|
1
57
58 package com.sun.org.apache.xerces.internal.impl.xpath.regex;
59
60 import java.text.CharacterIterator  ;
61
62
65
66 public final class REUtil {
67 private REUtil() {
68 }
69
70 static final int composeFromSurrogates(int high, int low) {
71 return 0x10000 + ((high-0xd800)<<10) + low-0xdc00;
72 }
73
74 static final boolean isLowSurrogate(int ch) {
75 return (ch & 0xfc00) == 0xdc00;
76 }
77
78 static final boolean isHighSurrogate(int ch) {
79 return (ch & 0xfc00) == 0xd800;
80 }
81
82 static final String decomposeToSurrogates(int ch) {
83 char[] chs = new char[2];
84 ch -= 0x10000;
85 chs[0] = (char)((ch>>10)+0xd800);
86 chs[1] = (char)((ch&0x3ff)+0xdc00);
87 return new String (chs);
88 }
89
90 static final String substring(CharacterIterator iterator, int begin, int end) {
91 char[] src = new char[end-begin];
92 for (int i = 0; i < src.length; i ++)
93 src[i] = iterator.setIndex(i+begin);
94 return new String (src);
95 }
96
97
99 static final int getOptionValue(int ch) {
100 int ret = 0;
101 switch (ch) {
102 case 'i':
103 ret = RegularExpression.IGNORE_CASE;
104 break;
105 case 'm':
106 ret = RegularExpression.MULTIPLE_LINES;
107 break;
108 case 's':
109 ret = RegularExpression.SINGLE_LINE;
110 break;
111 case 'x':
112 ret = RegularExpression.EXTENDED_COMMENT;
113 break;
114 case 'u':
115 ret = RegularExpression.USE_UNICODE_CATEGORY;
116 break;
117 case 'w':
118 ret = RegularExpression.UNICODE_WORD_BOUNDARY;
119 break;
120 case 'F':
121 ret = RegularExpression.PROHIBIT_FIXED_STRING_OPTIMIZATION;
122 break;
123 case 'H':
124 ret = RegularExpression.PROHIBIT_HEAD_CHARACTER_OPTIMIZATION;
125 break;
126 case 'X':
127 ret = RegularExpression.XMLSCHEMA_MODE;
128 break;
129 case ',':
130 ret = RegularExpression.SPECIAL_COMMA;
131 break;
132 default:
133 }
134 return ret;
135 }
136
137 static final int parseOptions(String opts) throws ParseException {
138 if (opts == null) return 0;
139 int options = 0;
140 for (int i = 0; i < opts.length(); i ++) {
141 int v = getOptionValue(opts.charAt(i));
142 if (v == 0)
143 throw new ParseException("Unknown Option: "+opts.substring(i), -1);
144 options |= v;
145 }
146 return options;
147 }
148
149 static final String createOptionString(int options) {
150 StringBuffer sb = new StringBuffer (9);
151 if ((options & RegularExpression.PROHIBIT_FIXED_STRING_OPTIMIZATION) != 0)
152 sb.append((char)'F');
153 if ((options & RegularExpression.PROHIBIT_HEAD_CHARACTER_OPTIMIZATION) != 0)
154 sb.append((char)'H');
155 if ((options & RegularExpression.XMLSCHEMA_MODE) != 0)
156 sb.append((char)'X');
157 if ((options & RegularExpression.IGNORE_CASE) != 0)
158 sb.append((char)'i');
159 if ((options & RegularExpression.MULTIPLE_LINES) != 0)
160 sb.append((char)'m');
161 if ((options & RegularExpression.SINGLE_LINE) != 0)
162 sb.append((char)'s');
163 if ((options & RegularExpression.USE_UNICODE_CATEGORY) != 0)
164 sb.append((char)'u');
165 if ((options & RegularExpression.UNICODE_WORD_BOUNDARY) != 0)
166 sb.append((char)'w');
167 if ((options & RegularExpression.EXTENDED_COMMENT) != 0)
168 sb.append((char)'x');
169 if ((options & RegularExpression.SPECIAL_COMMA) != 0)
170 sb.append((char)',');
171 return sb.toString().intern();
172 }
173
174
176 static String stripExtendedComment(String regex) {
177 int len = regex.length();
178 StringBuffer buffer = new StringBuffer (len);
179 int offset = 0;
180 while (offset < len) {
181 int ch = regex.charAt(offset++);
182 if (ch == '\t' || ch == '\n' || ch == '\f' || ch == '\r' || ch == ' ')
184 continue;
185
186 if (ch == '#') { while (offset < len) {
188 ch = regex.charAt(offset++);
189 if (ch == '\r' || ch == '\n')
190 break;
191 }
192 continue;
193 }
194
195 int next; if (ch == '\\' && offset < len) {
197 if ((next = regex.charAt(offset)) == '#'
198 || next == '\t' || next == '\n' || next == '\f'
199 || next == '\r' || next == ' ') {
200 buffer.append((char)next);
201 offset ++;
202 } else { buffer.append((char)'\\');
204 buffer.append((char)next);
205 offset ++;
206 }
207 } else buffer.append((char)ch);
209 }
210 return buffer.toString();
211 }
212
213
215
219 public static void main(String [] argv) {
220 String pattern = null;
221 try {
222 String options = "";
223 String target = null;
224 if( argv.length == 0 ) {
225 System.out.println( "Error:Usage: java REUtil -i|-m|-s|-u|-w|-X regularExpression String" );
226 System.exit( 0 );
227 }
228 for (int i = 0; i < argv.length; i ++) {
229 if (argv[i].length() == 0 || argv[i].charAt(0) != '-') {
230 if (pattern == null)
231 pattern = argv[i];
232 else if (target == null)
233 target = argv[i];
234 else
235 System.err.println("Unnecessary: "+argv[i]);
236 } else if (argv[i].equals("-i")) {
237 options += "i";
238 } else if (argv[i].equals("-m")) {
239 options += "m";
240 } else if (argv[i].equals("-s")) {
241 options += "s";
242 } else if (argv[i].equals("-u")) {
243 options += "u";
244 } else if (argv[i].equals("-w")) {
245 options += "w";
246 } else if (argv[i].equals("-X")) {
247 options += "X";
248 } else {
249 System.err.println("Unknown option: "+argv[i]);
250 }
251 }
252 RegularExpression reg = new RegularExpression(pattern, options);
253 System.out.println("RegularExpression: "+reg);
254 Match match = new Match();
255 reg.matches(target, match);
256 for (int i = 0; i < match.getNumberOfGroups(); i ++) {
257 if (i == 0 ) System.out.print("Matched range for the whole pattern: ");
258 else System.out.print("["+i+"]: ");
259 if (match.getBeginning(i) < 0)
260 System.out.println("-1");
261 else {
262 System.out.print(match.getBeginning(i)+", "+match.getEnd(i)+", ");
263 System.out.println("\""+match.getCapturedText(i)+"\"");
264 }
265 }
266 } catch (ParseException pe) {
267 if (pattern == null) {
268 pe.printStackTrace();
269 } else {
270 System.err.println("com.sun.org.apache.xerces.internal.utils.regex.ParseException: "+pe.getMessage());
271 String indent = " ";
272 System.err.println(indent+pattern);
273 int loc = pe.getLocation();
274 if (loc >= 0) {
275 System.err.print(indent);
276 for (int i = 0; i < loc; i ++) System.err.print("-");
277 System.err.println("^");
278 }
279 }
280 } catch (Exception e) {
281 e.printStackTrace();
282 }
283 }
284
285 static final int CACHESIZE = 20;
286 static final RegularExpression[] regexCache = new RegularExpression[CACHESIZE];
287
293 public static RegularExpression createRegex(String pattern, String options)
294 throws ParseException {
295 RegularExpression re = null;
296 int intOptions = REUtil.parseOptions(options);
297 synchronized (REUtil.regexCache) {
298 int i;
299 for (i = 0; i < REUtil.CACHESIZE; i ++) {
300 RegularExpression cached = REUtil.regexCache[i];
301 if (cached == null) {
302 i = -1;
303 break;
304 }
305 if (cached.equals(pattern, intOptions)) {
306 re = cached;
307 break;
308 }
309 }
310 if (re != null) {
311 if (i != 0) {
312 System.arraycopy(REUtil.regexCache, 0, REUtil.regexCache, 1, i);
313 REUtil.regexCache[0] = re;
314 }
315 } else {
316 re = new RegularExpression(pattern, options);
317 System.arraycopy(REUtil.regexCache, 0, REUtil.regexCache, 1, REUtil.CACHESIZE-1);
318 REUtil.regexCache[0] = re;
319 }
320 }
321 return re;
322 }
323
324
328 public static boolean matches(String regex, String target) throws ParseException {
329 return REUtil.createRegex(regex, null).matches(target);
330 }
331
332
336 public static boolean matches(String regex, String options, String target) throws ParseException {
337 return REUtil.createRegex(regex, options).matches(target);
338 }
339
340
342
345 public static String quoteMeta(String literal) {
346 int len = literal.length();
347 StringBuffer buffer = null;
348 for (int i = 0; i < len; i ++) {
349 int ch = literal.charAt(i);
350 if (".*+?{[()|\\^$".indexOf(ch) >= 0) {
351 if (buffer == null) {
352 buffer = new StringBuffer (i+(len-i)*2);
353 if (i > 0) buffer.append(literal.substring(0, i));
354 }
355 buffer.append((char)'\\');
356 buffer.append((char)ch);
357 } else if (buffer != null)
358 buffer.append((char)ch);
359 }
360 return buffer != null ? buffer.toString() : literal;
361 }
362
363
365 static void dumpString(String v) {
366 for (int i = 0; i < v.length(); i ++) {
367 System.out.print(Integer.toHexString(v.charAt(i)));
368 System.out.print(" ");
369 }
370 System.out.println();
371 }
372 }
373 |
Popular Tags |
Java API By Example, From Geeks To Geeks.