|
1
15
16
17 package org.python.modules.sre;
18
19 import java.util.*;
20 import org.python.core.*;
21
22 public class PatternObject extends PyObject {
23 char[] code;
24 public String  pattern;
25 public int groups;
26 public org.python.core.PyObject groupindex;
27 public int flags;
28 org.python.core.PyObject indexgroup;
29 public int codesize;
30
31
32 public PatternObject(PyString pattern, int flags, char[] code,
33 int groups, PyObject groupindex, PyObject indexgroup) {
34
35 if (pattern != null)
36 this.pattern = pattern.toString();
37 this.flags = flags;
38 this.code = code;
39 this.codesize = code.length;
40 this.groups = groups;
41 this.groupindex = groupindex;
42 this.indexgroup = indexgroup;
43 }
44
45 public MatchObject match(String string) {
46 return match(string, 0, Integer.MAX_VALUE);
47 }
48
49 public MatchObject match(String string, int start) {
50 return match(string, start, Integer.MAX_VALUE);
51 }
52
53 public MatchObject match(String string, int start, int end) {
54 SRE_STATE state = new SRE_STATE(string, start, end, flags);
55
56 state.ptr = state.start;
57 int status = state.SRE_MATCH(code, 0, 1);
58
59 return _pattern_new_match(state, string, status);
60 }
61
62
63
64
65 public MatchObject search(PyObject[] args, String [] kws) {
66 ArgParser ap = new ArgParser("search", args, kws,
67 "pattern", "pos", "endpos");
68 String string = ap.getString(0);
69 int start = ap.getInt(1, 0);
70 int end = ap.getInt(2, string.length());
71
72 SRE_STATE state = new SRE_STATE(string, start, end, flags);
73
74 int status = state.SRE_SEARCH(code, 0);
75
76 return _pattern_new_match(state, string, status);
77 }
78
79
80 public PyObject sub(PyObject[] args, String [] kws) {
81 ArgParser ap = new ArgParser("sub", args, kws,
82 "repl", "string", "count");
83 PyObject template = ap.getPyObject(0);
84 String string = ap.getString(1);
85 int count = ap.getInt(2, 0);
86
87 return subx(template, string, count, false);
88 }
89
90
91
92 public PyObject subn(PyObject[] args, String [] kws) {
93 ArgParser ap = new ArgParser("subn", args, kws,
94 "repl", "string", "count");
95 PyObject template = ap.getPyObject(0);
96 String string = ap.getString(1);
97 int count = ap.getInt(2, 0);
98
99 return subx(template, string, count, true);
100 }
101
102
103 private PyObject subx(PyObject template, String string, int count,
104 boolean subn)
105 {
106 PyObject filter = null;
107 boolean filter_is_callable = false;
108 if (template.isCallable()) {
109 filter = template;
110 filter_is_callable = true;
111 } else {
112 boolean literal = false;
113 if (template instanceof PyString) {
114 literal = template.toString().indexOf('\\') < 0;
115 }
116 if (literal) {
117 filter = template;
118 filter_is_callable = false;
119 } else {
120 filter = call("sre", "_subx", new PyObject[] {
121 this, template});
122 filter_is_callable = filter.isCallable();
123 }
124 }
125
126 SRE_STATE state = new SRE_STATE(string, 0, Integer.MAX_VALUE, flags);
127
128 StringBuffer buf = new StringBuffer ();
129
130 int n = 0;
131 int i = 0;
132
133 while (count == 0 || n < count) {
134 state.state_reset();
135 state.ptr = state.start;
136 int status = state.SRE_SEARCH(code, 0);
137 if (status <= 0) {
138 if (status == 0)
139 break;
140 _error(status);
141 }
142 int b = state.start;
143 int e = state.ptr;
144
145 if (i < b) {
146
147 buf.append(string.substring(i, b));
148 }
149 if (! (i == b && i == e && n > 0)) {
150 PyObject item;
151 if (filter_is_callable) {
152
153 MatchObject match = _pattern_new_match(state, string, 1);
154 item = filter.__call__(match);
155 } else {
156 item = filter;
157 }
158
159 if (item != Py.None) {
160 buf.append(item.toString());
161 }
162 i = e;
163 n++;
164 }
165
166
167 if (state.ptr == state.start)
168 state.start = state.ptr + 1;
169 else
170 state.start = state.ptr;
171 }
172 if (i < state.endpos) {
173 buf.append(string.substring(i, state.endpos));
174 }
175
176 if (subn)
177 return new PyTuple(new PyObject[] {
178 Py.newString(buf.toString()), Py.newInteger(n)
179 });
180 else
181 return Py.newString(buf.toString());
182 }
183
184
185 public PyObject split(PyObject[] args, String [] kws) {
186 ArgParser ap = new ArgParser("split", args, kws,
187 "source", "maxsplit");
188 String string = ap.getString(0);
189 int maxsplit = ap.getInt(1, 0);
190
191 SRE_STATE state = new SRE_STATE(string, 0, Integer.MAX_VALUE, flags);
192
193 PyList list = new PyList();
194
195 int n = 0;
196 int last = state.start;
197 while (maxsplit == 0 || n < maxsplit) {
198 state.state_reset();
199 state.ptr = state.start;
200 int status = state.SRE_SEARCH(code, 0);
201 if (status <= 0) {
202 if (status == 0)
203 break;
204 _error(status);
205 }
206 if (state.start == state.ptr) {
207 if (last == state.end)
208 break;
209
210 state.start = state.ptr + 1;
211 continue;
212 }
213
214
215 PyObject item = Py.newString(string.substring(last, state.start));
216 list.append(item);
217
218 for (int i = 0; i < groups; i++) {
219 String s = state.getslice(i+1, string, false);
220 if (s != null)
221 list.append(Py.newString(s));
222 else
223 list.append(Py.None);
224 }
225 n += 1;
226 last = state.start = state.ptr;
227 }
228
229 PyObject item = Py.newString(string.substring(last, state.endpos));
230 list.append(item);
231
232 return list;
233 }
234
235 private PyObject call(String module, String function, PyObject[] args) {
236 PyObject sre = imp.importName(module, true);
237 return sre.invoke(function, args);
238 }
239
240
241
242 public PyObject findall(PyObject[] args, String [] kws) {
243 ArgParser ap = new ArgParser("findall", args, kws,
244 "source", "pos", "endpos");
245 String string = ap.getString(0);
246 int start = ap.getInt(1, 0);
247 int end = ap.getInt(2, Integer.MAX_VALUE);
248
249 SRE_STATE state = new SRE_STATE(string, start, end, flags);
250
251 Vector list = new Vector();
252
253 while (state.start <= state.end) {
254 state.state_reset();
255 state.ptr = state.start;
256 int status = state.SRE_SEARCH(code, 0);
257 if (status > 0) {
258 PyObject item;
259
260
261 switch (groups) {
262 case 0:
263 item = Py.newString(
264 string.substring(state.start, state.ptr));
265 break;
266 case 1:
267 item = Py.newString(state.getslice(1, string, true));
268 break;
269 default:
270 PyObject[] t = new PyObject[groups];
271 for (int i = 0; i < groups; i++)
272 t[i] = Py.newString(state.getslice(i+1, string, true));
273 item = new PyTuple(t);
274 break;
275 }
276
277 list.addElement(item);
278
279 if (state.ptr == state.start)
280 state.start = state.ptr + 1;
281 else
282 state.start = state.ptr;
283 } else {
284
285 if (status == 0)
286 break;
287
288 _error(status);
289 }
290 }
291 return new PyList(list);
292 }
293
294
295 public PyObject finditer(String string) {
296 return finditer(string, 0, Integer.MAX_VALUE);
297 }
298
299 public PyObject finditer(String string, int start) {
300 return finditer(string, start, Integer.MAX_VALUE);
301 }
302
303 public PyObject finditer(String string, int start, int end) {
304 ScannerObject scanner = scanner(string, start, end);
305 PyObject search = scanner.__findattr__("search");
306 return new PyCallIter(search, Py.None);
307 }
308
309 public ScannerObject scanner(String string) {
310 return scanner(string, 0, Integer.MAX_VALUE);
311 }
312
313 public ScannerObject scanner(String string, int start) {
314 return scanner(string, start, Integer.MAX_VALUE);
315 }
316
317 public ScannerObject scanner(String string, int start, int end) {
318 ScannerObject self = new ScannerObject();
319 self.state = new SRE_STATE(string, start, end, flags);
320 self.pattern = this;
321 self.string = string;
322 return self;
323 }
324
325
326 private void _error(int status) {
327 if (status == SRE_STATE.SRE_ERROR_RECURSION_LIMIT)
328 throw Py.RuntimeError("maximum recursion limit exceeded");
329
330 throw Py.RuntimeError("internal error in regular expression engine");
331 }
332
333
334 MatchObject _pattern_new_match(SRE_STATE state, String string,
335 int status)
336 {
337
338
339
341 if (status > 0) {
342
343 MatchObject match = new MatchObject();
344 match.pattern = this;
345 match.string = string;
346 match.regs = null;
347 match.groups = groups+1;
348
349 int base = state.beginning;
350
351 match.mark = new int[match.groups*2];
352 match.mark[0] = state.start - base;
353 match.mark[1] = state.ptr - base;
354
355
356 int i, j;
357 for (i = j = 0; i < groups; i++, j+=2) {
358 if (j+1 <= state.lastmark && state.mark[j] != -1 &&
359 state.mark[j+1] != -1) {
360 match.mark[j+2] = state.mark[j] - base;
361 match.mark[j+3] = state.mark[j+1] - base;
362 } else
363 match.mark[j+2] = match.mark[j+3] = -1;
364 }
365 match.pos = state.pos;
366 match.endpos = state.endpos;
367 match.lastindex = state.lastindex;
368
369 return match;
370 } else if (status == 0) {
371 return null;
372 }
373
374 _error(status);
375 return null;
376 }
377 }
378
379
380 |
Popular Tags |
Java API By Example, From Geeks To Geeks.