Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 1 | /* _bz2 - Low-level Python interface to libbzip2. */ |
| 2 | |
| 3 | #define PY_SSIZE_T_CLEAN |
| 4 | |
| 5 | #include "Python.h" |
| 6 | #include "structmember.h" |
| 7 | |
| 8 | #ifdef WITH_THREAD |
| 9 | #include "pythread.h" |
| 10 | #endif |
| 11 | |
| 12 | #include <bzlib.h> |
| 13 | #include <stdio.h> |
| 14 | |
| 15 | |
| 16 | #ifndef BZ_CONFIG_ERROR |
| 17 | #define BZ2_bzCompress bzCompress |
| 18 | #define BZ2_bzCompressInit bzCompressInit |
| 19 | #define BZ2_bzCompressEnd bzCompressEnd |
| 20 | #define BZ2_bzDecompress bzDecompress |
| 21 | #define BZ2_bzDecompressInit bzDecompressInit |
| 22 | #define BZ2_bzDecompressEnd bzDecompressEnd |
| 23 | #endif /* ! BZ_CONFIG_ERROR */ |
| 24 | |
| 25 | |
| 26 | #ifdef WITH_THREAD |
| 27 | #define ACQUIRE_LOCK(obj) do { \ |
| 28 | if (!PyThread_acquire_lock((obj)->lock, 0)) { \ |
| 29 | Py_BEGIN_ALLOW_THREADS \ |
| 30 | PyThread_acquire_lock((obj)->lock, 1); \ |
| 31 | Py_END_ALLOW_THREADS \ |
| 32 | } } while (0) |
| 33 | #define RELEASE_LOCK(obj) PyThread_release_lock((obj)->lock) |
| 34 | #else |
| 35 | #define ACQUIRE_LOCK(obj) |
| 36 | #define RELEASE_LOCK(obj) |
| 37 | #endif |
| 38 | |
| 39 | |
| 40 | typedef struct { |
| 41 | PyObject_HEAD |
| 42 | bz_stream bzs; |
| 43 | int flushed; |
| 44 | #ifdef WITH_THREAD |
| 45 | PyThread_type_lock lock; |
| 46 | #endif |
| 47 | } BZ2Compressor; |
| 48 | |
| 49 | typedef struct { |
| 50 | PyObject_HEAD |
| 51 | bz_stream bzs; |
| 52 | char eof; /* T_BOOL expects a char */ |
| 53 | PyObject *unused_data; |
| 54 | #ifdef WITH_THREAD |
| 55 | PyThread_type_lock lock; |
| 56 | #endif |
| 57 | } BZ2Decompressor; |
| 58 | |
Larry Hastings | f256c22 | 2014-01-25 21:30:37 -0800 | [diff] [blame] | 59 | static PyTypeObject BZ2Compressor_Type; |
| 60 | static PyTypeObject BZ2Decompressor_Type; |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 61 | |
| 62 | /* Helper functions. */ |
| 63 | |
| 64 | static int |
| 65 | catch_bz2_error(int bzerror) |
| 66 | { |
| 67 | switch(bzerror) { |
| 68 | case BZ_OK: |
| 69 | case BZ_RUN_OK: |
| 70 | case BZ_FLUSH_OK: |
| 71 | case BZ_FINISH_OK: |
| 72 | case BZ_STREAM_END: |
| 73 | return 0; |
| 74 | |
| 75 | #ifdef BZ_CONFIG_ERROR |
| 76 | case BZ_CONFIG_ERROR: |
| 77 | PyErr_SetString(PyExc_SystemError, |
| 78 | "libbzip2 was not compiled correctly"); |
| 79 | return 1; |
| 80 | #endif |
| 81 | case BZ_PARAM_ERROR: |
| 82 | PyErr_SetString(PyExc_ValueError, |
| 83 | "Internal error - " |
| 84 | "invalid parameters passed to libbzip2"); |
| 85 | return 1; |
| 86 | case BZ_MEM_ERROR: |
| 87 | PyErr_NoMemory(); |
| 88 | return 1; |
| 89 | case BZ_DATA_ERROR: |
| 90 | case BZ_DATA_ERROR_MAGIC: |
| 91 | PyErr_SetString(PyExc_IOError, "Invalid data stream"); |
| 92 | return 1; |
| 93 | case BZ_IO_ERROR: |
| 94 | PyErr_SetString(PyExc_IOError, "Unknown I/O error"); |
| 95 | return 1; |
| 96 | case BZ_UNEXPECTED_EOF: |
| 97 | PyErr_SetString(PyExc_EOFError, |
| 98 | "Compressed file ended before the logical " |
| 99 | "end-of-stream was detected"); |
| 100 | return 1; |
| 101 | case BZ_SEQUENCE_ERROR: |
| 102 | PyErr_SetString(PyExc_RuntimeError, |
| 103 | "Internal error - " |
| 104 | "Invalid sequence of commands sent to libbzip2"); |
| 105 | return 1; |
| 106 | default: |
| 107 | PyErr_Format(PyExc_IOError, |
| 108 | "Unrecognized error from libbzip2: %d", bzerror); |
| 109 | return 1; |
| 110 | } |
| 111 | } |
| 112 | |
| 113 | #if BUFSIZ < 8192 |
| 114 | #define SMALLCHUNK 8192 |
| 115 | #else |
| 116 | #define SMALLCHUNK BUFSIZ |
| 117 | #endif |
| 118 | |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 119 | static int |
| 120 | grow_buffer(PyObject **buf) |
| 121 | { |
Nadeem Vawda | 72d6a13 | 2011-10-13 13:38:14 +0200 | [diff] [blame] | 122 | /* Expand the buffer by an amount proportional to the current size, |
| 123 | giving us amortized linear-time behavior. Use a less-than-double |
| 124 | growth factor to avoid excessive allocation. */ |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 125 | size_t size = PyBytes_GET_SIZE(*buf); |
Nadeem Vawda | 18b7fcc | 2012-10-21 21:16:58 +0200 | [diff] [blame] | 126 | size_t new_size = size + (size >> 3) + 6; |
| 127 | if (new_size > size) { |
| 128 | return _PyBytes_Resize(buf, new_size); |
| 129 | } else { /* overflow */ |
| 130 | PyErr_SetString(PyExc_OverflowError, |
| 131 | "Unable to allocate buffer - output too large"); |
| 132 | return -1; |
| 133 | } |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 134 | } |
| 135 | |
| 136 | |
| 137 | /* BZ2Compressor class. */ |
| 138 | |
| 139 | static PyObject * |
| 140 | compress(BZ2Compressor *c, char *data, size_t len, int action) |
| 141 | { |
| 142 | size_t data_size = 0; |
| 143 | PyObject *result; |
| 144 | |
| 145 | result = PyBytes_FromStringAndSize(NULL, SMALLCHUNK); |
| 146 | if (result == NULL) |
| 147 | return NULL; |
Nadeem Vawda | 57cb81d | 2013-01-02 23:05:56 +0100 | [diff] [blame] | 148 | |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 149 | c->bzs.next_in = data; |
Nadeem Vawda | 57cb81d | 2013-01-02 23:05:56 +0100 | [diff] [blame] | 150 | c->bzs.avail_in = 0; |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 151 | c->bzs.next_out = PyBytes_AS_STRING(result); |
Victor Stinner | fbf50d4 | 2013-06-04 23:18:48 +0200 | [diff] [blame] | 152 | c->bzs.avail_out = SMALLCHUNK; |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 153 | for (;;) { |
| 154 | char *this_out; |
| 155 | int bzerror; |
| 156 | |
Nadeem Vawda | 57cb81d | 2013-01-02 23:05:56 +0100 | [diff] [blame] | 157 | /* On a 64-bit system, len might not fit in avail_in (an unsigned int). |
| 158 | Do compression in chunks of no more than UINT_MAX bytes each. */ |
Nadeem Vawda | ea4b46f | 2011-04-12 23:02:42 +0200 | [diff] [blame] | 159 | if (c->bzs.avail_in == 0 && len > 0) { |
Victor Stinner | fbf50d4 | 2013-06-04 23:18:48 +0200 | [diff] [blame] | 160 | c->bzs.avail_in = (unsigned int)Py_MIN(len, UINT_MAX); |
Nadeem Vawda | ea4b46f | 2011-04-12 23:02:42 +0200 | [diff] [blame] | 161 | len -= c->bzs.avail_in; |
| 162 | } |
| 163 | |
Nadeem Vawda | 57cb81d | 2013-01-02 23:05:56 +0100 | [diff] [blame] | 164 | /* In regular compression mode, stop when input data is exhausted. */ |
| 165 | if (action == BZ_RUN && c->bzs.avail_in == 0) |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 166 | break; |
| 167 | |
| 168 | if (c->bzs.avail_out == 0) { |
Nadeem Vawda | 18b7fcc | 2012-10-21 21:16:58 +0200 | [diff] [blame] | 169 | size_t buffer_left = PyBytes_GET_SIZE(result) - data_size; |
| 170 | if (buffer_left == 0) { |
| 171 | if (grow_buffer(&result) < 0) |
| 172 | goto error; |
| 173 | c->bzs.next_out = PyBytes_AS_STRING(result) + data_size; |
| 174 | buffer_left = PyBytes_GET_SIZE(result) - data_size; |
| 175 | } |
Victor Stinner | fbf50d4 | 2013-06-04 23:18:48 +0200 | [diff] [blame] | 176 | c->bzs.avail_out = (unsigned int)Py_MIN(buffer_left, UINT_MAX); |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 177 | } |
Nadeem Vawda | 57cb81d | 2013-01-02 23:05:56 +0100 | [diff] [blame] | 178 | |
| 179 | Py_BEGIN_ALLOW_THREADS |
| 180 | this_out = c->bzs.next_out; |
| 181 | bzerror = BZ2_bzCompress(&c->bzs, action); |
| 182 | data_size += c->bzs.next_out - this_out; |
| 183 | Py_END_ALLOW_THREADS |
| 184 | if (catch_bz2_error(bzerror)) |
| 185 | goto error; |
| 186 | |
| 187 | /* In flushing mode, stop when all buffered data has been flushed. */ |
| 188 | if (action == BZ_FINISH && bzerror == BZ_STREAM_END) |
| 189 | break; |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 190 | } |
| 191 | if (data_size != PyBytes_GET_SIZE(result)) |
| 192 | if (_PyBytes_Resize(&result, data_size) < 0) |
| 193 | goto error; |
| 194 | return result; |
| 195 | |
| 196 | error: |
| 197 | Py_XDECREF(result); |
| 198 | return NULL; |
| 199 | } |
| 200 | |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 201 | /*[clinic input] |
| 202 | output preset file |
| 203 | module _bz2 |
Larry Hastings | c204726 | 2014-01-25 20:43:29 -0800 | [diff] [blame] | 204 | class _bz2.BZ2Compressor "BZ2Compressor *" "&BZ2Compressor_Type" |
| 205 | class _bz2.BZ2Decompressor "BZ2Decompressor *" "&BZ2Decompressor_Type" |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 206 | [clinic start generated code]*/ |
Larry Hastings | 581ee36 | 2014-01-28 05:00:08 -0800 | [diff] [blame] | 207 | /*[clinic end generated code: output=da39a3ee5e6b4b0d input=e3b139924f5e18cc]*/ |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 208 | |
Larry Hastings | f256c22 | 2014-01-25 21:30:37 -0800 | [diff] [blame] | 209 | #include "clinic/_bz2module.c.h" |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 210 | |
| 211 | /*[clinic input] |
| 212 | _bz2.BZ2Compressor.compress |
| 213 | |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 214 | data: Py_buffer |
| 215 | / |
| 216 | |
| 217 | Provide data to the compressor object. |
| 218 | |
| 219 | Returns a chunk of compressed data if possible, or b'' otherwise. |
| 220 | |
| 221 | When you have finished providing data to the compressor, call the |
| 222 | flush() method to finish the compression process. |
| 223 | [clinic start generated code]*/ |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 224 | |
| 225 | static PyObject * |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 226 | _bz2_BZ2Compressor_compress_impl(BZ2Compressor *self, Py_buffer *data) |
Larry Hastings | 581ee36 | 2014-01-28 05:00:08 -0800 | [diff] [blame] | 227 | /*[clinic end generated code: output=59365426e941fbcc input=85c963218070fc4c]*/ |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 228 | { |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 229 | PyObject *result = NULL; |
| 230 | |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 231 | ACQUIRE_LOCK(self); |
| 232 | if (self->flushed) |
| 233 | PyErr_SetString(PyExc_ValueError, "Compressor has been flushed"); |
| 234 | else |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 235 | result = compress(self, data->buf, data->len, BZ_RUN); |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 236 | RELEASE_LOCK(self); |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 237 | return result; |
| 238 | } |
| 239 | |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 240 | /*[clinic input] |
| 241 | _bz2.BZ2Compressor.flush |
| 242 | |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 243 | Finish the compression process. |
| 244 | |
| 245 | Returns the compressed data left in internal buffers. |
| 246 | |
| 247 | The compressor object may not be used after this method is called. |
| 248 | [clinic start generated code]*/ |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 249 | |
| 250 | static PyObject * |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 251 | _bz2_BZ2Compressor_flush_impl(BZ2Compressor *self) |
Larry Hastings | 581ee36 | 2014-01-28 05:00:08 -0800 | [diff] [blame] | 252 | /*[clinic end generated code: output=3ef03fc1b092a701 input=d64405d3c6f76691]*/ |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 253 | { |
| 254 | PyObject *result = NULL; |
| 255 | |
| 256 | ACQUIRE_LOCK(self); |
| 257 | if (self->flushed) |
| 258 | PyErr_SetString(PyExc_ValueError, "Repeated call to flush()"); |
| 259 | else { |
| 260 | self->flushed = 1; |
| 261 | result = compress(self, NULL, 0, BZ_FINISH); |
| 262 | } |
| 263 | RELEASE_LOCK(self); |
| 264 | return result; |
| 265 | } |
| 266 | |
Nadeem Vawda | 3797065 | 2013-10-28 21:35:23 +0100 | [diff] [blame] | 267 | static PyObject * |
| 268 | BZ2Compressor_getstate(BZ2Compressor *self, PyObject *noargs) |
| 269 | { |
| 270 | PyErr_Format(PyExc_TypeError, "cannot serialize '%s' object", |
| 271 | Py_TYPE(self)->tp_name); |
| 272 | return NULL; |
| 273 | } |
| 274 | |
Victor Stinner | 5064a52 | 2013-07-07 16:50:27 +0200 | [diff] [blame] | 275 | static void* |
| 276 | BZ2_Malloc(void* ctx, int items, int size) |
| 277 | { |
| 278 | if (items < 0 || size < 0) |
| 279 | return NULL; |
| 280 | if ((size_t)items > (size_t)PY_SSIZE_T_MAX / (size_t)size) |
| 281 | return NULL; |
| 282 | /* PyMem_Malloc() cannot be used: compress() and decompress() |
| 283 | release the GIL */ |
| 284 | return PyMem_RawMalloc(items * size); |
| 285 | } |
| 286 | |
| 287 | static void |
| 288 | BZ2_Free(void* ctx, void *ptr) |
| 289 | { |
Victor Stinner | b7f1f65 | 2013-07-07 17:10:34 +0200 | [diff] [blame] | 290 | PyMem_RawFree(ptr); |
Victor Stinner | 5064a52 | 2013-07-07 16:50:27 +0200 | [diff] [blame] | 291 | } |
| 292 | |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 293 | /*[clinic input] |
| 294 | _bz2.BZ2Compressor.__init__ |
| 295 | |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 296 | compresslevel: int = 9 |
| 297 | Compression level, as a number between 1 and 9. |
| 298 | / |
| 299 | |
| 300 | Create a compressor object for compressing data incrementally. |
| 301 | |
| 302 | For one-shot compression, use the compress() function instead. |
| 303 | [clinic start generated code]*/ |
| 304 | |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 305 | static int |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 306 | _bz2_BZ2Compressor___init___impl(BZ2Compressor *self, int compresslevel) |
Larry Hastings | 581ee36 | 2014-01-28 05:00:08 -0800 | [diff] [blame] | 307 | /*[clinic end generated code: output=c4e6adfd02963827 input=4e1ff7b8394b6e9a]*/ |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 308 | { |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 309 | int bzerror; |
| 310 | |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 311 | if (!(1 <= compresslevel && compresslevel <= 9)) { |
| 312 | PyErr_SetString(PyExc_ValueError, |
| 313 | "compresslevel must be between 1 and 9"); |
| 314 | return -1; |
| 315 | } |
| 316 | |
| 317 | #ifdef WITH_THREAD |
| 318 | self->lock = PyThread_allocate_lock(); |
| 319 | if (self->lock == NULL) { |
| 320 | PyErr_SetString(PyExc_MemoryError, "Unable to allocate lock"); |
| 321 | return -1; |
| 322 | } |
| 323 | #endif |
| 324 | |
Victor Stinner | 5064a52 | 2013-07-07 16:50:27 +0200 | [diff] [blame] | 325 | self->bzs.opaque = NULL; |
| 326 | self->bzs.bzalloc = BZ2_Malloc; |
| 327 | self->bzs.bzfree = BZ2_Free; |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 328 | bzerror = BZ2_bzCompressInit(&self->bzs, compresslevel, 0, 0); |
| 329 | if (catch_bz2_error(bzerror)) |
| 330 | goto error; |
| 331 | |
| 332 | return 0; |
| 333 | |
| 334 | error: |
| 335 | #ifdef WITH_THREAD |
| 336 | PyThread_free_lock(self->lock); |
| 337 | self->lock = NULL; |
| 338 | #endif |
| 339 | return -1; |
| 340 | } |
| 341 | |
| 342 | static void |
| 343 | BZ2Compressor_dealloc(BZ2Compressor *self) |
| 344 | { |
| 345 | BZ2_bzCompressEnd(&self->bzs); |
| 346 | #ifdef WITH_THREAD |
| 347 | if (self->lock != NULL) |
| 348 | PyThread_free_lock(self->lock); |
| 349 | #endif |
| 350 | Py_TYPE(self)->tp_free((PyObject *)self); |
| 351 | } |
| 352 | |
| 353 | static PyMethodDef BZ2Compressor_methods[] = { |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 354 | _BZ2_BZ2COMPRESSOR_COMPRESS_METHODDEF |
| 355 | _BZ2_BZ2COMPRESSOR_FLUSH_METHODDEF |
Nadeem Vawda | 3797065 | 2013-10-28 21:35:23 +0100 | [diff] [blame] | 356 | {"__getstate__", (PyCFunction)BZ2Compressor_getstate, METH_NOARGS}, |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 357 | {NULL} |
| 358 | }; |
| 359 | |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 360 | |
| 361 | static PyTypeObject BZ2Compressor_Type = { |
| 362 | PyVarObject_HEAD_INIT(NULL, 0) |
| 363 | "_bz2.BZ2Compressor", /* tp_name */ |
| 364 | sizeof(BZ2Compressor), /* tp_basicsize */ |
| 365 | 0, /* tp_itemsize */ |
| 366 | (destructor)BZ2Compressor_dealloc, /* tp_dealloc */ |
| 367 | 0, /* tp_print */ |
| 368 | 0, /* tp_getattr */ |
| 369 | 0, /* tp_setattr */ |
| 370 | 0, /* tp_reserved */ |
| 371 | 0, /* tp_repr */ |
| 372 | 0, /* tp_as_number */ |
| 373 | 0, /* tp_as_sequence */ |
| 374 | 0, /* tp_as_mapping */ |
| 375 | 0, /* tp_hash */ |
| 376 | 0, /* tp_call */ |
| 377 | 0, /* tp_str */ |
| 378 | 0, /* tp_getattro */ |
| 379 | 0, /* tp_setattro */ |
| 380 | 0, /* tp_as_buffer */ |
| 381 | Py_TPFLAGS_DEFAULT, /* tp_flags */ |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 382 | _bz2_BZ2Compressor___init____doc__, /* tp_doc */ |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 383 | 0, /* tp_traverse */ |
| 384 | 0, /* tp_clear */ |
| 385 | 0, /* tp_richcompare */ |
| 386 | 0, /* tp_weaklistoffset */ |
| 387 | 0, /* tp_iter */ |
| 388 | 0, /* tp_iternext */ |
| 389 | BZ2Compressor_methods, /* tp_methods */ |
| 390 | 0, /* tp_members */ |
| 391 | 0, /* tp_getset */ |
| 392 | 0, /* tp_base */ |
| 393 | 0, /* tp_dict */ |
| 394 | 0, /* tp_descr_get */ |
| 395 | 0, /* tp_descr_set */ |
| 396 | 0, /* tp_dictoffset */ |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 397 | _bz2_BZ2Compressor___init__, /* tp_init */ |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 398 | 0, /* tp_alloc */ |
| 399 | PyType_GenericNew, /* tp_new */ |
| 400 | }; |
| 401 | |
| 402 | |
| 403 | /* BZ2Decompressor class. */ |
| 404 | |
| 405 | static PyObject * |
| 406 | decompress(BZ2Decompressor *d, char *data, size_t len) |
| 407 | { |
| 408 | size_t data_size = 0; |
| 409 | PyObject *result; |
| 410 | |
| 411 | result = PyBytes_FromStringAndSize(NULL, SMALLCHUNK); |
| 412 | if (result == NULL) |
| 413 | return result; |
| 414 | d->bzs.next_in = data; |
Nadeem Vawda | ea4b46f | 2011-04-12 23:02:42 +0200 | [diff] [blame] | 415 | /* On a 64-bit system, len might not fit in avail_in (an unsigned int). |
| 416 | Do decompression in chunks of no more than UINT_MAX bytes each. */ |
Victor Stinner | fbf50d4 | 2013-06-04 23:18:48 +0200 | [diff] [blame] | 417 | d->bzs.avail_in = (unsigned int)Py_MIN(len, UINT_MAX); |
Nadeem Vawda | ea4b46f | 2011-04-12 23:02:42 +0200 | [diff] [blame] | 418 | len -= d->bzs.avail_in; |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 419 | d->bzs.next_out = PyBytes_AS_STRING(result); |
Victor Stinner | fbf50d4 | 2013-06-04 23:18:48 +0200 | [diff] [blame] | 420 | d->bzs.avail_out = SMALLCHUNK; |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 421 | for (;;) { |
| 422 | char *this_out; |
| 423 | int bzerror; |
| 424 | |
| 425 | Py_BEGIN_ALLOW_THREADS |
| 426 | this_out = d->bzs.next_out; |
| 427 | bzerror = BZ2_bzDecompress(&d->bzs); |
| 428 | data_size += d->bzs.next_out - this_out; |
| 429 | Py_END_ALLOW_THREADS |
| 430 | if (catch_bz2_error(bzerror)) |
| 431 | goto error; |
| 432 | if (bzerror == BZ_STREAM_END) { |
| 433 | d->eof = 1; |
Nadeem Vawda | ea4b46f | 2011-04-12 23:02:42 +0200 | [diff] [blame] | 434 | len += d->bzs.avail_in; |
| 435 | if (len > 0) { /* Save leftover input to unused_data */ |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 436 | Py_CLEAR(d->unused_data); |
Nadeem Vawda | ea4b46f | 2011-04-12 23:02:42 +0200 | [diff] [blame] | 437 | d->unused_data = PyBytes_FromStringAndSize(d->bzs.next_in, len); |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 438 | if (d->unused_data == NULL) |
| 439 | goto error; |
| 440 | } |
| 441 | break; |
| 442 | } |
Nadeem Vawda | ea4b46f | 2011-04-12 23:02:42 +0200 | [diff] [blame] | 443 | if (d->bzs.avail_in == 0) { |
| 444 | if (len == 0) |
| 445 | break; |
Victor Stinner | fbf50d4 | 2013-06-04 23:18:48 +0200 | [diff] [blame] | 446 | d->bzs.avail_in = (unsigned int)Py_MIN(len, UINT_MAX); |
Nadeem Vawda | ea4b46f | 2011-04-12 23:02:42 +0200 | [diff] [blame] | 447 | len -= d->bzs.avail_in; |
| 448 | } |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 449 | if (d->bzs.avail_out == 0) { |
Nadeem Vawda | 18b7fcc | 2012-10-21 21:16:58 +0200 | [diff] [blame] | 450 | size_t buffer_left = PyBytes_GET_SIZE(result) - data_size; |
| 451 | if (buffer_left == 0) { |
| 452 | if (grow_buffer(&result) < 0) |
| 453 | goto error; |
| 454 | d->bzs.next_out = PyBytes_AS_STRING(result) + data_size; |
| 455 | buffer_left = PyBytes_GET_SIZE(result) - data_size; |
| 456 | } |
Victor Stinner | fbf50d4 | 2013-06-04 23:18:48 +0200 | [diff] [blame] | 457 | d->bzs.avail_out = (unsigned int)Py_MIN(buffer_left, UINT_MAX); |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 458 | } |
| 459 | } |
| 460 | if (data_size != PyBytes_GET_SIZE(result)) |
| 461 | if (_PyBytes_Resize(&result, data_size) < 0) |
| 462 | goto error; |
| 463 | return result; |
| 464 | |
| 465 | error: |
| 466 | Py_XDECREF(result); |
| 467 | return NULL; |
| 468 | } |
| 469 | |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 470 | /*[clinic input] |
| 471 | _bz2.BZ2Decompressor.decompress |
| 472 | |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 473 | data: Py_buffer |
| 474 | / |
| 475 | |
| 476 | Provide data to the decompressor object. |
| 477 | |
| 478 | Returns a chunk of decompressed data if possible, or b'' otherwise. |
| 479 | |
| 480 | Attempting to decompress data after the end of stream is reached |
| 481 | raises an EOFError. Any data found after the end of the stream |
| 482 | is ignored and saved in the unused_data attribute. |
| 483 | [clinic start generated code]*/ |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 484 | |
| 485 | static PyObject * |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 486 | _bz2_BZ2Decompressor_decompress_impl(BZ2Decompressor *self, Py_buffer *data) |
Larry Hastings | 581ee36 | 2014-01-28 05:00:08 -0800 | [diff] [blame] | 487 | /*[clinic end generated code: output=086e4b99e60cb3f6 input=616c2a6db5269961]*/ |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 488 | { |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 489 | PyObject *result = NULL; |
| 490 | |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 491 | ACQUIRE_LOCK(self); |
| 492 | if (self->eof) |
| 493 | PyErr_SetString(PyExc_EOFError, "End of stream already reached"); |
| 494 | else |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 495 | result = decompress(self, data->buf, data->len); |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 496 | RELEASE_LOCK(self); |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 497 | return result; |
| 498 | } |
| 499 | |
Nadeem Vawda | 3797065 | 2013-10-28 21:35:23 +0100 | [diff] [blame] | 500 | static PyObject * |
| 501 | BZ2Decompressor_getstate(BZ2Decompressor *self, PyObject *noargs) |
| 502 | { |
| 503 | PyErr_Format(PyExc_TypeError, "cannot serialize '%s' object", |
| 504 | Py_TYPE(self)->tp_name); |
| 505 | return NULL; |
| 506 | } |
| 507 | |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 508 | /*[clinic input] |
| 509 | _bz2.BZ2Decompressor.__init__ |
| 510 | |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 511 | Create a decompressor object for decompressing data incrementally. |
| 512 | |
| 513 | For one-shot decompression, use the decompress() function instead. |
| 514 | [clinic start generated code]*/ |
| 515 | |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 516 | static int |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 517 | _bz2_BZ2Decompressor___init___impl(BZ2Decompressor *self) |
Larry Hastings | 581ee36 | 2014-01-28 05:00:08 -0800 | [diff] [blame] | 518 | /*[clinic end generated code: output=e4d2b9bb866ab8f1 input=95f6500dcda60088]*/ |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 519 | { |
| 520 | int bzerror; |
| 521 | |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 522 | #ifdef WITH_THREAD |
| 523 | self->lock = PyThread_allocate_lock(); |
| 524 | if (self->lock == NULL) { |
| 525 | PyErr_SetString(PyExc_MemoryError, "Unable to allocate lock"); |
| 526 | return -1; |
| 527 | } |
| 528 | #endif |
| 529 | |
| 530 | self->unused_data = PyBytes_FromStringAndSize("", 0); |
| 531 | if (self->unused_data == NULL) |
| 532 | goto error; |
| 533 | |
| 534 | bzerror = BZ2_bzDecompressInit(&self->bzs, 0, 0); |
| 535 | if (catch_bz2_error(bzerror)) |
| 536 | goto error; |
| 537 | |
| 538 | return 0; |
| 539 | |
| 540 | error: |
| 541 | Py_CLEAR(self->unused_data); |
| 542 | #ifdef WITH_THREAD |
| 543 | PyThread_free_lock(self->lock); |
| 544 | self->lock = NULL; |
| 545 | #endif |
| 546 | return -1; |
| 547 | } |
| 548 | |
| 549 | static void |
| 550 | BZ2Decompressor_dealloc(BZ2Decompressor *self) |
| 551 | { |
| 552 | BZ2_bzDecompressEnd(&self->bzs); |
| 553 | Py_CLEAR(self->unused_data); |
| 554 | #ifdef WITH_THREAD |
| 555 | if (self->lock != NULL) |
| 556 | PyThread_free_lock(self->lock); |
| 557 | #endif |
| 558 | Py_TYPE(self)->tp_free((PyObject *)self); |
| 559 | } |
| 560 | |
| 561 | static PyMethodDef BZ2Decompressor_methods[] = { |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 562 | _BZ2_BZ2DECOMPRESSOR_DECOMPRESS_METHODDEF |
Nadeem Vawda | 3797065 | 2013-10-28 21:35:23 +0100 | [diff] [blame] | 563 | {"__getstate__", (PyCFunction)BZ2Decompressor_getstate, METH_NOARGS}, |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 564 | {NULL} |
| 565 | }; |
| 566 | |
| 567 | PyDoc_STRVAR(BZ2Decompressor_eof__doc__, |
| 568 | "True if the end-of-stream marker has been reached."); |
| 569 | |
| 570 | PyDoc_STRVAR(BZ2Decompressor_unused_data__doc__, |
| 571 | "Data found after the end of the compressed stream."); |
| 572 | |
| 573 | static PyMemberDef BZ2Decompressor_members[] = { |
| 574 | {"eof", T_BOOL, offsetof(BZ2Decompressor, eof), |
| 575 | READONLY, BZ2Decompressor_eof__doc__}, |
| 576 | {"unused_data", T_OBJECT_EX, offsetof(BZ2Decompressor, unused_data), |
| 577 | READONLY, BZ2Decompressor_unused_data__doc__}, |
| 578 | {NULL} |
| 579 | }; |
| 580 | |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 581 | static PyTypeObject BZ2Decompressor_Type = { |
| 582 | PyVarObject_HEAD_INIT(NULL, 0) |
| 583 | "_bz2.BZ2Decompressor", /* tp_name */ |
| 584 | sizeof(BZ2Decompressor), /* tp_basicsize */ |
| 585 | 0, /* tp_itemsize */ |
| 586 | (destructor)BZ2Decompressor_dealloc,/* tp_dealloc */ |
| 587 | 0, /* tp_print */ |
| 588 | 0, /* tp_getattr */ |
| 589 | 0, /* tp_setattr */ |
| 590 | 0, /* tp_reserved */ |
| 591 | 0, /* tp_repr */ |
| 592 | 0, /* tp_as_number */ |
| 593 | 0, /* tp_as_sequence */ |
| 594 | 0, /* tp_as_mapping */ |
| 595 | 0, /* tp_hash */ |
| 596 | 0, /* tp_call */ |
| 597 | 0, /* tp_str */ |
| 598 | 0, /* tp_getattro */ |
| 599 | 0, /* tp_setattro */ |
| 600 | 0, /* tp_as_buffer */ |
| 601 | Py_TPFLAGS_DEFAULT, /* tp_flags */ |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 602 | _bz2_BZ2Decompressor___init____doc__, /* tp_doc */ |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 603 | 0, /* tp_traverse */ |
| 604 | 0, /* tp_clear */ |
| 605 | 0, /* tp_richcompare */ |
| 606 | 0, /* tp_weaklistoffset */ |
| 607 | 0, /* tp_iter */ |
| 608 | 0, /* tp_iternext */ |
| 609 | BZ2Decompressor_methods, /* tp_methods */ |
| 610 | BZ2Decompressor_members, /* tp_members */ |
| 611 | 0, /* tp_getset */ |
| 612 | 0, /* tp_base */ |
| 613 | 0, /* tp_dict */ |
| 614 | 0, /* tp_descr_get */ |
| 615 | 0, /* tp_descr_set */ |
| 616 | 0, /* tp_dictoffset */ |
Serhiy Storchaka | 1bc4bb2 | 2014-01-25 12:07:57 +0200 | [diff] [blame] | 617 | _bz2_BZ2Decompressor___init__, /* tp_init */ |
Antoine Pitrou | 37dc5f8 | 2011-04-03 17:05:46 +0200 | [diff] [blame] | 618 | 0, /* tp_alloc */ |
| 619 | PyType_GenericNew, /* tp_new */ |
| 620 | }; |
| 621 | |
| 622 | |
| 623 | /* Module initialization. */ |
| 624 | |
| 625 | static struct PyModuleDef _bz2module = { |
| 626 | PyModuleDef_HEAD_INIT, |
| 627 | "_bz2", |
| 628 | NULL, |
| 629 | -1, |
| 630 | NULL, |
| 631 | NULL, |
| 632 | NULL, |
| 633 | NULL, |
| 634 | NULL |
| 635 | }; |
| 636 | |
| 637 | PyMODINIT_FUNC |
| 638 | PyInit__bz2(void) |
| 639 | { |
| 640 | PyObject *m; |
| 641 | |
| 642 | if (PyType_Ready(&BZ2Compressor_Type) < 0) |
| 643 | return NULL; |
| 644 | if (PyType_Ready(&BZ2Decompressor_Type) < 0) |
| 645 | return NULL; |
| 646 | |
| 647 | m = PyModule_Create(&_bz2module); |
| 648 | if (m == NULL) |
| 649 | return NULL; |
| 650 | |
| 651 | Py_INCREF(&BZ2Compressor_Type); |
| 652 | PyModule_AddObject(m, "BZ2Compressor", (PyObject *)&BZ2Compressor_Type); |
| 653 | |
| 654 | Py_INCREF(&BZ2Decompressor_Type); |
| 655 | PyModule_AddObject(m, "BZ2Decompressor", |
| 656 | (PyObject *)&BZ2Decompressor_Type); |
| 657 | |
| 658 | return m; |
| 659 | } |