|
1
17
18
19
20 package org.apache.fop.hyphenation;
21
22 import org.xml.sax.XMLReader  ;
24 import org.xml.sax.InputSource ;
25 import org.xml.sax.SAXException ;
26 import org.xml.sax.SAXParseException ;
27 import org.xml.sax.helpers.DefaultHandler ;
28 import org.xml.sax.Attributes ;
29
30 import java.io.File ;
32 import java.io.FileNotFoundException ;
33 import java.io.IOException ;
34 import java.net.MalformedURLException ;
35 import java.util.ArrayList ;
36
37 import javax.xml.parsers.SAXParserFactory ;
38
39
45 public class PatternParser extends DefaultHandler implements PatternConsumer {
46
47 XMLReader parser;
48 int currElement;
49 PatternConsumer consumer;
50 StringBuffer token;
51 ArrayList exception;
52 char hyphenChar;
53 String errMsg;
54
55 static final int ELEM_CLASSES = 1;
56 static final int ELEM_EXCEPTIONS = 2;
57 static final int ELEM_PATTERNS = 3;
58 static final int ELEM_HYPHEN = 4;
59
60 public PatternParser() throws HyphenationException {
61 token = new StringBuffer ();
62 parser = createParser();
63 parser.setContentHandler(this);
64 parser.setErrorHandler(this);
65 hyphenChar = '-';
67 }
68
69 public PatternParser(PatternConsumer consumer)
70 throws HyphenationException {
71 this();
72 this.consumer = consumer;
73 }
74
75 public void setConsumer(PatternConsumer consumer) {
76 this.consumer = consumer;
77 }
78
79
84 public void parse(String filename) throws HyphenationException {
85 parse(new File (filename));
86 }
87
88
93 public void parse(File file) throws HyphenationException {
94 try {
95 InputSource src = new InputSource (file.toURL().toExternalForm());
96 parse(src);
97 } catch (MalformedURLException e) {
98 throw new HyphenationException("Error converting the File '" + file + "' to a URL: "
99 + e.getMessage());
100 }
101 }
102
103
108 public void parse(InputSource source) throws HyphenationException {
109 try {
110 parser.parse(source);
111 } catch (FileNotFoundException fnfe) {
112 throw new HyphenationException("File not found: " + fnfe.getMessage());
113 } catch (IOException ioe) {
114 throw new HyphenationException(ioe.getMessage());
115 } catch (SAXException e) {
116 throw new HyphenationException(errMsg);
117 }
118 }
119
120
124 static XMLReader createParser() {
125 try {
126 SAXParserFactory factory = SAXParserFactory.newInstance();
127 factory.setNamespaceAware(true);
128 return factory.newSAXParser().getXMLReader();
129 } catch (Exception e) {
130 throw new RuntimeException ("Couldn't create XMLReader: " + e.getMessage());
131 }
132 }
133
134 protected String readToken(StringBuffer chars) {
135 String word;
136 boolean space = false;
137 int i;
138 for (i = 0; i < chars.length(); i++) {
139 if (Character.isWhitespace(chars.charAt(i))) {
140 space = true;
141 } else {
142 break;
143 }
144 }
145 if (space) {
146 for (int countr = i; countr < chars.length(); countr++) {
148 chars.setCharAt(countr - i, chars.charAt(countr));
149 }
150 chars.setLength(chars.length() - i);
151 if (token.length() > 0) {
152 word = token.toString();
153 token.setLength(0);
154 return word;
155 }
156 }
157 space = false;
158 for (i = 0; i < chars.length(); i++) {
159 if (Character.isWhitespace(chars.charAt(i))) {
160 space = true;
161 break;
162 }
163 }
164 token.append(chars.toString().substring(0, i));
165 for (int countr = i; countr < chars.length(); countr++) {
167 chars.setCharAt(countr - i, chars.charAt(countr));
168 }
169 chars.setLength(chars.length() - i);
170 if (space) {
171 word = token.toString();
172 token.setLength(0);
173 return word;
174 }
175 token.append(chars);
176 return null;
177 }
178
179 protected static String getPattern(String word) {
180 StringBuffer pat = new StringBuffer ();
181 int len = word.length();
182 for (int i = 0; i < len; i++) {
183 if (!Character.isDigit(word.charAt(i))) {
184 pat.append(word.charAt(i));
185 }
186 }
187 return pat.toString();
188 }
189
190 protected ArrayList normalizeException(ArrayList ex) {
191 ArrayList res = new ArrayList ();
192 for (int i = 0; i < ex.size(); i++) {
193 Object item = ex.get(i);
194 if (item instanceof String ) {
195 String str = (String )item;
196 StringBuffer buf = new StringBuffer ();
197 for (int j = 0; j < str.length(); j++) {
198 char c = str.charAt(j);
199 if (c != hyphenChar) {
200 buf.append(c);
201 } else {
202 res.add(buf.toString());
203 buf.setLength(0);
204 char[] h = new char[1];
205 h[0] = hyphenChar;
206 res.add(new Hyphen(new String (h), null, null));
209 }
210 }
211 if (buf.length() > 0) {
212 res.add(buf.toString());
213 }
214 } else {
215 res.add(item);
216 }
217 }
218 return res;
219 }
220
221 protected String getExceptionWord(ArrayList ex) {
222 StringBuffer res = new StringBuffer ();
223 for (int i = 0; i < ex.size(); i++) {
224 Object item = ex.get(i);
225 if (item instanceof String ) {
226 res.append((String )item);
227 } else {
228 if (((Hyphen)item).noBreak != null) {
229 res.append(((Hyphen)item).noBreak);
230 }
231 }
232 }
233 return res.toString();
234 }
235
236 protected static String getInterletterValues(String pat) {
237 StringBuffer il = new StringBuffer ();
238 String word = pat + "a"; int len = word.length();
240 for (int i = 0; i < len; i++) {
241 char c = word.charAt(i);
242 if (Character.isDigit(c)) {
243 il.append(c);
244 i++;
245 } else {
246 il.append('0');
247 }
248 }
249 return il.toString();
250 }
251
252
256
259 public void startElement(String uri, String local, String raw,
260 Attributes attrs) {
261 if (local.equals("hyphen-char")) {
262 String h = attrs.getValue("value");
263 if (h != null && h.length() == 1) {
264 hyphenChar = h.charAt(0);
265 }
266 } else if (local.equals("classes")) {
267 currElement = ELEM_CLASSES;
268 } else if (local.equals("patterns")) {
269 currElement = ELEM_PATTERNS;
270 } else if (local.equals("exceptions")) {
271 currElement = ELEM_EXCEPTIONS;
272 exception = new ArrayList ();
273 } else if (local.equals("hyphen")) {
274 if (token.length() > 0) {
275 exception.add(token.toString());
276 }
277 exception.add(new Hyphen(attrs.getValue("pre"),
278 attrs.getValue("no"),
279 attrs.getValue("post")));
280 currElement = ELEM_HYPHEN;
281 }
282 token.setLength(0);
283 }
284
285
288 public void endElement(String uri, String local, String raw) {
289
290 if (token.length() > 0) {
291 String word = token.toString();
292 switch (currElement) {
293 case ELEM_CLASSES:
294 consumer.addClass(word);
295 break;
296 case ELEM_EXCEPTIONS:
297 exception.add(word);
298 exception = normalizeException(exception);
299 consumer.addException(getExceptionWord(exception),
300 (ArrayList )exception.clone());
301 break;
302 case ELEM_PATTERNS:
303 consumer.addPattern(getPattern(word),
304 getInterletterValues(word));
305 break;
306 case ELEM_HYPHEN:
307 break;
309 }
310 if (currElement != ELEM_HYPHEN) {
311 token.setLength(0);
312 }
313 }
314 if (currElement == ELEM_HYPHEN) {
315 currElement = ELEM_EXCEPTIONS;
316 } else {
317 currElement = 0;
318 }
319
320 }
321
322
325 public void characters(char ch[], int start, int length) {
326 StringBuffer chars = new StringBuffer (length);
327 chars.append(ch, start, length);
328 String word = readToken(chars);
329 while (word != null) {
330 switch (currElement) {
332 case ELEM_CLASSES:
333 consumer.addClass(word);
334 break;
335 case ELEM_EXCEPTIONS:
336 exception.add(word);
337 exception = normalizeException(exception);
338 consumer.addException(getExceptionWord(exception),
339 (ArrayList )exception.clone());
340 exception.clear();
341 break;
342 case ELEM_PATTERNS:
343 consumer.addPattern(getPattern(word),
344 getInterletterValues(word));
345 break;
346 }
347 word = readToken(chars);
348 }
349
350 }
351
352
356
359 public void warning(SAXParseException ex) {
360 errMsg = "[Warning] " + getLocationString(ex) + ": "
361 + ex.getMessage();
362 }
363
364
367 public void error(SAXParseException ex) {
368 errMsg = "[Error] " + getLocationString(ex) + ": " + ex.getMessage();
369 }
370
371
374 public void fatalError(SAXParseException ex) throws SAXException {
375 errMsg = "[Fatal Error] " + getLocationString(ex) + ": "
376 + ex.getMessage();
377 throw ex;
378 }
379
380
383 private String getLocationString(SAXParseException ex) {
384 StringBuffer str = new StringBuffer ();
385
386 String systemId = ex.getSystemId();
387 if (systemId != null) {
388 int index = systemId.lastIndexOf('/');
389 if (index != -1) {
390 systemId = systemId.substring(index + 1);
391 }
392 str.append(systemId);
393 }
394 str.append(':');
395 str.append(ex.getLineNumber());
396 str.append(':');
397 str.append(ex.getColumnNumber());
398
399 return str.toString();
400
401 }
403
404 public void addClass(String c) {
406 System.out.println("class: " + c);
407 }
408
409 public void addException(String w, ArrayList e) {
410 System.out.println("exception: " + w + " : " + e.toString());
411 }
412
413 public void addPattern(String p, String v) {
414 System.out.println("pattern: " + p + " : " + v);
415 }
416
417 public static void main(String [] args) throws Exception {
418 if (args.length > 0) {
419 PatternParser pp = new PatternParser();
420 pp.setConsumer(pp);
421 pp.parse(args[0]);
422 }
423 }
424
425 }
426 |
Popular Tags |
Java API By Example, From Geeks To Geeks.