While \file doesn't need an argument, it can't have another doxy command after it.
		
			
				
	
	
		
			669 lines
		
	
	
		
			18 KiB
		
	
	
	
		
			C
		
	
	
	
	
	
			
		
		
	
	
			669 lines
		
	
	
		
			18 KiB
		
	
	
	
		
			C
		
	
	
	
	
	
| /*
 | |
|  * This program is free software; you can redistribute it and/or
 | |
|  * modify it under the terms of the GNU General Public License
 | |
|  * as published by the Free Software Foundation; either version 2
 | |
|  * of the License, or (at your option) any later version.
 | |
|  *
 | |
|  * This program is distributed in the hope that it will be useful,
 | |
|  * but WITHOUT ANY WARRANTY; without even the implied warranty of
 | |
|  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 | |
|  * GNU General Public License for more details.
 | |
|  *
 | |
|  * You should have received a copy of the GNU General Public License
 | |
|  * along with this program; if not, write to the Free Software Foundation,
 | |
|  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
 | |
|  */
 | |
| 
 | |
| /** \file
 | |
|  * \ingroup pymathutils
 | |
|  */
 | |
| 
 | |
| #include <Python.h>
 | |
| 
 | |
| #include "mathutils.h"
 | |
| 
 | |
| #include "BLI_math.h"
 | |
| #include "BLI_utildefines.h"
 | |
| 
 | |
| #include "../generic/python_utildefines.h"
 | |
| 
 | |
| #ifndef MATH_STANDALONE
 | |
| #  include "BLI_dynstr.h"
 | |
| #endif
 | |
| 
 | |
| PyDoc_STRVAR(M_Mathutils_doc,
 | |
| "This module provides access to math operations.\n"
 | |
| "\n"
 | |
| ".. note::\n"
 | |
| "\n"
 | |
| "   Classes, methods and attributes that accept vectors also accept other numeric sequences,\n"
 | |
| "   such as tuples, lists."
 | |
| "\n\n"
 | |
| "Submodules:\n"
 | |
| "\n"
 | |
| ".. toctree::\n"
 | |
| "   :maxdepth: 1\n"
 | |
| "\n"
 | |
| "   mathutils.geometry.rst\n"
 | |
| "   mathutils.bvhtree.rst\n"
 | |
| "   mathutils.kdtree.rst\n"
 | |
| "   mathutils.interpolate.rst\n"
 | |
| "   mathutils.noise.rst\n"
 | |
| "\n"
 | |
| "The :mod:`mathutils` module provides the following classes:\n"
 | |
| "\n"
 | |
| "- :class:`Color`,\n"
 | |
| "- :class:`Euler`,\n"
 | |
| "- :class:`Matrix`,\n"
 | |
| "- :class:`Quaternion`,\n"
 | |
| "- :class:`Vector`,\n"
 | |
| );
 | |
| static int mathutils_array_parse_fast(float *array,
 | |
|                                       int size,
 | |
|                                       PyObject *value_fast,
 | |
|                                       const char *error_prefix)
 | |
| {
 | |
| 	PyObject *item;
 | |
| 	PyObject **value_fast_items = PySequence_Fast_ITEMS(value_fast);
 | |
| 
 | |
| 	int i;
 | |
| 
 | |
| 	i = size;
 | |
| 	do {
 | |
| 		i--;
 | |
| 		if (((array[i] = PyFloat_AsDouble((item = value_fast_items[i]))) == -1.0f) &&
 | |
| 		    PyErr_Occurred())
 | |
| 		{
 | |
| 			PyErr_Format(PyExc_TypeError,
 | |
| 			             "%.200s: sequence index %d expected a number, "
 | |
| 			             "found '%.200s' type, ",
 | |
| 			             error_prefix, i, Py_TYPE(item)->tp_name);
 | |
| 			size = -1;
 | |
| 			break;
 | |
| 		}
 | |
| 	} while (i);
 | |
| 
 | |
| 	return size;
 | |
| }
 | |
| 
 | |
| /**
 | |
|  * helper function that returns a Python ``__hash__``.
 | |
|  *
 | |
|  * \note consistent with the equivalent tuple of floats (CPython's 'tuplehash')
 | |
|  */
 | |
| Py_hash_t mathutils_array_hash(const float *array, size_t array_len)
 | |
| {
 | |
| 	int i;
 | |
| 	Py_uhash_t x;  /* Unsigned for defined overflow behavior. */
 | |
| 	Py_hash_t y;
 | |
| 	Py_uhash_t mult;
 | |
| 	Py_ssize_t len;
 | |
| 
 | |
| 	mult = _PyHASH_MULTIPLIER;
 | |
| 	len = array_len;
 | |
| 	x = 0x345678UL;
 | |
| 	i = 0;
 | |
| 	while (--len >= 0) {
 | |
| 		y = _Py_HashDouble((double)(array[i++]));
 | |
| 		if (y == -1)
 | |
| 			return -1;
 | |
| 		x = (x ^ y) * mult;
 | |
| 		/* the cast might truncate len; that doesn't change hash stability */
 | |
| 		mult += (Py_hash_t)(82520UL + len + len);
 | |
| 	}
 | |
| 	x += 97531UL;
 | |
| 	if (x == (Py_uhash_t)-1)
 | |
| 		x = -2;
 | |
| 	return x;
 | |
| }
 | |
| 
 | |
| /* helper function returns length of the 'value', -1 on error */
 | |
| int mathutils_array_parse(float *array, int array_min, int array_max, PyObject *value, const char *error_prefix)
 | |
| {
 | |
| 	const unsigned int flag = array_max;
 | |
| 	int size;
 | |
| 
 | |
| 	array_max &= ~MU_ARRAY_FLAGS;
 | |
| 
 | |
| #if 1 /* approx 6x speedup for mathutils types */
 | |
| 
 | |
| 	if ((size = VectorObject_Check(value)     ? ((VectorObject *)value)->size : 0) ||
 | |
| 	    (size = EulerObject_Check(value)      ? 3 : 0) ||
 | |
| 	    (size = QuaternionObject_Check(value) ? 4 : 0) ||
 | |
| 	    (size = ColorObject_Check(value)      ? 3 : 0))
 | |
| 	{
 | |
| 		if (BaseMath_ReadCallback((BaseMathObject *)value) == -1) {
 | |
| 			return -1;
 | |
| 		}
 | |
| 
 | |
| 		if (flag & MU_ARRAY_SPILL) {
 | |
| 			CLAMP_MAX(size, array_max);
 | |
| 		}
 | |
| 
 | |
| 		if (size > array_max || size < array_min) {
 | |
| 			if (array_max == array_min) {
 | |
| 				PyErr_Format(PyExc_ValueError,
 | |
| 				             "%.200s: sequence size is %d, expected %d",
 | |
| 				             error_prefix, size, array_max);
 | |
| 			}
 | |
| 			else {
 | |
| 				PyErr_Format(PyExc_ValueError,
 | |
| 				             "%.200s: sequence size is %d, expected [%d - %d]",
 | |
| 				             error_prefix, size, array_min, array_max);
 | |
| 			}
 | |
| 			return -1;
 | |
| 		}
 | |
| 
 | |
| 		memcpy(array, ((BaseMathObject *)value)->data, size * sizeof(float));
 | |
| 	}
 | |
| 	else
 | |
| #endif
 | |
| 	{
 | |
| 		PyObject *value_fast = NULL;
 | |
| 
 | |
| 		/* non list/tuple cases */
 | |
| 		if (!(value_fast = PySequence_Fast(value, error_prefix))) {
 | |
| 			/* PySequence_Fast sets the error */
 | |
| 			return -1;
 | |
| 		}
 | |
| 
 | |
| 		size = PySequence_Fast_GET_SIZE(value_fast);
 | |
| 
 | |
| 		if (flag & MU_ARRAY_SPILL) {
 | |
| 			CLAMP_MAX(size, array_max);
 | |
| 		}
 | |
| 
 | |
| 		if (size > array_max || size < array_min) {
 | |
| 			if (array_max == array_min) {
 | |
| 				PyErr_Format(PyExc_ValueError,
 | |
| 				             "%.200s: sequence size is %d, expected %d",
 | |
| 				             error_prefix, size, array_max);
 | |
| 			}
 | |
| 			else {
 | |
| 				PyErr_Format(PyExc_ValueError,
 | |
| 				             "%.200s: sequence size is %d, expected [%d - %d]",
 | |
| 				             error_prefix, size, array_min, array_max);
 | |
| 			}
 | |
| 			Py_DECREF(value_fast);
 | |
| 			return -1;
 | |
| 		}
 | |
| 
 | |
| 		size = mathutils_array_parse_fast(array, size, value_fast, error_prefix);
 | |
| 		Py_DECREF(value_fast);
 | |
| 	}
 | |
| 
 | |
| 	if (size != -1) {
 | |
| 		if (flag & MU_ARRAY_ZERO) {
 | |
| 			int size_left = array_max - size;
 | |
| 			if (size_left) {
 | |
| 				memset(&array[size], 0, sizeof(float) * size_left);
 | |
| 			}
 | |
| 		}
 | |
| 	}
 | |
| 
 | |
| 	return size;
 | |
| }
 | |
| 
 | |
| /* on error, -1 is returned and no allocation is made */
 | |
| int mathutils_array_parse_alloc(float **array, int array_min, PyObject *value, const char *error_prefix)
 | |
| {
 | |
| 	int size;
 | |
| 
 | |
| #if 1 /* approx 6x speedup for mathutils types */
 | |
| 
 | |
| 	if ((size = VectorObject_Check(value)     ? ((VectorObject *)value)->size : 0) ||
 | |
| 	    (size = EulerObject_Check(value)      ? 3 : 0) ||
 | |
| 	    (size = QuaternionObject_Check(value) ? 4 : 0) ||
 | |
| 	    (size = ColorObject_Check(value)      ? 3 : 0))
 | |
| 	{
 | |
| 		if (BaseMath_ReadCallback((BaseMathObject *)value) == -1) {
 | |
| 			return -1;
 | |
| 		}
 | |
| 
 | |
| 		if (size < array_min) {
 | |
| 			PyErr_Format(PyExc_ValueError,
 | |
| 			             "%.200s: sequence size is %d, expected > %d",
 | |
| 			             error_prefix, size, array_min);
 | |
| 			return -1;
 | |
| 		}
 | |
| 
 | |
| 		*array = PyMem_Malloc(size * sizeof(float));
 | |
| 		memcpy(*array, ((BaseMathObject *)value)->data, size * sizeof(float));
 | |
| 		return size;
 | |
| 	}
 | |
| 	else
 | |
| #endif
 | |
| 	{
 | |
| 		PyObject *value_fast = NULL;
 | |
| 		// *array = NULL;
 | |
| 		int ret;
 | |
| 
 | |
| 		/* non list/tuple cases */
 | |
| 		if (!(value_fast = PySequence_Fast(value, error_prefix))) {
 | |
| 			/* PySequence_Fast sets the error */
 | |
| 			return -1;
 | |
| 		}
 | |
| 
 | |
| 		size = PySequence_Fast_GET_SIZE(value_fast);
 | |
| 
 | |
| 		if (size < array_min) {
 | |
| 			Py_DECREF(value_fast);
 | |
| 			PyErr_Format(PyExc_ValueError,
 | |
| 			             "%.200s: sequence size is %d, expected > %d",
 | |
| 			             error_prefix, size, array_min);
 | |
| 			return -1;
 | |
| 		}
 | |
| 
 | |
| 		*array = PyMem_Malloc(size * sizeof(float));
 | |
| 
 | |
| 		ret = mathutils_array_parse_fast(*array, size, value_fast, error_prefix);
 | |
| 		Py_DECREF(value_fast);
 | |
| 
 | |
| 		if (ret == -1) {
 | |
| 			PyMem_Free(*array);
 | |
| 		}
 | |
| 
 | |
| 		return ret;
 | |
| 	}
 | |
| }
 | |
| 
 | |
| /* parse an array of vectors */
 | |
| int mathutils_array_parse_alloc_v(float **array, int array_dim, PyObject *value, const char *error_prefix)
 | |
| {
 | |
| 	PyObject *value_fast;
 | |
| 	const int array_dim_flag = array_dim;
 | |
| 	int i, size;
 | |
| 
 | |
| 	/* non list/tuple cases */
 | |
| 	if (!(value_fast = PySequence_Fast(value, error_prefix))) {
 | |
| 		/* PySequence_Fast sets the error */
 | |
| 		return -1;
 | |
| 	}
 | |
| 
 | |
| 	size = PySequence_Fast_GET_SIZE(value_fast);
 | |
| 
 | |
| 	if (size != 0) {
 | |
| 		PyObject **value_fast_items = PySequence_Fast_ITEMS(value_fast);
 | |
| 		float *fp;
 | |
| 
 | |
| 		array_dim &= ~MU_ARRAY_FLAGS;
 | |
| 
 | |
| 		fp = *array = PyMem_Malloc(size * array_dim * sizeof(float));
 | |
| 
 | |
| 		for (i = 0; i < size; i++, fp += array_dim) {
 | |
| 			PyObject *item = value_fast_items[i];
 | |
| 
 | |
| 			if (mathutils_array_parse(fp, array_dim, array_dim_flag, item, error_prefix) == -1) {
 | |
| 				PyMem_Free(*array);
 | |
| 				*array = NULL;
 | |
| 				size = -1;
 | |
| 				break;
 | |
| 			}
 | |
| 		}
 | |
| 	}
 | |
| 
 | |
| 	Py_DECREF(value_fast);
 | |
| 	return size;
 | |
| }
 | |
| 
 | |
| int mathutils_any_to_rotmat(float rmat[3][3], PyObject *value, const char *error_prefix)
 | |
| {
 | |
| 	if (EulerObject_Check(value)) {
 | |
| 		if (BaseMath_ReadCallback((BaseMathObject *)value) == -1) {
 | |
| 			return -1;
 | |
| 		}
 | |
| 		else {
 | |
| 			eulO_to_mat3(rmat, ((EulerObject *)value)->eul, ((EulerObject *)value)->order);
 | |
| 			return 0;
 | |
| 		}
 | |
| 	}
 | |
| 	else if (QuaternionObject_Check(value)) {
 | |
| 		if (BaseMath_ReadCallback((BaseMathObject *)value) == -1) {
 | |
| 			return -1;
 | |
| 		}
 | |
| 		else {
 | |
| 			float tquat[4];
 | |
| 			normalize_qt_qt(tquat, ((QuaternionObject *)value)->quat);
 | |
| 			quat_to_mat3(rmat, tquat);
 | |
| 			return 0;
 | |
| 		}
 | |
| 	}
 | |
| 	else if (MatrixObject_Check(value)) {
 | |
| 		if (BaseMath_ReadCallback((BaseMathObject *)value) == -1) {
 | |
| 			return -1;
 | |
| 		}
 | |
| 		else if (((MatrixObject *)value)->num_row < 3 || ((MatrixObject *)value)->num_col < 3) {
 | |
| 			PyErr_Format(PyExc_ValueError,
 | |
| 			             "%.200s: matrix must have minimum 3x3 dimensions",
 | |
| 			             error_prefix);
 | |
| 			return -1;
 | |
| 		}
 | |
| 		else {
 | |
| 			matrix_as_3x3(rmat, (MatrixObject *)value);
 | |
| 			normalize_m3(rmat);
 | |
| 			return 0;
 | |
| 		}
 | |
| 	}
 | |
| 	else {
 | |
| 		PyErr_Format(PyExc_TypeError,
 | |
| 		             "%.200s: expected a Euler, Quaternion or Matrix type, "
 | |
| 		             "found %.200s", error_prefix, Py_TYPE(value)->tp_name);
 | |
| 		return -1;
 | |
| 	}
 | |
| }
 | |
| 
 | |
| 
 | |
| /* ----------------------------------MATRIX FUNCTIONS-------------------- */
 | |
| 
 | |
| 
 | |
| /* Utility functions */
 | |
| 
 | |
| /* LomontRRDCompare4, Ever Faster Float Comparisons by Randy Dillon */
 | |
| /* XXX We may want to use 'safer' BLI's compare_ff_relative ultimately?
 | |
|  *     LomontRRDCompare4() is an optimized version of Dawson's AlmostEqual2sComplement() (see [1] and [2]).
 | |
|  *     Dawson himself now claims this is not a 'safe' thing to do (pushing ULP method beyond its limits),
 | |
|  *     an recommends using work from [3] instead, which is done in BLI func...
 | |
|  *
 | |
|  *     [1] http://www.randydillon.org/Papers/2007/everfast.htm
 | |
|  *     [2] http://www.cygnus-software.com/papers/comparingfloats/comparingfloats.htm
 | |
|  *     [3] https://randomascii.wordpress.com/2012/02/25/comparing-floating-point-numbers-2012-edition/ instead
 | |
|  */
 | |
| #define SIGNMASK(i) (-(int)(((unsigned int)(i)) >> 31))
 | |
| 
 | |
| int EXPP_FloatsAreEqual(float af, float bf, int maxDiff)
 | |
| {
 | |
| 	/* solid, fast routine across all platforms
 | |
| 	 * with constant time behavior */
 | |
| 	int ai = *(int *)(&af);
 | |
| 	int bi = *(int *)(&bf);
 | |
| 	int test = SIGNMASK(ai ^ bi);
 | |
| 	int diff, v1, v2;
 | |
| 
 | |
| 	assert((0 == test) || (0xFFFFFFFF == test));
 | |
| 	diff = (ai ^ (test & 0x7fffffff)) - bi;
 | |
| 	v1 = maxDiff + diff;
 | |
| 	v2 = maxDiff - diff;
 | |
| 	return (v1 | v2) >= 0;
 | |
| }
 | |
| 
 | |
| /*---------------------- EXPP_VectorsAreEqual -------------------------
 | |
|  * Builds on EXPP_FloatsAreEqual to test vectors */
 | |
| int EXPP_VectorsAreEqual(const float *vecA, const float *vecB, int size, int floatSteps)
 | |
| {
 | |
| 	int x;
 | |
| 	for (x = 0; x < size; x++) {
 | |
| 		if (EXPP_FloatsAreEqual(vecA[x], vecB[x], floatSteps) == 0)
 | |
| 			return 0;
 | |
| 	}
 | |
| 	return 1;
 | |
| }
 | |
| 
 | |
| #ifndef MATH_STANDALONE
 | |
| /* dynstr as python string utility functions, frees 'ds'! */
 | |
| PyObject *mathutils_dynstr_to_py(struct DynStr *ds)
 | |
| {
 | |
| 	const int ds_len = BLI_dynstr_get_len(ds); /* space for \0 */
 | |
| 	char *ds_buf     = PyMem_Malloc(ds_len + 1);
 | |
| 	PyObject *ret;
 | |
| 	BLI_dynstr_get_cstring_ex(ds, ds_buf);
 | |
| 	BLI_dynstr_free(ds);
 | |
| 	ret = PyUnicode_FromStringAndSize(ds_buf, ds_len);
 | |
| 	PyMem_Free(ds_buf);
 | |
| 	return ret;
 | |
| }
 | |
| #endif
 | |
| 
 | |
| /* Mathutils Callbacks */
 | |
| 
 | |
| /* for mathutils internal use only, eventually should re-alloc but to start with we only have a few users */
 | |
| #define MATHUTILS_TOT_CB 17
 | |
| static Mathutils_Callback *mathutils_callbacks[MATHUTILS_TOT_CB] = {NULL};
 | |
| 
 | |
| unsigned char Mathutils_RegisterCallback(Mathutils_Callback *cb)
 | |
| {
 | |
| 	unsigned char i;
 | |
| 
 | |
| 	/* find the first free slot */
 | |
| 	for (i = 0; mathutils_callbacks[i]; i++) {
 | |
| 		if (mathutils_callbacks[i] == cb) /* already registered? */
 | |
| 			return i;
 | |
| 	}
 | |
| 
 | |
| 	BLI_assert(i + 1 < MATHUTILS_TOT_CB);
 | |
| 
 | |
| 	mathutils_callbacks[i] = cb;
 | |
| 	return i;
 | |
| }
 | |
| 
 | |
| /* use macros to check for NULL */
 | |
| int _BaseMathObject_ReadCallback(BaseMathObject *self)
 | |
| {
 | |
| 	Mathutils_Callback *cb = mathutils_callbacks[self->cb_type];
 | |
| 	if (LIKELY(cb->get(self, self->cb_subtype) != -1)) {
 | |
| 		return 0;
 | |
| 	}
 | |
| 
 | |
| 	if (!PyErr_Occurred()) {
 | |
| 		PyErr_Format(PyExc_RuntimeError,
 | |
| 		             "%s read, user has become invalid",
 | |
| 		             Py_TYPE(self)->tp_name);
 | |
| 	}
 | |
| 	return -1;
 | |
| }
 | |
| 
 | |
| int _BaseMathObject_WriteCallback(BaseMathObject *self)
 | |
| {
 | |
| 	Mathutils_Callback *cb = mathutils_callbacks[self->cb_type];
 | |
| 	if (LIKELY(cb->set(self, self->cb_subtype) != -1)) {
 | |
| 		return 0;
 | |
| 	}
 | |
| 
 | |
| 	if (!PyErr_Occurred()) {
 | |
| 		PyErr_Format(PyExc_RuntimeError,
 | |
| 		             "%s write, user has become invalid",
 | |
| 		             Py_TYPE(self)->tp_name);
 | |
| 	}
 | |
| 	return -1;
 | |
| }
 | |
| 
 | |
| int _BaseMathObject_ReadIndexCallback(BaseMathObject *self, int index)
 | |
| {
 | |
| 	Mathutils_Callback *cb = mathutils_callbacks[self->cb_type];
 | |
| 	if (LIKELY(cb->get_index(self, self->cb_subtype, index) != -1)) {
 | |
| 		return 0;
 | |
| 	}
 | |
| 
 | |
| 	if (!PyErr_Occurred()) {
 | |
| 		PyErr_Format(PyExc_RuntimeError,
 | |
| 		             "%s read index, user has become invalid",
 | |
| 		             Py_TYPE(self)->tp_name);
 | |
| 	}
 | |
| 	return -1;
 | |
| }
 | |
| 
 | |
| int _BaseMathObject_WriteIndexCallback(BaseMathObject *self, int index)
 | |
| {
 | |
| 	Mathutils_Callback *cb = mathutils_callbacks[self->cb_type];
 | |
| 	if (LIKELY(cb->set_index(self, self->cb_subtype, index) != -1)) {
 | |
| 		return 0;
 | |
| 	}
 | |
| 
 | |
| 	if (!PyErr_Occurred()) {
 | |
| 		PyErr_Format(PyExc_RuntimeError,
 | |
| 		             "%s write index, user has become invalid",
 | |
| 		             Py_TYPE(self)->tp_name);
 | |
| 	}
 | |
| 	return -1;
 | |
| }
 | |
| 
 | |
| void _BaseMathObject_RaiseFrozenExc(const BaseMathObject *self)
 | |
| {
 | |
| 	PyErr_Format(PyExc_TypeError,
 | |
| 	             "%s is frozen (immutable)",
 | |
| 	             Py_TYPE(self)->tp_name);
 | |
| }
 | |
| 
 | |
| void _BaseMathObject_RaiseNotFrozenExc(const BaseMathObject *self)
 | |
| {
 | |
| 	PyErr_Format(PyExc_TypeError,
 | |
| 	             "%s is not frozen (mutable), call freeze first",
 | |
| 	             Py_TYPE(self)->tp_name);
 | |
| }
 | |
| 
 | |
| /* BaseMathObject generic functions for all mathutils types */
 | |
| char BaseMathObject_owner_doc[] = "The item this is wrapping or None  (read-only).";
 | |
| PyObject *BaseMathObject_owner_get(BaseMathObject *self, void *UNUSED(closure))
 | |
| {
 | |
| 	PyObject *ret = self->cb_user ? self->cb_user : Py_None;
 | |
| 	return Py_INCREF_RET(ret);
 | |
| }
 | |
| 
 | |
| char BaseMathObject_is_wrapped_doc[] = "True when this object wraps external data (read-only).\n\n:type: boolean";
 | |
| PyObject *BaseMathObject_is_wrapped_get(BaseMathObject *self, void *UNUSED(closure))
 | |
| {
 | |
| 	return PyBool_FromLong((self->flag & BASE_MATH_FLAG_IS_WRAP) != 0);
 | |
| }
 | |
| 
 | |
| char BaseMathObject_is_frozen_doc[] = "True when this object has been frozen (read-only).\n\n:type: boolean";
 | |
| PyObject *BaseMathObject_is_frozen_get(BaseMathObject *self, void *UNUSED(closure))
 | |
| {
 | |
| 	return PyBool_FromLong((self->flag & BASE_MATH_FLAG_IS_FROZEN) != 0);
 | |
| }
 | |
| 
 | |
| char BaseMathObject_freeze_doc[] =
 | |
| ".. function:: freeze()\n"
 | |
| "\n"
 | |
| "   Make this object immutable.\n"
 | |
| "\n"
 | |
| "   After this the object can be hashed, used in dictionaries & sets.\n"
 | |
| "\n"
 | |
| "   :return: An instance of this object.\n"
 | |
| ;
 | |
| PyObject *BaseMathObject_freeze(BaseMathObject *self)
 | |
| {
 | |
| 	if ((self->flag & BASE_MATH_FLAG_IS_WRAP) || (self->cb_user != NULL)) {
 | |
| 		PyErr_SetString(PyExc_TypeError, "Cannot freeze wrapped/owned data");
 | |
| 		return NULL;
 | |
| 	}
 | |
| 
 | |
| 	self->flag |= BASE_MATH_FLAG_IS_FROZEN;
 | |
| 
 | |
| 	return Py_INCREF_RET((PyObject *)self);
 | |
| }
 | |
| 
 | |
| int BaseMathObject_traverse(BaseMathObject *self, visitproc visit, void *arg)
 | |
| {
 | |
| 	Py_VISIT(self->cb_user);
 | |
| 	return 0;
 | |
| }
 | |
| 
 | |
| int BaseMathObject_clear(BaseMathObject *self)
 | |
| {
 | |
| 	Py_CLEAR(self->cb_user);
 | |
| 	return 0;
 | |
| }
 | |
| 
 | |
| void BaseMathObject_dealloc(BaseMathObject *self)
 | |
| {
 | |
| 	/* only free non wrapped */
 | |
| 	if ((self->flag & BASE_MATH_FLAG_IS_WRAP) == 0) {
 | |
| 		PyMem_Free(self->data);
 | |
| 	}
 | |
| 
 | |
| 	if (self->cb_user) {
 | |
| 		PyObject_GC_UnTrack(self);
 | |
| 		BaseMathObject_clear(self);
 | |
| 	}
 | |
| 
 | |
| 	Py_TYPE(self)->tp_free(self); // PyObject_DEL(self); // breaks subtypes
 | |
| }
 | |
| 
 | |
| /*----------------------------MODULE INIT-------------------------*/
 | |
| static struct PyMethodDef M_Mathutils_methods[] = {
 | |
| 	{NULL, NULL, 0, NULL},
 | |
| };
 | |
| 
 | |
| static struct PyModuleDef M_Mathutils_module_def = {
 | |
| 	PyModuleDef_HEAD_INIT,
 | |
| 	"mathutils",  /* m_name */
 | |
| 	M_Mathutils_doc,  /* m_doc */
 | |
| 	0,  /* m_size */
 | |
| 	M_Mathutils_methods,  /* m_methods */
 | |
| 	NULL,  /* m_reload */
 | |
| 	NULL,  /* m_traverse */
 | |
| 	NULL,  /* m_clear */
 | |
| 	NULL,  /* m_free */
 | |
| };
 | |
| 
 | |
| 
 | |
| /* submodules only */
 | |
| #include "mathutils_geometry.h"
 | |
| #include "mathutils_interpolate.h"
 | |
| #ifndef MATH_STANDALONE
 | |
| #  include "mathutils_bvhtree.h"
 | |
| #  include "mathutils_kdtree.h"
 | |
| #  include "mathutils_noise.h"
 | |
| #endif
 | |
| 
 | |
| PyMODINIT_FUNC PyInit_mathutils(void)
 | |
| {
 | |
| 	PyObject *mod;
 | |
| 	PyObject *submodule;
 | |
| 	PyObject *sys_modules = PyImport_GetModuleDict();
 | |
| 
 | |
| 	if (PyType_Ready(&vector_Type) < 0)
 | |
| 		return NULL;
 | |
| 	if (PyType_Ready(&matrix_Type) < 0)
 | |
| 		return NULL;
 | |
| 	if (PyType_Ready(&matrix_access_Type) < 0)
 | |
| 		return NULL;
 | |
| 	if (PyType_Ready(&euler_Type) < 0)
 | |
| 		return NULL;
 | |
| 	if (PyType_Ready(&quaternion_Type) < 0)
 | |
| 		return NULL;
 | |
| 	if (PyType_Ready(&color_Type) < 0)
 | |
| 		return NULL;
 | |
| 
 | |
| 	mod = PyModule_Create(&M_Mathutils_module_def);
 | |
| 
 | |
| 	/* each type has its own new() function */
 | |
| 	PyModule_AddObject(mod, vector_Type.tp_name,     (PyObject *)&vector_Type);
 | |
| 	PyModule_AddObject(mod, matrix_Type.tp_name,     (PyObject *)&matrix_Type);
 | |
| 	PyModule_AddObject(mod, euler_Type.tp_name,      (PyObject *)&euler_Type);
 | |
| 	PyModule_AddObject(mod, quaternion_Type.tp_name, (PyObject *)&quaternion_Type);
 | |
| 	PyModule_AddObject(mod, color_Type.tp_name,      (PyObject *)&color_Type);
 | |
| 
 | |
| 	/* submodule */
 | |
| 	PyModule_AddObject(mod, "geometry",       (submodule = PyInit_mathutils_geometry()));
 | |
| 	/* XXX, python doesn't do imports with this usefully yet
 | |
| 	 * 'from mathutils.geometry import PolyFill'
 | |
| 	 * ...fails without this. */
 | |
| 	PyDict_SetItem(sys_modules, PyModule_GetNameObject(submodule), submodule);
 | |
| 
 | |
| 	PyModule_AddObject(mod, "interpolate",    (submodule = PyInit_mathutils_interpolate()));
 | |
| 	/* XXX, python doesnt do imports with this usefully yet
 | |
| 	 * 'from mathutils.geometry import PolyFill'
 | |
| 	 * ...fails without this. */
 | |
| 	PyDict_SetItem(sys_modules, PyModule_GetNameObject(submodule), submodule);
 | |
| 
 | |
| #ifndef MATH_STANDALONE
 | |
| 	/* Noise submodule */
 | |
| 	PyModule_AddObject(mod, "noise", (submodule = PyInit_mathutils_noise()));
 | |
| 	PyDict_SetItem(sys_modules, PyModule_GetNameObject(submodule), submodule);
 | |
| 
 | |
| 	/* BVHTree submodule */
 | |
| 	PyModule_AddObject(mod, "bvhtree", (submodule = PyInit_mathutils_bvhtree()));
 | |
| 	PyDict_SetItem(sys_modules, PyModule_GetNameObject(submodule), submodule);
 | |
| 
 | |
| 	/* KDTree submodule */
 | |
| 	PyModule_AddObject(mod, "kdtree", (submodule = PyInit_mathutils_kdtree()));
 | |
| 	PyDict_SetItem(sys_modules, PyModule_GetNameObject(submodule), submodule);
 | |
| #endif
 | |
| 
 | |
| 	mathutils_matrix_row_cb_index = Mathutils_RegisterCallback(&mathutils_matrix_row_cb);
 | |
| 	mathutils_matrix_col_cb_index = Mathutils_RegisterCallback(&mathutils_matrix_col_cb);
 | |
| 	mathutils_matrix_translation_cb_index = Mathutils_RegisterCallback(&mathutils_matrix_translation_cb);
 | |
| 
 | |
| 	return mod;
 | |
| }
 |