Adding preprocessing on group by categories.

misoproject · Feb 12, 2012 · cce2e57 · cce2e57
1 parent 8fe24d1
commit cce2e57
Show file tree

Hide file tree

Showing 3 changed files with 58 additions and 15 deletions.
diff --git a/src/derived.js b/src/derived.js
@@ -16,21 +16,28 @@
     * group rows by values in a given column
     * @param {byColumn} column by which rows will be grouped
     * @param {columns} columns to be included
-    * @param {method} function to be applied, default addition
+    * @params {object} options 
+    *   method function to be applied, default addition
+    *   preprocess - specify a normalization function for the
+    * byColumn values if you need to group by some kind of derivation of 
+    * those values that are not just equality based.
     */
-    groupBy : function(byColumn, columns, method) {
+    groupBy : function(byColumn, columns, options) {
+
+      options = options || {};
+
       // TODO: should we check type match here?
       // default method is addition
-      method = method || function(array) {
-        return _.reduce(array, function(memo, num){ 
-          return memo + num; 
-        }, 0);
-      };
+      var method = options.method || _.sum;
 
       var d = {
         _columns : []
       };
 
+      if (options && options.preprocess) {
+        this.preprocess = options.preprocess;  
+      }
+
       var parser = new DS.Parsers();
 
       // copy columns we want - just types and names. No data.
@@ -53,7 +60,12 @@
 
       // bin all values by their categories
       for(var i = 0; i < this.length; i++) {
-        var category = this._columns[this._columnPositionByName[byColumn]].data[i];
+        var category = null;
+        if (this.preprocess) {
+          category = this.preprocess(this._columns[this._columnPositionByName[byColumn]].data[i]);
+        } else {
+          category = this._columns[this._columnPositionByName[byColumn]].data[i];  
+        }
 
         if (_.isUndefined(categoryPositions[category])) {
 

diff --git a/test/index.html b/test/index.html
@@ -42,9 +42,9 @@
   <script src="unit/products.js"></script>
   <script src="unit/views.js"></script>
   <script src="unit/events.js"></script>
-
-  <script src="unit/importers.js"></script>
   <script src="unit/derived.js"></script>
+  <script src="unit/importers.js"></script>
+
 
 </head>
 <body>

diff --git a/test/unit/derived.js b/test/unit/derived.js
@@ -65,14 +65,45 @@ test("base group by with diff modifier", function() {
   });
 
   var groupedData = ds.groupBy("state", 
-    ["count", "anothercount"],
-    function(array) {
-      return _.reduce(array, function(memo, num){ 
-        return memo * num; 
-      }, 1);
+    ["count", "anothercount"], {
+      method : function(array) {
+        return _.reduce(array, function(memo, num){ 
+          return memo * num; 
+        }, 1);
+      }
     });
 
   ok(_.isEqual(groupedData._columns[1].data, ["AZ", "MA"]), "states correct");
   ok(_.isEqual(groupedData._columns[2].data, [6,120]), "counts correct");
   ok(_.isEqual(groupedData._columns[3].data, [6000,120000]), "anothercounts correct" + groupedData._columns[3].data);
+});
+
+test("group by with preprocessing of categoeies", function() {
+  var ds = new DS.Dataset({
+    data : getData(),
+    strict: true
+  });
+
+  var groupedData = ds.groupBy("state", ["count", "anothercount"], {
+    preprocess : function(state) {
+      return state + state;
+    }
+  });
+
+  ok(_.isEqual(groupedData._columns[1].data, ["AZAZ", "MAMA"]), "states correct");
+  ok(_.isEqual(groupedData._columns[2].data, [6,15]), "counts correct");
+  ok(_.isEqual(groupedData._columns[3].data, [60,150]), "anothercounts correct" + groupedData._columns[3].data);
+
+  groupedData = ds.groupBy("state", ["count", "anothercount"], {
+    preprocess : function(state) {
+      return "A";
+    }
+  });
+
+  ok(_.isEqual(groupedData._columns[1].data, ["A"]), "states correct");
+  ok(_.isEqual(groupedData._columns[1].data.length, 1), "states correct");
+  ok(_.isEqual(groupedData._columns[2].data, [21]), "count correct");
+  ok(_.isEqual(groupedData._columns[3].data, [210]), "anothercounts correct" + groupedData._columns[3].data);
+
+
 });