Our job is to make your job easier




RapydScript (pronounced 'RapidScript') is a pre-compiler for JavaScript, similar to CoffeeScript, but with cleaner, more readable syntax. The syntax is very similar to Python, but allows JavaScript as well. This project was written as an alternative to Pyjamas for those wishing Python-like JavaScript without the extra overhead and complexity Pyjamas introduces.

Live Demos for every example can be found in the RapydScript Showcase

You can download RapydScript from our repository: https://github.com/atsepkov/RapydScript


Build Status npm

What is RapydScript?

RapydScript is a pre-compiler for JavaScript. The syntax is very similar to Python, but allows JavaScript as well. This project was written as a cleaner alternative to CoffeeScript. Here is a quick example of a high-performance Fibonacci function in RapydScript and the JavaScript it produces after compilation:

```python def memoize(f): memo = {} return def(x): if x not in memo: memo[x] = f(x) return memo[x]

@memoize def fib(n): if n == 0: return 0 elif n == 1: return 1 else: return fib(n-1) + fib(n-2) JavaScript:javascript function $rapyd$_in(val, arr) { if (arr instanceof Array || typeof arr === "string") { return arr.indexOf(val) != -1; } return arr.hasOwnProperty(val); } function memoize(f) { var memo = {}; return function(x) { if (!($rapyd$_in(x, memo))) { memo[x] = f(x); } return memo[x]; }; }

fib = memoize(function fib(n) { if (n === 0) { return 0; } else if (n === 1) { return 1; } else { return fib(n - 1) + fib(n - 2); } }); ```

Here are just a few examples of cleaner RapydScript syntax:

| RapydScript | JavaScript | | ------------------------ | ------------------------------------------------------------------------------ | | foo = 1 | var foo = 1; | | thing in stuff | stuff.indexOf(thing) != -1 | | a < b < c | a < b && b < c | | def(a, b='foo'): | function(a, b) { if (typeof b === 'undefined') b = 'foo'; } | | array[-1] | array[array.length-1] | | array[3:5] = [7, 8, 9] | [].splice.apply(array, [3, 5-3].concat([ 7, 8, 9 ])) | | [5 to 15] | Array.apply(null, {length: 11}).map(Number.call, function(n){ return n+5; }) | | [a**2 for a in array] | // Really, really long for-loop code... | | {a: "?"} == {a: "?"} | // Bet you didn't know deep equality with no overhead was possible |

RapydScript allows to write your JavaScript app in a language much closer to Python without the overhead that other similar frameworks introduce (the performance is the same as with pure JavaScript). To those familiar with CoffeeScript, RapydScript is like CoffeeScript, but inspired by Python's readability rather than Ruby's cleverness. To those familiar with Pyjamas, RapydScript brings many of the same features and support for Python syntax without the same overhead. Don't worry if you've never used either of the above-mentioned compilers, if you've ever had to write your code in pure JavaScript you'll appreciate RapydScript. RapydScript combines the best features of Python as well as JavaScript, bringing you features most other Pythonic JavaScript replacements overlook. Here are a few features of RapydScript:

  • == compiles to deep equality and uses clever optimizations to avoid performance overhead
  • type inference that allows for hybrid-typing similar to TypeScript
  • intelligent scoping (no need for repetitive var or new keywords)
  • intelligent code optimizations based on context
  • much cleaner code than native JavaScript
  • optional function arguments that work just like Python (func(third='foo'))
  • decorators, list comprehensions, dict comprehensions, verbose regex, starargs, kwargs, you name it
  • ability to use both, Python's methods/functions and JavaScript's alternatives
  • similar to above, ability to use both, Python's and JavaScript's tutorials (as well as widgets)
  • classes that work and feel very similar to Python
  • inheritance system that's both, more powerful than Python and cleaner than JavaScript (single inheritance w/ mixins);
  • pythonic import system (you can also use require())
  • support for object literals with anonymous functions, like in JavaScript
  • ability to invoke any JavaScript/DOM object/function/method as if it's part of the same framework, without the need for special syntax
  • it's self-hosting, the compiler is itself written in RapydScript and compiles into JavaScript

Let's not waste any more time with the introductions, however. The best way to learn a new language/framework is to dive in.

Table of Contents

<!-- START doctoc generated TOC please keep comment here to allow auto update --> <!-- DON'T EDIT THIS SECTION, INSTEAD RE-RUN doctoc TO UPDATE -->

<!-- END doctoc generated TOC please keep comment here to allow auto update -->


First make sure you have installed the latest version of node.js (You may need to restart your computer after this step).

From NPM for use as a command line app:

npm install rapydscript -g

From NPM for programmatic use:

npm install rapydscript

From Git:

git clone git://github.com/atsepkov/RapydScript.git
cd RapydScript
npm link .

If you're using OSX, you can probably use the same commands (let me know if that's not the case). If you're using Windows, you should be able to follow similar commands after installing node.js and git on your system.


If you have questions, bug reports, or feature requests, feel free to post them on our mailing list:

I bundled a few demos with RapydScript itself, but several members of the community put together much better demos. If you would like to take a look at them to see what's possible with RapydScript, here are some examples:

GlowScript is a WebGL-based environment and a physics engine, originally written in Python for the desktop, the author ported it to the browser using RapydScript compiler.

This includes the demos from RapydScript's examples directory, as well as a few others.

This is a collection of very cool demos, showcasing RapydScript's similarity to real Python and at the same time its ability to work with other JavaScript. It relies on a JavaScript port of NodeBox (which was originally written in Python). NodeBox was ported from Python to JavaScript to allow cross-platform compatibility. Ironically, the original demos from Python version of NodeBox now work with JavaScript version of NodeBox with few changes (and sometimes none at all) by using RapydScript.

RapydScript making use of GlowScript, another project done by a member of our community

https://github.com/adousen/RapydScript-pyjTransformer An in-browser compiler that allows you to use uncompiled RapydScript files in the browser directly via script tags: html <script type="text/pyj" otype="text/jsx" src="helloworld.pyj" async="false"></script>

My solutions to Project Euler challenges in RapydScript. For those unfamiliar with projecteuler.net, it's a collection of mathematical puzzles for developers testing their ability to come up with clever/efficient algorithms as well as brevity/elegance of their chosen language. While Python and Ruby are popular choices, barely any solutions are in JavaScript (probably due to the language's arcane syntax and error handling and very limited utility for mathematical challenges out of the box). RapydScript, however, does quite well - sometimes allowing for identical solution as Python, sometimes a more clever one. Execution speed is typically faster than Python, but in some cases lags behind (i.e. when Python version uses sets or optimized numpy logic).


NOTE: ES6 mode is getting stable enough where I feel comfortable making it the default soon, after which ES5 mode will be deprecated in favor of lighter codebase. To those needing support for older platforms in the future, I recommend sending RapydScript in conjunction with Babel.js.

Once you have installed RapydScript, compiling your application is as simple as running the following command:

rapydscript <location of main file> [options]

By default this will dump the output to STDOUT, but you can specify the output file using --output option. The generated file can then be referenced in your html page the same way as you would with a typical JavaScript file. If you're only using RapydScript for classes and functions, then you're all set. If you're using additional Python methods, such as range(), print(), list.append(), list.remove(), then you will want to link RapydScript's stdlib.js in your html page as well. There are two ways of doing this, one is to include it as a JavaScript file in your HTML, the other is to include it as an import in your source code and let RapydScript pull it in automatically.

RapydScript can take multiple input files. It's recommended that you pass the input files first, then pass the options. RapydScript will parse input files in sequence and apply any compression options. The files are parsed in the same global scope, that is, a reference from a file to some variable/function declared in another file will be matched properly.

If you want to read from STDIN instead, pass a single dash instead of input files.

The available options are:

-o, --output       Output file (default STDOUT).
-x, --execute      Execute the file in-place (no compiled output generated)
-b, --bare         Omit scope-protection wrapper around generated code
-p, --prettify     Beautify output/specify output options.
-V, --version      Print version number and exit.
-t, --test         Run unit tests, making sure the compiler produces usable code, you can specify a file or test everything
    --bench        Run performance tests, you can specify a file or test everything (note that these tests take a while to run)
-6, --es6          Build code for ES6, cleaner output with support for more features (EXPERIMENTAL)
-m, --omit-baselib Omit base library from generated code, make sure you're including baselib.js if you use this
-i, --auto-bind    Automatically bind methods to the class they belong to (more Pythonic, but could interfere with other JS libs)
-h, --help         Print usage and more information on each of these options
    --self         Compile the compiler itself
    --stats        Show compilation metrics in STDERR (time to parse, generate code, etc.)
    --dd           Drop specified decorators (takes a comma-separated list of decorator names)
    --di           Drop specified imports (takes a comma-separated list of import names)
-l, --lint         Check file for errors and compilation problems

You can also use RapydScript as your main system's scripting language (similar to how some prefer to write their system scripts in Python). To do so, add the following line to the top of your script files:

#!/usr/bin/env rapydscript -x

Now simply invoking the file (assuming it has execute permissions on it) should execute it.

Getting Started

As you read the following sections, I suggest you start a RapydScript shell (by typing rapydscript without arguments in your terminal) and follow along. You'll be able to see both, the generated JavaScript, and the output produced by the given RapydScript command.

You can also run the compiler in the browser. Just add a script tag linking back to lib/rapydscript.js and invoke the compiler as rapydscript.compile(stringOfCode, options).

Like JavaScript, RapydScript can be used to create anything from a quick function to a complex web-app. RapydScript can access anything regular JavaScript can, in the same manner. Let's say we want to write a function that greets us with a "Hello World" pop-up. The following code will do it:

python def greet(): alert("Hello World!")

Once compiled, the above code will turn into the following JavaScript:

javascript function greet() { alert("Hello World!"); }

Now you can reference this function from other JavaScript or the page itself (using "onclick", for example). For our next example, let's say you want a function that computes factorial of a number:

python def factorial(n): if n == 0: return 1 return n * factorial(n-1)

Now all we need is to tie it into our page so that it's interactive. Let's add an input field to the page body and a cell for displaying the factorial of the number in the input once the input loses focus.

html <input id="user-input" onblur="computeFactorial()"></input> <div id="result"></div>

NOTE: To complement RapydScript, I have also written RapydML (http://bitbucket.org/pyjeon/rapydml), which is a pre-compiler for HTML (just like RapydScript is a pre-compiler for JavaScript).

Now let's implement computeFactorial() function in RapydScript:

python def computeFactorial(): n = document.getElementById("user-input").value document.getElementById("result").innerHTML = factorial(n)

Again, notice that we have access to everything JavaScript has access to, including direct DOM manipulation. Once compiled, this function will look like this:

javascript function computeFactorial() { var n; n = document.getElementById("user-input").value; document.getElementById("result").innerHTML = factorial(n); }

Notice that RapydScript automatically declares variables in local scope when you try to assign to them. This not only makes your code shorter, but saves you from making common JavaScript mistake of overwriting a global. For more information on controlling variable scope, see Scope Control section.

Leveraging other APIs

Aside from Python-like stdlib, RapydScript does not have any of its own APIs. Nor does it need to, there are already good options available that we can leverage instead. If we wanted, for example, to rewrite the above factorial logic using jQuery, we could easily do so:

python def computeFactorial(): n = $("#user-input").val() $("#result").text(factorial(n))

Many of these external APIs, however, take object literals as input. Like with JavaScript, you can easily create those with RapydScript, the same way you would create one in JavaScript, or a dictionary in Python:

python styles = { 'background-color': '#ffe', 'border-left': '5px solid #ccc', 'width': 50, }

Now you can pass it to jQuery:


Another feature of RapydScript is ability to have functions as part of your object literal. JavaScript APIs often take callback/handler functions as part of their input parameters, and RapydScript lets you create such object literal without any quirks/hacks:

python params = { width: 50, height: 30, onclick: def(event): alert("you clicked me"), onmouseover: def(event): $(this).css('background', 'red') , onmouseout: def(event): # reset the background $(this).css('background', '') }

Note the comma on a new line following a function declaration, it needs to be there to let the compiler know there are more attributes in this object literal, yet it can't go on the same line as the function since it would get parsed as part of the function block. Like Python, however, RapydScript supports new-line shorthand using a ;, which you could use to place the comma on the same line:

python hash = { 'foo': def(): print('foo');, 'bar': def(): print('bar') }

It is because of easy integration with JavaScript's native libraries that RapydScript keeps its own libraries to a minimum. For example, it does not implement string interpolation, like native Python. However, by using sprintf.js library (https://github.com/alexei/sprintf.js) you can reproduce the same behavior in RapydScript:

python string = vsprintf('%d bottles of %s on the wall', (99, 'beer'))

Take a look at the examples directory to see RapydScript integration with jQuery, jQuery-UI, D3, and Google Charts.

Anonymous Functions

Like JavaScript, RapydScript allows the use of anonymous functions. In fact, you've already seen the use of anonymous functions in previous section when creating an object literal ('onmouseover' and 'onmouseout' assignments). This is similar to Python's lambda function, except that the syntax isn't awkward like lambda, and the function isn't limited to one line. The following two function declarations are equivalent:

```python def factorial(n): if n == 0: return 1 return n * factorial(n-1)

factorial = def(n): if n == 0: return 1 return n * factorial(n-1) ```

This might not seem like much at first, but if you're familiar with JavaScript, you know that this can be extermely useful to the programmer, especially when dealing with nested functions, which are a bit syntactically awkward in Python (it's not immediatelly obvious that those can be copied and assigned to other objects). To illustrate the usefulness, let's create a method that creates and returns an element that changes color while the user keeps the mouse pressed on it.

python def makeDivThatTurnsGreen(): div = $('<div></div>') turnGreen = def(event): div.css('background', 'green') div.mousedown(turnGreen) resetColor = def(event): div.css('background', '') div.mouseup(resetColor) return div

At first glance, anonymous functions might not seem that useful. We could have easily created nested functions and assigned them instead. By using anonymous functions, however, we can quickly identify that these functions will be bound to a different object. They belong to the div, not the main function that created them, nor the logic that invoked it. The best use case for these is creating an element inside another function/object without getting confused which object the function belongs to.

Additionally, as you already noticed in the previous section, anonymous functions can be used to avoid creating excessive temporary variables and make your code cleaner:

math_ops = {
    add:    def(a, b): return a+b;,
    sub:    def(a, b): return a-b;,
    mul:    def(a, b): return a*b;,
    div:    def(a, b): return a/b;,
    roots:  def(a, b, c):
        r = Math.sqrt(b*b - 4*a*c)
        d = 2*a
        return (-b + r)/d, (-b - r)/d

I'm sure you will agree that the above code is cleaner than declaring 5 temporary variables first and assigning them to the object literal keys after. Note that the example puts the function header (def()) and content on the same line. I'll refer to it as function inlining. This is meant as a feature of RapydScript to make the code cleaner in cases like the example above. While you can use it in longer functions by chaining statements together using ;, a good rule of thumb (to keep your code clean) is if your function needs semi-colons ask yourself whether you should be inlining, and if it needs more than 2 semi-colons, the answer is probably no (note that you can also use semi-colons as newline separators within functions that aren't inlined, as in the example in the previous section).


Like Python, RapydScript supports function decorators.

```python def makebold(fn): def wrapped(): return "<b>" + fn() + "</b>" return wrapped

def makeitalic(fn): def wrapped(): return "<i>" + fn() + "</i>" return wrapped

@makebold @makeitalic def hello(): return "hello world"

hello() # returns "<b><i>hello world</i></b>"

def multiple_args(front, back, repeat): return def(f): return def(): string = front + f() + back container = [] for i in range(repeat): container.push(string) return container.join(',')

@multiple_args('{{', '}}', 3) def foo(): return 'foo'

foo() # returns '{{foo}},{{foo}},{{foo}}' ```

Function Annotations

Like Python 3 and TypeScript, RapydScript allows function annotations. They look like this:

```python def foo(a: Number, b: String) -> String: return b + str(a)

def bar(x: "x coordinate of the point", y: "y coordinate of the point") -> "distance from 0, 0": return Math.sqrt(x2 + y2) ```

Python is more flexible about them since they're treated as a fancy comment. In TypeScript they are actually significant and used to enforce static typing. In current version of RapydScript they are significant if they resolve to a static type. See "Static Typing" section for more details.

Hybrid And Static Typing

RapydScript allows you to use both dynamic and static typing (similar to TypeScript or Flow). Unlike TypeScript, however, RapydScript goes a step further with error detection and code optimization when you use static typing. I call this approach hybrid-typing.

The idea behind hybrid-typing is to combine the best of dynamic typing and static typing. Dynamic typing gives more flexibility to the user and the code can more easily be reused in other places. Static typing gives the compiler more information about the context, allowing for better linting, error checking and optimizations.

Some examples of logic the compiler can do for statically-typed code are:

  • raise errors when the type of a statically declared variable is violated
  • raise errors when an incompatible operation is performed on two variables (no more {} + [])
  • inline certain functions/operations
  • precompute certain values, assuming enough context/information about their state

You may wonder why compilers like TypeScript can't do the same. There are two parts to the answer:

  • TypeScript tries to be more compatible with native JavaScript and as a result has to be more conservative with its optimizations
  • RapydScript uses some non-JavaScript design patterns, as well as banning some traditional JavaScript anti-patterns, this stricter language syntax allows RapydScript to be more aggressive with its optimizations and errors

You don't need to do anything to benefit from RapydScript's type inference, your variable types are automatically resolved behind the scenes. However, you can give the compiler even more hints about variables types (and as a result benefit from its optimizations and error detection even more). Do do so, simply label function inputs and outputs via type annotations as mentioned in previous section:

python def optimizeMe(a: Number, b:Number) -> Number: ...

The compiler will now check that the arguments passed to the function are indeed numbers and the output is also a number, throwing an error if it detects an incorrect input or return. Additionally, the compiler will use inputs and outputs of this function to resolve the types of other variables in your code, further assisting itself in optimizations and error checking.

The best way to use hybrid-typing is to start out with dynamically-typed code and add types to it once you finalize and refactor the function. While static typing is optional, I definitely recommend you do it. The best way to think about static typing is as a free unit test for your logic.

Self-Executing Functions

RapydScript wouldn't be useful if it required work-arounds for things that JavaScript handled easily. If you've worked with JavaScript or jQuery before, you've probably seen the following syntax:

    // some logic here

This code calls the function immediately after declaring it instead of assigning it to a variable. Python doesn't have any way of doing this. The closest work-around is this:

def tmp(args):
    # some logic here

While it's not horrible, it did litter our namespace with a temporary variable. If we have to do this repeatedly, this pattern does get annoying. This is where RapydScript decided to be a little unorthodox and implement the JavaScript-like solution:

    # some logic here

A close cousin of the above is the following code (passing current scope to the function being called):

    // some logic here

With RapydScript equivalent of:

    # some logic here

There is also a third alternative, that will pass the arguments as an array:

def(a, b):
    # some logic here
.apply(this, [a, b])

Chaining Blocks

As seen in previous section, RapydScript will bind any lines beginning with . to the outside of the block with the matching indentation. This logic isn't limited to the .call() method, you can use it with .apply() or any other method/property the function has assigned to it. This can be used for jQuery as well:

.css('background-color', 'red')

The only limitation is that the indentation has to match, if you prefer to indent your chained calls, you can still do so by using the \ delimiter:

    .css('background-color', 'red')\

You may use this feature to define and call the function immediately:

def(one, two):
.call(this, 1, 2)

JavaScript's parenthesized syntax will work as well:

(def(one, two):
)(1, 2)

Some of you might welcome this feature, some of you might not. RapydScript always aims to make its unique features unobtrusive to regular Python, which means that you don't have to use them if you disagree with them. RapydScript implements do/while loops via similar syntax as well:

a = 0
    a += 1
.while a < 1

Function calling with optional arguments

RapydScript supports the same function calling format as Python. You can have named optional arguments, create functions with variable numbers of arguments and variable number of named arguments. Some examples will illustrate this best:

def foo(a, b=2):
    return [a, b]
foo(1) == foo(1, 2) == [1, 2]

def bar(a, b, c):
    return [a, b, c]
bar(3, 2, 1) == bar(c=1, b=2, a=3) == [3, 2, 1]

def baz(a, *args):
    return [a, args]
baz(1, 2, 3) == [1, [2, 3]]

def qux(a, b=2, **kwargs):
    return [a, b, kwargs]
qux(1, b=3, c=4) == [1, 3, {c:4}]

RapydScript is not as strict as Python when it comes to validating function arguments. This is both for performance and to make it easier to interoperate with other JavaScript libraries. So if you do not pass enough arguments when calling a function, the extra arguments will be set to undefined instead of raising a TypeError, as would be the case in Python. Similarly, when mixing **kwargs and optional arguments, RapydScript will not complain if an optional argument is specified twice, it will use the one specified in kwargs.

RapydScript will also create a separate object for an optional argument each time the function is called. This makes it slightly less efficient, but prevents the common bug in python caused by using a mutable object literal as the default value for an optional argument.

Inferred Tuple Packing/Unpacking

Like Python, RapydScript allows inferred tuple packing/unpacking and assignment. While inferred/implicit logic is usually bad, it can sometimes make the code cleaner, and based on the order of statements in the Zen of Python, 'beautiful' takes priority over 'explicit'. For example, if you wanted to swap two variables, the following looks cleaner than explicitly declaring a temporary variable:

a, b = b, a

Likewise, if a function returns multiple variables, it's cleaner to say:

a, b, c = fun()

rather than:

tmp = fun()
a = tmp[0]
b = tmp[1]
c = tmp[2]

Since JavaScript doesn't have tuples, RapydScript uses arrays for tuple packing/unpacking behind the scenes, but the functionality stays the same. Note that unpacking only occurs when you're assigning to multiple arguments:

a, b, c = fun()     # gets unpacked
tmp = fun()         # no unpacking, tmp will store an array of length 3

Unpacking can also be done in for loops (which you can read about in later section):

for index, value in enumerate(items):
    print(index+': '+value)

Tuple packing is the reverse operation, and is done to the variables being assigned, rather than the ones being assigned to. This can occur during assignment or function return:

def fun():
    return 1, 2, 3

To summarize packing and unpacking, it's basically just syntax sugar to remove obvious assignment logic that would just litter the code. For example, the swap operation shown in the beginning of this section is equivalent to the following code:

tmp = [b, a]
a = tmp[0]
b = tmp[1]

Python vs JavaScript

RapydScript allows you to use both, Python and JavaScript names for the methods. For example, we can 'push()' a value to array, as well as 'append()' it:

arr = []
print(arr) # outputs [2,4]

In order to use Python's methods, you will have to include RapydScript's stdlib.js. There are two ways of doing this. One is by adding the following line in your html page:

<script type="text/javascript" src='stdlib.js'></script>

The other way is to include the following line at the top of your main RapydScript file:

import stdlib

The advantage of the second method is that the library will automatically be pulled in without having to manually copy it over into your JavaScript directory. The first include method, however, might be useful when you have multiple independently compiled RapydScript programs on your page and don't want the overhead of the same stdlib included in each one.

In order to simulate Python-like methods, I had to bastardize a couple of JavaScript's own methods. For example, array.pop() has been overwritten to work like Python's pop() (or JavaScript's splice()):

arr.pop()       # removes last element (expected behavior in JavaScript and Python)
arr.pop(2)      # removes third element (expected behavior in Python, but not JavaScript)
arr.splice(2,1) # removes third element (expected behavior in JavaScript, but not Python)

There is a subtle difference between the last 2 lines above, arr.pop(2) will return a single element, arr.splice(2,1) will return an array containing that single element. Most of the keywords are interchangeable as well:

RapydScript     JavaScript

None/null       null
False/false     false
True/true       true
undefined       undefined
this            this
NaN             NaN
Infinity        Infinity

The basic operators like +, +=, -, -=, /, /=, *, *= work the same way in both languages as well. The JavaScript operators, however, are not supported. You will have to use Python versions of those. If you're unfamiliar with them, here is the mapping RapydScript uses:

RapydScript     JavaScript

==              deep equality (JS has no equivalent)
!=              deep inequality (JS has no equivalent)
and             &&
or              ||
not             !
is              ===
is not          !==
+=1             ++
-=1             --
**              Math.pow()
//              floor division

Admittedly, is is not exactly the same thing in Python as === in JavaScript, but JavaScript is quirky when it comes to comparing objects anyway.

One important feature of RapydScript is its deep equality operators (==, !=). They have been carefully crafted through a few compilation tricks to experience no overhead over JavaScript's traditional operators yet return correct result when comparing copies of objects.

```python [1,2,3] == [1,[2,3]] # False [[1,2],3] == [1,[2,3]] # False [1,[2,3]] == [1,[2,3]] # True

class Stuff: def init(self, data): self.data = data

Stuff(1) == Stuff(1) # True Stuff(1) == Stuff(2) # False Stuff(1) == {data: 1} # False ```

In rare cases RapydScript might not allow you to do what you need to, and you need access to pure JavaScript. When that's the case, you can wrap your JavaScript in a string, passing it to JS() method. Code inside JS() method is not a sandbox, you can still interact with it from normal RapydScript:

JS('a = {foo: "bar", baz: 1};')
print(a.foo)    # prints "bar"

One last thing to note is the difference between print() and console.log, print() is part of RapydScript's stdlib, and is designed to work similar to Python's print statement, console.log() is JavaScript's version of debug output, which is more powerful but also more tedious when you just want a quick output. Here are some examples:

arr = [1,2,3,4]
print(arr)          # [1,2,3,4]
console.log(arr)    # [1,2,3,4]

arr2 = [[1,2],[3,4]]
print(arr2)         # [[1,2],[3,4]]
console.log(arr2)   # [Array[2], Array[2]]

hash = {'dogs': 1, 'cats': 2}
print(hash)         # {"dogs":1,"cats":2}
console.log(hash)   # Object

One other topic I debated for a while is handling of conditionals. In Python, if you wanted a one-liner conditional that gets assigned to a variable, you would write something like this:

foo = bar if baz else 10

In JavaScript, the equivalent logic would be written as follows:

var foo = baz ? bar : 10

Which one looks cleaner is subject to personal preference (I'm used to seeing condition first in most of the if statements, so the second way makes more sense to me). But where the second approach wins is when dealing with anonymous functions. Indeed, Python doesn't have to handle them, yet RapydScript does. As a result, I decided to go with JavaScript's approach here. This also allows me to assign functions here, same way as JavaScript would:

foo = baz ? def(): return bar; : def(): return 10


RapydScript loops work like Python, not JavaScript. You can't, for example, use 'for(i=0;i<max;i++)' syntax. You can, however, loop through arrays using 'for ... in' syntax without worrying about the extra irrelevant attributes regular JavaScript returns.

animals = ['cat', 'dog', 'mouse', 'horse']
for animal in animals:
    print('I have a '+animal)

If you need to use the index in the loop as well, you can do so by using enumerate():

for index, animal in enumerate(animals):
    print("index:"+index, "animal:"+animal)

Like in Python, if you just want the index, you can use range:

for index in range(len(animals)):           # or range(animals.length)
    print("animal "+index+" is a "+animals[index])

When possible, RapydScript will automatically optimize the loop for you into JavaScript basic syntax, so you're not missing much by not being able to call it directly. However, if for some reason you really need to, you always can:

for JS('i = 0; i < 50; i++'):

Dict and List Comprehensions

RapydScript also supports list comprehensions, using Python syntax. Instead of the following, for example:

myArray = []
for index in range(1,20):
    if index*index % 3 == 0:

You could write this:

myArray = [i*i for i in range(1,20) if i*i%3 == 0]

Dict comprehensions are also supported:

alphabet = {String.charFromCode(64+n): n for n in [1 to 26]}

Inclusive/Exclusive Sequences

Like Python, RapydScript has a range() function. While powerful, the result it generates isn't immediately obvious when looking at the code. It's a minor pet peeve, but the couple extra seconds trying to visually parse it and remember that it's not inclusive can detract from the code flow. To remedy this, RapydScript borrows to/til operators from LiveScript (also known as human-readable versions of Ruby's ../...). The following 4 lines of code are equivalent, for example:

a = [3 to 8]
a = [3 til 9]
a = range(3, 9)
a = [3, 4, 5, 6, 7, 8]

You can also use sequences within loops:

for i in [1 to 5]:

Or in list comprehensions:

[i*i for i in [1 to 6] if i%2 == 0]

The to/til statement gets converted to range() at compile time, and therefore can support variables or even expressions for start and end of the range:

num = 5
rng = [num to num * 2]

The to/til operators bind less tightly than arithmetic operators, so parentheses are optional.


This is where RapydScript really starts to shine. JavaScript is known for having really crappy class implementation (it's basically a hack on top of a normal function, most experienced users suggest using external libraries for creating those instead of creating them in pure JavaScript). Luckily RapydScript fixes that. Let's imagine we want a special text field that takes in a user color string and changes color based on it. Let's create such field via a class.

class ColorfulTextField:
    def __init__(self):
        field = $('<input></input>')
        changeColor = def(event):
            field.css('backround', field.val())
        self.widget = field

This class abuses DOM's tolerant behavior, where it will default to the original setting when the passed-in color is invalid (saving us the extra error-checking logic). To append this field to our page we can run the following code:

textfield = ColorfulTextField()

If you're used to JavaScript, the code above probably set off a few red flags in your head. In pure JavaScript, you can't create an object without using a 'new' operator. Don't worry, the above code will compile to the following:

var textfield;
textfield = new ColorfulTextField()

RapydScript will automatically handle appending the 'new' keyword for you, assuming you used 'class' to create the class for your object. This also holds when creating an object inside a list or returning it as well. You could easily do the following, for example:

fields = [ColorfulTextField(), ColorfulTextField(), ColorfulTextField()]

This is very useful for avoiding a common JavaScript error of creating 'undefined' objects by forgetting this keyword. One other point to note here is that regular DOM/JavaScript objects are also covered by this. So if you want to create a DOM image element, you should not use the 'new' keyword either:

myImage = Image()

But RapydScript's capability doesn't end here. Like Python, RapydScript allows inheritance. Let's say, for example, we want a new field, which works similar to the one above. But in addition to changing color of the field, it allows us to change the color of a different item, with ID of 'target' after we press the 'apply' button, located right next to it. Not a problem, let's implement this guy:

class TextFieldAffectingOthers(ColorfulTextField):
    def __init__(self):
        field = self.widget
        submit = $('<button type="button">apply</button>')
        applyColor = def(event):
            $('#target').css('background', field.val())
        self.widget = $('<div></div>')\

A couple things to note here. We can invoke methods from the parent class the same way we would in Python, by using Parent.method(self, ...) syntax. This allows us to control when and how (assuming it requires additional arguments) the parent method gets executed. Also note the use of \ operator to break up a line. This is something Python allows for keeping each line short and legible. Likewise, RapydScript, being indentation-based, allows the same.

An important distinction between Python and RapydScript is that RapydScript does not allow multiple inheritance. This is intentional, both for performance reasons and because multiple inheritance creates ambiguity. When using multiple inheritance, we usually only care about a few methods from the alternative classes (also see mixin section below). Leveraging JavaScript prototypical inheritance, RapydScript allows us to reuse methods from another class without even inheriting from it:

class Something(Parent):
    def method(self, var):
        Parent.method(self, var)
        SomethingElse.method(self, var)

Notice that Something class has no __init__ method. Like in Python, this method is optional for classes. If you omit it, an empty constructor will automatically get created for you by RapydScript (or when inheriting, the parent's constructor will be used). Also notice that we never inherited from SomethingElse class, yet we can invoke its methods. This brings us to the next point, the only real advantage of inheriting from another class (which you can't gain by calling the other classes method as shown above) is that the omitted methods are automatically copied from the parent. Admittedly, we might also care about isinstance() method, to have it work with the non-main parent, which is equivalent to JavaScript's instanceof operator.

To reuse portions of other classes similar to how one would expect via multiple inheritance, RapydScript also allows mixins. To declare class with a few mixins simply use a decorator:

@mixin(Predator, Maneuverable)
class Shark(Fish):
    def __init__(self):
        self.bites = True
        self.damage = 999

There is also a convenience function in RapydScript called merge that lets you assign all methods of a given class to another, it gives you a bit more control than a mixin at the expense of prettier syntax:

merge(Snake, Animal, false)     # add Animal's methods to Snake, don't overwrite ones already declared in Snake
merge(Snake, Animal, true)      # add Animal's methods to Snake, overwriting ones already declared in Snake

To summarize classes, assume they work the same way as in Python, plus a few bonus cases. The following, for example, are equivalent:

class Aclass:
    def __init__(self):

    def method(self):

class Aclass:
    def __init__(self):
        self.method = def():

The variable self in the above example is not a keyword. Like in Python, it can be replaced with any other variable name. Also, like in Python, this variable will be tied to the class, unlike this keyword of JavaScript. RapydScript still treats this keyword the same way JavaScript does:

class Main:
    def __init__(s):
        main = this
        method = def():

    def doSomething(s):

Or, leveraging Pythonic binding to first argument, the same can be shortened to:

class Main:
    def __init__(s):
        method = def():

    def doSomething(s):

Like Python, RapydScript allows static methods. Marking the method static with @staticmethod decorator will compile that method such that it's not bound to the object instance, and ensure all calls to this method compile into static method calls:

class Test:
    def normalMethod(self):
        return 1

    def staticMethod(a):
        return a+1

Some methods in the native JavaScript classes, such as String.fromCharCode() have also been marked as static to make things easier for the developer.

External Classes

RapydScript will automatically detect classes declared within the same scope (as long as the declaration occurs before use), as well as classes properly imported into the module (each module making use of a certain class should explicitly import the module containing that class). RapydScript will also properly detect native JavaScript classes (String, Array, Date, etc.). Unfortunately, RapydScript has no way of detecting classes from third-party libraries. In those cases, you could use the new keyword every time you create an object from such class. Alternatively, you could mark the class as external.

Marking a class as external is done via external decorator. You do not need to fill in the contents of the class, a simple pass statement will do:

class Alpha:

RapydScript will now treat Alpha as if it was declared within the same scope, auto-prepending the new keyword when needed and using prototype to access its methods (see casperjs example in next section to see how this can be used in practice). You don't need to pre-declare the methods of this class (unless you decide to for personal reference, the compiler will simply ignore them) unless you want to mark certain methods as static:

class Alpha:
    def one():

Alpha.one is now a static method, every other method invoked on Alpha will still be treated as a regular class method. While not mandatory, you could pre-declare other methods you plan to use from Alpha class as well, to make your code easier to read for other developers, in which case this external declaration would also serve as a table of contents for Alpha:

class Alpha:
    def two(): pass
    def three(): pass

    def one(): pass

As mentioned earlier, this is simply for making your code easier to read. The compiler itself will ignore all method declarations except ones marked with staticmethod decorator.

You could also use external decorator to bypass improperly imported RapydScript modules. However, if you actually have control of these modules, the better solution would be to fix those imports.

Method Binding

By default, RapydScript does not bind methods to the classes they're declared under. This behavior is unlike Python, but very much like the rest of JavaScript. For example, consider this code:

class Boy:
    def __init__(self, name):
        self.name = name

    def greet(self):
        print('My name is' + self.name)

tod = Boy('Tod')
tod.greet()                 # Hello, my name is Tod
getattr(tod, 'greet')()     # Hello, my name is undefined

In some cases, however, you may wish for the functions to remain bound to the object they were retrieved from. For those cases, RapydScript has bind function. Unlike regular JavaScript Function.prototype.bind, RapydScript's bind can rebind methods that have already been bound. The binding we wanted to see in the above example can be achieved as follows:

bound = bind(getattr(tod, 'greet'), tod)
bound()                     # Hello, my name is Tod

To unbind a bound method, you can call pass false as a second argument instead of an object you wish to bind to. You can also auto-bind all methods of the class by calling rebind_all:

class Boy:
    def __init__(self, name):
        self.name = name

    def greet(self):
        print('My name is' + self.name)

tod = Boy('Tod')
tod.greet()                 # Hello, my name is Tod
getattr(tod, 'greet')()     # Hello, my name is Tod

Likewise, rebind_all(self, false) will unbind all methods. It's not recommended to auto-bind classes that inherit from 3rd party libraries. For example, casperjs has Casper class, which RapydScript can easily inherit and extend:

class Casper:

class Scraper(Casper):
    def __init__(self):

s = Scraper()
    def(): this.echo(this.getTitle())

Including rebind_all call in the constructor, however, will break Casper. It is for that reason that rebind_all isn't added to the constructor by default by RapydScript. You could, however use --auto-bind compile flag to have RapydScript rebind automatically for you. There is a bit more that this flag does behind the scenes, which ensures that class binding behaves identical to Python, at the expense of some performance and compatibility with libraries like casperjs.


RapydScript's module system works almost exactly like Python's. Modules are files ending with the suffix .pyj and packages are directories containing an __init__.pyj file. The only caveat is that star imports are not currently supported (this is by design, star imports are easily abused).

For those unfamiliar with Python, here are a few examples that will cover all the user cases you will encounter:

import foo
from baz import anotherMethod, andAnotherMethod
import bar as qux


RapydScript first tries to search the current directory of the file for the import. If that fails, it tries to search user-defined imports (which can be set via RAPYDSCRIPT_PATH environment variable as colon-separated directories, or semi-colon separated on Windows), and finally in the RapydScript builtin directory. This search order allows the user to easily override inbuilt functions.

Exception Handling

Like Python and JavaScript, RapydScript has exception handling logic. The following, for example, will warn the user if variable foo is not defined:

    print("Foo wasn't declared yet")

It's a good practice, however, to only catch exceptions we expect. Imagine, for example, if foo was defined, but as a circular structure (with one of its attributes referencing itself):

foo = {}
foo.bar = foo

We would still trigger an exception, but for a completely different reason. A better way to rewrite our try/except block would be:

except ReferenceError:
    print("Foo wasn't declared yet")

We could also handle circular structure exception, if we needed to:

except ReferenceError:
    print("Foo wasn't declared yet")
except TypeError:
    print("One of foo's attributes references foo")

Or we could just dump the error back to the user:

except as err:
    print(err.name + ':' + err.message)

In this example, err is a JavaScript error object, it has name and message attributes, more information can be found at https://developer.mozilla.org/en-US/docs/JavaScript/Reference/Global_Objects/Error. You can inherit from this object as if it was a class to create custom errors, just like you would in Python:

class MyError(Error):
    def __init__(self, message):
        self.name = Error
        self.message = message

raise MyError('This is a custom error!')

You can lump multiple errors in the same except block as well:

except ReferenceError, TypeError as e:
    print(e.name + ':' + e.message)

Basically, try/except/finally in RapydScript works very similar to the way it does in Python 3, lacking only the else directive (it didn't seem useful enough to implement). Like in Python and JavaScript, you can nest multiple exceptions inside each other, and use raise to throw the error back at the user:

except ReferenceError as e:
    if e.message == 'foo is not defined':
        raise e
    # reset foo
    foo = 'bar'

Like in Python (but unlike in regular JavaScript), you can use raise keyword by itself from within except block to reraise last-caught error.

Scope Control

Scope refers to the context of a variable. For example, a variable declared inside a function is not seen by the code outside the function. This variable's scope is local to the function. JavaScript controls scope via var keyword. Any variable that you start using without declaring with a var first will try to reference inner-most variable with the same name, if one doesn't exist, it will create a global variable with that name. For example, the following JavaScript code will not only return a incremented by 1, but also overwrite the global variable a with a+1:

a = 1;
a_plus_1 = function() {
    return ++a;

Basically, JavaScript defaults to outer or global scope if you omit var. This behavior can introduce some very frustrating bugs in large applications. To avoid this problem, RapydScript's scope preference works in reverse (same as Python's). RapydScript will prefer local-most scope, always creating a local variable if you perform any sort of assignment on it in a function (this is called variable shadowing). Shadowing can create another annoyance, however, of function's variable changes getting discarded. For example, at first, it looks like the following code will set a to 2:

a = 1
b = 1
increment = def():
    a += b

When executed, however, increment() function will discard any changes to a. This is because, like Python, RapydScript will not allow you to edit variables declared in outer scope. As soon as you use any sort of assignment with a in the inner scope, RapydScript will declare it as an internal variable, shadowing a in the outer scope. One way around this is to use the global keyword, declaring a as a global variable. This, however, must be done in every function that edits a. It also litters global scope, which it frowned upon because it can accidentally overwrite an unrelated variable with the same name (declared by someone else or another library). RapydScript solves this by introducing nonlocal keyword (just like Python 3):

a = 1
b = 1
increment = def():
    nonlocal a
    a += b

Note that b is not affected by shadowing. It's the assignment operator that triggers shadowing, you can read outer-scope variables without having to use nonlocal. You can combine multiple non-local arguments by separating them with a comma: nonlocal a, b, c. You can also chain nonlocal declarations to escape multiple scopes:

def fun1():
    a = 5
    b = fun2():
        nonlocal a
        a *= 2
        c = fun3():
            nonlocal a
            a += 1

Shadowing is preferred in most cases, since it can't accidently damage outside logic, and if you want to edit an external variable, you're usually better off assigning function's return value to it. There are cases, however, when using nonlocal makes the code cleaner. There is also global, but it is rarely a good solution, and use of nonlocal is preferred to it.


Like Python, RapydScript allows you to import additional modules into your code.

For those unfamiliar with importing, let's imagine we're writing a very large program. This program is several thousand lines of code. We could dump it all into the same file, but that wouldn't be too clean (especially when we have multiple developers working on it). Alternatively, we could separate different chunks of the program into different files. Let's imagine, for example, that we're writing a videogame. We've already written a module that implements 'BasicCharacter' class, used by NPCs, monsters, and the main character. We've saved this class to Basic.pyj (that's the extension RapydScript prefers). Now let's create the main character in a different module:

from Basic import BasicCharacter

class MainCharacter(BasicCharacter):
    This is the main character class, similar to basic but it implements attack and hp
    def __init__(self):
        self.hp = 100
        self.damage = 10

    def attack(self, something):

    def getHurt(self, damage):
        self.hp -= damage

With RapydScript's importing, you don't need to worry about including each JavaScript file individually in your html. All your .pyj files will get concatenated into a single .js file that you can then include in your page. The name of the .js file will match the name of the file you used input for the RapydScript compiler.

RapydScript already adds several modules you can import by default, like stdlib and yql. Check out the examples or the documentation in the modules themselves for usage.

Regular Expressions

RapydScript supports Pythonic and JavaScript way of declaring regular expressions. Additionally, like CoffeeScript, RapydScript supports verbose mode (which works like Perl's /x modifier or Python's VERBOSE modifier) - flags, if any, are at the end:

# regular
op = /^(?:[-=]>|[-+*/%<>&|^!?=]=|>>>=?|([-+:])\1|([&|<>])\2=?|\?\.|\.{2,3})/

# verbose
op = /// ^ (
    ?: [-=]>             # function
     | [-+*/%<>&|^!?=]=  # compound assign / compare
     | >>>=?             # zero-fill right shift
     | ([-+:])\1         # doubles
     | ([&|<>])\2=?      # logic / shift
     | \?\.              # soak access
     | \.{2,3}           # range or splat
) ///

And the Pythonic alternative (note that importing re module also brings in other logic into your code, you'll end up with more functionality at the expense of performance - unless you need that extra functionality, I recommend JavaScript version shown above):

# pythonic
import re
op = re.compile('^(?:[-=]>|[-+*/%<>&|^!?=]=|>>>=?|([-+:])\1|([&|<>])\2=?|\?\.|\.{2,3})')

# verbose pythonic (re.VERBOSE is not implemented, since VERBOSE conversion in current engine happens at
# compile time, for Pythonic re.VERBOSE syntax we'd need to duplicate this logic in re module to run at
# runtime instead)
op = re.compile(/// ^ (
    ?: [-=]>             # function
     | [-+*/%<>&|^!?=]=  # compound assign / compare
     | >>>=?             # zero-fill right shift
     | ([-+:])\1         # doubles
     | ([&|<>])\2=?      # logic / shift
     | \?\.              # soak access
     | \.{2,3}           # range or splat
) ///)

As in JavaScript, you can also use the RegExp class:

regex = /ab+c/i
regex = RegExp('ab+c', 'i')
regex = RegExp(/ab+c/, 'i')

ES6 Features

Most of the features in the compiler work fine with any version of JavaScript. Some features, however, are only available when you compile with -es6 flag. Implementing them in older versions of JavaScript is simply not the best use of my time with libraries like Babel.js doing a good job with it. Here are some features that work differently in ES6 mode:

  • Function generators (only supported with ES6 flags)
  • Computed dictionary/object literal keys ((foo**2): val, only supported with ES6 flag)
  • Spread operators in arrays (['one', *splat, 'ten'], only supported with ES6 flag)
  • Spread operators in object literals ({foo: "bar", *splat, baz: "qux"}, only supported with ES6 flag)
  • Setters and getters for classes (only supported with ES6 flag)
  • Template literals

The remaining features are all supported with regular mode but take advantage of ES6 features when compiled with -es6 flag:

  • Cleaner optional function arguments
  • Cleaner array to variable unpacking
  • Cleaner class declaration logic
  • Starargs produce cleaner compilation

Out of the new features mentioned above, function generator syntax is the same as in Python:

```python def firstN(n): num = 0 while num < n: yield num num += 1

print(sum(firstN(1e6))) ```

The syntax for computed keys is slightly different from Python, since JavaScript already uses the unquoted key syntax as alias for quoted keys and RapydScript has been doing the same for a while. But since the array-like syntax of ES6 is ugly/confusing, RapydScript follows the convention it has itself set for a while of using parentheses (this convention is consistent with self-executing function convention and using class as a function convention):

computed = "qux"
key = "Quux"
hash = {
    regularKey: 1
    'regular-key': "foo",
    (computed + key): "baz"

Spread operators are not supported in Python but are part of ES6 syntax. These allow you to splat one array/hash inside another. For now RapydScript uses the *args convention set by Python for function calls (since this operation is similar in spirit), although the ...args operator JavaScript uses seems more intuitive to me, so there is a good chance it may change (or at least support both). Example use cases:

a = [5 to 20]
b = [100 til 500]
c = [1, 2, *a, 27, 28, *b, 999]

props = {
    foo: "bar",
    baz: 2
moreProps = {
    qux: "quux"

Setters and getters for classes are implemented in a way that's more similar to JavaScript than Python, mainly because their syntax is so horrible in Python. As an added bonus, they actually feel more consistent with RapydScript's def keyword than they do in JavaScript itself:

```python class Item: _item = None

def unit(self):

get item(self):
    return self._item

set item(self, val):
    self._item = val


Template literals are new type of string in ES6 that auto-interpolates variables for you. Just like Python 3.6, RapydScript uses the f prefix for these:

python name = 'Simon' greeting = f"Hello, my name is ${name}."

This will auto-fill name into the greeting, after which it will function as a regular string. This will eventually be the preferred approach over concatenating strings, and integer/string concatenation will probably be marked as an error. It is, however, possible that the ${} syntax, which was taken from ES6, may change (Python's own syntax is {} by themselves).

Available Libraries

One of Python's main strengths is the number of libraries available to the developer. This is something very few other Python-in-a-browser frameworks understand. In the browser JavaScript is king, and no matter how many libraries the community for the given project will write, the readily-available JavaScript libraries will always outnumber them. This is why RapydScript was designed with JavaScript and DOM integration in mind from the beginning. Indeed, plugging underscore.js in place of RapydScript's stdlib will work just as well, and some developers may choose to do so, after all, underscore.js is very Pythonic and very complete. Likewise, sprintf.js (https://npmjs.org/package/sprintf-js) can be used with RapydScript to replicate Python's string interpolation.

It is for that reason that I try to keep RapydScript bells and whistles to a minimum. RapydScript's main strength is easy integration with JavaScript and DOM, which allows me to stay sane and not rewrite my own versions of the libraries that are already available. That doesn't mean, however, that pythonic libraries can't be written for RapydScript. To prove that, I have implemented lightweight clones of several popular Python libraries and bundled them into RapydScript, you can find them in src directory. The following libraries are included:

stdlib/stdlib2      # see stdlib section
math                # replicates almost all of the functionality from Python's math library
re                  # replicates almost all of the functionality from Python's re library
unittest            # replicates almost all of the functionality from Python's unittest library
random              # replicates most of the functionality from Python's random library
yql                 # lightweight library for performing Yahoo Query Language requests

For the most part, the logic implemented in these libraries functions identically to the Python versions. One notable exception is that unittest library requires that classes be bound to the global (nodejs) or window (browser) object to be picked up by unittest.main(). An example in unitetest.pyj shows this usage. I'd be happy to include more libraries, if other members of the community want to implement them (it's fun to do, re.pyj is a good example), but I want to reemphasize that unlike most other Python-to-JavaScript compilers, RapydScript doesn't need them to be complete since there are already tons of available JavaScript libraries that it can use natively.

Advanced Usage Topics

This section contains various topics which might be of interest to the programmer writing large projects using RapydScript, but might not be relevant to a programmer who is just getting started with RapydScript. The topics in this section focus on coding conventions to keep your code clean, optimizations, and additional libraries that come with RapydScript, as well as suggestions for writing your own libraries.

System Scripts

I typically have a language I strongly prefer over others for writing miscellaneous system scripts, things like moving files around or automating certain tasks on my home (or work) machine. Bash tends to work well for simple tasks without too much conditional logic. For other things I used to prefer Python. Today, I use RapydScript instead. JavaScript has a powerful ecosystem, and it would be a shame to let it go to waste. You can have your RapydScript files run natively on your OS as well. Do do so, you can include the following line at the top of your file and chmod a+x it:

#!/usr/bin/env rapydscript -x

This is identical to the following terminal operation:

rapydscript --execute [filename]

It will trigger the script, omitting the compiled code. You can include the --pretty option to include output of the compiled code as well.

Unit Testing

You're free to use any unit testing library you wish, but RapydScript comes with a dev library out of the box. It's a convenient way to test your functions without having to build out a test harness yourself. To use it, simply import dev, set UTEST to true on global root object and use dev.utest decorator:

```python import dev

root.UTEST = True

@dev.utest({input: [1, 2], output: 3}) @dev.utest({input: [1, 2, 3], output: 3}) @dev.utest({input: [Infinity, 2], output: Infinity}) def foo(a, b): return a + b

@dev.utest({input: [], error: /qux/}) def bar(): raise Error('qux') `` The function is tested at load time, not when it's executed, which means these test decorators can stay in your code without affecting execution after. You may disable the test, however, by unsetting theUTESTvariable. Aside from running at initialization, you won't even experience function-call overhead with these decorators since they pass the original function through. However, if you want to remove them altogether in your production environment, you can usedrop_decoratorsanddrop_imports` flags:

bash rapydscript file.pyj --dd dev.utest --di dev

Browser Compatibility

By default, RapydScript compiles your logic such that it will work on modern browsers running HTML5. Previously I generated code that was compatible with older versions of IE, but have since decided that it wasn't worth it. It prevented me from making use of sensible JavaScript features many developers take for granted (setters, getters, strict mode, etc.), forced special cases on me, and required overly verbose JavaScript with unnecessary polyfill. RapydScript no longer supports versions of IE before 9, but you can easily bring that support back into RapydScript with the help of a tool like Modernizr or Babel.

Easter Eggs

While working with RapydScript, and interesting idea to try came to me:

```python array = [i**2 for i in [1 to 100]] # just an array, initialized whichever way you want

@array.forEach def f(e, i): print('lolz ' + e) ``` The above logic will do exactly what it looks like. It will auto-trigger on function creation, and repeat the action for each element in the array. This wasn't intentional addition to the language, more of a side-effect of how decorators are implemented, but I can assure you it's not going anywhere in case you want to make use of it for something of value.

Code Conventions

It's not hard to see that RapydScript is a cleaner language than JavaScript. However, like with all dynamically-typed languages (including Python), it's still easy to shoot yourself in the foot if you don't follow some sort of code conventions. Needless to say, they're called conventions for a reason, feel free to ignore them if you already have a set of conventions you follow or if you disagree with some.

Tabs vs Spaces

This seems to be a very old debate. Python code conventions suggest 4-space indent, most of the bundled RapydScript files use 1-tab for indentation. The old version of RapydScript relied on tabs, new one uses spaces since that seems to be more consistent in both Python and JavaScript communities. Use whichever one you prefer, as long as you stay consistent. If you intend to submit your code to RapydScript, it must use spaces to be consistent with the rest of the code int he repository.

Object Literals vs Hashes/Dicts

JavaScript treats object literals and hashes as the same thing. I'm not a fan of this policy. Some of the problems you can see resulting from this is Google Closure compiler's ADVANACED_OPTIMIZATIONS breaking a lot of seemingly-good JavaScript code. The main problem for most of the code that breaks seems to be renaming of methods/variables in one place and not another. My suggestion is to ALWAYS treat object literals as object literals and ALWAYS treat hashes as hashes, basically be consistent about quoting your keys. As an added bonus, your code will have a much better chance of compiling correctly via Closure compiler. For example:

obj = {
    foo:    1,
    bar:    def(): print('bar' + str(foo))
hash = {
    'foo':  1,
    'bar':  def(): print('bar' + str(foo))

obj.bar()       # good
obj['bar']()    # bad

hash.bar()      # bad
hash['bar']()   # good


Don't abuse semi-colons. They're meant as a way to group related logic together, not to fit your entire web-app on one line. The following is fine:

X = 0; Y = 1

Anything that requires more than a couple semi-colons, however, or involves long mathematical computations, is better off on its own line. Use your discretion, if the logic requires more than one visual pass-through from the programmer to understand the flow, you probably shouldn't use semi-colons. A Fibanacci function, as shown below, would probably be the upper limit of the kind of logic you could sanely represent with semi-colons:

fib = def(x): if x<=1: return 1; return fib(x-1)+fib(x-2)

Even for this example, however, I'd personally prefer to use multiple lines.

Raw JavaScript

Occasionally you may need to use raw JavaScript due to a limitation of RapydScript. To keep code legible and consistent, I suggest minimizing the chunk of code contained within JS, keeping most of the logic within RapydScript itself. For example, instead of accessing a variable that shares reserved keyword as JS('module.exports') use JS('module').exports. This will make it more visually-obvious what you're trying to do as well as allow your syntax highlighter to do its job.

jQuery-wrapped Elements

If you use jQuery with your app, you will probably be storing these into variables a lot. If you've written a decently sized app, you've probably mistaken a bare element with wrapped element at least once. This is especially true of objects like canvas, where you need to access object's attributes and methods directly. My solution for these is simple, prepend jQuery-wrapped elements with $:

$canvas = ('<canvas></canvas>')
canvas = $canvas.get(0)
ctx = canvas.getContext('2d')

This is especially useful with function definitions, since you will immediately know what kind of object the function takes in just by skimming its signature.


I recommend that developers rely on native JavaScript logic when possible, rather than libraries like math.pyj and re.pyj. While they mimic Python without problems and work quite well, they introduce extra overhead that your web app doesn't need. Additionally, I think re module in Python is unnecessarily complex, and JavaScript's RegExp object is much easier to use. With that said, these libraries can be extremely useful for porting large applications from Python to the web, but if you're writing new code, it will probably be easier to maintain if you decide to use native JavaScript alternatives (such as Math and RegExp) instead.

External Libraries and Classes

RapydScript will pick up any classes you declare yourself as well as native JavaScript classes. It will not, however, pick up class-like objects created by outside frameworks. There are two approaches for dealing with those. One is via @external decorator, the other is via new operator when declaring such object. To keep code legible and consistent, I strongly prefer the use of @external decorator over the new operator for several reasons, even if it may be more verbose:

  • @external decorator makes classes declared externally obvious to anyone looking at your code
  • class declaration that uses @external decorator can be exported into a reusable module
  • developers are much more likely to forget a single instance of new operator when declaring an object than to forget an import, the errors due to omitted new keyword are also likely to be more subtle and devious to debug


Performance emphasis is at the core of RapydScript. I don't want to feel like I'm a second-class citizen compared to native JavaScript developers and neither should other RapydScript users. For that reason RapydScript rejected many Pythonic features that I myself would like to see in favor of performance. Some of these include operator overloading and better error catching. However, when performance hit is negligible and allows additional productivity to the developer, that's the kind of win-win RapydScript attempts to capture. For that reason, the compiler comes bundled with a mini benchmark suite. You can invoke it by using --bench command. Since it takes a while to run performance tests, I recommend you select a particular file (in test/perf/) to bench. A few of these runs will give you a pretty good idea what kind of performance sacrifices you're making with different functions/operators. Feel free to throw your own tests into the same directory.


In a perfect world, software works flawlessly and doesn't have any special cases requiring workarounds on the user's part. In a less-perfect world, all quirks are due to the software itself and can be fixed. In our world, we not only have to deal with the quirks of the software we're using, but other software it interacts with. RapydScript is no exception, in addition to its own quirks there are a few quirks brought upon us by the browser as well as a few bugs in jQuery that affect us. Here is a list of things you need to be aware of:

  • RapydScript automatically appends 'new' keyword when using classes generated by it, native JavaScript objects like Image and RegExp and classes from other libraries marked as external. This means that you should be using @external decorator from classes you're bringing in from the outside that the compiler isn't aware of, or use new when declaring them manually.

  • Automatic new insertion depends on the compiler being able to detect that a symbol resolves to a class. Because of the dynamic nature of JavaScript this is not possible to do with 100% accuracy (this is usually only a problem when you rely on duck-typing to conceal a class constructor in a variable). In those cases you should append the new keyword yourself. Similarly, the compiler will try to convert SomeClass.method() into SomeClass.prototype.method() for you, but will fail in some cases. Declaring this variable as a class with @external decorator should fix this issue.

  • Truthiness in JavaScript is very different from Python. Empty lists and dicts are False in Python but True in JavaScript. The compiler could work around that, but not without a significant performance cost, so it is best to just get used to checking the length instead of the object directly.

  • Operators in JavaScript are very different from Python. 1 + '1' would be an error in Python, but results in '11' in JavaScript. Keep that in mind as you write code. When possible, RapydScript tries to detect type and warn you ahead of time of nonsensical operations such as [] + {}, but sometimes it can't do its magic.

  • Method binding in RS is not automatic. So someobj.somemethod() will do the right thing, but x = someobj.somethod; x() will not. RS could work around it, but at significant performance cost. See the section above on method binding for details. To work around it, you could assign it instead as x = bind(someobj.somemethod, someobj). Alternatively, you can use --auto-bind flag, and enjoy automatic method binding at the expense of performance.

  • jQuery erroneously assumes that no other library will be modifying JavaScript's 'Object', and fails to do object.hasOwnProperty() check in multiple places where it should. To avoid breaking it, I had to implement stdlib such that dictionary methods are methods of a different object. Regular Python allows you to call hash.keys() as well as dict.keys(hash). RapydScript only supports the second notation - which is admittedly a bit more awkward.

  • Negative indexes are only supported for constant numbers. If the compiler detects that you're using a negative index (array[-1]), it will automatically convert it to array[array.length-n]. Otherwise, if the index is masked in a variable, you will have to convert it yourself.

  • Operator overloading is not supported, yet

Rapydscript Examples

Asteroids Example

Copyright 2013 Pyjeon Software LLC
Author: Charles Law (based on Pyjamas Asteroids)
License: Creative Commons: Attribution + Noncommerial + ShareAlike


RapydScript Asteroids

The asteroids game was originally written as a Pyjs (formally Pyjamas) example. With some relatively minor changes, the code was ported over to RapydScript. The RapydScript verion has the full functionality of the game but with a significantly smaller footprint. The important metrics from our perspective are:

Download Size: 18 kb   vs 1200 kb
Download Time: 1.3 sec vs  17 sec
LOC:             700   vs   32900

It is also worth noting that there are currently 6 cache files generated for Pyjs so the app takes significantly more storage space.

Example in action: RapydScript Asteroids

Pyjs Asteroids

The Pyjs Asteroids game builds to approximately 1200kb for a single cache file. End users will have to wait significantly longer to load a webpage.

For developers, Pyjs can make applications appear slow and use more bandwidth to transfer. It is also worth noting that Pyjs generates 6 cache files that all must be deployed.

Example in Action: Pyjs Example



The view contains the canvas and has all the logic for drawing the game state
on the screen. The original Pyjs code used multiple libraries here. Of the 3
files, this one required the most changes to port to RapydScript.

import stdlib
import game_model
import game_controller

SHOT_COLOR = '#fff'


class View:
    def __init__(self, w, h, canvas):
        self.width = w
        self.height = h
        self.canvas = canvas

        self.model = Model(w, h)
        self.controller = Controller(self.model)

        self.number_images_loaded = 0
        self.view_loaded = False

        document.addEventListener('keydown', def (event):
            self.setKey(event.keyCode, True)
        document.addEventListener('keyup', def (event):
            self.setKey(event.keyCode, False)

        #Load images
        imgsrcs = ['./images/ship1.png',

        self.img = [None, None, None]
        for i in range(3):
            self.img[i] = Image()
            self.img[i].src = imgsrcs[i]
            self.img[i].onload = def():
                self.number_images_loaded += 1
                if self.number_images_loaded >= 3:
                    self.view_loaded = True

    def setKey(self, k, set):
        if k == 38:
            self.controller.key_up = set
        elif k == 40:
            self.controller.key_down = set
        elif k == 37:
            self.controller.key_left = set
        elif k == 39:
            self.controller.key_right = set
        elif k == 32:
            self.controller.key_fire = set

    def draw_asteroid(self, ctx, asteroid):
        ctx.translate(asteroid.x, asteroid.y)
        ctx.drawImage(self.img[2], -(ASTEROID_SIZE/2), -(ASTEROID_SIZE/2))

    def draw_shot(self, ctx, shot):
        ctx.fillStyle = SHOT_COLOR
        ctx.fillRect(Math.ceil(shot.x - 1), Math.ceil(shot.y - 1), 3, 3)

    def draw_ship(self, ctx, ship):
        ctx.translate(ship.x, ship.y)
        if self.controller.key_up:
            img = self.img[1]
            img = self.img[0]
        ctx.drawImage(img, -15, -12)

    def draw(self):
        ctx = self.canvas.getContext('2d')

        #fill the background
        ctx.fillStyle = '#000'
        ctx.fillRect(0, 0, self.width, self.height)

        for i in range(len(self.model.asteroids)):
            self.draw_asteroid(ctx, self.model.asteroids[i])
        for i in range(len(self.model.shots)):
            self.draw_shot(ctx, self.model.shots[i])

        self.draw_ship(ctx, self.model.ship)

window.runGame = def():
    # The view tells when the controller to start and passes in a reference
    # to itself
    canvas = document.getElementById('myCanvas')
    view = View(CANVAS_DIM_X, CANVAS_DIM_Y, canvas)


#   Copyright 2012 Charles Law
#   Licensed under the Apache License, Version 2.0 (the "License");
#   you may not use this file except in compliance with the License.
#   You may obtain a copy of the License at
#    http://www.apache.org/licenses/LICENSE-2.0
#   Unless required by applicable law or agreed to in writing, software
#   distributed under the License is distributed on an "AS IS" BASIS,
#   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
#   See the License for the specific language governing permissions and
#   limitations under the License.


The controller runs the game and modifies the Model. This required very few changes
to port from Pyjs.

import game_model

class Controller:
    def __init__(self, model):
        self.model = model
        self.key_up = False
        self.key_down = False
        self.key_left = False
        self.key_right = False
        self.key_fire = False

    def start_game(self, view):
        self.view = view

        #setup a timer
        , 1000/FPS)

    def update(self):

    def keyboard_updates(self):
        ship = self.model.ship

        drot = 0
        if self.key_left:
            drot -= ROTATE_SPEED
        if self.key_right:
            drot += ROTATE_SPEED
        if drot:

        if self.key_up:

        if self.key_fire:


#   Copyright 2009 Joe Rumsey
#   Copyright 2012 Charles Law
#   Licensed under the Apache License, Version 2.0 (the "License");
#   you may not use this file except in compliance with the License.
#   You may obtain a copy of the License at
#    http://www.apache.org/licenses/LICENSE-2.0
#   Unless required by applicable law or agreed to in writing, software
#   distributed under the License is distributed on an "AS IS" BASIS,
#   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
#   See the License for the specific language governing permissions and
#   limitations under the License.


The model contains the classes representing the various game objects - the asteroid, ship
and shots. These are used to store all the state information on these different objects
as well as helping modify them. The original Pyjs code did not use any libraries so
porting over this code required almost no changes to the code.

FPS = 30
THRUST = 0.2
ASTEROID_SIZES = [90.0, 45.0, 22.0, 11.0]

def randfloat(min, max):
    return Math.random() * (max - min) + min

def randint(min, max):
    return Math.floor(Math.random() * (max - min + 1)) + min

def distsq(x1,y1, x2,y2):
    return ((x1 - x2) * (x1 - x2)) + ((y1 - y2) * (y1 - y2))

class Asteroid:
    def __init__(self, model, x=None, y=None, size=0):
        self.model = model
        if x is None or y is None:
            self.x = model.x/2
            self.y = model.y/2
            while distsq(self.x, self.y, model.x / 2, model.y / 2) < (180*180):
                self.x = randint(0, model.x)
                self.y = randint(0, model.y)
            self.x = x
            self.y = y

        self.dx = randfloat(0, MAX_ASTEROID_SPEED)
        self.dy = randfloat(0, MAX_ASTEROID_SPEED)
        self.rot = randfloat(0, 2*Math.PI) - Math.PI
        self.rotspeed = randfloat(0, 0.1) - 0.05
        self.size = size
        self.radius = ASTEROID_SIZES[self.size]
        self.radius2 = self.radius*self.radius
        self.scale = (self.radius / ASTEROID_SIZE) * 2

    def update_dim(self, pos, d_dim, max_dim):
        pos += d_dim

        if d_dim < 0:
            if pos <= 0:
                pos = -pos
                d_dim = -d_dim
            if pos >= max_dim:
                d_dim = -d_dim
                pos = 2*max_dim - pos
        return pos, d_dim

    def move(self):
        self.x, self.dx = self.update_dim(self.x, self.dx, self.model.x)
        self.y, self.dy = self.update_dim(self.y, self.dy, self.model.y)

        self.rot += self.rotspeed

class Shot:
    def __init__(self, model, ship):
        self.model = model
        self.x = ship.x
        self.y = ship.y
        self.dx = ship.dx
        self.dy = ship.dy
        self.direction = ship.rot
        self.lifespan = SHOT_LIFESPAN

    def move(self):
        self.lifespan -= 1
        if self.lifespan <= 0:
            return False
        self.x = self.x + self.dx + SHOT_SPEED * Math.sin(self.direction)
        self.y = self.y + self.dy - SHOT_SPEED * Math.cos(self.direction)
        for i, a in enumerate(self.model.asteroids):
            if distsq(self.x, self.y, a.x, a.y) < (a.radius2):
                return False

        return True

class Ship:
    def __init__(self, cx, cy):
        self.cx = cx
        self.cy = cy

    def rotate_ship(self, drot):
        self.rot += drot

        if drot < 0-Math.PI:
            drot += 2*Math.PI
        elif drot > Math.PI:
            drot -= 2*Math.PI

    def thrust(self):
        self.dx += THRUST * Math.sin(self.rot)
        self.dy -= THRUST * Math.cos(self.rot)

    def friction(self):
        if Math.abs(self.dx) < 0.001 and Math.abs(self.dy) < 0.001:
            self.dx = 0
            self.dy = 0
            direction = Math.atan2(self.dx, self.dy)
            self.dx -= FRICTION * Math.sin(direction)
            self.dy -= FRICTION * Math.cos(direction)

    def move(self):
        self.shot_delay -= 1

        self.x += self.dx
        self.y += self.dy
        if self.dx > 0 and self.x >= self.cx:
            self.x -= self.cx
        elif self.dx < 0 and self.x < 0:
            self.x += self.cx
        if self.y > 0 and self.y >= self.cy:
            self.y -= self.cy
        elif self.dy < 0 and self.y < 0:
            self.y += self.cy

    def reset(self):
        self.x = self.cx/2
        self.y = self.cy/2
        self.dx = 0
        self.dy = 0
        self.rot = 0
        self.shot_delay = 0

class Model:
    def __init__(self, x, y):
        self.x = x
        self.y = y

        self.num_asteroids = NUM_ASTEROIDS
        self.ship = Ship(x, y)

    def start_game(self, view):
        self.view = view

    def update(self):
        # Make model updates
        for a in self.asteroids:
            if (distsq(self.ship.x, self.ship.y, a.x, a.y) < (a.radius2)):

        for i in reversed(range(len(self.shots))):
            if not self.shots[i].move():

                if len(self.asteroids) == 0:



    def start_next_level(self):
        self.num_asteroids += 1

    def destroyShip(self):
        self.num_asteroids = NUM_ASTEROIDS

    def reset(self):
        self.shots = []
        self.asteroids = [Asteroid(self) for i in range(self.num_asteroids)]

    def trigger_fire(self):
        if self.ship.shot_delay > 0:
            self.shots.append(Shot(self, self.ship))
            self.ship.shot_delay = SHOT_DELAY

    def split_asteroid(self, i):
        a = self.asteroids[i]
        if a.size < len(ASTEROID_SIZES) - 1:
            for j in range(2):
                self.asteroids.append(Asteroid(self, a.x, a.y, a.size + 1))

D3 TreeMap Example

Copyright 2013 Pyjeon Software LLC
Author: Charles Law License: Creative Commons: Attribution + Noncommerial + ShareAlike


This is an example using the D3.js (Data-Driven Documents) library taken from http://bl.ocks.org/mbostock/4063582. This example was a test case of the RapydScript decompiler.


margin = {
    top: 40,
    right: 10,
    bottom: 10,
    left: 10
width = 960 - margin.left - margin.right
height = 500 - margin.top - margin.bottom

color = d3.scale.category20c()

treemap = d3.layout.treemap().size([ width, height ]).sticky(True)\
    return d.size

div = d3.select("body").append("div").style("position", "relative").\
    style("width", width + margin.left + margin.right + "px").\
    style("height", height + margin.top + margin.bottom + "px").\
    style("left", margin.left + "px").style("top", margin.top + "px")

d3.json("flare.json", def(error, root):
    node = div.datum(root).selectAll(".node").data(treemap.nodes).\
        enter().append("div").attr("class", "node").call(position).\
        style("background", def(d):
        #Note: This is actually indented 1 level it's just not obvious
        if (d.children):
            return color(d.name)
            return None
        if (d.children):
            return None
            return d.name
    d3.selectAll("input").on("change", def change():
        if (this.value is "count"):
            value = def():
                return 1
            value = def(d):
                return d.size


def position():
    this.style("left", def(d): return d.x + "px"
    ).style("top", def(d): return d.y + "px"
    ).style("width", def(d): return Math.max(0, d.dx - 1) + "px"
    ).style("height", def(d): return Math.max(0, d.dy - 1) + "px"

Paint App Example

Copyright 2012 Pyjeon Software LLC
Author: Alexander Tsepkov
License: License: Creative Commons: Attribution + Noncommerial + ShareAlike


This paint app is meant to showcase RapydScript simplicity and power. It employs multiple features of RapydScript as well as RapydML and RapydCSS, it shows DRY (don't repeat yourself) coding style when writing RapydScript as well as building the webpage itself in RapydML. It uses multiple HTML5 features to cleanly separate CSS, HTML, and JavaScript. This app also showcases RapydScript's ability to integrate with other widgets and frameworks (jQuery, jQuery UI, and Farbtastic widget). It leverages many of JavaScript's design patterns (closures/function generators) while still keeping the code clean, as one would expect in Python. It might be hard to tell from this example alone, but it is also extremely efficient. The paint-bucket algorithm, for example, when implemented in Pyjamas, takes over a second to complete (and only works in Firefox), but is instantaneous here and works in all browsers.


# Copyright 2012 Pyjeon Software LLC
# Author:   Alexander Tsepkov
# License:  License: Creative Commons: Attribution + Noncommerial + ShareAlike

# modes we will be using for drawing
BRUSH   = 0
LINE    = 2
LASSO   = 5
RECT    = 6
TEXT    = 10

# misc
X       = 0
Y       = 1
CLICK   = 1

class Drawing:
    Controls the currently displayed image, and handles undo/redo logic
    def __init__(self):

        # these are meant to be private
        self._canvas = $('#perm-dwg').get(0)        # get actual DOM element
        self._ctx = ctx = self._canvas.getContext('2d')
        self._undoStack = []
        self._redoStack = []
        self._fillColor = None
        self._strokeColor = None

        # and these are not exposed at all
        $tmpCanvas = $('#temp-dwg')
        tmpCanvas = $tmpCanvas.get(0)
        tmpCtx = tmpCanvas.getContext('2d')
        canvasWidth = tmpCanvas.width
        canvasHeight = tmpCanvas.height

        dragging = False
        points = []
        selection = None
        lastPt = [0, 0]
        transparent_bg = True

        # utility functions
        getXY = def(obj, event):
            # +0.5 allows us to snap to the center of the pixel, preventing interpolation and
            # making our images more crisp
            absolute = $(obj).offset()
            return event.pageX - absolute.left, event.pageY - absolute.top
        normalize = def(x, y, width, height):
            if width < 0:
                width = -width
                x = x-width
            if height < 0:
                height = -height
                y = y-height
            return x, y, width, height
        ellipse = def(context, x, y, width, height):
            x, y, width, height = normalize(x, y, width, height)
            ctrX = (x+x+width)/2
            ctrY = (y+y+height)/2
            circ = Math.max(width, height)
            scaleX = width/circ
            scaleY = height/circ
            context.translate(ctrX, ctrY)
            context.scale(scaleX, scaleY)
            context.arc(0, 0, circ/2, 0, 2*Math.PI)
        drawSpline = def(context, lastPoint):
            if lastPoint is undefined:
                lastPoint = points[len(points)-1]
            context.moveTo(points[0][X], points[0][Y])
            if len(points) == 1:
                context.lineTo(lastPoint[X], lastPoint[Y])
            elif len(points) == 2:
                context.quadraticCurveTo(points[1][X], points[1][Y], lastPoint[X], lastPoint[Y])
                    points[1][X], points[1][Y], \
                    points[2][X], points[2][Y], \
                    lastPoint[X], lastPoint[Y])
        sample = def(x, y, click):
            data = ctx.getImageData(x, y, 1, 1).data
            if data[3]:
                color = 'rgb('+data[0]+','+data[1]+','+data[2]+')'
                color = 'transparent'
            if click == CLICK:
                tag = '#stroke'
                tag = '#fill'

            if color == 'transparent':
                color = None
                $(tag).css('background', '')
                $(tag).css('background', color)
            if click == CLICK:
                self._brushColor = color
                self._fillColor = color
        matchStartColor = def(colorLayer, pixelPos, startPixel):
            r = colorLayer[pixelPos]
            g = colorLayer[pixelPos+1]
            b = colorLayer[pixelPos+2]
            a = colorLayer[pixelPos+3]
            return r == startPixel[0] and g == startPixel[1] and \
                    b == startPixel[2] and bool(a) == bool(startPixel[3])
        eachPixel = def(imageData, callback):
            offset = 0
            length = imageData.height * imageData.width * 4
            while offset < length:
                offset += 4
        self._clear = clear = def(context):
            context.clearRect(0, 0, canvasWidth, canvasHeight)

        # user input
        onMouseDown = def(event):
            nonlocal dragging, points, lastPt, selection

            event.preventDefault()  # prevents change to mouse cursor
            dragging = True
            x, y = getXY(this, event)
            self._undoStack.append(ctx.getImageData(0, 0, canvasWidth, canvasHeight))
            self._redoStack = []
            if self._mode in [BRUSH, ERASER]:
                ctx.lineWidth = self._lineWidth
                if self._mode == BRUSH:
                    ctx.strokeStyle = self._brushColor
                    ctx.globalCompositeOperation = 'destination-out'
                ctx.moveTo(x, y)
            elif self._mode in [RECT, ELLIPSE]:
                tmpCtx.lineWidth = self._lineWidth
                tmpCtx.strokeStyle = self._brushColor
                tmpCtx.fillStyle = self._fillColor
                points.append([x, y])
            elif self._mode == LINE:
                tmpCtx.lineWidth = self._lineWidth
                tmpCtx.strokeStyle = self._brushColor
                points.append([x, y])
            elif self._mode == SAMPLER:
                sample(x, y, event.which)
            elif self._mode == BUCKET and self._fillColor:
                # paint-bucket algorithm based on one from this website:
                # http://www.williammalone.com/articles/html5-canvas-javascript-paint-bucket-tool/
                pixelStack = [(x, y)]
                colorLayer = ctx.getImageData(0, 0, canvasWidth, canvasHeight)
                data = colorLayer.data
                startPixel = ctx.getImageData(x, y, 1, 1).data

                # convert swatch color to imageData (we're using canvas to avoid dealing with literal
                # color names like 'aquamarine')
                tmpCtx.fillStyle = self._fillColor
                locX, locY = 50, 50 # pick point far enough from the corner so it's not affected by border style
                tmpCtx.fillRect(locX, locY, 1, 1)
                swatchPixel = tmpCtx.getImageData(locX, locY, 1, 1).data
                tmpCtx.clearRect(locX, locY, 1, 1)
                if not matchStartColor(swatchPixel, 0, startPixel):
                    while len(pixelStack):
                        newPos = pixelStack.pop()
                        x = newPos[X]
                        y = newPos[Y]
                        pixelPos = (y*canvasWidth + x)*4
                        while y >= 0 and matchStartColor(data, pixelPos, startPixel):
                            y -= 1
                            pixelPos -= canvasWidth*4
                        pixelPos += canvasWidth*4
                        y += 1
                        reachLeft = False
                        reachRight = False
                        while y < canvasHeight-1 and matchStartColor(data, pixelPos, startPixel):
                            y += 1
                            data[pixelPos] = swatchPixel[0]
                            data[pixelPos+1] = swatchPixel[1]
                            data[pixelPos+2] = swatchPixel[2]
                            data[pixelPos+3] = swatchPixel[3]
                            if x > 0:
                                if matchStartColor(data, pixelPos-4, startPixel):
                                    if not reachLeft:
                                        pixelStack.append((x-1, y))
                                        reachLeft = True
                                elif reachLeft:
                                    reachLeft = False
                            if x < canvasWidth-1:
                                if matchStartColor(data, pixelPos+4, startPixel):
                                    if not reachRight:
                                        pixelStack.append((x+1, y))
                                elif reachRight:
                                    reachRight = False
                            pixelPos += canvasWidth*4
                    ctx.putImageData(colorLayer, 0, 0)
            elif self._mode == SELECT:
                lastPt = [x, y]
                if selection is None:
                    tmpCtx.lineWidth = 1
                    tmpCtx.strokeStyle = 'rgb(0,255,0)'
                    points.append([x, y])
                elif not (points[0][X] < x and x < points[1][X] \
                and points[0][Y] < y and y < points[1][Y]):
                    # we just deselected our selection, dump it back to main canvas
                    if transparent_bg:
                        # background is transparent, merge two images together
                        tmp = ctx.getImageData(points[0][X], points[0][Y], \
                            points[1][X]-points[0][X], points[1][Y]-points[0][Y]).data
                        data = selection.data
                        merge = def(offset):
                            if not data[offset+3]:
                                data[offset] = tmp[offset]
                                data[offset+1] = tmp[offset+1]
                                data[offset+2] = tmp[offset+2]
                                data[offset+3] = tmp[offset+3]
                        eachPixel(selection, merge)
                    ctx.putImageData(selection, points[0][X], points[0][Y])
                    selection = None
                    points = [[x, y]]

        onMouseMove = def(event):
            nonlocal lastPt

            if self._mode == LINE and len(points):
                x, y = getXY(this, event)
                drawSpline(tmpCtx, (x, y))
            elif dragging:
                x, y = getXY(this, event)
                if self._mode in [BRUSH, ERASER]:
                    ctx.lineTo(x, y)
                elif self._mode in [RECT, ELLIPSE]:
                    if self._mode == RECT:
                        tmpCtx.rect(points[0][X], points[0][Y], x-points[0][X], y-points[0][Y])
                        ellipse(tmpCtx, points[0][X], points[0][Y], x-points[0][X], y-points[0][Y])

                    if self._fillColor is not None:
                    if self._brushColor is not None:
                elif self._mode == SAMPLER:
                    sample(x, y, event.which)
                elif self._mode == SELECT:
                    if selection is not None:
                        for point in points:
                            point[0] += x-lastPt[X]
                            point[1] += y-lastPt[Y]
                        lastPt = [x, y]
                        x = points[1][X]
                        y = points[1][Y]
                        tmpCtx.putImageData(selection, points[0][X], points[0][Y])
                    tmpCtx.strokeRect(points[0][X], points[0][Y], x-points[0][X], y-points[0][Y])

        onMouseUp = def(event):
            nonlocal dragging, points, selection

            if self._mode in [RECT, ELLIPSE]:
                x, y = getXY(this, event)
                ctx.lineWidth = self._lineWidth
                ctx.strokeStyle = self._brushColor
                ctx.fillStyle = self._fillColor
                if self._mode == RECT:
                    ctx.rect(points[0][X], points[0][Y], x-points[0][X], y-points[0][Y])
                    ellipse(ctx, points[0][X], points[0][Y], x-points[0][X], y-points[0][Y])

                if self._fillColor is not None:
                if self._brushColor is not None:
            elif self._mode == LINE and event.which == CLICK and len(points) > 1:
                ctx.lineWidth = self._lineWidth
                ctx.strokeStyle = self._brushColor
            elif self._mode == SELECT:
                x, y = getXY(this, event)
                if selection is None and x != points[0][X] and y != points[0][Y]:
                    # we just selected something
                    sx = points[0][X]
                    sy = points[0][Y]
                    width = x-sx
                    height = y-sy
                    sx, sy, width, height = normalize(sx, sy, width, height)
                    points = [[sx, sy]]
                    x = sx+width
                    y = sy+height
                    selection = ctx.getImageData(sx, sy, width, height)
                    ctx.clearRect(sx, sy, width, height)
                    tmpCtx.putImageData(selection, points[0][X], points[0][Y])
                    tmpCtx.strokeRect(points[0][X], points[0][Y], x-points[0][X], y-points[0][Y])
                    points.append([x, y])

            dragging = False
            if (not selection or (x == points[0][X] and y == points[0][Y])) \
            and (self._mode != LINE or (event.which == CLICK and len(points) > 1)):
                points = []

        # to be used by selection manipulation filters
        self._filter = def(callback):
            nonlocal selection
            if selection:
                pixels = selection
                pixels = ctx.getImageData(0, 0, canvasWidth, canvasHeight)
            data = pixels.data
            invoke = def(offset):
                callback(data, offset)
            eachPixel(pixels, invoke)
            if selection:
                tmpCtx.putImageData(pixels, points[0][X], points[0][Y])
                tmpCtx.strokeRect(points[0][X], points[0][Y], points[1][X]-points[0][X], points[1][Y]-points[0][Y])
                ctx.putImageData(pixels, 0, 0)

        onMouseLeave = def(event):
            if dragging:

        self._ctx.lineJoin = tmpCtx.lineJoin = self._ctx.lineCap = tmpCtx.lineCap = 'round'

    def undo(self):
        state = self._undoStack.pop()
        if state:
            self._ctx.putImageData(state, 0, 0)

    def redo(self):
        state = self._redoStack.pop()
        if state:
            self._ctx.putImageData(state, 0, 0)

    # invert declared inside __init__() since it needs access to hidden 'selection' variable

    def setMode(self, mode):
        self._mode = mode

    def setStroke(self, style):
        self._brushColor = style

    def setFill(self, style):
        self._fillColor = style

    def setWidth(self, value):
        self._lineWidth = value

    def clear(self):

    def exportDwg(self):
        return self._canvas.toDataURL()

    def invert(self):
        invert = def(data, offset):
            data[offset] = 255-data[offset]
            data[offset+1] = 255-data[offset+1]
            data[offset+2] = 255-data[offset+2]

    def redFilter(self):
        remove = def(data, offset):
            data[offset] = 0

    def greenFilter(self):
        remove = def(data, offset):
            data[offset+1] = 0

    def blueFilter(self):
        remove = def(data, offset):
            data[offset+2] = 0

    def darken(self):
        darken = def(data, offset):
            data[offset] /= 2
            data[offset+1] /= 2
            data[offset+2] /= 2

    def lighten(self):
        lighten = def(data, offset):
            data[offset] = Math.min(data[offset] * 2, 255)
            data[offset+1] = Math.min(data[offset+1] * 2, 255)
            data[offset+2] = Math.min(data[offset+2] * 2, 255)


# Copyright 2012 Pyjeon Software LLC
# Author:   Alexander Tsepkov
# License:  License: Creative Commons: Attribution + Noncommerial + ShareAlike

import Drawing

class ColorSwatch:
    Color selection and preview widget
    def __init__(self):
        fg = $('<div></div>')\
            .css({'position': 'absolute', 'border': '1px solid black'})
        fg.change = def(color):
            $(self).css('background', color)
        bg = fg.clone()

def main():
    dwg = Drawing()

    # control brush sizes
    # NOTE: jQuery UI spinner implementation is inconsistent with its documentation, since their
    # implementation was buggy, this required some testing and work-arounds on my end, in particular
    # .value() method doesn't exist (.val() does), and .change() doesn't trigger when using the
    # arrows to update the value, instead I had to add my own manual trigger for it
    onChange = def():
    $brushWidget = $('#brush-size').spinner({'min': 1, 'max': 40})
    triggerChange = def():

    # map toolbox buttons to drawing modes
    $tools = $('.toolbox-item')
    makeMode = def(mode):
        setMode = def(event):
        return setMode
    $('#toolbox-brush').click()     # default to brush

    # set up color selection menus
    $stroke = $('#stroke')
    $fill = $('#fill')
    setStroke = def(style):
        $stroke.css('background', style)
    setFill = def(style):
        $fill.css('background', style)

    # farbtastic is an open-source color-picker we will be leveraging to avoid reinventing the wheel:
    # http://acko.net/blog/farbtastic-jquery-color-picker-plug-in/
    $colorPickers = $('.colorpicker')
    for $swatch, idtag, resetid, callback in [($stroke, '#fg-color', '#no-stroke', setStroke), \
                                            ($fill, '#bg-color', '#no-fill', setFill)]:
        $colorPicker = $(idtag)

        # a function generator allows us to create a handler tied to current values, not the final ones
        # that the loop terminates with (useful for keeping color-pickers independent)
        makeHandler = def($target, $popup):
            showColorpicker = def(event):
                popupUnder(event, $target, $popup)
            return showColorpicker
        $swatch.click(makeHandler($swatch, $colorPicker))

        # now add a button that resets color to transparent
        makeReset = def($target, setFunction):
            reset = def():
                event.stopPropagation() # prevent colorpicker from showing
                $target.css('background', '')
            return reset
        $(resetid).click(makeReset($swatch, callback))

    # set up top menubar
    popupUnder = def(event, $element, $popup):
        event.stopPropagation() # prevent document from closing us
        absolute = $element.offset()
        $popup.css({'left': absolute.left, 'top': absolute.top+$element.outerHeight()}).show()
    $menus = $('.menubar-menu')
    makeMenu = def():
        $this = $(this)
        idtag = $this.attr('id').split('-')[2]
        showMenu = def(event):
            popupUnder(event, $this, $('#menubar-menu-'+idtag))

    # hide menus if we click anywhere outside the menu div
    hideMenus = def():

    $menus.menu().removeClass('ui-widget')  # ui-widget adds ugly font, we don't want that

    # attach actual functionality to menu items
        # start new drawing
        def(): dwg.clear()

        # export drawing to PNG
            url = dwg.exportDwg()
            window.open(url, '_blank')

        # undo last action
        def(): dwg.undo()

        # redo last action
        def(): dwg.redo()

        # invert colors
        def(): dwg.invert()

        # filter out red channel
        def(): dwg.redFilter()

        # filter out green channel
        def(): dwg.greenFilter()

        # filter out blue channel
        def(): dwg.blueFilter()

        # make image half as bright
        def(): dwg.darken()

        # make image twice as bright
        def(): dwg.lighten()

    # pre-cache the element, so we don't have to 'select' it repeatedly
    $about = $('#about').dialog({'modal': True})
        # show info about this app
        def(): $about.dialog('open')

    # disable right-click menu
    window.oncontextmenu = def(): return False


Ray-Tracing Example

Author: Salvatore DI DIO License: License: Creative Commons: Attribution + Noncommerial + ShareAlike

This demo shows Ray-Tracing done on a canvas element in RapydScript. This demo is based on logic described on this website. You can also play with this demo's source code live here.


# Based on:
# Documentation :
# http://www.scratchapixel.com
# Thanks to Alexander Tsepkov the creator of RapydScript, and
# Charles Law for allowing testing RapydScript online
# Post comments and suggestions on RapydScript Google Group
# You can modify the script and see the modifications
# in direct live (quite)
# When modifying you can reduce WIDTH and HEIGHT to avoid lag
# Authored by Salvatore DI DIO (email : salvatore dot didio at gmail)
# Best performance with Firefox
# On my mac book pro : 1280 x 800 rendered in less than a minute

WIDTH = 440
HEIGHT = 240

class Vector:
    def __init__(self, x=0, y=0, z=0):
        self.x = x
        self.y = y
        self.z = z

    def copy(self):
        return Vector(self.x, self.y, self.z)

    def length(self):
        return Math.sqrt(self.x*self.x + self.y*self.y + self.z*self.z)

    def sqrLength(self):
        return (self.x*self.x + self.y*self.y + self.z*self.z)

    def normalize(self):
        inv = 1.0/self.length()
        return Vector(self.x*inv, self.y*inv,self.z*inv)

    def negate(self):
        return Vector(-self.x, -self.y, -self.z)

    def add(self,v):
        return Vector(self.x + v.x, self.y + v.y, self.z + v.z)

    def sub(self,v):
        return Vector(self.x - v.x, self.y - v.y, self.z - v.z)

    def times(self, k):
        return Vector(k*self.x, k*self.y, k*self.z)

    def dot(self,t):
        return self.x*t.x + self.y*t.y+self.z*t.z

    def cross(self,w):
        v = Vector()
        v.x =  self.y * w.z - self.z * w.y
        v.y = -self.x * w.z + self.z * w.x
        v.z = self.x * w.y - self.y * w.x
        return v

    def zero():
        return Vector()

class Ray:
    def __init__(self, vectOrigin, vectDest):
        self.origin = vectOrigin
        self.direction = vectDest

    def getPoint(self, t):
        # p = self.origin + t * self.direction
        return self.origin.add(self.direction.times(t))

class IntersectionResult:
    def __init__(self):
        self.sceneobject = null
        self.distance = 0
        self.position = Vector.zero()
        self.normal = Vector.zero()

    def nohit():
        return IntersectionResult()

class Sphere:
    def __init__(self, center = Vector(0,2,-1), radius = 1):
        self.center = center
        self.radius = radius
        self.name = "Sphere"

    def initialize(self):
        self.sqrRadius = self.radius * self.radius

    def double(self):
        return 100

    def intersect(self, ray):    
        v = ray.origin.sub(self.center)
        a0 = v.sqrLength() - self.sqrRadius
        DdotV = ray.direction.dot(v)

        if DdotV <= 0: 
            discr = DdotV * DdotV - a0   
            if discr >= 0:
                r = IntersectionResult()
                r.sceneobject = self
                r.distance =  -DdotV - Math.sqrt(discr)
                r.position = ray.getPoint(r.distance)
                r.normal =  r.position.sub(self.center)
                return r
        return IntersectionResult.nohit()

class Camera:
    def __init__(self,eye, front, up, fov,aspectratio):
        self.eye = eye
        self.front = front 
        self.RefUp = up
        self.fov = fov
        self.aspectratio = aspectratio

    def initialize(self):
        self.right =  self.front.cross(self.RefUp)
        self.up = self.right.cross(self.front)
        self.fovScale = Math.tan(self.fov * 0.5 * Math.PI/180)*2

    def generateRay(self,x,y):
        r = self.right.times((x - 0.5) * self.fovScale*self.aspectratio)
        u = self.up.times((y - 0.5) * self.fovScale)
        ray = Ray(self.eye, self.front.add(r).add(u).normalize())
        return ray

class Union:
    def __init__(self,geometries):
        self.geometries = geometries

    def initialize(self):
        for geo in self.geometries:

    def intersect(self, ray):
        minDistance = Infinity
        minResult = IntersectionResult()
        for geo in self.geometries:
            result = geo.intersect(ray)
            if ((result.sceneobject != null) and (result.distance < minDistance)):        
                minDistance = result.distance
                minResult = result
        return minResult

class Plane:
    def __init__(self, normal, distance):
        self.normal = normal
        self.d = distance
        self.name = "Plane"

    def initialize(self):
        self.position = self.normal.times(self.d)

    def intersect(self, ray):
        a =  ray.direction.dot(self.normal)
        if a >= 0:
            return IntersectionResult.nohit()

        b = self.normal.dot(ray.origin.sub(self.position))
        result = IntersectionResult()
        result.sceneobject = self
        result.distance = -b/a
        result.position =  ray.getPoint(result.distance)
        result.normal = self.normal
        return result

class Color:
    def __init__(self, r=0, g=0, b=0):
        self.r = r
        self.g = g
        self.b = b

    def add(self,c):
       return Color(self.r + c.r, self.g + c.g, self.b + c.b)

    def times(self, k):
        return Color(k*self.r, k*self.g, k*self.b)

    def modulate(self, c):
        return Color(self.r * c.r, self.g * c.g, self.b * c.b)

Color.black = Color()
Color.red = Color(1,0,0)
Color.white = Color(1,1,1)
Color.green = Color(0,1,0)
Color.blue = Color(0,0,1)
Color.gold = Color(1,0.8745098039215686,0)
Color.cyan = Color(0,1,1)

class CheckerMaterial:
    def __init__(self, scale, reflectiveness):
        self.scale = scale
        self.reflectiveness = reflectiveness

    def sample(self,ray,position, normal):
        v =  Math.abs((Math.floor(position.x * 0.1) \
           + Math.floor(position.z * self.scale)) % 2)
        if v < 1:
            return Color.black
            return Color.white

lightDir =  Vector(6, 8.5, 15).normalize()
lightColor = Color(1,1,1)

class PhongMaterial:
    def __init__(self,diffuse, specular, shininess, reflectiveness):
        self.diffuse = diffuse
        self.specular = specular
        self.shininess = shininess
        self.reflectiveness = reflectiveness

    def sample(self,ray,position,normal):
        NdotL = normal.dot(lightDir)
        H = (lightDir.sub(ray.direction)).normalize()
        NdotH = normal.dot(H)
        diffuseTerm = self.diffuse.times(Math.max(NdotL, 0))        
        specularTerm =  self.specular.times(Math.pow(Math.max(NdotH, 0), self.shininess))
        return lightColor.modulate(diffuseTerm.add(specularTerm))

def rayTraceRecursive(scene, ray, maxReflect):
    result = scene.intersect(ray)
    if result.sceneobject:
        reflectiveness = result.sceneobject.material.reflectiveness
        mat = result.sceneobject.material
        color = mat.sample(ray, result.position,result.normal)
        color.times(1 - reflectiveness)
        if (reflectiveness > 0 and maxReflect > 0):
            r = result.normal.times(-2 * result.normal.dot(ray.direction)).add(ray.direction)
            ray = Ray(result.position, r)
            reflectedColor = rayTraceRecursive(scene, ray, maxReflect - 1)
            color = color.add(reflectedColor.times(reflectiveness))
        return color
    return Color(0,0,0)

def main():
    canvas = document.getElementById("canvas")
    ctx = canvas.getContext("2d")
    canvas.width = WIDTH
    canvas.height = HEIGHT
    w  = canvas.width
    h = canvas.height
    ctx.fillStyle = "rgb(255,255,255)"
    ctx.fillRect(0, 0, w, h)

    aspectratio = w/h
    camera = Camera(Vector(0,0.56,8),Vector(0,0,-1),Vector(0,1,0),30,aspectratio)

    sphere = Sphere(Vector(1.5,0,-1),1)
    sphere.material = PhongMaterial(Color.red, Color.white, 500,0.25)

    sphere2 = Sphere(Vector(-1.5,0,-1),1)
    sphere2.material = PhongMaterial(Color.gold, Color.white, 500,0.25)

    sphere3 = Sphere(Vector(0,0,-2),1)
    sphere3.material = PhongMaterial(Color.blue, Color.white, 500,0.25)

    plane = Plane(Vector(0,1,0),-1)
        plane.material = CheckerMaterial(0.1,0.9);
        plane.material = PhongMaterial(Color.cyan, Color.white, 500,1.0)

    scene = Union([plane,sphere,sphere2,sphere3])

    maxDepth = 20

    imgdata = ctx.createImageData(w,h)
    for y in range(h):
        for x in range(w):
            sy = 1- y/h
            sx = x/w
            index = (x + y * w) * 4
            ray = camera.generateRay(sx, sy)
            result = scene.intersect(ray)
            if (result.sceneobject != null):
                col =  rayTraceRecursive(scene, ray, 4)
                imgdata.data[index + 0] = col.r * 255
                imgdata.data[index + 1] = col.g * 255
                imgdata.data[index + 2] = col.b * 255
                imgdata.data[index + 3] = 255
                imgdata.data[index + 0] = 0
                imgdata.data[index + 1] = 0
                imgdata.data[index + 2] = 0
                imgdata.data[index + 3] = 255


toggle = document.getElementById('toggle')
toggle.onclick = def():
    nonlocal CHECKER_PATTERN

Stocks App Example

Copyright 2013 Pyjeon Software LLC
Author: Alexander Tsepkov
License: Creative Commons: Attribution + Noncommerial + ShareAlike


This is a web app for retrieving stock information from Google Finance and displaying it in a chart. The app also allows the user to modify the way stock data is presented. This app showcases:

  • Including stdlib as an import instead of a separate js file
  • Integration with Google Charts API
  • Ability to wrap/abstract entire APIs in separate classes (YQL and Google Charts)
  • Ability to create Pyjamas-like widgets that wrap around existing elements and extend their functionality (stock picker input widget, stock chart widget)
  • Ability to work with Flash (Google Charts API currently uses Flash for rendering)
  • Pulling data from outside sources via YQL (NASDAQ and Google Finance)
  • Asynchronous coding via RapydScript
  • RapydScript integration with RapydML and RapydCSS/SASS


# Copyright 2013 Pyjeon Software LLC
# Author:   Alexander Tsepkov
# License: Creative Commons: Attribution + Noncommerial + ShareAlike

import stdlib
import yql

class Stock:
    Class responsible for tracking stock information. It handles tracking current stock name
    as well as performing the YQL query to retrieve stock data from Google and storing the
    def __init__(self, symbols, callback=None):

        self.callback = callback

        # pre-cache these so we're not retrieving the element again on every blur/enter event
        $start = $('#start-date')
        $end = $('#end-date')
        onUpdate = def($event, ui):
            # select triggers before the update occurs, so we have to use 'select' name instead of .val()
            if $event.type == 'autocompleteselect':
            startdate = $start.val()
            enddate = $end.val()
            self.get(startdate, enddate)

        self.$widget = $('<input></input>').autocomplete({
            'minLength':    2,
            'source':       symbols,
            'select':       onUpdate,
        self.data = None

        ENTER = 13
        onKeypress = def($event):
            code = $event.keyCode or $event.which
            if code == ENTER:

    def get(self, startDate, endDate):
        get method for this stock, which retrieves info from google finance
        name = self.$widget.val()
        if self.symbol == name and self.startDate == startDate and self.endDate == endDate:
            # don't trigger a refresh unless the parameters have been updated
        self.symbol = name
        self.startDate = startDate
        self.endDate = endDate

        format = def(date):
            mmddyyyy = date.split('/')
            return '-'.join([mmddyyyy[2], mmddyyyy[0], mmddyyyy[1]])

        # google's historical data is only available in CSV format, but we need it in
        # JSON if we want to work with it in JavaScript. We will use Yahoo's YQL to
        # convert it. We will also need to create a callback function, since the request
        # is asynchronous
        onUpdate = def(query):
            if query['results'] is None:
                # bad symbol
                self.data = None
                self.$widget.css('background', '#fdd')
                self.data = query['results']['row']
                self.$widget.css('background', '')
            if self.callback is not None:
                self.callback(name, self.data)

        if self.symbol != '':
            # we will use csv version of the data to avoid dealing with unnecessary html
            url = 'http://www.google.com/finance/historical?q=' + name \
                + '&startdate=' + format(startDate) \
                + '&enddate=' + format(endDate) + '&output=csv'

            YQL('select * from csv where url="' + url + '"', onUpdate).fetch()
            # box is blank, don't try to request YQL for it
            self.data = None
            self.$widget.css('background', '')
            self.callback(name, self.data)

class StockChart:
    This is a wrapper around Google Charts. It handles displaying and updating the chart,
    abstracting Google Charts away from the rest of the app. It also handles compiling
    the data in a representation Google Charts supports since the format of the API is
    somewhat clunky compared to other plotting tools like jqPlot. The only thing it doesn't
    wrap around is the loading logic for Google Charts API, unfortunately, which is due to
    awkward loading implementation on Google's part.
    def __init__(self):
        $options = $('#chart-options')

        # create various technical indicator filters the user can play with
        self._filters = {}
        self._filterLogic = {}
        addFilter = def(name, callback):
            $button = $('<input type="checkbox" value="'
                    + name.replace(' ', '-') + '">' + name + '</input>')
            self._filterLogic[name] = callback
            setFilter = def():
                self._filters[name] = $(this).is(':checked')

        normalize = def(cols, rows):
            # divide all stocks by their starting value so we can compare them better
            normalized = []
            orig = rows[len(rows)-1]
            for day, row in enumerate(rows):
                for num, stock in enumerate(row[1:]):
            return cols, normalized
        makeMovingAvg = def(name, days, ema):
            sum = def(a, b):
                return a + b
            return name, def(cols, rows):
                for idx, col in enumerate(cols):
                    if col[len(col)-1] != ')':
                        avgs = [{}]
                        moving = []
                        if ema:
                            alpha = 2/(days+1)
                            moving = rows[len(rows)-1][idx+1]
                        for row in reversed(rows):
                            if ema:
                                moving = alpha * row[idx+1] + (1-alpha)*moving
                                avgs.unshift({'col4': moving})
                                avgs.unshift({'col4': moving.reduce(sum)/len(moving)})
                                if len(moving) >= days:
                        self.add(col + ' (' + name + ')', avgs, cols, rows)
                return cols, rows
        rsi = def(cols, rows):
            # relative strength index filter
            tmp, ema15 = makeMovingAvg('', 15, True)
            for idx, col in enumerate(cols):
                if col[len(col)-1] != ')':
                    rsis = [{}]
                    ups = []
                    downs = []
                    prev = rows[len(rows)-1][idx+1]
                    for row in reversed(rows):
                        current = row[idx+1]
                        DATE = 0    # fake date placeholder
                        if current > prev:
                            ups.unshift([DATE, current-prev])
                            downs.unshift([DATE, 0])
                            ups.unshift([DATE, 0])
                            downs.unshift([DATE, prev-current])
                        prev = current
                    tmp, ups = ema15(['up'], ups)
                    tmp, downs = ema15(['down'], downs)
                    for index in range(len(ups)):
                        if downs[index][2]:
                            rs = ups[index][2]/downs[index][2]
                            rs = 100
                        rsis.append({'col4': 100-100/(1+rs)})
                    self.add(col + ' (RSI)', rsis, cols, rows)
            return cols, rows

        for name, fun in [\
                ('Normalize', normalize), \
                makeMovingAvg('15-Day SMA', 15, False), \
                makeMovingAvg('50-Day SMA', 50, False), \
                makeMovingAvg('15-Day EMA', 15, True), \
                makeMovingAvg('50-Day EMA', 50, True), \
                ('15-Day RSI', rsi)]:
            addFilter(name, fun)

        # this is an external API, so we have to use the 'new' keyword
        self.annotatedTimeline = new google.visualization.AnnotatedTimeLine($('#chart').get(0))

    def clear(self):
        self._cols = []
        self._rows = []

    def add(self, symbol, data, cols=self._cols, rows=self._rows):
        if data is not None:
            if not len(cols):
                # this is the first symbol, create rows and append date
                for item in data[1:]:
                    rows.append([Date(item['col0']), float(item['col4'])])
                # rows have already been generated, just append to them
                for index, item in enumerate(data[1:]):

    def redraw(self):
        # make a deep-copies of the data
        cols = $.extend(True, [], self._cols)
        rows = $.extend(True, [], self._rows)
        for key, val in dict.items(self._filters):
            if val:
                cols, rows = self._filterLogic[key](cols, rows)

        data = new google.visualization.DataTable()
        data.addColumn('date', 'Date')
        for col in cols:
            data.addColumn('number', col)
        self.annotatedTimeline.draw(data, {})

def main():
    stockFields = []

    # declare a dummy method to avoid errors in case it is triggered before Google Charts finish loading
    updateChart = def():

    # load charting widget
    onChartLoad = def():
        nonlocal updateChart
        chart = StockChart()
        updateChart = def(symbol, data):
            for stock in stockFields:
                chart.add(stock.symbol, stock.data)
    google.load('visualization', '1', {'packages':['annotatedtimeline'], 'callback': onChartLoad})

    # set up stock picker and dates
    triggerChange = def():
    $start = $('#start-date').datepicker({'onSelect': triggerChange})
    $end = $('#end-date').datepicker({'onSelect': triggerChange})
    $start.datepicker('setDate', -90)
    $end.datepicker('setDate', Date())

    symbols = []
    $stocks = $('#stock-input')

    # add new input widget
    newStock = def():
        symbol = Stock(symbols)
        get = def():
            startdate = $start.val()
            enddate = $end.val()
            symbol.get(startdate, enddate)

        # handle addition and removal of new widgets
        # new widgets will get added when there are no more empty widgets
        # empty widgets will get removed when there is more than one
        onWidgetUpdate = def(label, data):
            value = symbol.symbol
            isLast = symbol.$widget.is(':last-child')
            if isLast and value != '':
            elif value == '' and not isLast:
            updateChart(label, data)
        symbol.callback = onWidgetUpdate

    # NASDAQ returns stock symbols in CSV format as well, YQL will convert them for us
    # it also seems to store companies from other stock exchanges, which makes work easier for us
    sync = 0
    exchanges = ['nyse', 'nasdaq', 'lon']
    onUpdate = def(query):
        nonlocal sync, symbols
        for symbol in query['results']['row']:
        sync += 1

        if sync == len(exchanges):
            # all exchanges loaded, remove duplicates and create stock picker
            unique = def(element, index):
                return this.index(element) == index
            symbols = symbols.filter(unique, symbols)

    $stocks.text('Loading Symbols from Stock Exchanges')
    for exchange in exchanges:
        YQL('select col0 from csv where url="http://www.nasdaq.com/screening/companies-by-name.aspx?letter=0&exchange=' + exchange + '&render=download"', onUpdate).fetch()


Mailing List

This is a general group for anything RapydScript related, including questions, suggestions, comments, and showing your own projects.

RapydScript Mailing List

Community Projects

Flask RapydScript: Automatically compile RapydScript files when developing for the Flask Framework

RapydScript Decompiler: Convert JavaScript files to RapydScript

RapydScript Showcases: RapydScript user Salvatore DI DIO setup a showcase of RapydScript examples.

Vim RapydScript: RapydScript plugin for Vim