics

Koto programs contain a series of expressions that are evaluated by Koto's runtime.

As an example, this simple script prints a friendly greeting.

name = 'World'
print 'Hello, {name}!'

Comments

Single-line comments start with a #.

# This is a comment, everything until the end of the line is ignored.

Multi-line comments start with #- and end with -#.

#- 
This is a 
multi-line 
comment.
-#

Numbers

Numbers and arithmetic are expressed in a familiar way.

1
# -> 1

1 + 1
# -> 2

-1 - 10
# -> -11

3 * 4
# -> 12

9 / 2
# -> 4.5

12 % 5
# -> 2

Parentheses

Arithmetic operations follow the conventional order of precedence. Parentheses can be used to group expressions as needed.

# Without parentheses, multiplication is performed before addition
1 + 2 * 3 + 4
# -> 11
# With parentheses, the additions are performed first
(1 + 2) * (3 + 4)
# -> 21

Booleans

Booleans are declared with the true and false keywords, and combined using the and and or operators.

true and false
# -> false

true or false
# -> true

Booleans can be negated with the not operator.

not true
# -> false

not false
# -> true

Values can be compared for equality with the == and != operators.

1 + 1 == 2
# -> true

99 != 100
# -> true

Null

The null keyword is used to declare a value of type Null, which indicates the absence of a value.

null
# -> null

Truthiness

In boolean contexts (such as logical operations), null is treated as being equivalent to false. Every other value in Koto evaluates as true.

not null
# -> true

null or 42
# -> 42

Assigning Variables

Values are assigned to named identifiers with =, and can be freely reassigned. Named values like this are known as variables.

# Assign the value `42` to `x`
x = 42
x
# -> 42

# Replace the existing value of `x` 
x = true
x
# -> true

Compound assignment operators are also available. For example, x *= y is a simpler way of writing x = x * y.

a = 100
a += 11
# -> 111
a
# -> 111

a *= 10
# -> 1110
a
# -> 1110

Debug

The debug keyword allows you to quickly display a value while working on a program.

It prints the result of an expression, prefixed with its line number and the original expression as a string.

x = 10 + 20
debug x / 10
# -> [2] x / 10: 3.0

When using debug, the displayed value is also the result of the expression, which can be useful if you want to quickly get feedback during development.

x = debug 2 + 2
# -> [1] 2 + 2: 4
x
# -> 4

Lists

Lists in Koto are created with [] square brackets and can contain a mix of different value types.

Access list elements by index using square brackets, starting from 0.

x = [99, null, true]
x[0]
# -> 99
x[1]
# -> null

x[2] = false
x[2]
# -> false

Once a list has been created, its underlying data is shared between other instances of the same list. Changes to one instance of the list are reflected in the other.

# Assign a list to x
x = [10, 20, 30]

# Assign another instance of the list to y
y = x

# Modify the list through y
y[1] = 99

# The change to y is also reflected in x
x 
# -> [10, 99, 30]

Joining Lists

The + operator allows lists to be joined together, creating a new list that contains their concatenated elements.

a = [98, 99, 100]
b = a + [1, 2, 3]
b
# -> [98, 99, 100, 1, 2, 3]

Tuples

Tuples in Koto are similiar to lists, but are designed for sequences of values that have a fixed structure.

Unlike lists, tuples can't be resized after creation, and values that are contained in the tuple can't be replaced.

Tuples are declared with a series of expressions separated by commas.

x = 100, true, -1
x
# -> (100, true, -1)

Parentheses can be used for grouping to avoid ambiguity.

(1, 2, 3), (4, 5, 6)
# -> ((1, 2, 3), (4, 5, 6))

Access tuple elements by index using square brackets, starting from 0.

x = false, 10
# -> (false, 10)
x[0]
# -> false
x[1]
# -> 10

y = true, 20
# -> (true, 20)
x, y
# -> ((false, 10), (true, 20))

Joining Tuples

The + operator allows tuples to be joined together, creating a new tuple containing their concatenated elements.

a = 1, 2, 3
b = a + (4, 5, 6)
b
# -> (1, 2, 3, 4, 5, 6)

Creating Empty Tuples

An empty pair of parentheses in Koto resolves to null. If an empty tuple is needed then use a single , inside parentheses.

# An empty pair of parentheses resolves to null
() 
# -> null

# A comma inside parentheses creates a tuple 
(,) 
# -> ()

Tuple Mutability

While tuples have a fixed structure and its contained elements can't be replaced, mutable value types (like lists) can be modified while they're contained in tuples.

# A Tuple containing two lists
x = ([1, 2, 3], [4, 5, 6])

# Modify the second list in the tuple
x[1][0] = 99
x
# -> ([1, 2, 3], [99, 5, 6])

Strings

Strings in Koto contain a sequence of UTF-8 encoded characters, and can be declared using ' or " quotes.

'Hello, World!'
# -> Hello, World!

"Welcome to Koto πŸ‘‹"
# -> Welcome to Koto πŸ‘‹

Strings can start on one line and finish on another.

'This is a string
that spans
several lines.'
# -> This is a string
# -> that spans
# -> several lines.

Strings can be joined together with the + operator.

'a' + 'Bc' + 'Def'
# -> aBcDef

String Interpolation

Variables can be easily included in a string by surrounding them with {} curly braces.

xyz = 123
'The value of xyz is {xyz}'
# -> The value of xyz is 123

Including variables in a string this way is known as string interpolation.

Simple expressions can also be interpolated using the same syntax.

'2 plus 3 is {2 + 3}.'
# -> 2 plus 3 is 5.

String Escape Codes

Strings can contain the following escape codes to define special characters, all of which start with a \.

  • \n: Newline
  • \r: Carriage Return
  • \t: Tab
  • \': Single quote
  • \": Double quote
  • \\: Backslash
  • \{: Interpolation start
  • \u{NNNNNN}: Unicode character
    • Up to 6 hexadecimal digits can be included within the {} braces. The maximum value is \u{10ffff}.
  • \xNN: ASCII character
    • Exactly 2 hexadecimal digits follow the \x.
'\{\'\"}'
# -> {'"}
'Hi \u{1F44B}'
# -> Hi πŸ‘‹

Continuing a Long Line

The end of a line can be escaped with a \, which will skip the newline and any leading whitespace on the next line.

foo = "This string \
       doesn't contain \
       newlines."
foo
# -> This string doesn't contain newlines.

Single or Double Quotes

Both single ' and double " quotes are valid for defining strings in Koto and can be used interchangeably.

A practical reason to choose one over the other is that the alternate quote type can be used in a string without needing to use escape characters.

print 'This string has to escape its \'single quotes\'.'
# -> This string has to escape its 'single quotes'.

print "This string contains unescaped 'single quotes'."
# -> This string contains unescaped 'single quotes'.

String Indexing

Individual bytes of a string can be accessed via indexing with [] braces.

'abcdef'[3]
# -> d
'xyz'[1..]
# -> yz

Care must be taken when using indexing with strings that could contain non-ASCII data. If the indexed bytes would produce invalid UTF-8 data then an error will be thrown. To access unicode characters see string.chars.

Raw Strings

When a string contains a lot of special characters, it can be preferable to use a raw string.

Raw strings ignore escape characters and interpolated expressions, providing the raw contents of the string between its delimiters.

Raw strings use single or double quotes as the delimiter, prefixed with an r.

print r'This string contains special characters: {foo}\n\t.'
# -> This string contains special characters: {foo}\n\t.

For more complex string contents, the delimiter can be extended using up to 255 # characters after the r prefix,

print r#'This string contains "both" 'quote' types.'#
# -> This string contains "both" 'quote' types.

print r##'This string also includes a '#' symbol.'##
# -> This string also includes a '#' symbol.

Functions

Functions in Koto are created using a pair of vertical bars (||), with the function's arguments listed between the bars. The body of the function follows the vertical bars.

hi = || 'Hello!'
add = |x, y| x + y

Functions are called with arguments contained in () parentheses. The body of the function is evaluated and the result is returned to the caller.

hi = || 'Hello!'
hi()
# -> Hello!

add = |x, y| x + y
add(50, 5)
# -> 55

The parentheses for arguments when calling a function are optional and can be ommitted in simple expressions.

square = |x| x * x
square 8
# -> 64

add = |x, y| x + y
add 2, 3
# -> 5

# Equivalent to square(add(2, 3))
square add 2, 3 
# -> 25

A function's body can be an indented block, where the last expression in the body is evaluated as the function's result.

f = |x, y, z|
  x *= 100
  y *= 10
  x + y + z
f 2, 3, 4
# -> 234

Return

When the function should be exited early, the return keyword can be used.

f = |n|
  return 42
  # This expression won't be reached
  n * n
f -1
# -> 42

If a value isn't provided to return, then the returned value is null.

f = |n|
  return
  n * n
f 123
# -> null

Function Piping

The pipe operator (>>) can be used to pass the result of one function to another, working from left to right. This is known as function piping, and can aid readability when working with a long chain of function calls.

add = |x, y| x + y
multiply = |x, y| x * y
square = |x| x * x

# Chained function calls can be a bit hard to follow for the reader.
x = multiply 2, square add 1, 3
x
# -> 32

# Parentheses don't help all that much...
x = multiply(2, square(add(1, 3)))
x
# -> 32

# Piping allows for a left-to-right flow of results.
x = add(1, 3) >> square >> multiply 2
x
# -> 32

# Call chains can also be broken across lines.
x = add 1, 3
  >> square 
  >> multiply 2
x
# -> 32

Maps

Maps in Koto are containers that contain a series of entries with keys that correspond to associated values.

The . dot operator returns the value associated with a particular key.

Maps can be created using inline syntax with {} braces:

m = {apples: 42, oranges: 99, lemons: 63}

# Get the value associated with the `oranges` key
m.oranges
# -> 99

...or using block syntax with indented entries:

m = 
  apples: 42
  oranges: 99
  lemons: 63
m.apples
# -> 42

Once a map has been created, its underlying data is shared between other instances of the same map. Changes to one instance are reflected in the other.

# Create a map and assign it to `a`.
a = {foo: 99} 
a.foo
# -> 99

# Assign a new instance of the map to `z`.
z = a

# Modifying the data via `z` is reflected in `a`.
z.foo = 'Hi!'
a.foo
# -> Hi!

Shorthand Values

Koto supports a shorthand notation when creating maps with inline syntax. If a value isn't provided for a key, then Koto will look for a value in scope that matches the key's name, and if one is found then it will be used as that entry's value.

bar = 'hi!'
m = {foo: 42, bar}
m.bar
# -> hi!

Maps and Self

Maps can store any type of value, including functions, which provides a convenient way to group functions together.

m = 
  hello: |name| 'Hello, {name}!'
  bye: |name| 'Bye, {name}!'

m.hello 'World'
# -> Hello, World!
m.bye 'Friend'
# -> Bye, Friend!

self is a special identifier that refers to the instance of the container in which the function is contained.

self allows functions to access and modify data from the map, enabling object-like behaviour.

m = 
  name: 'World'
  say_hello: || 'Hello, {self.name}!'

m.say_hello()
# -> Hello, World!

m.name = 'Friend'
m.say_hello()
# -> Hello, Friend!

Joining Maps

The + operator allows maps to be joined together, creating a new map that combines their entries.

a = {red: 100, blue: 150}
b = {green: 200, blue: 99}
c = a + b
c
# -> {red: 100, blue: 99, green: 200}

Quoted Map Keys

Map keys are usually defined and accessed without quotes, but they are stored in the map as strings. Quotes can be used if a key needs to be defined that would be otherwise be disallowed by Koto syntax rules (e.g. a keyword, or using characters that aren't allowed in an identifier). Quoted keys also allow dynamic keys to be generated by using string interpolation.

x = 99
m =
  'true': 42
  'key{x}': x
m.'true'
# -> 42
m.key99
# -> 99

Map Key Types

Although map keys are typically strings, other value types can be used as keys by using the map.insert and map.get functions.

Map keys are typically strings, but any immutable value can be used as a map key.

The immutable value types in Koto are strings, numbers, booleans, ranges, and null. A tuple is also considered to be immutable when its contained elements are also immutable.

Core Library

The Core Library provides a collection of fundamental functions and values for working with the Koto language, organized within modules.

# Get the size of a string
string.to_lowercase 'HELLO'
# -> hello

# Return the first element of the list
list.first [99, -1, 3]
# -> 99

Values in Koto automatically have access to their corresponding core modules via . access.

'xyz'.to_uppercase()
# -> XYZ

['abc', 123].first()
# -> abc

(7 / 2).round()
# -> 4

{apples: 42, pears: 99}.contains_key 'apples'
# -> true

The documentation for the Core library (along with this guide) are available in the help command of the Koto CLI.

Prelude

Koto's prelude is a collection of core library items that are automatically made available in a Koto script without the need for first calling import.

The modules that make up the core library are all included by default in the prelude. The following functions are also added to the prelude by default:

print 'io.print is available without needing to be imported'
# -> io.print is available without needing to be imported

Conditional Expressions

Koto includes several ways of producing values that depend on conditions.

if

if expressions come in two flavours; single-line:

x = 99
if x % 2 == 0 then print 'even' else print 'odd'
# -> odd

...and multi-line using indented blocks:

x = 24
if x < 0
  print 'negative'
else if x > 24
  print 'no way!'
else 
  print 'ok'
# -> ok

The result of an if expression is the final expression in the branch that gets executed.

x = if 1 + 1 == 2 then 3 else -1
x, x
# -> (3, 3)

# Assign the result of the if expression to foo
foo = if x > 0
  y = x * 10
  y + 3
else 
  y = x * 100
  y * y

foo, foo 
# -> (33, 33)

switch

switch expressions can be used as a cleaner alternative to if/else if/else cascades.

fib = |n|
  switch
    n <= 0 then 0
    n == 1 then 1
    else (fib n - 1) + (fib n - 2)

fib 7
# -> 13

match

match expressions can be used to match a value against a series of patterns, with the matched pattern causing a specific branch of code to be executed.

Patterns can be literals or identifiers. An identifier will accept any value, so they're often used with if conditions to refine the match.

match 40 + 2
  0 then 'zero'
  1 then 'one'
  x if x < 10 then 'less than 10: {x}'
  x if x < 50 then 'less than 50: {x}'
  x then 'other: {x}'
# -> less than 50: 42

The _ wildcard match can be used to match against any value (when the matched value itself can be ignored), and else can be used for fallback branches.

fizz_buzz = |n|
  match n % 3, n % 5
    0, 0 then "Fizz Buzz"
    0, _ then "Fizz"
    _, 0 then "Buzz"
    else n

(10, 11, 12, 13, 14, 15)
  .each |n| fizz_buzz n
  .to_tuple()
# -> ('Buzz', 11, 'Fizz', 13, 14, 'Fizz Buzz')

List and tuple entries can be matched against by using parentheses, with ... available for capturing the rest of the sequence.

match ['a', 'b', 'c'].extend [1, 2, 3]
  ('a', 'b') then "A list containing 'a' and 'b'"
  (1, ...) then "Starts with '1'"
  (..., 'y', last) then "Ends with 'y' followed by '{last}'"
  ('a', x, others...) then
    "Starts with 'a', followed by '{x}', then {size others} others"
  unmatched then "other: {unmatched}"
# -> Starts with 'a', followed by 'b', then 4 others

Loops

Koto includes several ways of evaluating expressions repeatedly in a loop.

while

while loops continue to repeat while a condition is true.

x = 0
while x < 5
  x += 1
x
# -> 5

until

until loops continue to repeat until a condition is true.

z = [1, 2, 3]
until z.is_empty()
  # Remove the last element of the list
  print z.pop()
# -> 3
# -> 2
# -> 1

break

Loops can be terminated with the break keyword.

x = 0
while x < 100000
  if x >= 3
    # Break out of the loop when x is greater or equal to 3
    break
  x += 1
x
# -> 3

A value can be provided to break, which is then used as the result of the loop.

x = 0
y = while x < 100000
  if x >= 3
    # Break out of the loop, providing x + 100 as the loop's result
    break x + 100
  x += 1
y
# -> 103

loop

loop creates a loop that will repeat indefinitely.

x = 0
y = loop
  x += 1
  # Stop looping when x is greater than 4
  if x > 4
    break x * x
y
# -> 25

for

for loops are repeated for each element in a sequence, such as a list or tuple.

for n in [10, 20, 30]
  print n
# -> 10
# -> 20
# -> 30

continue

continue skips the remaining part of a loop's body and proceeds with the next repetition of the loop.

for n in (-2, -1, 1, 2)
  # Skip over any values less than 0
  if n < 0
    continue
  print n
# -> 1
# -> 2

Iterators

The elements of a sequence can be accessed sequentially with an iterator, created using the .iter() function.

An iterator yields values via .next() until the end of the sequence is reached, when null is returned.

i = [10, 20].iter()

i.next()
# -> IteratorOutput(10)
i.next()
# -> IteratorOutput(20)
i.next()
# -> null

Iterator Generators

The iterator module contains iterator generators like once and repeat that generate output values lazily during iteration.

# Create an iterator that repeats ! twice
i = iterator.repeat('!', 2)
i.next()
# -> IteratorOutput(!)
i.next()
# -> IteratorOutput(!)
i.next()
# -> null

Iterator Adaptors

The output of an iterator can be modified using adaptors from the iterator module.

# Create an iterator that keeps any value above 3
x = [1, 2, 3, 4, 5].keep |n| n > 3

x.next()
# -> IteratorOutput(4)
x.next()
# -> IteratorOutput(5)
x.next()
# -> null

Using iterators with for

for loops accept any iterable value as input, including adapted iterators.

for x in 'abacad'.keep |c| c != 'a'
  print x
# -> b
# -> c
# -> d

Iterator Chains

Iterator adaptors can be passed into other adaptors, creating iterator chains that act as data processing pipelines.

i = (1, 2, 3, 4, 5)
  .skip 1
  .each |n| n * 10
  .keep |n| n <= 40
  .intersperse '--'

for x in i
  print x
# -> 20
# -> --
# -> 30
# -> --
# -> 40

Iterator Consumers

Iterators can also be consumed using functions like .to_list() and .to_tuple(), allowing the output of an iterator to be easily captured in a container.

[1, 2, 3]
  .each |n| n * 2
  .to_tuple()
# -> (2, 4, 6)

(1, 2, 3, 4)
  .keep |n| n % 2 == 0
  .each |n| n * 11
  .to_list()
# -> [22, 44]

Value Unpacking

Multiple assignments can be performed in a single expression by separating the variable names with commas.

a, b = 10, 20
a, b
# -> (10, 20)

If there's a single value being assigned, and the value is iterable, then it gets unpacked into the target variables.

my_tuple = 1, 2
x, y = my_tuple
y, x
# -> (2, 1)

Unpacking works with any iterable value, including adapted iterators.

a, b, c = [1, 2, 3, 4, 5]
a, b, c
# -> (1, 2, 3)

x, y, z = 'a-b-c'.split '-'
x, y, z
# -> ('a', 'b', 'c')

If the value being unpacked doesn't contain enough values for the assignment, then null is assigned to any remaining variables.

a, b, c = [-1, -2]
a, b, c
# -> (-1, -2, null)

x, y, z = 42
x, y, z
# -> (42, null, null)

Unpacking can also be used in for loops, which is particularly useful when looping over the contents of a map.

my_map = {foo: 42, bar: 99}
for key, value in my_map
  print key, value
# -> ('foo', 42)
# -> ('bar', 99)

Generators

Generators are iterators that are made by calling generator functions, which are any functions that contain a yield expression.

The generator is paused each time yield is encountered, waiting for the caller to continue execution.

my_first_generator = ||
  yield 1
  yield 2

x = my_first_generator()
x.next()
# -> IteratorOutput(1)
x.next()
# -> IteratorOutput(2)
x.next()
# -> null

Generator functions can accept arguments like any other function, and each time they're called a new generator is created.

As with any other iterable value, the iterator module's functions are made available to generators.

make_generator = |x|
  for y in (1, 2, 3)
    yield x + y

make_generator(0).to_tuple()
# -> (1, 2, 3)
make_generator(10)
  .keep |n| n % 2 == 1
  .to_list()
# -> [11, 13]

Custom Iterator Adaptors

Generators can also serve as iterator adaptors by modifying the output of another iterator.

Inserting a generator into the iterator module makes it available in any iterator chain.

# Make an iterator adaptor that yields every
# other value from the adapted iterator
iterator.every_other = ||
  n = 0
  # When the generator is created, self is initialized with the previous
  # iterator in the chain, allowing its output to be adapted.
  for output in self
    # If n is even, then yield a value
    if n % 2 == 0
      yield output
    n += 1

(1, 2, 3, 4, 5)
  .each |n| n * 10
  .every_other() # Skip over every other value in the iterator chain
  .to_list()
# -> [10, 30, 50]

Ranges

Ranges of integers can be created with .. or ..=.

.. creates a non-inclusive range, which defines a range up to but not including the end of the range.

# Create a range from 10 to 20, not including 20
r = 10..20
# -> 10..20
r.start()
# -> 10
r.end()
# -> 20
r.contains 20
# -> false

..= creates an inclusive range, which includes the end of the range.

# Create a range from 10 to 20, including 20
r = 10..=20
# -> 10..=20
r.contains 20
# -> true

If a value is missing from either side of the range operator then an unbounded range is created.

# Create an unbounded range starting from 10
r = 10..
r.start()
# -> 10
r.end()
# -> null
 
# Create an unbounded range up to and including 100
r = ..=100
r.start()
# -> null
r.end()
# -> 100

Bounded ranges are declared as iterable, so they can be used in for loops and with the iterator module.

for x in 1..=3
  print x
# -> 1
# -> 2
# -> 3

(0..5).to_list()
# -> [0, 1, 2, 3, 4]

Slices

Ranges can be used to create a slice of a container's data.

x = (10, 20, 30, 40, 50)
x[1..=3] 
# -> (20, 30, 40)

For immutable containers like tuples and strings, slices share the original value's data, with no copies being made.

For mutable containers like lists, creating a slice makes a copy of the sliced portion of the underlying data.

x = 'abcdef'
# No copies are made when a string is sliced
y = x[3..6]
# -> def

a = [1, 2, 3]
# When a list is sliced, the sliced elements get copied into a new list
b = a[0..2]
# -> [1, 2]
b[0] = 42
# -> 42
a[0]
# -> 1

When creating a slice with an unbounded range, if the start of the range if ommitted then the slice starts from the beginning of the container. If the end of the range is ommitted, then the slice includes all remaining elements in the container.

z = 'HΓ«llΓΈ'.to_tuple()
z[..2]
# -> ('H', 'Γ«')
z[2..]
# -> ('l', 'l', 'ΓΈ')

String Formatting

Interpolated string expressions can be formatted using formatting options similar to Rust's.

Inside an interpolated expression, options are provided after a : separator.

'{number.pi:πœ‹^8.2}'
# -> πœ‹πœ‹3.14πœ‹πœ‹

Minimum Width and Alignment

A minimum width can be specified, ensuring that the formatted value takes up at least that many characters.

foo = "abcd"
'_{foo:8}_'
# -> _abcd    _

The minimum width can be prefixed with an alignment modifier:

  • < - left-aligned
  • ^ - centered
  • > - right-aligned
foo = "abcd"
'_{foo:^8}_'
# -> _  abcd  _

All values are left-aligned if an alignment modifier isn't specified, except for numbers which are right-aligned by default.

x = 1.2
'_{x:8}_'
# -> _     1.2_

The alignment modifier can be prefixed with a character which will be used to fill any empty space in the formatted string (the default character being ).

x = 1.2
'_{x:~<8}_'
# -> _1.2~~~~~_

For numbers, the minimum width can be prefixed with 0, which will pad the number to the specified width with zeroes.

x = 1.2
'{x:06}'
# -> 0001.2

Maximum Width / Precision

A maximum width for the interpolated expression can be specified following a . character.

foo = "abcd"
'{foo:_^8.2}'
# -> ___ab___

For numbers, the maximum width acts as a 'precision' value, or in other words, the number of decimal places that will be rendered for the number.

x = 1 / 3
'{x:.4}'
# -> 0.3333

Advanced Functions

Functions in Koto have some advanced features that are worth exploring.

Captured Variables

When a variable is accessed in a function that wasn't declared locally, then it gets captured by copying it into the function.

x = 1

my_function = |n| 
  # x is assigned outside the function,
  # so it gets captured when the function is created.
  n + x 

# Reassigning x here doesn't modify the value 
# of x that was captured when my_function was created.
x = 100

my_function 2
# -> 3

This behavior is different to many other languages, where captures are often taken by reference rather than by copy.

It's also worth noting that capture variables will have the same starting value each time the function is called.

x = 99
f = || 
  # Modifying x only happens with a local copy during a function call.
  # The value of x at the start of the call matches when the value it had when 
  # it was captured.
  x += 1

f(), f(), f()
# -> (100, 100, 100)

To modify captured state that can be modified, use a container (like a map) to hold on to mutable values.

data = {x: 99}

f = || 
  # The data map gets captured by the function, 
  # and its contained values can be modified between calls.
  data.x += 1

f(), f(), f()
# -> (100, 101, 102)

Optional Arguments

When calling a function, any missing arguments will be replaced by null.

f = |a, b, c|
  print a, b, c

f 1
# -> (1, null, null)
f 1, 2
# -> (1, 2, null)
f 1, 2, 3
# -> (1, 2, 3)

Missing arguments can be replaced with default values by using or.

f = |a, b, c|
  print a or -1, b or -2, c or -3

f 42
# -> (42, -2, -3)
f 99, 100
# -> (99, 100, -3)

or will reject false, so if false would be a valid input then a direct comparison against null can be used instead.

f = |a| 
  print if a == null then -1 else a

f()
# -> -1
f false
# -> false

Variadic Functions

A variadic function can be created by appending ... to the last argument. When the function is called any extra arguments will be collected into a tuple.

f = |a, b, others...|
  print "a: {a}, b: {b}, others: {others}"

f 1, 2, 3, 4, 5
# -> a: 1, b: 2, others: (3, 4, 5)

Argument Unpacking

Functions that expect containers as arguments can unpack the contained elements directly in the argument declaration by using parentheses.

# A function that sums a container with three contained values
f = |(a, b, c)| a + b + c

x = [100, 10, 1]
f x
# -> 111

Any container that supports indexing operations (like lists and tuples) with a matching number of elements will be unpacked, otherwise an error will be thrown.

Unpacked arguments can also be nested.

# A function that sums elements from nested containers
f = |((a, b), (c, d, e))| 
  a + b + c + d + e
x = ([1, 2], [3, 4, 5])
f x
# -> 15

Ellipses can be used to unpack any number of elements at the start or end of a container.

f = |(..., last)| last * last
x = (1, 2, 3, 4)
f x
# -> 16

A name can be added to ellipses to assign the unpacked elements.

f = |(first, others...)| first * others.sum()
x = (10, 1, 2, 3)
f x
# -> 60

Ignoring Arguments

The wildcard _ can be used to ignore function arguments.

# A function that sums the first and third elements of a container
f = |(a, _, c)| a + c

f [100, 10, 1]
# -> 101

If you would like to keep the name of the ignored value as a reminder, then _ can be used as a prefix for an identifier. Identifiers starting with _ can be written to, but can't be accessed.

my_map = {foo_a: 1, bar_a: 2, foo_b: 3, bar_b: 4}
my_map
  .keep |(key, _value)| key.starts_with 'foo'
  .to_tuple()
# -> (('foo_a', 1), ('foo_b', 3))

Objects and Metamaps

Value types with custom behaviour can be defined in Koto through the concept of objects.

An object is any map that includes one or more metakeys (keys prefixed with @), that are stored in the object's metamap. Whenever operations are performed on the object, the runtime checks its metamap for corresponding metakeys.

In the following example, addition and subtraction operators are overridden for a custom Foo object:

# Declare a function that makes Foo objects
foo = |n|
  data: n

  # Overriding the addition operator
  @+: |other|
    # A new Foo is made using the result 
    # of adding the two data values together
    foo self.data + other.data

  # Overriding the subtraction operator
  @-: |other|
    foo self.data - other.data

  # Overriding the multiply-assignment operator
  @*=: |other|
    self.data *= other.data
    self

a = foo 10
b = foo 20

(a + b).data
# -> 30
(a - b).data
# -> -10
a *= b
a.data
# -> 200

Meta Operators

All of the binary arithmetic and logic operators (*, <, >=, etc) can be implemented following this pattern.

Additionally, the following metakeys can also be defined:

@negate

The @negate metakey overrides the negation operator.

foo = |n|
  data: n
  @negate: || foo -self.data

x = -foo(100)
x.data
# -> -100

@not

The @not metakey overrides the not operator.

foo = |n|
  data: n
  @not: || self.data == 0

not (foo 10)
# -> false

@size and @[]

The @size metakey defines how the object should report its size, while the @[] metakey defines what values should be returned when indexing is performed on the object.

If @size is implemented, then @[] should also be implemented.

The @[] implementation can support indexing by any input values that make sense for your object type, but for argument unpacking to work correctly the runtime expects that indexing by both single indices and ranges should be supported.

foo = |data|
  data: data
  @size: || size self.data
  @[]: |index| self.data[index]

x = foo (100, 200, 300)
size x
# -> 3
x[1]
# -> 200

# Unpack the first two elements in the argument and multiply them
multiply_first_two = |(a, b, ...)| a * b
multiply_first_two x
# -> 20000

# Inspect the first element in the object
match x
  (first, others...) then 'first: {first}, remaining: {size others}'
# -> first: 100, remaining: 2

@||

The @|| metakey defines how the object should behave when its called as a function.

foo = |n|
  data: n
  @||: || 
    self.data *= 2
    self.data

x = foo 2
x()
# -> 4
x()
# -> 8

@iterator

The @iterator metakey defines how iterators should be created when the object is used in an iterable context. When called, @iterator should return an iterable value that will then be used for iterator operations.

foo = |n|
  # Return a generator that yields the three numbers following n
  @iterator: || 
    yield n + 1
    yield n + 2
    yield n + 3

(foo 0).to_tuple()
# -> (1, 2, 3)

(foo 100).to_list()
# -> [101, 102, 103]

Note that this key will be ignored if the object also implements @next, which implies that the object is already an iterator.

@next

The @next metakey allows for objects to behave as iterators.

Whenever the runtime needs to produce an iterator from an object, it will first check the metamap for an implementation of @next, before looking for @iterator.

The @next function will be called repeatedly during iteration, with the returned value being used as the iterator's output. When the returned value is null then the iterator will stop producing output.

foo = |start, end|
  start: start
  end: end
  @next: || 
    if self.start < self.end
      result = self.start
      self.start += 1
      result
    else 
      null
  
foo(10, 15).to_tuple()
# -> (10, 11, 12, 13, 14)

@next_back

The @next_back metakey is used by iterator.reversed when producing a reversed iterator.

The runtime will only look for @next_back if @next is implemented.

foo =
  n: 0
  @next: || self.n += 1
  @next_back: || self.n -= 1

foo
  .skip 3 # 0, 1, 2
  .reversed()
  .take 3 # 2, 1, 0
  .to_tuple()
# -> (2, 1, 0)

@display

The @display metakey defines how the object should be represented when displaying the object as a string.

foo = |n|
  data: n
  @display: || 'Foo({self.data})'

foo 42
# -> Foo(42)

x = foo -1
"The value of x is '{x}'"
# -> The value of x is 'Foo(-1)'

@type

The @type metakey takes a string as a value which is used when checking the value's type, e.g. with koto.type

foo = |n|
  data: n
  @type: "Foo"

koto.type (foo 42)
# -> Foo

@base

Objects can inherit properties and behavior from other values, establishing a base value through the @base metakey. This allows objects to share common functionality while maintaining their own unique attributes.

In the following example, two kinds of animals are created that share the speak function from their base value.

animal = |name|
  name: name
  speak: || '{self.noise}! My name is {self.name}!'

dog = |name|
  @base: animal name
  noise: 'Woof'

cat = |name|
  @base: animal name
  noise: 'Meow'

dog('Fido').speak()
# -> Woof! My name is Fido!

cat('Smudge').speak()
# -> Meow! My name is Smudge!

@meta

The @meta metakey allows named metakeys to be added to the metamap. Metakeys defined with @meta are accessible via . access, similar to regular object keys, but they don't appear as part of the object's main data entries when treated as a regular map.

foo = |n|
  data: n
  @meta hello: "Hello!"
  @meta get_info: || 
    info = match self.data 
      0 then "zero"
      n if n < 0 then "negative"
      else "positive"
    "{self.data} is {info}"

x = foo -1
x.hello
# -> Hello!

print x.get_info()
# -> -1 is negative

print map.keys(x).to_tuple()
# -> ('data')

Sharing Metamaps

Metamaps can be shared between objects by using Map.with_meta, which helps to avoid inefficient duplication when creating a lot of objects.

In the following example, behavior is overridden in a single metamap, which is then shared between object instances.

# Create an empty map for global values
global = {}

# Define a function that makes a Foo object
foo = |data|
  # Make a new map that contains `data`, 
  # and then attach a shared copy of the metamap from foo_meta.
  {data}.with_meta global.foo_meta

# Define some metakeys in foo_meta
global.foo_meta =
  # Override the + operator
  @+: |other| foo self.data + other.data

  # Define how the object should be displayed 
  @display: || "Foo({self.data})"

(foo 10) + (foo 20)
# -> Foo(30)

Error Handling

Errors can be thrown in the Koto runtime, which then cause the runtime to stop execution.

A try / catch expression can be used to catch any thrown errors, allowing execution to continue. An optional finally block can be used for cleanup actions that need to performed whether or not an error was caught.

x = [1, 2, 3]
try
  # Accessing an invalid index will throw an error
  print x[100]
catch error 
  print "Caught an error"
finally
  print "...and finally"
# -> Caught an error
# -> ...and finally

throw can be used to explicity throw an error when an exceptional condition has occurred.

throw accepts strings or objects that implement @display.

f = || throw "!Error!"

try
  f()
catch error
  print "Caught an error: '{error}'"
# -> Caught an error: '!Error!'

Testing

Koto includes a simple testing framework that help you to check that your code is behaving as you expect through automated checks.

Assertions

The core library includes a collection of assertion functions in the test module, which are included by default in the prelude.

try 
  assert 1 + 1 == 3
catch error
  print 'An assertion failed'
# -> An assertion failed

try 
  assert_eq 'hello', 'goodbye'
catch error
  print 'An assertion failed'
# -> An assertion failed

Organizing Tests

Tests can be organized by collecting @test functions in an object.

The tests can then be run manually with test.run_tests. For automatic testing, see the description of exporting @tests in the following section.

basic_tests = 
  @test add: || assert_eq 1 + 1, 2 
  @test subtract: || assert_eq 1 - 1, 0 

test.run_tests basic_tests

For setup and cleanup operations shared across tests, @pre_test and @post_test metakeys can be implemented. @pre_test will be run before each @test, and @post_test will be run after.

make_x = |n|
  data: n
  @+: |other| make_x self.data + other.data
  @-: |other| make_x self.data - other.data

x_tests =
  @pre_test: || 
    self.x1 = make_x 100
    self.x2 = make_x 200
      
  @post_test: ||
    print 'Test complete'

  @test addition: ||
    print 'Testing addition'
    assert_eq self.x1 + self.x2, make_x 300

  @test subtraction: ||
    print 'Testing subtraction'
    assert_eq self.x1 - self.x2, make_x -100

  @test failing_test: ||
    print 'About to fail'
    assert false

try
  test.run_tests x_tests
catch _
  print 'A test failed'
# -> Testing addition
# -> Test complete
# -> Testing subtraction
# -> Test complete
# -> About to fail
# -> A test failed

Modules

Koto includes a module system that helps you to organize and re-use your code when your program grows too large for a single file.

import

Items from modules can be brought into the current scope using import.

from list import last
from number import abs

x = [1, 2, 3]
last x
# -> 3

abs -42
# -> 42

Multiple items from a single module can be imported at the same time.

from tuple import contains, first, last

x = 'a', 'b', 'c'
first x
# -> a
last x
# -> c
contains x, 'b'
# -> true

Imported items can be renamed using as for clarity or to avoid conflicts.

from list import first as list_first
from tuple import first as tuple_first
list_first [1, 2]
# -> 1
tuple_first (3, 2, 1)
# -> 3

export

export is used to add values to the current module's exports map.

Single values can be assigned to and exported at the same time:

##################
# my_module.koto #
##################

export say_hello = |name| 'Hello, {name}!'

##################
##################

from my_module import say_hello

say_hello 'Koto'
# -> 'Hello, Koto!' 

When exporting multiple values, it can be convenient to use map syntax:

##################
# my_module.koto #
##################

# Define some local values
a, b, c = 1, 2, 3

# Inline maps allow for shorthand syntax
export { a, b, c, foo: 42 }

# Map blocks can also be used with export
export 
  bar: 99
  baz: 'baz'

@tests and @main

A module can export a @tests object containing @test functions, which will be automatically run after the module has been compiled and initialized.

Additionally, a module can export a @main function. The @main function will be called after the module has been compiled and initialized, and after exported @tests have been successfully run.

Note that because metakeys can't be assigned locally, the use of export is optional when adding entries to the module's metamap.

##################
# my_module.koto #
##################

export say_hello = |name| 'Hello, {name}!'

@main = || # Equivalent to export @main =
  print '`my_module` initialized'

@tests =
  @test hello_world: ||
    print 'Testing...'
    assert_eq (say_hello 'World'), 'Hello, World!'

##################
##################

from my_module import say_hello
# -> Testing...
# -> Successfully initialized `my_module`

say_hello 'Koto'
# -> 'Hello, Koto!' 

Module Paths

When looking for a module, import will look for a .koto file with a matching name, or for a folder with a matching name that contains a main.koto file.

e.g. When an import foo expression is run, then a foo.koto file will be looked for in the same location as the current script, and if foo.koto isn't found then the runtime will look for foo/main.koto.