REST API: Failure to index docs that have their ids URL encoded and contain `/`, closes #681.

modules/elasticsearch/src/main/java/org/elasticsearch/common/path/PathTrie.java

@@ -30,20 +30,37 @@ import static org.elasticsearch.common.collect.MapBuilder.*;
  * @author kimchy (Shay Banon)
  */
 public class PathTrie<T> {
+
+    public static interface Decoder {
+        String decode(String value);
+    }
+
+    public static final Decoder NO_DECODER = new Decoder() {
+        @Override public String decode(String value) {
+            return value;
+        }
+    };
+
+    private final Decoder decoder;
     private final TrieNode<T> root;
     private final Pattern pattern;
     private T rootValue;
 
     public PathTrie() {
-        this("/", "*");
+        this("/", "*", NO_DECODER);
     }
 
-    public PathTrie(String separator, String wildcard) {
+    public PathTrie(Decoder decoder) {
+        this("/", "*", decoder);
+    }
+
+    public PathTrie(String separator, String wildcard, Decoder decoder) {
+        this.decoder = decoder;
         pattern = Pattern.compile(separator);
         root = new TrieNode<T>(separator, null, null, wildcard);
     }
 
-    public static class TrieNode<T> {
+    public class TrieNode<T> {
         private transient String key;
         private transient T value;
         private boolean isWildcard;
@@ -169,6 +186,10 @@ public class PathTrie<T> {
 
             return res;
         }
+
+        private void put(Map<String, String> params, String key, String value) {
+            params.put(key, decoder.decode(value));
+        }
     }
 
     public void insert(String path, T value) {
@@ -204,8 +225,4 @@ public class PathTrie<T> {
         }
         return root.retrieve(strings, index, params);
     }
-
-    private static void put(Map<String, String> params, String key, String value) {
-        params.put(key, value);
-    }
 }

modules/elasticsearch/src/main/java/org/elasticsearch/http/HttpServer.java

@@ -29,6 +29,7 @@ import org.elasticsearch.rest.RestController;
 import org.elasticsearch.rest.RestRequest;
 import org.elasticsearch.rest.StringRestResponse;
 import org.elasticsearch.rest.XContentThrowableRestResponse;
+import org.elasticsearch.rest.support.RestUtils;
 import org.elasticsearch.threadpool.ThreadPool;
 
 import java.io.IOException;
@@ -48,12 +49,12 @@ public class HttpServer extends AbstractLifecycleComponent<HttpServer> {
 
     private final TransportNodesInfoAction nodesInfoAction;
 
-    private final PathTrie<HttpServerHandler> getHandlers = new PathTrie<HttpServerHandler>();
+    private final PathTrie<HttpServerHandler> getHandlers = new PathTrie<HttpServerHandler>(RestUtils.REST_DECODER);
-    private final PathTrie<HttpServerHandler> postHandlers = new PathTrie<HttpServerHandler>();
+    private final PathTrie<HttpServerHandler> postHandlers = new PathTrie<HttpServerHandler>(RestUtils.REST_DECODER);
-    private final PathTrie<HttpServerHandler> putHandlers = new PathTrie<HttpServerHandler>();
+    private final PathTrie<HttpServerHandler> putHandlers = new PathTrie<HttpServerHandler>(RestUtils.REST_DECODER);
-    private final PathTrie<HttpServerHandler> deleteHandlers = new PathTrie<HttpServerHandler>();
+    private final PathTrie<HttpServerHandler> deleteHandlers = new PathTrie<HttpServerHandler>(RestUtils.REST_DECODER);
-    private final PathTrie<HttpServerHandler> headHandlers = new PathTrie<HttpServerHandler>();
+    private final PathTrie<HttpServerHandler> headHandlers = new PathTrie<HttpServerHandler>(RestUtils.REST_DECODER);
-    private final PathTrie<HttpServerHandler> optionsHandlers = new PathTrie<HttpServerHandler>();
+    private final PathTrie<HttpServerHandler> optionsHandlers = new PathTrie<HttpServerHandler>(RestUtils.REST_DECODER);
 
     @Inject public HttpServer(Settings settings, HttpServerTransport transport, ThreadPool threadPool,
                               RestController restController, TransportNodesInfoAction nodesInfoAction) {
@@ -166,6 +167,9 @@ public class HttpServer extends AbstractLifecycleComponent<HttpServer> {
     }
 
     private String getPath(HttpRequest request) {
-        return request.path();
+        // we use rawPath since we don't want to decode it while processing the path resolution
+        // so we can handle things like:
+        // my_index/my_type/http%3A%2F%2Fwww.google.com
+        return request.rawPath();
     }
 }

modules/elasticsearch/src/main/java/org/elasticsearch/http/netty/NettyHttpRequest.java

@@ -39,7 +39,7 @@ public class NettyHttpRequest extends AbstractRestRequest implements HttpRequest
 
     private final Map<String, String> params;
 
-    private final String path;
+    private final String rawPath;
 
     private byte[] cachedData;
 
@@ -50,9 +50,9 @@ public class NettyHttpRequest extends AbstractRestRequest implements HttpRequest
         String uri = request.getUri();
         int pathEndPos = uri.indexOf('?');
         if (pathEndPos < 0) {
-            this.path = RestUtils.decodeComponent(uri);
+            this.rawPath = uri;
         } else {
-            this.path = RestUtils.decodeComponent(uri.substring(0, pathEndPos));
+            this.rawPath = uri.substring(0, pathEndPos);
             RestUtils.decodeQueryString(uri, pathEndPos + 1, params);
         }
     }
@@ -86,8 +86,8 @@ public class NettyHttpRequest extends AbstractRestRequest implements HttpRequest
         return request.getUri();
     }
 
-    @Override public String path() {
+    @Override public String rawPath() {
-        return path;
+        return rawPath;
     }
 
     @Override public Map<String, String> params() {

modules/elasticsearch/src/main/java/org/elasticsearch/rest/RestController.java

@@ -25,6 +25,7 @@ import org.elasticsearch.common.component.AbstractLifecycleComponent;
 import org.elasticsearch.common.inject.Inject;
 import org.elasticsearch.common.path.PathTrie;
 import org.elasticsearch.common.settings.Settings;
+import org.elasticsearch.rest.support.RestUtils;
 
 import java.io.IOException;
 
@@ -33,12 +34,12 @@ import java.io.IOException;
  */
 public class RestController extends AbstractLifecycleComponent<RestController> {
 
-    private final PathTrie<RestHandler> getHandlers = new PathTrie<RestHandler>();
+    private final PathTrie<RestHandler> getHandlers = new PathTrie<RestHandler>(RestUtils.REST_DECODER);
-    private final PathTrie<RestHandler> postHandlers = new PathTrie<RestHandler>();
+    private final PathTrie<RestHandler> postHandlers = new PathTrie<RestHandler>(RestUtils.REST_DECODER);
-    private final PathTrie<RestHandler> putHandlers = new PathTrie<RestHandler>();
+    private final PathTrie<RestHandler> putHandlers = new PathTrie<RestHandler>(RestUtils.REST_DECODER);
-    private final PathTrie<RestHandler> deleteHandlers = new PathTrie<RestHandler>();
+    private final PathTrie<RestHandler> deleteHandlers = new PathTrie<RestHandler>(RestUtils.REST_DECODER);
-    private final PathTrie<RestHandler> headHandlers = new PathTrie<RestHandler>();
+    private final PathTrie<RestHandler> headHandlers = new PathTrie<RestHandler>(RestUtils.REST_DECODER);
-    private final PathTrie<RestHandler> optionsHandlers = new PathTrie<RestHandler>();
+    private final PathTrie<RestHandler> optionsHandlers = new PathTrie<RestHandler>(RestUtils.REST_DECODER);
 
     @Inject public RestController(Settings settings) {
         super(settings);
@@ -116,6 +117,9 @@ public class RestController extends AbstractLifecycleComponent<RestController> {
     }
 
     private String getPath(RestRequest request) {
-        return request.path();
+        // we use rawPath since we don't want to decode it while processing the path resolution
+        // so we can handle things like:
+        // my_index/my_type/http%3A%2F%2Fwww.google.com
+        return request.rawPath();
     }
 }

modules/elasticsearch/src/main/java/org/elasticsearch/rest/RestRequest.java

@@ -43,7 +43,12 @@ public interface RestRequest extends ToXContent.Params {
     String uri();
 
     /**
-     * The path part of the URI (without the query string).
+     * The non decoded, raw path provided.
+     */
+    String rawPath();
+
+    /**
+     * The path part of the URI (without the query string), decoded.
      */
     String path();
 

modules/elasticsearch/src/main/java/org/elasticsearch/rest/support/AbstractRestRequest.java

@@ -37,6 +37,10 @@ public abstract class AbstractRestRequest implements RestRequest {
 
     private static final Pattern commaPattern = Pattern.compile(",");
 
+    @Override public final String path() {
+        return RestUtils.decodeComponent(rawPath());
+    }
+
     @Override public float paramAsFloat(String key, float defaultValue) {
         String sValue = param(key);
         if (sValue == null) {

modules/elasticsearch/src/main/java/org/elasticsearch/rest/support/RestUtils.java

@@ -20,6 +20,7 @@
 package org.elasticsearch.rest.support;
 
 import org.elasticsearch.common.base.Charsets;
+import org.elasticsearch.common.path.PathTrie;
 
 import java.nio.charset.Charset;
 import java.util.Map;
@@ -29,6 +30,12 @@ import java.util.Map;
  */
 public class RestUtils {
 
+    public static PathTrie.Decoder REST_DECODER = new PathTrie.Decoder() {
+        @Override public String decode(String value) {
+            return RestUtils.decodeComponent(value);
+        }
+    };
+
     public static void decodeQueryString(String s, int fromIndex, Map<String, String> params) {
         if (fromIndex < 0) {
             return;

plugins/transport/memcached/src/main/java/org/elasticsearch/memcached/MemcachedRestRequest.java

@@ -45,7 +45,7 @@ public class MemcachedRestRequest extends AbstractRestRequest {
 
     private final Map<String, String> params;
 
-    private final String path;
+    private final String rawPath;
 
     private byte[] data;
 
@@ -62,9 +62,9 @@ public class MemcachedRestRequest extends AbstractRestRequest {
         this.params = new HashMap<String, String>();
         int pathEndPos = uri.indexOf('?');
         if (pathEndPos < 0) {
-            this.path = uri;
+            this.rawPath = uri;
         } else {
-            this.path = uri.substring(0, pathEndPos);
+            this.rawPath = uri.substring(0, pathEndPos);
             RestUtils.decodeQueryString(uri, pathEndPos + 1, params);
         }
     }
@@ -77,8 +77,8 @@ public class MemcachedRestRequest extends AbstractRestRequest {
         return this.uri;
     }
 
-    @Override public String path() {
+    @Override public String rawPath() {
-        return this.path;
+        return this.rawPath;
     }
 
     public byte[] getUriBytes() {

plugins/transport/thrift/src/main/java/org/elasticsearch/thrift/ThriftRestRequest.java

@@ -36,7 +36,7 @@ public class ThriftRestRequest extends AbstractRestRequest implements org.elasti
 
     private final org.elasticsearch.thrift.RestRequest request;
 
-    private final String path;
+    private final String rawPath;
 
     private final Map<String, String> params;
 
@@ -46,9 +46,9 @@ public class ThriftRestRequest extends AbstractRestRequest implements org.elasti
 
         int pathEndPos = request.getUri().indexOf('?');
         if (pathEndPos < 0) {
-            this.path = request.getUri();
+            this.rawPath = request.getUri();
         } else {
-            this.path = request.getUri().substring(0, pathEndPos);
+            this.rawPath = request.getUri().substring(0, pathEndPos);
             RestUtils.decodeQueryString(request.getUri(), pathEndPos + 1, params);
         }
     }
@@ -75,8 +75,8 @@ public class ThriftRestRequest extends AbstractRestRequest implements org.elasti
         return request.getUri();
     }
 
-    @Override public String path() {
+    @Override public String rawPath() {
-        return this.path;
+        return this.rawPath;
     }
 
     @Override public boolean hasContent() {