Changeset 6202

Timestamp:

2003-12-10T16:31:18+13:00 (20 years ago)

Author:

jmt12

Message:

As well as implementing a removeExtractedNodes method, also fixed error where not all metadata was removed when a file was

File:

• trunk/gli/src/org/greenstone/gatherer/msm/GDMDocument.java (modified) (9 diffs)

trunk/gli/src/org/greenstone/gatherer/msm/GDMDocument.java

@@ -170 +170 @@
                 // It appears that its possible that we can be asked to add the same metadata twice (especially after a copy action is cancelled then repeated). So we check if we have been asked to add exactly the same value twice.
                 if(sibling_metadata_element.getAttribute(StaticStrings.NAME_ATTRIBUTE).equals(metadata_element.getAttribute(StaticStrings.NAME_ATTRIBUTE))) {
-                    // Check the values and return if they are the same.
+                    // Check the values and return if they are the same.
                    if(metadata.getAbsoluteValue().equals(MSMUtils.getValue(sibling_metadata_element))) {
                       return;
@@ -184 +184 @@
         sibling_description_elements = null;
         if(will_accumulate || force_accumulate) { //mode.equals(ACCUMULATE)) {
-        metadata_element.setAttribute(MODE_ATTRIBUTE, ACCUMULATE);
+            metadata_element.setAttribute(MODE_ATTRIBUTE, ACCUMULATE);
         }
         // As we can't possibly store all the metadata in memory, nor can we ensure that the indexes written to file remain the same until the new time we look at this file, and to avoid having to open a rewrite every collection document whenever any value tree changes, I'm writing the value out as a full path string
@@ -250 +250 @@
     return base_document;
     }
+
     /** Get all of the metadata, including directory level, associated with this file. */
     public ArrayList getMetadata(String filename, boolean remove, ArrayList metadatum_so_far, File file, boolean append_folder_level) {
@@ -257 +258 @@
     public ArrayList getMetadata(String filename, boolean remove, ArrayList metadatum_so_far, File file, boolean append_folder_level, boolean purge) {
     Gatherer.println("Get metadata for " + filename);
+    Gatherer.println("append_folder_level: " + append_folder_level);
+    Gatherer.println("purge: " + purge);
     ArrayList metadatum = null;
+    ArrayList queued_for_removal = new ArrayList();
     if(metadatum_so_far == null) {
         metadatum = new ArrayList();
@@ -267 +271 @@
         // Retrieve the document element.
         Element directorymetadata_element = base_document.getDocumentElement();
-        // Iterate through the filesets, checking the FileName child element against    the target file's name using regular expression matching.
+        // Iterate through the filesets, checking the FileName child element against the target file's name using regular expression matching.
         NodeList fileset_elements = directorymetadata_element.getElementsByTagName(FILESET_ELEMENT);
         for(int i = 0; i < fileset_elements.getLength(); i++) {
@@ -368 +372 @@
                     // Having found our metadata check if the value from the xml matches the one from the gvaluenode. If not update it. This happens whenever hierarchy information is involved (indexes rapidly become obsolete).
                     // If remove was set, remove it. We can only remove pure file level metadata, or folder level iff we were asked for folder level.
+                    ///atherer.println("Have we been asked to remove the metadata: " + metadata);
+                    ///atherer.println("Given:");
+                    ///atherer.println("\tremove = " + remove);
+                    ///atherer.println("\tfilename = " + filename);
+                    ///atherer.println("\tfilename_text = " + filename_text + "?");
                     if(remove && ((filename != null && filename.matches(filename_text) && !filename_text.equals(DIRECTORY_FILENAME)) || (filename == null && filename_text.equals(DIRECTORY_FILENAME)))) {
-                        ///ystem.err.println("Removing " + metadata + " from " + file);
-                        description_element.removeChild(metadata_element);
-                        // Remove the description element if empty.
-                        if(!description_element.hasChildNodes()) {
-                        fileset_element.removeChild(description_element);
-                        }
+                        ///atherer.println("Yes! Queuing for Removal.");
+                        queued_for_removal.add(metadata_element);
                     }
                     else {
-                        //String current_value = metadata.getAbsoluteValue();
-                        //String current_value = Codec.transform(metadata.getValueNode().getFullPath(false), Codec.TEXT_TO_DOM);
+                        ///atherer.println("No. Updating.");
                         String current_value = metadata.getValueNode().getFullPath(false);
                         ///ystem.err.println("Checking the current mdv path: " + current_value);
@@ -399 +403 @@
                 metadata_elements = null;
                 }
+
+                // Now we remove any elements that have been queued for deletion
+                for(int a = 0; a < queued_for_removal.size(); a++) {
+                Element metadata_element = (Element) queued_for_removal.get(a);
+                description_element.removeChild(metadata_element);
+                up_to_date = false; 
+                }
+                queued_for_removal.clear();
+                queued_for_removal = null;
+
+                        // If the description_element no longer has any children remove it
+                NodeList metadata_elements = description_element.getElementsByTagName(StaticStrings.METADATA_ELEMENT);
+                NodeList extracted_elements = description_element.getElementsByTagName(StaticStrings.EXTRACTED_METADATA_ELEMENT);
+                if(metadata_elements.getLength() == 0 && extracted_elements.getLength() == 0) {
+                fileset_element.removeChild(description_element);
+                up_to_date = false;
+                }
                 description_element = null;
             }
@@ -410 +431 @@
         if(description_elements.getLength() == 0) {
             directorymetadata_element.removeChild(fileset_element);
+            up_to_date = false;
         }
         description_elements = null;
@@ -437 +459 @@
     return ((doctype_name.equals("GreenstoneDirectoryMetadata") && root_name.equals("GreenstoneDirectoryMetadata")) || (doctype_name.equals("DirectoryMetadata") && root_name.equals("DirectoryMetadata")));
     }
+
+    /** Remove all of the extracted metadata (XMetadata) from this document. */
+    public void removeExtractedMetadata() {
+    try {
+        Element document_element = base_document.getDocumentElement();
+        NodeList extracted_metadata_elements = document_element.getElementsByTagName(StaticStrings.EXTRACTED_METADATA_ELEMENT);
+        document_element = null;
+        for(int i = extracted_metadata_elements.getLength(); i != 0; i--) {
+        Element extracted_metadata_element = (Element) extracted_metadata_elements.item(i - 1);
+        String element_name = extracted_metadata_element.getAttribute(StaticStrings.NAME_ATTRIBUTE);
+        ElementWrapper element = Gatherer.c_man.getCollection().msm.getElement(element_name);
+        if(element != null) {
+            element.dec();
+        }
+        element = null;
+        ///ystem.err.println("Removing extracted metadata: " + element_name + "=" + MSMUtils.getValue(extracted_metadata_element));
+        element_name = null;
+        Node parent_node = extracted_metadata_element.getParentNode();
+        parent_node.removeChild(extracted_metadata_element);
+        parent_node = null;
+        extracted_metadata_element = null;
+        }
+        extracted_metadata_elements = null;
+        up_to_date = false;
+    }
+    catch(Exception exception) {
+        Gatherer.println("Exception in GDMDocument.removeExtractedMetadata() - unexpected");
+        Gatherer.printStackTrace(exception);
+    }
+    }
+
     /** Remove the given directory level metadata from this document. All directory level metadata is available under the FileSet with filename '.*'. There is at least one nasty case to consider, where the first overwriting metadata entry, of several with the same element, is removed. In this case the next entry must become overwrite to ensure proper inheritance. */
     public void removeMetadata(String filename, Metadata metadata) {
     Gatherer.println("Remove metadata: " + metadata + "\nFrom filename: " + filename);
     try {
-        boolean found = false;
-        boolean first_metadata_element_found = true;
-        boolean make_next_metadata_element_overwrite = false;
-        boolean remove_fileset = false;
-        // Retrieve the document element.
-        Element directorymetadata_element = base_document.getDocumentElement();
-        // Iterate through the filesets looking for the directory level one.
-        NodeList fileset_elements = directorymetadata_element.getElementsByTagName(FILESET_ELEMENT);
-        for(int i = 0; !found && i < fileset_elements.getLength(); i++) {
+        boolean found = false;
+        boolean first_metadata_element_found = true;
+        boolean make_next_metadata_element_overwrite = false;
+        boolean remove_fileset = false;
+        // Retrieve the document element.
+        Element directorymetadata_element = base_document.getDocumentElement();
+        // Iterate through the filesets looking for the directory level one.
+        NodeList fileset_elements = directorymetadata_element.getElementsByTagName(FILESET_ELEMENT);
+        for(int i = 0; !found && i < fileset_elements.getLength(); i++) {
         Element fileset_element = (Element) fileset_elements.item(i);
         NodeList filename_elements = fileset_element.getElementsByTagName(FILENAME_ELEMENT);
         for(int j = 0; !found && j < filename_elements.getLength(); j++) {
-            Element filename_element = (Element) filename_elements.item(j);
-            String filename_text = MSMUtils.getValue(filename_element);
-            if((filename != null && filename.matches(filename_text) && !filename.equals(DIRECTORY_FILENAME)) || (filename == null && filename_text.equals(DIRECTORY_FILENAME))) {
+            Element filename_element = (Element) filename_elements.item(j);
+            String filename_text = MSMUtils.getValue(filename_element);
+            if((filename != null && filename.matches(filename_text) && !filename.equals(DIRECTORY_FILENAME)) || (filename == null && filename_text.equals(DIRECTORY_FILENAME))) {
             // Retrieve the Metadata Element for this fileset, and iterate through them looking for the one which we are to remove.
             NodeList description_elements = fileset_element.getElementsByTagName("Description");
             for(int k = 0; !found && k < description_elements.getLength(); k++) {
-                Element description_element = (Element) description_elements.item(k);
-                // We have to do this for each type of metadata
-                for(int z = 0; z < ALL_METADATA_TYPES.length; z++) {
-                     NodeList metadata_elements = description_element.getElementsByTagName(ALL_METADATA_TYPES[z]);
-                     for(int l = 0; (!found || !make_next_metadata_element_overwrite) && l < metadata_elements.getLength(); l++) {
-                     Element metadata_element = (Element) metadata_elements.item(l);
-                     String element = metadata_element.getAttribute("name");
-                     String value = MSMUtils.getValue(metadata_element);
-                     // See if this is the metadata we wish to remove
-                     if(element.equals(metadata.getElement().getName())) {
-                         if(value.equals(metadata.getAbsoluteValue())) {
-                             // Remove it
-                             ///ystem.err.println("Remove " + element + "-" + value);
-                         description_element.removeChild(metadata_element);
-                         found = true;
-                         // If this was the first metadata with this element found, and it was set to overwrite, then we have to ensure that the next metadata with this element found (if any) is changed to be overwrite now.
-                         if(first_metadata_element_found && !metadata.accumulates()) {
-                             ///ystem.err.println("First of this element found!");
-                             make_next_metadata_element_overwrite = true;
-                         }
-                         }
-                         // If this was the first metadata we've found with the element of the one to be removed set first found to false.
-                         else if(first_metadata_element_found) {
-                         ///ystem.err.println("Found a matching element: " + element + "=" + value);
-                         first_metadata_element_found = false;
-                         }
-                         // Otherwise we should make this metadata overwrite as requested.
-                         else if(make_next_metadata_element_overwrite) {
-                         ///ystem.err.println("Changing to overwrite: " + element + "=" + value);
-                         metadata_element.setAttribute(MODE_ATTRIBUTE, "");
-                         }
-                     }
-                     value = null;
-                     element = null;
-                     metadata_element = null;
-                     }
-                     NodeList normal_metadata_elements = description_element.getElementsByTagName(ALL_METADATA_TYPES[0]);
-                     NodeList extracted_metadata_elements = description_element.getElementsByTagName(ALL_METADATA_TYPES[1]);
-                     // If we found it, removed it, and now the description tag has no children, mark the fileset for removal
-                     if(normal_metadata_elements.getLength() == 0 && extracted_metadata_elements.getLength() == 0) {
-                         remove_fileset = true;
-                     }
-                     extracted_metadata_elements = null;
-                     normal_metadata_elements = null;
-                     metadata_elements = null;
+                Element description_element = (Element) description_elements.item(k);
+                // We have to do this for each type of metadata
+                for(int z = 0; z < ALL_METADATA_TYPES.length; z++) {
+                NodeList metadata_elements = description_element.getElementsByTagName(ALL_METADATA_TYPES[z]);
+                for(int l = 0; (!found || !make_next_metadata_element_overwrite) && l < metadata_elements.getLength(); l++) {
+                    Element metadata_element = (Element) metadata_elements.item(l);
+                    String element = metadata_element.getAttribute("name");
+                    String value = MSMUtils.getValue(metadata_element);
+                    // See if this is the metadata we wish to remove
+                    if(element.equals(metadata.getElement().getName())) {
+                    if(value.equals(metadata.getValueNode().getFullPath(false))) {
+                        // Remove it
+                        ///ystem.err.println("Remove " + element + "-" + value);
+                        description_element.removeChild(metadata_element);
+                        found = true;
+                        // If this was the first metadata with this element found, and it was set to overwrite, then we have to ensure that the next metadata with this element found (if any) is changed to be overwrite now.
+                        if(first_metadata_element_found && !metadata.accumulates()) {
+                        ///ystem.err.println("First of this element found!");
+                        make_next_metadata_element_overwrite = true;
+                        }
+                    }
+                    // If this was the first metadata we've found with the element of the one to be removed set first found to false.
+                    else if(first_metadata_element_found) {
+                        ///ystem.err.println("Found a matching element: " + element + "=" + value);
+                        first_metadata_element_found = false;
+                    }
+                    // Otherwise we should make this metadata overwrite as requested.
+                    else if(make_next_metadata_element_overwrite) {
+                        ///ystem.err.println("Changing to overwrite: " + element + "=" + value);
+                        metadata_element.setAttribute(MODE_ATTRIBUTE, "");
+                    }
+                    }
+                    value = null;
+                    element = null;
+                    metadata_element = null;
                 }
-                description_element = null;
+                NodeList normal_metadata_elements = description_element.getElementsByTagName(ALL_METADATA_TYPES[0]);
+                NodeList extracted_metadata_elements = description_element.getElementsByTagName(ALL_METADATA_TYPES[1]);
+                // If we found it, removed it, and now the description tag has no children, mark the fileset for removal
+                if(normal_metadata_elements.getLength() == 0 && extracted_metadata_elements.getLength() == 0) {
+                    remove_fileset = true;
+                }
+                extracted_metadata_elements = null;
+                normal_metadata_elements = null;
+                metadata_elements = null;
+                }
+                description_element = null;
             }
             description_elements = null;
-            }
-            filename_text = null;
-            filename_element = null;
+            }
+            filename_text = null;
+            filename_element = null;
         }
         filename_elements = null;
         if(found && remove_fileset) {
-            directorymetadata_element.removeChild(fileset_element);
+            directorymetadata_element.removeChild(fileset_element);
         }
         fileset_element = null;
-        }
-        fileset_elements = null;
-        directorymetadata_element = null;
-        up_to_date = false;
+        }
+        fileset_elements = null;
+        directorymetadata_element = null;
+        up_to_date = false;
     }
     catch (Exception error) {
-        Gatherer.printStackTrace(error);
-    }
-    }
-
-    /** Change the up to date flag. */
-    public void setUpToDate(boolean up_to_date) {
+        Gatherer.printStackTrace(error);
+    }
+    }
+    
+    /** Change the up to date flag. 
+     * @param up_to_date true if the document on the filesystem is the same as the one in memory, false otherwise
+     */
+    public void setUpToDate(boolean up_to_date) {
     this.up_to_date = up_to_date;
-    }
+    }
 }