_speedups.c 7.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262
  1. #include <Python/Python.h>
  2. #include <Python/structmember.h>
  3. static unsigned long int
  4. checksum(const unsigned char *data, int len, unsigned long int sum)
  5. {
  6. unsigned long int s1, s2, i;
  7. s1 = sum & 0xffff;
  8. s2 = sum >> 16;
  9. for(i=0; i < len; i++)
  10. {
  11. s1 += data[i] + 1;
  12. s2 += s1;
  13. }
  14. return ((s2 & 0xffff) << 16) | (s1 & 0xffff);
  15. }
  16. static unsigned long int
  17. roll_checksum(unsigned long int sum, unsigned char remove, unsigned char add, int len)
  18. {
  19. unsigned long int s1, s2;
  20. s1 = sum & 0xffff;
  21. s2 = sum >> 16;
  22. s1 -= remove - add;
  23. s2 -= len * (remove + 1) - s1;
  24. return ((s2 & 0xffff) << 16) | (s1 & 0xffff);
  25. }
  26. typedef struct {
  27. PyObject_HEAD
  28. int chunk_size, i, full_sum, done, buf_size, data_len;
  29. PyObject *chunks, *fd, *extra;
  30. unsigned long sum;
  31. unsigned char *data, add, remove;
  32. } ChunkifyIter;
  33. static PyObject*
  34. ChunkifyIter_iter(PyObject *self)
  35. {
  36. Py_INCREF(self);
  37. return self;
  38. }
  39. static void
  40. ChunkifyIter_dealloc(PyObject *self)
  41. {
  42. ChunkifyIter *c = (ChunkifyIter *)self;
  43. Py_DECREF(c->fd);
  44. Py_DECREF(c->chunks);
  45. free(c->data);
  46. self->ob_type->tp_free(self);
  47. }
  48. static PyObject*
  49. ChunkifyIter_iternext(PyObject *self)
  50. {
  51. ChunkifyIter *c = (ChunkifyIter *)self;
  52. PyObject *pysum;
  53. int o = 0;
  54. if(c->done)
  55. {
  56. PyErr_SetNone(PyExc_StopIteration);
  57. return NULL;
  58. }
  59. if(c->extra)
  60. {
  61. c->done = 1;
  62. return c->extra;
  63. }
  64. for(;;)
  65. {
  66. if(c->i > c->buf_size - c->chunk_size)
  67. {
  68. memmove(c->data, c->data + c->i - o, c->data_len - c->i + o);
  69. c->data_len -= c->i - o;
  70. c->i = o;
  71. }
  72. if(c->data_len - c->i < c->chunk_size)
  73. {
  74. PyObject *data = PyObject_CallMethod(c->fd, "read", "i", c->buf_size - c->data_len);
  75. int n = PyString_Size(data);
  76. memcpy(c->data + c->data_len, PyString_AsString(data), n);
  77. c->data_len += n;
  78. Py_DECREF(data);
  79. }
  80. if(c->i == c->data_len)
  81. {
  82. PyErr_SetNone(PyExc_StopIteration);
  83. return NULL;
  84. }
  85. if(c->data_len - c->i < c->chunk_size) /* EOF ? */
  86. {
  87. if(o == 1)
  88. {
  89. c->done = 1;
  90. return PyString_FromStringAndSize((char *)(c->data + c->i - 1), c->data_len - c->i + 1);
  91. }
  92. else if(o > 1)
  93. {
  94. c->extra = PyString_FromStringAndSize((char *)(c->data + c->i - 1), c->chunk_size);
  95. return PyString_FromStringAndSize((char *)(c->data + c->i - o), o - 1);
  96. }
  97. else
  98. {
  99. c->done = 1;
  100. return PyString_FromStringAndSize((char *)(c->data + c->i), c->data_len - c->i);
  101. }
  102. }
  103. if(o == c->chunk_size)
  104. {
  105. return PyString_FromStringAndSize((char *)(c->data + c->i - c->chunk_size), c->chunk_size);
  106. }
  107. if(c->full_sum || c->i + c->chunk_size > c->data_len)
  108. {
  109. c->full_sum = 0;
  110. c->sum = checksum(c->data + c->i, c->chunk_size, 0);
  111. }
  112. else
  113. {
  114. c->sum = roll_checksum(c->sum, c->remove, c->data[c->i + c->chunk_size - 1], c->chunk_size);
  115. }
  116. c->remove = c->data[c->i];
  117. pysum = PyInt_FromLong(c->sum);
  118. if(PySequence_Contains(c->chunks, pysum) == 1)
  119. {
  120. Py_DECREF(pysum);
  121. c->full_sum = 1;
  122. if(o > 0)
  123. {
  124. return PyString_FromStringAndSize((char *)(c->data + c->i - o), o);
  125. }
  126. else
  127. {
  128. c->i += c->chunk_size;
  129. return PyString_FromStringAndSize((char *)(c->data + c->i - c->chunk_size), c->chunk_size);
  130. }
  131. }
  132. Py_DECREF(pysum);
  133. o++;
  134. c->i++;
  135. }
  136. PyErr_SetNone(PyExc_StopIteration);
  137. return NULL;
  138. }
  139. static PyTypeObject ChunkifyIterType = {
  140. PyObject_HEAD_INIT(NULL)
  141. 0, /*ob_size*/
  142. "_chunkifier._ChunkifyIter", /*tp_name*/
  143. sizeof(ChunkifyIter), /*tp_basicsize*/
  144. 0, /*tp_itemsize*/
  145. ChunkifyIter_dealloc, /*tp_dealloc*/
  146. 0, /*tp_print*/
  147. 0, /*tp_getattr*/
  148. 0, /*tp_setattr*/
  149. 0, /*tp_compare*/
  150. 0, /*tp_repr*/
  151. 0, /*tp_as_number*/
  152. 0, /*tp_as_sequence*/
  153. 0, /*tp_as_mapping*/
  154. 0, /*tp_hash */
  155. 0, /*tp_call*/
  156. 0, /*tp_str*/
  157. 0, /*tp_getattro*/
  158. 0, /*tp_setattro*/
  159. 0, /*tp_as_buffer*/
  160. Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_ITER,
  161. /* tp_flags: Py_TPFLAGS_HAVE_ITER tells python to
  162. use tp_iter and tp_iternext fields. */
  163. "", /* tp_doc */
  164. 0, /* tp_traverse */
  165. 0, /* tp_clear */
  166. 0, /* tp_richcompare */
  167. 0, /* tp_weaklistoffset */
  168. ChunkifyIter_iter, /* tp_iter: __iter__() method */
  169. ChunkifyIter_iternext /* tp_iternext: next() method */
  170. };
  171. static PyObject *
  172. chunkify(PyObject *self, PyObject *args)
  173. {
  174. PyObject *fd, *chunks;
  175. long int chunk_size;
  176. ChunkifyIter *c;
  177. if (!PyArg_ParseTuple(args, "OiO", &fd, &chunk_size, &chunks))
  178. {
  179. return NULL;
  180. }
  181. if (!(c = PyObject_New(ChunkifyIter, &ChunkifyIterType)))
  182. {
  183. return NULL;
  184. }
  185. PyObject_Init((PyObject *)c, &ChunkifyIterType);
  186. c->buf_size = chunk_size * 10;
  187. c->data = malloc(c->buf_size);
  188. c->data_len = 0;
  189. c->i = 0;
  190. c->full_sum = 1;
  191. c->done = 0;
  192. c->extra = NULL;
  193. c->fd = fd;
  194. c->chunk_size = chunk_size;
  195. c->chunks = chunks;
  196. Py_INCREF(fd);
  197. Py_INCREF(chunks);
  198. return (PyObject *)c;
  199. }
  200. static PyObject *
  201. py_checksum(PyObject *self, PyObject *args)
  202. {
  203. PyObject *data;
  204. unsigned long int sum = 0;
  205. if(!PyArg_ParseTuple(args, "O|k", &data, &sum)) return NULL;
  206. if(!PyString_Check(data))
  207. {
  208. PyErr_SetNone(PyExc_TypeError);
  209. return NULL;
  210. }
  211. return PyInt_FromLong(checksum((unsigned char *)PyString_AsString(data),
  212. PyString_Size(data), sum));
  213. }
  214. static PyObject *
  215. py_roll_checksum(PyObject *self, PyObject *args)
  216. {
  217. unsigned long int sum = 0, len, a, r;
  218. PyObject *add, *remove;
  219. if (!PyArg_ParseTuple(args, "kOOk", &sum, &remove, &add, &len)) return NULL;
  220. if(!PyString_Check(remove) || !PyString_Check(add) ||
  221. PyString_Size(remove) != 1 || PyString_Size(add) != 1)
  222. {
  223. PyErr_SetNone(PyExc_TypeError);
  224. return NULL;
  225. }
  226. a = *((const unsigned char *)PyString_AsString(add));
  227. r = *((const unsigned char *)PyString_AsString(remove));
  228. return PyInt_FromLong(roll_checksum(sum, r, a, len));
  229. }
  230. static PyMethodDef ChunkifierMethods[] = {
  231. {"chunkify", chunkify, METH_VARARGS, ""},
  232. {"checksum", py_checksum, METH_VARARGS, ""},
  233. {"roll_checksum", py_roll_checksum, METH_VARARGS, ""},
  234. {NULL, NULL, 0, NULL} /* Sentinel */
  235. };
  236. PyMODINIT_FUNC
  237. init_speedups(void)
  238. {
  239. PyObject* m;
  240. ChunkifyIterType.tp_new = PyType_GenericNew;
  241. if (PyType_Ready(&ChunkifyIterType) < 0) return;
  242. m = Py_InitModule("_speedups", ChunkifierMethods);
  243. Py_INCREF(&ChunkifyIterType);
  244. PyModule_AddObject(m, "_ChunkifyIter", (PyObject *)&ChunkifyIterType);
  245. }