Refactor of CreateProjectCommand.java and Importers

The code for determining if an importer is suitable to import a file is now in each respective importer rather than in CreateProjectCommand.  There is an additional method, canImportData, on the Importer interface to support this.

CreateProjectCommand registers Importers from a Hashtable (this is a copy of Tom's code for registering commands in Gridworks Servlet).  Plugging in new importers should be simpler.

git-svn-id: http://google-refine.googlecode.com/svn/trunk@861 7d457c2a-affb-35e4-300a-418c747d4874

src/main/java/com/metaweb/gridworks/commands/project/CreateProjectCommand.java

@@ -45,12 +45,8 @@ import com.metaweb.gridworks.Gridworks;
 import com.metaweb.gridworks.ProjectManager;
 import com.metaweb.gridworks.ProjectMetadata;
 import com.metaweb.gridworks.commands.Command;
-import com.metaweb.gridworks.importers.ExcelImporter;
 import com.metaweb.gridworks.importers.Importer;
-import com.metaweb.gridworks.importers.MarcImporter;
-import com.metaweb.gridworks.importers.RdfTripleImporter;
 import com.metaweb.gridworks.importers.TsvCsvImporter;
-import com.metaweb.gridworks.importers.XmlImporter;
 import com.metaweb.gridworks.model.Project;
 import com.metaweb.gridworks.util.IOUtils;
 import com.metaweb.gridworks.util.ParsingUtilities;
@@ -59,6 +55,71 @@ public class CreateProjectCommand extends Command {
 
     final static Logger logger = LoggerFactory.getLogger("create-project_command");
 
+    static final private Map<String, Importer> importers = new HashMap<String, Importer>();
+
+    private static final String[][] importerNames = {
+        {"ExcelImporter", "com.metaweb.gridworks.importers.ExcelImporter"},
+        {"XmlImporter", "com.metaweb.gridworks.importers.XmlImporter"},
+        {"RdfTripleImporter", "com.metaweb.gridworks.importers.RdfTripleImporter"},
+        {"MarcImporter", "com.metaweb.gridworks.importers.MarcImporter"},
+        {"TsvCsvImporter", "com.metaweb.gridworks.importers.TsvCsvImporter"},
+    };
+
+    static {
+        registerImporters(importerNames);
+    }
+
+    static public boolean registerImporters(String[][] importers) {
+        boolean status = true;
+        for (String[] importer : importerNames) {
+            String importerName = importer[0];
+            String className = importer[1];
+            logger.debug("Loading command " + importerName + " class: " + className);
+            Importer cmd;
+            try {
+                // TODO: May need to use the servlet container's class loader here
+                cmd = (Importer) Class.forName(className).newInstance();
+            } catch (InstantiationException e) {
+                logger.error("Failed to load importer class " + className, e);
+                status = false;
+                continue;
+            } catch (IllegalAccessException e) {
+                logger.error("Failed to load importer class " + className, e);
+                status = false;
+                continue;
+            } catch (ClassNotFoundException e) {
+                logger.error("Failed to load importer class " + className, e);
+                status = false;
+                continue;
+            }
+            status |= registerImporter(importerName, cmd);
+        }
+        return status;
+    }
+
+    /**
+     * Register a single importer.
+     *
+     * @param name
+     *            importer verb for importer
+     * @param commandObject
+     *            object implementing the importer
+     * @return true if importer was loaded and registered successfully
+     */
+    static public boolean registerImporter(String name,
+            Importer importerObject) {
+        if (importers.containsKey(name)) {
+            return false;
+        }
+        importers.put(name, importerObject);
+        return true;
+    }
+
+    // Currently only for test purposes
+    static protected boolean unregisterImporter(String verb) {
+        return importers.remove(verb) != null;
+    }
+
     @Override
     public void doPost(HttpServletRequest request, HttpServletResponse response)
             throws ServletException, IOException {
@@ -301,7 +362,7 @@ public class CreateProjectCommand extends Command {
     }
 
     private void load(Project project, Properties options, String fileName, InputStream inputStream) throws Exception {
-        Importer importer = guessImporter(options, null, fileName);
+        Importer importer = guessImporter(null, fileName);
         internalInvokeImporter(project, importer, options, inputStream, null);
     }
 
@@ -366,7 +427,6 @@ public class CreateProjectCommand extends Command {
 
         try {
             Importer importer = guessImporter(
-                options, 
                 connection.getContentType(),
                 url.getPath()
             );
@@ -438,57 +498,13 @@ public class CreateProjectCommand extends Command {
         importer.read(reader, project, options);
     }
 
-    protected Importer guessImporter(
-            Properties options, String contentType, String fileName) {
-        
-        if (contentType != null) {
-            contentType = contentType.toLowerCase().trim();
-            
-            if ("application/msexcel".equals(contentType) ||
-                "application/x-msexcel".equals(contentType) ||
-                "application/x-ms-excel".equals(contentType) ||
-                "application/vnd.ms-excel".equals(contentType) ||
-                "application/x-excel".equals(contentType) ||
-                "application/xls".equals(contentType)) {
-                
-                return new ExcelImporter(false);
-            } else if("application/x-xls".equals(contentType)) {
-                return new ExcelImporter(true); 
-            } else if("application/xml".equals(contentType) ||
-                      "text/xml".equals(contentType) ||
-                      "application/rss+xml".equals(contentType) ||
-                      "application/atom+xml".equals(contentType)) {
-                return new XmlImporter();
-            } else if("application/rdf+xml".equals(contentType)) {
-            	return new RdfTripleImporter();
-            } else if ("application/marc".equals(contentType)) {
-                return new MarcImporter();
-            }
-        } else if (fileName != null) {
-            fileName = fileName.toLowerCase();
-            if (fileName.endsWith(".xls")) {
-                return new ExcelImporter(false); 
-            } else if (fileName.endsWith(".xlsx")) {
-                return new ExcelImporter(true); 
-            } else if (
-                    fileName.endsWith(".xml") ||
-                    fileName.endsWith(".atom") ||
-                    fileName.endsWith(".rss")
-                ) {
-                return new XmlImporter(); 
-            } else if (
-                    fileName.endsWith(".rdf")) {
-            	return new RdfTripleImporter();
-            } else if (
-                    fileName.endsWith(".mrc") || 
-                    fileName.endsWith(".marc") || 
-                    fileName.contains(".mrc.") || 
-                    fileName.contains(".marc.")
-                ) {
-                return new MarcImporter();
+    protected Importer guessImporter(String contentType, String fileName) {
+        for(Importer i : importers.values()){
+            if(i.canImportData(contentType, fileName)){
+                return i;
             }
         }
 
-        return new TsvCsvImporter();
+        return new TsvCsvImporter(); //default
     }
 }

src/main/java/com/metaweb/gridworks/importers/ExcelImporter.java

@@ -27,11 +27,7 @@ import com.metaweb.gridworks.model.Row;
 import com.metaweb.gridworks.model.Recon.Judgment;
 
 public class ExcelImporter implements Importer {
-    final protected boolean _xmlBased;
-    
-    public ExcelImporter(boolean xmlBased) {
-        _xmlBased = xmlBased;
-    }
+    protected boolean _xmlBased;
 
     public boolean takesReader() {
         return false;
@@ -247,4 +243,32 @@ public class ExcelImporter implements Importer {
             }
         }
     }
+
+    public boolean canImportData(String contentType, String fileName) {
+        if (contentType != null) {
+            contentType = contentType.toLowerCase().trim();
+            if ("application/msexcel".equals(contentType) ||
+                "application/x-msexcel".equals(contentType) ||
+                "application/x-ms-excel".equals(contentType) ||
+                "application/vnd.ms-excel".equals(contentType) ||
+                "application/x-excel".equals(contentType) ||
+                "application/xls".equals(contentType)) {
+                this._xmlBased = false;
+                return true;
+            } else if("application/x-xls".equals(contentType)) {
+                this._xmlBased = true;
+                return true;
+            }
+        } else if (fileName != null) {
+            fileName = fileName.toLowerCase();
+            if (fileName.endsWith(".xls")) {
+                this._xmlBased = false;
+                return true;
+            } else if (fileName.endsWith(".xlsx")) {
+                this._xmlBased = true;
+                return true;
+            }
+        }
+        return false;
+    }
 }

src/main/java/com/metaweb/gridworks/importers/Importer.java

@@ -11,4 +11,6 @@ public interface Importer {
 
     public void read(Reader reader, Project project, Properties options) throws Exception;
     public void read(InputStream inputStream, Project project, Properties options) throws Exception;
+
+    public boolean canImportData(String contentType, String fileName);
 }

src/main/java/com/metaweb/gridworks/importers/MarcImporter.java

@@ -75,4 +75,25 @@ public class MarcImporter implements Importer {
             tempFile.delete();
         }
     }
+
+    public boolean canImportData(String contentType, String fileName) {
+        if (contentType != null) {
+            contentType = contentType.toLowerCase().trim();
+
+            if ("application/marc".equals(contentType)) {
+                return true;
+            }
+        } else if (fileName != null) {
+            fileName = fileName.toLowerCase();
+            if (
+                    fileName.endsWith(".mrc") ||
+                    fileName.endsWith(".marc") ||
+                    fileName.contains(".mrc.") ||
+                    fileName.contains(".marc.")
+                ) {
+                return true;
+            }
+        }
+        return false;
+    }
 }

src/main/java/com/metaweb/gridworks/importers/RdfTripleImporter.java

@@ -120,4 +120,21 @@ public class RdfTripleImporter implements Importer{
         return true;
     }
 
+    public boolean canImportData(String contentType, String fileName) {
+        if (contentType != null) {
+            contentType = contentType.toLowerCase().trim();
+
+            if("application/rdf+xml".equals(contentType)) {
+                return true;
+            }
+        } else if (fileName != null) {
+            fileName = fileName.toLowerCase();
+            if (
+                    fileName.endsWith(".rdf")) {
+                return true;
+            }
+        }
+        return false;
+    }
+
 }

src/main/java/com/metaweb/gridworks/importers/TsvCsvImporter.java

@@ -2,6 +2,7 @@ package com.metaweb.gridworks.importers;
 
 import java.io.IOException;
 import java.io.InputStream;
+import java.io.InputStreamReader;
 import java.io.LineNumberReader;
 import java.io.Reader;
 import java.util.ArrayList;
@@ -151,10 +152,25 @@ public class TsvCsvImporter implements Importer {
     }
 
     public void read(InputStream inputStream, Project project, Properties options) throws Exception {
-        throw new UnsupportedOperationException();
+        read(new InputStreamReader(inputStream), project, options);
     }
 
     public boolean takesReader() {
         return true;
     }
+
+    public boolean canImportData(String contentType, String fileName) {
+        if (contentType != null) {
+            contentType = contentType.toLowerCase().trim();
+            return false;
+        } else if (fileName != null) {
+            fileName = fileName.toLowerCase();
+            if (fileName.endsWith(".tsv")) {
+                return true;
+            }else if (fileName.endsWith(".csv")){
+                return true;
+            }
+        }
+        return false;
+    }
 }

src/main/java/com/metaweb/gridworks/importers/XmlImporter.java

@@ -59,4 +59,27 @@ public class XmlImporter implements Importer {
         project.columnModel.update();
     }
 
+    public boolean canImportData(String contentType, String fileName) {
+        if (contentType != null) {
+            contentType = contentType.toLowerCase().trim();
+
+            if("application/xml".equals(contentType) ||
+                      "text/xml".equals(contentType) ||
+                      "application/rss+xml".equals(contentType) ||
+                      "application/atom+xml".equals(contentType)) {
+                return true;
+            }
+        } else if (fileName != null) {
+            fileName = fileName.toLowerCase();
+            if (
+                    fileName.endsWith(".xml") ||
+                    fileName.endsWith(".atom") ||
+                    fileName.endsWith(".rss")
+                ) {
+                return true;
+            }
+        }
+        return false;
+    }
+
 }