jdk/src/java.base/share/classes/sun/misc/MetaIndex.java
changeset 25859 3317bb8137f4
parent 14342 8435a30053c1
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/jdk/src/java.base/share/classes/sun/misc/MetaIndex.java	Sun Aug 17 15:54:13 2014 +0100
@@ -0,0 +1,274 @@
+/*
+ * Copyright (c) 2005, 2012, Oracle and/or its affiliates. All rights reserved.
+ * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
+ *
+ * This code is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 only, as
+ * published by the Free Software Foundation.  Oracle designates this
+ * particular file as subject to the "Classpath" exception as provided
+ * by Oracle in the LICENSE file that accompanied this code.
+ *
+ * This code is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * version 2 for more details (a copy is included in the LICENSE file that
+ * accompanied this code).
+ *
+ * You should have received a copy of the GNU General Public License version
+ * 2 along with this work; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
+ *
+ * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
+ * or visit www.oracle.com if you need additional information or have any
+ * questions.
+ */
+
+package sun.misc;
+
+import java.io.BufferedReader;
+import java.io.FileReader;
+import java.io.File;
+import java.io.IOException;
+import java.util.ArrayList;
+import java.util.Collections;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+
+/*
+ * MetaIndex is intended to decrease startup time (in particular cold
+ * start, when files are not yet in the disk cache) by providing a
+ * quick reject mechanism for probes into jar files. The on-disk
+ * representation of the meta-index is a flat text file with per-jar
+ * entries indicating (generally speaking) prefixes of package names
+ * contained in the jar. As an example, here is an edited excerpt of
+ * the meta-index generated for jre/lib in the current build:
+ *
+<PRE>
+% VERSION 1
+# charsets.jar
+sun/
+# jce.jar
+javax/
+! jsse.jar
+sun/
+com/sun/net/
+javax/
+com/sun/security/
+@ resources.jar
+com/sun/xml/
+com/sun/rowset/
+com/sun/org/
+sun/
+com/sun/imageio/
+javax/
+com/sun/java/swing/
+META-INF/services/
+com/sun/java/util/jar/pack/
+com/sun/corba/
+com/sun/jndi/
+! rt.jar
+org/w3c/
+com/sun/imageio/
+javax/
+java/
+sun/
+...
+</PRE>
+ * <p> A few notes about the design of the meta-index:
+ *
+ * <UL>
+ *
+ * <LI> It contains entries for multiple jar files. This is
+ * intentional, to reduce the number of disk accesses that need to be
+ * performed during startup.
+ *
+ * <LI> It is only intended to act as a fast reject mechanism to
+ * prevent application and other classes from forcing all jar files on
+ * the boot and extension class paths to be opened. It is not intended
+ * as a precise index of the contents of the jar.
+ *
+ * <LI> It should be as small as possible to reduce the amount of time
+ * required to parse it during startup. For example, adding on the
+ * secondary package element to java/ and javax/ packages
+ * ("javax/swing/", for example) causes the meta-index to grow
+ * significantly. This is why substrings of the packages have been
+ * chosen as the principal contents.
+ *
+ * <LI> It is versioned, and optional, to prevent strong dependencies
+ * between the JVM and JDK. It is also potentially applicable to more
+ * than just the boot and extension class paths.
+ *
+ * <LI> Precisely speaking, it plays different role in JVM and J2SE
+ * side.  On the JVM side, meta-index file is used to speed up locating the
+ * class files only while on the J2SE side, meta-index file is used to speed
+ * up the resources file & class file.
+ * To help the JVM and J2SE code to better utilize the information in meta-index
+ * file, we mark the jar file differently. Here is the current rule we use.
+ * For jar file containing only class file, we put '!' before the jar file name;
+ * for jar file containing only resources file, we put '@' before the jar file name;
+ * for jar file containing both resources and class file, we put '#' before the
+ * jar name.
+ * Notice the fact that every jar file contains at least the manifest file, so when
+ * we say "jar file containing only class file", we don't include that file.
+ *
+ * </UL>
+ *
+ * <p> To avoid changing the behavior of the current application
+ * loader and other loaders, the current MetaIndex implementation in
+ * the JDK requires that the directory containing the meta-index be
+ * registered with the MetaIndex class before construction of the
+ * associated URLClassPath. This prevents the need for automatic
+ * searching for the meta-index in the URLClassPath code and potential
+ * changes in behavior for non-core ClassLoaders.
+ *
+ * This class depends on make/tools/MetaIndex/BuildMetaIndex.java and
+ * is used principally by sun.misc.URLClassPath.
+ */
+
+public class MetaIndex {
+    // Maps jar file names in registered directories to meta-indices
+    private static volatile Map<File, MetaIndex> jarMap;
+
+    // List of contents of this meta-index
+    private String[] contents;
+
+    // Indicate whether the coresponding jar file is a pure class jar file or not
+    private boolean isClassOnlyJar;
+
+    //----------------------------------------------------------------------
+    // Registration of directories (which can cause parsing of the
+    // meta-index file if it is present), and fetching of parsed
+    // meta-indices
+    // jarMap is not strictly thread-safe when the meta index mechanism
+    // is extended for user-provided jar files in future.
+
+    public static MetaIndex forJar(File jar) {
+        return getJarMap().get(jar);
+    }
+
+    // 'synchronized' is added to protect the jarMap from being modified
+    // by multiple threads.
+    public static synchronized void registerDirectory(File dir) {
+        // Note that this does not currently check to see whether the
+        // directory has previously been registered, since the meta-index
+        // in a particular directory creates multiple entries in the
+        // jarMap. If this mechanism is extended beyond the boot and
+        // extension class paths (for example, automatically searching for
+        // meta-index files in directories containing jars which have been
+        // explicitly opened) then this code should be generalized.
+        //
+        // This method must be called from a privileged context.
+        File indexFile = new File(dir, "meta-index");
+        if (indexFile.exists()) {
+            try {
+                BufferedReader reader = new BufferedReader(new FileReader(indexFile));
+                String line = null;
+                String curJarName = null;
+                boolean isCurJarContainClassOnly = false;
+                List<String> contents = new ArrayList<String>();
+                Map<File, MetaIndex> map = getJarMap();
+
+                /* Convert dir into canonical form. */
+                dir = dir.getCanonicalFile();
+                /* Note: The first line should contain the version of
+                 * the meta-index file. We have to match the right version
+                 * before trying to parse this file. */
+                line = reader.readLine();
+                if (line == null ||
+                    !line.equals("% VERSION 2")) {
+                    reader.close();
+                    return;
+                }
+                while ((line = reader.readLine()) != null) {
+                    switch (line.charAt(0)) {
+                    case '!':
+                    case '#':
+                    case '@': {
+                        // Store away current contents, if any
+                        if ((curJarName != null) && (contents.size() > 0)) {
+                            map.put(new File(dir, curJarName),
+                                    new MetaIndex(contents,
+                                                  isCurJarContainClassOnly));
+
+                            contents.clear();
+                        }
+                        // Fetch new current jar file name
+                        curJarName = line.substring(2);
+                        if (line.charAt(0) == '!') {
+                            isCurJarContainClassOnly = true;
+                        } else if (isCurJarContainClassOnly) {
+                            isCurJarContainClassOnly = false;
+                        }
+
+                        break;
+                    }
+                    case '%':
+                        break;
+                    default: {
+                        contents.add(line);
+                    }
+                    }
+                }
+                // Store away current contents, if any
+                if ((curJarName != null) && (contents.size() > 0)) {
+                    map.put(new File(dir, curJarName),
+                            new MetaIndex(contents, isCurJarContainClassOnly));
+                }
+
+                reader.close();
+
+            } catch (IOException e) {
+                // Silently fail for now (similar behavior to elsewhere in
+                // extension and core loaders)
+            }
+        }
+    }
+
+    //----------------------------------------------------------------------
+    // Public APIs
+    //
+
+    public boolean mayContain(String entry) {
+        // Ask non-class file from class only jar returns false
+        // This check is important to avoid some class only jar
+        // files such as rt.jar are opened for resource request.
+        if  (isClassOnlyJar && !entry.endsWith(".class")){
+            return false;
+        }
+
+        String[] conts = contents;
+        for (int i = 0; i < conts.length; i++) {
+            if (entry.startsWith(conts[i])) {
+                return true;
+            }
+        }
+        return false;
+    }
+
+
+    //----------------------------------------------------------------------
+    // Implementation only below this point
+    // @IllegalArgumentException if entries is null.
+    private MetaIndex(List<String> entries, boolean isClassOnlyJar)
+        throws IllegalArgumentException {
+        if (entries == null) {
+            throw new IllegalArgumentException();
+        }
+
+        contents = entries.toArray(new String[0]);
+        this.isClassOnlyJar = isClassOnlyJar;
+    }
+
+    private static Map<File, MetaIndex> getJarMap() {
+        if (jarMap == null) {
+            synchronized (MetaIndex.class) {
+                if (jarMap == null) {
+                    jarMap = new HashMap<File, MetaIndex>();
+                }
+            }
+        }
+        assert jarMap != null;
+        return jarMap;
+    }
+}