now supporting the effect of limit in cases with and without query.

variety.js

@@ -152,6 +152,7 @@ function serializeDoc(doc, maxDepth){
 
 var interimResults = {}; //hold results here until converted to final format
 // main cursor
+var numDocuments = 0;
 db[collection].find(query).sort(sort).limit(limit).forEach(function(obj) {
 	//printjson(obj)
 	flattened = serializeDoc(obj, maxDepth);
@@ -174,9 +175,9 @@ db[collection].find(query).sort(sort).limit(limit).forEach(function(obj) {
 			interimResults[key]['totalOccurrences']++;
 		}
 	}
+    numDocuments++;
 });
 
-
 var varietyResults = [];
 //now convert the interimResults into the proper format
 for(var key in interimResults){
@@ -189,8 +190,6 @@ for(var key in interimResults){
 	varietyResults.push(newEntry);
 }
 
-var numDocuments = db[collection].count(query);
-
 // We throw away keys which end in an array index, since they are not useful
 // for our analysis. (We still keep the key of their parent array, though.) -JC
 var filter = function(item) {
@@ -207,9 +206,7 @@ var map = function(item) {
   }
   // we don't need to set it if limit isn't being used. (it's set above.)
   if(limit < numDocuments) {
-    var existsQuery = query;
-    existsQuery[keyName] = {$exists: true};
-    item.totalOccurrences = db[collection].count(existsQuery);
+      item.totalOccurrences = db[collection].count(query);
   }
   item.percentContaining = (item.totalOccurrences / numDocuments) * 100.0;
   return item;