1 | package org.greenstone.gsdl3.gs3build.metadata;
|
---|
2 |
|
---|
3 | import java.io.PrintWriter;
|
---|
4 |
|
---|
5 | import java.util.List;
|
---|
6 | import java.util.ArrayList;
|
---|
7 |
|
---|
8 | import java.net.URL;
|
---|
9 | import java.net.URLConnection;
|
---|
10 |
|
---|
11 | import java.sql.ResultSet;
|
---|
12 | import java.sql.SQLException;
|
---|
13 | import java.sql.Statement;
|
---|
14 |
|
---|
15 | import org.w3c.dom.Document;
|
---|
16 | import org.w3c.dom.Element;
|
---|
17 | import org.w3c.dom.NamedNodeMap;
|
---|
18 | import org.w3c.dom.Node;
|
---|
19 | import org.w3c.dom.NodeList;
|
---|
20 | import org.w3c.dom.Text;
|
---|
21 |
|
---|
22 | import org.greenstone.gsdl3.gs3build.util.HTTPTools;
|
---|
23 | import org.greenstone.gsdl3.gs3build.util.XMLTools;
|
---|
24 | import org.greenstone.gsdl3.gs3build.database.*;
|
---|
25 |
|
---|
26 | /**
|
---|
27 | * An individual METS component file for a document. Many features of METS are not
|
---|
28 | * currently fully supported/used here, but member variables exist for all of the
|
---|
29 | * standard members.
|
---|
30 | */
|
---|
31 |
|
---|
32 | public class METSFile
|
---|
33 | {
|
---|
34 | METSFileID id;
|
---|
35 | METSFilePos location;
|
---|
36 | String MIMEType;
|
---|
37 | int sequenceNo;
|
---|
38 | long size;
|
---|
39 | METSDate created;
|
---|
40 | long checkSum;
|
---|
41 | List adminRefs;
|
---|
42 | List describeRefs;
|
---|
43 | METSFileGroup group;
|
---|
44 |
|
---|
45 | /**
|
---|
46 | * Create a new METSFile object, with the given properties. The remaining
|
---|
47 | * fields are filled in with default values.
|
---|
48 | *
|
---|
49 | * @param <code>String</code> the id of the file
|
---|
50 | * @param <code>METSFilePos</code> the location of the file
|
---|
51 | * @param <code>String</code> the MIME type of the file
|
---|
52 | */
|
---|
53 | public METSFile(METSFileID id, METSFilePos location, String mimeType)
|
---|
54 | {
|
---|
55 | this.id = id; // required
|
---|
56 | this.location = location; // the location
|
---|
57 | this.MIMEType = mimeType; // the MIME type;
|
---|
58 | this.sequenceNo = -1; // -1 indicates unset
|
---|
59 | this.size = -1; // -1 indicates unset
|
---|
60 | this.checkSum = -1; // -1 indicates unset
|
---|
61 | this.created = null;
|
---|
62 | this.adminRefs = new ArrayList();
|
---|
63 | this.describeRefs = new ArrayList();
|
---|
64 | this.group = null;
|
---|
65 | }
|
---|
66 |
|
---|
67 | public boolean equals(URL url)
|
---|
68 | {
|
---|
69 | return this.location.equals(url);
|
---|
70 | }
|
---|
71 |
|
---|
72 | /**
|
---|
73 | * Set the parent group
|
---|
74 | *
|
---|
75 | * @param <code>METSFileGroup</code> the new holding file group
|
---|
76 | */
|
---|
77 | public void setGroup(METSFileGroup group)
|
---|
78 | {
|
---|
79 | this.group = group;
|
---|
80 | }
|
---|
81 |
|
---|
82 | /**
|
---|
83 | * Get the parent group of this file
|
---|
84 | *
|
---|
85 | * @return <code>METSFileGroup</code> the holding file group
|
---|
86 | */
|
---|
87 | public METSFileGroup getGroup()
|
---|
88 | {
|
---|
89 | return this.group;
|
---|
90 | }
|
---|
91 |
|
---|
92 | /**
|
---|
93 | * Get the identifier of this file
|
---|
94 | *
|
---|
95 | * @return <code>METSFileID</code> the file identifier
|
---|
96 | */
|
---|
97 | public METSFileID getID()
|
---|
98 | {
|
---|
99 | return this.id;
|
---|
100 | }
|
---|
101 |
|
---|
102 | /**
|
---|
103 | * Set the identifier of this file
|
---|
104 | *
|
---|
105 | * @param <code>METSFileID</code> the file identifier
|
---|
106 | */
|
---|
107 | public void setID(METSFileID id)
|
---|
108 | {
|
---|
109 | this.id = id;
|
---|
110 | }
|
---|
111 |
|
---|
112 | /**
|
---|
113 | * @return <code>String</code> the MIME (content) type of the file
|
---|
114 | */
|
---|
115 | public String getMIMEType()
|
---|
116 | {
|
---|
117 | return this.MIMEType;
|
---|
118 | }
|
---|
119 |
|
---|
120 | /**
|
---|
121 | * @return <code>URL</code> the location of the file. This will often be in
|
---|
122 | * "file://" form.
|
---|
123 | */
|
---|
124 | public URL getLocation()
|
---|
125 | {
|
---|
126 | return this.location.getLocation();
|
---|
127 | }
|
---|
128 |
|
---|
129 | /**
|
---|
130 | * Take a url and make a METSFile object out of it - works out
|
---|
131 | * details such as the Mime type, identifiers and other necessary
|
---|
132 | * information to call the constructor for METSFile
|
---|
133 | *
|
---|
134 | * @param <code>URL</code> the url of the object - this may be a
|
---|
135 | * reference to a file on the local filestore
|
---|
136 | *
|
---|
137 | * @return <code>METSFile</code> the METS file object for the corresponding
|
---|
138 | * document component.
|
---|
139 | */
|
---|
140 | public static METSFile makeMETSFile(URL url, String mimeType)
|
---|
141 | {
|
---|
142 | METSFile reply = new METSFile(new METSFileID(), new METSFilePos(url), mimeType);
|
---|
143 | return reply;
|
---|
144 | }
|
---|
145 |
|
---|
146 | public static METSFile makeMETSFile(URL url)
|
---|
147 | {
|
---|
148 | String mimeType;
|
---|
149 |
|
---|
150 | if (url.getProtocol().equals("file")){
|
---|
151 | // TODO: Work out the MIME type
|
---|
152 | mimeType = URLConnection.getFileNameMap().getContentTypeFor(url.toString().substring(7));
|
---|
153 | if (mimeType == null) {
|
---|
154 | mimeType = "text/plain";
|
---|
155 | }
|
---|
156 | } else {
|
---|
157 | // TODO: look up the MIME type through the pertinent connection
|
---|
158 | mimeType = HTTPTools.getMIMEType(url);
|
---|
159 | }
|
---|
160 | // TODO: make the identifier...
|
---|
161 |
|
---|
162 | return makeMETSFile(url, mimeType);
|
---|
163 | }
|
---|
164 |
|
---|
165 | /**
|
---|
166 | * Write the METS file in an XML to a text-output sink
|
---|
167 | *
|
---|
168 | * @param <code>PrintWriter</code> the destination of the output
|
---|
169 | */
|
---|
170 | public void write(PrintWriter writer)
|
---|
171 | {
|
---|
172 | String tag = XMLTools.getOpenTag("mets", "file");
|
---|
173 | tag = XMLTools.addAttribute(tag, "MIMETYPE", this.MIMEType);
|
---|
174 | tag = XMLTools.addAttribute(tag, "ID", this.id.toString());
|
---|
175 | writer.print(" "); //indentation
|
---|
176 | writer.println(tag);
|
---|
177 |
|
---|
178 | tag = XMLTools.getOpenTag("mets", "FLocat");
|
---|
179 | tag = XMLTools.addAttribute(tag, "LOCTYPE", this.location.getType());
|
---|
180 | tag = XMLTools.addAttribute(tag, "xlink:href", this.location.getLocation().toString());
|
---|
181 |
|
---|
182 | tag = XMLTools.makeSingleton(tag);
|
---|
183 | writer.print(" "); //indentation
|
---|
184 | writer.println(tag);
|
---|
185 |
|
---|
186 | writer.print(" "); //indentation
|
---|
187 | writer.println(XMLTools.getCloseTag("mets", "file"));
|
---|
188 | }
|
---|
189 |
|
---|
190 |
|
---|
191 | public boolean writeSQL(int groupReference, GS3SQLConnection connection)
|
---|
192 | {
|
---|
193 | // check if this file is in the table already...
|
---|
194 | GS3SQLAction action;
|
---|
195 |
|
---|
196 | GS3SQLSelect select = new GS3SQLSelect("files");
|
---|
197 | select.addField("*");
|
---|
198 | GS3SQLWhereItem whereItem = new GS3SQLWhereItem("FileGroupRef", "=", Integer.toString(groupReference),
|
---|
199 | GS3SQLField.INTEGER_TYPE);
|
---|
200 | GS3SQLWhere where = new GS3SQLWhere(whereItem);
|
---|
201 | whereItem = new GS3SQLWhereItem("FileID","=", this.id.toString());
|
---|
202 | where.add(whereItem);
|
---|
203 | select.setWhere(where);
|
---|
204 |
|
---|
205 | Statement statement = null;
|
---|
206 |
|
---|
207 | // if not, then make an insert action
|
---|
208 | try {
|
---|
209 | statement = connection.createStatement();
|
---|
210 | ResultSet results = statement.executeQuery(select.toString());
|
---|
211 | if (!results.first()){
|
---|
212 | GS3SQLInsert insert = new GS3SQLInsert("files");
|
---|
213 |
|
---|
214 | insert.addValue("FileGroupRef", Integer.toString(groupReference), GS3SQLField.INTEGER_TYPE);
|
---|
215 | insert.addValue("FileID", this.id.toString());
|
---|
216 |
|
---|
217 | action = insert;
|
---|
218 | }
|
---|
219 | else {
|
---|
220 | GS3SQLUpdate update = new GS3SQLUpdate("files");
|
---|
221 |
|
---|
222 | update.setWhere(where);
|
---|
223 | action = update;
|
---|
224 | }
|
---|
225 | }
|
---|
226 | catch (SQLException ex){
|
---|
227 | System.err.println("METSFile.writeSQL(): "+ex);
|
---|
228 | return false;
|
---|
229 | }
|
---|
230 | action.addValue("FileLocType", this.location.getType());
|
---|
231 | action.addValue("FileLocation", this.location.getLocation().toString());
|
---|
232 | action.addValue("MIMEType", this.MIMEType);
|
---|
233 |
|
---|
234 | try {
|
---|
235 | statement.execute(action.toString());
|
---|
236 | statement.close();
|
---|
237 |
|
---|
238 | } catch (SQLException e) {
|
---|
239 | System.err.println("METSFile.writeSQL():"+e);
|
---|
240 | return false;
|
---|
241 | }
|
---|
242 | return true;
|
---|
243 | }
|
---|
244 |
|
---|
245 | public static METSFile readSQL(GS3SQLConnection connection, ResultSet parentSet)
|
---|
246 | {
|
---|
247 | try {
|
---|
248 | String locType = parentSet.getString("FileLocType");
|
---|
249 | String location = parentSet.getString("FileLocation");
|
---|
250 | String mimeType = parentSet.getString("MIMEType");
|
---|
251 | String id = parentSet.getString("FileID");
|
---|
252 |
|
---|
253 | METSFileID metsID = new METSFileID(id);
|
---|
254 | METSFilePos metsFilePos = new METSFilePos(locType, location);
|
---|
255 |
|
---|
256 | METSFile reply = new METSFile(metsID, metsFilePos, mimeType);
|
---|
257 | return reply;
|
---|
258 | }
|
---|
259 | catch (SQLException ex){
|
---|
260 | System.out.println("METSFile.readSQL(): "+ex);
|
---|
261 | }
|
---|
262 | catch (java.net.MalformedURLException urlEx){
|
---|
263 | System.out.println("METSFile.readSQL(): "+urlEx);
|
---|
264 | }
|
---|
265 | return null;
|
---|
266 | }
|
---|
267 |
|
---|
268 | public static METSFile parse_file(Element element, METSFileGroup group, String parseFilePath)
|
---|
269 | {
|
---|
270 | NodeList children = element.getChildNodes();
|
---|
271 |
|
---|
272 | METSFile thisFile = null;
|
---|
273 |
|
---|
274 | for (int c = 0; c < children.getLength(); c++){
|
---|
275 | if (children.item(c).getNodeType() != org.w3c.dom.Node.ELEMENT_NODE){
|
---|
276 | continue;
|
---|
277 | }
|
---|
278 | Element child = (Element) children.item(c);
|
---|
279 | String childName = child.getNodeName();
|
---|
280 |
|
---|
281 | if (childName.equals("mets:file")){
|
---|
282 | METSFile.parse_file(child, group, parseFilePath);
|
---|
283 | } else if (childName.equals("mets:FLocat")) {
|
---|
284 | String mimeType = element.getAttribute("MIMETYPE");
|
---|
285 | String fileId = element.getAttribute("ID");
|
---|
286 |
|
---|
287 | METSFilePos thisFilePos = METSFile.parse_flocateXML(child, parseFilePath, fileId);
|
---|
288 | thisFile = new METSFile(new METSFileID(fileId), thisFilePos, mimeType);
|
---|
289 | group.addFile(thisFile);
|
---|
290 | } else {
|
---|
291 | System.err.println("Warning: unrecognised tag " +childName);
|
---|
292 | }
|
---|
293 | }
|
---|
294 | return thisFile;
|
---|
295 | }
|
---|
296 |
|
---|
297 | public static METSFilePos parse_flocateXML(Element element, String parseFilePath, String fileId)
|
---|
298 | {
|
---|
299 | METSFilePos thisFilePos = null;
|
---|
300 |
|
---|
301 | // get most of the information from the child FLocat node
|
---|
302 | String locationType = element.getAttribute("LOCTYPE");
|
---|
303 | String href = element.getAttribute("xlink:href");
|
---|
304 | //String fileId = element.getAttribute("ID");
|
---|
305 | try {
|
---|
306 | thisFilePos = new METSFilePos(locationType, href);
|
---|
307 |
|
---|
308 | if (fileId.startsWith("default.")) {
|
---|
309 | //retrieve the string after ..import/
|
---|
310 | int importPosition = href.indexOf("import/")+7;
|
---|
311 | href.substring(importPosition);
|
---|
312 | if (fileId.equals("default.1")) {
|
---|
313 | href = "file:" + parseFilePath+ "doctxt.xml";
|
---|
314 | } else {
|
---|
315 | href = "file:" + parseFilePath + href.substring(importPosition);
|
---|
316 | }
|
---|
317 | }
|
---|
318 | thisFilePos = new METSFilePos(locationType, href);
|
---|
319 | } catch (java.net.MalformedURLException ex) {
|
---|
320 | // TODO: raise error
|
---|
321 | System.err.println("METSFile.parse_flocateXML(): "+ex);
|
---|
322 | }
|
---|
323 | return thisFilePos;
|
---|
324 | }
|
---|
325 |
|
---|
326 |
|
---|
327 | /**
|
---|
328 | * Get the location of this file as a URL
|
---|
329 | */
|
---|
330 | public URL getURL()
|
---|
331 | {
|
---|
332 | return this.location.getLocation();
|
---|
333 | }
|
---|
334 |
|
---|
335 | /**
|
---|
336 | * Get modified file date
|
---|
337 | */
|
---|
338 | public long getModifiedDatestamp()
|
---|
339 | {
|
---|
340 | return this.location.getModifiedDatestamp();
|
---|
341 | }
|
---|
342 |
|
---|
343 | /**
|
---|
344 | * Overridden <code>toString</code> for convenience - returns the location of the file.
|
---|
345 | *
|
---|
346 | * @return <code>String</code> the location of the file as a string
|
---|
347 | */
|
---|
348 | public String toString()
|
---|
349 | {
|
---|
350 | return this.location.toString();
|
---|
351 | }
|
---|
352 | }
|
---|