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