1 | /**
|
---|
2 | *############################################################################
|
---|
3 | * A component of the Greenstone Librarian Interface, part of the Greenstone
|
---|
4 | * digital library suite from the New Zealand Digital Library Project at the
|
---|
5 | * University of Waikato, New Zealand.
|
---|
6 | *
|
---|
7 | * Author: Michael Dewsnip, NZDL Project, University of Waikato, NZ
|
---|
8 | *
|
---|
9 | * Copyright (C) 2004 New Zealand Digital Library Project
|
---|
10 | *
|
---|
11 | * This program is free software; you can redistribute it and/or modify
|
---|
12 | * it under the terms of the GNU General Public License as published by
|
---|
13 | * the Free Software Foundation; either version 2 of the License, or
|
---|
14 | * (at your option) any later version.
|
---|
15 | *
|
---|
16 | * This program is distributed in the hope that it will be useful,
|
---|
17 | * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
18 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
19 | * GNU General Public License for more details.
|
---|
20 | *
|
---|
21 | * You should have received a copy of the GNU General Public License
|
---|
22 | * along with this program; if not, write to the Free Software
|
---|
23 | * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
---|
24 | *############################################################################
|
---|
25 | */
|
---|
26 |
|
---|
27 | package org.greenstone.gatherer.metadata;
|
---|
28 |
|
---|
29 |
|
---|
30 | import java.io.*;
|
---|
31 | import java.util.*;
|
---|
32 | import java.net.URLDecoder;
|
---|
33 | import org.greenstone.gatherer.DebugStream;
|
---|
34 | import org.greenstone.gatherer.util.Utility;
|
---|
35 |
|
---|
36 |
|
---|
37 | /** This class represents one doc.xml file */
|
---|
38 |
|
---|
39 | public abstract class DocXMLFile extends File
|
---|
40 | {
|
---|
41 | protected HashMap source_file_name_to_description_elements_mapping = new HashMap();
|
---|
42 |
|
---|
43 | protected final String MetadataWrap;
|
---|
44 | protected final String MetadataItem;
|
---|
45 |
|
---|
46 | public DocXMLFile(String doc_xml_file_path, String metaWrap, String metaItem)
|
---|
47 | {
|
---|
48 | super(doc_xml_file_path);
|
---|
49 | this.MetadataWrap = metaWrap;
|
---|
50 | this.MetadataItem = metaItem;
|
---|
51 | }
|
---|
52 |
|
---|
53 |
|
---|
54 | public ArrayList getMetadataExtractedFromFile(File file)
|
---|
55 | {
|
---|
56 | // Build up a list of metadata extracted from this file
|
---|
57 | ArrayList metadata_values = new ArrayList();
|
---|
58 |
|
---|
59 | String file_relative_path = file.getAbsolutePath();
|
---|
60 | int import_index = file_relative_path.indexOf("import");
|
---|
61 | if (import_index != -1) {
|
---|
62 | file_relative_path = file_relative_path.substring(import_index + "import".length() + 1);
|
---|
63 | }
|
---|
64 |
|
---|
65 | // Check whether this file (i.e. doc.xml or docmets.xml on inheritance) file contains extracted metadata for the specified file
|
---|
66 | ArrayList description_elements_list = (ArrayList) source_file_name_to_description_elements_mapping.get(file_relative_path);
|
---|
67 | if (description_elements_list == null) {
|
---|
68 | // ...it doesn't
|
---|
69 | return metadata_values;
|
---|
70 | }
|
---|
71 |
|
---|
72 | MetadataSet extracted_metadata_set = MetadataSetManager.getMetadataSet(MetadataSetManager.EXTRACTED_METADATA_NAMESPACE);
|
---|
73 |
|
---|
74 | // Parse the file
|
---|
75 | DebugStream.println("Applicable file: " + this);
|
---|
76 | try {
|
---|
77 | BufferedReader buffered_reader = new BufferedReader(new FileReader(this));
|
---|
78 |
|
---|
79 | int description_element_num = 0;
|
---|
80 | int next_description_element_start = ((Integer) description_elements_list.get(description_element_num)).intValue();
|
---|
81 | boolean in_relevant_description_element = false;
|
---|
82 |
|
---|
83 | String line = null;
|
---|
84 | for (int line_num = 0; (line = buffered_reader.readLine()) != null; line_num++) {
|
---|
85 | // Check if this line contains the start of a relevant "Description" element
|
---|
86 | // (mets:xmlData in METS parlance, Description in GreenstoneArchive format)
|
---|
87 | if (line_num == next_description_element_start) {
|
---|
88 | in_relevant_description_element = true;
|
---|
89 | continue;
|
---|
90 | }
|
---|
91 |
|
---|
92 | // If we're not in a relevant Description element we don't care about anything
|
---|
93 | if (in_relevant_description_element == false) {
|
---|
94 | continue;
|
---|
95 | }
|
---|
96 |
|
---|
97 | // Check if this line contains the end of the relevant Description element
|
---|
98 | if (line.indexOf("</"+MetadataWrap+">") != -1) {
|
---|
99 | description_element_num++;
|
---|
100 | if (description_element_num == description_elements_list.size()) {
|
---|
101 | break;
|
---|
102 | }
|
---|
103 |
|
---|
104 | next_description_element_start = ((Integer) description_elements_list.get(description_element_num)).intValue();
|
---|
105 | in_relevant_description_element = false;
|
---|
106 | continue;
|
---|
107 | }
|
---|
108 |
|
---|
109 | // If this line doesn't contain a complete Metadata element, we're not interested
|
---|
110 | if (line.indexOf("<"+MetadataItem+" ") == -1 || line.indexOf("</"+MetadataItem+">") == -1) {
|
---|
111 | continue;
|
---|
112 | }
|
---|
113 |
|
---|
114 | // Extract the metadata element name
|
---|
115 | int name_index = line.indexOf(" name=\"") + " name=\"".length();
|
---|
116 | String metadata_element_name_full = line.substring(name_index, line.indexOf("\"", name_index));
|
---|
117 |
|
---|
118 | // If the metadata has a namespace it isn't extracted metadata, so we're not interested
|
---|
119 | String metadata_set_namespace = MetadataTools.getMetadataSetNamespace(metadata_element_name_full);
|
---|
120 | if (!metadata_set_namespace.equals("")) {
|
---|
121 | continue;
|
---|
122 | }
|
---|
123 |
|
---|
124 | // Extracted metadata!
|
---|
125 | String metadata_element_name = metadata_element_name_full;
|
---|
126 |
|
---|
127 | // We completely ignore bibliographic data
|
---|
128 | if (metadata_element_name.equals("SourceSegment")) {
|
---|
129 | buffered_reader.close();
|
---|
130 | return new ArrayList();
|
---|
131 | }
|
---|
132 |
|
---|
133 | // Ignore metadata starting with gsdl (used to be lowercase metadata and /metadata)
|
---|
134 | if (metadata_element_name.startsWith("gsdl")) {
|
---|
135 | continue;
|
---|
136 | }
|
---|
137 |
|
---|
138 | MetadataElement metadata_element = extracted_metadata_set.getMetadataElementWithName(metadata_element_name);
|
---|
139 |
|
---|
140 | // Value trees are not stored for extracted metadata, so create a new value tree node now
|
---|
141 | int value_index = line.indexOf(">", name_index) + ">".length();
|
---|
142 | String metadata_element_value = line.substring(value_index, line.lastIndexOf("</"+MetadataItem+">"));
|
---|
143 |
|
---|
144 | metadata_element.addMetadataValue(metadata_element_value);
|
---|
145 | MetadataValueTreeNode metadata_value_tree_node = metadata_element.getMetadataValueTreeNode(metadata_element_value);
|
---|
146 |
|
---|
147 | // Add the new metadata value to the list
|
---|
148 | MetadataValue metadata_value = new MetadataValue(metadata_element, metadata_value_tree_node);
|
---|
149 | metadata_values.add(metadata_value);
|
---|
150 | }
|
---|
151 |
|
---|
152 | buffered_reader.close();
|
---|
153 | }
|
---|
154 | catch (FileNotFoundException exception) {
|
---|
155 | DebugStream.printStackTrace(exception);
|
---|
156 | }
|
---|
157 | catch (IOException exception) {
|
---|
158 | DebugStream.printStackTrace(exception);
|
---|
159 | }
|
---|
160 |
|
---|
161 | return metadata_values;
|
---|
162 | }
|
---|
163 |
|
---|
164 |
|
---|
165 |
|
---|
166 |
|
---|
167 | /**
|
---|
168 | * Every file must be skimmed when a collection is opened, for two reasons:
|
---|
169 | * - To build a mapping from source file to its corresponding doc.xml file
|
---|
170 | * - To get a complete list of all extracted metadata elements
|
---|
171 | */
|
---|
172 | public void skimFile()
|
---|
173 | {
|
---|
174 | MetadataSet extracted_metadata_set = MetadataSetManager.getMetadataSet(MetadataSetManager.EXTRACTED_METADATA_NAMESPACE);
|
---|
175 |
|
---|
176 | // Skim the file as quickly as possible (don't parse as XML), looking at the Metadata elements
|
---|
177 | DebugStream.println("Skimming " + this + "...");
|
---|
178 | try {
|
---|
179 | BufferedReader buffered_reader = new BufferedReader(new FileReader(this));
|
---|
180 | int description_element_start = -1;
|
---|
181 |
|
---|
182 | String line = null;
|
---|
183 | for (int line_num = 0; (line = buffered_reader.readLine()) != null; line_num++) {
|
---|
184 | // This line contains the start of a "MetadataWrap" element
|
---|
185 | // (mets:xmlData in METS parlance, Description in GreenstoneArchive format)
|
---|
186 | if (line.indexOf("<"+MetadataWrap+">") != -1) {
|
---|
187 | if (description_element_start != -1) {
|
---|
188 | System.err.println("Parse error: previous " + MetadataWrap + " element unfinished!");
|
---|
189 | }
|
---|
190 | description_element_start = line_num;
|
---|
191 | continue;
|
---|
192 | }
|
---|
193 |
|
---|
194 | // This line contains the end of a "MetadataWrap" element
|
---|
195 | if (line.indexOf("</"+MetadataWrap+">") != -1) {
|
---|
196 | if (description_element_start == -1) {
|
---|
197 | System.err.println("Parse error: "+MetadataWrap+" element unstarted!");
|
---|
198 | }
|
---|
199 | description_element_start = -1;
|
---|
200 | continue;
|
---|
201 | }
|
---|
202 |
|
---|
203 | // If we're not in a"MetadataWrap" element there shouldn't be any Metadata elements
|
---|
204 | if (description_element_start == -1) {
|
---|
205 | continue;
|
---|
206 | }
|
---|
207 |
|
---|
208 | // This line doesn't contain a Metadata element, so we're not interested
|
---|
209 | if (line.indexOf("<"+MetadataItem+" ") == -1) {
|
---|
210 | DebugStream.println("Warning: "+MetadataWrap+" element line doesn't contain Metadata element.");
|
---|
211 | continue;
|
---|
212 | }
|
---|
213 |
|
---|
214 | // Extract the metadata element name
|
---|
215 | int name_index = line.indexOf(" name=\"") + " name=\"".length();
|
---|
216 | String metadata_element_name_full = line.substring(name_index, line.indexOf("\"", name_index));
|
---|
217 |
|
---|
218 | // If the metadata has a namespace it isn't extracted metadata, so we're not interested
|
---|
219 | String metadata_set_namespace = MetadataTools.getMetadataSetNamespace(metadata_element_name_full);
|
---|
220 | if (!metadata_set_namespace.equals("")) {
|
---|
221 | continue;
|
---|
222 | }
|
---|
223 |
|
---|
224 | // Extracted metadata!
|
---|
225 | String metadata_element_name = metadata_element_name_full;
|
---|
226 |
|
---|
227 | // Note which file this is for
|
---|
228 | if (metadata_element_name.equals("gsdlsourcefilename")) {
|
---|
229 | // Extract the gsdlsourcefilename element value
|
---|
230 | int value_index = line.indexOf(">", name_index) + ">".length();
|
---|
231 | String gsdlsourcefilename_value = line.substring(value_index, line.indexOf("<", value_index));
|
---|
232 |
|
---|
233 | // We're only interested in the path relative to the import folder
|
---|
234 | int import_index = gsdlsourcefilename_value.indexOf("import");
|
---|
235 | if (import_index != -1) {
|
---|
236 | gsdlsourcefilename_value = gsdlsourcefilename_value.substring(import_index + "import".length());
|
---|
237 |
|
---|
238 | boolean is_unix_path = gsdlsourcefilename_value.startsWith("/");
|
---|
239 | gsdlsourcefilename_value = gsdlsourcefilename_value.substring(1);
|
---|
240 |
|
---|
241 | // URL decode gsdlsourcefilename. Need to set the decoder to use the default system encoding
|
---|
242 | // This is stored in the System's file.encoding property.
|
---|
243 | gsdlsourcefilename_value = URLDecoder.decode(gsdlsourcefilename_value, System.getProperty("file.encoding"));
|
---|
244 |
|
---|
245 | // Make sure the path matches the OS that is running
|
---|
246 | if (is_unix_path && Utility.isWindows()) {
|
---|
247 | // Convert path from Unix to Windows
|
---|
248 | gsdlsourcefilename_value = gsdlsourcefilename_value.replaceAll("\\/", "\\\\");
|
---|
249 | }
|
---|
250 | else if (!is_unix_path && !Utility.isWindows()) {
|
---|
251 | // Convert path from Windows to Unix
|
---|
252 | gsdlsourcefilename_value = gsdlsourcefilename_value.replaceAll("\\\\", "/");
|
---|
253 | }
|
---|
254 |
|
---|
255 | // Remember this for quick access later
|
---|
256 | if (source_file_name_to_description_elements_mapping.get(gsdlsourcefilename_value) == null) {
|
---|
257 | source_file_name_to_description_elements_mapping.put(gsdlsourcefilename_value, new ArrayList());
|
---|
258 | }
|
---|
259 |
|
---|
260 | ((ArrayList) source_file_name_to_description_elements_mapping.get(gsdlsourcefilename_value)).add(new Integer(description_element_start));
|
---|
261 | }
|
---|
262 |
|
---|
263 | // Warn about an odd gsdlsourcefilename, except if it is the Greenstone "tmp" directory or
|
---|
264 | // (as in the case of using FLI) if it is the etc/collect.cfg or etc/collectionConfig.xml file
|
---|
265 | // which are the gsdlsourcefilenames for the fedora digital object representing a collection.
|
---|
266 | // This (tmp dir) is true when the source files come from a zip file processed by ZIPPlug, for example
|
---|
267 | else if (gsdlsourcefilename_value.indexOf("tmp") == -1
|
---|
268 | && !gsdlsourcefilename_value.endsWith("collect.cfg")
|
---|
269 | && !gsdlsourcefilename_value.endsWith("collectionConfig.xml")) {
|
---|
270 | // We don't really know what is going on...
|
---|
271 | System.err.println("Warning: Could not understand gsdlsourcefilename " + gsdlsourcefilename_value);
|
---|
272 | }
|
---|
273 | }
|
---|
274 |
|
---|
275 | // Ignore metadata starting with gsdl (used to be lowercase metadata and /metadata)
|
---|
276 | if (metadata_element_name.startsWith("gsdl")) {
|
---|
277 | continue;
|
---|
278 | }
|
---|
279 |
|
---|
280 | MetadataElement metadata_element = extracted_metadata_set.getMetadataElementWithName(metadata_element_name);
|
---|
281 | if (metadata_element == null) {
|
---|
282 | // This element isn't defined in ex.mds, so create it for this session
|
---|
283 | DebugStream.println("Extracted metadata element not defined: " + metadata_element_name);
|
---|
284 | extracted_metadata_set.addMetadataElementForThisSession(metadata_element_name);
|
---|
285 | }
|
---|
286 | }
|
---|
287 |
|
---|
288 | buffered_reader.close();
|
---|
289 | }
|
---|
290 | catch (FileNotFoundException exception) {
|
---|
291 | DebugStream.printStackTrace(exception);
|
---|
292 | }
|
---|
293 | catch (IOException exception) {
|
---|
294 | DebugStream.printStackTrace(exception);
|
---|
295 | }
|
---|
296 | }
|
---|
297 |
|
---|
298 |
|
---|
299 | /*
|
---|
300 | public ArrayList getMetadataExtractedFromFile(File file)
|
---|
301 | {
|
---|
302 | // Build up a list of metadata extracted from this file
|
---|
303 | ArrayList metadata_values = new ArrayList();
|
---|
304 |
|
---|
305 | String file_relative_path = file.getAbsolutePath();
|
---|
306 | int import_index = file_relative_path.indexOf("import");
|
---|
307 | if (import_index != -1) {
|
---|
308 | file_relative_path = file_relative_path.substring(import_index + "import".length() + 1);
|
---|
309 | }
|
---|
310 |
|
---|
311 | // Check whether this doc.xml file contains extracted metadata for the specified file
|
---|
312 | ArrayList description_elements_list = (ArrayList) source_file_name_to_description_elements_mapping.get(file_relative_path);
|
---|
313 | if (description_elements_list == null) {
|
---|
314 | // ...it doesn't
|
---|
315 | return metadata_values;
|
---|
316 | }
|
---|
317 |
|
---|
318 | MetadataSet extracted_metadata_set = MetadataSetManager.getMetadataSet(MetadataSetManager.EXTRACTED_METADATA_NAMESPACE);
|
---|
319 |
|
---|
320 | // Parse the doc.xml file
|
---|
321 | DebugStream.println("Applicable doc.xml file: " + this);
|
---|
322 | try {
|
---|
323 | BufferedReader buffered_reader = new BufferedReader(new FileReader(this));
|
---|
324 |
|
---|
325 | int description_element_num = 0;
|
---|
326 | int next_description_element_start = ((Integer) description_elements_list.get(description_element_num)).intValue();
|
---|
327 | boolean in_relevant_description_element = false;
|
---|
328 |
|
---|
329 | String line = null;
|
---|
330 | for (int line_num = 0; (line = buffered_reader.readLine()) != null; line_num++) {
|
---|
331 | // Check if this line contains the start of a relevant Description element
|
---|
332 | if (line_num == next_description_element_start) {
|
---|
333 | in_relevant_description_element = true;
|
---|
334 | continue;
|
---|
335 | }
|
---|
336 |
|
---|
337 | // If we're not in a relevant Description element we don't care about anything
|
---|
338 | if (in_relevant_description_element == false) {
|
---|
339 | continue;
|
---|
340 | }
|
---|
341 |
|
---|
342 | // Check if this line contains the end of the relevant Description element
|
---|
343 | if (line.indexOf("</Description>") != -1) {
|
---|
344 | description_element_num++;
|
---|
345 | if (description_element_num == description_elements_list.size()) {
|
---|
346 | break;
|
---|
347 | }
|
---|
348 |
|
---|
349 | next_description_element_start = ((Integer) description_elements_list.get(description_element_num)).intValue();
|
---|
350 | in_relevant_description_element = false;
|
---|
351 | continue;
|
---|
352 | }
|
---|
353 |
|
---|
354 | // If this line doesn't contain a complete Metadata element, we're not interested
|
---|
355 | if (line.indexOf("<Metadata ") == -1 || line.indexOf("</Metadata>") == -1) {
|
---|
356 | continue;
|
---|
357 | }
|
---|
358 |
|
---|
359 | // Extract the metadata element name
|
---|
360 | int name_index = line.indexOf(" name=\"") + " name=\"".length();
|
---|
361 | String metadata_element_name_full = line.substring(name_index, line.indexOf("\"", name_index));
|
---|
362 |
|
---|
363 | // If the metadata has a namespace it isn't extracted metadata, so we're not interested
|
---|
364 | String metadata_set_namespace = MetadataTools.getMetadataSetNamespace(metadata_element_name_full);
|
---|
365 | if (!metadata_set_namespace.equals("")) {
|
---|
366 | continue;
|
---|
367 | }
|
---|
368 |
|
---|
369 | // Extracted metadata!
|
---|
370 | String metadata_element_name = metadata_element_name_full;
|
---|
371 |
|
---|
372 | // We completely ignore bibliographic data
|
---|
373 | if (metadata_element_name.equals("SourceSegment")) {
|
---|
374 | buffered_reader.close();
|
---|
375 | return new ArrayList();
|
---|
376 | }
|
---|
377 |
|
---|
378 | // Ignore metadata starting with gsdl (used to be lowercase metadata and /metadata)
|
---|
379 | if (metadata_element_name.startsWith("gsdl")) {
|
---|
380 | continue;
|
---|
381 | }
|
---|
382 |
|
---|
383 | MetadataElement metadata_element = extracted_metadata_set.getMetadataElementWithName(metadata_element_name);
|
---|
384 |
|
---|
385 | // Value trees are not stored for extracted metadata, so create a new value tree node now
|
---|
386 | int value_index = line.indexOf(">", name_index) + ">".length();
|
---|
387 | String metadata_element_value = line.substring(value_index, line.lastIndexOf("</Metadata>"));
|
---|
388 |
|
---|
389 | metadata_element.addMetadataValue(metadata_element_value);
|
---|
390 | MetadataValueTreeNode metadata_value_tree_node = metadata_element.getMetadataValueTreeNode(metadata_element_value);
|
---|
391 |
|
---|
392 | // Add the new metadata value to the list
|
---|
393 | MetadataValue metadata_value = new MetadataValue(metadata_element, metadata_value_tree_node);
|
---|
394 | metadata_values.add(metadata_value);
|
---|
395 | }
|
---|
396 |
|
---|
397 | buffered_reader.close();
|
---|
398 | }
|
---|
399 | catch (FileNotFoundException exception) {
|
---|
400 | DebugStream.printStackTrace(exception);
|
---|
401 | }
|
---|
402 | catch (IOException exception) {
|
---|
403 | DebugStream.printStackTrace(exception);
|
---|
404 | }
|
---|
405 |
|
---|
406 | return metadata_values;
|
---|
407 | }
|
---|
408 |
|
---|
409 | */
|
---|
410 |
|
---|
411 | /**
|
---|
412 | * Every doc.xml file must be skimmed when a collection is opened, for two reasons:
|
---|
413 | * - To build a mapping from source file to its corresponding doc.xml file
|
---|
414 | * - To get a complete list of all extracted metadata elements
|
---|
415 | */
|
---|
416 | /*
|
---|
417 | public void skimFile()
|
---|
418 | {
|
---|
419 | MetadataSet extracted_metadata_set = MetadataSetManager.getMetadataSet(MetadataSetManager.EXTRACTED_METADATA_NAMESPACE);
|
---|
420 |
|
---|
421 | // Skim the doc.xml file as quickly as possible (don't parse as XML), looking at the Metadata elements
|
---|
422 | DebugStream.println("Skimming " + this + "...");
|
---|
423 | try {
|
---|
424 | BufferedReader buffered_reader = new BufferedReader(new FileReader(this));
|
---|
425 | int description_element_start = -1;
|
---|
426 |
|
---|
427 | String line = null;
|
---|
428 | for (int line_num = 0; (line = buffered_reader.readLine()) != null; line_num++) {
|
---|
429 | // This line contains the start of a Description element
|
---|
430 | if (line.indexOf("<Description>") != -1) {
|
---|
431 | if (description_element_start != -1) {
|
---|
432 | System.err.println("Parse error: previous Description element unfinished!");
|
---|
433 | }
|
---|
434 | description_element_start = line_num;
|
---|
435 | continue;
|
---|
436 | }
|
---|
437 |
|
---|
438 | // This line contains the end of a Description element
|
---|
439 | if (line.indexOf("</Description>") != -1) {
|
---|
440 | if (description_element_start == -1) {
|
---|
441 | System.err.println("Parse error: Description element unstarted!");
|
---|
442 | }
|
---|
443 | description_element_start = -1;
|
---|
444 | continue;
|
---|
445 | }
|
---|
446 |
|
---|
447 | // If we're not in a Description element there shouldn't be any Metadata elements
|
---|
448 | if (description_element_start == -1) {
|
---|
449 | continue;
|
---|
450 | }
|
---|
451 |
|
---|
452 | // This line doesn't contain a Metadata element, so we're not interested
|
---|
453 | if (line.indexOf("<Metadata ") == -1) {
|
---|
454 | DebugStream.println("Warning: Description element line doesn't contain Metadata element.");
|
---|
455 | continue;
|
---|
456 | }
|
---|
457 |
|
---|
458 | // Extract the metadata element name
|
---|
459 | int name_index = line.indexOf(" name=\"") + " name=\"".length();
|
---|
460 | String metadata_element_name_full = line.substring(name_index, line.indexOf("\"", name_index));
|
---|
461 |
|
---|
462 | // If the metadata has a namespace it isn't extracted metadata, so we're not interested
|
---|
463 | String metadata_set_namespace = MetadataTools.getMetadataSetNamespace(metadata_element_name_full);
|
---|
464 | if (!metadata_set_namespace.equals("")) {
|
---|
465 | continue;
|
---|
466 | }
|
---|
467 |
|
---|
468 | // Extracted metadata!
|
---|
469 | String metadata_element_name = metadata_element_name_full;
|
---|
470 |
|
---|
471 | // Note which file this doc.xml is for
|
---|
472 | if (metadata_element_name.equals("gsdlsourcefilename")) {
|
---|
473 | // Extract the gsdlsourcefilename element value
|
---|
474 | int value_index = line.indexOf(">", name_index) + ">".length();
|
---|
475 | String gsdlsourcefilename_value = line.substring(value_index, line.indexOf("<", value_index));
|
---|
476 |
|
---|
477 | // We're only interested in the path relative to the import folder
|
---|
478 | int import_index = gsdlsourcefilename_value.indexOf("import");
|
---|
479 | if (import_index != -1) {
|
---|
480 | gsdlsourcefilename_value = gsdlsourcefilename_value.substring(import_index + "import".length());
|
---|
481 |
|
---|
482 | boolean is_unix_path = gsdlsourcefilename_value.startsWith("/");
|
---|
483 | gsdlsourcefilename_value = gsdlsourcefilename_value.substring(1);
|
---|
484 |
|
---|
485 | // URL decode gsdlsourcefilename. Need to set the decoder to use the default system encoding
|
---|
486 | // This is stored in the System's file.encoding property.
|
---|
487 | gsdlsourcefilename_value = URLDecoder.decode(gsdlsourcefilename_value, System.getProperty("file.encoding"));
|
---|
488 |
|
---|
489 | // Make sure the path matches the OS that is running
|
---|
490 | if (is_unix_path && Utility.isWindows()) {
|
---|
491 | // Convert path from Unix to Windows
|
---|
492 | gsdlsourcefilename_value = gsdlsourcefilename_value.replaceAll("\\/", "\\\\");
|
---|
493 | }
|
---|
494 | else if (!is_unix_path && !Utility.isWindows()) {
|
---|
495 | // Convert path from Windows to Unix
|
---|
496 | gsdlsourcefilename_value = gsdlsourcefilename_value.replaceAll("\\\\", "/");
|
---|
497 | }
|
---|
498 |
|
---|
499 | // Remember this for quick access later
|
---|
500 | if (source_file_name_to_description_elements_mapping.get(gsdlsourcefilename_value) == null) {
|
---|
501 | source_file_name_to_description_elements_mapping.put(gsdlsourcefilename_value, new ArrayList());
|
---|
502 | }
|
---|
503 |
|
---|
504 | ((ArrayList) source_file_name_to_description_elements_mapping.get(gsdlsourcefilename_value)).add(new Integer(description_element_start));
|
---|
505 | }
|
---|
506 |
|
---|
507 | // Warn about an odd gsdlsourcefilename, except if it is the Greenstone "tmp" directory
|
---|
508 | // This is true when the source files come from a zip file processed by ZIPPlug, for example
|
---|
509 | else if (gsdlsourcefilename_value.indexOf("tmp") == -1) {
|
---|
510 | // We don't really know what is going on...
|
---|
511 | System.err.println("Warning: Could not understand gsdlsourcefilename " + gsdlsourcefilename_value);
|
---|
512 | }
|
---|
513 | }
|
---|
514 |
|
---|
515 | // Ignore metadata starting with gsdl (used to be lowercase metadata and /metadata)
|
---|
516 | if (metadata_element_name.startsWith("gsdl")) {
|
---|
517 | continue;
|
---|
518 | }
|
---|
519 |
|
---|
520 | MetadataElement metadata_element = extracted_metadata_set.getMetadataElementWithName(metadata_element_name);
|
---|
521 | if (metadata_element == null) {
|
---|
522 | // This element isn't defined in ex.mds, so create it for this session
|
---|
523 | DebugStream.println("Extracted metadata element not defined: " + metadata_element_name);
|
---|
524 | extracted_metadata_set.addMetadataElementForThisSession(metadata_element_name);
|
---|
525 | }
|
---|
526 | }
|
---|
527 |
|
---|
528 | buffered_reader.close();
|
---|
529 | }
|
---|
530 | catch (FileNotFoundException exception) {
|
---|
531 | DebugStream.printStackTrace(exception);
|
---|
532 | }
|
---|
533 | catch (IOException exception) {
|
---|
534 | DebugStream.printStackTrace(exception);
|
---|
535 | }
|
---|
536 | }
|
---|
537 | */
|
---|
538 |
|
---|
539 | }
|
---|