Make workspace saving and loading more robust - fixes #528

- don't overwrite old files if we get an error writing new ones
- don't write unchanged data
- keep backup files around until next write rather than deleting
immediately
- attempt to recreate missing metadata as best as possible

main/src/com/google/refine/ProjectManager.java

@@ -62,6 +62,13 @@ public abstract class ProjectManager {
     // last n expressions used across all projects
     static protected final int s_expressionHistoryMax = 100;
 
+    // If a project has been idle this long, flush it from memory
+    static protected final int PROJECT_FLUSH_DELAY = 1000 * 60 * 15; // 15 minutes
+    
+    // Don't spend more than this much time saving projects if doing a quick save
+    static protected final int QUICK_SAVE_MAX_TIME = 1000 * 30; // 30 secs
+
+
     protected Map<Long, ProjectMetadata> _projectsMetadata;
     protected PreferenceStore            _preferenceStore;
 
@@ -202,7 +209,6 @@ public abstract class ProjectManager {
     public void save(boolean allModified) {
         if (allModified || _busy == 0) {
             saveProjects(allModified);
-            // TODO: Only save workspace if it's dirty
             saveWorkspace();
         }
     }
@@ -226,9 +232,6 @@ public abstract class ProjectManager {
         }
     }
 
-    static protected final int s_projectFlushDelay = 1000 * 60 * 15; // 15 minutes
-    static protected final int s_quickSaveTimeout = 1000 * 30; // 30 secs
-
     /**
      * Saves all projects to the data store
      * @param allModified
@@ -255,7 +258,7 @@ public abstract class ProjectManager {
                         records.add(new SaveRecord(project, msecsOverdue));
 
                     } else if (!project.getProcessManager().hasPending()
-                              && startTimeOfSave.getTime() - project.getLastSave().getTime() > s_projectFlushDelay) {
+                              && startTimeOfSave.getTime() - project.getLastSave().getTime() > PROJECT_FLUSH_DELAY) {
                         
                         /*
                          *  It's been a while since the project was last saved and it hasn't been
@@ -288,7 +291,7 @@ public abstract class ProjectManager {
 
             for (int i = 0;
                  i < records.size() &&
-                    (allModified || (new Date().getTime() - startTimeOfSave.getTime() < s_quickSaveTimeout));
+                    (allModified || (new Date().getTime() - startTimeOfSave.getTime() < QUICK_SAVE_MAX_TIME));
                  i++) {
 
                 try {

main/src/com/google/refine/ProjectMetadata.java

@@ -54,6 +54,7 @@ import com.google.refine.util.ParsingUtilities;
 public class ProjectMetadata implements Jsonizable {
     private final Date     _created;
     private Date           _modified;
+    private Date written = null;
     private String         _name;
     private String         _password;
 
@@ -71,11 +72,16 @@ public class ProjectMetadata implements Jsonizable {
     }
 
     public ProjectMetadata() {
-        _created = new Date();
+        this(new Date());
         _modified = _created;
-        preparePreferenceStore(_preferenceStore);
     }
 
+    public ProjectMetadata(Date created, Date modified, String name) {
+        this(created);
+        _modified = modified;
+        _name = name;
+    }
+    
     @Override
     public void write(JSONWriter writer, Properties options)
             throws JSONException {
@@ -103,16 +109,36 @@ public class ProjectMetadata implements Jsonizable {
         }
         
         writer.endObject();
+        
+        if ("save".equals(options.getProperty("mode"))) {
+            written = new Date();
+        }
+    }
+
+    public boolean isDirty() {
+        return written == null || _modified.after(written);
     }
     
     public void write(JSONWriter jsonWriter) throws JSONException  {
-        Properties options = new Properties();
+        write(jsonWriter, false);
-        options.setProperty("mode", "save");
+    }
+    
+    /**
+     * @param jsonWriter writer to save metadatea to
+     * @param onlyIfDirty true to not write unchanged metadata
+     * @throws JSONException
+     */
+    public void write(JSONWriter jsonWriter, boolean onlyIfDirty) throws JSONException  {
+        if (!onlyIfDirty || isDirty()) {
+            Properties options = new Properties();
+            options.setProperty("mode", "save");
 
-        write(jsonWriter, options);
+            write(jsonWriter, options);
+        }
     }
     
     static public ProjectMetadata loadFromJSON(JSONObject obj) {
+        // TODO: Is this correct?  It's using modified date for creation date
         ProjectMetadata pm = new ProjectMetadata(JSONUtilities.getDate(obj, "modified", new Date()));
 
         pm._modified = JSONUtilities.getDate(obj, "modified", new Date());
@@ -156,6 +182,8 @@ public class ProjectMetadata implements Jsonizable {
                 // ignore
             }
         }
+
+        pm.written = new Date(); // Mark it as not needing writing until modified
         
         return pm;
     }

main/src/com/google/refine/io/FileProjectManager.java

@@ -62,7 +62,7 @@ import com.google.refine.model.Project;
 import com.google.refine.preference.TopList;
 
 public class FileProjectManager extends ProjectManager {
-    final static protected String s_projectDirNameSuffix = ".project";
+    final static protected String PROJECT_DIR_SUFFIX = ".project";
 
     protected File                       _workspaceDir;
 
@@ -72,6 +72,8 @@ public class FileProjectManager extends ProjectManager {
         if (singleton == null) {
             logger.info("Using workspace directory: {}", dir.getAbsolutePath());
             singleton = new FileProjectManager(dir);
+            // This needs our singleton set, thus the unconventional control flow
+            ((FileProjectManager) singleton).recover();
         }
 
     }
@@ -85,7 +87,6 @@ public class FileProjectManager extends ProjectManager {
         }
 
         load();
-        recover();
     }
 
     public File getWorkspaceDir() {
@@ -93,7 +94,7 @@ public class FileProjectManager extends ProjectManager {
     }
 
     static public File getProjectDir(File workspaceDir, long projectID) {
-        File dir = new File(workspaceDir, projectID + s_projectDirNameSuffix);
+        File dir = new File(workspaceDir, projectID + PROJECT_DIR_SUFFIX);
         if (!dir.exists()) {
             dir.mkdir();
         }
@@ -114,6 +115,9 @@ public class FileProjectManager extends ProjectManager {
     public boolean loadProjectMetadata(long projectID) {
         synchronized (this) {
             ProjectMetadata metadata = ProjectMetadataUtilities.load(getProjectDir(projectID));
+            if (metadata == null) {
+                metadata = ProjectMetadataUtilities.recover(getProjectDir(projectID), projectID);
+            }
             if (metadata != null) {
                 _projectsMetadata.put(projectID, metadata);
                 return true;
@@ -227,20 +231,21 @@ public class FileProjectManager extends ProjectManager {
     }
 
 
-
     /**
      * Save the workspace's data out to file in a safe way: save to a temporary file first
      * and rename it to the real file.
-     * <p>
-     * FIXME: Even though this attempts to be safe by writing new file and renaming,
-     * it's still possible for it to corrupt things.
      */
     @Override
     protected void saveWorkspace() {
         synchronized (this) {
             File tempFile = new File(_workspaceDir, "workspace.temp.json");
             try {
-                saveToFile(tempFile);
+                if (!saveToFile(tempFile)) {
+                    // If the save wasn't really needed, just keep what we had
+                    tempFile.delete();
+                    logger.info("Skipping unnecessary workspace save");
+                    return;
+                }
             } catch (Exception e) {
                 e.printStackTrace();
 
@@ -251,21 +256,23 @@ public class FileProjectManager extends ProjectManager {
             File file = new File(_workspaceDir, "workspace.json");
             File oldFile = new File(_workspaceDir, "workspace.old.json");
 
+            if (oldFile.exists()) {
+                oldFile.delete();
+            }
+            
             if (file.exists()) {
                 file.renameTo(oldFile);
             }
 
             tempFile.renameTo(file);
-            if (oldFile.exists()) {
-                oldFile.delete();
-            }
 
             logger.info("Saved workspace");
         }
     }
 
-    protected void saveToFile(File file) throws IOException, JSONException {
+    protected boolean saveToFile(File file) throws IOException, JSONException {
         FileWriter writer = new FileWriter(file);
+        boolean saveWasNeeded = false;
         try {
             JSONWriter jsonWriter = new JSONWriter(writer);
             jsonWriter.object();
@@ -275,20 +282,24 @@ public class FileProjectManager extends ProjectManager {
                 ProjectMetadata metadata = _projectsMetadata.get(id);
                 if (metadata != null) {
                     jsonWriter.value(id);
-
+                    if (metadata.isDirty()) {
-                    ProjectMetadataUtilities.save(metadata, getProjectDir(id));
+                        ProjectMetadataUtilities.save(metadata, getProjectDir(id));
+                        saveWasNeeded = true;
+                    }
                 }
             }
             jsonWriter.endArray();
             writer.write('\n');
 
             jsonWriter.key("preferences");
+            saveWasNeeded |= _preferenceStore.isDirty();
             _preferenceStore.write(jsonWriter, new Properties());
 
             jsonWriter.endObject();
         } finally {
             writer.close();
         }
+        return saveWasNeeded;
     }
 
 
@@ -385,11 +396,12 @@ public class FileProjectManager extends ProjectManager {
     }
 
     protected void recover() {
+        boolean recovered = false;
         for (File file : _workspaceDir.listFiles()) {
             if (file.isDirectory() && !file.isHidden()) {
-                String name = file.getName();
+                String dirName = file.getName();
-                if (file.getName().endsWith(s_projectDirNameSuffix)) {
+                if (file.getName().endsWith(PROJECT_DIR_SUFFIX)) {
-                    String idString = name.substring(0, name.length() - s_projectDirNameSuffix.length());
+                    String idString = dirName.substring(0, dirName.length() - PROJECT_DIR_SUFFIX.length());
                     long id = -1;
                     try {
                         id = Long.parseLong(idString);
@@ -399,19 +411,22 @@ public class FileProjectManager extends ProjectManager {
 
                     if (id > 0 && !_projectsMetadata.containsKey(id)) {
                         if (loadProjectMetadata(id)) {
-                            logger.info(
+                            logger.info("Recovered project named " 
-                                    "Recovered project named " + 
+                                    + getProjectMetadata(id).getName()
-                                            getProjectMetadata(id).getName() +
+                                    + " in directory " + dirName);
-                                            " in directory " + name);
+                            recovered = true;
                         } else {
-                            logger.warn("Failed to recover project in directory " + name);
+                            logger.warn("Failed to recover project in directory " + dirName);
 
-                            file.renameTo(new File(file.getParentFile(), name + ".corrupted"));
+                            file.renameTo(new File(file.getParentFile(), dirName + ".corrupted"));
                         }
                     }
                 }
             }
         }
+        if (recovered) {
+            saveWorkspace();
+        }
     }
 
     @Override

main/src/com/google/refine/io/ProjectMetadataUtilities.java

@@ -39,7 +39,10 @@ import java.io.FileReader;
 import java.io.IOException;
 import java.io.OutputStreamWriter;
 import java.io.Writer;
+import java.util.Date;
+import java.util.List;
 
+import org.apache.commons.lang.StringUtils;
 import org.json.JSONException;
 import org.json.JSONObject;
 import org.json.JSONTokener;
@@ -48,6 +51,7 @@ import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
 import com.google.refine.ProjectMetadata;
+import com.google.refine.model.Project;
 
 
 public class ProjectMetadataUtilities {
@@ -60,14 +64,15 @@ public class ProjectMetadataUtilities {
         File file = new File(projectDir, "metadata.json");
         File oldFile = new File(projectDir, "metadata.old.json");
 
+        if (oldFile.exists()) {
+            oldFile.delete();
+        }
+        
         if (file.exists()) {
             file.renameTo(oldFile);
         }
 
         tempFile.renameTo(file);
-        if (oldFile.exists()) {
-            oldFile.delete();
-        }
     }
 
     protected static void saveToFile(ProjectMetadata projectMeta, File metadataFile) throws JSONException, IOException   {
@@ -98,6 +103,45 @@ public class ProjectMetadataUtilities {
 
         return null;
     }
+    
+    /**
+     * Reconstruct the project metadata on a best efforts basis.  The name is
+     * gone, so build something descriptive from the column names.  Recover the
+     * creation and modification times based on whatever files are available.
+     * 
+     * @param projectDir the project directory
+     * @param id the proejct id
+     * @return
+     */
+    static public ProjectMetadata recover(File projectDir, long id) {
+        ProjectMetadata pm = null;
+        Project p = ProjectUtilities.load(projectDir, id);
+        if (p != null) {
+            List<String> columnNames = p.columnModel.getColumnNames();
+            String tempName = "<recovered project> - " + columnNames.size() 
+                    + " cols X " + p.rows.size() + " rows - "
+                    + StringUtils.join(columnNames,'|');
+            p.dispose();
+            long ctime = System.currentTimeMillis();
+            long mtime = 0;
+
+            File dataFile = new File(projectDir, "data.zip");
+            ctime = mtime = dataFile.lastModified();
+
+            File historyDir = new File(projectDir,"history");
+            File[] files = historyDir.listFiles();
+            if (files != null) {
+                for (File f : files) {
+                    long time = f.lastModified();
+                    ctime = Math.min(ctime, time);
+                    mtime = Math.max(mtime, time);
+                }
+            }
+            pm = new ProjectMetadata(new Date(ctime),new Date(mtime), tempName);
+            logger.error("Partially recovered missing metadata project in directory " + projectDir + " - " + tempName);
+        }
+        return pm;
+    }
 
     static protected ProjectMetadata loadFromFile(File metadataFile) throws Exception {
         FileReader reader = new FileReader(metadataFile);

main/src/com/google/refine/model/Project.java

@@ -194,6 +194,7 @@ public class Project {
     ) throws Exception {
         long start = System.currentTimeMillis();
         
+        // version of Refine which wrote the file
         /* String version = */ reader.readLine();
         
         Project project = new Project(id);