21 |
|
|
22 |
package nongnu.cashews.rdf; |
package nongnu.cashews.rdf; |
23 |
|
|
24 |
|
import java.io.File; |
25 |
|
import java.io.FileInputStream; |
26 |
|
import java.io.FileNotFoundException; |
27 |
|
import java.io.IOException; |
28 |
|
import java.net.URI; |
29 |
|
import java.net.URISyntaxException; |
30 |
|
import java.util.logging.ConsoleHandler; |
31 |
|
import java.util.logging.Handler; |
32 |
|
import java.util.logging.Level; |
33 |
|
import java.util.logging.Logger; |
34 |
|
|
35 |
|
import org.xml.sax.Attributes; |
36 |
|
import org.xml.sax.InputSource; |
37 |
|
import org.xml.sax.SAXException; |
38 |
|
import org.xml.sax.XMLReader; |
39 |
|
import org.xml.sax.helpers.DefaultHandler; |
40 |
|
import org.xml.sax.helpers.XMLReaderFactory; |
41 |
|
|
42 |
/** |
/** |
43 |
* This class deals with the parsing of RDF in XML form, as defined by the |
* This class deals with the parsing of RDF in XML form, as defined by the |
44 |
* <a href="http://www.w3.org/TR/rdf-syntax-grammar/">RDFXML </a> syntax |
* <a href="http://www.w3.org/TR/rdf-syntax-grammar/">RDFXML </a> syntax |
51 |
public class XMLParser |
public class XMLParser |
52 |
{ |
{ |
53 |
|
|
54 |
|
/** |
55 |
|
* The RDF namespace. |
56 |
|
*/ |
57 |
|
public static final String |
58 |
|
RDF_NAMESPACE = "http://www.w3.org/1999/02/22-rdf-syntax-ns#"; |
59 |
|
|
60 |
|
/** |
61 |
|
* The reader, which interprets the XML file. |
62 |
|
* |
63 |
|
* @serial the XML reader. |
64 |
|
*/ |
65 |
|
private XMLReader reader; |
66 |
|
|
67 |
|
/** |
68 |
|
* Constructs a new XML-based RDF parser, using the specified handler |
69 |
|
* for messages. |
70 |
|
* |
71 |
|
* @param handler the handler to use for log messages. |
72 |
|
* @throws SAXException if the parser fails to obtain an |
73 |
|
* <code>XMLReader</code> instance. |
74 |
|
*/ |
75 |
|
public XMLParser(Handler handler) |
76 |
|
throws SAXException |
77 |
|
{ |
78 |
|
reader = XMLReaderFactory.createXMLReader(); |
79 |
|
reader.setContentHandler(new RDFHandler(handler)); |
80 |
|
} |
81 |
|
|
82 |
|
/** |
83 |
|
* Parse an RDF document from the specified system identifier (URI). |
84 |
|
* |
85 |
|
* @param systemId the URI of the RDF document. |
86 |
|
* @throws SAXException if an error occurs in parsing. |
87 |
|
* @throws IOException if an error occurs in the underlying input. |
88 |
|
*/ |
89 |
|
public void parse(String systemId) |
90 |
|
throws IOException, SAXException |
91 |
|
{ |
92 |
|
reader.parse(systemId); |
93 |
|
} |
94 |
|
|
95 |
|
/** |
96 |
|
* Parse an RDF document from the specified input source. |
97 |
|
* |
98 |
|
* @param source the source of the RDF document. |
99 |
|
* @throws SAXException if an error occurs in parsing. |
100 |
|
* @throws IOException if an error occurs in the underlying input. |
101 |
|
*/ |
102 |
|
public void parse(InputSource source) |
103 |
|
throws IOException, SAXException |
104 |
|
{ |
105 |
|
reader.parse(source); |
106 |
|
} |
107 |
|
|
108 |
|
/** |
109 |
|
* Parse an RDF document from the specified file. |
110 |
|
* |
111 |
|
* @param file the file containing the RDF document. |
112 |
|
* @throws SAXException if an error occurs in parsing. |
113 |
|
* @throws IOException if an error occurs in the underlying input. |
114 |
|
*/ |
115 |
|
public void parse(File file) |
116 |
|
throws IOException, SAXException |
117 |
|
{ |
118 |
|
parse(new InputSource(new FileInputStream(file))); |
119 |
|
} |
120 |
|
|
121 |
|
/** |
122 |
|
* A simple test harness to parse the RDF files supplied on the command |
123 |
|
* line. |
124 |
|
* |
125 |
|
* @param args the command line arguments. |
126 |
|
* @throws SAXException if a <code>XMLReader</code> can't be obtained |
127 |
|
* or an error occurs during parsing. |
128 |
|
* @throws IOException if an error occurs in the underlying input. |
129 |
|
* @throws FileNotFoundException if one of the specified files can't |
130 |
|
* be found. |
131 |
|
*/ |
132 |
|
public static void main(String[] args) |
133 |
|
throws SAXException, IOException, FileNotFoundException |
134 |
|
{ |
135 |
|
Handler handler = new ConsoleHandler(); |
136 |
|
handler.setLevel(Level.FINE); |
137 |
|
XMLParser parser = new XMLParser(handler); |
138 |
|
for (int a = 0; a < args.length; ++a) |
139 |
|
parser.parse(new File(args[a])); |
140 |
|
} |
141 |
|
|
142 |
|
/** |
143 |
|
* Handles capturing the XML events and turning them into an RDF |
144 |
|
* <code>Graph</code>. |
145 |
|
* |
146 |
|
* @author Andrew John Hughes (gnu_andrew@member.fsf.org) |
147 |
|
*/ |
148 |
|
private class RDFHandler |
149 |
|
extends DefaultHandler |
150 |
|
{ |
151 |
|
|
152 |
|
/** |
153 |
|
* A <code>Logger</code> instance to log events generated by the |
154 |
|
* parsing process. |
155 |
|
*/ |
156 |
|
private Logger logger; |
157 |
|
|
158 |
|
/** |
159 |
|
* Flag to indicate whether we are inside an RDF document |
160 |
|
* or not. There is not a one-to-one correspondence between |
161 |
|
* an RDF document and the XML document. RDF is to usually |
162 |
|
* to be expected inside an RDF element. However, a single |
163 |
|
* <code>Description</code> element may occur on its own, outside |
164 |
|
* this context. In either of these cases, this flag should |
165 |
|
* be set. |
166 |
|
*/ |
167 |
|
private boolean inRDF; |
168 |
|
|
169 |
|
/** |
170 |
|
* Flag to indicate whether we are inside a subject node or |
171 |
|
* not. |
172 |
|
*/ |
173 |
|
private boolean inSubject; |
174 |
|
|
175 |
|
/** |
176 |
|
* Flag to indicate whether we are inside a predicate node or |
177 |
|
* not. |
178 |
|
*/ |
179 |
|
private boolean inPredicate; |
180 |
|
|
181 |
|
/** |
182 |
|
* The current triple. |
183 |
|
*/ |
184 |
|
private Triple triple; |
185 |
|
|
186 |
|
/** |
187 |
|
* The current subject in the current RDF triple. |
188 |
|
*/ |
189 |
|
private Subject subject; |
190 |
|
|
191 |
|
/** |
192 |
|
* The current predicate in the current RDF triple. |
193 |
|
*/ |
194 |
|
private Predicate predicate; |
195 |
|
|
196 |
|
/** |
197 |
|
* The current object in the current RDF triple. |
198 |
|
*/ |
199 |
|
private RDFObject object; |
200 |
|
|
201 |
|
/** |
202 |
|
* The String form of the URI of the current predicate. |
203 |
|
*/ |
204 |
|
private String predicateURI; |
205 |
|
|
206 |
|
/** |
207 |
|
* Constructs a new <code>RDFHandler</code>, using the specified |
208 |
|
* handler for log messages. |
209 |
|
* |
210 |
|
* @param handler the handler to use for log messages. |
211 |
|
*/ |
212 |
|
public RDFHandler(Handler handler) |
213 |
|
{ |
214 |
|
logger = Logger.getLogger("nongnu.cashews.rdf.XMLParser"); |
215 |
|
logger.addHandler(handler); |
216 |
|
logger.setLevel(handler.getLevel()); |
217 |
|
} |
218 |
|
|
219 |
|
/** |
220 |
|
* Captures the start of the document and sets up the initial |
221 |
|
* state. |
222 |
|
*/ |
223 |
|
public void startDocument() |
224 |
|
{ |
225 |
|
inSubject = false; |
226 |
|
inRDF = false; |
227 |
|
inPredicate = false; |
228 |
|
predicateURI = null; |
229 |
|
} |
230 |
|
|
231 |
|
/** |
232 |
|
* Captures the start of an XML element. |
233 |
|
* |
234 |
|
* @param uri the namespace URI. |
235 |
|
* @param localName the local name of the element inside the namespace. |
236 |
|
* @param qName the local name qualified with the namespace URI. This |
237 |
|
* may or may not be provided, as we don't ask for namespace |
238 |
|
* prefixes. |
239 |
|
* @param attributes the attributes of this element. |
240 |
|
* @throws SAXException if some error occurs in parsing. |
241 |
|
*/ |
242 |
|
public void startElement(String uri, String localName, |
243 |
|
String qName, Attributes attributes) |
244 |
|
throws SAXException |
245 |
|
{ |
246 |
|
if (uri.equals(RDF_NAMESPACE)) |
247 |
|
{ |
248 |
|
if (localName.equals("RDF")) |
249 |
|
{ |
250 |
|
inRDF = true; |
251 |
|
logger.finer("Start of RDF block"); |
252 |
|
} |
253 |
|
else if (localName.equals("Description")) |
254 |
|
{ |
255 |
|
logger.finer("Start of RDF description block"); |
256 |
|
inRDF = true; |
257 |
|
if (!inSubject) |
258 |
|
{ |
259 |
|
String value = attributes.getValue(RDF_NAMESPACE, "about"); |
260 |
|
if (value != null) |
261 |
|
{ |
262 |
|
subject = parseRDFURI(value); |
263 |
|
} |
264 |
|
inSubject = true; |
265 |
|
logger.fine("Created subject: " + subject); |
266 |
|
} |
267 |
|
else |
268 |
|
logger.warning("Invalid use of RDF namespace: " + uri + |
269 |
|
localName); |
270 |
|
} |
271 |
|
} |
272 |
|
else |
273 |
|
{ |
274 |
|
if (inSubject) |
275 |
|
{ |
276 |
|
predicateURI = uri + localName; |
277 |
|
logger.finer("Start of predicate: " + predicateURI); |
278 |
|
predicate = parseRDFURI(predicateURI); |
279 |
|
logger.fine("Created predicate: " + predicate); |
280 |
|
inSubject = false; |
281 |
|
inPredicate = true; |
282 |
|
} |
283 |
|
} |
284 |
|
} |
285 |
|
|
286 |
|
/** |
287 |
|
* Attempts to returns the RDF URI parsed from the specified |
288 |
|
* <code>String</code>. |
289 |
|
* |
290 |
|
* @param value the value to parse. |
291 |
|
* @return an RDF URI. |
292 |
|
* @throws SAXException if the URI can't be parsed. |
293 |
|
*/ |
294 |
|
private RDFURI parseRDFURI(String value) |
295 |
|
throws SAXException |
296 |
|
{ |
297 |
|
try |
298 |
|
{ |
299 |
|
return new RDFURI(new URI(value)); |
300 |
|
} |
301 |
|
catch (URISyntaxException e) |
302 |
|
{ |
303 |
|
throw new SAXException("Failed to parse URI: " + e, e); |
304 |
|
} |
305 |
|
} |
306 |
|
|
307 |
|
/** |
308 |
|
* Captures characters within an XML element. |
309 |
|
* |
310 |
|
* @param ch the array of characters. |
311 |
|
* @param start the start index of the characters to use. |
312 |
|
* @param length the number of characters to use from the start index on. |
313 |
|
* @throws SAXException if some error occurs in parsing. |
314 |
|
*/ |
315 |
|
public void characters(char[] ch, int start, int length) |
316 |
|
throws SAXException |
317 |
|
{ |
318 |
|
String value = new String(ch, start, length).trim(); |
319 |
|
if (value.length() == 0) |
320 |
|
return; |
321 |
|
logger.finer("Characters: " + value); |
322 |
|
if (inPredicate) |
323 |
|
{ |
324 |
|
object = new Literal(value); |
325 |
|
logger.fine("Created object: " + object); |
326 |
|
} |
327 |
|
} |
328 |
|
|
329 |
|
/** |
330 |
|
* Captures the end of an XML element. |
331 |
|
* |
332 |
|
* @param uri the namespace URI. |
333 |
|
* @param localName the local name of the element inside the namespace. |
334 |
|
* @param qName the local name qualified with the namespace URI. This |
335 |
|
* may or may not be provided, as we don't ask for namespace |
336 |
|
* prefixes. |
337 |
|
* @throws SAXException if some error occurs in parsing. |
338 |
|
*/ |
339 |
|
public void endElement(String uri, String localName, |
340 |
|
String qName) |
341 |
|
throws SAXException |
342 |
|
{ |
343 |
|
if (uri.equals(RDF_NAMESPACE)) |
344 |
|
{ |
345 |
|
if (localName.equals("RDF")) |
346 |
|
{ |
347 |
|
inRDF = false; |
348 |
|
logger.finer("End of RDF block"); |
349 |
|
} |
350 |
|
else if (localName.equals("Description")) |
351 |
|
{ |
352 |
|
logger.finer("End of description block"); |
353 |
|
triple = new Triple(subject, predicate, object); |
354 |
|
logger.fine("Created triple: " + triple); |
355 |
|
inSubject = false; |
356 |
|
} |
357 |
|
} |
358 |
|
else if (inPredicate && predicateURI.equals(uri + localName)) |
359 |
|
{ |
360 |
|
inPredicate = false; |
361 |
|
inSubject = true; |
362 |
|
predicateURI = null; |
363 |
|
logger.finer("End of predicate block"); |
364 |
|
} |
365 |
|
} |
366 |
|
|
367 |
|
} |
368 |
|
|
369 |
} |
} |