js-weighted-list
v0.1.1
Published
Small weighted-probability list library (node.js port fork)
Downloads
34,339
Readme
js-weighted-list
This is a smallish library which implements a weighted list, from which elements can be picked out at random with a probability dependent on their weight. The list implements random sampling without replacement.
As an example:
var data = [['a', 10],
['b', 1],
['c', 1],
['d', 5],
['e', 3]];
var wl = new WeightedList(data);
// wl.peek() returns items at random from the list, and does not modify the list.
var result = wl.peek(); // Ex: ['a']
var result = wl.peek(); // Ex: ['e']
var result = wl.peek(3); // Ex: ['a', 'c', 'd']
var result = wl.peek(3); // Ex: ['e', 'a', 'd']
var result = wl.peek(3); // Ex: ['a', 'b', 'e']
// wl.pop() returns random items from the list and removes the items it found
var result = wl.pop(2); // Ex: ['a', 'd'], after which wl consists of [ ['b', 1], ['c', 1], ['e', 3] ]
// wl.push() adds new data into the set
// note that despite the terms push and pop, the weighted list has no natural order
wl.push('f', 6); // wl is now [ ['b', 1], ['c', 1], ['e', 3], ['f', 6] ]
// wl.addWeight() will increase the weight of a list item (or decrease it if the user passes a negative number)
wl.addWeight('b', 4); // wl is now [ ['b', 5], ['c', 1], ['e', 3], ['f', 6] ]
// wl.shuffle() will return the entire list in random order.
wl.shuffle(); // Ex: ['b', 'f', 'c', 'e']
wl.shuffle(); // Ex: ['f', 'e', 'b', 'c']
wl.shuffle(); // Ex: ['f', 'b', 'e', 'c']
This project initially grew from a javascript reimplementation of this Stack Overflow answer by Jason Orendorff, and it still uses basically the same algorithm.
js-weighted-list has no external dependencies and is licensed under the MIT License.
Usage
Include the file:
<script type="text/javascript" src="js-weighted-list.js"></script>
(Eventually, the node.js equivalent should be require('js-weighted-list')
, but I
haven't verified that the package structure is correct for npm yet.)
The basic object which is exported is a WeightedList, which you can instantiate like this:
// Initially empty list
var wl = new WeightedList();
// Array of arrays
var wl = new WeightedList([ ['a': 1], ['b': 3], ['c': 1] ]);
// Array of objects
var wl = new WeightedList([ {'key': 'a', 'weight': 1},
{'key': 'b', 'weight': 3},
{'key': 'c', 'weight': 1} ]);
Every item on the list has a key, a weight, and optionally, some other data attached (see below). Keys must be unique and must be usable as keys in an object (eg, strings or integers).
Attaching Data
In addition to just returning key values in a random order, a WeightedList instance can associate arbitrary data with particular keys, which is then returned along with the keys.
Additional data can be passed in as a third element in the constructor list,
or as a third argument to the addItem()
method:
var sandwich = new WeightedList(
[ ['marv', 10, { name: 'Marvin', sandwich: 'roast beef' } ],
['bob', 1, { name: 'Bob', sandwich: 'turkey' } ] ]);
// Equivalent to above
var sandwich = new WeightedList();
sandwich.push('marv', 10, { name: 'Marvin', sandwich: 'roast beef' });
sandwich.push('bob', 1, { name: 'Bob', sandwich: 'turkey' });
If a WeightedList instance contains data elements as described above, methods
which retrieve elements from it will return objects of the form
{'key': k, 'data': d}
instead of simply key
. For example:
sandwich.peek(); // Ex: [ {'key': 'marv', 'data': {'name': 'Marvin', 'sandwich': 'roast beef'}} ]
sandwich.shuffle(); // Ex: [ {'key': 'marv', 'data': {'name': 'Marvin', 'sandwich': 'roast beef'}},
// {'key': 'bob', 'data': {'name': 'Bob', 'sandwich': 'turkey' }} ]
You can also pass in an {'key': 'k', 'weight': 12,
instead of an array if you'd prefer to:
var planets = new WeightedList(
[ {'key': 'earth', 'weight': 10, 'data': {'orbit': 3}},
{'key': 'mars', 'weight': 7, 'data': {'orbit': 4}} ]);
planets.push({'key': 'pluto', 'weight': 1, 'data': {'orbit': 9}});
sandwich.peek(); // Ex: [ {'key': 'earth', 'data': {'orbit': 3}} ]
As with using arrays as parameters, the 'data'
key is optional, and if no key
in the list has data attached to it, return values from operations that retrieve
values from the list will return simple lists of keys instead of {'key': k, 'data': d}
pairs.
TODO
better error handling (stack underflow, etc)
expand qunit unit tests and look into mocha
node.jsify. In particular, figure out AMD and maybe use amdefine. The goal is to be able to run mocha tests from the command-line as well as in the browser.
Minified version?
'use strict';
browser tests which graph lists, stats
Not entirely sure if the whole "data" thing is really neccesary