Fix documents percentages, if the limit is higher than real count of the documents #76

test/src/test/java/com/github/variety/test/LimitResultsAnalysisTest.java

@@ -35,4 +35,20 @@ public class LimitResultsAnalysisTest {
         analysis.validate("name", 1, 100, "String");
         analysis.validate("someBinData", 1, 100, "BinData-old");
     }
+
+    @Test
+    public void verifyLimitOverMaxResults() throws Exception {
+
+        // limit is set to higher number, that the actual number of documents in collection
+        // analysis should compute percentages based on the real number of documents, not on the
+        // number provided in the limit var.
+        final ResultsValidator analysis = variety.withLimit(10).runDatabaseAnalysis();
+
+        analysis.validate("_id", 5, 100, "ObjectId");
+        analysis.validate("name", 5, 100, "String");
+        analysis.validate("bio", 3, 60, "String");
+        analysis.validate("pets", 2, 40, "String", "Array");
+        analysis.validate("someBinData", 1, 20, "BinData-old");
+        analysis.validate("someWeirdLegacyKey", 1, 20, "String");
+    }
 }

variety.js

@@ -205,7 +205,7 @@ var mergeDocument = function(docResult, interimResults) {
   }
 };
 
-var convertResults = function(interimResults) {
+var convertResults = function(interimResults, documentsCount) {
   var getKeys = function(obj) {
     var keys = [];
     for(var key in obj) {
@@ -221,7 +221,7 @@ var convertResults = function(interimResults) {
         '_id': {'key':key},
         'value': {'types':getKeys(entry.types)},
         'totalOccurrences': entry['totalOccurrences'],
-        'percentContaining': entry['totalOccurrences'] * 100 / $limit
+        'percentContaining': entry['totalOccurrences'] * 100 / documentsCount
     });
   }
   return varietyResults;
@@ -255,13 +255,9 @@ DBQuery.prototype.reduce = function(callback, initialValue) {
   return result;
 };
 
-var interimResults = db[collection]
-  .find($query)
-  .sort($sort)
-  .limit($limit)
-  .reduce(reduceDocuments, {});
-
-var varietyResults = convertResults(interimResults)
+var cursor = db[collection].find($query).sort($sort).limit($limit);
+var interimResults = cursor.reduce(reduceDocuments, {});
+var varietyResults = convertResults(interimResults, cursor.size())
   .filter(filter)
   .sort(comparator);