|
1
16
17 package org.apache.xerces.impl.xpath.regex;
18
19 import java.text.CharacterIterator  ;
20
21
26 public final class REUtil {
27 private REUtil() {
28 }
29
30 static final int composeFromSurrogates(int high, int low) {
31 return 0x10000 + ((high-0xd800)<<10) + low-0xdc00;
32 }
33
34 static final boolean isLowSurrogate(int ch) {
35 return (ch & 0xfc00) == 0xdc00;
36 }
37
38 static final boolean isHighSurrogate(int ch) {
39 return (ch & 0xfc00) == 0xd800;
40 }
41
42 static final String decomposeToSurrogates(int ch) {
43 char[] chs = new char[2];
44 ch -= 0x10000;
45 chs[0] = (char)((ch>>10)+0xd800);
46 chs[1] = (char)((ch&0x3ff)+0xdc00);
47 return new String (chs);
48 }
49
50 static final String substring(CharacterIterator iterator, int begin, int end) {
51 char[] src = new char[end-begin];
52 for (int i = 0; i < src.length; i ++)
53 src[i] = iterator.setIndex(i+begin);
54 return new String (src);
55 }
56
57
59 static final int getOptionValue(int ch) {
60 int ret = 0;
61 switch (ch) {
62 case 'i':
63 ret = RegularExpression.IGNORE_CASE;
64 break;
65 case 'm':
66 ret = RegularExpression.MULTIPLE_LINES;
67 break;
68 case 's':
69 ret = RegularExpression.SINGLE_LINE;
70 break;
71 case 'x':
72 ret = RegularExpression.EXTENDED_COMMENT;
73 break;
74 case 'u':
75 ret = RegularExpression.USE_UNICODE_CATEGORY;
76 break;
77 case 'w':
78 ret = RegularExpression.UNICODE_WORD_BOUNDARY;
79 break;
80 case 'F':
81 ret = RegularExpression.PROHIBIT_FIXED_STRING_OPTIMIZATION;
82 break;
83 case 'H':
84 ret = RegularExpression.PROHIBIT_HEAD_CHARACTER_OPTIMIZATION;
85 break;
86 case 'X':
87 ret = RegularExpression.XMLSCHEMA_MODE;
88 break;
89 case ',':
90 ret = RegularExpression.SPECIAL_COMMA;
91 break;
92 default:
93 }
94 return ret;
95 }
96
97 static final int parseOptions(String opts) throws ParseException {
98 if (opts == null) return 0;
99 int options = 0;
100 for (int i = 0; i < opts.length(); i ++) {
101 int v = getOptionValue(opts.charAt(i));
102 if (v == 0)
103 throw new ParseException("Unknown Option: "+opts.substring(i), -1);
104 options |= v;
105 }
106 return options;
107 }
108
109 static final String createOptionString(int options) {
110 StringBuffer sb = new StringBuffer (9);
111 if ((options & RegularExpression.PROHIBIT_FIXED_STRING_OPTIMIZATION) != 0)
112 sb.append((char)'F');
113 if ((options & RegularExpression.PROHIBIT_HEAD_CHARACTER_OPTIMIZATION) != 0)
114 sb.append((char)'H');
115 if ((options & RegularExpression.XMLSCHEMA_MODE) != 0)
116 sb.append((char)'X');
117 if ((options & RegularExpression.IGNORE_CASE) != 0)
118 sb.append((char)'i');
119 if ((options & RegularExpression.MULTIPLE_LINES) != 0)
120 sb.append((char)'m');
121 if ((options & RegularExpression.SINGLE_LINE) != 0)
122 sb.append((char)'s');
123 if ((options & RegularExpression.USE_UNICODE_CATEGORY) != 0)
124 sb.append((char)'u');
125 if ((options & RegularExpression.UNICODE_WORD_BOUNDARY) != 0)
126 sb.append((char)'w');
127 if ((options & RegularExpression.EXTENDED_COMMENT) != 0)
128 sb.append((char)'x');
129 if ((options & RegularExpression.SPECIAL_COMMA) != 0)
130 sb.append((char)',');
131 return sb.toString().intern();
132 }
133
134
136 static String stripExtendedComment(String regex) {
137 int len = regex.length();
138 StringBuffer buffer = new StringBuffer (len);
139 int offset = 0;
140 while (offset < len) {
141 int ch = regex.charAt(offset++);
142 if (ch == '\t' || ch == '\n' || ch == '\f' || ch == '\r' || ch == ' ')
144 continue;
145
146 if (ch == '#') { while (offset < len) {
148 ch = regex.charAt(offset++);
149 if (ch == '\r' || ch == '\n')
150 break;
151 }
152 continue;
153 }
154
155 int next; if (ch == '\\' && offset < len) {
157 if ((next = regex.charAt(offset)) == '#'
158 || next == '\t' || next == '\n' || next == '\f'
159 || next == '\r' || next == ' ') {
160 buffer.append((char)next);
161 offset ++;
162 } else { buffer.append((char)'\\');
164 buffer.append((char)next);
165 offset ++;
166 }
167 } else buffer.append((char)ch);
169 }
170 return buffer.toString();
171 }
172
173
175
179 public static void main(String [] argv) {
180 String pattern = null;
181 try {
182 String options = "";
183 String target = null;
184 if( argv.length == 0 ) {
185 System.out.println( "Error:Usage: java REUtil -i|-m|-s|-u|-w|-X regularExpression String" );
186 System.exit( 0 );
187 }
188 for (int i = 0; i < argv.length; i ++) {
189 if (argv[i].length() == 0 || argv[i].charAt(0) != '-') {
190 if (pattern == null)
191 pattern = argv[i];
192 else if (target == null)
193 target = argv[i];
194 else
195 System.err.println("Unnecessary: "+argv[i]);
196 } else if (argv[i].equals("-i")) {
197 options += "i";
198 } else if (argv[i].equals("-m")) {
199 options += "m";
200 } else if (argv[i].equals("-s")) {
201 options += "s";
202 } else if (argv[i].equals("-u")) {
203 options += "u";
204 } else if (argv[i].equals("-w")) {
205 options += "w";
206 } else if (argv[i].equals("-X")) {
207 options += "X";
208 } else {
209 System.err.println("Unknown option: "+argv[i]);
210 }
211 }
212 RegularExpression reg = new RegularExpression(pattern, options);
213 System.out.println("RegularExpression: "+reg);
214 Match match = new Match();
215 reg.matches(target, match);
216 for (int i = 0; i < match.getNumberOfGroups(); i ++) {
217 if (i == 0 ) System.out.print("Matched range for the whole pattern: ");
218 else System.out.print("["+i+"]: ");
219 if (match.getBeginning(i) < 0)
220 System.out.println("-1");
221 else {
222 System.out.print(match.getBeginning(i)+", "+match.getEnd(i)+", ");
223 System.out.println("\""+match.getCapturedText(i)+"\"");
224 }
225 }
226 } catch (ParseException pe) {
227 if (pattern == null) {
228 pe.printStackTrace();
229 } else {
230 System.err.println("org.apache.xerces.utils.regex.ParseException: "+pe.getMessage());
231 String indent = " ";
232 System.err.println(indent+pattern);
233 int loc = pe.getLocation();
234 if (loc >= 0) {
235 System.err.print(indent);
236 for (int i = 0; i < loc; i ++) System.err.print("-");
237 System.err.println("^");
238 }
239 }
240 } catch (Exception e) {
241 e.printStackTrace();
242 }
243 }
244
245 static final int CACHESIZE = 20;
246 static final RegularExpression[] regexCache = new RegularExpression[CACHESIZE];
247
253 public static RegularExpression createRegex(String pattern, String options)
254 throws ParseException {
255 RegularExpression re = null;
256 int intOptions = REUtil.parseOptions(options);
257 synchronized (REUtil.regexCache) {
258 int i;
259 for (i = 0; i < REUtil.CACHESIZE; i ++) {
260 RegularExpression cached = REUtil.regexCache[i];
261 if (cached == null) {
262 i = -1;
263 break;
264 }
265 if (cached.equals(pattern, intOptions)) {
266 re = cached;
267 break;
268 }
269 }
270 if (re != null) {
271 if (i != 0) {
272 System.arraycopy(REUtil.regexCache, 0, REUtil.regexCache, 1, i);
273 REUtil.regexCache[0] = re;
274 }
275 } else {
276 re = new RegularExpression(pattern, options);
277 System.arraycopy(REUtil.regexCache, 0, REUtil.regexCache, 1, REUtil.CACHESIZE-1);
278 REUtil.regexCache[0] = re;
279 }
280 }
281 return re;
282 }
283
284
288 public static boolean matches(String regex, String target) throws ParseException {
289 return REUtil.createRegex(regex, null).matches(target);
290 }
291
292
296 public static boolean matches(String regex, String options, String target) throws ParseException {
297 return REUtil.createRegex(regex, options).matches(target);
298 }
299
300
302
305 public static String quoteMeta(String literal) {
306 int len = literal.length();
307 StringBuffer buffer = null;
308 for (int i = 0; i < len; i ++) {
309 int ch = literal.charAt(i);
310 if (".*+?{[()|\\^$".indexOf(ch) >= 0) {
311 if (buffer == null) {
312 buffer = new StringBuffer (i+(len-i)*2);
313 if (i > 0) buffer.append(literal.substring(0, i));
314 }
315 buffer.append((char)'\\');
316 buffer.append((char)ch);
317 } else if (buffer != null)
318 buffer.append((char)ch);
319 }
320 return buffer != null ? buffer.toString() : literal;
321 }
322
323
325 static void dumpString(String v) {
326 for (int i = 0; i < v.length(); i ++) {
327 System.out.print(Integer.toHexString(v.charAt(i)));
328 System.out.print(" ");
329 }
330 System.out.println();
331 }
332 }
333 |
Popular Tags |
Java API By Example, From Geeks To Geeks.