1 | package org.greenstone.gatherer.metadata;
|
---|
2 |
|
---|
3 |
|
---|
4 | import java.io.*;
|
---|
5 | import java.util.*;
|
---|
6 | import org.greenstone.gatherer.util.XMLTools;
|
---|
7 | import org.w3c.dom.*;
|
---|
8 |
|
---|
9 |
|
---|
10 | /** This class represents one metadata.xml file */
|
---|
11 | public class MetadataXMLFile
|
---|
12 | extends File
|
---|
13 | {
|
---|
14 | static final private String DESCRIPTION_ELEMENT = "Description";
|
---|
15 | static final private String DIRECTORY_FILENAME = ".*";
|
---|
16 | static final private String FILENAME_ELEMENT = "FileName";
|
---|
17 | static final private String FILESET_ELEMENT = "FileSet";
|
---|
18 | static final private String METADATA_ELEMENT = "Metadata";
|
---|
19 |
|
---|
20 |
|
---|
21 | public MetadataXMLFile(String metadata_xml_file_path)
|
---|
22 | {
|
---|
23 | super(metadata_xml_file_path);
|
---|
24 | }
|
---|
25 |
|
---|
26 |
|
---|
27 | public void addMetadata(File file, MetadataValue metadata_value)
|
---|
28 | {
|
---|
29 | // Parse the metadata.xml file
|
---|
30 | Document document = XMLTools.parseXMLFile(this);
|
---|
31 | if (document == null) {
|
---|
32 | System.err.println("Error: Could not parse metadata.xml file " + getAbsolutePath());
|
---|
33 | return;
|
---|
34 | }
|
---|
35 |
|
---|
36 | // Determine the file's path relative to the location of the metadata.xml file
|
---|
37 | File metadata_xml_file_directory = getParentFile();
|
---|
38 | String file_relative_path = file.getAbsolutePath().substring(metadata_xml_file_directory.getAbsolutePath().length());
|
---|
39 | if (file_relative_path.startsWith(File.separator)) {
|
---|
40 | file_relative_path = file_relative_path.substring(File.separator.length());
|
---|
41 | }
|
---|
42 |
|
---|
43 | // Convert the file path into a regular expression that will match it
|
---|
44 | String file_path_regexp = file_relative_path.replaceAll("\\.", "\\\\.");
|
---|
45 | if (file_relative_path.equals("")) {
|
---|
46 | file_path_regexp = DIRECTORY_FILENAME;
|
---|
47 | }
|
---|
48 |
|
---|
49 | // Square brackets need to be escaped because they are a special character in Greenstone
|
---|
50 | String metadata_value_string = metadata_value.getFullValue();
|
---|
51 | metadata_value_string = metadata_value_string.replaceAll("\\[", "[");
|
---|
52 | metadata_value_string = metadata_value_string.replaceAll("\\]", "]");
|
---|
53 |
|
---|
54 | // Create a new Metadata element to record this metadata
|
---|
55 | Element new_metadata_value_element = document.createElement(METADATA_ELEMENT);
|
---|
56 | new_metadata_value_element.setAttribute("name", metadata_value.getMetadataElement().getFullName());
|
---|
57 | new_metadata_value_element.setAttribute("mode", (metadata_value.isAccumulatingMetadata() ? "accumulate" : "override"));
|
---|
58 | new_metadata_value_element.appendChild(document.createTextNode(metadata_value_string));
|
---|
59 | boolean have_added_metadata = false;
|
---|
60 |
|
---|
61 | // Read all the FileSet elements in the file
|
---|
62 | NodeList fileset_elements_nodelist = document.getElementsByTagName(FILESET_ELEMENT);
|
---|
63 | for (int i = 0; i < fileset_elements_nodelist.getLength(); i++) {
|
---|
64 | Element current_fileset_element = (Element) fileset_elements_nodelist.item(i);
|
---|
65 |
|
---|
66 | // Check the FileName elements of the FileSet to see if we have a match
|
---|
67 | NodeList filename_elements_nodelist = current_fileset_element.getElementsByTagName(FILENAME_ELEMENT);
|
---|
68 | for (int j = 0; j < filename_elements_nodelist.getLength(); j++) {
|
---|
69 | Element current_filename_element = (Element) filename_elements_nodelist.item(j);
|
---|
70 | String current_filename_element_value = XMLTools.getElementTextValue(current_filename_element);
|
---|
71 |
|
---|
72 | // Only exact matches can be extended with new metadata
|
---|
73 | if (current_filename_element_value.equals(file_path_regexp)) {
|
---|
74 | // Append the new Metadata element to the Description element of this FileSet
|
---|
75 | Element description_element = (Element) current_fileset_element.getElementsByTagName(DESCRIPTION_ELEMENT).item(0);
|
---|
76 |
|
---|
77 | // Accumulating metadata: add at the end
|
---|
78 | if (metadata_value.isAccumulatingMetadata()) {
|
---|
79 | description_element.appendChild(new_metadata_value_element);
|
---|
80 | }
|
---|
81 | // Override metadata: add at the start (so it overrides inherited metadata without affecting other assigned metadata)
|
---|
82 | else {
|
---|
83 | description_element.insertBefore(new_metadata_value_element, description_element.getFirstChild());
|
---|
84 | }
|
---|
85 |
|
---|
86 | have_added_metadata = true;
|
---|
87 | break;
|
---|
88 | }
|
---|
89 | }
|
---|
90 | }
|
---|
91 |
|
---|
92 | // Check if the metadata was added to an existing FileSet
|
---|
93 | if (!have_added_metadata) {
|
---|
94 | // It wasn't, so create a new FileSet element for it
|
---|
95 | Element new_fileset_element = document.createElement(FILESET_ELEMENT);
|
---|
96 |
|
---|
97 | Element new_filename_element = document.createElement(FILENAME_ELEMENT);
|
---|
98 | new_filename_element.appendChild(document.createTextNode(file_path_regexp));
|
---|
99 | new_fileset_element.appendChild(new_filename_element);
|
---|
100 |
|
---|
101 | // Append the new Metadata element to the Description element of this FileSet
|
---|
102 | Element new_description_element = document.createElement(DESCRIPTION_ELEMENT);
|
---|
103 | new_description_element.appendChild(new_metadata_value_element);
|
---|
104 | new_fileset_element.appendChild(new_description_element);
|
---|
105 |
|
---|
106 | document.getDocumentElement().appendChild(new_fileset_element);
|
---|
107 | }
|
---|
108 |
|
---|
109 | // Rewrite the metadata.xml file
|
---|
110 | XMLTools.writeXMLFile(this, document);
|
---|
111 | }
|
---|
112 |
|
---|
113 |
|
---|
114 | public ArrayList getMetadataAssignedToFile(File file)
|
---|
115 | {
|
---|
116 | // Parse the metadata.xml file
|
---|
117 | Document document = XMLTools.parseXMLFile(this);
|
---|
118 | if (document == null) {
|
---|
119 | System.err.println("Error: Could not parse metadata.xml file " + getAbsolutePath());
|
---|
120 | return null;
|
---|
121 | }
|
---|
122 |
|
---|
123 | // Determine the file's path relative to the location of the metadata.xml file
|
---|
124 | File metadata_xml_file_directory = getParentFile();
|
---|
125 | String file_relative_path = file.getAbsolutePath().substring(metadata_xml_file_directory.getAbsolutePath().length());
|
---|
126 | if (file_relative_path.startsWith(File.separator)) {
|
---|
127 | file_relative_path = file_relative_path.substring(File.separator.length());
|
---|
128 | }
|
---|
129 |
|
---|
130 | // Build up a list of metadata assigned to this file
|
---|
131 | ArrayList metadata_values = new ArrayList();
|
---|
132 |
|
---|
133 | // Read all the FileSet elements in the file
|
---|
134 | NodeList fileset_elements_nodelist = document.getElementsByTagName(FILESET_ELEMENT);
|
---|
135 | for (int i = 0; i < fileset_elements_nodelist.getLength(); i++) {
|
---|
136 | Element current_fileset_element = (Element) fileset_elements_nodelist.item(i);
|
---|
137 | boolean current_fileset_matches = false;
|
---|
138 | File folder_metadata_inherited_from = null;
|
---|
139 |
|
---|
140 | // Check the FileName elements of the FileSet to see if we have a match
|
---|
141 | NodeList filename_elements_nodelist = current_fileset_element.getElementsByTagName(FILENAME_ELEMENT);
|
---|
142 | for (int j = 0; j < filename_elements_nodelist.getLength(); j++) {
|
---|
143 | Element current_filename_element = (Element) filename_elements_nodelist.item(j);
|
---|
144 | String current_filename_element_value = XMLTools.getElementTextValue(current_filename_element);
|
---|
145 |
|
---|
146 | // This fileset specifies metadata for the file
|
---|
147 | if (file_relative_path.matches(current_filename_element_value)) {
|
---|
148 | current_fileset_matches = true;
|
---|
149 | if (!file_relative_path.equals("") && current_filename_element_value.equals(DIRECTORY_FILENAME)) {
|
---|
150 | folder_metadata_inherited_from = metadata_xml_file_directory;
|
---|
151 | }
|
---|
152 | break;
|
---|
153 | }
|
---|
154 |
|
---|
155 | // This fileset specifies metadata for the folder the file is in
|
---|
156 | if (file_relative_path.startsWith(current_filename_element_value + File.separator)) {
|
---|
157 | current_fileset_matches = true;
|
---|
158 | folder_metadata_inherited_from = new File(metadata_xml_file_directory, current_filename_element_value);
|
---|
159 | break;
|
---|
160 | }
|
---|
161 | }
|
---|
162 |
|
---|
163 | // The FileSet doesn't apply, so move onto the next one
|
---|
164 | if (current_fileset_matches == false) {
|
---|
165 | continue;
|
---|
166 | }
|
---|
167 |
|
---|
168 | // Read all the Metadata elements in the fileset
|
---|
169 | NodeList metadata_elements_nodelist = current_fileset_element.getElementsByTagName(METADATA_ELEMENT);
|
---|
170 | for (int k = 0; k < metadata_elements_nodelist.getLength(); k++) {
|
---|
171 | Element current_metadata_element = (Element) metadata_elements_nodelist.item(k);
|
---|
172 | String metadata_element_name_full = current_metadata_element.getAttribute("name");
|
---|
173 |
|
---|
174 | String metadata_set_namespace = MetadataTools.getMetadataSetNamespace(metadata_element_name_full);
|
---|
175 | MetadataSet metadata_set = MetadataSetManager.getMetadataSet(metadata_set_namespace);
|
---|
176 |
|
---|
177 | // If the metadata set isn't loaded give the option of mapping the element into a loaded set
|
---|
178 | if (metadata_set == null) {
|
---|
179 | String target_metadata_element_name_full = MetadataSetManager.mapUnloadedMetadataElement(metadata_element_name_full);
|
---|
180 | if (target_metadata_element_name_full == null || target_metadata_element_name_full.equals("")) {
|
---|
181 | // Skip this element if we still don't have a loaded element for it
|
---|
182 | continue;
|
---|
183 | }
|
---|
184 |
|
---|
185 | metadata_element_name_full = target_metadata_element_name_full;
|
---|
186 | metadata_set_namespace = MetadataTools.getMetadataSetNamespace(metadata_element_name_full);
|
---|
187 | metadata_set = MetadataSetManager.getMetadataSet(metadata_set_namespace);
|
---|
188 | }
|
---|
189 |
|
---|
190 | String metadata_element_name = MetadataTools.getMetadataElementName(metadata_element_name_full);
|
---|
191 | MetadataElement metadata_element = metadata_set.getMetadataElement(metadata_element_name);
|
---|
192 |
|
---|
193 | // If the element doesn't exist in the metadata set, we're not interested
|
---|
194 | if (metadata_element == null) {
|
---|
195 | continue;
|
---|
196 | }
|
---|
197 |
|
---|
198 | // Square brackets need to be escaped because they are a special character in Greenstone
|
---|
199 | String metadata_element_value = XMLTools.getElementTextValue(current_metadata_element);
|
---|
200 | metadata_element_value = metadata_element_value.replaceAll("[", "[");
|
---|
201 | metadata_element_value = metadata_element_value.replaceAll("]", "]");
|
---|
202 |
|
---|
203 | MetadataValueTreeNode metadata_value_tree_node = metadata_element.getMetadataValueTreeNode(metadata_element_value);
|
---|
204 |
|
---|
205 | // If there is no metadata value tree node for this value, create it
|
---|
206 | if (metadata_value_tree_node == null) {
|
---|
207 | System.err.println("Note: No value tree node for metadata value \"" + metadata_element_value + "\"");
|
---|
208 | metadata_element.addMetadataValue(metadata_element_value);
|
---|
209 | metadata_value_tree_node = metadata_element.getMetadataValueTreeNode(metadata_element_value);
|
---|
210 | }
|
---|
211 |
|
---|
212 | MetadataValue metadata_value = new MetadataValue(metadata_element, metadata_value_tree_node);
|
---|
213 | metadata_value.inheritsMetadataFromFolder(folder_metadata_inherited_from);
|
---|
214 |
|
---|
215 | // Is this accumulating metadata?
|
---|
216 | if (current_metadata_element.getAttribute("mode").equals("accumulate")) {
|
---|
217 | metadata_value.setIsAccumulatingMetadata(true);
|
---|
218 | }
|
---|
219 |
|
---|
220 | // Add the new metadata value to the list
|
---|
221 | metadata_values.add(metadata_value);
|
---|
222 | }
|
---|
223 | }
|
---|
224 |
|
---|
225 | return metadata_values;
|
---|
226 | }
|
---|
227 |
|
---|
228 |
|
---|
229 | public void removeMetadata(File file, MetadataValue metadata_value)
|
---|
230 | {
|
---|
231 | // Parse the metadata.xml file
|
---|
232 | Document document = XMLTools.parseXMLFile(this);
|
---|
233 | if (document == null) {
|
---|
234 | System.err.println("Error: Could not parse metadata.xml file " + getAbsolutePath());
|
---|
235 | return;
|
---|
236 | }
|
---|
237 |
|
---|
238 | // Determine the file's path relative to the location of the metadata.xml file
|
---|
239 | File metadata_xml_file_directory = getParentFile();
|
---|
240 | String file_relative_path = file.getAbsolutePath().substring(metadata_xml_file_directory.getAbsolutePath().length());
|
---|
241 | if (file_relative_path.startsWith(File.separator)) {
|
---|
242 | file_relative_path = file_relative_path.substring(File.separator.length());
|
---|
243 | }
|
---|
244 |
|
---|
245 | // Convert the file path into a regular expression that will match it
|
---|
246 | String file_path_regexp = file_relative_path.replaceAll("\\.", "\\\\.");
|
---|
247 | if (file_relative_path.equals("")) {
|
---|
248 | file_path_regexp = DIRECTORY_FILENAME;
|
---|
249 | }
|
---|
250 |
|
---|
251 | // Read all the FileSet elements in the file
|
---|
252 | NodeList fileset_elements_nodelist = document.getElementsByTagName(FILESET_ELEMENT);
|
---|
253 | for (int i = 0; i < fileset_elements_nodelist.getLength(); i++) {
|
---|
254 | Element current_fileset_element = (Element) fileset_elements_nodelist.item(i);
|
---|
255 | boolean current_fileset_matches = false;
|
---|
256 |
|
---|
257 | // Check the FileName elements of the FileSet to see if we have a match
|
---|
258 | NodeList filename_elements_nodelist = current_fileset_element.getElementsByTagName(FILENAME_ELEMENT);
|
---|
259 | for (int j = 0; j < filename_elements_nodelist.getLength(); j++) {
|
---|
260 | Element current_filename_element = (Element) filename_elements_nodelist.item(j);
|
---|
261 | String current_filename_element_value = XMLTools.getElementTextValue(current_filename_element);
|
---|
262 |
|
---|
263 | // Only exact matches can be edited
|
---|
264 | if (current_filename_element_value.equals(file_path_regexp)) {
|
---|
265 | current_fileset_matches = true;
|
---|
266 | break;
|
---|
267 | }
|
---|
268 | }
|
---|
269 |
|
---|
270 | // The FileSet doesn't apply, so move onto the next one
|
---|
271 | if (current_fileset_matches == false) {
|
---|
272 | continue;
|
---|
273 | }
|
---|
274 |
|
---|
275 | // Find the Metadata element to delete from the fileset
|
---|
276 | String metadata_element_name_full = metadata_value.getMetadataElement().getFullName();
|
---|
277 | String metadata_element_value = metadata_value.getFullValue();
|
---|
278 | NodeList metadata_elements_nodelist = current_fileset_element.getElementsByTagName(METADATA_ELEMENT);
|
---|
279 | for (int k = 0; k < metadata_elements_nodelist.getLength(); k++) {
|
---|
280 | Element current_metadata_element = (Element) metadata_elements_nodelist.item(k);
|
---|
281 |
|
---|
282 | // Check the metadata element name matches
|
---|
283 | String current_metadata_element_name_full = current_metadata_element.getAttribute("name");
|
---|
284 | if (!current_metadata_element_name_full.equals(metadata_element_name_full)) {
|
---|
285 | continue;
|
---|
286 | }
|
---|
287 |
|
---|
288 | // Check the metadata element value matches
|
---|
289 | String current_metadata_element_value = XMLTools.getElementTextValue(current_metadata_element);
|
---|
290 | if (!current_metadata_element_value.equals(metadata_element_value)) {
|
---|
291 | continue;
|
---|
292 | }
|
---|
293 |
|
---|
294 | // Remove this Metadata element
|
---|
295 | current_metadata_element.getParentNode().removeChild(current_metadata_element);
|
---|
296 | }
|
---|
297 | }
|
---|
298 |
|
---|
299 | // Rewrite the metadata.xml file
|
---|
300 | XMLTools.writeXMLFile(this, document);
|
---|
301 | }
|
---|
302 |
|
---|
303 |
|
---|
304 | public void skimFile()
|
---|
305 | {
|
---|
306 | boolean file_changed = false;
|
---|
307 |
|
---|
308 | // Parse the metadata.xml file
|
---|
309 | System.err.println("Skimming metadata.xml file " + this + "...");
|
---|
310 | Document document = XMLTools.parseXMLFile(this);
|
---|
311 | if (document == null) {
|
---|
312 | System.err.println("Error: Could not parse metadata.xml file " + getAbsolutePath());
|
---|
313 | return;
|
---|
314 | }
|
---|
315 |
|
---|
316 | // Read all the Metadata elements in the file
|
---|
317 | NodeList metadata_elements_nodelist = document.getElementsByTagName(METADATA_ELEMENT);
|
---|
318 | for (int i = 0; i < metadata_elements_nodelist.getLength(); i++) {
|
---|
319 | Element current_metadata_element = (Element) metadata_elements_nodelist.item(i);
|
---|
320 | String metadata_element_name_full = current_metadata_element.getAttribute("name");
|
---|
321 |
|
---|
322 | String metadata_set_namespace = MetadataTools.getMetadataSetNamespace(metadata_element_name_full);
|
---|
323 | MetadataSet metadata_set = MetadataSetManager.getMetadataSet(metadata_set_namespace);
|
---|
324 |
|
---|
325 | // If the metadata set isn't loaded give the option of mapping the element into a loaded set
|
---|
326 | if (metadata_set == null) {
|
---|
327 | String target_metadata_element_name_full = MetadataSetManager.mapUnloadedMetadataElement(metadata_element_name_full);
|
---|
328 | if (target_metadata_element_name_full == null || target_metadata_element_name_full.equals("")) {
|
---|
329 | // Skip this element if we still don't have a loaded element for it
|
---|
330 | continue;
|
---|
331 | }
|
---|
332 |
|
---|
333 | // Update the metadata.xml file to have the new element name
|
---|
334 | current_metadata_element.setAttribute("name", target_metadata_element_name_full);
|
---|
335 | file_changed = true;
|
---|
336 |
|
---|
337 | metadata_element_name_full = target_metadata_element_name_full;
|
---|
338 | metadata_set_namespace = MetadataTools.getMetadataSetNamespace(metadata_element_name_full);
|
---|
339 | metadata_set = MetadataSetManager.getMetadataSet(metadata_set_namespace);
|
---|
340 | }
|
---|
341 |
|
---|
342 | String metadata_element_name = MetadataTools.getMetadataElementName(metadata_element_name_full);
|
---|
343 | MetadataElement metadata_element = metadata_set.getMetadataElement(metadata_element_name);
|
---|
344 |
|
---|
345 | // If the element doesn't exist in the metadata set, add it
|
---|
346 | if (metadata_element == null) {
|
---|
347 | metadata_element = metadata_set.addMetadataElementForThisSession(metadata_element_name);
|
---|
348 | }
|
---|
349 |
|
---|
350 | String metadata_element_value = XMLTools.getElementTextValue(current_metadata_element);
|
---|
351 | metadata_element.addMetadataValue(metadata_element_value);
|
---|
352 | }
|
---|
353 |
|
---|
354 | // Rewrite the metadata.xml file if it has changed
|
---|
355 | if (file_changed) {
|
---|
356 | XMLTools.writeXMLFile(this, document);
|
---|
357 | }
|
---|
358 | }
|
---|
359 | }
|
---|