Work on logical backups.
[lgpl/argeo-commons.git] / org.argeo.maintenance / src / org / argeo / maintenance / backup / LogicalBackup.java
index 4f7a2cfea561e7806a867a69268b9dcd2921b86f..60e8f8e5d89d13ec5cbceb728aa8a6f33a16e0f6 100644 (file)
@@ -18,6 +18,7 @@ import java.util.Enumeration;
 import java.util.HashMap;
 import java.util.Map;
 import java.util.Set;
+import java.util.TreeMap;
 import java.util.concurrent.ExecutionException;
 import java.util.concurrent.ExecutorService;
 import java.util.concurrent.Executors;
@@ -42,6 +43,8 @@ import javax.jcr.Session;
 import org.apache.commons.io.IOUtils;
 import org.apache.commons.logging.Log;
 import org.apache.commons.logging.LogFactory;
+import org.apache.jackrabbit.api.JackrabbitSession;
+import org.apache.jackrabbit.api.JackrabbitValue;
 import org.argeo.api.NodeConstants;
 import org.argeo.api.NodeUtils;
 import org.argeo.jackrabbit.client.ClientDavexRepositoryFactory;
@@ -50,7 +53,6 @@ import org.argeo.jcr.JcrException;
 import org.argeo.jcr.JcrUtils;
 import org.osgi.framework.Bundle;
 import org.osgi.framework.BundleContext;
-import org.xml.sax.SAXException;
 
 /**
  * Performs a backup of the data based only on programmatic interfaces. Useful
@@ -64,6 +66,10 @@ public class LogicalBackup implements Runnable {
        public final static String WORKSPACES_BASE = "workspaces/";
        public final static String FILES_BASE = "files/";
        public final static String OSGI_BASE = "share/osgi/";
+
+       public final static String JCR_SYSTEM = "jcr:system";
+       public final static String JCR_VERSION_STORAGE_PATH = "/jcr:system/jcr:versionStorage";
+
        private final Repository repository;
        private String defaultWorkspace;
        private final BundleContext bundleContext;
@@ -73,13 +79,19 @@ public class LogicalBackup implements Runnable {
 
        private ExecutorService executorService;
 
+       private boolean performSoftwareBackup = false;
+
+       private Map<String, String> checksums = new TreeMap<>();
+
+       private int threadCount = 5;
+
+       private boolean backupFailed = false;
+
        public LogicalBackup(BundleContext bundleContext, Repository repository, Path basePath) {
                this.repository = repository;
                this.zout = null;
                this.basePath = basePath;
                this.bundleContext = bundleContext;
-
-               executorService = Executors.newFixedThreadPool(3);
        }
 
        @Override
@@ -95,9 +107,12 @@ public class LogicalBackup implements Runnable {
        }
 
        public void perform() throws RepositoryException, IOException {
+               if (executorService != null && !executorService.isTerminated())
+                       throw new IllegalStateException("Another backup is running");
+               executorService = Executors.newFixedThreadPool(threadCount);
                long begin = System.currentTimeMillis();
                // software backup
-               if (bundleContext != null)
+               if (bundleContext != null && performSoftwareBackup)
                        executorService.submit(() -> performSoftwareBackup(bundleContext));
 
                // data backup
@@ -117,18 +132,37 @@ public class LogicalBackup implements Runnable {
                                executorService.awaitTermination(24, TimeUnit.HOURS);
                        } catch (InterruptedException e) {
                                // silent
+                               throw new IllegalStateException("Backup was interrupted before completion", e);
+                       }
+               }
+               // versions
+               executorService = Executors.newFixedThreadPool(threadCount);
+               try {
+                       performVersionsBackup();
+               } finally {
+                       executorService.shutdown();
+                       try {
+                               executorService.awaitTermination(24, TimeUnit.HOURS);
+                       } catch (InterruptedException e) {
+                               // silent
+                               throw new IllegalStateException("Backup was interrupted before completion", e);
                        }
                }
                long duration = System.currentTimeMillis() - begin;
-               log.info("System logical backup completed in " + (duration / 60000) + "min " + (duration / 1000) + "s");
+               if (isBackupFailed())
+                       log.info("System logical backup failed after " + (duration / 60000) + "min " + (duration / 1000) + "s");
+               else
+                       log.info("System logical backup completed in " + (duration / 60000) + "min " + (duration / 1000) + "s");
        }
 
        protected void performDataBackup(String workspaceName) throws RepositoryException, IOException {
                Session session = login(workspaceName);
                try {
                        nodes: for (NodeIterator nit = session.getRootNode().getNodes(); nit.hasNext();) {
+                               if (isBackupFailed())
+                                       return;
                                Node nodeToExport = nit.nextNode();
-                               if ("jcr:system".equals(nodeToExport.getName()) && !workspaceName.equals(defaultWorkspace))
+                               if (JCR_SYSTEM.equals(nodeToExport.getName()))
                                        continue nodes;
                                String nodePath = nodeToExport.getPath();
                                Future<Set<String>> contentPathsFuture = executorService
@@ -140,33 +174,51 @@ public class LogicalBackup implements Runnable {
                }
        }
 
+       protected void performVersionsBackup() throws RepositoryException, IOException {
+               Session session = login(defaultWorkspace);
+               Node versionStorageNode = session.getNode(JCR_VERSION_STORAGE_PATH);
+               try {
+                       for (NodeIterator nit = versionStorageNode.getNodes(); nit.hasNext();) {
+                               Node nodeToExport = nit.nextNode();
+                               String nodePath = nodeToExport.getPath();
+                               if (isBackupFailed())
+                                       return;
+                               Future<Set<String>> contentPathsFuture = executorService
+                                               .submit(() -> performNodeBackup(defaultWorkspace, nodePath));
+                               executorService.submit(() -> performFilesBackup(defaultWorkspace, contentPathsFuture));
+                       }
+               } finally {
+                       Jcr.logout(session);
+               }
+
+       }
+
        protected Set<String> performNodeBackup(String workspaceName, String nodePath) {
                Session session = login(workspaceName);
                try {
                        Node nodeToExport = session.getNode(nodePath);
-                       String nodeName = nodeToExport.getName();
+//                     String nodeName = nodeToExport.getName();
 //             if (nodeName.startsWith("jcr:") || nodeName.startsWith("rep:"))
 //                     continue nodes;
 //             // TODO make it more robust / configurable
 //             if (nodeName.equals("user"))
 //                     continue nodes;
-                       String relativePath = WORKSPACES_BASE + workspaceName + "/" + nodeName + ".xml";
+                       String relativePath = WORKSPACES_BASE + workspaceName + nodePath + ".xml";
                        OutputStream xmlOut = openOutputStream(relativePath);
                        BackupContentHandler contentHandler;
                        try (Writer writer = new BufferedWriter(new OutputStreamWriter(xmlOut, StandardCharsets.UTF_8))) {
-                               contentHandler = new BackupContentHandler(writer, session);
+                               contentHandler = new BackupContentHandler(writer, nodeToExport);
                                session.exportSystemView(nodeToExport.getPath(), contentHandler, true, false);
                                if (log.isDebugEnabled())
-                                       log.debug(workspaceName + ":/" + nodeName + " metadata exported to " + relativePath);
+                                       log.debug(workspaceName + ":" + nodePath + " metadata exported to " + relativePath);
                        }
 
                        // Files
                        Set<String> contentPaths = contentHandler.getContentPaths();
                        return contentPaths;
-               } catch (IOException | SAXException e) {
-                       throw new RuntimeException("Cannot backup node " + workspaceName + ":" + nodePath, e);
-               } catch (RepositoryException e) {
-                       throw new JcrException("Cannot backup node " + workspaceName + ":" + nodePath, e);
+               } catch (Exception e) {
+                       markBackupFailed("Cannot backup node " + workspaceName + ":" + nodePath, e);
+                       throw new ThreadDeath();
                } finally {
                        Jcr.logout(session);
                }
@@ -177,7 +229,8 @@ public class LogicalBackup implements Runnable {
                try {
                        contentPaths = contentPathsFuture.get(24, TimeUnit.HOURS);
                } catch (InterruptedException | ExecutionException | TimeoutException e1) {
-                       throw new RuntimeException("Cannot retrieve content paths for workspace " + workspaceName);
+                       markBackupFailed("Cannot retrieve content paths for workspace " + workspaceName, e1);
+                       return;
                }
                if (contentPaths == null || contentPaths.size() == 0)
                        return;
@@ -185,23 +238,55 @@ public class LogicalBackup implements Runnable {
                try {
                        String workspacesFilesBasePath = FILES_BASE + workspaceName;
                        for (String path : contentPaths) {
+                               if (isBackupFailed())
+                                       return;
                                Node contentNode = session.getNode(path);
-                               Binary binary = contentNode.getProperty(Property.JCR_DATA).getBinary();
-                               String fileRelativePath = workspacesFilesBasePath + contentNode.getParent().getPath();
-                               try (InputStream in = binary.getStream(); OutputStream out = openOutputStream(fileRelativePath)) {
-                                       IOUtils.copy(in, out);
-                                       if (log.isTraceEnabled())
-                                               log.trace("Workspace " + workspaceName + ": file content exported to " + fileRelativePath);
+                               Binary binary = null;
+                               try {
+                                       binary = contentNode.getProperty(Property.JCR_DATA).getBinary();
+                                       String fileRelativePath = workspacesFilesBasePath + contentNode.getParent().getPath();
+
+                                       // checksum
+                                       boolean skip = false;
+                                       String checksum = null;
+                                       if (session instanceof JackrabbitSession) {
+                                               JackrabbitValue value = (JackrabbitValue) contentNode.getProperty(Property.JCR_DATA).getValue();
+//                                     ReferenceBinary referenceBinary = (ReferenceBinary) binary;
+                                               checksum = value.getContentIdentity();
+                                       }
+                                       if (checksum != null) {
+                                               if (!checksums.containsKey(checksum)) {
+                                                       checksums.put(checksum, fileRelativePath);
+                                               } else {
+                                                       skip = true;
+                                                       String sourcePath = checksums.get(checksum);
+                                                       if (log.isTraceEnabled())
+                                                               log.trace(fileRelativePath + " : already " + sourcePath + " with checksum " + checksum);
+                                                       createLink(sourcePath, fileRelativePath);
+                                                       try (Writer writerSum = new OutputStreamWriter(
+                                                                       openOutputStream(fileRelativePath + ".sha256"), StandardCharsets.UTF_8)) {
+                                                               writerSum.write(checksum);
+                                                       }
+                                               }
+                                       }
+
+                                       // copy file
+                                       if (!skip)
+                                               try (InputStream in = binary.getStream();
+                                                               OutputStream out = openOutputStream(fileRelativePath)) {
+                                                       IOUtils.copy(in, out);
+                                                       if (log.isTraceEnabled())
+                                                               log.trace("Workspace " + workspaceName + ": file content exported to "
+                                                                               + fileRelativePath);
+                                               }
                                } finally {
                                        JcrUtils.closeQuietly(binary);
                                }
                        }
                        if (log.isDebugEnabled())
                                log.debug(workspaceName + ":" + contentPaths.size() + " files exported to " + workspacesFilesBasePath);
-               } catch (RepositoryException e) {
-                       throw new JcrException("Cannot backup files from " + workspaceName + ":", e);
-               } catch (IOException e) {
-                       throw new RuntimeException("Cannot backup files from " + workspaceName + ":", e);
+               } catch (Exception e) {
+                       markBackupFailed("Cannot backup files from " + workspaceName + ":", e);
                } finally {
                        Jcr.logout(session);
                }
@@ -221,6 +306,21 @@ public class LogicalBackup implements Runnable {
                }
        }
 
+       protected void createLink(String source, String target) throws IOException {
+               if (zout != null) {
+                       // TODO implement for zip
+                       throw new UnsupportedOperationException();
+               } else if (basePath != null) {
+                       Path sourcePath = basePath.resolve(Paths.get(source));
+                       Path targetPath = basePath.resolve(Paths.get(target));
+                       Path relativeSource = targetPath.getParent().relativize(sourcePath);
+                       Files.createDirectories(targetPath.getParent());
+                       Files.createSymbolicLink(targetPath, relativeSource);
+               } else {
+                       throw new UnsupportedOperationException();
+               }
+       }
+
        protected void closeOutputStream(String relativePath, OutputStream out) throws IOException {
                if (zout != null) {
                        zout.closeEntry();
@@ -335,4 +435,15 @@ public class LogicalBackup implements Runnable {
 
        }
 
+       protected synchronized void markBackupFailed(Object message, Exception e) {
+               log.error(message, e);
+               backupFailed = true;
+               notifyAll();
+               if (executorService != null)
+                       executorService.shutdownNow();
+       }
+
+       protected boolean isBackupFailed() {
+               return backupFailed;
+       }
 }