|
a/src/python/recoll/pyrclextract.cpp |
|
b/src/python/recoll/pyrclextract.cpp |
|
... |
|
... |
49 |
static void
|
49 |
static void
|
50 |
Extractor_dealloc(rclx_ExtractorObject *self)
|
50 |
Extractor_dealloc(rclx_ExtractorObject *self)
|
51 |
{
|
51 |
{
|
52 |
LOGDEB(("Extractor_dealloc\n"));
|
52 |
LOGDEB(("Extractor_dealloc\n"));
|
53 |
delete self->xtr;
|
53 |
delete self->xtr;
|
54 |
self->ob_type->tp_free((PyObject*)self);
|
54 |
Py_TYPE(self)->tp_free((PyObject*)self);
|
55 |
}
|
55 |
}
|
56 |
|
56 |
|
57 |
static PyObject *
|
57 |
static PyObject *
|
58 |
Extractor_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
|
58 |
Extractor_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
|
59 |
{
|
59 |
{
|
|
... |
|
... |
189 |
PyErr_SetString(PyExc_AttributeError, "interntofile failure");
|
189 |
PyErr_SetString(PyExc_AttributeError, "interntofile failure");
|
190 |
return 0;
|
190 |
return 0;
|
191 |
}
|
191 |
}
|
192 |
if (outfile.empty())
|
192 |
if (outfile.empty())
|
193 |
temp->setnoremove(1);
|
193 |
temp->setnoremove(1);
|
194 |
PyObject *result = outfile.empty() ? PyString_FromString(temp->filename()) :
|
194 |
PyObject *result = outfile.empty() ? PyBytes_FromString(temp->filename()) :
|
195 |
PyString_FromString(outfile.c_str());
|
195 |
PyBytes_FromString(outfile.c_str());
|
196 |
return (PyObject *)result;
|
196 |
return (PyObject *)result;
|
197 |
}
|
197 |
}
|
198 |
|
198 |
|
199 |
static PyMethodDef Extractor_methods[] = {
|
199 |
static PyMethodDef Extractor_methods[] = {
|
200 |
{"textextract", (PyCFunction)Extractor_textextract,
|
200 |
{"textextract", (PyCFunction)Extractor_textextract,
|
|
... |
|
... |
209 |
"\n"
|
209 |
"\n"
|
210 |
"An Extractor object can extract data from a native simple or compound\n"
|
210 |
"An Extractor object can extract data from a native simple or compound\n"
|
211 |
"object.\n"
|
211 |
"object.\n"
|
212 |
);
|
212 |
);
|
213 |
static PyTypeObject rclx_ExtractorType = {
|
213 |
static PyTypeObject rclx_ExtractorType = {
|
214 |
PyObject_HEAD_INIT(NULL)
|
214 |
PyVarObject_HEAD_INIT(NULL, 0)
|
215 |
0, /*ob_size*/
|
|
|
216 |
"rclextract.Extractor", /*tp_name*/
|
215 |
"rclextract.Extractor", /*tp_name*/
|
217 |
sizeof(rclx_ExtractorObject), /*tp_basicsize*/
|
216 |
sizeof(rclx_ExtractorObject), /*tp_basicsize*/
|
218 |
0, /*tp_itemsize*/
|
217 |
0, /*tp_itemsize*/
|
219 |
(destructor)Extractor_dealloc, /*tp_dealloc*/
|
218 |
(destructor)Extractor_dealloc, /*tp_dealloc*/
|
220 |
0, /*tp_print*/
|
219 |
0, /*tp_print*/
|
|
... |
|
... |
251 |
0, /* tp_alloc */
|
250 |
0, /* tp_alloc */
|
252 |
Extractor_new, /* tp_new */
|
251 |
Extractor_new, /* tp_new */
|
253 |
};
|
252 |
};
|
254 |
|
253 |
|
255 |
///////////////////////////////////// Module-level stuff
|
254 |
///////////////////////////////////// Module-level stuff
|
256 |
static PyMethodDef rclxMethods[] = {
|
255 |
static PyMethodDef rclextract_methods[] = {
|
257 |
{NULL, NULL, 0, NULL} /* Sentinel */
|
256 |
{NULL, NULL, 0, NULL} /* Sentinel */
|
258 |
};
|
257 |
};
|
259 |
PyDoc_STRVAR(rclx_doc_string,
|
258 |
PyDoc_STRVAR(rclx_doc_string,
|
260 |
"This is an interface to the Recoll text extraction features.");
|
259 |
"This is an interface to the Recoll text extraction features.");
|
261 |
|
260 |
|
262 |
#ifndef PyMODINIT_FUNC /* declarations for DLL import/export */
|
261 |
struct module_state {
|
263 |
#define PyMODINIT_FUNC void
|
262 |
PyObject *error;
|
|
|
263 |
};
|
|
|
264 |
|
|
|
265 |
#if PY_MAJOR_VERSION >= 3
|
|
|
266 |
#define GETSTATE(m) ((struct module_state*)PyModule_GetState(m))
|
|
|
267 |
#else
|
|
|
268 |
#define GETSTATE(m) (&_state)
|
|
|
269 |
static struct module_state _state;
|
264 |
#endif
|
270 |
#endif
|
|
|
271 |
|
|
|
272 |
#if PY_MAJOR_VERSION >= 3
|
|
|
273 |
static int rclextract_traverse(PyObject *m, visitproc visit, void *arg) {
|
|
|
274 |
Py_VISIT(GETSTATE(m)->error);
|
|
|
275 |
return 0;
|
|
|
276 |
}
|
|
|
277 |
|
|
|
278 |
static int rclextract_clear(PyObject *m) {
|
|
|
279 |
Py_CLEAR(GETSTATE(m)->error);
|
|
|
280 |
return 0;
|
|
|
281 |
}
|
|
|
282 |
|
|
|
283 |
static struct PyModuleDef moduledef = {
|
|
|
284 |
PyModuleDef_HEAD_INIT,
|
|
|
285 |
"rclextract",
|
|
|
286 |
NULL,
|
|
|
287 |
sizeof(struct module_state),
|
|
|
288 |
rclextract_methods,
|
|
|
289 |
NULL,
|
|
|
290 |
rclextract_traverse,
|
|
|
291 |
rclextract_clear,
|
|
|
292 |
NULL
|
|
|
293 |
};
|
|
|
294 |
|
|
|
295 |
#define INITERROR return NULL
|
|
|
296 |
|
|
|
297 |
extern "C" PyObject *
|
|
|
298 |
PyInit_rclextract(void)
|
|
|
299 |
|
|
|
300 |
#else
|
|
|
301 |
#define INITERROR return
|
265 |
PyMODINIT_FUNC
|
302 |
PyMODINIT_FUNC
|
266 |
initrclextract(void)
|
303 |
initrclextract(void)
|
|
|
304 |
#endif
|
267 |
{
|
305 |
{
|
268 |
// We run recollinit. It's responsible for initializing some static data
|
306 |
// We run recollinit. It's responsible for initializing some static data
|
269 |
// which is distinct from pyrecoll's as we're separately dlopened
|
307 |
// which is distinct from pyrecoll's as we're separately dlopened
|
270 |
string reason;
|
308 |
string reason;
|
271 |
rclconfig = recollinit(0, 0, reason, 0);
|
309 |
rclconfig = recollinit(0, 0, reason, 0);
|
272 |
if (rclconfig == 0) {
|
310 |
if (rclconfig == 0) {
|
273 |
PyErr_SetString(PyExc_EnvironmentError, reason.c_str());
|
311 |
PyErr_SetString(PyExc_EnvironmentError, reason.c_str());
|
274 |
return;
|
312 |
INITERROR;
|
275 |
}
|
313 |
}
|
276 |
if (!rclconfig->ok()) {
|
314 |
if (!rclconfig->ok()) {
|
277 |
PyErr_SetString(PyExc_EnvironmentError,
|
315 |
PyErr_SetString(PyExc_EnvironmentError,
|
278 |
"Recoll init error: bad environment ?");
|
316 |
"Recoll init error: bad environment ?");
|
279 |
return;
|
317 |
INITERROR;
|
280 |
}
|
318 |
}
|
281 |
|
319 |
|
|
|
320 |
#if PY_MAJOR_VERSION >= 3
|
|
|
321 |
PyObject *module = PyModule_Create(&moduledef);
|
|
|
322 |
#else
|
282 |
PyObject* m = Py_InitModule("rclextract", rclxMethods);
|
323 |
PyObject *module = Py_InitModule("rclextract", rclextract_methods);
|
|
|
324 |
#endif
|
|
|
325 |
if (module == NULL)
|
|
|
326 |
INITERROR;
|
|
|
327 |
|
|
|
328 |
struct module_state *st = GETSTATE(module);
|
|
|
329 |
// The first parameter is a char *. Hopefully we don't initialize
|
|
|
330 |
// modules too often...
|
|
|
331 |
st->error = PyErr_NewException(strdup("rclextract.Error"), NULL, NULL);
|
|
|
332 |
if (st->error == NULL) {
|
|
|
333 |
Py_DECREF(module);
|
|
|
334 |
INITERROR;
|
|
|
335 |
}
|
|
|
336 |
|
283 |
PyModule_AddStringConstant(m, "__doc__", rclx_doc_string);
|
337 |
PyModule_AddStringConstant(module, "__doc__", rclx_doc_string);
|
284 |
|
338 |
|
285 |
if (PyType_Ready(&rclx_ExtractorType) < 0)
|
339 |
if (PyType_Ready(&rclx_ExtractorType) < 0)
|
286 |
return;
|
340 |
INITERROR;
|
287 |
Py_INCREF(&rclx_ExtractorType);
|
341 |
Py_INCREF(&rclx_ExtractorType);
|
288 |
PyModule_AddObject(m, "Extractor", (PyObject *)&rclx_ExtractorType);
|
342 |
PyModule_AddObject(module, "Extractor", (PyObject *)&rclx_ExtractorType);
|
289 |
|
343 |
|
290 |
#if PY_MAJOR_VERSION >= 2 && PY_MINOR_VERSION >= 7
|
344 |
#if PY_MAJOR_VERSION >= 3 || (PY_MAJOR_VERSION >= 2 && PY_MINOR_VERSION >= 7)
|
291 |
recoll_DocType = (PyObject*)PyCapsule_Import(PYRECOLL_PACKAGE "recoll.doctypeptr", 0);
|
345 |
recoll_DocType = (PyObject*)PyCapsule_Import(PYRECOLL_PACKAGE "recoll.doctypeptr", 0);
|
292 |
#else
|
346 |
#else
|
293 |
PyObject *module = PyImport_ImportModule(PYRECOLL_PACKAGE "recoll");
|
347 |
PyObject *module = PyImport_ImportModule(PYRECOLL_PACKAGE "recoll");
|
294 |
if (module != NULL) {
|
348 |
if (module != NULL) {
|
295 |
PyObject *cobject = PyObject_GetAttrString(module, "_C_API");
|
349 |
PyObject *cobject = PyObject_GetAttrString(module, "_C_API");
|
296 |
if (cobject == NULL)
|
350 |
if (cobject == NULL)
|
297 |
return;
|
351 |
INITERROR;
|
298 |
if (PyCObject_Check(cobject))
|
352 |
if (PyCObject_Check(cobject))
|
299 |
recoll_DocType = (PyObject*)PyCObject_AsVoidPtr(cobject);
|
353 |
recoll_DocType = (PyObject*)PyCObject_AsVoidPtr(cobject);
|
300 |
Py_DECREF(cobject);
|
354 |
Py_DECREF(cobject);
|
301 |
}
|
355 |
}
|
302 |
#endif
|
356 |
#endif
|
|
|
357 |
|
|
|
358 |
#if PY_MAJOR_VERSION >= 3
|
|
|
359 |
return module;
|
|
|
360 |
#endif
|
303 |
}
|
361 |
}
|