moonscript/docs/reference.md
2011-10-02 00:08:13 -07:00

26 KiB

target: reference/index template: reference title: MoonScript v0.2.0

MoonScript is a programming language that compiles to Lua. This guide expects the reader to have basic familiarity with Lua. For each code snippet below, the MoonScript is on the left and the compiled Lua is on right right.

The Language

Assignment

Unlike Lua, there is no local keyword. All assignments to names that are not already defined will be declared as local to the scope of that declaration. If you wish to create a global variable it must be done using the export keyword.

hello = "world"
a,b,c = 1, 2, 3

Update Assignment

+=, -=, /=, *=, %=, ..= operators have been added for updating a value by a certain amount. They are aliases for their expanded equivalents.

x = 0
x += 10

s = "hello "
s ..= "world"

Comments

Like Lua, comments start with -- and continue to the end of the line. Comments are not written to the output.

-- I am a comment

Literals & Operators

MoonScript supports all the same primitive literals as Lua and uses the same syntax. This applies to numbers, strings, booleans, and nil.

MoonScript also supports all the same binary and unary operators. Additionally != is as an alias for ~=.

Function Literals

All functions are created using a function expression. A simple function is denoted using the arrow: ->

my_function = ->
my_function() -- call the empty function

The body of the function can either be one statement placed directly after the arrow, or it can be a series of statements indented on the following lines:

func_a = -> print "hello world"

func_b = ->
  value = 100
  print "The value:", value

If a function has no arguments, it can be called using the ! operator, instead of empty parentheses. The ! invocation is the preferred way to call functions with no arguments.

func_a!
func_b()

Functions with arguments can be created by preceding the arrow with a list of argument names in parentheses:

sum = (x, y) -> print "sum", x + y

Functions can be called by listing the values of the arguments after the name of the variable where the function is stored. When chaining together function calls, the arguments are applied to the closest function to the left.

sum 10, 20
print sum 10, 20

a b c "a", "b", "c"

In order to avoid ambiguity in when calling functions, parentheses can also be used to surround the arguments. This is required here in order to make sure the right arguments get sent to the right functions.

print "x:", sum(10, 20), "y:", sum(30, 40)

Functions will coerce the last statement in their body into a return statement, this is called implicit return:

sum = (x, y) -> x + y
print "The sum is ", sum 10, 20

And if you need to explicitly return, you can use the return keyword:

sum = (x, y) -> return x + y

Just like in Lua, functions can return multiple values. The last statement must be a list of values separated by commas:

mystery = (x, y) -> x + y, x - y
a,b = mystery 10, 20

Fat Arrows

Because it is an idiom in Lua to send an object as the first argument when calling a method, a special syntax is provided for creating functions which automatically includes a self argument.

func = (num) => self.value + num

Argument Defaults

It is possible to provide default values for the arguments of a function. An argument is determined to be empty if it's value is nil. Any nil arguments that have a default value will be replace before the body of the function is run.

my_function = (name="something", height=100) ->
  print "Hello I am", name
  print "My height is", height

An argument default value expression is evaluated in the body of the function in the order of the argument declarations. For this reason default values have access to previously declared arguments.

some_args = (x=100, y=x+1000) ->
  print x + y

Considerations

Because of the expressive parentheses-less way of calling functions, some restrictions must be put in place to avoid parsing ambiguity involving whitespace.

The minus sign plays two roles, a unary negation operator and a binary subtraction operator. In order to force subtraction a space must be placed after the - operator. In order to force a negation, no space must follow the -. Consider the examples below.

a = x - 10
b = x-10
c = x -y

The precedence of the first argument of a function call can also be controlled using whitespace if the argument is a literal string.In Lua, it is common to leave off parentheses when calling a function with a single string literal.

When there is no space between a variable and a string literal, the function call takes precedence over any following expressions. No other arguments can be passed to the function when it is called this way.

Where there is a space following a variable and a string literal, the function call acts as show above. The string literal belongs to any following expressions (if they exist), which serves as the argument list.

x = func"hello" + 100
y = func "hello" + 100

Multi-line arguments

When calling functions that take a large number of arguments, it is convenient to split the argument list over multiple lines. Because of the white-space sensitive nature of the language, care must be taken when splitting up the argument list.

If an argument list is to be continued onto the next line, the current line must end in a comma. And the following line must be indented more than the current indentation. Once indented, all other argument lines must be at the same level of indentation to be part of the argument list

my_func 5,4,3,
  8,9,10

cool_func 1,2,
  3,4,
  5,6,
  7,8

This type of invocation can be nested. The level of indentation is used to determine to which function the arguments belong to.

my_func 5,6,7,
  6, another_func 6,7,8,
    9,1,2,
  5,4

Because tables also use the comma as a delimiter, this indentation syntax is helpful for letting values be part of the argument list instead of being part of the table.

x = {
  1,2,3,4, a_func 4,5,
    5,6,
  8,9,10
}

Although uncommon, notice how we can give a deeper indentation for function arguments if we know we will be using a lower indentation futher on.

y = { my_func 1,2,3,
   4,5,
  5,6,7
}

The same thing can be done with other block level statements like conditionals. We can use indentation level to determine what statement a value belongs to:

if func 1,2,3,
  "hello",
  "world"
    print "hello"
	print "I am inside if"

if func 1,2,3,
    "hello",
    "world"
  print "hello"
  print "I am inside if"

Table Literals

Like in Lua, tables are delimited in curly braces.

some_values = { 1, 2, 3, 4 }

Unlike Lua, assigning a value to a key in a table is done with : (instead of =).

some_values = {
  name: "Bill",
  age: 200,
  ["favorite food"]: "rice"
}

The curly braces can be left off if a single table of key value pairs is being assigned.

profile = 
  height: "4 feet",
  shoe_size: 13,
  favorite_foods: {"ice cream", "donuts"}

Newlines can be used to delimit values instead of a comma (or both):

values = {
  1,2,3,4
  5,6,7,8
  name: "superman"
  occupation: "crime fighting"
}

When creating a single line table literal, the curly braces can also be left off:

my_function dance: "Tango", partner: "none"

y = type: "dog", legs: 4, tails: 1

The keys of a table literal can be language keywords without being escaped:

tbl = {
  do: "something"
  end: "hunger"
}

If you are constructing a table out of variables and wish the keys to be the same as the variable names, then the : prefix operator can be used:

hair = "golden"
height = 200
person = { :hair, :height, shoe_size: 40 }

print_table :hair, :height

Table Comprehensions

Table comprehensions provide a quick way to iterate over a table's values while applying a statement and accumulating the result.

The following creates a copy of the items table but with all the values doubled.

items = { 1, 2, 3, 4 }
doubled = [item * 2 for i, item in ipairs items]

The items included in the new table can be restricted with a when clause:

iter = ipairs items
slice = [item for i, item in iter when i > 1 and i < 3]

Because it is common to iterate over the values of a numerically indexed table, an * operator is introduced. The doubled example can be rewritten as:

doubled = [item * 2 for item in *items]

The for and when clauses can be chained as much as desired. The only requirement is that a comprehension has at least one for clause.

Using multiple for clauses is the same as using nested loops:

x_coords = {4, 5, 6, 7}
y_coords = {9, 2, 3}

points = [{x,y} for x in *x_coords for y in *y_coords]

Slicing

A special syntax is provided to restrict the items that are iterated over when using the * operator. This is equivalent to setting the iteration bounds and a step size in a for loop.

Here we can set the minimum and maximum bounds, taking all items with indexes between 1 and 5 inclusive:

slice = [item for item in *items[1:5]]

Any of the slice arguments can be left off to use a sensible default. In this example, if the max index is left off it defaults to the length of the table. This will take everything but the first element:

slice = [item for item in *items[2:]]

If the minimum bound is left out, it defaults to 1. Here we only provide a step size and leave the other bounds blank. This takes all odd indexed items: (1, 3, 5, ...)

slice = [item for items in *items[::2]]

For Loop

There are two for loop forms, just like in Lua. A numeric one and a generic one:

for i = 10, 20
  print i

for k = 1,15,2 -- an optional step provided
  print k 

for key, value in pairs object
  print key, value

The slicing and * operators can be used, just like with table comprehensions:

for item in *items[2:4]
  print item

A shorter syntax is also available for all variations when the body is only a single line:

for item in *items do print item

for j = 1,10,3 do print j

A for loop can also be used an expression. The last statement in the body of the for loop is coerced into an expression and appended to an accumulating table if the value of that expression is not nil.

Doubling every even number:

doubled_evens = for i=1,20
  if i % 2 == 0
    i * 2
  else
    i

Filtering out odd numbers:

my_numbers = {1,2,3,4,5,6}
odds = for x in *my_numbers
  if x % 2 == 1 then x

For loops at the end of a function body are not accumulated into a table for a return value (Instead the function will return nil). Either an explicit return statement can be used, or the loop can be converted into a list comprehension.

func_a = -> for i=1,10 do i
func_b = -> return for i=1,10 do i

print func_a! -- prints nil
print func_b! -- prints table object

This is done to avoid the needless creation of tables for functions that don't need to return the results of the loop.

While Loop

The while loop also comes in two variations:

i = 10
while i > 0
  print i
  i -= 1

while running == true do my_function!

Like for loops, the while loop can also be used an expression. Additionally, for a function to return the accumulated value of a while loop, the statement must be explicitly returned.

Conditionals

have_coins = false
if have_coins
  print "Got coins"
else
  print "No coins"

A short syntax for single statements can also be used:

have_coins = false
  if have_coins then print "Got coins" else print "No coins"

Because if statements can be used as expressions, this can able be written as:

have_coins = false
print if have_coins then "Got coins" else "No coins"

Conditionals can also be used in return statements and assignments:

is_tall = (name) ->
  if name == "Rob"
    true
  else
    false

message = if is_tall "Rob"
  "I am very tall"
else
  "I am not so tall"

print message -- prints: I am very tall

Line Decorators

For convenience, the for loop and if statement can be applied to single statements at the end of the line:

print "hello world" if name == "Rob"

And with basic loops:

print "item: ", item for item in *items

Object Oriented Programming

In these examples, the generated Lua code may appear overwhelming. It is best to focus on the meaning of the MoonScript code at first, then look into the Lua code if you wish to know the implementation details.

A simple class:

class Inventory
  new: =>
    @items = {}

  add_item: (name) =>
    if @items[name]
      @items[name] += 1
    else
      @items[name] = 1

A class is declared with a class statement followed by a table-like declaration where all of the methods and properties are listed.

The new property is special in that it will become the constructor.

Notice how all the methods in the class use the fat arrow function syntax. When calling methods on a instance, the instance itself is sent in as the first argument. The fat arrow handles the creation of a self argument.

The @ prefix on a variable name is shorthand for self.. @items becomes self.items.

Creating an instance of the class is done by calling the name of the class as a function.

inv = Inventory!
inv\add_item "t-shirt"
inv\add_item "pants"

Because the instance of the class needs to be sent to the methods when they are called, the '' operator is used.

All properties of a class are shared among the instances. This is fine for functions, but for other types of objects, undesired results may occur.

Consider the example below, the clothes property is shared amongst all instances, so modifications to it in one instance will show up in another:

class Person
  clothes: {}
  give_item: (name) =>
    table.insert @clothes, name

a = Person!
b = Person!

a\give_item "pants"
b\give_item "shirt"

-- will print both pants and shirt
print item for item in *a.clothes

The proper way to avoid this problem is to create the mutable state of the object in the constructor:

class Person
  new: =>
    @clothes = {}

Inheritance

The extends keyword can be used in a class declaration to inherit the properties and methods from another class.

class BackPack extends Inventory
  size: 10
  add_item: (name) =>
    if #@items > size then error "backpack is full"
    super name

Here we extend our Inventory class, and limit the amount of items it can carry. The super keyword can be called as a function to call the function of the same name in the super class. It can also be accessed like an object in order to retrieve values in the parent class that might have been shadowed by the child class.

Types

Every instance of a class carries its type with it. This is stored in the special __class property. This property holds the class object. The class object is what we call to build a new instance. We can also index the class object to retrieve class methods and properties.

b = BackPack!
assert b.__class == BackPack

print BackPack.size -- prints 10

Export Statement

Because, by default, all assignments to variables that are not lexically visible will be declared as local, special syntax is required to declare a variable globally.

The export keyword makes it so any following assignments to the specified names will not be assigned locally.

export var_name, var_name2
var_name, var_name3 = "hello", "world"

This is especially useful when declaring what will be externally visible in a module:

-- my_module.moon
module "my_module", package.seeall
export print_result

length = (x, y) -> math.sqrt x*x + y*y

print_result = (x, y) ->
  print "Length is ", length x, y

-- main.moon
require "my_module"

my_module.print_result 4, 5 -- prints the result

print my_module.length 6, 7 -- errors, `length` not visible

Export All & Export Proper

The export statement can also take special symbols * and ^.

export * is used to export any name declared after the statement in the current scope. export ^ will export all proper names, names that begin with a capital letter.

Import Statement

Often you want to bring some values from a table into the current scope as local variables by their name. The import statement lets us accomplish this:

import insert from table

The multiple names can be given, each separated by a comma:

import C, Ct, Cmt from lpeg

Sometimes a function requires that the table be sent in as the first argument (when using the \ syntax). As a shortcut, we can prefix the name with a \ to bind it to that table:

-- some object
my_module =
    state: 100
    add: (value) =>
        self.state + value

import \add from my_module

print add(22) -- equivalent to calling my_module\get 22

With Statement

A common pattern involving the creation of an object is calling a series of functions and setting a series of properties immediately after creating it.

This results in repeating the name of the object multiple times in code, adding unnecessary noise. A common solution to this is to pass a table in as an argument which contains a collection of keys and values to overwrite. The downside to this is that the constructor of this object must support this form.

The with block helps to alleviate this. Within a with block we can use a special statements that begin with either . or \ which represent those operations applied to the object we are using with on.

For example, we work with a newly created object:

with Person!
  .name = "Oswald"
  \add_relative my_dad
  \save!
  print .name

The with statement can also be used as an expression which returns the value it has been giving access to.

file = with File "favorite_foods.txt"
  \set_encoding "utf8"

Or...

create_person = (name,  relatives) ->
  with Person!
    .name = name
    \add_relative relative for relative in *relatives

me = create_person "Leaf", {dad, mother, sister}

Function Stubs

It is common to pass a function from an object around as a value, for example, passing an instance method into a function as a callback. If the function expects the object it is operating on as the first argument then you must somehow bundle that object with the function so it can be called properly.

The function stub syntax is a shorthand for creating a new closure function that bundles both the object and function. This new function calls the wrapped function in the correct context of the object.

Its syntax is the same as calling an instance method with the \ operator but with no argument list provided.

my_object = {
  value: 1000
  write: => print "the value:", @value
}

run_callback (func) ->
  print "running callback..."
  func!

-- this will not work:
-- the function has to no reference to my_object
run_callback my_object.write

-- function stub syntax
-- lets us bundle the object into a new function
run_callback my_object\write

The Using Clause; Controlling Destructive Assignment

While lexical scoping can be a great help in reducing the complexity of the code we write, things can get unwieldy as the code size increases. Consider the following snippet:

i = 100

-- many lines of code...

my_func = ->
    i = 10
    while i > 0
        print i
        i -= 1

my_func()

print i -- will print 0

In my_func, we've overwritten the value of i mistakenly. In this example it is quite obvious, but consider a large, or foreign code base where it isn't clear what names have already been declared.

It would be helpful to say which variables from the enclosing scope we intend on change, in order to prevent us from changing others by accident.

The using keyword lets us do that. using nil makes sure that no closed variables are overwritten in assignment. The using clause is placed after the argument list in a function, or in place of it if there are no arguments.

i = 100

my_func = (using nil) ->
    i = "hello" -- a new local variable is created here

my_func()
print i -- prints 100, i is unaffected

Multiple names can be separated by commas. Closure values can still be accessed, they just cant be modified:

tmp = 1213
i, k = 100, 50

my_func = (add using k,i) ->
    tmp = tmp + add -- a new local tmp is created 
    i += tmp
    k += tmp

my_func(22)
print i,k -- these have been updated

MoonScript API

moonscript Module

Upon installing MoonScript, a moonscript module is made available. The best use of this module is making your Lua's require function MoonScript aware.

require "moonscript"

After moonscript is required, Lua's package loader is updated to search for .moon files on any subsequent calls to require. The search path for .moon files is based on the current package.path value in Lua when moonscript is required. Any search paths in package.path ending in .lua are copied, rewritten to end in .moon, and then inserted in package.moonpath.

The moonloader is the function that is responsible for searching package.moonpath for a file available to be included. It is inserted in the second position of the package.loaders table. This means that a matching .moon file will be loaded over a matching .lua file that has the same base name.

For more information on Lua's package.loaders see Lua Reference Manual — package.loaders

The moonloader, when finding a valid path to a .moon file, will parse and compile the file in memory. The code is then turned into a function using the built in load function, which is run as the module.

Error Rewriting

Runtime errors are given special attention when using the moonloader. Because we start off as MoonScript, but run code as Lua, errors that happen during runtime report their line numbers as they are in the compiled file. This can make debugging particularly difficult.

Consider the following file with a bug:

add_numbers = (x,y) -> x + z
print add_numbers 10,0

The following error is generated:

moon:err.moon:1: attempt to perform arithmetic on global 'z' (a nil value)
stack traceback:
    err.moon:1: in function 'add_numbers'
    err.moon:2: in main chunk

Instead of the error being reported on line number 3, where it appears in the Lua file, it is reported on line 1, where the faulty line originated. The entire stack trace is rewritten in addition to the error.

Programmatically Compiling

The MoonScript module also contains methods for parsing MoonScript text into an abstract syntax tree, and compiling an instance of a tree into Lua source code.

Knowledge of this API may be useful for creating tools to aid the generation of Lua code from MoonScript code.

Here is a quick example of how you would compile a MoonScript string to a Lua String:

require "moonscript.parse"
require "moonscript.compile"

import parse, compile from moonscript

moon_code = [[(-> print "hello world")!]]

tree, err = parse.string moon_code
if not tree
  error "Parse error: " .. err

lua_code, err, pos = compile.tree tree
if not lua_code
  error compile.format_error err, pos, moon_code

-- our code is ready
print lua_code

Command Line Use

Two tools are installed with MoonScript, moon and moonc.

moonc is for compiling MoonScript code to Lua.
moon is for running MoonsScript code directly.

moon

moon can be used to run MoonsScript files directly from the command line, without needing a separate compile step. All MoonsScript files are compiled in memory as they are run.

~> moon my_script.moon

Any MoonScript files that are required will also be compiled and run automatically.

When an error occurs during runtime, the stack trace is rewritten to give line numbers from the original .moon file.

If you want to disable error rewriting, you can pass the -d flag. A full list of flags can be seen by passing the -h or --help flag.

moonc

moonc is used for transforming MoonsScript files into Lua files. It takes a list of files, compiles them all, and creates the associated .lua files in the same directories.

~> moonc my_script1.moon my_script2.moon ...

You can control where the compiled files are put using the -t flag, followed by a directory.

moonc can also take a directory as an argument, and it will recursively scan for all MoonScript files and compile them.

moonc can write to standard out by passing the -p flag.

Combined with linotify on linux, the -w flag can be used to watch all files that match the given search path for changes, and then compile them only when required.

A full list of flags can be seen by passing the -h or --help flag.

License (MIT)

Copyright (C) 2011 by Leaf Corcoran

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.