Merge pull request #39 from rnagpal/master

Java SDK 1.5.1 release

Author: rnagpal

README.md

@@ -14,7 +14,7 @@ To get the binaries of this library as distributed by Microsoft, ready for use w
     <dependency>
     	<groupId>com.microsoft.azure</groupId>
     	<artifactId>azure-documentdb</artifactId>
-    	<version>1.5.0</version>
+    	<version>1.5.1</version>
     </dependency>
 
 ###Option 2: Source Via Git

changelog.md

@@ -1,3 +1,7 @@
+## Changes in 1.5.1 : ##
+
+- Fixed a bug in HashPartitionResolver to generate hash values in little-endian order to be consistent with other SDKs.
+
 ## Changes in 1.5.0 : ##
 
 - Added Client-side sharding framework to the SDK. Implemented HashPartionResolver and RangePartitionResolver classes.

pom.xml

@@ -3,7 +3,7 @@
   <modelVersion>4.0.0</modelVersion>
   <groupId>com.microsoft.azure</groupId>
   <artifactId>azure-documentdb</artifactId>
-  <version>1.5.0</version>
+  <version>1.5.1</version>
   <name>${project.groupId}:${project.artifactId}</name>
   <description>Java SDK for Microsoft Azure DocumentDB</description>
   <url>http://azure.microsoft.com/en-us/services/documentdb/</url>

src/com/microsoft/azure/documentdb/ConsistentHashRing.java

@@ -24,26 +24,31 @@ of this software and associated documentation files (the "Software"), to deal
 package com.microsoft.azure.documentdb;
 
 import java.io.UnsupportedEncodingException;
+import java.nio.ByteBuffer;
+import java.nio.ByteOrder;
+import java.util.AbstractMap;
 import java.util.ArrayList;
 import java.util.Arrays;
+import java.util.List;
+import java.util.Map;
 
 /**
- * The ConsistentHashRing class internally implements a consistent hash ring using the hash function specified
+ * The ConsistentHashRing class implements a consistent hash ring using the hash function specified
  */
 final class ConsistentHashRing {
     private HashGenerator hashGenerator;
     private Partition[] partitions;
     private ArrayList<String> collectionLinks = new ArrayList<String>();
 
     /**
-     * ConsistentHashRing constructor taking in the collection links, total number of partitions 
+     * ConsistentHashRing constructor taking in the collection links, number of partitions per node 
      * and hash generator to initialize the ring.
      * 
      * @param collectionLinks the links of collections participating in partitioning.
-     * @param totalPartitions the total number of partitions.
+     * @param partitionsPerNode number of partitions per node.
      * @param hashGenerator the hash generator to be used for hashing algorithm.
      */
-    public ConsistentHashRing(Iterable<String> collectionLinks, int totalPartitions, HashGenerator hashGenerator) {
+    public ConsistentHashRing(Iterable<String> collectionLinks, int partitionsPerNode, HashGenerator hashGenerator) {
         if(collectionLinks == null) {
             throw new IllegalArgumentException("collectionLinks");
         }
@@ -52,16 +57,16 @@ public ConsistentHashRing(Iterable<String> collectionLinks, int totalPartitions,
             this.collectionLinks.add(collectionLink);
         }
 
-        if(totalPartitions < this.collectionLinks.size()) {
-            throw new IllegalArgumentException("The total number of partitions must be at least the number of collections.");
+        if(partitionsPerNode <= 0) {
+            throw new IllegalArgumentException("The partitions per node must greater than 0.");
         }
 
         if(hashGenerator == null) {
             throw new IllegalArgumentException("hashGenerator");
         }
 
         this.hashGenerator = hashGenerator;
-        this.partitions = this.constructPartitions(this.collectionLinks, totalPartitions);
+        this.partitions = this.constructPartitions(this.collectionLinks, partitionsPerNode);
     }
 
     /**
@@ -112,23 +117,18 @@ private static byte[] getBytes(Object partitionKey) {
      * using the hashing algorithm and then finally sorting the partitions based on the hash value.
      * 
      */
-    private Partition[] constructPartitions(ArrayList<String> collectionLinks, int totalPartitions) {
+    private Partition[] constructPartitions(ArrayList<String> collectionLinks, int partitionsPerNode) {
         int collectionsNodeCount = collectionLinks.size();
-        Partition[] partitions = new Partition[totalPartitions];
-        
-        int partitionsPerNode = totalPartitions/collectionsNodeCount;
-        int extraPartitions = totalPartitions - (partitionsPerNode * collectionsNodeCount);
+        Partition[] partitions = new Partition[partitionsPerNode * collectionsNodeCount];
 
         int index = 0;
         for(String collectionNode : collectionLinks) {
             byte[] hashValue = this.hashGenerator.computeHash(getBytes(collectionNode));
 
-            for(int i=0; i < partitionsPerNode + (extraPartitions > 0 ? 1 : 0); ++i) {
+            for(int i=0; i < partitionsPerNode; ++i) {
                 partitions[index++] = new Partition(hashValue, collectionNode);
                 hashValue = this.hashGenerator.computeHash(hashValue);
             }
-            
-            extraPartitions--;
         }
 
         Arrays.sort(partitions);
@@ -157,5 +157,21 @@ private static int lowerBoundSearch(Partition[] partitions, byte[] hashValue) {
         }
 
         return partitions.length-1;
-    }   
+    }
+    
+    /**
+     * Gets the serialized version of the consistentRing. Added this helper for the test code.
+     * 
+     */
+    List<Map.Entry<String,Long>> getSerializedPartitionList() {
+        List<Map.Entry<String,Long>> partitionList= new ArrayList<>();
+        
+        for(int i=0; i<this.partitions.length; i++) {
+            ByteBuffer wrapped = ByteBuffer.wrap(partitions[i].getHashValue()).order(ByteOrder.LITTLE_ENDIAN);
+            int num = wrapped.getInt();
+            partitionList.add(new AbstractMap.SimpleEntry<>(partitions[i].getNode(), (long)num & 0x0FFFFFFFFL));
+        }
+         
+        return partitionList;
+    }
 }

src/com/microsoft/azure/documentdb/HashPartitionResolver.java

@@ -24,6 +24,8 @@ of this software and associated documentation files (the "Software"), to deal
 package com.microsoft.azure.documentdb;
 
 import java.util.ArrayList;
+import java.util.List;
+import java.util.Map;
 
 /**
  * HashPartitionResolver implements partitioning based on the value of a hash function, allowing you to evenly 
@@ -105,7 +107,7 @@ public HashPartitionResolver(PartitionKeyExtractor partitionKeyExtractor, Iterab
         }
 
         // Initialize the consistent ring to be used for the hashing algorithm by placing the virtual nodes along the ring
-        this.consistentHashRing = new ConsistentHashRing(this.collectionLinks, this.collectionLinks.size() * numberOfVirtualNodesPerCollection, hashGenerator);
+        this.consistentHashRing = new ConsistentHashRing(this.collectionLinks, numberOfVirtualNodesPerCollection, hashGenerator);
     }
 
     /**
@@ -121,10 +123,6 @@ public String resolveForCreate(Object document) {
             throw new IllegalArgumentException("document");
         }
 
-        if(this.partitionKeyExtractor == null) {
-            throw new IllegalStateException("Unable to extract partition key from document. Ensure that you have provided a valid PartitionKeyExtractor function.");
-        }
-        
         Object partitionKey = this.partitionKeyExtractor.getPartitionKey(document);
         return this.consistentHashRing.getCollectionNode(partitionKey);
     }
@@ -147,4 +145,13 @@ public Iterable<String> resolveForRead(Object partitionKey) {
         collectionLinks.add(this.consistentHashRing.getCollectionNode(partitionKey));
         return collectionLinks;
     }
+    
+    /**
+     * Gets the serialized version of the consistentRing. Added this helper for the test code.
+     * 
+     */
+    @SuppressWarnings("unused") // used only by test code
+    private List<Map.Entry<String,Long>> getSerializedPartitionList() {
+        return this.consistentHashRing.getSerializedPartitionList();
+    } 
 }

src/com/microsoft/azure/documentdb/HttpConstants.java

@@ -161,7 +161,7 @@ public static class HttpHeaders {
 
     public static class Versions {
         public static String CURRENT_VERSION = "2015-08-06";
-        public static String USER_AGENT = "documentdb-java-sdk-1.5.0";
+        public static String USER_AGENT = "documentdb-java-sdk-1.5.1";
     }
 
     public static class StatusCodes {

src/com/microsoft/azure/documentdb/MurmurHash.java

@@ -24,6 +24,7 @@ of this software and associated documentation files (the "Software"), to deal
 package com.microsoft.azure.documentdb;
 
 import java.nio.ByteBuffer;
+import java.nio.ByteOrder;
 
 /**
  *  The MurmurHash3 algorithm was created by Austin Appleby and placed in the public domain.
@@ -46,7 +47,10 @@ public byte[] computeHash(byte[] data) {
         }
 
         int hashValue = computeHash(data, data.length, 0);
-        return ByteBuffer.allocate(4).putInt(hashValue).array();
+        // Java's default "Endianess" is BigEndian but for all other SDKs  
+        // the default is LittleEndian, so changing the ByteOrder to be LittleEndian
+        // here so that we can be consistent across all SDKs.
+        return ByteBuffer.allocate(4).order(ByteOrder.LITTLE_ENDIAN).putInt(hashValue).array();
     }
 
     /** Returns the MurmurHash3_x86_32 hash. */

src/com/microsoft/azure/documentdb/Partition.java

@@ -60,11 +60,16 @@ private static int CompareHashValues(byte[] hash1, byte[] hash2) {
         if(hash1.length != hash2.length)
             throw new IllegalArgumentException("Length of hashes doesn't match.");
 
+        // Casting "byte" which is 8-bit signed data type to a "char" which is a 16 bit unsigned data type in Java, 
+        // so that they are compared the same way in all SDKs which have the native 8-bit types as unsigned.
+        
+        // The hash byte array that is returned from ComputeHash method has the MSB at the end of the array
+        // so comparing the bytes from the end for compare operations.
         for (int i = 0; i < hash1.length; i++) {
-            if (hash1[i] < hash2[i]) {
+            if ((char)hash1[hash1.length - i - 1] < (char)hash2[hash1.length - i - 1]) {
                 return -1;
             }
-            else if (hash1[i] > hash2[i]) {
+            else if ((char)hash1[hash1.length - i - 1] > (char)hash2[hash1.length - i - 1]) {
                 return 1;
             }
         }

src/com/microsoft/azure/documentdb/RangePartitionResolver.java

@@ -37,8 +37,7 @@ public class RangePartitionResolver<T extends Comparable<T>> implements Partitio
     private Map<Range<T>, String> partitionMap;
 
     /**
-     * RangePartitionResolver constructor taking in the PartitionKeyExtractor, a map of Ranges to collection links
-     * with the type of class used as the partition key.
+     * RangePartitionResolver constructor taking in the PartitionKeyExtractor, a map of Ranges to collection links.
      * 
      * @param partitionKeyExtractor an instance of class that implements PartitionKeyExtractor interface.
      * @param partitionMap the map of ranges to collection links.
@@ -69,10 +68,6 @@ public String resolveForCreate(Object document) {
           throw new IllegalArgumentException("document");
         }
 
-        if(this.partitionKeyExtractor == null) {
-            throw new UnsupportedOperationException("Unable to extract partition key from document. Ensure that you have provided a valid PartitionKeyExtractor function.");
-        }
-        
         Object partitionKey = this.partitionKeyExtractor.getPartitionKey(document);
         Range<T> containingRange = this.getContainingRange(partitionKey);
 
@@ -92,14 +87,7 @@ public String resolveForCreate(Object document) {
      */
     @Override
     public Iterable<String> resolveForRead(Object partitionKey) {
-        Set<Range<T>> intersectingRanges = null;
-        
-        if (partitionKey == null) {
-            intersectingRanges = this.partitionMap.keySet();
-        }
-        else {
-            intersectingRanges = this.getIntersectingRanges(partitionKey);
-        }
+        Set<Range<T>> intersectingRanges = this.getIntersectingRanges(partitionKey);
 
         ArrayList<String> collectionsLinks = new ArrayList<String>();
         for(Range<T> range : intersectingRanges) {
@@ -137,6 +125,10 @@ private Set<Range<T>> getIntersectingRanges(Object partitionKey) {
         Set<Range<T>> intersectingRanges = new HashSet<Range<T>>();
         Set<Range<T>> partitionKeyRanges = new HashSet<Range<T>>();
 
+        if (partitionKey == null) {
+            return this.partitionMap.keySet();
+        }
+        
         try {
             // Check the type of partitionKey to be Range<T>. In Java the type information is erased at runtime due to type erasure.
             // We can only check for Range<?> in this case