Merge branch 'rhino' into beta

TODO

@@ -1,3 +1,7 @@
-
-* better lifecycle management (couchdb keeps killing our java process when it thinks a query is too slow, this prevents ever building a decent FieldCache for sorting, etc).
-* use javascript to let couchdb user to program output transformation rather than current blanket strategy
+* indexer as updater process
+* searcher as external
+* sort cache can take too long to build, respond with keep-alive messages.
+* revisit numeric handling (it's too surprising)
+* document rhino integration
+* code tidy in general.
+* add a _status uri that lets you see index state/progress.

pom.xml

@@ -31,6 +31,11 @@
       <artifactId>tika</artifactId>
       <version>0.2</version>
     </dependency>
+    <dependency>
+      <groupId>rhino</groupId>
+      <artifactId>js</artifactId>
+      <version>1.7R1</version>
+    </dependency>
     <dependency>
       <groupId>log4j</groupId>
       <artifactId>log4j</artifactId>

src/main/java/org/apache/couchdb/lucene/Database.java

@@ -54,7 +54,10 @@ public JSONObject getAllDocsBySeq(final String dbname, final long from, final in
 	}
 
 	public JSONObject getDoc(final String dbname, final String id, final String rev) throws HttpException, IOException {
-		return JSONObject.fromObject(get(String.format("%s/%s?rev=%s", dbname, id, rev)));
+		if (rev == null)
+			return JSONObject.fromObject(get(String.format("%s/%s", dbname, id)));
+		else
+			return JSONObject.fromObject(get(String.format("%s/%s?rev=%s", dbname, id, rev)));
 	}
 
 	public JSONObject getDocs(final String dbname, final String... ids) throws HttpException, IOException {
@@ -96,7 +99,7 @@ private String post(final String path, final String body) throws HttpException,
 
 	private synchronized String execute(final HttpMethodBase method) throws HttpException, IOException {
 		try {
-			CLIENT.executeMethod(method);
+			CLIENT.executeMethod(method);			
 			final InputStream in = method.getResponseBodyAsStream();
 			try {
 				final StringWriter writer = new StringWriter(2048);

src/main/java/org/apache/couchdb/lucene/Index.java

@@ -79,6 +79,8 @@ private void openReader() throws IOException {
 
 	private class IndexUpdateTask extends TimerTask {
 
+		private Rhino filter = null;
+
 		@Override
 		public void run() {
 			IndexWriter writer = null;
@@ -87,6 +89,18 @@ public void run() {
 				final String[] dbnames = db.getAllDatabases();
 				writer = newWriter();
 				for (final String dbname : dbnames) {
+					// Database might supply a filter.
+					final JSONObject designDoc = db.getDoc(dbname, "_design/lucene", null);
+					if (designDoc.containsKey("filter")) {
+						String filterFun = designDoc.getString("filter");
+						// Strip start and end double quotes.
+						filterFun = filterFun.replaceAll("^\"*", "");
+						filterFun = filterFun.replaceAll("\"*$", "");
+						filter = new Rhino(filterFun);
+					} else {
+						filter = null;
+					}
+
 					commit |= updateDatabase(writer, dbname);
 				}
 			} catch (final IOException e) {
@@ -167,13 +181,18 @@ private boolean updateDatabase(final IndexWriter writer, final String dbname) th
 		private void updateDocument(final IndexWriter writer, final String dbname, final JSONObject obj)
 				throws IOException {
 			final Document doc = new Document();
-			final JSONObject json = obj.getJSONObject("doc");
+			JSONObject json = obj.getJSONObject("doc");
 
 			// Skip design documents.
 			if (json.getString(Config.ID).startsWith("_design")) {
 				return;
 			}
 
+			// Filter, if supplied.
+			if (filter != null) {
+				json = JSONObject.fromObject(filter.parse(json.toString()));
+			}
+
 			// Standard properties.
 			doc.add(token(Config.DB, dbname, false));
 			final String id = (String) json.remove(Config.ID);

src/main/java/org/apache/couchdb/lucene/Rhino.java

@@ -0,0 +1,65 @@
+package org.apache.couchdb.lucene;
+
+import java.io.IOException;
+import java.io.InputStream;
+
+import org.apache.commons.io.IOUtils;
+import org.mozilla.javascript.Context;
+import org.mozilla.javascript.ContextFactory;
+import org.mozilla.javascript.Function;
+import org.mozilla.javascript.Scriptable;
+
+public final class Rhino {
+
+	private final ContextFactory contextFactory = new ContextFactory();
+
+	private final Context context;
+
+	private final Scriptable scope;
+
+	private final Function userFun;
+
+	private final Function systemFun;
+
+	private final String fun;
+
+	public Rhino(final String fun) throws IOException {
+		this.fun = fun;
+		this.context = contextFactory.enterContext();
+		context.setOptimizationLevel(9);
+		scope = context.initStandardObjects();
+
+		// compile user-defined function.
+		this.userFun = context.compileFunction(scope, fun, "userFun", 0, null);
+
+		// compile system function.
+		this.systemFun = context.compileFunction(scope,
+				"function(json,filter) { var doc=JSON.parse(json); doc=filter(doc); return JSON.stringify(doc); }",
+				"systemFun", 0, null);
+
+		// add JSON parser.
+		context.evaluateString(scope, loadJSONParser(), "json2", 0, null);
+	}
+
+	private String loadJSONParser() throws IOException {
+		final InputStream in = Rhino.class.getClassLoader().getResourceAsStream("json2.js");
+		try {
+			return IOUtils.toString(in, "UTF-8");
+		} finally {
+			in.close();
+		}
+	}
+
+	public String parse(final String doc) {
+		return (String) systemFun.call(context, scope, null, new Object[] { doc, userFun });
+	}
+
+	public void close() {
+		Context.exit();
+	}
+
+	public String toString() {
+		return fun;
+	}
+
+}