1 | #include "Python.h"
|
---|
2 | #include "structmember.h" /* for offsetof() */
|
---|
3 | #include "_iomodule.h"
|
---|
4 |
|
---|
5 | typedef struct {
|
---|
6 | PyObject_HEAD
|
---|
7 | char *buf;
|
---|
8 | Py_ssize_t pos;
|
---|
9 | Py_ssize_t string_size;
|
---|
10 | size_t buf_size;
|
---|
11 | PyObject *dict;
|
---|
12 | PyObject *weakreflist;
|
---|
13 | } bytesio;
|
---|
14 |
|
---|
15 | #define CHECK_CLOSED(self) \
|
---|
16 | if ((self)->buf == NULL) { \
|
---|
17 | PyErr_SetString(PyExc_ValueError, \
|
---|
18 | "I/O operation on closed file."); \
|
---|
19 | return NULL; \
|
---|
20 | }
|
---|
21 |
|
---|
22 | /* Internal routine to get a line from the buffer of a BytesIO
|
---|
23 | object. Returns the length between the current position to the
|
---|
24 | next newline character. */
|
---|
25 | static Py_ssize_t
|
---|
26 | get_line(bytesio *self, char **output)
|
---|
27 | {
|
---|
28 | char *n;
|
---|
29 | const char *str_end;
|
---|
30 | Py_ssize_t len;
|
---|
31 |
|
---|
32 | assert(self->buf != NULL);
|
---|
33 |
|
---|
34 | /* Move to the end of the line, up to the end of the string, s. */
|
---|
35 | str_end = self->buf + self->string_size;
|
---|
36 | for (n = self->buf + self->pos;
|
---|
37 | n < str_end && *n != '\n';
|
---|
38 | n++);
|
---|
39 |
|
---|
40 | /* Skip the newline character */
|
---|
41 | if (n < str_end)
|
---|
42 | n++;
|
---|
43 |
|
---|
44 | /* Get the length from the current position to the end of the line. */
|
---|
45 | len = n - (self->buf + self->pos);
|
---|
46 | *output = self->buf + self->pos;
|
---|
47 |
|
---|
48 | assert(len >= 0);
|
---|
49 | assert(self->pos < PY_SSIZE_T_MAX - len);
|
---|
50 | self->pos += len;
|
---|
51 |
|
---|
52 | return len;
|
---|
53 | }
|
---|
54 |
|
---|
55 | /* Internal routine for changing the size of the buffer of BytesIO objects.
|
---|
56 | The caller should ensure that the 'size' argument is non-negative. Returns
|
---|
57 | 0 on success, -1 otherwise. */
|
---|
58 | static int
|
---|
59 | resize_buffer(bytesio *self, size_t size)
|
---|
60 | {
|
---|
61 | /* Here, unsigned types are used to avoid dealing with signed integer
|
---|
62 | overflow, which is undefined in C. */
|
---|
63 | size_t alloc = self->buf_size;
|
---|
64 | char *new_buf = NULL;
|
---|
65 |
|
---|
66 | assert(self->buf != NULL);
|
---|
67 |
|
---|
68 | /* For simplicity, stay in the range of the signed type. Anyway, Python
|
---|
69 | doesn't allow strings to be longer than this. */
|
---|
70 | if (size > PY_SSIZE_T_MAX)
|
---|
71 | goto overflow;
|
---|
72 |
|
---|
73 | if (size < alloc / 2) {
|
---|
74 | /* Major downsize; resize down to exact size. */
|
---|
75 | alloc = size + 1;
|
---|
76 | }
|
---|
77 | else if (size < alloc) {
|
---|
78 | /* Within allocated size; quick exit */
|
---|
79 | return 0;
|
---|
80 | }
|
---|
81 | else if (size <= alloc * 1.125) {
|
---|
82 | /* Moderate upsize; overallocate similar to list_resize() */
|
---|
83 | alloc = size + (size >> 3) + (size < 9 ? 3 : 6);
|
---|
84 | }
|
---|
85 | else {
|
---|
86 | /* Major upsize; resize up to exact size */
|
---|
87 | alloc = size + 1;
|
---|
88 | }
|
---|
89 |
|
---|
90 | if (alloc > ((size_t)-1) / sizeof(char))
|
---|
91 | goto overflow;
|
---|
92 | new_buf = (char *)PyMem_Realloc(self->buf, alloc * sizeof(char));
|
---|
93 | if (new_buf == NULL) {
|
---|
94 | PyErr_NoMemory();
|
---|
95 | return -1;
|
---|
96 | }
|
---|
97 | self->buf_size = alloc;
|
---|
98 | self->buf = new_buf;
|
---|
99 |
|
---|
100 | return 0;
|
---|
101 |
|
---|
102 | overflow:
|
---|
103 | PyErr_SetString(PyExc_OverflowError,
|
---|
104 | "new buffer size too large");
|
---|
105 | return -1;
|
---|
106 | }
|
---|
107 |
|
---|
108 | /* Internal routine for writing a string of bytes to the buffer of a BytesIO
|
---|
109 | object. Returns the number of bytes written, or -1 on error. */
|
---|
110 | static Py_ssize_t
|
---|
111 | write_bytes(bytesio *self, const char *bytes, Py_ssize_t len)
|
---|
112 | {
|
---|
113 | assert(self->buf != NULL);
|
---|
114 | assert(self->pos >= 0);
|
---|
115 | assert(len >= 0);
|
---|
116 |
|
---|
117 | if ((size_t)self->pos + len > self->buf_size) {
|
---|
118 | if (resize_buffer(self, (size_t)self->pos + len) < 0)
|
---|
119 | return -1;
|
---|
120 | }
|
---|
121 |
|
---|
122 | if (self->pos > self->string_size) {
|
---|
123 | /* In case of overseek, pad with null bytes the buffer region between
|
---|
124 | the end of stream and the current position.
|
---|
125 |
|
---|
126 | 0 lo string_size hi
|
---|
127 | | |<---used--->|<----------available----------->|
|
---|
128 | | | <--to pad-->|<---to write---> |
|
---|
129 | 0 buf position
|
---|
130 | */
|
---|
131 | memset(self->buf + self->string_size, '\0',
|
---|
132 | (self->pos - self->string_size) * sizeof(char));
|
---|
133 | }
|
---|
134 |
|
---|
135 | /* Copy the data to the internal buffer, overwriting some of the existing
|
---|
136 | data if self->pos < self->string_size. */
|
---|
137 | memcpy(self->buf + self->pos, bytes, len);
|
---|
138 | self->pos += len;
|
---|
139 |
|
---|
140 | /* Set the new length of the internal string if it has changed. */
|
---|
141 | if (self->string_size < self->pos) {
|
---|
142 | self->string_size = self->pos;
|
---|
143 | }
|
---|
144 |
|
---|
145 | return len;
|
---|
146 | }
|
---|
147 |
|
---|
148 | static PyObject *
|
---|
149 | bytesio_get_closed(bytesio *self)
|
---|
150 | {
|
---|
151 | if (self->buf == NULL) {
|
---|
152 | Py_RETURN_TRUE;
|
---|
153 | }
|
---|
154 | else {
|
---|
155 | Py_RETURN_FALSE;
|
---|
156 | }
|
---|
157 | }
|
---|
158 |
|
---|
159 | PyDoc_STRVAR(readable_doc,
|
---|
160 | "readable() -> bool. Returns True if the IO object can be read.");
|
---|
161 |
|
---|
162 | PyDoc_STRVAR(writable_doc,
|
---|
163 | "writable() -> bool. Returns True if the IO object can be written.");
|
---|
164 |
|
---|
165 | PyDoc_STRVAR(seekable_doc,
|
---|
166 | "seekable() -> bool. Returns True if the IO object can be seeked.");
|
---|
167 |
|
---|
168 | /* Generic getter for the writable, readable and seekable properties */
|
---|
169 | static PyObject *
|
---|
170 | return_not_closed(bytesio *self)
|
---|
171 | {
|
---|
172 | CHECK_CLOSED(self);
|
---|
173 | Py_RETURN_TRUE;
|
---|
174 | }
|
---|
175 |
|
---|
176 | PyDoc_STRVAR(flush_doc,
|
---|
177 | "flush() -> None. Does nothing.");
|
---|
178 |
|
---|
179 | static PyObject *
|
---|
180 | bytesio_flush(bytesio *self)
|
---|
181 | {
|
---|
182 | CHECK_CLOSED(self);
|
---|
183 | Py_RETURN_NONE;
|
---|
184 | }
|
---|
185 |
|
---|
186 | PyDoc_STRVAR(getval_doc,
|
---|
187 | "getvalue() -> bytes.\n"
|
---|
188 | "\n"
|
---|
189 | "Retrieve the entire contents of the BytesIO object.");
|
---|
190 |
|
---|
191 | static PyObject *
|
---|
192 | bytesio_getvalue(bytesio *self)
|
---|
193 | {
|
---|
194 | CHECK_CLOSED(self);
|
---|
195 | return PyBytes_FromStringAndSize(self->buf, self->string_size);
|
---|
196 | }
|
---|
197 |
|
---|
198 | PyDoc_STRVAR(isatty_doc,
|
---|
199 | "isatty() -> False.\n"
|
---|
200 | "\n"
|
---|
201 | "Always returns False since BytesIO objects are not connected\n"
|
---|
202 | "to a tty-like device.");
|
---|
203 |
|
---|
204 | static PyObject *
|
---|
205 | bytesio_isatty(bytesio *self)
|
---|
206 | {
|
---|
207 | CHECK_CLOSED(self);
|
---|
208 | Py_RETURN_FALSE;
|
---|
209 | }
|
---|
210 |
|
---|
211 | PyDoc_STRVAR(tell_doc,
|
---|
212 | "tell() -> current file position, an integer\n");
|
---|
213 |
|
---|
214 | static PyObject *
|
---|
215 | bytesio_tell(bytesio *self)
|
---|
216 | {
|
---|
217 | CHECK_CLOSED(self);
|
---|
218 | return PyLong_FromSsize_t(self->pos);
|
---|
219 | }
|
---|
220 |
|
---|
221 | PyDoc_STRVAR(read_doc,
|
---|
222 | "read([size]) -> read at most size bytes, returned as a string.\n"
|
---|
223 | "\n"
|
---|
224 | "If the size argument is negative, read until EOF is reached.\n"
|
---|
225 | "Return an empty string at EOF.");
|
---|
226 |
|
---|
227 | static PyObject *
|
---|
228 | bytesio_read(bytesio *self, PyObject *args)
|
---|
229 | {
|
---|
230 | Py_ssize_t size, n;
|
---|
231 | char *output;
|
---|
232 | PyObject *arg = Py_None;
|
---|
233 |
|
---|
234 | CHECK_CLOSED(self);
|
---|
235 |
|
---|
236 | if (!PyArg_ParseTuple(args, "|O:read", &arg))
|
---|
237 | return NULL;
|
---|
238 |
|
---|
239 | if (PyNumber_Check(arg)) {
|
---|
240 | size = PyNumber_AsSsize_t(arg, PyExc_OverflowError);
|
---|
241 | if (size == -1 && PyErr_Occurred())
|
---|
242 | return NULL;
|
---|
243 | }
|
---|
244 | else if (arg == Py_None) {
|
---|
245 | /* Read until EOF is reached, by default. */
|
---|
246 | size = -1;
|
---|
247 | }
|
---|
248 | else {
|
---|
249 | PyErr_Format(PyExc_TypeError, "integer argument expected, got '%s'",
|
---|
250 | Py_TYPE(arg)->tp_name);
|
---|
251 | return NULL;
|
---|
252 | }
|
---|
253 |
|
---|
254 | /* adjust invalid sizes */
|
---|
255 | n = self->string_size - self->pos;
|
---|
256 | if (size < 0 || size > n) {
|
---|
257 | size = n;
|
---|
258 | if (size < 0)
|
---|
259 | size = 0;
|
---|
260 | }
|
---|
261 |
|
---|
262 | assert(self->buf != NULL);
|
---|
263 | output = self->buf + self->pos;
|
---|
264 | self->pos += size;
|
---|
265 |
|
---|
266 | return PyBytes_FromStringAndSize(output, size);
|
---|
267 | }
|
---|
268 |
|
---|
269 |
|
---|
270 | PyDoc_STRVAR(read1_doc,
|
---|
271 | "read1(size) -> read at most size bytes, returned as a string.\n"
|
---|
272 | "\n"
|
---|
273 | "If the size argument is negative or omitted, read until EOF is reached.\n"
|
---|
274 | "Return an empty string at EOF.");
|
---|
275 |
|
---|
276 | static PyObject *
|
---|
277 | bytesio_read1(bytesio *self, PyObject *n)
|
---|
278 | {
|
---|
279 | PyObject *arg, *res;
|
---|
280 |
|
---|
281 | arg = PyTuple_Pack(1, n);
|
---|
282 | if (arg == NULL)
|
---|
283 | return NULL;
|
---|
284 | res = bytesio_read(self, arg);
|
---|
285 | Py_DECREF(arg);
|
---|
286 | return res;
|
---|
287 | }
|
---|
288 |
|
---|
289 | PyDoc_STRVAR(readline_doc,
|
---|
290 | "readline([size]) -> next line from the file, as a string.\n"
|
---|
291 | "\n"
|
---|
292 | "Retain newline. A non-negative size argument limits the maximum\n"
|
---|
293 | "number of bytes to return (an incomplete line may be returned then).\n"
|
---|
294 | "Return an empty string at EOF.\n");
|
---|
295 |
|
---|
296 | static PyObject *
|
---|
297 | bytesio_readline(bytesio *self, PyObject *args)
|
---|
298 | {
|
---|
299 | Py_ssize_t size, n;
|
---|
300 | char *output;
|
---|
301 | PyObject *arg = Py_None;
|
---|
302 |
|
---|
303 | CHECK_CLOSED(self);
|
---|
304 |
|
---|
305 | if (!PyArg_ParseTuple(args, "|O:readline", &arg))
|
---|
306 | return NULL;
|
---|
307 |
|
---|
308 | if (PyNumber_Check(arg)) {
|
---|
309 | size = PyNumber_AsSsize_t(arg, PyExc_OverflowError);
|
---|
310 | if (size == -1 && PyErr_Occurred())
|
---|
311 | return NULL;
|
---|
312 | }
|
---|
313 | else if (arg == Py_None) {
|
---|
314 | /* No size limit, by default. */
|
---|
315 | size = -1;
|
---|
316 | }
|
---|
317 | else {
|
---|
318 | PyErr_Format(PyExc_TypeError, "integer argument expected, got '%s'",
|
---|
319 | Py_TYPE(arg)->tp_name);
|
---|
320 | return NULL;
|
---|
321 | }
|
---|
322 |
|
---|
323 | n = get_line(self, &output);
|
---|
324 |
|
---|
325 | if (size >= 0 && size < n) {
|
---|
326 | size = n - size;
|
---|
327 | n -= size;
|
---|
328 | self->pos -= size;
|
---|
329 | }
|
---|
330 |
|
---|
331 | return PyBytes_FromStringAndSize(output, n);
|
---|
332 | }
|
---|
333 |
|
---|
334 | PyDoc_STRVAR(readlines_doc,
|
---|
335 | "readlines([size]) -> list of strings, each a line from the file.\n"
|
---|
336 | "\n"
|
---|
337 | "Call readline() repeatedly and return a list of the lines so read.\n"
|
---|
338 | "The optional size argument, if given, is an approximate bound on the\n"
|
---|
339 | "total number of bytes in the lines returned.\n");
|
---|
340 |
|
---|
341 | static PyObject *
|
---|
342 | bytesio_readlines(bytesio *self, PyObject *args)
|
---|
343 | {
|
---|
344 | Py_ssize_t maxsize, size, n;
|
---|
345 | PyObject *result, *line;
|
---|
346 | char *output;
|
---|
347 | PyObject *arg = Py_None;
|
---|
348 |
|
---|
349 | CHECK_CLOSED(self);
|
---|
350 |
|
---|
351 | if (!PyArg_ParseTuple(args, "|O:readlines", &arg))
|
---|
352 | return NULL;
|
---|
353 |
|
---|
354 | if (PyNumber_Check(arg)) {
|
---|
355 | maxsize = PyNumber_AsSsize_t(arg, PyExc_OverflowError);
|
---|
356 | if (maxsize == -1 && PyErr_Occurred())
|
---|
357 | return NULL;
|
---|
358 | }
|
---|
359 | else if (arg == Py_None) {
|
---|
360 | /* No size limit, by default. */
|
---|
361 | maxsize = -1;
|
---|
362 | }
|
---|
363 | else {
|
---|
364 | PyErr_Format(PyExc_TypeError, "integer argument expected, got '%s'",
|
---|
365 | Py_TYPE(arg)->tp_name);
|
---|
366 | return NULL;
|
---|
367 | }
|
---|
368 |
|
---|
369 | size = 0;
|
---|
370 | result = PyList_New(0);
|
---|
371 | if (!result)
|
---|
372 | return NULL;
|
---|
373 |
|
---|
374 | while ((n = get_line(self, &output)) != 0) {
|
---|
375 | line = PyBytes_FromStringAndSize(output, n);
|
---|
376 | if (!line)
|
---|
377 | goto on_error;
|
---|
378 | if (PyList_Append(result, line) == -1) {
|
---|
379 | Py_DECREF(line);
|
---|
380 | goto on_error;
|
---|
381 | }
|
---|
382 | Py_DECREF(line);
|
---|
383 | size += n;
|
---|
384 | if (maxsize > 0 && size >= maxsize)
|
---|
385 | break;
|
---|
386 | }
|
---|
387 | return result;
|
---|
388 |
|
---|
389 | on_error:
|
---|
390 | Py_DECREF(result);
|
---|
391 | return NULL;
|
---|
392 | }
|
---|
393 |
|
---|
394 | PyDoc_STRVAR(readinto_doc,
|
---|
395 | "readinto(bytearray) -> int. Read up to len(b) bytes into b.\n"
|
---|
396 | "\n"
|
---|
397 | "Returns number of bytes read (0 for EOF), or None if the object\n"
|
---|
398 | "is set not to block as has no data to read.");
|
---|
399 |
|
---|
400 | static PyObject *
|
---|
401 | bytesio_readinto(bytesio *self, PyObject *args)
|
---|
402 | {
|
---|
403 | Py_buffer buf;
|
---|
404 | Py_ssize_t len, n;
|
---|
405 |
|
---|
406 | CHECK_CLOSED(self);
|
---|
407 |
|
---|
408 | if (!PyArg_ParseTuple(args, "w*", &buf))
|
---|
409 | return NULL;
|
---|
410 |
|
---|
411 | len = buf.len;
|
---|
412 | /* adjust invalid sizes */
|
---|
413 | n = self->string_size - self->pos;
|
---|
414 | if (len > n) {
|
---|
415 | len = n;
|
---|
416 | if (len < 0)
|
---|
417 | len = 0;
|
---|
418 | }
|
---|
419 |
|
---|
420 | memcpy(buf.buf, self->buf + self->pos, len);
|
---|
421 | assert(self->pos + len < PY_SSIZE_T_MAX);
|
---|
422 | assert(len >= 0);
|
---|
423 | self->pos += len;
|
---|
424 |
|
---|
425 | PyBuffer_Release(&buf);
|
---|
426 | return PyLong_FromSsize_t(len);
|
---|
427 | }
|
---|
428 |
|
---|
429 | PyDoc_STRVAR(truncate_doc,
|
---|
430 | "truncate([size]) -> int. Truncate the file to at most size bytes.\n"
|
---|
431 | "\n"
|
---|
432 | "Size defaults to the current file position, as returned by tell().\n"
|
---|
433 | "The current file position is unchanged. Returns the new size.\n");
|
---|
434 |
|
---|
435 | static PyObject *
|
---|
436 | bytesio_truncate(bytesio *self, PyObject *args)
|
---|
437 | {
|
---|
438 | Py_ssize_t size;
|
---|
439 | PyObject *arg = Py_None;
|
---|
440 |
|
---|
441 | CHECK_CLOSED(self);
|
---|
442 |
|
---|
443 | if (!PyArg_ParseTuple(args, "|O:truncate", &arg))
|
---|
444 | return NULL;
|
---|
445 |
|
---|
446 | if (PyNumber_Check(arg)) {
|
---|
447 | size = PyNumber_AsSsize_t(arg, PyExc_OverflowError);
|
---|
448 | if (size == -1 && PyErr_Occurred())
|
---|
449 | return NULL;
|
---|
450 | }
|
---|
451 | else if (arg == Py_None) {
|
---|
452 | /* Truncate to current position if no argument is passed. */
|
---|
453 | size = self->pos;
|
---|
454 | }
|
---|
455 | else {
|
---|
456 | PyErr_Format(PyExc_TypeError, "integer argument expected, got '%s'",
|
---|
457 | Py_TYPE(arg)->tp_name);
|
---|
458 | return NULL;
|
---|
459 | }
|
---|
460 |
|
---|
461 | if (size < 0) {
|
---|
462 | PyErr_Format(PyExc_ValueError,
|
---|
463 | "negative size value %zd", size);
|
---|
464 | return NULL;
|
---|
465 | }
|
---|
466 |
|
---|
467 | if (size < self->string_size) {
|
---|
468 | self->string_size = size;
|
---|
469 | if (resize_buffer(self, size) < 0)
|
---|
470 | return NULL;
|
---|
471 | }
|
---|
472 |
|
---|
473 | return PyLong_FromSsize_t(size);
|
---|
474 | }
|
---|
475 |
|
---|
476 | static PyObject *
|
---|
477 | bytesio_iternext(bytesio *self)
|
---|
478 | {
|
---|
479 | char *next;
|
---|
480 | Py_ssize_t n;
|
---|
481 |
|
---|
482 | CHECK_CLOSED(self);
|
---|
483 |
|
---|
484 | n = get_line(self, &next);
|
---|
485 |
|
---|
486 | if (!next || n == 0)
|
---|
487 | return NULL;
|
---|
488 |
|
---|
489 | return PyBytes_FromStringAndSize(next, n);
|
---|
490 | }
|
---|
491 |
|
---|
492 | PyDoc_STRVAR(seek_doc,
|
---|
493 | "seek(pos, whence=0) -> int. Change stream position.\n"
|
---|
494 | "\n"
|
---|
495 | "Seek to byte offset pos relative to position indicated by whence:\n"
|
---|
496 | " 0 Start of stream (the default). pos should be >= 0;\n"
|
---|
497 | " 1 Current position - pos may be negative;\n"
|
---|
498 | " 2 End of stream - pos usually negative.\n"
|
---|
499 | "Returns the new absolute position.");
|
---|
500 |
|
---|
501 | static PyObject *
|
---|
502 | bytesio_seek(bytesio *self, PyObject *args)
|
---|
503 | {
|
---|
504 | PyObject *posobj;
|
---|
505 | Py_ssize_t pos;
|
---|
506 | int mode = 0;
|
---|
507 |
|
---|
508 | CHECK_CLOSED(self);
|
---|
509 |
|
---|
510 | if (!PyArg_ParseTuple(args, "O|i:seek", &posobj, &mode))
|
---|
511 | return NULL;
|
---|
512 |
|
---|
513 | pos = PyNumber_AsSsize_t(posobj, PyExc_OverflowError);
|
---|
514 | if (pos == -1 && PyErr_Occurred())
|
---|
515 | return NULL;
|
---|
516 |
|
---|
517 | if (pos < 0 && mode == 0) {
|
---|
518 | PyErr_Format(PyExc_ValueError,
|
---|
519 | "negative seek value %zd", pos);
|
---|
520 | return NULL;
|
---|
521 | }
|
---|
522 |
|
---|
523 | /* mode 0: offset relative to beginning of the string.
|
---|
524 | mode 1: offset relative to current position.
|
---|
525 | mode 2: offset relative the end of the string. */
|
---|
526 | if (mode == 1) {
|
---|
527 | if (pos > PY_SSIZE_T_MAX - self->pos) {
|
---|
528 | PyErr_SetString(PyExc_OverflowError,
|
---|
529 | "new position too large");
|
---|
530 | return NULL;
|
---|
531 | }
|
---|
532 | pos += self->pos;
|
---|
533 | }
|
---|
534 | else if (mode == 2) {
|
---|
535 | if (pos > PY_SSIZE_T_MAX - self->string_size) {
|
---|
536 | PyErr_SetString(PyExc_OverflowError,
|
---|
537 | "new position too large");
|
---|
538 | return NULL;
|
---|
539 | }
|
---|
540 | pos += self->string_size;
|
---|
541 | }
|
---|
542 | else if (mode != 0) {
|
---|
543 | PyErr_Format(PyExc_ValueError,
|
---|
544 | "invalid whence (%i, should be 0, 1 or 2)", mode);
|
---|
545 | return NULL;
|
---|
546 | }
|
---|
547 |
|
---|
548 | if (pos < 0)
|
---|
549 | pos = 0;
|
---|
550 | self->pos = pos;
|
---|
551 |
|
---|
552 | return PyLong_FromSsize_t(self->pos);
|
---|
553 | }
|
---|
554 |
|
---|
555 | PyDoc_STRVAR(write_doc,
|
---|
556 | "write(bytes) -> int. Write bytes to file.\n"
|
---|
557 | "\n"
|
---|
558 | "Return the number of bytes written.");
|
---|
559 |
|
---|
560 | static PyObject *
|
---|
561 | bytesio_write(bytesio *self, PyObject *obj)
|
---|
562 | {
|
---|
563 | Py_ssize_t n = 0;
|
---|
564 | Py_buffer buf;
|
---|
565 | PyObject *result = NULL;
|
---|
566 |
|
---|
567 | CHECK_CLOSED(self);
|
---|
568 |
|
---|
569 | if (PyObject_GetBuffer(obj, &buf, PyBUF_CONTIG_RO) < 0)
|
---|
570 | return NULL;
|
---|
571 |
|
---|
572 | if (buf.len != 0)
|
---|
573 | n = write_bytes(self, buf.buf, buf.len);
|
---|
574 | if (n >= 0)
|
---|
575 | result = PyLong_FromSsize_t(n);
|
---|
576 |
|
---|
577 | PyBuffer_Release(&buf);
|
---|
578 | return result;
|
---|
579 | }
|
---|
580 |
|
---|
581 | PyDoc_STRVAR(writelines_doc,
|
---|
582 | "writelines(sequence_of_strings) -> None. Write strings to the file.\n"
|
---|
583 | "\n"
|
---|
584 | "Note that newlines are not added. The sequence can be any iterable\n"
|
---|
585 | "object producing strings. This is equivalent to calling write() for\n"
|
---|
586 | "each string.");
|
---|
587 |
|
---|
588 | static PyObject *
|
---|
589 | bytesio_writelines(bytesio *self, PyObject *v)
|
---|
590 | {
|
---|
591 | PyObject *it, *item;
|
---|
592 | PyObject *ret;
|
---|
593 |
|
---|
594 | CHECK_CLOSED(self);
|
---|
595 |
|
---|
596 | it = PyObject_GetIter(v);
|
---|
597 | if (it == NULL)
|
---|
598 | return NULL;
|
---|
599 |
|
---|
600 | while ((item = PyIter_Next(it)) != NULL) {
|
---|
601 | ret = bytesio_write(self, item);
|
---|
602 | Py_DECREF(item);
|
---|
603 | if (ret == NULL) {
|
---|
604 | Py_DECREF(it);
|
---|
605 | return NULL;
|
---|
606 | }
|
---|
607 | Py_DECREF(ret);
|
---|
608 | }
|
---|
609 | Py_DECREF(it);
|
---|
610 |
|
---|
611 | /* See if PyIter_Next failed */
|
---|
612 | if (PyErr_Occurred())
|
---|
613 | return NULL;
|
---|
614 |
|
---|
615 | Py_RETURN_NONE;
|
---|
616 | }
|
---|
617 |
|
---|
618 | PyDoc_STRVAR(close_doc,
|
---|
619 | "close() -> None. Disable all I/O operations.");
|
---|
620 |
|
---|
621 | static PyObject *
|
---|
622 | bytesio_close(bytesio *self)
|
---|
623 | {
|
---|
624 | if (self->buf != NULL) {
|
---|
625 | PyMem_Free(self->buf);
|
---|
626 | self->buf = NULL;
|
---|
627 | }
|
---|
628 | Py_RETURN_NONE;
|
---|
629 | }
|
---|
630 |
|
---|
631 | /* Pickling support.
|
---|
632 |
|
---|
633 | Note that only pickle protocol 2 and onward are supported since we use
|
---|
634 | extended __reduce__ API of PEP 307 to make BytesIO instances picklable.
|
---|
635 |
|
---|
636 | Providing support for protocol < 2 would require the __reduce_ex__ method
|
---|
637 | which is notably long-winded when defined properly.
|
---|
638 |
|
---|
639 | For BytesIO, the implementation would similar to one coded for
|
---|
640 | object.__reduce_ex__, but slightly less general. To be more specific, we
|
---|
641 | could call bytesio_getstate directly and avoid checking for the presence of
|
---|
642 | a fallback __reduce__ method. However, we would still need a __newobj__
|
---|
643 | function to use the efficient instance representation of PEP 307.
|
---|
644 | */
|
---|
645 |
|
---|
646 | static PyObject *
|
---|
647 | bytesio_getstate(bytesio *self)
|
---|
648 | {
|
---|
649 | PyObject *initvalue = bytesio_getvalue(self);
|
---|
650 | PyObject *dict;
|
---|
651 | PyObject *state;
|
---|
652 |
|
---|
653 | if (initvalue == NULL)
|
---|
654 | return NULL;
|
---|
655 | if (self->dict == NULL) {
|
---|
656 | Py_INCREF(Py_None);
|
---|
657 | dict = Py_None;
|
---|
658 | }
|
---|
659 | else {
|
---|
660 | dict = PyDict_Copy(self->dict);
|
---|
661 | if (dict == NULL)
|
---|
662 | return NULL;
|
---|
663 | }
|
---|
664 |
|
---|
665 | state = Py_BuildValue("(OnN)", initvalue, self->pos, dict);
|
---|
666 | Py_DECREF(initvalue);
|
---|
667 | return state;
|
---|
668 | }
|
---|
669 |
|
---|
670 | static PyObject *
|
---|
671 | bytesio_setstate(bytesio *self, PyObject *state)
|
---|
672 | {
|
---|
673 | PyObject *result;
|
---|
674 | PyObject *position_obj;
|
---|
675 | PyObject *dict;
|
---|
676 | Py_ssize_t pos;
|
---|
677 |
|
---|
678 | assert(state != NULL);
|
---|
679 |
|
---|
680 | /* We allow the state tuple to be longer than 3, because we may need
|
---|
681 | someday to extend the object's state without breaking
|
---|
682 | backward-compatibility. */
|
---|
683 | if (!PyTuple_Check(state) || Py_SIZE(state) < 3) {
|
---|
684 | PyErr_Format(PyExc_TypeError,
|
---|
685 | "%.200s.__setstate__ argument should be 3-tuple, got %.200s",
|
---|
686 | Py_TYPE(self)->tp_name, Py_TYPE(state)->tp_name);
|
---|
687 | return NULL;
|
---|
688 | }
|
---|
689 | /* Reset the object to its default state. This is only needed to handle
|
---|
690 | the case of repeated calls to __setstate__. */
|
---|
691 | self->string_size = 0;
|
---|
692 | self->pos = 0;
|
---|
693 |
|
---|
694 | /* Set the value of the internal buffer. If state[0] does not support the
|
---|
695 | buffer protocol, bytesio_write will raise the appropriate TypeError. */
|
---|
696 | result = bytesio_write(self, PyTuple_GET_ITEM(state, 0));
|
---|
697 | if (result == NULL)
|
---|
698 | return NULL;
|
---|
699 | Py_DECREF(result);
|
---|
700 |
|
---|
701 | /* Set carefully the position value. Alternatively, we could use the seek
|
---|
702 | method instead of modifying self->pos directly to better protect the
|
---|
703 | object internal state against errneous (or malicious) inputs. */
|
---|
704 | position_obj = PyTuple_GET_ITEM(state, 1);
|
---|
705 | if (!PyIndex_Check(position_obj)) {
|
---|
706 | PyErr_Format(PyExc_TypeError,
|
---|
707 | "second item of state must be an integer, not %.200s",
|
---|
708 | Py_TYPE(position_obj)->tp_name);
|
---|
709 | return NULL;
|
---|
710 | }
|
---|
711 | pos = PyNumber_AsSsize_t(position_obj, PyExc_OverflowError);
|
---|
712 | if (pos == -1 && PyErr_Occurred())
|
---|
713 | return NULL;
|
---|
714 | if (pos < 0) {
|
---|
715 | PyErr_SetString(PyExc_ValueError,
|
---|
716 | "position value cannot be negative");
|
---|
717 | return NULL;
|
---|
718 | }
|
---|
719 | self->pos = pos;
|
---|
720 |
|
---|
721 | /* Set the dictionary of the instance variables. */
|
---|
722 | dict = PyTuple_GET_ITEM(state, 2);
|
---|
723 | if (dict != Py_None) {
|
---|
724 | if (!PyDict_Check(dict)) {
|
---|
725 | PyErr_Format(PyExc_TypeError,
|
---|
726 | "third item of state should be a dict, got a %.200s",
|
---|
727 | Py_TYPE(dict)->tp_name);
|
---|
728 | return NULL;
|
---|
729 | }
|
---|
730 | if (self->dict) {
|
---|
731 | /* Alternatively, we could replace the internal dictionary
|
---|
732 | completely. However, it seems more practical to just update it. */
|
---|
733 | if (PyDict_Update(self->dict, dict) < 0)
|
---|
734 | return NULL;
|
---|
735 | }
|
---|
736 | else {
|
---|
737 | Py_INCREF(dict);
|
---|
738 | self->dict = dict;
|
---|
739 | }
|
---|
740 | }
|
---|
741 |
|
---|
742 | Py_RETURN_NONE;
|
---|
743 | }
|
---|
744 |
|
---|
745 | static void
|
---|
746 | bytesio_dealloc(bytesio *self)
|
---|
747 | {
|
---|
748 | _PyObject_GC_UNTRACK(self);
|
---|
749 | if (self->buf != NULL) {
|
---|
750 | PyMem_Free(self->buf);
|
---|
751 | self->buf = NULL;
|
---|
752 | }
|
---|
753 | Py_CLEAR(self->dict);
|
---|
754 | if (self->weakreflist != NULL)
|
---|
755 | PyObject_ClearWeakRefs((PyObject *) self);
|
---|
756 | Py_TYPE(self)->tp_free(self);
|
---|
757 | }
|
---|
758 |
|
---|
759 | static PyObject *
|
---|
760 | bytesio_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
|
---|
761 | {
|
---|
762 | bytesio *self;
|
---|
763 |
|
---|
764 | assert(type != NULL && type->tp_alloc != NULL);
|
---|
765 | self = (bytesio *)type->tp_alloc(type, 0);
|
---|
766 | if (self == NULL)
|
---|
767 | return NULL;
|
---|
768 |
|
---|
769 | /* tp_alloc initializes all the fields to zero. So we don't have to
|
---|
770 | initialize them here. */
|
---|
771 |
|
---|
772 | self->buf = (char *)PyMem_Malloc(0);
|
---|
773 | if (self->buf == NULL) {
|
---|
774 | Py_DECREF(self);
|
---|
775 | return PyErr_NoMemory();
|
---|
776 | }
|
---|
777 |
|
---|
778 | return (PyObject *)self;
|
---|
779 | }
|
---|
780 |
|
---|
781 | static int
|
---|
782 | bytesio_init(bytesio *self, PyObject *args, PyObject *kwds)
|
---|
783 | {
|
---|
784 | char *kwlist[] = {"initial_bytes", NULL};
|
---|
785 | PyObject *initvalue = NULL;
|
---|
786 |
|
---|
787 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "|O:BytesIO", kwlist,
|
---|
788 | &initvalue))
|
---|
789 | return -1;
|
---|
790 |
|
---|
791 | /* In case, __init__ is called multiple times. */
|
---|
792 | self->string_size = 0;
|
---|
793 | self->pos = 0;
|
---|
794 |
|
---|
795 | if (initvalue && initvalue != Py_None) {
|
---|
796 | PyObject *res;
|
---|
797 | res = bytesio_write(self, initvalue);
|
---|
798 | if (res == NULL)
|
---|
799 | return -1;
|
---|
800 | Py_DECREF(res);
|
---|
801 | self->pos = 0;
|
---|
802 | }
|
---|
803 |
|
---|
804 | return 0;
|
---|
805 | }
|
---|
806 |
|
---|
807 | static PyObject *
|
---|
808 | bytesio_sizeof(bytesio *self, void *unused)
|
---|
809 | {
|
---|
810 | Py_ssize_t res;
|
---|
811 |
|
---|
812 | res = sizeof(bytesio);
|
---|
813 | if (self->buf)
|
---|
814 | res += self->buf_size;
|
---|
815 | return PyLong_FromSsize_t(res);
|
---|
816 | }
|
---|
817 |
|
---|
818 | static int
|
---|
819 | bytesio_traverse(bytesio *self, visitproc visit, void *arg)
|
---|
820 | {
|
---|
821 | Py_VISIT(self->dict);
|
---|
822 | return 0;
|
---|
823 | }
|
---|
824 |
|
---|
825 | static int
|
---|
826 | bytesio_clear(bytesio *self)
|
---|
827 | {
|
---|
828 | Py_CLEAR(self->dict);
|
---|
829 | return 0;
|
---|
830 | }
|
---|
831 |
|
---|
832 |
|
---|
833 | static PyGetSetDef bytesio_getsetlist[] = {
|
---|
834 | {"closed", (getter)bytesio_get_closed, NULL,
|
---|
835 | "True if the file is closed."},
|
---|
836 | {NULL}, /* sentinel */
|
---|
837 | };
|
---|
838 |
|
---|
839 | static struct PyMethodDef bytesio_methods[] = {
|
---|
840 | {"readable", (PyCFunction)return_not_closed, METH_NOARGS, readable_doc},
|
---|
841 | {"seekable", (PyCFunction)return_not_closed, METH_NOARGS, seekable_doc},
|
---|
842 | {"writable", (PyCFunction)return_not_closed, METH_NOARGS, writable_doc},
|
---|
843 | {"close", (PyCFunction)bytesio_close, METH_NOARGS, close_doc},
|
---|
844 | {"flush", (PyCFunction)bytesio_flush, METH_NOARGS, flush_doc},
|
---|
845 | {"isatty", (PyCFunction)bytesio_isatty, METH_NOARGS, isatty_doc},
|
---|
846 | {"tell", (PyCFunction)bytesio_tell, METH_NOARGS, tell_doc},
|
---|
847 | {"write", (PyCFunction)bytesio_write, METH_O, write_doc},
|
---|
848 | {"writelines", (PyCFunction)bytesio_writelines, METH_O, writelines_doc},
|
---|
849 | {"read1", (PyCFunction)bytesio_read1, METH_O, read1_doc},
|
---|
850 | {"readinto", (PyCFunction)bytesio_readinto, METH_VARARGS, readinto_doc},
|
---|
851 | {"readline", (PyCFunction)bytesio_readline, METH_VARARGS, readline_doc},
|
---|
852 | {"readlines", (PyCFunction)bytesio_readlines, METH_VARARGS, readlines_doc},
|
---|
853 | {"read", (PyCFunction)bytesio_read, METH_VARARGS, read_doc},
|
---|
854 | {"getvalue", (PyCFunction)bytesio_getvalue, METH_NOARGS, getval_doc},
|
---|
855 | {"seek", (PyCFunction)bytesio_seek, METH_VARARGS, seek_doc},
|
---|
856 | {"truncate", (PyCFunction)bytesio_truncate, METH_VARARGS, truncate_doc},
|
---|
857 | {"__getstate__", (PyCFunction)bytesio_getstate, METH_NOARGS, NULL},
|
---|
858 | {"__setstate__", (PyCFunction)bytesio_setstate, METH_O, NULL},
|
---|
859 | {"__sizeof__", (PyCFunction)bytesio_sizeof, METH_NOARGS, NULL},
|
---|
860 | {NULL, NULL} /* sentinel */
|
---|
861 | };
|
---|
862 |
|
---|
863 | PyDoc_STRVAR(bytesio_doc,
|
---|
864 | "BytesIO([buffer]) -> object\n"
|
---|
865 | "\n"
|
---|
866 | "Create a buffered I/O implementation using an in-memory bytes\n"
|
---|
867 | "buffer, ready for reading and writing.");
|
---|
868 |
|
---|
869 | PyTypeObject PyBytesIO_Type = {
|
---|
870 | PyVarObject_HEAD_INIT(NULL, 0)
|
---|
871 | "_io.BytesIO", /*tp_name*/
|
---|
872 | sizeof(bytesio), /*tp_basicsize*/
|
---|
873 | 0, /*tp_itemsize*/
|
---|
874 | (destructor)bytesio_dealloc, /*tp_dealloc*/
|
---|
875 | 0, /*tp_print*/
|
---|
876 | 0, /*tp_getattr*/
|
---|
877 | 0, /*tp_setattr*/
|
---|
878 | 0, /*tp_reserved*/
|
---|
879 | 0, /*tp_repr*/
|
---|
880 | 0, /*tp_as_number*/
|
---|
881 | 0, /*tp_as_sequence*/
|
---|
882 | 0, /*tp_as_mapping*/
|
---|
883 | 0, /*tp_hash*/
|
---|
884 | 0, /*tp_call*/
|
---|
885 | 0, /*tp_str*/
|
---|
886 | 0, /*tp_getattro*/
|
---|
887 | 0, /*tp_setattro*/
|
---|
888 | 0, /*tp_as_buffer*/
|
---|
889 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE |
|
---|
890 | Py_TPFLAGS_HAVE_GC, /*tp_flags*/
|
---|
891 | bytesio_doc, /*tp_doc*/
|
---|
892 | (traverseproc)bytesio_traverse, /*tp_traverse*/
|
---|
893 | (inquiry)bytesio_clear, /*tp_clear*/
|
---|
894 | 0, /*tp_richcompare*/
|
---|
895 | offsetof(bytesio, weakreflist), /*tp_weaklistoffset*/
|
---|
896 | PyObject_SelfIter, /*tp_iter*/
|
---|
897 | (iternextfunc)bytesio_iternext, /*tp_iternext*/
|
---|
898 | bytesio_methods, /*tp_methods*/
|
---|
899 | 0, /*tp_members*/
|
---|
900 | bytesio_getsetlist, /*tp_getset*/
|
---|
901 | 0, /*tp_base*/
|
---|
902 | 0, /*tp_dict*/
|
---|
903 | 0, /*tp_descr_get*/
|
---|
904 | 0, /*tp_descr_set*/
|
---|
905 | offsetof(bytesio, dict), /*tp_dictoffset*/
|
---|
906 | (initproc)bytesio_init, /*tp_init*/
|
---|
907 | 0, /*tp_alloc*/
|
---|
908 | bytesio_new, /*tp_new*/
|
---|
909 | };
|
---|