Skip to content
/ mochi Public
forked from i2y/mochi

Dynamically typed functional programming language

License

Notifications You must be signed in to change notification settings

pya/mochi

This branch is 2 commits ahead of, 23 commits behind i2y/mochi:master.

Folders and files

NameName
Last commit message
Last commit date

Latest commit

a6fcc9a · Jul 25, 2015
May 10, 2015
Jul 18, 2015
Jul 20, 2015
Jul 21, 2015
Jun 14, 2015
Jun 14, 2015
Jul 21, 2015
Apr 20, 2015
May 10, 2015
Apr 21, 2015
Jan 18, 2015
Apr 26, 2015
Jul 21, 2015
Jul 21, 2015
Jul 25, 2015
Jul 21, 2015
Jul 20, 2015
Dec 3, 2014
Apr 26, 2015
Apr 15, 2015

Repository files navigation

Mochi

Mochi is a dynamically typed programming language for functional programming and actor-style programming.

Its interpreter is written in Python3. The interpreter translates a program written in Mochi to Python3's AST / bytecode.

Features

  • Python-like syntax
  • Tail recursion optimization (self tail recursion only), and no loop syntax
  • Re-assignments are not allowed in function definition.
  • Basic collection type is a persistent data structure. (using Pyrsistent)
  • Pattern matching / Data types, like algebraic data types
  • Pipeline operator
  • Syntax sugar of anonymous function definition
  • Actor, like the actor of Erlang (using Eventlet)
  • Macro, like the traditional macro of Lisp
  • Builtin functions includes functions exported by itertools module, recipes, functools module and operator module

Examples

Factorial

def factorial(n, m):
    if n == 1:
        m
    else:
        factorial(n - 1, n * m)


factorial(10000, 1)
# => 28462596809170545189064132121198688...

# Or

def factorial:
    n: factorial(n, 1)
    0, acc: acc
    n, acc: factorial(n - 1, acc * n)

factorial(10000)
# => 28462596809170545189064132121198688...

FizzBuzz

def fizzbuzz(n):
    match [n % 3, n % 5]:
        [0, 0]: "fizzbuzz"
        [0, _]: "fizz"
        [_, 0]: "buzz"
        _: n

range(1, 31)
|> map(fizzbuzz)
|> pvector()
|> print()

Actor

def show():
    receive:
        message:
            print(message)
            show()

actor = spawn(show)

send('foo', actor)
actor ! 'bar' # send('bar', actor)

sleep(1)
# -> foo
# -> bar


'foo' !> spawn(show)

sleep(1)
# -> foo

['foo', 'bar'] !&> spawn(show)
# The meaning of the above is the same as the meaning of the following.
# spawn(show) ! 'foo'
# spawn(show) ! 'bar'

sleep(1)
# -> foo
# -> bar

def show_loop():
    receive:
        [tag, value]:
            print(tag, value)
            show_loop()

actor2 = spawn(show_loop)

actor2 ! ["bar", 2000]
sleep(1)
# -> bar 2000

['foo', 1000] !> spawn(show_loop)
sleep(1)
# -> foo 1000

[['foo', 1000],['bar', 2000]] !&> spawn(show_loop)
sleep(1)
# -> foo 1000
# -> bar 2000

Distributed Computing

# comsumer.mochi
from mochi.actor.mailbox import KombuMailbox, ZmqInbox

def consumer():
    receive:
        'exit':
            print('exit!')
        other:
            print(other)
            consumer()

mailbox = KombuMailbox('sqs://<access_key_id>@<secret_access_key>:80//',
                       '<queue_name>',
                       dict(region='<region>'))
spawn_with_mailbox(consumer, mailbox)

mailbox = ZmqInbox('tcp://*:9999')
spawn_with_mailbox(consumer, mailbox)

wait_all()
# producer.mochi
from mochi.actor.mailbox import KombuMailbox, ZmqOutbox

mailbox = KombuMailbox('sqs://<access_key_id>@<secret_access_key>:80//',
                       '<queue_name>',
                       dict(region='<region>'))
mailbox ! [1, 2, 3]
mailbox ! 'exit'

mailbox = ZmqOutbox('tcp://localhost:9999')
mailbox ! [4, 5, 6]
mailbox ! 'exit'

Flask

from flask import Flask

app = Flask('demo')

@app.route('/')
def hello():
    'Hello World!'

app.run()

RxPY

# usage: mochi -no-mp timer.mochi
# original:
# https://github.com/ReactiveX/RxPY/blob/master/examples/parallel/timer.py

import rx
import concurrent.futures
import time

seconds = [5, 1, 2, 4, 3]


def sleep(t):
    time.sleep(t)
    return t


def output(result):
    print('%d seconds' % result)


with concurrent.futures.ProcessPoolExecutor(5) as executor:
    rx.Observable.from_(seconds)
                 .flat_map((s) -> executor.submit(sleep, s))
                 .subscribe(output)

# 1 seconds
# 2 seconds
# 3 seconds
# 4 seconds
# 5 seconds

aif (Anaphoric macro)

macro aif(test, true_expr, false_expr):
    quasi_quote:
        it = unquote(test)
        if it:
            unquote(true_expr)
        else:
            unquote(false_expr)

aif([], first(it), "empty")
# => "empty"
aif([10, 20], first(it), "empty")
# => 10

Requirements

  • CPython >= 3.2 or PyPy >= 3.2.1
  • rply >= 0.7.2
  • pyrsistent >= 0.10.1
  • pathlib >= 1.0.1
  • eventlet >= 0.17.1
  • msgpack-python >= 0.4.6
  • typeannotations >= 0.1.0
  • astunparse>=1.2.2

Installation

$ pip3 install mochi

Optional Installation

$ pip3 install flask Flask-RESTful Pillow RxPY  # to run the examples
$ pip3 install pyzmq # to use ZmqInbox and ZmqOutbox
$ pip3 install kombu # to use KombuMailbox
$ pip3 install boto # to use SQS as transport of KombuMailbox

The following may occur when you run Mochi on PyPy.

ImportError: Importing zmq.backend.cffi failed with version mismatch, 0.8.2 != 0.9.2

In this case, please change the version of cffi to 0.8.2 using pip on PyPy.

$ pip3 uninstall cffi
$ pip3 install cffi==0.8.2

Usage

REPL

$ mochi
>>>

loading and running a file

$ cat kinako.mochi
print('kinako')
$ mochi kinako.mochi
kinako
$ mochi -no-mp kinako.mochi  # not apply eventlet's monkey patching
kinako

byte compilation

$ mochi -c kinako.mochi > kinako.mochic

running a byte-compiled file

$ mochi -e kinako.mochic
kinako
$ mochi -e -no-mp kinako.mochic  # not apply eventlet's monkey patching
kinako

generating .pyc

$ ls
kagami.mochi
$ cat kagami.mochi
print('kagami')
$ mochi
>>> import kagami
kagami
>>> exit()
$ ls
kagami.mochi kagami.pyc
$ python3 kagami.pyc
kagami

Or

$ mochi -pyc kagami.mochi > kagami.pyc
$ python3 kagami.pyc
kagami
$ mochi -pyc -no-mp kagami.mochi > kagami.pyc  # not apply eventlet's monkey patching
$ python3 kagami.pyc
kagami

Examples for each feature

Persistent data structures

[1, 2, 3]
# => pvector([1, 2, 3])

v(1, 2, 3)
# => pvector([1, 2, 3])

vec = [1, 2, 3]
vec2 = vec.set(0, 8)
# => pvector([8, 2, 3]
vec
# => pvector([1, 2, 3])
[x, y, z] = vec
x # => 1
y # => 2
z # => 3

get(vec, 0) # => 1
get(vec, 0, 2) # => [1, 2]

vec[0] # => 1
vec[0:2] # => [1, 2]

{'x': 100, 'y': 200}
# => pmap({'y': 200, 'x': 100})

ma = {'x': 100, 'y': 200}
ma.get('x') # => 100
ma.x # => 100
ma['x'] # => 100
ma2 = ma.set('x', 10000)
# => pmap({'y': 200, 'x': 10000})
ma # => pmap({'y': 200, 'x': 100})
get(ma, 'y') # => 200
ma['y'] # => 200

m(x=100, y=200)
# => pmap({'y': 200, 'x': 100})

s(1, 2, 3)
# => pset([1, 2, 3])

b(1, 2, 3)
# => pbag([1, 2, 3])

Function definitions

def hoge(x):
    'hoge' + str(x)

hoge(3)
# => hoge3

Pattern matching

lis = [1, 2, 3]

# Sequence pattern
match lis:
    [1, 2, x]: x
    _: None
# => 3

match lis:
    [1, &rest]: rest
    _: None

# => pvector([2, 3])


foo_map = {'foo' : 'bar'}

# Mapping pattern
match foo_map:
    {'foo' : value}: value
    _: None
# => 'bar'


# Type pattern
# <name of variable refers to type> <pattern>: <action>
match 10:
    int x: 'int'
    float x: 'float'
    str x: 'str'
    bool x: 'bool'
    _: 'other'
# => 'int'

match [1, 2, 3]:
    [1, str x, 3]: 'str'
    [1, int x, 3]: 'int'
    _: 'other'
# => 'int'

num = union(int, float)
vector nums[num]
vector strs[str]

match nums([1, 2, 3]):
    nums[x, y, z]: z
    strs[x, y, z]: x
# => 3

Positive = predicate(-> $1 > 0)
Even = predicate(-> $1 % 2 == 0)
EvenAndPositive = predicate(-> ($1 % 2 == 0) and ($1 >= 0))

match 10:
    EvenAndPositive n: str(n) + ':Even and Positive'
    Even n: str(n) + ':Even'
    Positive n: str(n) + ':Positive'

# => 10:Even and Positive


# Or pattern
match ['foo', 100]:
    ['foo' or 'bar', value]: value
    _: 10000
# => 100

match ['foo', 100]:
    [str x or int x, value]: value
    _: 10000
# => 100


# Record pattern
record Person(name, age)

foo = Person('foo', 32)

match foo:
    Person('bar', age):
        'bar:' + str(age)
    Person('foo', age):
        'foo:' + str(age)
    _: None
# => 'foo:32'

Records

record Mochi
record AnkoMochi(anko) < Mochi
record KinakoMochi(kinako) < Mochi

anko_mochi = AnkoMochi(anko=3)

isinstance(anko_mochi, Mochi)
# => True
isinstance(anko_mochi, AnkoMochi)
# => True
isinstance(anko_mochi, KinakoMochi)
# => False

match anko_mochi:
    KinakoMochi(kinako): 'kinako ' * kinako + ' mochi'
    AnkoMochi(anko): 'anko ' * anko + 'mochi'
    Mochi(_): 'mochi'
# => 'anko anko anko mochi'


record Person(name, age):
    def show(self):
        print(self.name + ': ' + self.age)

foo = Person('foo', '32')
foo.show()
# -> foo: 32

# runtime type checking
record Point(x:int, y:int, z:optional(int))
Point(1, 2, None)
# => Point(x=1, y=2, z=None)
Point(1, 2, 3)
# => Point(x=1, y=2, z=3)
Point(1, None, 3)
# => TypeError

Bindings

x = 3000
# => 3000

[a, b] = [1, 2]
a
# => 1
b
# => 2

[c, &d] = [1, 2, 3]
c
# => 1
d
# => pvector([2, 3])

Data types, like algebraic data types

data Point:
    Point2D(x, y)
    Point3D(x, y, z)

# The meaning of the above is the same as the meaning of the following.
# record Point
# record Point2D(x, y) < Point
# record Point3D(x, y, z) < Point

p1 = Point2D(x=1, y=2)
# => Point2D(x=1, y=2)

p2 = Point2D(3, 4)
# => Point2D(x=3, y=4)

p1.x
# => 1

Pattern-matching function definitions

data Point:
    Point2D(x, y)
    Point3D(x, y, z)

def offset:
    Point2D(x1, y1), Point2D(x2, y2):
        Point2D(x1 + x2, y1 + y2)
    Point3D(x1, y1, z1), Point3D(x2, y2, z2):
        Point3D(x1 + x2, y1 + y2, z1 + z2)
    _: None

offset(Point2D(1, 2), Point2D(3, 4))
# => Point2D(x=4, y=6)
offset(Point3D(1, 2, 3), Point3D(4, 5, 6))
# => Point3D(x=5, y=7, z=9)

def show:
    int x, message: print('int', x, message)
    float x, message: print('float', x, message)
    _: None

show(1.0, 'msg')
# -> float 1.0 msg
# => None

FileMode = options('r', 'w', 'a', 'r+', 'w+', 'a+')

def open_file:
    str path, FileMode mode:
        open(path, mode)
    str path:
        open(path, 'r')

Anonymous function

# Arrow expression.
add = (x, y) -> x + y
add(1, 2)
# => 3

add = -> $1 + $2
add(1, 2)
# => 3

foo = (x, y) ->
    if x == 0:
        y
    else:
        x

foo(1, 2)
# => 1

foo(0, 2)
# => 2

pvector(map(-> $1 * 2, [1, 2, 3]))
# => pvector([2, 4, 6])

Pipeline operator

add = -> $1 + $2
2 |> add(10) |> add(12)
# => 24
None |>? add(10) |>? add(12)
# => None

Lazy sequences

def fizzbuzz(n):
    match [n % 3, n % 5]:
        [0, 0]: "fizzbuzz"
        [0, _]: "fizz"
        [_, 0]: "buzz"
        _: n


result = range(1, 31) |> map(fizzbuzz)
pvector(result)
# => pvector([1, 2, fizz, 4, 'buzz', 'fizz', 7, 8, 'fizz', 'buzz', 11, 'fizz', 13, 14, 'fizzbuzz', 16, 17, 'fizz', 19, 'buzz', 'fizz', 22, 23, 'fizz', 'buzz', 26, 'fizz', 28, 29, 'fizzbuzz'])
pvector(result)
# => pvector([])
pvector(result)
# => pvector([])


# Iterator -> lazyseq
lazy_result = range(1, 31) |> map(fizzbuzz) |> lazyseq()
pvector(lazy_result)
# => pvector([1, 2, fizz, 4, 'buzz', 'fizz', 7, 8, 'fizz', 'buzz', 11, 'fizz', 13, 14, 'fizzbuzz', 16, 17, 'fizz', 19, 'buzz', 'fizz', 22, 23, 'fizz', 'buzz', 26, 'fizz', 28, 29, 'fizzbuzz'])
pvector(lazy_result)
# => pvector([1, 2, fizz, 4, 'buzz', 'fizz', 7, 8, 'fizz', 'buzz', 11, 'fizz', 13, 14, 'fizzbuzz', 16, 17, 'fizz', 19, 'buzz', 'fizz', 22, 23, 'fizz', 'buzz', 26, 'fizz', 28, 29, 'fizzbuzz'])
pvector(lazy_result)
# => pvector([1, 2, fizz, 4, 'buzz', 'fizz', 7, 8, 'fizz', 'buzz', 11, 'fizz', 13, 14, 'fizzbuzz', 16, 17, 'fizz', 19, 'buzz', 'fizz', 22, 23, 'fizz', 'buzz', 26, 'fizz', 28, 29, 'fizzbuzz'])

Trailing closures

# The following trailing closure expression is passed to a function as the function’s first argument.
result = map([1, 2, 3]) ->
    print($1)
    $1 * 2

print(doall(result))

# -> 1
# -> 2
# -> 3
# => pvector([2, 4, 6])


def foreach(closure, seq):
    doall(filter(closure, seq))

# The following trailing closure expression is passed to a function as the function’s first argument.
foreach([1, 2, 3]) (item) ->
    new_item = item * 100
    print(new_item)

# -> 100
# -> 200
# -> 300
# => pvector([])

# Or

def foreach(seq, closure):
    doall(filter(closure, seq))

# The following trailing closure expression is passed to a function as the function’s final argument.
foreach([1, 2, 3]) @ (item) ->
    new_item = item * 100
    print(new_item)

# -> 100
# -> 200
# -> 300
# => pvector([])

Short form for keyword arguments and dict keys

def foo(a, b, c):
    a + b + c

a = 1
b = 2
c = 3

# This is the same as foo(a=a, b=b, c=c)
foo(=a, =b, =c))
# => 6

# This is the same as {'a': a, 'b': b}
{=a, =b}
# => pmap({'a': 1, 'b': 2})

Macros

macro rest_if_first_is_true(first, &args):
     match first:
         quote(True): quasi_quote(v(unquote_splicing(args)))
         _: quote(False)

rest_if_first_is_true(True, 1, 2, 3)
# => pvector([1, 2, 3])
rest_if_first_is_true("foo", 1, 2, 3)
# => False

macro pipeline(&args):
    [Symbol('|>')] + args

pipeline([1, 2, 3],
         map(-> $1 * 2),
         filter(-> $1 != 2),
         pvector())
# => pvector([4, 6])

Including a file at compile time

$ cat anko.mochi
x = 10000
y = 20000
require 'anko.mochi'
x
# => 10000

x = 30000

require 'anko.mochi' # include once at compile time
x
# => 30000

Module

module Math:
    export add, sub

    def add(x, y):
        x + y

    def sub(x, y):
        x - y

Math.add(1, 2)
# => 3
$ cat foobar.mochi
foo = 'foo'
bar = 'bar'
require 'foobar.mochi'
[foo, bar]
# => pvector(['foo', 'bar'])

foo = 'foofoofoo'

module X:
    export foobar
    require 'foobar.mochi'
    def foobar:
        [foo, bar]

X.foobar()
# => pvector(['foo', 'bar'])

[foo, bar]
# => pvector(['foofoofoo', 'bar'])

TODO

  • Improve documentation
  • Improve parsing
  • Support type annotation

License

MIT License

Contributors

https://github.com/i2y/mochi/graphs/contributors

About

Dynamically typed functional programming language

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Python 98.9%
  • Shell 1.1%