1 | /* |
---|
2 | * pymigemo.c - C/Migemo Python binding |
---|
3 | * Copyright(C) 2005-2012, Atzm WATANABE <atzm@atzm.org> |
---|
4 | * |
---|
5 | * $Id$ |
---|
6 | */ |
---|
7 | |
---|
8 | #include <Python.h> |
---|
9 | #include <structmember.h> |
---|
10 | #include <migemo.h> |
---|
11 | #include <stdbool.h> |
---|
12 | #include <string.h> |
---|
13 | #include <stdlib.h> |
---|
14 | #include <errno.h> |
---|
15 | #include <sys/types.h> |
---|
16 | #include <sys/stat.h> |
---|
17 | #include <fcntl.h> |
---|
18 | #include <unistd.h> |
---|
19 | |
---|
20 | #define PYMIGEMO_VERSION "0.4" |
---|
21 | |
---|
22 | #if (PY_MAJOR_VERSION == 3 && PY_MINOR_VERSION >= 1) || PY_MAJOR_VERSION > 3 |
---|
23 | # define PYTHON3 |
---|
24 | #endif |
---|
25 | |
---|
26 | #ifndef Py_TYPE |
---|
27 | # define Py_TYPE(ob) (((PyObject *)(ob))->ob_type) |
---|
28 | #endif |
---|
29 | |
---|
30 | /* for dereference migemo object members */ |
---|
31 | struct _migemo { |
---|
32 | int enable; |
---|
33 | void *mtree; |
---|
34 | int charset; |
---|
35 | void *roma2hira; |
---|
36 | void *hira2kata; |
---|
37 | void *han2zen; |
---|
38 | void *zen2han; |
---|
39 | void *rx; |
---|
40 | void *addword; |
---|
41 | void *char2int; |
---|
42 | }; |
---|
43 | |
---|
44 | typedef struct { |
---|
45 | PyObject_HEAD |
---|
46 | migemo *migemo_obj; |
---|
47 | } Migemo; |
---|
48 | |
---|
49 | static bool |
---|
50 | get_encoding(char *encoding, size_t size, int charset) |
---|
51 | { |
---|
52 | char *enc; |
---|
53 | |
---|
54 | switch(charset) { |
---|
55 | case 1: |
---|
56 | enc = "cp932"; |
---|
57 | break; |
---|
58 | case 2: |
---|
59 | enc = "euc_jp"; |
---|
60 | break; |
---|
61 | case 3: |
---|
62 | enc = "utf_8"; |
---|
63 | break; |
---|
64 | default: |
---|
65 | enc = "ascii"; |
---|
66 | } |
---|
67 | |
---|
68 | if (strlen(enc) < size) { |
---|
69 | strcpy(encoding, enc); |
---|
70 | return true; |
---|
71 | } |
---|
72 | |
---|
73 | return false; |
---|
74 | } |
---|
75 | |
---|
76 | static int |
---|
77 | isloadable(const char *path) |
---|
78 | { |
---|
79 | struct stat st; |
---|
80 | int ret = 0; |
---|
81 | int fd = open(path, O_RDONLY); |
---|
82 | |
---|
83 | if (fd < 0) { |
---|
84 | return errno; |
---|
85 | } |
---|
86 | |
---|
87 | if (fstat(fd, &st) < 0) { |
---|
88 | ret = errno; |
---|
89 | goto isloadable_end; |
---|
90 | } |
---|
91 | if (S_ISDIR(st.st_mode)) { |
---|
92 | ret = EISDIR; |
---|
93 | goto isloadable_end; |
---|
94 | } |
---|
95 | |
---|
96 | isloadable_end: |
---|
97 | if (close(fd) < 0) { |
---|
98 | ret = errno; |
---|
99 | } |
---|
100 | return ret; |
---|
101 | } |
---|
102 | |
---|
103 | static void |
---|
104 | Migemo_dealloc(Migemo *self) |
---|
105 | { |
---|
106 | if (self->migemo_obj) { |
---|
107 | migemo_close(self->migemo_obj); |
---|
108 | } |
---|
109 | |
---|
110 | Py_TYPE(self)->tp_free((PyObject *)self); |
---|
111 | } |
---|
112 | |
---|
113 | static PyObject * |
---|
114 | Migemo_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
---|
115 | { |
---|
116 | Migemo *self; |
---|
117 | |
---|
118 | self = (Migemo *)type->tp_alloc(type, 0); |
---|
119 | |
---|
120 | if (self != NULL) { |
---|
121 | self->migemo_obj = NULL; |
---|
122 | } |
---|
123 | |
---|
124 | return (PyObject *)self; |
---|
125 | } |
---|
126 | |
---|
127 | static int |
---|
128 | Migemo_init(Migemo *self, PyObject *args, PyObject *kwds) |
---|
129 | { |
---|
130 | migemo *migemo_obj; |
---|
131 | char *dictionary; |
---|
132 | |
---|
133 | static char *kwlist[] = {"dictionary", NULL}; |
---|
134 | |
---|
135 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "s", kwlist, &dictionary)) { |
---|
136 | return -1; |
---|
137 | } |
---|
138 | |
---|
139 | if (dictionary) { |
---|
140 | int ret = isloadable(dictionary); |
---|
141 | |
---|
142 | if (ret != 0) { |
---|
143 | PyErr_SetString(PyExc_ValueError, strerror(ret)); |
---|
144 | return -1; |
---|
145 | } |
---|
146 | |
---|
147 | if (self->migemo_obj) { |
---|
148 | migemo_close(self->migemo_obj); |
---|
149 | } |
---|
150 | |
---|
151 | migemo_obj = migemo_open(dictionary); |
---|
152 | |
---|
153 | if (migemo_obj) { |
---|
154 | self->migemo_obj = migemo_obj; |
---|
155 | } |
---|
156 | else { |
---|
157 | PyErr_SetString(PyExc_AssertionError, "migemo_open() failed"); |
---|
158 | return -1; |
---|
159 | } |
---|
160 | } |
---|
161 | |
---|
162 | return 0; |
---|
163 | } |
---|
164 | |
---|
165 | static PyObject * |
---|
166 | Migemo_get_encoding(Migemo *self) |
---|
167 | { |
---|
168 | char encoding[7]; |
---|
169 | |
---|
170 | if (!get_encoding(encoding, sizeof(encoding), self->migemo_obj->charset)) { |
---|
171 | PyErr_SetString(PyExc_AssertionError, "get_encoding() failed"); |
---|
172 | return NULL; |
---|
173 | } |
---|
174 | |
---|
175 | return PyBytes_FromString(encoding); |
---|
176 | } |
---|
177 | |
---|
178 | static PyObject * |
---|
179 | Migemo_query(Migemo *self, PyObject *args, PyObject *kwds) |
---|
180 | { |
---|
181 | PyObject *result, *pyquery, *pyrestr; |
---|
182 | char encoding[7], *query = NULL; |
---|
183 | unsigned char *regex; |
---|
184 | |
---|
185 | static char *kwlist[] = {"query", NULL}; |
---|
186 | |
---|
187 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O", kwlist, &pyquery)) { |
---|
188 | return NULL; |
---|
189 | } |
---|
190 | |
---|
191 | if (!get_encoding(encoding, sizeof(encoding), self->migemo_obj->charset)) { |
---|
192 | PyErr_SetString(PyExc_AssertionError, "get_encoding() failed"); |
---|
193 | return NULL; |
---|
194 | } |
---|
195 | |
---|
196 | if (PyUnicode_Check(pyquery)) { |
---|
197 | PyObject *q = PyUnicode_AsEncodedString(pyquery, encoding, "strict"); |
---|
198 | |
---|
199 | if (q == NULL) { |
---|
200 | return NULL; |
---|
201 | } |
---|
202 | |
---|
203 | query = strdup(PyBytes_AS_STRING(q)); |
---|
204 | Py_DECREF(q); |
---|
205 | |
---|
206 | if (query == NULL) { |
---|
207 | return PyErr_NoMemory(); |
---|
208 | } |
---|
209 | } |
---|
210 | else if (PyBytes_Check(pyquery)) { |
---|
211 | query = strdup(PyBytes_AS_STRING(pyquery)); |
---|
212 | |
---|
213 | if (query == NULL) { |
---|
214 | return PyErr_NoMemory(); |
---|
215 | } |
---|
216 | } |
---|
217 | else { |
---|
218 | PyErr_SetString(PyExc_ValueError, "argument must be string"); |
---|
219 | return NULL; |
---|
220 | } |
---|
221 | |
---|
222 | regex = migemo_query(self->migemo_obj, query); |
---|
223 | free(query); |
---|
224 | if (regex == NULL) { |
---|
225 | PyErr_SetString(PyExc_AssertionError, "migemo_query() failed"); |
---|
226 | return NULL; |
---|
227 | } |
---|
228 | |
---|
229 | pyrestr = PyBytes_FromString(regex); |
---|
230 | migemo_release(self->migemo_obj, regex); |
---|
231 | if (pyrestr == NULL) { |
---|
232 | return NULL; |
---|
233 | } |
---|
234 | |
---|
235 | result = PyUnicode_FromEncodedObject(pyrestr, encoding, "strict"); |
---|
236 | Py_DECREF(pyrestr); |
---|
237 | return result; |
---|
238 | } |
---|
239 | |
---|
240 | static PyObject * |
---|
241 | Migemo_set_operator(Migemo *self, PyObject *args, PyObject *kwds) |
---|
242 | { |
---|
243 | PyObject *result = NULL; |
---|
244 | char *op; |
---|
245 | int index; |
---|
246 | |
---|
247 | static char *kwlist[] = {"index", "op", NULL}; |
---|
248 | |
---|
249 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "is", kwlist, &index, &op)) { |
---|
250 | return NULL; |
---|
251 | } |
---|
252 | |
---|
253 | if (op) { |
---|
254 | result = PyBool_FromLong((long)migemo_set_operator(self->migemo_obj, index, op)); |
---|
255 | } |
---|
256 | |
---|
257 | return result; |
---|
258 | } |
---|
259 | |
---|
260 | static PyObject * |
---|
261 | Migemo_get_operator(Migemo *self, PyObject *args, PyObject *kwds) |
---|
262 | { |
---|
263 | PyObject *result = NULL; |
---|
264 | const unsigned char *op; |
---|
265 | int index; |
---|
266 | |
---|
267 | static char *kwlist[] = {"index", NULL}; |
---|
268 | |
---|
269 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "i", kwlist, &index)) { |
---|
270 | return NULL; |
---|
271 | } |
---|
272 | |
---|
273 | if (op = migemo_get_operator(self->migemo_obj, index)) { |
---|
274 | result = PyBytes_FromString(op); |
---|
275 | } |
---|
276 | else { |
---|
277 | PyErr_SetString(PyExc_ValueError, "invalid opindex"); |
---|
278 | } |
---|
279 | |
---|
280 | return result; |
---|
281 | } |
---|
282 | |
---|
283 | static PyObject * |
---|
284 | Migemo_load(Migemo *self, PyObject *args, PyObject *kwds) |
---|
285 | { |
---|
286 | PyObject *result = NULL; |
---|
287 | char *dict_file; |
---|
288 | int dict_id; |
---|
289 | |
---|
290 | static char *kwlist[] = {"dict_id", "dict_file", NULL}; |
---|
291 | |
---|
292 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "is", kwlist, &dict_id, &dict_file)) { |
---|
293 | return NULL; |
---|
294 | } |
---|
295 | |
---|
296 | if (dict_file) { |
---|
297 | int ret = isloadable(dict_file); |
---|
298 | |
---|
299 | if (ret != 0) { |
---|
300 | PyErr_SetString(PyExc_ValueError, strerror(ret)); |
---|
301 | return NULL; |
---|
302 | } |
---|
303 | |
---|
304 | result = PyLong_FromLong((long)migemo_load(self->migemo_obj, dict_id, dict_file)); |
---|
305 | } |
---|
306 | |
---|
307 | return result; |
---|
308 | } |
---|
309 | |
---|
310 | static PyObject * |
---|
311 | Migemo_is_enable(Migemo *self) |
---|
312 | { |
---|
313 | return PyBool_FromLong((long)migemo_is_enable(self->migemo_obj)); |
---|
314 | } |
---|
315 | |
---|
316 | static PyMethodDef Migemo_methods[] = { |
---|
317 | {"query", (PyCFunction)Migemo_query, METH_VARARGS | METH_KEYWORDS, |
---|
318 | "return regex from romaji string\n\ |
---|
319 | \n\ |
---|
320 | def query(query)\n\ |
---|
321 | query: romaji string (str or unicode)\n\ |
---|
322 | \n\ |
---|
323 | returns: regex string as Unicode object"}, |
---|
324 | {"set_operator", (PyCFunction)Migemo_set_operator, METH_VARARGS | METH_KEYWORDS, |
---|
325 | "set operator string as the meta character of regex\n\ |
---|
326 | \n\ |
---|
327 | def set_operator(index, op):\n\ |
---|
328 | index: (OPINDEX_NEST_IN|OPINDEX_NEST_OUT|OPINDEX_NEWLINE|\n\ |
---|
329 | OPINDEX_OR|OPINDEX_SELECT_IN|OPINDEX_SELECT_OUT)\n\ |
---|
330 | op: operator string (str)\n\ |
---|
331 | \n\ |
---|
332 | returns: boolean value"}, |
---|
333 | {"get_operator", (PyCFunction)Migemo_get_operator, METH_VARARGS | METH_KEYWORDS, |
---|
334 | "get operator string as the meta character of regex\n\ |
---|
335 | \n\ |
---|
336 | def get_operator(index)\n\ |
---|
337 | index: (OPINDEX_NEST_IN|OPINDEX_NEST_OUT|OPINDEX_NEWLINE|\n\ |
---|
338 | OPINDEX_OR|OPINDEX_SELECT_IN|OPINDEX_SELECT_OUT)\n\ |
---|
339 | \n\ |
---|
340 | returns: operator string (str)"}, |
---|
341 | {"load", (PyCFunction)Migemo_load, METH_VARARGS | METH_KEYWORDS, |
---|
342 | "add dictionary to Migemo object\n\ |
---|
343 | \n\ |
---|
344 | def load(dict_id, dict_file)\n\ |
---|
345 | dict_id: (DICTID_HAN2ZEN|DICTID_HIRA2KATA|DICTID_MIGEMO|\n\ |
---|
346 | DICTID_ROMA2HIRA|DICTID_ZEN2HAN)\n\ |
---|
347 | dict_file: path to dictionary file (str)\n\ |
---|
348 | \n\ |
---|
349 | returns: ID of loaded dictionary"}, |
---|
350 | {"is_enable", (PyCFunction)Migemo_is_enable, METH_NOARGS, |
---|
351 | "check internal migemo_dict\n\ |
---|
352 | \n\ |
---|
353 | def is_enable()\n\ |
---|
354 | returns: boolean value"}, |
---|
355 | {"get_encoding", (PyCFunction)Migemo_get_encoding, METH_NOARGS, |
---|
356 | "get dictionary encoding\n\ |
---|
357 | \n\ |
---|
358 | def get_encoding()\n\ |
---|
359 | returns: encoding string (str)"}, |
---|
360 | {NULL} /* Sentinel */ |
---|
361 | }; |
---|
362 | |
---|
363 | static PyMemberDef Migemo_members[] = { |
---|
364 | {NULL} /* Sentinel */ |
---|
365 | }; |
---|
366 | |
---|
367 | #ifndef PyVarObject_HEAD_INIT |
---|
368 | # define PyVarObject_HEAD_INIT(type, size) PyObject_HEAD_INIT(type) size, |
---|
369 | #endif |
---|
370 | |
---|
371 | static PyTypeObject MigemoType = { |
---|
372 | PyVarObject_HEAD_INIT(NULL, 0) |
---|
373 | "migemo.Migemo", /*tp_name*/ |
---|
374 | sizeof(Migemo), /*tp_basicsize*/ |
---|
375 | 0, /*tp_itemsize*/ |
---|
376 | (destructor)Migemo_dealloc, /*tp_dealloc*/ |
---|
377 | 0, /*tp_print*/ |
---|
378 | 0, /*tp_getattr*/ |
---|
379 | 0, /*tp_setattr*/ |
---|
380 | 0, /*tp_compare*/ |
---|
381 | 0, /*tp_repr*/ |
---|
382 | 0, /*tp_as_number*/ |
---|
383 | 0, /*tp_as_sequence*/ |
---|
384 | 0, /*tp_as_mapping*/ |
---|
385 | 0, /*tp_hash */ |
---|
386 | 0, /*tp_call*/ |
---|
387 | 0, /*tp_str*/ |
---|
388 | 0, /*tp_getattro*/ |
---|
389 | 0, /*tp_setattro*/ |
---|
390 | 0, /*tp_as_buffer*/ |
---|
391 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE, /*tp_flags*/ |
---|
392 | "", /* tp_doc */ |
---|
393 | 0, /* tp_traverse */ |
---|
394 | 0, /* tp_clear */ |
---|
395 | 0, /* tp_richcompare */ |
---|
396 | 0, /* tp_weaklistoffset */ |
---|
397 | 0, /* tp_iter */ |
---|
398 | 0, /* tp_iternext */ |
---|
399 | Migemo_methods, /* tp_methods */ |
---|
400 | Migemo_members, /* tp_members */ |
---|
401 | 0, /* tp_getset */ |
---|
402 | 0, /* tp_base */ |
---|
403 | 0, /* tp_dict */ |
---|
404 | 0, /* tp_descr_get */ |
---|
405 | 0, /* tp_descr_set */ |
---|
406 | 0, /* tp_dictoffset */ |
---|
407 | (initproc)Migemo_init, /* tp_init */ |
---|
408 | 0, /* tp_alloc */ |
---|
409 | Migemo_new, /* tp_new */ |
---|
410 | }; |
---|
411 | |
---|
412 | #define PYMIGEMO_MODULEDOC "C/Migemo Python binding" |
---|
413 | |
---|
414 | static PyMethodDef module_methods[] = { |
---|
415 | {NULL} /* Sentinel */ |
---|
416 | }; |
---|
417 | |
---|
418 | #ifdef PYTHON3 |
---|
419 | static struct PyModuleDef moduledef = { |
---|
420 | PyModuleDef_HEAD_INIT, |
---|
421 | "migemo", /* m_name */ |
---|
422 | PYMIGEMO_MODULEDOC, /* m_doc */ |
---|
423 | -1, /* m_size */ |
---|
424 | module_methods, /* m_methods */ |
---|
425 | NULL, /* m_reload */ |
---|
426 | NULL, /* m_traverse */ |
---|
427 | NULL, /* m_clear */ |
---|
428 | NULL, /* m_free */ |
---|
429 | }; |
---|
430 | #endif |
---|
431 | |
---|
432 | #ifdef PYTHON3 |
---|
433 | # define MOD_INIT(name) PyObject *PyInit_##name(void) |
---|
434 | #else |
---|
435 | # define MOD_INIT(name) void init##name(void) |
---|
436 | #endif |
---|
437 | |
---|
438 | MOD_INIT(migemo) |
---|
439 | { |
---|
440 | PyObject* m; |
---|
441 | |
---|
442 | if (PyType_Ready(&MigemoType) < 0) |
---|
443 | return; |
---|
444 | |
---|
445 | #ifdef PYTHON3 |
---|
446 | m = PyModule_Create(&moduledef); |
---|
447 | #else |
---|
448 | m = Py_InitModule3("migemo", module_methods, PYMIGEMO_MODULEDOC); |
---|
449 | #endif |
---|
450 | |
---|
451 | if (m == NULL) { |
---|
452 | #ifdef PYTHON3 |
---|
453 | return NULL; |
---|
454 | #else |
---|
455 | return; |
---|
456 | #endif |
---|
457 | } |
---|
458 | |
---|
459 | Py_INCREF(&MigemoType); |
---|
460 | PyModule_AddObject(m, "Migemo", (PyObject *)&MigemoType); |
---|
461 | PyModule_AddObject(m, "PYMIGEMO_VERSION", Py_BuildValue("s", PYMIGEMO_VERSION)); |
---|
462 | |
---|
463 | PyModule_AddObject(m, "MIGEMO_VERSION", Py_BuildValue("s", MIGEMO_VERSION)); |
---|
464 | |
---|
465 | PyModule_AddObject(m, "DICTID_INVALID", Py_BuildValue("i", MIGEMO_DICTID_INVALID)); |
---|
466 | PyModule_AddObject(m, "DICTID_MIGEMO", Py_BuildValue("i", MIGEMO_DICTID_MIGEMO)); |
---|
467 | PyModule_AddObject(m, "DICTID_ROMA2HIRA", Py_BuildValue("i", MIGEMO_DICTID_ROMA2HIRA)); |
---|
468 | PyModule_AddObject(m, "DICTID_HIRA2KATA", Py_BuildValue("i", MIGEMO_DICTID_HIRA2KATA)); |
---|
469 | PyModule_AddObject(m, "DICTID_HAN2ZEN", Py_BuildValue("i", MIGEMO_DICTID_HAN2ZEN)); |
---|
470 | PyModule_AddObject(m, "DICTID_ZEN2HAN", Py_BuildValue("i", MIGEMO_DICTID_ZEN2HAN)); |
---|
471 | |
---|
472 | PyModule_AddObject(m, "OPINDEX_OR", Py_BuildValue("i", MIGEMO_OPINDEX_OR)); |
---|
473 | PyModule_AddObject(m, "OPINDEX_NEST_IN", Py_BuildValue("i", MIGEMO_OPINDEX_NEST_IN)); |
---|
474 | PyModule_AddObject(m, "OPINDEX_NEST_OUT", Py_BuildValue("i", MIGEMO_OPINDEX_NEST_OUT)); |
---|
475 | PyModule_AddObject(m, "OPINDEX_SELECT_IN", Py_BuildValue("i", MIGEMO_OPINDEX_SELECT_IN)); |
---|
476 | PyModule_AddObject(m, "OPINDEX_SELECT_OUT", Py_BuildValue("i", MIGEMO_OPINDEX_SELECT_OUT)); |
---|
477 | PyModule_AddObject(m, "OPINDEX_NEWLINE", Py_BuildValue("i", MIGEMO_OPINDEX_NEWLINE)); |
---|
478 | |
---|
479 | #ifdef PYTHON3 |
---|
480 | return m; |
---|
481 | #endif |
---|
482 | } |
---|