index.html

<!DOCTYPE HTML>
<html>
<head>
  <meta http-equiv="content-type" content="text/html;charset=UTF-8" />
  <meta http-equiv="X-UA-Compatible" content="chrome=1">
  <title>Underscore.js</title>
  <style>
    body {
      font-size: 16px;
      line-height: 24px;
      background: #f0f0e5;
      color: #252519;
      font-family: "Palatino Linotype", "Book Antiqua", Palatino, FreeSerif, serif;
    }
    div.container {
      width: 720px;
      margin: 50px 0 50px 50px;
    }
    p {
      width: 550px;
    }
      #documentation p {
        margin-bottom: 4px;
      }
    a, a:visited {
      padding: 0 2px;
      text-decoration: none;
      background: #dadaba;
      color: #252519;
    }
    a:active, a:hover {
      color: #000;
      background: #f0c095;
    }
    h1, h2, h3, h4, h5, h6 {
      margin-top: 40px;
    }
    b.header {
      font-size: 18px;
    }
    span.alias {
      font-size: 14px;
      font-style: italic;
      margin-left: 20px;
    }
    table, tr, td {
      margin: 0; padding: 0;
    }
      td {
        padding: 2px 12px 2px 0;
      }
    code, pre, tt {
      font-family: Monaco, Consolas, "Lucida Console", monospace;
      font-size: 12px;
      line-height: 18px;
      color: #555529;
    }
      code {
        margin-left: 20px;
      }
      pre {
        font-size: 12px;
        padding: 2px 0 2px 12px;
        border-left: 6px solid #aaaa99;
        margin: 0px 0 30px;
      }
  </style>
</head>
<body>

  <div class="container">

    <h1>Underscore.js</h1>

    <p>
      <a href="http://github.com/documentcloud/underscore/">Underscore</a> is a
      utility-belt library for JavaScript that provides a lot of the
      functional programming support that you would expect in
      <a href="http://prototypejs.org/api">Prototype.js</a>
      (or <a href="http://www.ruby-doc.org/core/classes/Enumerable.html">Ruby</a>),
      but without extending any of the built-in JavaScript objects. It's the
      tie to go along with <a href="http://docs.jquery.com">jQuery</a>'s tux.
    </p>

    <p>
      Underscore provides 60-odd functions that support both the usual
      functional suspects: <b>map</b>, <b>select</b>, <b>invoke</b> &mdash;
      as well as more specialized helpers: function binding, javascript
      templating, deep equality testing, and so on. It delegates to built-in
      functions, if present, so modern browsers will use the
      native implementations of <b>forEach</b>, <b>map</b>, <b>reduce</b>,
      <b>filter</b>, <b>every</b>, <b>some</b> and <b>indexOf</b>.
    </p>

    <p>
      A complete <a href="test/test.html">Test &amp; Benchmark Suite</a>
      is included for your perusal.
    </p>
    
    <p>
      You may also read through the <a href="docs/underscore.html">annotated source code</a>.
    </p>

    <p>
      The project is
      <a href="http://github.com/documentcloud/underscore/">hosted on GitHub</a>.
      You can report bugs and discuss features on the
      <a href="http://github.com/documentcloud/underscore/issues">issues page</a>,
      on Freenode in the <tt>#documentcloud</tt> channel,
      or send tweets to <a href="http://twitter.com/documentcloud">@documentcloud</a>.
    </p>

    <p>
      <i>Underscore is an open-source component of <a href="http://documentcloud.org/">DocumentCloud</a>.</i>
    </p>

    <h2>Downloads <i style="padding-left: 12px; font-size:12px;">(Right-click, and use "Save As")</i></h2>

    <table>
      <tr>
        <td><a href="underscore.js">Development Version (1.1.7)</a></td>
        <td><i>28kb, Uncompressed with Comments</i></td>
      </tr>
      <tr>
        <td><a href="underscore-min.js">Production Version (1.1.7)</a></td>
        <td><i>3kb, Minified and Gzipped</i></td>
      </tr>
    </table>

    <h2>Table of Contents</h2>
    
    <a href="#styles">Object-Oriented and Functional Styles</a>

    <p>
      <b>Collections</b>
      <br />
      <span class="methods"><a href="#each">each</a>, <a href="#map">map</a>,
      <a href="#reduce">reduce</a>, <a href="#reduceRight">reduceRight</a>,
      <a href="#detect">detect</a>, <a href="#select">select</a>,
      <a href="#reject">reject</a>, <a href="#all">all</a>,
      <a href="#any">any</a>, <a href="#include">include</a>,
      <a href="#invoke">invoke</a>, <a href="#pluck">pluck</a>,
      <a href="#max">max</a>, <a href="#min">min</a>,
      <a href="#sortBy">sortBy</a>, <a href="#groupBy">groupBy</a>, <a href="#sortedIndex">sortedIndex</a>,
      <a href="#toArray">toArray</a>, <a href="#size">size</a></span>
    </p>

    <p>
      <b>Arrays</b>
      <br />
      <span class="methods"><a href="#first">first</a>, <a href="#rest">rest</a>, <a href="#last">last</a>,
      <a href="#compact">compact</a>, <a href="#flatten">flatten</a>, <a href="#without">without</a>, 
      <a href="#union">union</a>, <a href="#intersection">intersection</a>, <a href="#difference">difference</a>, 
      <a href="#uniq">uniq</a>, <a href="#zip">zip</a>, <a href="#indexOf">indexOf</a>,
      <a href="#lastIndexOf">lastIndexOf</a>, <a href="#range">range</a></span>
    </p>

    <p>
      <b>Functions</b>
      <br />
      <span class="methods"><a href="#bind">bind</a>, <a href="#bindAll">bindAll</a>, 
      <a href="#memoize">memoize</a>, <a href="#delay">delay</a>, <a href="#defer">defer</a>, 
      <a href="#throttle">throttle</a>, <a href="#debounce">debounce</a>, 
      <a href="#once">once</a>, <a href="#after">after</a>, <a href="#wrap">wrap</a>, <a href="#compose">compose</a></span>
    </p>

    <p>
      <b>Objects</b>
      <br />
      <span class="methods"><a href="#keys">keys</a>, <a href="#values">values</a>,
      <a href="#functions">functions</a>, <a href="#extend">extend</a>, <a href="#defaults">defaults</a>, <a href="#clone">clone</a>, <a href="#tap">tap</a>,
      <a href="#isEqual">isEqual</a>, <a href="#isEmpty">isEmpty</a>, <a href="#isElement">isElement</a>,
      <a href="#isArray">isArray</a>, <a href="#isArguments">isArguments</a>, <a href="#isFunction">isFunction</a>, <a href="#isString">isString</a>,
      <a href="#isNumber">isNumber</a>, <a href="#isBoolean">isBoolean</a>, <a href="#isDate">isDate</a>, <a href="#isRegExp">isRegExp</a>
      <a href="#isNaN">isNaN</a>, <a href="#isNull">isNull</a>,
      <a href="#isUndefined">isUndefined</a>
      </span>
    </p>

    <p>
      <b>Utility</b>
      <br />
      <span class="methods"><a href="#noConflict">noConflict</a>,
      <a href="#identity">identity</a>, <a href="#times">times</a>, 
      <a href="#mixin">mixin</a>, <a href="#uniqueId">uniqueId</a>, 
      <a href="#template">template</a></span>
    </p>

    <p>
      <b>Chaining</b>
      <br />
      <span class="methods"><a href="#chain">chain</a>, <a href="#value">value</a>
    </p>

    <div id="documentation">
      
      <h2 id="styles">Object-Oriented and Functional Styles</h2>

      <p>
        You can use Underscore in either an object-oriented or a functional style,
        depending on your preference. The following two lines of code are
        identical ways to double a list of numbers.
      </p>

    <pre>
_.map([1, 2, 3], function(n){ return n * 2; });
_([1, 2, 3]).map(function(n){ return n * 2; });</pre>

      <p>
        Using the object-oriented style allows you to chain together methods. Calling
        <tt>chain</tt> on a wrapped object will cause all future method calls to
        return wrapped objects as well. When you've finished the computation,
        use <tt>value</tt> to retrieve the final value. Here's an example of chaining
        together a <b>map/flatten/reduce</b>, in order to get the word count of
        every word in a song.
      </p>

<pre>
var lyrics = [
  {line : 1, words : "I'm a lumberjack and I'm okay"},
  {line : 2, words : "I sleep all night and I work all day"},
  {line : 3, words : "He's a lumberjack and he's okay"},
  {line : 4, words : "He sleeps all night and he works all day"}
];

_(lyrics).chain()
  .map(function(line) { return line.words.split(' '); })
  .flatten()
  .reduce(function(counts, word) {
    counts[word] = (counts[word] || 0) + 1;
    return counts;
}, {}).value();

=&gt; {lumberjack : 2, all : 4, night : 2 ... }</pre>

      <p>
        In addition, the
        <a href="https://developer.mozilla.org/en/JavaScript/Reference/Global_Objects/Array/prototype">Array prototype's methods</a>
        are proxied through the chained Underscore object, so you can slip a
        <tt>reverse</tt> or a <tt>push</tt> into your chain, and continue to
        modify the array.
      </p>

      <h2>Collection Functions (Arrays or Objects)</h2>

      <p id="each">
        <b class="header">each</b><code>_.each(list, iterator, [context])</code>
        <span class="alias">Alias: <b>forEach</b></span>
        <br />
        Iterates over a <b>list</b> of elements, yielding each in turn to an <b>iterator</b>
        function. The <b>iterator</b> is bound to the <b>context</b> object, if one is
        passed. Each invocation of <b>iterator</b> is called with three arguments:
        <tt>(element, index, list)</tt>. If <b>list</b> is a JavaScript object, <b>iterator</b>'s
        arguments will be <tt>(value, key, list)</tt>. Delegates to the native
        <b>forEach</b> function if it exists.
      </p>
      <pre>
_.each([1, 2, 3], function(num){ alert(num); });
=&gt; alerts each number in turn...
_.each({one : 1, two : 2, three : 3}, function(num, key){ alert(num); });
=&gt; alerts each number in turn...</pre>

      <p id="map">
        <b class="header">map</b><code>_.map(list, iterator, [context])</code>
        <br />
        Produces a new array of values by mapping each value in <b>list</b>
        through a transformation function (<b>iterator</b>). If the native <b>map</b> method
        exists, it will be used instead. If <b>list</b> is a JavaScript object,
        <b>iterator</b>'s arguments will be <tt>(value, key, list)</tt>.
      </p>
      <pre>
_.map([1, 2, 3], function(num){ return num * 3; });
=&gt; [3, 6, 9]
_.map({one : 1, two : 2, three : 3}, function(num, key){ return num * 3; });
=&gt; [3, 6, 9]</pre>

      <p id="reduce">
        <b class="header">reduce</b><code>_.reduce(list, iterator, memo, [context])</code>
        <span class="alias">Aliases: <b>inject, foldl</b></span>
        <br />
        Also known as <b>inject</b> and <b>foldl</b>, <b>reduce</b> boils down a
        <b>list</b> of values into a single value. <b>Memo</b> is the initial state
        of the reduction, and each successive step of it should be returned by
        <b>iterator</b>.
      </p>
      <pre>
var sum = _.reduce([1, 2, 3], function(memo, num){ return memo + num; }, 0);
=&gt; 6
</pre>

      <p id="reduceRight">
        <b class="header">reduceRight</b><code>_.reduceRight(list, iterator, memo, [context])</code>
        <span class="alias">Alias: <b>foldr</b></span>
        <br />
        The right-associative version of <b>reduce</b>. Delegates to the
        JavaScript 1.8 version of <b>reduceRight</b>, if it exists. <b>Foldr</b>
        is not as useful in JavaScript as it would be in a language with lazy
        evaluation.
      </p>
      <pre>
var list = [[0, 1], [2, 3], [4, 5]];
var flat = _.reduceRight(list, function(a, b) { return a.concat(b); }, []);
=&gt; [4, 5, 2, 3, 0, 1]
</pre>

      <p id="detect">
        <b class="header">detect</b><code>_.detect(list, iterator, [context])</code>
        <br />
        Looks through each value in the <b>list</b>, returning the first one that
        passes a truth test (<b>iterator</b>). The function returns as
        soon as it finds an acceptable element, and doesn't traverse the
        entire list.
      </p>
      <pre>
var even = _.detect([1, 2, 3, 4, 5, 6], function(num){ return num % 2 == 0; });
=&gt; 2
</pre>

      <p id="select">
        <b class="header">select</b><code>_.select(list, iterator, [context])</code>
        <span class="alias">Alias: <b>filter</b></span>
        <br />
        Looks through each value in the <b>list</b>, returning an array of all
        the values that pass a truth test (<b>iterator</b>). Delegates to the
        native <b>filter</b> method, if it exists.
      </p>
      <pre>
var evens = _.select([1, 2, 3, 4, 5, 6], function(num){ return num % 2 == 0; });
=&gt; [2, 4, 6]
</pre>

      <p id="reject">
        <b class="header">reject</b><code>_.reject(list, iterator, [context])</code>
        <br />
        Returns the values in <b>list</b> without the elements that the truth
        test (<b>iterator</b>) passes. The opposite of <b>select</b>.
      </p>
      <pre>
var odds = _.reject([1, 2, 3, 4, 5, 6], function(num){ return num % 2 == 0; });
=&gt; [1, 3, 5]
</pre>

      <p id="all">
        <b class="header">all</b><code>_.all(list, iterator, [context])</code>
        <span class="alias">Alias: <b>every</b></span>
        <br />
        Returns <i>true</i> if all of the values in the <b>list</b> pass the <b>iterator</b>
        truth test. Delegates to the native method <b>every</b>, if present.
      </p>
      <pre>
_.all([true, 1, null, 'yes'], _.identity);
=&gt; false
</pre>

      <p id="any">
        <b class="header">any</b><code>_.any(list, [iterator], [context])</code>
        <span class="alias">Alias: <b>some</b></span>
        <br />
        Returns <i>true</i> if any of the values in the <b>list</b> pass the
        <b>iterator</b> truth test. Short-circuits and stops traversing the list
        if a true element is found. Delegates to the native method <b>some</b>,
        if present.
      </p>
      <pre>
_.any([null, 0, 'yes', false]);
=&gt; true
</pre>

      <p id="include">
        <b class="header">include</b><code>_.include(list, value)</code>
        <span class="alias">Alias: <b>contains</b></span>
        <br />
        Returns <i>true</i> if the <b>value</b> is present in the <b>list</b>, using
        <i>===</i> to test equality. Uses <b>indexOf</b> internally, if <b>list</b>
        is an Array.
      </p>
      <pre>
_.include([1, 2, 3], 3);
=&gt; true
</pre>

      <p id="invoke">
        <b class="header">invoke</b><code>_.invoke(list, methodName, [*arguments])</code>
        <br />
        Calls the method named by <b>methodName</b> on each value in the <b>list</b>.
        Any extra arguments passed to <b>invoke</b> will be forwarded on to the
        method invocation.
      </p>
      <pre>
_.invoke([[5, 1, 7], [3, 2, 1]], 'sort');
=&gt; [[1, 5, 7], [1, 2, 3]]
</pre>

      <p id="pluck">
        <b class="header">pluck</b><code>_.pluck(list, propertyName)</code>
        <br />
        A convenient version of what is perhaps the most common use-case for
        <b>map</b>: extracting a list of property values.
      </p>
      <pre>
var stooges = [{name : 'moe', age : 40}, {name : 'larry', age : 50}, {name : 'curly', age : 60}];
_.pluck(stooges, 'name');
=&gt; ["moe", "larry", "curly"]
</pre>

      <p id="max">
        <b class="header">max</b><code>_.max(list, [iterator], [context])</code>
        <br />
        Returns the maximum value in <b>list</b>. If <b>iterator</b> is passed,
        it will be used on each value to generate the criterion by which the
        value is ranked.
      </p>
      <pre>
var stooges = [{name : 'moe', age : 40}, {name : 'larry', age : 50}, {name : 'curly', age : 60}];
_.max(stooges, function(stooge){ return stooge.age; });
=&gt; {name : 'curly', age : 60};
</pre>

      <p id="min">
        <b class="header">min</b><code>_.min(list, [iterator], [context])</code>
        <br />
        Returns the minimum value in <b>list</b>. If <b>iterator</b> is passed,
        it will be used on each value to generate the criterion by which the
        value is ranked.
      </p>
      <pre>
var numbers = [10, 5, 100, 2, 1000];
_.min(numbers);
=&gt; 2
</pre>

      <p id="sortBy">
        <b class="header">sortBy</b><code>_.sortBy(list, iterator, [context])</code>
        <br />
        Returns a sorted copy of <b>list</b>, ranked by the results of running 
        each value through <b>iterator</b>.
      </p>
      <pre>
_.sortBy([1, 2, 3, 4, 5, 6], function(num){ return Math.sin(num); });
=&gt; [5, 4, 6, 3, 1, 2]
</pre>

      <p id="groupBy">
        <b class="header">groupBy</b><code>_.groupBy(list, iterator)</code>
        <br />
        Splits a collection into sets, grouped by the result of running each
        value through <b>iterator</b>.
      </p>
      <pre>
_.groupBy([1.3, 2.1, 2.4], function(num){ return Math.floor(num); });
=&gt; {1: [1.3], 2: [2.1, 2.4]}
</pre>

      <p id="sortedIndex">
        <b class="header">sortedIndex</b><code>_.sortedIndex(list, value, [iterator])</code>
        <br />
        Uses a binary search to determine the index at which the <b>value</b>
        <i>should</i> be inserted into the <b>list</b> in order to maintain the <b>list</b>'s
        sorted order. If an <b>iterator</b> is passed, it will be used to compute
        the sort ranking of each value.
      </p>
      <pre>
_.sortedIndex([10, 20, 30, 40, 50], 35);
=&gt; 3
</pre>

      <p id="toArray">
        <b class="header">toArray</b><code>_.toArray(list)</code>
        <br />
        Converts the <b>list</b> (anything that can be iterated over), into a
        real Array. Useful for transmuting the <b>arguments</b> object.
      </p>
      <pre>
(function(){ return _.toArray(arguments).slice(0); })(1, 2, 3);
=&gt; [1, 2, 3]
</pre>

      <p id="size">
        <b class="header">size</b><code>_.size(list)</code>
        <br />
        Return the number of values in the <b>list</b>.
      </p>
      <pre>
_.size({one : 1, two : 2, three : 3});
=&gt; 3
</pre>

      <h2>Array Functions</h2>

      <p>
        <i>Note: All array functions will also work on the <b>arguments</b> object.</i>
      </p>

      <p id="first">
        <b class="header">first</b><code>_.first(array, [n])</code>
        <span class="alias">Alias: <b>head</b></span>
        <br />
        Returns the first element of an <b>array</b>. Passing <b>n</b> will
        return the first <b>n</b> elements of the array.
      </p>
      <pre>
_.first([5, 4, 3, 2, 1]);
=&gt; 5
</pre>

      <p id="rest">
        <b class="header">rest</b><code>_.rest(array, [index])</code>
        <span class="alias">Alias: <b>tail</b></span>
        <br />
        Returns the <b>rest</b> of the elements in an array. Pass an <b>index</b>
        to return the values of the array from that index onward.
      </p>
      <pre>
_.rest([5, 4, 3, 2, 1]);
=&gt; [4, 3, 2, 1]
</pre>

      <p id="last">
        <b class="header">last</b><code>_.last(array)</code>
        <br />
        Returns the last element of an <b>array</b>.
      </p>
      <pre>
_.last([5, 4, 3, 2, 1]);
=&gt; 1
</pre>

      <p id="compact">
        <b class="header">compact</b><code>_.compact(array)</code>
        <br />
        Returns a copy of the <b>array</b> with all falsy values removed.
        In JavaScript, <i>false</i>, <i>null</i>, <i>0</i>, <i>""</i>,
        <i>undefined</i> and <i>NaN</i> are all falsy.
      </p>
      <pre>
_.compact([0, 1, false, 2, '', 3]);
=&gt; [1, 2, 3]
</pre>

      <p id="flatten">
        <b class="header">flatten</b><code>_.flatten(array)</code>
        <br />
        Flattens a nested <b>array</b> (the nesting can be to any depth).
      </p>
      <pre>
_.flatten([1, [2], [3, [[[4]]]]]);
=&gt; [1, 2, 3, 4];
</pre>

      <p id="without">
        <b class="header">without</b><code>_.without(array, [*values])</code>
        <br />
        Returns a copy of the <b>array</b> with all instances of the <b>values</b>
        removed. <i>===</i> is used for the equality test.
      </p>
      <pre>
_.without([1, 2, 1, 0, 3, 1, 4], 0, 1);
=&gt; [2, 3, 4]
</pre>

      <p id="union">
        <b class="header">union</b><code>_.union(*arrays)</code>
        <br />
        Computes the union of the passed-in <b>arrays</b>: the list of unique items,
        in order, that are present in one or more of the <b>arrays</b>.
      </p>
      <pre>
_.union([1, 2, 3], [101, 2, 1, 10], [2, 1]);
=&gt; [1, 2, 3, 101, 10]
</pre>

      <p id="intersection">
        <b class="header">intersection</b><code>_.intersection(*arrays)</code>
        <br />
        Computes the list of values that are the intersection of all the <b>arrays</b>.
        Each value in the result is present in each of the <b>arrays</b>.
      </p>
      <pre>
_.intersection([1, 2, 3], [101, 2, 1, 10], [2, 1]);
=&gt; [1, 2]
</pre>

      <p id="difference">
        <b class="header">difference</b><code>_.difference(array, other)</code>
        <br />
        Similar to <b>without</b>, but returns the values from <b>array</b> that
        are not present in <b>other</b>.
      </p>
      <pre>
_.difference([1, 2, 3, 4, 5], [5, 2, 10]);
=&gt; [1, 3, 4]
</pre>

      <p id="uniq">
        <b class="header">uniq</b><code>_.uniq(array, [isSorted])</code>
        <span class="alias">Alias: <b>unique</b></span>
        <br />
        Produces a duplicate-free version of the <b>array</b>, using <i>===</i> to test
        object equality. If you know in advance that the <b>array</b> is sorted,
        passing <i>true</i> for <b>isSorted</b> will run a much faster algorithm.
      </p>
      <pre>
_.uniq([1, 2, 1, 3, 1, 4]);
=&gt; [1, 2, 3, 4]
</pre>

      <p id="zip">
        <b class="header">zip</b><code>_.zip(*arrays)</code>
        <br />
        Merges together the values of each of the <b>arrays</b> with the
        values at the corresponding position. Useful when you have separate
        data sources that are coordinated through matching array indexes.
        If you're working with a matrix of nested arrays, <b>zip.apply</b>
        can transpose the matrix in a similar fashion.
      </p>
      <pre>
_.zip(['moe', 'larry', 'curly'], [30, 40, 50], [true, false, false]);
=&gt; [["moe", 30, true], ["larry", 40, false], ["curly", 50, false]]
</pre>

      <p id="indexOf">
        <b class="header">indexOf</b><code>_.indexOf(array, value, [isSorted])</code>
        <br />
        Returns the index at which <b>value</b> can be found in the <b>array</b>,
        or <i>-1</i> if value is not present in the <b>array</b>. Uses the native
        <b>indexOf</b> function unless it's missing. If you're working with a 
        large array, and you know that the array is already sorted, pass <tt>true</tt>
        for <b>isSorted</b> to use a faster binary search.
      </p>
      <pre>
_.indexOf([1, 2, 3], 2);
=&gt; 1
</pre>

      <p id="lastIndexOf">
        <b class="header">lastIndexOf</b><code>_.lastIndexOf(array, value)</code>
        <br />
        Returns the index of the last occurrence of <b>value</b> in the <b>array</b>,
        or <i>-1</i> if value is not present. Uses the native <b>lastIndexOf</b>
        function if possible.
      </p>
      <pre>
_.lastIndexOf([1, 2, 3, 1, 2, 3], 2);
=&gt; 4
</pre>

      <p id="range">
        <b class="header">range</b><code>_.range([start], stop, [step])</code>
        <br />
        A function to create flexibly-numbered lists of integers, handy for
        <tt>each</tt> and <tt>map</tt> loops. <b>start</b>, if omitted, defaults
        to <i>0</i>; <b>step</b> defaults to <i>1</i>. Returns a list of integers
        from <b>start</b> to <b>stop</b>, incremented (or decremented) by <b>step</b>,
        exclusive.
      </p>
      <pre>
_.range(10);
=&gt; [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]
_.range(1, 11);
=&gt; [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
_.range(0, 30, 5);
=&gt; [0, 5, 10, 15, 20, 25]
_.range(0, -10, -1);
=&gt; [0, -1, -2, -3, -4, -5, -6, -7, -8, -9]
_.range(0);
=&gt; []
</pre>

      <h2>Function (uh, ahem) Functions</h2>

      <p id="bind">
        <b class="header">bind</b><code>_.bind(function, object, [*arguments])</code>
        <br />
        Bind a <b>function</b> to an <b>object</b>, meaning that whenever
        the function is called, the value of <i>this</i> will be the <b>object</b>.
        Optionally, bind <b>arguments</b> to the <b>function</b> to pre-fill them,
        also known as <b>currying</b>.
      </p>
      <pre>
var func = function(greeting){ return greeting + ': ' + this.name };
func = _.bind(func, {name : 'moe'}, 'hi');
func();
=&gt; 'hi: moe'
</pre>

      <p id="bindAll">
        <b class="header">bindAll</b><code>_.bindAll(object, [*methodNames])</code>
        <br />
        Binds a number of methods on the <b>object</b>, specified by
        <b>methodNames</b>, to be run in the context of that object whenever they
        are invoked. Very handy for binding functions that are going to be used
        as event handlers, which would otherwise be invoked with a fairly useless
        <i>this</i>. If no <b>methodNames</b> are provided, all of the object's
        function properties will be bound to it.
      </p>
      <pre>
var buttonView = {
  label   : 'underscore',
  onClick : function(){ alert('clicked: ' + this.label); },
  onHover : function(){ console.log('hovering: ' + this.label); }
};
_.bindAll(buttonView);
jQuery('#underscore_button').bind('click', buttonView.onClick);
=&gt; When the button is clicked, this.label will have the correct value...
</pre>

      <p id="memoize">
        <b class="header">memoize</b><code>_.memoize(function, [hashFunction])</code>
        <br />
        Memoizes a given <b>function</b> by caching the computed result. Useful
        for speeding up slow-running computations. If passed an optional 
        <b>hashFunction</b>, it will be used to compute the hash key for storing
        the result, based on the arguments to the original function. The default
        <b>hashFunction</b> just uses the first argument to the memoized function
        as the key.
      </p>
      <pre>
var fibonacci = function(n) {
  return n &lt; 2 ? n : fibonacci(n - 1) + fibonacci(n - 2);
};
var fastFibonacci = _.memoize(fibonacci);
</pre>

      <p id="delay">
        <b class="header">delay</b><code>_.delay(function, wait, [*arguments])</code>
        <br />
        Much like <b>setTimeout</b>, invokes <b>function</b> after <b>wait</b>
        milliseconds. If you pass the optional <b>arguments</b>, they will be
        forwarded on to the <b>function</b> when it is invoked.
      </p>
      <pre>
var log = _.bind(console.log, console);
_.delay(log, 1000, 'logged later');
=&gt; 'logged later' // Appears after one second.
</pre>

      <p id="defer">
        <b class="header">defer</b><code>_.defer(function)</code>
        <br />
        Defers invoking the <b>function</b> until the current call stack has cleared,
        similar to using <b>setTimeout</b> with a delay of 0. Useful for performing
        expensive computations or HTML rendering in chunks without blocking the UI thread
        from updating.
      </p>
      <pre>
_.defer(function(){ alert('deferred'); });
// Returns from the function before the alert runs.
</pre>

      <p id="throttle">
        <b class="header">throttle</b><code>_.throttle(function, wait)</code>
        <br />
        Returns a throttled version of the function, that, when invoked repeatedly, 
        will only actually call the wrapped function at most once per every <b>wait</b>
        milliseconds. Useful for rate-limiting events that occur faster than you
        can keep up with.
      </p>
      <pre>
var throttled = _.throttle(updatePosition, 100);
$(window).scroll(throttled);
</pre>

      <p id="debounce">
        <b class="header">debounce</b><code>_.debounce(function, wait)</code>
        <br />
        Repeated calls to a debounced function will postpone it's execution
        until after <b>wait</b> milliseconds have elapsed. Useful for implementing
        behavior that should only happen <i>after</i> the input has stopped arriving.
        For example: rendering a preview of a Markdown comment, recalculating a
        layout after the window has stopped being resized...
      </p>
      <pre>
var lazyLayout = _.debounce(calculateLayout, 300);
$(window).resize(lazyLayout);
</pre>

      <p id="once">
        <b class="header">once</b><code>_.once(function)</code>
        <br />
        Creates a version of the function that can only be called one time. 
        Repeated calls to the modified function will have no effect, returning
        the value from the original call. Useful for initialization functions,
        instead of having to set a boolean flag and then check it later.
      </p>
      <pre>
var initialize = _.once(createApplication);
initialize();
initialize();
// Application is only created once.
</pre>

      <p id="after">
        <b class="header">after</b><code>_.after(count, function)</code>
        <br />
        Creates a version of the function that will only be run after first 
        being called <b>count</b> times. Useful for grouping asynchronous responses,
        where you want to be sure that all the async calls have finished, before
        proceeding.
      </p>
      <pre>
var renderNotes = _.after(notes.length, render);
_.each(notes, function(note) {
  note.asyncSave({success: renderNotes}); 
});
// renderNotes is run once, after all notes have saved.
</pre>

      <p id="wrap">
        <b class="header">wrap</b><code>_.wrap(function, wrapper)</code>
        <br />
        Wraps the first <b>function</b> inside of the <b>wrapper</b> function,
        passing it as the first argument. This allows the <b>wrapper</b> to
        execute code before and after the <b>function</b> runs, adjust the arguments,
        and execute it conditionally.
      </p>
      <pre>
var hello = function(name) { return "hello: " + name; };
hello = _.wrap(hello, function(func) {
  return "before, " + func("moe") + ", after";
});
hello();
=&gt; 'before, hello: moe, after'
</pre>

      <p id="compose">
        <b class="header">compose</b><code>_.compose(*functions)</code>
        <br />
        Returns the composition of a list of <b>functions</b>, where each function
        consumes the return value of the function that follows. In math terms,
        composing the functions <i>f()</i>, <i>g()</i>, and <i>h()</i> produces
        <i>f(g(h()))</i>.
      </p>
      <pre>
var greet    = function(name){ return "hi: " + name; };
var exclaim  = function(statement){ return statement + "!"; };
var welcome = _.compose(exclaim, greet);
welcome('moe');
=&gt; 'hi: moe!'
</pre>

      <h2>Object Functions</h2>

      <p id="keys">
        <b class="header">keys</b><code>_.keys(object)</code>
        <br />
        Retrieve all the names of the <b>object</b>'s properties.
      </p>
      <pre>
_.keys({one : 1, two : 2, three : 3});
=&gt; ["one", "two", "three"]
</pre>

      <p id="values">
        <b class="header">values</b><code>_.values(object)</code>
        <br />
        Return all of the values of the <b>object</b>'s properties.
      </p>
      <pre>
_.values({one : 1, two : 2, three : 3});
=&gt; [1, 2, 3]
</pre>

      <p id="functions">
        <b class="header">functions</b><code>_.functions(object)</code>
        <span class="alias">Alias: <b>methods</b></span>
        <br />
        Returns a sorted list of the names of every method in an object &mdash;
        that is to say, the name of every function property of the object.
      </p>
      <pre>
_.functions(_);
=&gt; ["all", "any", "bind", "bindAll", "clone", "compact", "compose" ...
</pre>

      <p id="extend">
        <b class="header">extend</b><code>_.extend(destination, *sources)</code>
        <br />
        Copy all of the properties in the <b>source</b> objects over to the
        <b>destination</b> object. It's in-order, to the last source will override
        properties of the same name in previous arguments.
      </p>
      <pre>
_.extend({name : 'moe'}, {age : 50});
=&gt; {name : 'moe', age : 50}
</pre>

      <p id="defaults">
        <b class="header">defaults</b><code>_.defaults(object, *defaults)</code>
        <br />
        Fill in missing properties in <b>object</b> with default values from the
        <b>defaults</b> objects. As soon as the property is filled, further defaults
        will have no effect.
      </p>
      <pre>
var iceCream = {flavor : "chocolate"};
_.defaults(iceCream, {flavor : "vanilla", sprinkles : "lots"});
=&gt; {flavor : "chocolate", sprinkles : "lots"}
</pre>

      <p id="clone">
        <b class="header">clone</b><code>_.clone(object)</code>
        <br />
        Create a shallow-copied clone of the <b>object</b>. Any nested objects
        or arrays will be copied by reference, not duplicated.
      </p>
      <pre>
_.clone({name : 'moe'});
=&gt; {name : 'moe'};
</pre>

      <p id="tap">
        <b class="header">tap</b><code>_.tap(object, interceptor)</code>
        <br />
        Invokes <b>interceptor</b> with the <b>object</b>, and then returns <b>object</b>.
        The primary purpose of this method is to "tap into" a method chain, in order to perform operations on intermediate results within the chain.
      </p>
      <pre>
_([1,2,3,200]).chain().
  select(function(num) { return num % 2 == 0; }).
  tap(console.log).
  map(function(num) { return num * num }).
  value();
=&gt; [2, 200]
=&gt; [4, 40000]
</pre>

      <p id="isEqual">
        <b class="header">isEqual</b><code>_.isEqual(object, other)</code>
        <br />
        Performs an optimized deep comparison between the two objects, to determine
        if they should be considered equal.
      </p>
      <pre>
var moe   = {name : 'moe', luckyNumbers : [13, 27, 34]};
var clone = {name : 'moe', luckyNumbers : [13, 27, 34]};
moe == clone;
=&gt; false
_.isEqual(moe, clone);
=&gt; true
</pre>

      <p id="isEmpty">
        <b class="header">isEmpty</b><code>_.isEmpty(object)</code>
        <br />
        Returns <i>true</i> if <b>object</b> contains no values.
      </p>
      <pre>
_.isEmpty([1, 2, 3]);
=&gt; false
_.isEmpty({});
=&gt; true
</pre>

      <p id="isElement">
        <b class="header">isElement</b><code>_.isElement(object)</code>
        <br />
        Returns <i>true</i> if <b>object</b> is a DOM element.
      </p>
      <pre>
_.isElement(jQuery('body')[0]);
=&gt; true
</pre>

      <p id="isArray">
        <b class="header">isArray</b><code>_.isArray(object)</code>
        <br />
        Returns <i>true</i> if <b>object</b> is an Array.
      </p>
      <pre>
(function(){ return _.isArray(arguments); })();
=&gt; false
_.isArray([1,2,3]);
=&gt; true
</pre>

      <p id="isArguments">
        <b class="header">isArguments</b><code>_.isArguments(object)</code>
        <br />
        Returns <i>true</i> if <b>object</b> is an Arguments object.
      </p>
      <pre>
(function(){ return _.isArguments(arguments); })(1, 2, 3);
=&gt; true
_.isArguments([1,2,3]);
=&gt; false
</pre>

      <p id="isFunction">
        <b class="header">isFunction</b><code>_.isFunction(object)</code>
        <br />
        Returns <i>true</i> if <b>object</b> is a Function.
      </p>
      <pre>
_.isFunction(alert);
=&gt; true
</pre>

      <p id="isString">
        <b class="header">isString</b><code>_.isString(object)</code>
        <br />
        Returns <i>true</i> if <b>object</b> is a String.
      </p>
      <pre>
_.isString("moe");
=&gt; true
</pre>

      <p id="isNumber">
        <b class="header">isNumber</b><code>_.isNumber(object)</code>
        <br />
        Returns <i>true</i> if <b>object</b> is a Number.
      </p>
      <pre>
_.isNumber(8.4 * 5);
=&gt; true
</pre>

      <p id="isBoolean">
        <b class="header">isBoolean</b><code>_.isBoolean(object)</code>
        <br />
        Returns <i>true</i> if <b>object</b> is either <i>true</i> or <i>false</i>.
      </p>
      <pre>
_.isBoolean(null);
=&gt; false
</pre>

      <p id="isDate">
        <b class="header">isDate</b><code>_.isDate(object)</code>
        <br />
        Returns <i>true</i> if <b>object</b> is a Date.
      </p>
      <pre>
_.isDate(new Date());
=&gt; true
</pre>

      <p id="isRegExp">
        <b class="header">isRegExp</b><code>_.isRegExp(object)</code>
        <br />
        Returns <i>true</i> if <b>object</b> is a RegExp.
      </p>
      <pre>
_.isRegExp(/moe/);
=&gt; true
</pre>

      <p id="isNaN">
        <b class="header">isNaN</b><code>_.isNaN(object)</code>
        <br />
        Returns <i>true</i> if <b>object</b> is <i>NaN</i>.<br /> Note: this is not
        the same as the native <b>isNaN</b> function, which will also return
        true if the variable is <i>undefined</i>.
      </p>
      <pre>
_.isNaN(NaN);
=&gt; true
isNaN(undefined);
=&gt; true
_.isNaN(undefined);
=&gt; false
</pre>

      <p id="isNull">
        <b class="header">isNull</b><code>_.isNull(object)</code>
        <br />
        Returns <i>true</i> if the value of <b>object</b> is <i>null</i>.
      </p>
      <pre>
_.isNull(null);
=&gt; true
_.isNull(undefined);
=&gt; false
</pre>

      <p id="isUndefined">
        <b class="header">isUndefined</b><code>_.isUndefined(variable)</code>
        <br />
        Returns <i>true</i> if <b>variable</b> is <i>undefined</i>.
      </p>
      <pre>
_.isUndefined(window.missingVariable);
=&gt; true
</pre>

      <h2>Utility Functions</h2>

      <p id="noConflict">
        <b class="header">noConflict</b><code>_.noConflict()</code>
        <br />
        Give control of the "_" variable back to its previous owner. Returns
        a reference to the <b>Underscore</b> object.
      </p>
      <pre>
var underscore = _.noConflict();</pre>

      <p id="identity">
        <b class="header">identity</b><code>_.identity(value)</code>
        <br />
        Returns the same value that is used as the argument. In math:
        <tt>f(x) = x</tt><br />
        This function looks useless, but is used throughout Underscore as
        a default iterator.
      </p>
      <pre>
var moe = {name : 'moe'};
moe === _.identity(moe);
=&gt; true</pre>

      <p id="times">
        <b class="header">times</b><code>_.times(n, iterator)</code>
        <br />
        Invokes the given iterator function <b>n</b> times.
      </p>
      <pre>
_(3).times(function(){ genie.grantWish(); });</pre>

      <p id="mixin">
        <b class="header">mixin</b><code>_.mixin(object)</code>
        <br />
        Allows you to extend Underscore with your own utility functions. Pass
        a hash of <tt>{name: function}</tt> definitions to have your functions 
        added to the Underscore object, as well as the OOP wrapper.
      </p>
      <pre>
_.mixin({
  capitalize : function(string) {
    return string.charAt(0).toUpperCase() + string.substring(1).toLowerCase();
  }
});
_("fabio").capitalize();
=&gt; "Fabio"
</pre>

      <p id="uniqueId">
        <b class="header">uniqueId</b><code>_.uniqueId([prefix])</code>
        <br />
        Generate a globally-unique id for client-side models or DOM elements
        that need one. If <b>prefix</b> is passed, the id will be appended to it.
      </p>
      <pre>
_.uniqueId('contact_');
=&gt; 'contact_104'</pre>

      <p id="template">
        <b class="header">template</b><code>_.template(templateString, [context])</code>
        <br />
        Compiles JavaScript templates into functions that can be evaluated
        for rendering. Useful for rendering complicated bits of HTML from JSON
        data sources. Template functions can both interpolate variables, using<br />
        <tt>&lt;%= &hellip; %&gt;</tt>, as well as execute arbitrary JavaScript code, with
        <tt>&lt;% &hellip; %&gt;</tt>. When you evaluate a template function, pass in a
        <b>context</b> object that has properties corresponding to the template's free
        variables. If you're writing a one-off, you can pass the <b>context</b>
        object as the second parameter to <b>template</b> in order to render
        immediately instead of returning a template function.
      </p>
      <pre>
var compiled = _.template("hello: &lt;%= name %&gt;");
compiled({name : 'moe'});
=&gt; "hello: moe"

var list = "&lt;% _.each(people, function(name) { %&gt; &lt;li&gt;&lt;%= name %&gt;&lt;/li&gt; &lt;% }); %&gt;";
_.template(list, {people : ['moe', 'curly', 'larry']});
=&gt; "&lt;li&gt;moe&lt;/li&gt;&lt;li&gt;curly&lt;/li&gt;&lt;li&gt;larry&lt;/li&gt;"</pre>

      <p>
        You can also use <tt>print</tt> from within JavaScript code.  This is
        sometimes more convenient than using <tt>&lt;%= ... %&gt;</tt>.
      </p>
      
      <pre>
var compiled = _.template("&lt;% print('Hello ' + epithet); %&gt;");
compiled({epithet: "stooge"});
=&gt; "Hello stooge."</pre>

      <p>
        If ERB-style delimiters aren't your cup of tea, you can change Underscore's
        template settings to use different symbols to set off interpolated code.
        Define an <b>interpolate</b> regex, and an (optional) <b>evaluate</b> regex
        to match expressions that should be inserted and evaluated, respectively.
        If no <b>evaluate</b> regex is provided, your templates will only be 
        capable of interpolating values. 
        For example, to perform
        <a href="http://github.com/janl/mustache.js#readme">Mustache.js</a>
        style templating:
      </p>

      <pre>
_.templateSettings = {
  interpolate : /\{\{(.+?)\}\}/g
};

var template = _.template("Hello {{ name }}!");
template({name : "Mustache"});
=&gt; "Hello Mustache!"</pre>

      <h2>Chaining</h2>

      <p id="chain">
        <b class="header">chain</b><code>_(obj).chain()</code>
        <br />
        Returns a wrapped object. Calling methods on this object will continue
        to return wrapped objects until <tt>value</tt> is used. (
        <a href="#styles">A more realistic example.</a>)
      </p>
      <pre>
var stooges = [{name : 'curly', age : 25}, {name : 'moe', age : 21}, {name : 'larry', age : 23}];
var youngest = _(stooges).chain()
  .sortBy(function(stooge){ return stooge.age; })
  .map(function(stooge){ return stooge.name + ' is ' + stooge.age; })
  .first()
  .value();
=&gt; "moe is 21"
</pre>

      <p id="value">
        <b class="header">value</b><code>_(obj).value()</code>
        <br />
        Extracts the value of a wrapped object.
      </p>
      <pre>
_([1, 2, 3]).value();
=&gt; [1, 2, 3]
</pre>

      <h2 id="duck_typing">Duck Typing</h2>

      <p>
        The <b>isType</b> (<tt>isArray</tt>, <tt>isFunction</tt>, <tt>isString</tt> ...) family of type-checking
        functions use property detection to do their work, which, although
        orders of magnitude faster than the alternative, isn't entirely safe when dealing
        with objects that are used as hashes, where arbitrary strings are being
        set for the keys. It's entirely possible for an object to masquerade as
        another type, if you're setting properties with names like "concat" and
        "charCodeAt". So be aware.
      </p>


      <h2>Links &amp; Suggested Reading</h2>

      <p>
        <a href="http://mirven.github.com/underscore.lua/">Underscore.lua</a>,
        a Lua port of the functions that are applicable in both languages.
        Includes OOP-wrapping and chaining.
        The <a href="http://github.com/mirven/underscore.lua">source</a> is
        available on GitHub.
      </p>
      
      <p>
        <a href="https://github.com/edtsech/underscore.string">Underscore.string</a>,
        an Underscore extension that adds functions for string-manipulation: 
        <tt>trim</tt>, <tt>startsWith</tt>, <tt>contains</tt>, <tt>capitalize</tt>,
        <tt>reverse</tt>, <tt>sprintf</tt>, and more.
      </p>

      <p>
        Ruby's <a href="http://ruby-doc.org/core/classes/Enumerable.html">Enumerable</a> module.
      </p>

      <p>
        <a href="http://www.prototypejs.org/">Prototype.js</a>, which provides
        JavaScript with collection functions in the manner closest to Ruby's Enumerable.
      </p>

      <p>
        Oliver Steele's
        <a href="http://osteele.com/sources/javascript/functional/">Functional JavaScript</a>,
        which includes comprehensive higher-order function support as well as string lambdas.
      </p>
      
      <p>
        Michael Aufreiter's <a href="http://substance.io/#michael/data-js">Data.js</a>,
        a data manipulation + persistence library for JavaScript.
      </p>

      <p>
        Python's <a href="http://docs.python.org/library/itertools.html">itertools</a>.
      </p>

      <h2 id="changelog">Change Log</h2>
      
      <p>
        <b class="header">1.1.7</b> &mdash; <small><i>July 13, 2011</i></small><br />
        Added <tt>_.groupBy</tt>, which aggregates a collection into groups of like items.
        Added <tt>_.union</tt> and <tt>_.difference</tt>, to complement the 
        (re-named) <tt>_.intersection</tt>.
        Various improvements for support of sparse arrays.
        <tt>_.toArray</tt> now returns a clone, if directly passed an array.
        <tt>_.functions</tt> now also returns the names of functions that are present
        in the prototype chain.
      </p>
            
      <p>
        <b class="header">1.1.6</b> &mdash; <small><i>April 18, 2011</i></small><br />
        Added <tt>_.after</tt>, which will return a function that only runs after
        first being called a specified number of times.
        <tt>_.invoke</tt> can now take a direct function reference.
        <tt>_.every</tt> now requires an iterator function to be passed, which
        mirrors the ECMA5 API.
        <tt>_.extend</tt> no longer copies keys when the value is undefined.
        <tt>_.bind</tt> now errors when trying to bind an undefined value.
      </p>
      
      <p>
        <b class="header">1.1.5</b> &mdash; <small><i>Mar 20, 2011</i></small><br />
        Added an <tt>_.defaults</tt> function, for use merging together JS objects
        representing default options.
        Added an <tt>_.once</tt> function, for manufacturing functions that should
        only ever execute a single time.
        <tt>_.bind</tt> now delegates to the native ECMAScript 5 version, 
        where available.
        <tt>_.keys</tt> now throws an error when used on non-Object values, as in
        ECMAScript 5.
        Fixed a bug with <tt>_.keys</tt> when used over sparse arrays.
      </p>
      
      <p>
        <b class="header">1.1.4</b> &mdash; <small><i>Jan 9, 2011</i></small><br />
        Improved compliance with ES5's Array methods when passing <tt>null</tt> 
        as a value. <tt>_.wrap</tt> now correctly sets <tt>this</tt> for the
        wrapped function. <tt>_.indexOf</tt> now takes an optional flag for
        finding the insertion index in an array that is guaranteed to already
        be sorted. Avoiding the use of <tt>.callee</tt>, to allow <tt>_.isArray</tt>
        to work properly in ES5's strict mode.
      </p>
      
      <p>
        <b class="header">1.1.3</b> &mdash; <small><i>Dec 1, 2010</i></small><br />
        In CommonJS, Underscore may now be required with just: <br />
        <tt>var _ = require("underscore")</tt>.
        Added <tt>_.throttle</tt> and <tt>_.debounce</tt> functions.
        Removed <tt>_.breakLoop</tt>, in favor of an ECMA5-style un-<i>break</i>-able
        each implementation &mdash; this removes the try/catch, and you'll now have
        better stack traces for exceptions that are thrown within an Underscore iterator.
        Improved the <b>isType</b> family of functions for better interoperability
        with Internet Explorer host objects.
        <tt>_.template</tt> now correctly escapes backslashes in templates.
        Improved <tt>_.reduce</tt> compatibility with the ECMA5 version: 
        if you don't pass an initial value, the first item in the collection is used.
        <tt>_.each</tt> no longer returns the iterated collection, for improved
        consistency with ES5's <tt>forEach</tt>.
      </p>
      
      <p>
        <b class="header">1.1.2</b><br />
        Fixed <tt>_.contains</tt>, which was mistakenly pointing at 
        <tt>_.intersect</tt> instead of <tt>_.include</tt>, like it should 
        have been. Added <tt>_.unique</tt> as an alias for <tt>_.uniq</tt>.
      </p>
      
      <p>
        <b class="header">1.1.1</b><br />
        Improved the speed of <tt>_.template</tt>, and its handling of multiline
        interpolations. Ryan Tenney contributed optimizations to many Underscore 
        functions. An annotated version of the source code is now available.
      </p>
      
      <p>
        <b class="header">1.1.0</b><br />
        The method signature of <tt>_.reduce</tt> has been changed to match
        the ECMAScript 5 signature, instead of the Ruby/Prototype.js version.
        This is a backwards-incompatible change. <tt>_.template</tt> may now be
        called with no arguments, and preserves whitespace. <tt>_.contains</tt>
        is a new alias for <tt>_.include</tt>.
      </p>
      
      <p>
        <b class="header">1.0.4</b><br />
        <a href="http://themoell.com/">Andri Möll</a> contributed the <tt>_.memoize</tt> 
        function, which can be used to speed up expensive repeated computations 
        by caching the results.
      </p>
      
      <p>
        <b class="header">1.0.3</b><br />
        Patch that makes <tt>_.isEqual</tt> return <tt>false</tt> if any property
        of the compared object has a <tt>NaN</tt> value. Technically the correct
        thing to do, but of questionable semantics. Watch out for NaN comparisons.
      </p>
      
      <p>
        <b class="header">1.0.2</b><br />
        Fixes <tt>_.isArguments</tt> in recent versions of Opera, which have
        arguments objects as real Arrays.
      </p>
      
      <p>
        <b class="header">1.0.1</b><br />
        Bugfix for <tt>_.isEqual</tt>, when comparing two objects with the same 
        number of undefined keys, but with different names.
      </p>
      
      <p>
        <b class="header">1.0.0</b><br />
        Things have been stable for many months now, so Underscore is now
        considered to be out of beta, at <b>1.0</b>. Improvements since <b>0.6</b>
        include <tt>_.isBoolean</tt>, and the ability to have <tt>_.extend</tt>
        take multiple source objects.
      </p>
      
      <p>
        <b class="header">0.6.0</b><br />
        Major release. Incorporates a number of 
        <a href="http://github.com/ratbeard">Mile Frawley's</a> refactors for
        safer duck-typing on collection functions, and cleaner internals. A new
        <tt>_.mixin</tt> method that allows you to extend Underscore with utility
        functions of your own. Added <tt>_.times</tt>, which works the same as in 
        Ruby or Prototype.js. Native support for ECMAScript 5's <tt>Array.isArray</tt>, 
        and <tt>Object.keys</tt>.
      </p>

      <p>
        <b class="header">0.5.8</b><br />
        Fixed Underscore's collection functions to work on
        <a href="https://developer.mozilla.org/En/DOM/NodeList">NodeLists</a> and
        <a href="https://developer.mozilla.org/En/DOM/HTMLCollection">HTMLCollections</a>
        once more, thanks to
        <a href="http://github.com/jmtulloss">Justin Tulloss</a>.
      </p>

      <p>
        <b class="header">0.5.7</b><br />
        A safer implementation of <tt>_.isArguments</tt>, and a
        faster <tt>_.isNumber</tt>,<br />thanks to
        <a href="http://jedschmidt.com/">Jed Schmidt</a>.
      </p>

      <p>
        <b class="header">0.5.6</b><br />
        Customizable delimiters for <tt>_.template</tt>, contributed by
        <a href="http://github.com/iamnoah">Noah Sloan</a>.
      </p>

      <p>
        <b class="header">0.5.5</b><br />
        Fix for a bug in MobileSafari's OOP-wrapper, with the arguments object.
      </p>

      <p>
        <b class="header">0.5.4</b><br />
        Fix for multiple single quotes within a template string for
        <tt>_.template</tt>. See:
        <a href="http://www.west-wind.com/Weblog/posts/509108.aspx">Rick Strahl's blog post</a>.
      </p>

      <p>
        <b class="header">0.5.2</b><br />
        New implementations of <tt>isArray</tt>, <tt>isDate</tt>, <tt>isFunction</tt>,
        <tt>isNumber</tt>, <tt>isRegExp</tt>, and <tt>isString</tt>, thanks to
        a suggestion from
        <a href="http://www.broofa.com/">Robert Kieffer</a>.
        Instead of doing <tt>Object#toString</tt>
        comparisons, they now check for expected properties, which is less safe,
        but more than an order of magnitude faster. Most other Underscore
        functions saw minor speed improvements as a result.
        <a href="http://dolzhenko.org/">Evgeniy Dolzhenko</a>
        contributed <tt>_.tap</tt>,
        <a href="http://ruby-doc.org/core-1.9/classes/Object.html#M000191">similar to Ruby 1.9's</a>,
        which is handy for injecting side effects (like logging) into chained calls.
      </p>

      <p>
        <b class="header">0.5.1</b><br />
        Added an <tt>_.isArguments</tt> function. Lots of little safety checks
        and optimizations contributed by
        <a href="http://github.com/iamnoah/">Noah Sloan</a> and 
        <a href="http://themoell.com/">Andri Möll</a>.
      </p>

      <p>
        <b class="header">0.5.0</b><br />
        <b>[API Changes]</b> <tt>_.bindAll</tt> now takes the context object as
        its first parameter. If no method names are passed, all of the context
        object's methods are bound to it, enabling chaining and easier binding.
        <tt>_.functions</tt> now takes a single argument and returns the names
        of its Function properties. Calling <tt>_.functions(_)</tt> will get you
        the previous behavior.
        Added <tt>_.isRegExp</tt> so that <tt>isEqual</tt> can now test for RegExp equality.
        All of the "is" functions have been shrunk down into a single definition.
        <a href="http://github.com/grayrest/">Karl Guertin</a> contributed patches.
      </p>

      <p>
        <b class="header">0.4.7</b><br />
        Added <tt>isDate</tt>, <tt>isNaN</tt>, and <tt>isNull</tt>, for completeness.
        Optimizations for <tt>isEqual</tt> when checking equality between Arrays
        or Dates. <tt>_.keys</tt> is now <small><i><b>25%&ndash;2X</b></i></small> faster (depending on your
        browser) which speeds up the functions that rely on it, such as <tt>_.each</tt>.
      </p>

      <p>
        <b class="header">0.4.6</b><br />
        Added the <tt>range</tt> function, a port of the
        <a href="http://docs.python.org/library/functions.html#range">Python
        function of the same name</a>, for generating flexibly-numbered lists
        of integers. Original patch contributed by
        <a href="http://github.com/kylichuku">Kirill Ishanov</a>.
      </p>

      <p>
        <b class="header">0.4.5</b><br />
        Added <tt>rest</tt> for Arrays and arguments objects, and aliased
        <tt>first</tt> as <tt>head</tt>, and <tt>rest</tt> as <tt>tail</tt>,
        thanks to <a href="http://github.com/lukesutton/">Luke Sutton</a>'s patches.
        Added tests ensuring that all Underscore Array functions also work on
        <i>arguments</i> objects.
      </p>

      <p>
        <b class="header">0.4.4</b><br />
        Added <tt>isString</tt>, and <tt>isNumber</tt>, for consistency. Fixed
        <tt>_.isEqual(NaN, NaN)</tt> to return <i>true</i> (which is debatable).
      </p>

      <p>
        <b class="header">0.4.3</b><br />
        Started using the native <tt>StopIteration</tt> object in browsers that support it.
        Fixed Underscore setup for CommonJS environments.
      </p>

      <p>
        <b class="header">0.4.2</b><br />
        Renamed the unwrapping function to <tt>value</tt>, for clarity.
      </p>

      <p>
        <b class="header">0.4.1</b><br />
        Chained Underscore objects now support the Array prototype methods, so
        that you can perform the full range of operations on a wrapped array
        without having to break your chain. Added a <tt>breakLoop</tt> method
        to <b>break</b> in the middle of any Underscore iteration. Added an
        <tt>isEmpty</tt> function that works on arrays and objects.
      </p>

      <p>
        <b class="header">0.4.0</b><br />
        All Underscore functions can now be called in an object-oriented style,
        like so: <tt>_([1, 2, 3]).map(...);</tt>. Original patch provided by
        <a href="http://macournoyer.com/">Marc-André Cournoyer</a>.
        Wrapped objects can be chained through multiple
        method invocations. A <a href="#functions"><tt>functions</tt></a> method
        was added, providing a sorted list of all the functions in Underscore.
      </p>

      <p>
        <b class="header">0.3.3</b><br />
        Added the JavaScript 1.8 function <tt>reduceRight</tt>. Aliased it
        as <tt>foldr</tt>, and aliased <tt>reduce</tt> as <tt>foldl</tt>.
      </p>

      <p>
        <b class="header">0.3.2</b><br />
        Now runs on stock <a href="http://www.mozilla.org/rhino/">Rhino</a>
        interpreters with: <tt>load("underscore.js")</tt>.
        Added <a href="#identity"><tt>identity</tt></a> as a utility function.
      </p>

      <p>
        <b class="header">0.3.1</b><br />
        All iterators are now passed in the original collection as their third
        argument, the same as JavaScript 1.6's <b>forEach</b>. Iterating over
        objects is now called with <tt>(value, key, collection)</tt>, for details
        see <a href="#each"><tt>_.each</tt></a>.
      </p>

      <p>
        <b class="header">0.3.0</b><br />
        Added <a href="http://github.com/dmitryBaranovskiy">Dmitry Baranovskiy</a>'s
        comprehensive optimizations, merged in
        <a href="http://github.com/kriskowal/">Kris Kowal</a>'s patches to make Underscore
        <a href="http://wiki.commonjs.org/wiki/CommonJS">CommonJS</a> and
        <a href="http://narwhaljs.org/">Narwhal</a> compliant.
      </p>

      <p>
        <b class="header">0.2.0</b><br />
        Added <tt>compose</tt> and <tt>lastIndexOf</tt>, renamed <tt>inject</tt> to
        <tt>reduce</tt>, added aliases for <tt>inject</tt>, <tt>filter</tt>,
        <tt>every</tt>, <tt>some</tt>, and <tt>forEach</tt>.
      </p>

      <p>
        <b class="header">0.1.1</b><br />
        Added <tt>noConflict</tt>, so that the "Underscore" object can be assigned to
        other variables.
      </p>

      <p>
        <b class="header">0.1.0</b><br />
        Initial release of Underscore.js.
      </p>

      <p>
        <a href="http://documentcloud.org/" title="A DocumentCloud Project" style="background:none;">
          <img src="http://jashkenas.s3.amazonaws.com/images/a_documentcloud_project.png" alt="A DocumentCloud Project" />
        </a>
      </p>

    </div>

  </div>

  <!-- Include Underscore, so you can play with it in the console. -->
  <script type="text/javascript" src="underscore.js"></script>

</body>
</html>