Add optimizations for large arrays to _.difference, _.intersection, and _.without.

Former-commit-id: 26d55a6a3340e77b5269b2003d20def3fe77bca9

lodash.js

@@ -439,6 +439,41 @@
 
   /*--------------------------------------------------------------------------*/
 
+  /**
+   * Creates a new function optimized for searching large arrays for a given `value`,
+   * starting at `fromIndex`, using strict equality for comparisons, i.e. `===`.
+   *
+   * @private
+   * @param {Array} array The array to search.
+   * @param {Mixed} value The value to search for.
+   * @param {Number} [fromIndex=0] The index to start searching from.
+   * @param {Number} [largeSize=30] The length at which an array is considered large.
+   * @returns {Boolean} Returns `true` if `value` is found, else `false`.
+   */
+  function cachedContains(array, fromIndex, largeSize) {
+    fromIndex || (fromIndex = 0);
+
+    var length = array.length,
+        isLarge = (length - fromIndex) >= (largeSize || 30),
+        cache = isLarge ? {} : array;
+
+    if (isLarge) {
+      // init value cache
+      var value,
+          index = fromIndex - 1;
+
+      while (++index < length) {
+        value = array[index];
+        (hasOwnProperty.call(cache, value) ? cache[value] : (cache[value] = [])).push(value);
+      }
+    }
+    return function(value) {
+      return isLarge
+        ? hasOwnProperty.call(cache, value) && indexOf(cache[value], value) > -1
+        : indexOf(cache, value, fromIndex) > -1;
+    }
+  }
+
   /**
    * Creates compiled iteration functions. The iteration function will be created
    * to iterate over only objects if the first argument of `options.args` is
@@ -1224,10 +1259,11 @@
     }
     var index = -1,
         length = array.length,
-        flattened = concat.apply(result, arguments);
+        flattened = concat.apply(result, arguments),
+        contains = cachedContains(flattened, length);
 
     while (++index < length) {
-      if (indexOf(flattened, array[index], length) < 0) {
+      if (!contains(array[index])) {
         result.push(array[index]);
       }
     }
@@ -1390,12 +1426,15 @@
     var value,
         index = -1,
         length = array.length,
-        others = slice.call(arguments, 1);
+        others = slice.call(arguments, 1),
+        cache = [];
 
     while (++index < length) {
       value = array[index];
       if (indexOf(result, value) < 0 &&
-          every(others, function(other) { return indexOf(other, value) > -1; })) {
+          every(others, function(other, index) {
+            return (cache[index] || (cache[index] = cachedContains(other)))(value);
+          })) {
         result.push(value);
       }
     }
@@ -1847,10 +1886,11 @@
       return result;
     }
     var index = -1,
-        length = array.length;
+        length = array.length,
+        contains = cachedContains(arguments, 1, 20);
 
     while (++index < length) {
-      if (indexOf(arguments, array[index], 1) < 0) {
+      if (!contains(array[index])) {
         result.push(array[index]);
       }
     }
@@ -2774,7 +2814,8 @@
           }
         }
       }
-    } else {
+    }
+    else {
       // objects with different constructors are not equivalent
       if ('constructor' in a != 'constructor' in b || a.constructor != b.constructor) {
         return false;

perf/perf.js

@@ -95,7 +95,6 @@
           belt = this.name == 'Lo-Dash' ? lodash : _;
 
       var index,
-          length,
           limit = 20,
           object = {},
           objects = Array(limit),
@@ -104,7 +103,7 @@
           nestedNumbers = [1, [2], [3, [[4]]]],
           twoNumbers = [12, 21];
 
-      for (index = 0, length = limit; index < length; index++) {
+      for (index = 0; index < limit; index++) {
         numbers[index] = index;
         object['key' + index] = index;
         objects[index] = { 'num': index };
@@ -132,7 +131,7 @@
             funcNames = belt.functions(lodash);
 
         // potentially expensive
-        for (index = 0, length = this.count; index < length; index++) {
+        for (index = 0; index < this.count; index++) {
           bindAllObjects[index] = belt.clone(lodash);
         }
       }
@@ -187,13 +186,50 @@
             numbers2 = Array(limit),
             nestedNumbers2 = [1, [2], [3, [[4]]]];
 
-        for (index = 0, length = limit; index < length; index++) {
+        for (index = 0; index < limit; index++) {
           numbers2[index] = index;
           object2['key' + index] = index;
           objects2[index] = { 'num': index };
         }
       }
 
+      if (typeof multiArrays != 'undefined') {
+        var twentyFiveValues = Array(25),
+            twentyFiveValues2 = Array(25),
+            fiftyValues = Array(50),
+            fiftyValues2 = Array(50),
+            seventyFiveValues = Array(75),
+            seventyFiveValues2 = Array(75),
+            lowerChars = 'abcdefghijklmnopqrstuvwxyz'.split(''),
+            upperChars = 'ABCDEFGHIJKLMNOPQRSTUVWXYZ'.split('');
+
+        for (index = 0; index < 75; index++) {
+          if (index < 26) {
+            if (index < 20) {
+              twentyFiveValues[index] = lowerChars[index];
+              twentyFiveValues2[index] = upperChars[index];
+            }
+            else if (index < 25) {
+              twentyFiveValues[index] =
+              twentyFiveValues2[index] = index;
+            }
+            fiftyValues[index] =
+            seventyFiveValues[index] = lowerChars[index];
+
+            fiftyValues2[index] =
+            seventyFiveValues2[index] = upperChars[index];
+          }
+          else {
+            if (index < 50) {
+              fiftyValues[index] = index;
+              fiftyValues2[index] = index + (index < 40 ? 75 : 0);
+            }
+            seventyFiveValues[index] = index;
+            seventyFiveValues2[index] = index + (index < 60 ? 75 : 0);
+          }
+        }
+      }
+
       if (typeof template != 'undefined') {
         var tplData = {
           'header1': 'Header1',
@@ -489,6 +525,54 @@
       })
   );
 
+  suites.push(
+    Benchmark.Suite('`_.difference` iterating 25 elements')
+      .add('Lo-Dash', {
+        'fn': function() {
+          lodash.difference(twentyFiveValues, twentyFiveValues2);
+        },
+        'teardown': 'function multiArrays(){}'
+      })
+      .add('Underscore', {
+        'fn': function() {
+          _.difference(twentyFiveValues, twentyFiveValues2);
+        },
+        'teardown': 'function multiArrays(){}'
+      })
+  );
+
+  suites.push(
+    Benchmark.Suite('`_.difference` iterating 50 elements')
+      .add('Lo-Dash', {
+        'fn': function() {
+          lodash.difference(fiftyValues, fiftyValues2);
+        },
+        'teardown': 'function multiArrays(){}'
+      })
+      .add('Underscore', {
+        'fn': function() {
+          _.difference(fiftyValues, fiftyValues2);
+        },
+        'teardown': 'function multiArrays(){}'
+      })
+  );
+
+  suites.push(
+    Benchmark.Suite('`_.difference` iterating 75 elements')
+      .add('Lo-Dash', {
+        'fn': function() {
+          lodash.difference(seventyFiveValues, seventyFiveValues2);
+        },
+        'teardown': 'function multiArrays(){}'
+      })
+      .add('Underscore', {
+        'fn': function() {
+          _.difference(seventyFiveValues, seventyFiveValues2);
+        },
+        'teardown': 'function multiArrays(){}'
+      })
+  );
+
   /*--------------------------------------------------------------------------*/
 
   suites.push(
@@ -725,6 +809,54 @@
       })
   );
 
+  suites.push(
+    Benchmark.Suite('`_.intersection` iterating 25 elements')
+      .add('Lo-Dash', {
+        'fn': function() {
+          lodash.intersection(twentyFiveValues, twentyFiveValues2);
+        },
+        'teardown': 'function multiArrays(){}'
+      })
+      .add('Underscore', {
+        'fn': function() {
+          _.intersection(twentyFiveValues, twentyFiveValues2);
+        },
+        'teardown': 'function multiArrays(){}'
+      })
+  );
+
+  suites.push(
+    Benchmark.Suite('`_.intersection` iterating 50 elements')
+      .add('Lo-Dash', {
+        'fn': function() {
+          lodash.intersection(fiftyValues, fiftyValues2);
+        },
+        'teardown': 'function multiArrays(){}'
+      })
+      .add('Underscore', {
+        'fn': function() {
+          _.intersection(fiftyValues, fiftyValues2);
+        },
+        'teardown': 'function multiArrays(){}'
+      })
+  );
+
+  suites.push(
+    Benchmark.Suite('`_.intersection` iterating 75 elements')
+      .add('Lo-Dash', {
+        'fn': function() {
+          lodash.intersection(seventyFiveValues, seventyFiveValues2);
+        },
+        'teardown': 'function multiArrays(){}'
+      })
+      .add('Underscore', {
+        'fn': function() {
+          _.intersection(seventyFiveValues, seventyFiveValues2);
+        },
+        'teardown': 'function multiArrays(){}'
+      })
+  );
+
   /*--------------------------------------------------------------------------*/
 
   suites.push(
@@ -759,7 +891,6 @@
 
   /*--------------------------------------------------------------------------*/
 
-
   suites.push(
     Benchmark.Suite('`_.isEqual` comparing primitives and objects (edge case)')
       .add('Lo-Dash', {
@@ -1218,6 +1349,54 @@
       })
   );
 
+  suites.push(
+    Benchmark.Suite('`_.union` iterating an array of 25 elements')
+      .add('Lo-Dash', {
+        'fn': function() {
+          lodash.union(twentyFiveValues, twentyFiveValues2);
+        },
+        'teardown': 'function multiArrays(){}'
+      })
+      .add('Underscore', {
+        'fn': function() {
+          _.union(twentyFiveValues, twentyFiveValues2);
+        },
+        'teardown': 'function multiArrays(){}'
+      })
+  );
+
+  suites.push(
+    Benchmark.Suite('`_.union` iterating an array of 50 elements')
+      .add('Lo-Dash', {
+        'fn': function() {
+          lodash.union(fiftyValues, fiftyValues2);
+        },
+        'teardown': 'function multiArrays(){}'
+      })
+      .add('Underscore', {
+        'fn': function() {
+          _.union(fiftyValues, fiftyValues2);
+        },
+        'teardown': 'function multiArrays(){}'
+      })
+  );
+
+  suites.push(
+    Benchmark.Suite('`_.union` iterating an array of 75 elements')
+      .add('Lo-Dash', {
+        'fn': function() {
+          lodash.union(seventyFiveValues, seventyFiveValues2);
+        },
+        'teardown': 'function multiArrays(){}'
+      })
+      .add('Underscore', {
+        'fn': function() {
+          _.union(seventyFiveValues, seventyFiveValues2);
+        },
+        'teardown': 'function multiArrays(){}'
+      })
+  );
+
   /*--------------------------------------------------------------------------*/
 
   suites.push(
@@ -1258,6 +1437,66 @@
 
   /*--------------------------------------------------------------------------*/
 
+  suites.push(
+    Benchmark.Suite('`_.without`')
+      .add('Lo-Dash', function() {
+        lodash.without(numbers, 9, 12, 14, 15);
+      })
+      .add('Underscore', function() {
+        _.without(numbers, 9, 12, 14, 15);
+      })
+  );
+
+  suites.push(
+    Benchmark.Suite('`_.without` iterating an array of 25 elements')
+      .add('Lo-Dash', {
+        'fn': function() {
+          lodash.without.apply(lodash, [twentyFiveValues].concat(twentyFiveValues2));
+        },
+        'teardown': 'function multiArrays(){}'
+      })
+      .add('Underscore', {
+        'fn': function() {
+          _.without.apply(_, [twentyFiveValues].concat(twentyFiveValues2));
+        },
+        'teardown': 'function multiArrays(){}'
+      })
+  );
+
+  suites.push(
+    Benchmark.Suite('`_.without` iterating an array of 50 elements')
+      .add('Lo-Dash', {
+        'fn': function() {
+          lodash.without.apply(lodash, [fiftyValues].concat(fiftyValues2));
+        },
+        'teardown': 'function multiArrays(){}'
+      })
+      .add('Underscore', {
+        'fn': function() {
+          _.without.apply(_, [fiftyValues].concat(fiftyValues2));
+        },
+        'teardown': 'function multiArrays(){}'
+      })
+  );
+
+  suites.push(
+    Benchmark.Suite('`_.without` iterating an array of 75 elements')
+      .add('Lo-Dash', {
+        'fn': function() {
+          lodash.without.apply(lodash, [seventyFiveValues].concat(seventyFiveValues2));
+        },
+        'teardown': 'function multiArrays(){}'
+      })
+      .add('Underscore', {
+        'fn': function() {
+          _.without.apply(_, [seventyFiveValues].concat(seventyFiveValues2));
+        },
+        'teardown': 'function multiArrays(){}'
+      })
+  );
+
+  /*--------------------------------------------------------------------------*/
+
   if (Benchmark.platform + '') {
     log(Benchmark.platform + '');
   }

test/test.js

@@ -166,6 +166,25 @@
 
   /*--------------------------------------------------------------------------*/
 
+  QUnit.module('lodash.difference');
+
+  (function() {
+    test('should work correctly when using `cachedContains`', function() {
+      var array1 = _.range(27),
+          array2 = array1.slice(),
+          a = {},
+          b = {},
+          c = {};
+
+      array1.push(a, b, c);
+      array2.push(b, c, a);
+
+      deepEqual(_.difference(array1, array2), []);
+    });
+  }());
+
+  /*--------------------------------------------------------------------------*/
+
   QUnit.module('lodash.escape');
 
   (function() {