blob: 8987ce5a0e8c130da225d00d98f6558dd3aa3796 [file] [log] [blame]
Victor Stinner4e314432010-10-07 21:45:39 +00001#include "Python.h"
Stefan Krah6df5cae2012-11-12 20:14:36 +01002#include "osdefs.h"
Stefan Krah6c01e382014-01-20 15:31:08 +01003#include <locale.h>
4
Victor Stinnerb306d752010-10-07 22:09:40 +00005#ifdef MS_WINDOWS
Steve Dowerd81431f2015-03-06 14:47:02 -08006# include <malloc.h>
Victor Stinnerb306d752010-10-07 22:09:40 +00007# include <windows.h>
Steve Dower8fc89802015-04-12 00:26:27 -04008extern int winerror_to_errno(int);
Victor Stinnerb306d752010-10-07 22:09:40 +00009#endif
Victor Stinner4e314432010-10-07 21:45:39 +000010
Brett Cannonefb00c02012-02-29 18:31:31 -050011#ifdef HAVE_LANGINFO_H
12#include <langinfo.h>
13#endif
14
Victor Stinnerdaf45552013-08-28 00:53:59 +020015#ifdef HAVE_SYS_IOCTL_H
16#include <sys/ioctl.h>
17#endif
18
19#ifdef HAVE_FCNTL_H
20#include <fcntl.h>
21#endif /* HAVE_FCNTL_H */
22
Victor Stinnere2623772012-11-12 23:04:02 +010023#ifdef __APPLE__
24extern wchar_t* _Py_DecodeUTF8_surrogateescape(const char *s, Py_ssize_t size);
25#endif
26
Victor Stinnerdaf45552013-08-28 00:53:59 +020027#ifdef O_CLOEXEC
Victor Stinnerb034eee2013-09-07 10:36:04 +020028/* Does open() support the O_CLOEXEC flag? Possible values:
Victor Stinnerdaf45552013-08-28 00:53:59 +020029
30 -1: unknown
31 0: open() ignores O_CLOEXEC flag, ex: Linux kernel older than 2.6.23
32 1: open() supports O_CLOEXEC flag, close-on-exec is set
33
Victor Stinnera555cfc2015-03-18 00:22:14 +010034 The flag is used by _Py_open(), _Py_open_noraise(), io.FileIO
35 and os.open(). */
Victor Stinnerdaf45552013-08-28 00:53:59 +020036int _Py_open_cloexec_works = -1;
37#endif
38
Brett Cannonefb00c02012-02-29 18:31:31 -050039PyObject *
40_Py_device_encoding(int fd)
41{
Victor Stinner14b9b112013-06-25 00:37:25 +020042#if defined(MS_WINDOWS)
Brett Cannonefb00c02012-02-29 18:31:31 -050043 UINT cp;
44#endif
Steve Dower8fc89802015-04-12 00:26:27 -040045 int valid;
46 _Py_BEGIN_SUPPRESS_IPH
47 valid = _PyVerify_fd(fd) && isatty(fd);
48 _Py_END_SUPPRESS_IPH
49 if (!valid)
Brett Cannonefb00c02012-02-29 18:31:31 -050050 Py_RETURN_NONE;
Steve Dower8fc89802015-04-12 00:26:27 -040051
Victor Stinner14b9b112013-06-25 00:37:25 +020052#if defined(MS_WINDOWS)
Brett Cannonefb00c02012-02-29 18:31:31 -050053 if (fd == 0)
54 cp = GetConsoleCP();
55 else if (fd == 1 || fd == 2)
56 cp = GetConsoleOutputCP();
57 else
58 cp = 0;
59 /* GetConsoleCP() and GetConsoleOutputCP() return 0 if the application
60 has no console */
61 if (cp != 0)
62 return PyUnicode_FromFormat("cp%u", (unsigned int)cp);
63#elif defined(CODESET)
64 {
65 char *codeset = nl_langinfo(CODESET);
66 if (codeset != NULL && codeset[0] != 0)
67 return PyUnicode_FromString(codeset);
68 }
69#endif
70 Py_RETURN_NONE;
71}
72
Victor Stinnerd45c7f82012-12-04 01:34:47 +010073#if !defined(__APPLE__) && !defined(MS_WINDOWS)
74extern int _Py_normalize_encoding(const char *, char *, size_t);
75
76/* Workaround FreeBSD and OpenIndiana locale encoding issue with the C locale.
77 On these operating systems, nl_langinfo(CODESET) announces an alias of the
78 ASCII encoding, whereas mbstowcs() and wcstombs() functions use the
79 ISO-8859-1 encoding. The problem is that os.fsencode() and os.fsdecode() use
80 locale.getpreferredencoding() codec. For example, if command line arguments
81 are decoded by mbstowcs() and encoded back by os.fsencode(), we get a
82 UnicodeEncodeError instead of retrieving the original byte string.
83
84 The workaround is enabled if setlocale(LC_CTYPE, NULL) returns "C",
85 nl_langinfo(CODESET) announces "ascii" (or an alias to ASCII), and at least
86 one byte in range 0x80-0xff can be decoded from the locale encoding. The
87 workaround is also enabled on error, for example if getting the locale
88 failed.
89
Philip Jenvey215c49a2013-01-15 13:24:12 -080090 Values of force_ascii:
Victor Stinnerd45c7f82012-12-04 01:34:47 +010091
Victor Stinnerf6a271a2014-08-01 12:28:48 +020092 1: the workaround is used: Py_EncodeLocale() uses
93 encode_ascii_surrogateescape() and Py_DecodeLocale() uses
Victor Stinnerd45c7f82012-12-04 01:34:47 +010094 decode_ascii_surrogateescape()
Victor Stinnerf6a271a2014-08-01 12:28:48 +020095 0: the workaround is not used: Py_EncodeLocale() uses wcstombs() and
96 Py_DecodeLocale() uses mbstowcs()
Victor Stinnerd45c7f82012-12-04 01:34:47 +010097 -1: unknown, need to call check_force_ascii() to get the value
98*/
99static int force_ascii = -1;
100
101static int
102check_force_ascii(void)
103{
104 char *loc;
105#if defined(HAVE_LANGINFO_H) && defined(CODESET)
106 char *codeset, **alias;
107 char encoding[100];
108 int is_ascii;
109 unsigned int i;
110 char* ascii_aliases[] = {
111 "ascii",
112 "646",
113 "ansi-x3.4-1968",
114 "ansi-x3-4-1968",
115 "ansi-x3.4-1986",
116 "cp367",
117 "csascii",
118 "ibm367",
119 "iso646-us",
120 "iso-646.irv-1991",
121 "iso-ir-6",
122 "us",
123 "us-ascii",
124 NULL
125 };
126#endif
127
128 loc = setlocale(LC_CTYPE, NULL);
129 if (loc == NULL)
130 goto error;
131 if (strcmp(loc, "C") != 0) {
132 /* the LC_CTYPE locale is different than C */
133 return 0;
134 }
135
136#if defined(HAVE_LANGINFO_H) && defined(CODESET)
137 codeset = nl_langinfo(CODESET);
138 if (!codeset || codeset[0] == '\0') {
139 /* CODESET is not set or empty */
140 goto error;
141 }
142 if (!_Py_normalize_encoding(codeset, encoding, sizeof(encoding)))
143 goto error;
144
145 is_ascii = 0;
146 for (alias=ascii_aliases; *alias != NULL; alias++) {
147 if (strcmp(encoding, *alias) == 0) {
148 is_ascii = 1;
149 break;
150 }
151 }
152 if (!is_ascii) {
153 /* nl_langinfo(CODESET) is not "ascii" or an alias of ASCII */
154 return 0;
155 }
156
157 for (i=0x80; i<0xff; i++) {
158 unsigned char ch;
159 wchar_t wch;
160 size_t res;
161
162 ch = (unsigned char)i;
163 res = mbstowcs(&wch, (char*)&ch, 1);
164 if (res != (size_t)-1) {
165 /* decoding a non-ASCII character from the locale encoding succeed:
166 the locale encoding is not ASCII, force ASCII */
167 return 1;
168 }
169 }
170 /* None of the bytes in the range 0x80-0xff can be decoded from the locale
171 encoding: the locale encoding is really ASCII */
172 return 0;
173#else
174 /* nl_langinfo(CODESET) is not available: always force ASCII */
175 return 1;
176#endif
177
178error:
179 /* if an error occured, force the ASCII encoding */
180 return 1;
181}
182
183static char*
184encode_ascii_surrogateescape(const wchar_t *text, size_t *error_pos)
185{
186 char *result = NULL, *out;
187 size_t len, i;
188 wchar_t ch;
189
190 if (error_pos != NULL)
191 *error_pos = (size_t)-1;
192
193 len = wcslen(text);
194
195 result = PyMem_Malloc(len + 1); /* +1 for NUL byte */
196 if (result == NULL)
197 return NULL;
198
199 out = result;
200 for (i=0; i<len; i++) {
201 ch = text[i];
202
203 if (ch <= 0x7f) {
204 /* ASCII character */
205 *out++ = (char)ch;
206 }
207 else if (0xdc80 <= ch && ch <= 0xdcff) {
208 /* UTF-8b surrogate */
209 *out++ = (char)(ch - 0xdc00);
210 }
211 else {
212 if (error_pos != NULL)
213 *error_pos = i;
214 PyMem_Free(result);
215 return NULL;
216 }
217 }
218 *out = '\0';
219 return result;
220}
221#endif /* !defined(__APPLE__) && !defined(MS_WINDOWS) */
222
223#if !defined(__APPLE__) && (!defined(MS_WINDOWS) || !defined(HAVE_MBRTOWC))
224static wchar_t*
225decode_ascii_surrogateescape(const char *arg, size_t *size)
226{
227 wchar_t *res;
228 unsigned char *in;
229 wchar_t *out;
Benjamin Petersonf18bf6f2015-01-04 16:03:17 -0600230 size_t argsize = strlen(arg) + 1;
Victor Stinnerd45c7f82012-12-04 01:34:47 +0100231
Benjamin Petersonf18bf6f2015-01-04 16:03:17 -0600232 if (argsize > PY_SSIZE_T_MAX/sizeof(wchar_t))
233 return NULL;
Benjamin Peterson10ecaa22015-01-04 16:05:39 -0600234 res = PyMem_RawMalloc(argsize*sizeof(wchar_t));
Victor Stinnerd45c7f82012-12-04 01:34:47 +0100235 if (!res)
236 return NULL;
237
238 in = (unsigned char*)arg;
239 out = res;
240 while(*in)
241 if(*in < 128)
242 *out++ = *in++;
243 else
244 *out++ = 0xdc00 + *in++;
245 *out = 0;
246 if (size != NULL)
247 *size = out - res;
248 return res;
249}
250#endif
251
Victor Stinner4e314432010-10-07 21:45:39 +0000252
253/* Decode a byte string from the locale encoding with the
Victor Stinnerf6a271a2014-08-01 12:28:48 +0200254 surrogateescape error handler: undecodable bytes are decoded as characters
255 in range U+DC80..U+DCFF. If a byte sequence can be decoded as a surrogate
Victor Stinner4e314432010-10-07 21:45:39 +0000256 character, escape the bytes using the surrogateescape error handler instead
257 of decoding them.
258
Victor Stinnerf6a271a2014-08-01 12:28:48 +0200259 Return a pointer to a newly allocated wide character string, use
260 PyMem_RawFree() to free the memory. If size is not NULL, write the number of
261 wide characters excluding the null character into *size
Victor Stinner4e314432010-10-07 21:45:39 +0000262
Victor Stinnerf6a271a2014-08-01 12:28:48 +0200263 Return NULL on decoding error or memory allocation error. If *size* is not
264 NULL, *size is set to (size_t)-1 on memory error or set to (size_t)-2 on
265 decoding error.
Victor Stinner19de4c32010-11-08 23:30:46 +0000266
Victor Stinnerf6a271a2014-08-01 12:28:48 +0200267 Decoding errors should never happen, unless there is a bug in the C
268 library.
269
270 Use the Py_EncodeLocale() function to encode the character string back to a
271 byte string. */
Victor Stinner4e314432010-10-07 21:45:39 +0000272wchar_t*
Victor Stinnerf6a271a2014-08-01 12:28:48 +0200273Py_DecodeLocale(const char* arg, size_t *size)
Victor Stinner4e314432010-10-07 21:45:39 +0000274{
Victor Stinnere2623772012-11-12 23:04:02 +0100275#ifdef __APPLE__
276 wchar_t *wstr;
277 wstr = _Py_DecodeUTF8_surrogateescape(arg, strlen(arg));
Victor Stinner0d92c4f2012-11-12 23:32:21 +0100278 if (size != NULL) {
279 if (wstr != NULL)
280 *size = wcslen(wstr);
281 else
282 *size = (size_t)-1;
283 }
Victor Stinnere2623772012-11-12 23:04:02 +0100284 return wstr;
285#else
Victor Stinner4e314432010-10-07 21:45:39 +0000286 wchar_t *res;
Victor Stinnerd45c7f82012-12-04 01:34:47 +0100287 size_t argsize;
Victor Stinner4e314432010-10-07 21:45:39 +0000288 size_t count;
Victor Stinner313f10c2013-05-07 23:48:56 +0200289#ifdef HAVE_MBRTOWC
Victor Stinner4e314432010-10-07 21:45:39 +0000290 unsigned char *in;
291 wchar_t *out;
Victor Stinner4e314432010-10-07 21:45:39 +0000292 mbstate_t mbs;
293#endif
Victor Stinnerd45c7f82012-12-04 01:34:47 +0100294
295#ifndef MS_WINDOWS
296 if (force_ascii == -1)
297 force_ascii = check_force_ascii();
298
299 if (force_ascii) {
300 /* force ASCII encoding to workaround mbstowcs() issue */
301 res = decode_ascii_surrogateescape(arg, size);
302 if (res == NULL)
303 goto oom;
304 return res;
305 }
306#endif
307
308#ifdef HAVE_BROKEN_MBSTOWCS
309 /* Some platforms have a broken implementation of
310 * mbstowcs which does not count the characters that
311 * would result from conversion. Use an upper bound.
312 */
313 argsize = strlen(arg);
314#else
315 argsize = mbstowcs(NULL, arg, 0);
316#endif
Victor Stinner4e314432010-10-07 21:45:39 +0000317 if (argsize != (size_t)-1) {
Benjamin Petersonf18bf6f2015-01-04 16:03:17 -0600318 if (argsize == PY_SSIZE_T_MAX)
319 goto oom;
320 argsize += 1;
321 if (argsize > PY_SSIZE_T_MAX/sizeof(wchar_t))
322 goto oom;
Benjamin Peterson10ecaa22015-01-04 16:05:39 -0600323 res = (wchar_t *)PyMem_RawMalloc(argsize*sizeof(wchar_t));
Victor Stinner4e314432010-10-07 21:45:39 +0000324 if (!res)
325 goto oom;
Benjamin Petersonf18bf6f2015-01-04 16:03:17 -0600326 count = mbstowcs(res, arg, argsize);
Victor Stinner4e314432010-10-07 21:45:39 +0000327 if (count != (size_t)-1) {
328 wchar_t *tmp;
329 /* Only use the result if it contains no
330 surrogate characters. */
331 for (tmp = res; *tmp != 0 &&
Victor Stinner76df43d2012-10-30 01:42:39 +0100332 !Py_UNICODE_IS_SURROGATE(*tmp); tmp++)
Victor Stinner4e314432010-10-07 21:45:39 +0000333 ;
Victor Stinner168e1172010-10-16 23:16:16 +0000334 if (*tmp == 0) {
335 if (size != NULL)
336 *size = count;
Victor Stinner4e314432010-10-07 21:45:39 +0000337 return res;
Victor Stinner168e1172010-10-16 23:16:16 +0000338 }
Victor Stinner4e314432010-10-07 21:45:39 +0000339 }
Victor Stinner1a7425f2013-07-07 16:25:15 +0200340 PyMem_RawFree(res);
Victor Stinner4e314432010-10-07 21:45:39 +0000341 }
342 /* Conversion failed. Fall back to escaping with surrogateescape. */
343#ifdef HAVE_MBRTOWC
344 /* Try conversion with mbrtwoc (C99), and escape non-decodable bytes. */
345
346 /* Overallocate; as multi-byte characters are in the argument, the
347 actual output could use less memory. */
348 argsize = strlen(arg) + 1;
Benjamin Petersonf18bf6f2015-01-04 16:03:17 -0600349 if (argsize > PY_SSIZE_T_MAX/sizeof(wchar_t))
350 goto oom;
Victor Stinner1a7425f2013-07-07 16:25:15 +0200351 res = (wchar_t*)PyMem_RawMalloc(argsize*sizeof(wchar_t));
Victor Stinner19de4c32010-11-08 23:30:46 +0000352 if (!res)
353 goto oom;
Victor Stinner4e314432010-10-07 21:45:39 +0000354 in = (unsigned char*)arg;
355 out = res;
356 memset(&mbs, 0, sizeof mbs);
357 while (argsize) {
358 size_t converted = mbrtowc(out, (char*)in, argsize, &mbs);
359 if (converted == 0)
360 /* Reached end of string; null char stored. */
361 break;
362 if (converted == (size_t)-2) {
363 /* Incomplete character. This should never happen,
364 since we provide everything that we have -
365 unless there is a bug in the C library, or I
366 misunderstood how mbrtowc works. */
Victor Stinner1a7425f2013-07-07 16:25:15 +0200367 PyMem_RawFree(res);
Victor Stinneraf02e1c2011-12-16 23:56:01 +0100368 if (size != NULL)
369 *size = (size_t)-2;
Victor Stinner4e314432010-10-07 21:45:39 +0000370 return NULL;
371 }
372 if (converted == (size_t)-1) {
373 /* Conversion error. Escape as UTF-8b, and start over
374 in the initial shift state. */
375 *out++ = 0xdc00 + *in++;
376 argsize--;
377 memset(&mbs, 0, sizeof mbs);
378 continue;
379 }
Victor Stinner76df43d2012-10-30 01:42:39 +0100380 if (Py_UNICODE_IS_SURROGATE(*out)) {
Victor Stinner4e314432010-10-07 21:45:39 +0000381 /* Surrogate character. Escape the original
382 byte sequence with surrogateescape. */
383 argsize -= converted;
384 while (converted--)
385 *out++ = 0xdc00 + *in++;
386 continue;
387 }
388 /* successfully converted some bytes */
389 in += converted;
390 argsize -= converted;
391 out++;
392 }
Victor Stinnerd45c7f82012-12-04 01:34:47 +0100393 if (size != NULL)
394 *size = out - res;
Victor Stinnere2623772012-11-12 23:04:02 +0100395#else /* HAVE_MBRTOWC */
Victor Stinner4e314432010-10-07 21:45:39 +0000396 /* Cannot use C locale for escaping; manually escape as if charset
397 is ASCII (i.e. escape all bytes > 128. This will still roundtrip
398 correctly in the locale's charset, which must be an ASCII superset. */
Victor Stinnerd45c7f82012-12-04 01:34:47 +0100399 res = decode_ascii_surrogateescape(arg, size);
400 if (res == NULL)
Victor Stinneraf02e1c2011-12-16 23:56:01 +0100401 goto oom;
Victor Stinnere2623772012-11-12 23:04:02 +0100402#endif /* HAVE_MBRTOWC */
Victor Stinner4e314432010-10-07 21:45:39 +0000403 return res;
404oom:
Victor Stinneraf02e1c2011-12-16 23:56:01 +0100405 if (size != NULL)
406 *size = (size_t)-1;
Victor Stinner4e314432010-10-07 21:45:39 +0000407 return NULL;
Victor Stinnere2623772012-11-12 23:04:02 +0100408#endif /* __APPLE__ */
Victor Stinner4e314432010-10-07 21:45:39 +0000409}
410
Victor Stinnerf6a271a2014-08-01 12:28:48 +0200411/* Encode a wide character string to the locale encoding with the
412 surrogateescape error handler: surrogate characters in the range
413 U+DC80..U+DCFF are converted to bytes 0x80..0xFF.
Victor Stinner4e314432010-10-07 21:45:39 +0000414
Victor Stinnerf6a271a2014-08-01 12:28:48 +0200415 Return a pointer to a newly allocated byte string, use PyMem_Free() to free
416 the memory. Return NULL on encoding or memory allocation error.
Victor Stinner4e314432010-10-07 21:45:39 +0000417
Victor Stinnerf6a271a2014-08-01 12:28:48 +0200418 If error_pos is not NULL, *error_pos is set to the index of the invalid
419 character on encoding error, or set to (size_t)-1 otherwise.
Victor Stinner2f02a512010-11-08 22:43:46 +0000420
Victor Stinnerf6a271a2014-08-01 12:28:48 +0200421 Use the Py_DecodeLocale() function to decode the bytes string back to a wide
422 character string. */
Victor Stinner4e314432010-10-07 21:45:39 +0000423char*
Victor Stinnerf6a271a2014-08-01 12:28:48 +0200424Py_EncodeLocale(const wchar_t *text, size_t *error_pos)
Victor Stinner4e314432010-10-07 21:45:39 +0000425{
Victor Stinnere2623772012-11-12 23:04:02 +0100426#ifdef __APPLE__
427 Py_ssize_t len;
428 PyObject *unicode, *bytes = NULL;
429 char *cpath;
430
431 unicode = PyUnicode_FromWideChar(text, wcslen(text));
Victor Stinner0d92c4f2012-11-12 23:32:21 +0100432 if (unicode == NULL)
Victor Stinnere2623772012-11-12 23:04:02 +0100433 return NULL;
Victor Stinnere2623772012-11-12 23:04:02 +0100434
435 bytes = _PyUnicode_AsUTF8String(unicode, "surrogateescape");
436 Py_DECREF(unicode);
437 if (bytes == NULL) {
438 PyErr_Clear();
Victor Stinner0d92c4f2012-11-12 23:32:21 +0100439 if (error_pos != NULL)
440 *error_pos = (size_t)-1;
Victor Stinnere2623772012-11-12 23:04:02 +0100441 return NULL;
442 }
443
444 len = PyBytes_GET_SIZE(bytes);
445 cpath = PyMem_Malloc(len+1);
446 if (cpath == NULL) {
Victor Stinner0d92c4f2012-11-12 23:32:21 +0100447 PyErr_Clear();
Victor Stinnere2623772012-11-12 23:04:02 +0100448 Py_DECREF(bytes);
Victor Stinner0d92c4f2012-11-12 23:32:21 +0100449 if (error_pos != NULL)
450 *error_pos = (size_t)-1;
Victor Stinnere2623772012-11-12 23:04:02 +0100451 return NULL;
452 }
453 memcpy(cpath, PyBytes_AsString(bytes), len + 1);
454 Py_DECREF(bytes);
455 return cpath;
456#else /* __APPLE__ */
Victor Stinner4e314432010-10-07 21:45:39 +0000457 const size_t len = wcslen(text);
458 char *result = NULL, *bytes = NULL;
459 size_t i, size, converted;
460 wchar_t c, buf[2];
461
Victor Stinnerd45c7f82012-12-04 01:34:47 +0100462#ifndef MS_WINDOWS
463 if (force_ascii == -1)
464 force_ascii = check_force_ascii();
465
466 if (force_ascii)
467 return encode_ascii_surrogateescape(text, error_pos);
468#endif
469
Victor Stinner4e314432010-10-07 21:45:39 +0000470 /* The function works in two steps:
471 1. compute the length of the output buffer in bytes (size)
472 2. outputs the bytes */
473 size = 0;
474 buf[1] = 0;
475 while (1) {
476 for (i=0; i < len; i++) {
477 c = text[i];
478 if (c >= 0xdc80 && c <= 0xdcff) {
479 /* UTF-8b surrogate */
480 if (bytes != NULL) {
481 *bytes++ = c - 0xdc00;
482 size--;
483 }
484 else
485 size++;
486 continue;
487 }
488 else {
489 buf[0] = c;
490 if (bytes != NULL)
491 converted = wcstombs(bytes, buf, size);
492 else
493 converted = wcstombs(NULL, buf, 0);
494 if (converted == (size_t)-1) {
495 if (result != NULL)
496 PyMem_Free(result);
Victor Stinner2f02a512010-11-08 22:43:46 +0000497 if (error_pos != NULL)
498 *error_pos = i;
Victor Stinner4e314432010-10-07 21:45:39 +0000499 return NULL;
500 }
501 if (bytes != NULL) {
502 bytes += converted;
503 size -= converted;
504 }
505 else
506 size += converted;
507 }
508 }
509 if (result != NULL) {
Victor Stinnerd45c7f82012-12-04 01:34:47 +0100510 *bytes = '\0';
Victor Stinner4e314432010-10-07 21:45:39 +0000511 break;
512 }
513
514 size += 1; /* nul byte at the end */
515 result = PyMem_Malloc(size);
Victor Stinner0d92c4f2012-11-12 23:32:21 +0100516 if (result == NULL) {
517 if (error_pos != NULL)
518 *error_pos = (size_t)-1;
Victor Stinner4e314432010-10-07 21:45:39 +0000519 return NULL;
Victor Stinner0d92c4f2012-11-12 23:32:21 +0100520 }
Victor Stinner4e314432010-10-07 21:45:39 +0000521 bytes = result;
522 }
523 return result;
Victor Stinnere2623772012-11-12 23:04:02 +0100524#endif /* __APPLE__ */
Victor Stinner4e314432010-10-07 21:45:39 +0000525}
526
Victor Stinner6672d0c2010-10-07 22:53:43 +0000527
Steve Dowerf2f373f2015-02-21 08:44:05 -0800528#ifdef MS_WINDOWS
529static __int64 secs_between_epochs = 11644473600; /* Seconds between 1.1.1601 and 1.1.1970 */
530
531static void
532FILE_TIME_to_time_t_nsec(FILETIME *in_ptr, time_t *time_out, int* nsec_out)
533{
534 /* XXX endianness. Shouldn't matter, as all Windows implementations are little-endian */
535 /* Cannot simply cast and dereference in_ptr,
536 since it might not be aligned properly */
537 __int64 in;
538 memcpy(&in, in_ptr, sizeof(in));
539 *nsec_out = (int)(in % 10000000) * 100; /* FILETIME is in units of 100 nsec. */
540 *time_out = Py_SAFE_DOWNCAST((in / 10000000) - secs_between_epochs, __int64, time_t);
541}
542
543void
Steve Dowerbf1f3762015-02-21 15:26:02 -0800544_Py_time_t_to_FILE_TIME(time_t time_in, int nsec_in, FILETIME *out_ptr)
Steve Dowerf2f373f2015-02-21 08:44:05 -0800545{
546 /* XXX endianness */
547 __int64 out;
548 out = time_in + secs_between_epochs;
549 out = out * 10000000 + nsec_in / 100;
550 memcpy(out_ptr, &out, sizeof(out));
551}
552
553/* Below, we *know* that ugo+r is 0444 */
554#if _S_IREAD != 0400
555#error Unsupported C library
556#endif
557static int
558attributes_to_mode(DWORD attr)
559{
560 int m = 0;
561 if (attr & FILE_ATTRIBUTE_DIRECTORY)
562 m |= _S_IFDIR | 0111; /* IFEXEC for user,group,other */
563 else
564 m |= _S_IFREG;
565 if (attr & FILE_ATTRIBUTE_READONLY)
566 m |= 0444;
567 else
568 m |= 0666;
569 return m;
570}
571
Steve Dowerbf1f3762015-02-21 15:26:02 -0800572void
Victor Stinnere134a7f2015-03-30 10:09:31 +0200573_Py_attribute_data_to_stat(BY_HANDLE_FILE_INFORMATION *info, ULONG reparse_tag,
574 struct _Py_stat_struct *result)
Steve Dowerf2f373f2015-02-21 08:44:05 -0800575{
576 memset(result, 0, sizeof(*result));
577 result->st_mode = attributes_to_mode(info->dwFileAttributes);
578 result->st_size = (((__int64)info->nFileSizeHigh)<<32) + info->nFileSizeLow;
579 result->st_dev = info->dwVolumeSerialNumber;
580 result->st_rdev = result->st_dev;
581 FILE_TIME_to_time_t_nsec(&info->ftCreationTime, &result->st_ctime, &result->st_ctime_nsec);
582 FILE_TIME_to_time_t_nsec(&info->ftLastWriteTime, &result->st_mtime, &result->st_mtime_nsec);
583 FILE_TIME_to_time_t_nsec(&info->ftLastAccessTime, &result->st_atime, &result->st_atime_nsec);
584 result->st_nlink = info->nNumberOfLinks;
585 result->st_ino = (((__int64)info->nFileIndexHigh)<<32) + info->nFileIndexLow;
586 if (reparse_tag == IO_REPARSE_TAG_SYMLINK) {
587 /* first clear the S_IFMT bits */
588 result->st_mode ^= (result->st_mode & S_IFMT);
589 /* now set the bits that make this a symlink */
590 result->st_mode |= S_IFLNK;
591 }
592 result->st_file_attributes = info->dwFileAttributes;
Steve Dowerf2f373f2015-02-21 08:44:05 -0800593}
594#endif
595
596/* Return information about a file.
597
598 On POSIX, use fstat().
599
600 On Windows, use GetFileType() and GetFileInformationByHandle() which support
601 files larger than 2 GB. fstat() may fail with EOVERFLOW on files larger
Martin Panter6f9b0102015-12-17 10:18:28 +0000602 than 2 GB because the file size type is a signed 32-bit integer: see issue
Steve Dowerf2f373f2015-02-21 08:44:05 -0800603 #23152.
Victor Stinnere134a7f2015-03-30 10:09:31 +0200604
605 On Windows, set the last Windows error and return nonzero on error. On
606 POSIX, set errno and return nonzero on error. Fill status and return 0 on
607 success. */
Steve Dowerf2f373f2015-02-21 08:44:05 -0800608int
Victor Stinnere134a7f2015-03-30 10:09:31 +0200609_Py_fstat_noraise(int fd, struct _Py_stat_struct *status)
Steve Dowerf2f373f2015-02-21 08:44:05 -0800610{
611#ifdef MS_WINDOWS
612 BY_HANDLE_FILE_INFORMATION info;
613 HANDLE h;
614 int type;
615
616 if (!_PyVerify_fd(fd))
617 h = INVALID_HANDLE_VALUE;
Steve Dower8fc89802015-04-12 00:26:27 -0400618 else {
619 _Py_BEGIN_SUPPRESS_IPH
Steve Dowerf2f373f2015-02-21 08:44:05 -0800620 h = (HANDLE)_get_osfhandle(fd);
Steve Dower8fc89802015-04-12 00:26:27 -0400621 _Py_END_SUPPRESS_IPH
622 }
Steve Dowerf2f373f2015-02-21 08:44:05 -0800623
624 if (h == INVALID_HANDLE_VALUE) {
Steve Dower8fc89802015-04-12 00:26:27 -0400625 /* errno is already set by _get_osfhandle, but we also set
626 the Win32 error for callers who expect that */
Steve Dower8acde7d2015-03-07 18:14:07 -0800627 SetLastError(ERROR_INVALID_HANDLE);
Steve Dowerf2f373f2015-02-21 08:44:05 -0800628 return -1;
629 }
Victor Stinnere134a7f2015-03-30 10:09:31 +0200630 memset(status, 0, sizeof(*status));
Steve Dowerf2f373f2015-02-21 08:44:05 -0800631
632 type = GetFileType(h);
633 if (type == FILE_TYPE_UNKNOWN) {
634 DWORD error = GetLastError();
Steve Dower8fc89802015-04-12 00:26:27 -0400635 if (error != 0) {
636 errno = winerror_to_errno(error);
Steve Dowerf2f373f2015-02-21 08:44:05 -0800637 return -1;
Steve Dower8fc89802015-04-12 00:26:27 -0400638 }
Steve Dowerf2f373f2015-02-21 08:44:05 -0800639 /* else: valid but unknown file */
640 }
641
642 if (type != FILE_TYPE_DISK) {
643 if (type == FILE_TYPE_CHAR)
Victor Stinnere134a7f2015-03-30 10:09:31 +0200644 status->st_mode = _S_IFCHR;
Steve Dowerf2f373f2015-02-21 08:44:05 -0800645 else if (type == FILE_TYPE_PIPE)
Victor Stinnere134a7f2015-03-30 10:09:31 +0200646 status->st_mode = _S_IFIFO;
Steve Dowerf2f373f2015-02-21 08:44:05 -0800647 return 0;
648 }
649
650 if (!GetFileInformationByHandle(h, &info)) {
Steve Dower8fc89802015-04-12 00:26:27 -0400651 /* The Win32 error is already set, but we also set errno for
652 callers who expect it */
653 errno = winerror_to_errno(GetLastError());
Steve Dowerf2f373f2015-02-21 08:44:05 -0800654 return -1;
655 }
656
Victor Stinnere134a7f2015-03-30 10:09:31 +0200657 _Py_attribute_data_to_stat(&info, 0, status);
Steve Dowerf2f373f2015-02-21 08:44:05 -0800658 /* specific to fstat() */
Victor Stinnere134a7f2015-03-30 10:09:31 +0200659 status->st_ino = (((__int64)info.nFileIndexHigh)<<32) + info.nFileIndexLow;
Steve Dowerf2f373f2015-02-21 08:44:05 -0800660 return 0;
661#else
Victor Stinnere134a7f2015-03-30 10:09:31 +0200662 return fstat(fd, status);
Steve Dowerf2f373f2015-02-21 08:44:05 -0800663#endif
664}
Steve Dowerf2f373f2015-02-21 08:44:05 -0800665
Victor Stinnere134a7f2015-03-30 10:09:31 +0200666/* Return information about a file.
667
668 On POSIX, use fstat().
669
670 On Windows, use GetFileType() and GetFileInformationByHandle() which support
671 files larger than 2 GB. fstat() may fail with EOVERFLOW on files larger
Martin Panter6f9b0102015-12-17 10:18:28 +0000672 than 2 GB because the file size type is a signed 32-bit integer: see issue
Victor Stinnere134a7f2015-03-30 10:09:31 +0200673 #23152.
674
675 Raise an exception and return -1 on error. On Windows, set the last Windows
676 error on error. On POSIX, set errno on error. Fill status and return 0 on
677 success.
678
Victor Stinner6f4fae82015-04-01 18:34:32 +0200679 Release the GIL to call GetFileType() and GetFileInformationByHandle(), or
680 to call fstat(). The caller must hold the GIL. */
Victor Stinnere134a7f2015-03-30 10:09:31 +0200681int
682_Py_fstat(int fd, struct _Py_stat_struct *status)
683{
684 int res;
685
686 Py_BEGIN_ALLOW_THREADS
687 res = _Py_fstat_noraise(fd, status);
688 Py_END_ALLOW_THREADS
689
690 if (res != 0) {
691#ifdef MS_WINDOWS
692 PyErr_SetFromWindowsErr(0);
693#else
694 PyErr_SetFromErrno(PyExc_OSError);
695#endif
696 return -1;
697 }
698 return 0;
699}
Steve Dowerf2f373f2015-02-21 08:44:05 -0800700
Victor Stinner6672d0c2010-10-07 22:53:43 +0000701/* Call _wstat() on Windows, or encode the path to the filesystem encoding and
702 call stat() otherwise. Only fill st_mode attribute on Windows.
703
Victor Stinnerbd0850b2011-12-18 20:47:30 +0100704 Return 0 on success, -1 on _wstat() / stat() error, -2 if an exception was
705 raised. */
Victor Stinner4e314432010-10-07 21:45:39 +0000706
707int
Victor Stinnera4a75952010-10-07 22:23:10 +0000708_Py_stat(PyObject *path, struct stat *statbuf)
Victor Stinner4e314432010-10-07 21:45:39 +0000709{
710#ifdef MS_WINDOWS
Victor Stinner4e314432010-10-07 21:45:39 +0000711 int err;
712 struct _stat wstatbuf;
Victor Stinneree587ea2011-11-17 00:51:38 +0100713 wchar_t *wpath;
Victor Stinner4e314432010-10-07 21:45:39 +0000714
Victor Stinneree587ea2011-11-17 00:51:38 +0100715 wpath = PyUnicode_AsUnicode(path);
716 if (wpath == NULL)
Victor Stinnerbd0850b2011-12-18 20:47:30 +0100717 return -2;
Victor Stinneree587ea2011-11-17 00:51:38 +0100718 err = _wstat(wpath, &wstatbuf);
Victor Stinner4e314432010-10-07 21:45:39 +0000719 if (!err)
720 statbuf->st_mode = wstatbuf.st_mode;
721 return err;
722#else
723 int ret;
Victor Stinnera4a75952010-10-07 22:23:10 +0000724 PyObject *bytes = PyUnicode_EncodeFSDefault(path);
Victor Stinner4e314432010-10-07 21:45:39 +0000725 if (bytes == NULL)
Victor Stinnerbd0850b2011-12-18 20:47:30 +0100726 return -2;
Victor Stinner4e314432010-10-07 21:45:39 +0000727 ret = stat(PyBytes_AS_STRING(bytes), statbuf);
728 Py_DECREF(bytes);
729 return ret;
730#endif
731}
732
Victor Stinnerd45c7f82012-12-04 01:34:47 +0100733
Antoine Pitrou409b5382013-10-12 22:41:17 +0200734static int
Victor Stinnerdaf45552013-08-28 00:53:59 +0200735get_inheritable(int fd, int raise)
736{
737#ifdef MS_WINDOWS
738 HANDLE handle;
739 DWORD flags;
Victor Stinner6672d0c2010-10-07 22:53:43 +0000740
Victor Stinnerdaf45552013-08-28 00:53:59 +0200741 if (!_PyVerify_fd(fd)) {
742 if (raise)
743 PyErr_SetFromErrno(PyExc_OSError);
744 return -1;
745 }
746
Steve Dower8fc89802015-04-12 00:26:27 -0400747 _Py_BEGIN_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +0200748 handle = (HANDLE)_get_osfhandle(fd);
Steve Dower8fc89802015-04-12 00:26:27 -0400749 _Py_END_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +0200750 if (handle == INVALID_HANDLE_VALUE) {
751 if (raise)
Steve Dower41e72442015-03-14 11:38:27 -0700752 PyErr_SetFromErrno(PyExc_OSError);
Victor Stinnerdaf45552013-08-28 00:53:59 +0200753 return -1;
754 }
755
756 if (!GetHandleInformation(handle, &flags)) {
757 if (raise)
758 PyErr_SetFromWindowsErr(0);
759 return -1;
760 }
761
762 return (flags & HANDLE_FLAG_INHERIT);
763#else
764 int flags;
765
766 flags = fcntl(fd, F_GETFD, 0);
767 if (flags == -1) {
768 if (raise)
769 PyErr_SetFromErrno(PyExc_OSError);
770 return -1;
771 }
772 return !(flags & FD_CLOEXEC);
773#endif
774}
775
776/* Get the inheritable flag of the specified file descriptor.
Victor Stinnerb034eee2013-09-07 10:36:04 +0200777 Return 1 if the file descriptor can be inherited, 0 if it cannot,
Victor Stinnerdaf45552013-08-28 00:53:59 +0200778 raise an exception and return -1 on error. */
779int
780_Py_get_inheritable(int fd)
781{
782 return get_inheritable(fd, 1);
783}
784
785static int
786set_inheritable(int fd, int inheritable, int raise, int *atomic_flag_works)
787{
788#ifdef MS_WINDOWS
789 HANDLE handle;
790 DWORD flags;
Victor Stinner282124b2014-09-02 11:41:04 +0200791#else
792#if defined(HAVE_SYS_IOCTL_H) && defined(FIOCLEX) && defined(FIONCLEX)
793 static int ioctl_works = -1;
Victor Stinnerdaf45552013-08-28 00:53:59 +0200794 int request;
795 int err;
Victor Stinner282124b2014-09-02 11:41:04 +0200796#endif
Victor Stinnerdaf45552013-08-28 00:53:59 +0200797 int flags;
798 int res;
799#endif
800
801 /* atomic_flag_works can only be used to make the file descriptor
802 non-inheritable */
803 assert(!(atomic_flag_works != NULL && inheritable));
804
805 if (atomic_flag_works != NULL && !inheritable) {
806 if (*atomic_flag_works == -1) {
Steve Dower41e72442015-03-14 11:38:27 -0700807 int isInheritable = get_inheritable(fd, raise);
808 if (isInheritable == -1)
Victor Stinnerdaf45552013-08-28 00:53:59 +0200809 return -1;
Steve Dower41e72442015-03-14 11:38:27 -0700810 *atomic_flag_works = !isInheritable;
Victor Stinnerdaf45552013-08-28 00:53:59 +0200811 }
812
813 if (*atomic_flag_works)
814 return 0;
815 }
816
817#ifdef MS_WINDOWS
818 if (!_PyVerify_fd(fd)) {
819 if (raise)
820 PyErr_SetFromErrno(PyExc_OSError);
821 return -1;
822 }
823
Steve Dower8fc89802015-04-12 00:26:27 -0400824 _Py_BEGIN_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +0200825 handle = (HANDLE)_get_osfhandle(fd);
Steve Dower8fc89802015-04-12 00:26:27 -0400826 _Py_END_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +0200827 if (handle == INVALID_HANDLE_VALUE) {
828 if (raise)
Steve Dower41e72442015-03-14 11:38:27 -0700829 PyErr_SetFromErrno(PyExc_OSError);
Victor Stinnerdaf45552013-08-28 00:53:59 +0200830 return -1;
831 }
832
833 if (inheritable)
834 flags = HANDLE_FLAG_INHERIT;
835 else
836 flags = 0;
837 if (!SetHandleInformation(handle, HANDLE_FLAG_INHERIT, flags)) {
838 if (raise)
839 PyErr_SetFromWindowsErr(0);
840 return -1;
841 }
842 return 0;
843
Victor Stinnerdaf45552013-08-28 00:53:59 +0200844#else
Victor Stinner282124b2014-09-02 11:41:04 +0200845
846#if defined(HAVE_SYS_IOCTL_H) && defined(FIOCLEX) && defined(FIONCLEX)
847 if (ioctl_works != 0) {
848 /* fast-path: ioctl() only requires one syscall */
849 if (inheritable)
850 request = FIONCLEX;
851 else
852 request = FIOCLEX;
853 err = ioctl(fd, request, NULL);
854 if (!err) {
855 ioctl_works = 1;
856 return 0;
857 }
858
Victor Stinner3116cc42016-05-19 16:46:18 +0200859 if (errno != ENOTTY && errno != EACCES) {
Victor Stinner282124b2014-09-02 11:41:04 +0200860 if (raise)
861 PyErr_SetFromErrno(PyExc_OSError);
862 return -1;
863 }
864 else {
865 /* Issue #22258: Here, ENOTTY means "Inappropriate ioctl for
866 device". The ioctl is declared but not supported by the kernel.
867 Remember that ioctl() doesn't work. It is the case on
Victor Stinner3116cc42016-05-19 16:46:18 +0200868 Illumos-based OS for example.
869
870 Issue #27057: When SELinux policy disallows ioctl it will fail
871 with EACCES. While FIOCLEX is safe operation it may be
872 unavailable because ioctl was denied altogether.
873 This can be the case on Android. */
Victor Stinner282124b2014-09-02 11:41:04 +0200874 ioctl_works = 0;
875 }
876 /* fallback to fcntl() if ioctl() does not work */
877 }
878#endif
879
880 /* slow-path: fcntl() requires two syscalls */
Victor Stinnerdaf45552013-08-28 00:53:59 +0200881 flags = fcntl(fd, F_GETFD);
882 if (flags < 0) {
883 if (raise)
884 PyErr_SetFromErrno(PyExc_OSError);
885 return -1;
886 }
887
888 if (inheritable)
889 flags &= ~FD_CLOEXEC;
890 else
891 flags |= FD_CLOEXEC;
892 res = fcntl(fd, F_SETFD, flags);
893 if (res < 0) {
894 if (raise)
895 PyErr_SetFromErrno(PyExc_OSError);
896 return -1;
897 }
898 return 0;
899#endif
900}
901
902/* Make the file descriptor non-inheritable.
Victor Stinnerb034eee2013-09-07 10:36:04 +0200903 Return 0 on success, set errno and return -1 on error. */
Victor Stinnerdaf45552013-08-28 00:53:59 +0200904static int
905make_non_inheritable(int fd)
906{
907 return set_inheritable(fd, 0, 0, NULL);
908}
909
910/* Set the inheritable flag of the specified file descriptor.
911 On success: return 0, on error: raise an exception if raise is nonzero
912 and return -1.
913
914 If atomic_flag_works is not NULL:
915
916 * if *atomic_flag_works==-1, check if the inheritable is set on the file
917 descriptor: if yes, set *atomic_flag_works to 1, otherwise set to 0 and
918 set the inheritable flag
919 * if *atomic_flag_works==1: do nothing
920 * if *atomic_flag_works==0: set inheritable flag to False
921
922 Set atomic_flag_works to NULL if no atomic flag was used to create the
923 file descriptor.
924
925 atomic_flag_works can only be used to make a file descriptor
926 non-inheritable: atomic_flag_works must be NULL if inheritable=1. */
927int
928_Py_set_inheritable(int fd, int inheritable, int *atomic_flag_works)
929{
930 return set_inheritable(fd, inheritable, 1, atomic_flag_works);
931}
932
Victor Stinnera555cfc2015-03-18 00:22:14 +0100933static int
934_Py_open_impl(const char *pathname, int flags, int gil_held)
Victor Stinnerdaf45552013-08-28 00:53:59 +0200935{
936 int fd;
Victor Stinnera47fc5c2015-03-18 09:52:54 +0100937 int async_err = 0;
Victor Stinnera555cfc2015-03-18 00:22:14 +0100938#ifndef MS_WINDOWS
Victor Stinnerdaf45552013-08-28 00:53:59 +0200939 int *atomic_flag_works;
Victor Stinnera555cfc2015-03-18 00:22:14 +0100940#endif
941
942#ifdef MS_WINDOWS
943 flags |= O_NOINHERIT;
944#elif defined(O_CLOEXEC)
Victor Stinnerdaf45552013-08-28 00:53:59 +0200945 atomic_flag_works = &_Py_open_cloexec_works;
946 flags |= O_CLOEXEC;
947#else
948 atomic_flag_works = NULL;
949#endif
Victor Stinnerdaf45552013-08-28 00:53:59 +0200950
Victor Stinnera555cfc2015-03-18 00:22:14 +0100951 if (gil_held) {
Victor Stinnera47fc5c2015-03-18 09:52:54 +0100952 do {
953 Py_BEGIN_ALLOW_THREADS
954 fd = open(pathname, flags);
955 Py_END_ALLOW_THREADS
956 } while (fd < 0
957 && errno == EINTR && !(async_err = PyErr_CheckSignals()));
958 if (async_err)
959 return -1;
Victor Stinnera555cfc2015-03-18 00:22:14 +0100960 if (fd < 0) {
961 PyErr_SetFromErrnoWithFilename(PyExc_OSError, pathname);
962 return -1;
963 }
964 }
965 else {
966 fd = open(pathname, flags);
967 if (fd < 0)
968 return -1;
969 }
970
971#ifndef MS_WINDOWS
972 if (set_inheritable(fd, 0, gil_held, atomic_flag_works) < 0) {
Victor Stinnerdaf45552013-08-28 00:53:59 +0200973 close(fd);
974 return -1;
975 }
Victor Stinnera555cfc2015-03-18 00:22:14 +0100976#endif
977
Victor Stinnerdaf45552013-08-28 00:53:59 +0200978 return fd;
979}
980
Victor Stinnera555cfc2015-03-18 00:22:14 +0100981/* Open a file with the specified flags (wrapper to open() function).
982 Return a file descriptor on success. Raise an exception and return -1 on
983 error.
984
985 The file descriptor is created non-inheritable.
986
Victor Stinnera47fc5c2015-03-18 09:52:54 +0100987 When interrupted by a signal (open() fails with EINTR), retry the syscall,
988 except if the Python signal handler raises an exception.
989
Victor Stinner6f4fae82015-04-01 18:34:32 +0200990 Release the GIL to call open(). The caller must hold the GIL. */
Victor Stinnera555cfc2015-03-18 00:22:14 +0100991int
992_Py_open(const char *pathname, int flags)
993{
Victor Stinnerbc5b80b2015-10-11 09:54:42 +0200994#ifdef WITH_THREAD
Victor Stinnera555cfc2015-03-18 00:22:14 +0100995 /* _Py_open() must be called with the GIL held. */
996 assert(PyGILState_Check());
Victor Stinnerbc5b80b2015-10-11 09:54:42 +0200997#endif
Victor Stinnera555cfc2015-03-18 00:22:14 +0100998 return _Py_open_impl(pathname, flags, 1);
999}
1000
1001/* Open a file with the specified flags (wrapper to open() function).
1002 Return a file descriptor on success. Set errno and return -1 on error.
1003
Victor Stinnera47fc5c2015-03-18 09:52:54 +01001004 The file descriptor is created non-inheritable.
1005
1006 If interrupted by a signal, fail with EINTR. */
Victor Stinnera555cfc2015-03-18 00:22:14 +01001007int
1008_Py_open_noraise(const char *pathname, int flags)
1009{
1010 return _Py_open_impl(pathname, flags, 0);
1011}
1012
Victor Stinnerdaf45552013-08-28 00:53:59 +02001013/* Open a file. Use _wfopen() on Windows, encode the path to the locale
Victor Stinnere42ccd22015-03-18 01:39:23 +01001014 encoding and use fopen() otherwise.
1015
Victor Stinnera47fc5c2015-03-18 09:52:54 +01001016 The file descriptor is created non-inheritable.
1017
1018 If interrupted by a signal, fail with EINTR. */
Victor Stinner4e314432010-10-07 21:45:39 +00001019FILE *
1020_Py_wfopen(const wchar_t *path, const wchar_t *mode)
1021{
Victor Stinner4e314432010-10-07 21:45:39 +00001022 FILE *f;
Victor Stinnerdaf45552013-08-28 00:53:59 +02001023#ifndef MS_WINDOWS
Victor Stinner4e314432010-10-07 21:45:39 +00001024 char *cpath;
1025 char cmode[10];
1026 size_t r;
1027 r = wcstombs(cmode, mode, 10);
1028 if (r == (size_t)-1 || r >= 10) {
1029 errno = EINVAL;
1030 return NULL;
1031 }
Victor Stinnerf6a271a2014-08-01 12:28:48 +02001032 cpath = Py_EncodeLocale(path, NULL);
Victor Stinner4e314432010-10-07 21:45:39 +00001033 if (cpath == NULL)
1034 return NULL;
1035 f = fopen(cpath, cmode);
1036 PyMem_Free(cpath);
Victor Stinner4e314432010-10-07 21:45:39 +00001037#else
Victor Stinnerdaf45552013-08-28 00:53:59 +02001038 f = _wfopen(path, mode);
Victor Stinner4e314432010-10-07 21:45:39 +00001039#endif
Victor Stinnerdaf45552013-08-28 00:53:59 +02001040 if (f == NULL)
1041 return NULL;
1042 if (make_non_inheritable(fileno(f)) < 0) {
1043 fclose(f);
1044 return NULL;
1045 }
1046 return f;
Victor Stinner4e314432010-10-07 21:45:39 +00001047}
1048
Victor Stinnere42ccd22015-03-18 01:39:23 +01001049/* Wrapper to fopen().
1050
Victor Stinnera47fc5c2015-03-18 09:52:54 +01001051 The file descriptor is created non-inheritable.
1052
1053 If interrupted by a signal, fail with EINTR. */
Victor Stinnerdaf45552013-08-28 00:53:59 +02001054FILE*
1055_Py_fopen(const char *pathname, const char *mode)
1056{
1057 FILE *f = fopen(pathname, mode);
1058 if (f == NULL)
1059 return NULL;
1060 if (make_non_inheritable(fileno(f)) < 0) {
1061 fclose(f);
1062 return NULL;
1063 }
1064 return f;
1065}
1066
1067/* Open a file. Call _wfopen() on Windows, or encode the path to the filesystem
Victor Stinnere42ccd22015-03-18 01:39:23 +01001068 encoding and call fopen() otherwise.
Victor Stinner6672d0c2010-10-07 22:53:43 +00001069
Victor Stinnere42ccd22015-03-18 01:39:23 +01001070 Return the new file object on success. Raise an exception and return NULL
1071 on error.
1072
1073 The file descriptor is created non-inheritable.
1074
Victor Stinnera47fc5c2015-03-18 09:52:54 +01001075 When interrupted by a signal (open() fails with EINTR), retry the syscall,
1076 except if the Python signal handler raises an exception.
1077
Victor Stinner6f4fae82015-04-01 18:34:32 +02001078 Release the GIL to call _wfopen() or fopen(). The caller must hold
1079 the GIL. */
Victor Stinner4e314432010-10-07 21:45:39 +00001080FILE*
Victor Stinnerdaf45552013-08-28 00:53:59 +02001081_Py_fopen_obj(PyObject *path, const char *mode)
Victor Stinner4e314432010-10-07 21:45:39 +00001082{
Victor Stinnerdaf45552013-08-28 00:53:59 +02001083 FILE *f;
Victor Stinnera47fc5c2015-03-18 09:52:54 +01001084 int async_err = 0;
Victor Stinner4e314432010-10-07 21:45:39 +00001085#ifdef MS_WINDOWS
Victor Stinneree587ea2011-11-17 00:51:38 +01001086 wchar_t *wpath;
Victor Stinner4e314432010-10-07 21:45:39 +00001087 wchar_t wmode[10];
1088 int usize;
Victor Stinner4e314432010-10-07 21:45:39 +00001089
Victor Stinnerbc5b80b2015-10-11 09:54:42 +02001090#ifdef WITH_THREAD
Victor Stinnere42ccd22015-03-18 01:39:23 +01001091 assert(PyGILState_Check());
Victor Stinnerbc5b80b2015-10-11 09:54:42 +02001092#endif
Victor Stinnere42ccd22015-03-18 01:39:23 +01001093
Antoine Pitrou0e576f12011-12-22 10:03:38 +01001094 if (!PyUnicode_Check(path)) {
1095 PyErr_Format(PyExc_TypeError,
1096 "str file path expected under Windows, got %R",
1097 Py_TYPE(path));
1098 return NULL;
1099 }
Victor Stinneree587ea2011-11-17 00:51:38 +01001100 wpath = PyUnicode_AsUnicode(path);
1101 if (wpath == NULL)
1102 return NULL;
1103
Victor Stinner4e314432010-10-07 21:45:39 +00001104 usize = MultiByteToWideChar(CP_ACP, 0, mode, -1, wmode, sizeof(wmode));
Victor Stinnere42ccd22015-03-18 01:39:23 +01001105 if (usize == 0) {
1106 PyErr_SetFromWindowsErr(0);
Victor Stinner4e314432010-10-07 21:45:39 +00001107 return NULL;
Victor Stinnere42ccd22015-03-18 01:39:23 +01001108 }
Victor Stinner4e314432010-10-07 21:45:39 +00001109
Victor Stinnera47fc5c2015-03-18 09:52:54 +01001110 do {
1111 Py_BEGIN_ALLOW_THREADS
1112 f = _wfopen(wpath, wmode);
1113 Py_END_ALLOW_THREADS
1114 } while (f == NULL
1115 && errno == EINTR && !(async_err = PyErr_CheckSignals()));
Victor Stinner4e314432010-10-07 21:45:39 +00001116#else
Antoine Pitrou2b1cc892011-12-19 18:19:06 +01001117 PyObject *bytes;
Victor Stinnere42ccd22015-03-18 01:39:23 +01001118 char *path_bytes;
1119
Victor Stinnerbc5b80b2015-10-11 09:54:42 +02001120#ifdef WITH_THREAD
Victor Stinnere42ccd22015-03-18 01:39:23 +01001121 assert(PyGILState_Check());
Victor Stinnerbc5b80b2015-10-11 09:54:42 +02001122#endif
Victor Stinnere42ccd22015-03-18 01:39:23 +01001123
Antoine Pitrou2b1cc892011-12-19 18:19:06 +01001124 if (!PyUnicode_FSConverter(path, &bytes))
Victor Stinner4e314432010-10-07 21:45:39 +00001125 return NULL;
Victor Stinnere42ccd22015-03-18 01:39:23 +01001126 path_bytes = PyBytes_AS_STRING(bytes);
1127
Victor Stinnera47fc5c2015-03-18 09:52:54 +01001128 do {
1129 Py_BEGIN_ALLOW_THREADS
1130 f = fopen(path_bytes, mode);
1131 Py_END_ALLOW_THREADS
1132 } while (f == NULL
1133 && errno == EINTR && !(async_err = PyErr_CheckSignals()));
Victor Stinnere42ccd22015-03-18 01:39:23 +01001134
Victor Stinner4e314432010-10-07 21:45:39 +00001135 Py_DECREF(bytes);
Victor Stinner4e314432010-10-07 21:45:39 +00001136#endif
Victor Stinnera47fc5c2015-03-18 09:52:54 +01001137 if (async_err)
1138 return NULL;
1139
Victor Stinnere42ccd22015-03-18 01:39:23 +01001140 if (f == NULL) {
1141 PyErr_SetFromErrnoWithFilenameObject(PyExc_OSError, path);
Victor Stinnerdaf45552013-08-28 00:53:59 +02001142 return NULL;
Victor Stinnere42ccd22015-03-18 01:39:23 +01001143 }
1144
1145 if (set_inheritable(fileno(f), 0, 1, NULL) < 0) {
Victor Stinnerdaf45552013-08-28 00:53:59 +02001146 fclose(f);
1147 return NULL;
1148 }
1149 return f;
Victor Stinner4e314432010-10-07 21:45:39 +00001150}
1151
Victor Stinner66aab0c2015-03-19 22:53:20 +01001152/* Read count bytes from fd into buf.
Victor Stinner82c3e452015-04-01 18:34:45 +02001153
1154 On success, return the number of read bytes, it can be lower than count.
1155 If the current file offset is at or past the end of file, no bytes are read,
1156 and read() returns zero.
1157
1158 On error, raise an exception, set errno and return -1.
1159
1160 When interrupted by a signal (read() fails with EINTR), retry the syscall.
1161 If the Python signal handler raises an exception, the function returns -1
1162 (the syscall is not retried).
1163
1164 Release the GIL to call read(). The caller must hold the GIL. */
Victor Stinner66aab0c2015-03-19 22:53:20 +01001165Py_ssize_t
1166_Py_read(int fd, void *buf, size_t count)
1167{
1168 Py_ssize_t n;
Victor Stinnera3c02022015-03-20 11:58:18 +01001169 int err;
Victor Stinner66aab0c2015-03-19 22:53:20 +01001170 int async_err = 0;
1171
1172 /* _Py_read() must not be called with an exception set, otherwise the
1173 * caller may think that read() was interrupted by a signal and the signal
1174 * handler raised an exception. */
1175 assert(!PyErr_Occurred());
1176
Victor Stinnerc1cf4f72015-03-19 23:53:04 +01001177 if (!_PyVerify_fd(fd)) {
Victor Stinnera3c02022015-03-20 11:58:18 +01001178 /* save/restore errno because PyErr_SetFromErrno() can modify it */
1179 err = errno;
Victor Stinner66aab0c2015-03-19 22:53:20 +01001180 PyErr_SetFromErrno(PyExc_OSError);
Victor Stinnera3c02022015-03-20 11:58:18 +01001181 errno = err;
Victor Stinner66aab0c2015-03-19 22:53:20 +01001182 return -1;
1183 }
1184
1185#ifdef MS_WINDOWS
1186 if (count > INT_MAX) {
1187 /* On Windows, the count parameter of read() is an int */
1188 count = INT_MAX;
1189 }
1190#else
1191 if (count > PY_SSIZE_T_MAX) {
1192 /* if count is greater than PY_SSIZE_T_MAX,
1193 * read() result is undefined */
1194 count = PY_SSIZE_T_MAX;
1195 }
1196#endif
1197
Steve Dower8fc89802015-04-12 00:26:27 -04001198 _Py_BEGIN_SUPPRESS_IPH
Victor Stinner66aab0c2015-03-19 22:53:20 +01001199 do {
1200 Py_BEGIN_ALLOW_THREADS
1201 errno = 0;
1202#ifdef MS_WINDOWS
1203 n = read(fd, buf, (int)count);
1204#else
1205 n = read(fd, buf, count);
1206#endif
Victor Stinnera3c02022015-03-20 11:58:18 +01001207 /* save/restore errno because PyErr_CheckSignals()
1208 * and PyErr_SetFromErrno() can modify it */
1209 err = errno;
Victor Stinner66aab0c2015-03-19 22:53:20 +01001210 Py_END_ALLOW_THREADS
Victor Stinnera3c02022015-03-20 11:58:18 +01001211 } while (n < 0 && err == EINTR &&
Victor Stinner66aab0c2015-03-19 22:53:20 +01001212 !(async_err = PyErr_CheckSignals()));
Steve Dower8fc89802015-04-12 00:26:27 -04001213 _Py_END_SUPPRESS_IPH
Victor Stinner66aab0c2015-03-19 22:53:20 +01001214
1215 if (async_err) {
1216 /* read() was interrupted by a signal (failed with EINTR)
1217 * and the Python signal handler raised an exception */
Victor Stinnera3c02022015-03-20 11:58:18 +01001218 errno = err;
1219 assert(errno == EINTR && PyErr_Occurred());
Victor Stinner66aab0c2015-03-19 22:53:20 +01001220 return -1;
1221 }
1222 if (n < 0) {
Victor Stinner66aab0c2015-03-19 22:53:20 +01001223 PyErr_SetFromErrno(PyExc_OSError);
Victor Stinnera3c02022015-03-20 11:58:18 +01001224 errno = err;
Victor Stinner66aab0c2015-03-19 22:53:20 +01001225 return -1;
1226 }
1227
1228 return n;
1229}
1230
Victor Stinner82c3e452015-04-01 18:34:45 +02001231static Py_ssize_t
1232_Py_write_impl(int fd, const void *buf, size_t count, int gil_held)
Victor Stinner66aab0c2015-03-19 22:53:20 +01001233{
1234 Py_ssize_t n;
Victor Stinnera3c02022015-03-20 11:58:18 +01001235 int err;
Victor Stinner66aab0c2015-03-19 22:53:20 +01001236 int async_err = 0;
1237
Victor Stinner66aab0c2015-03-19 22:53:20 +01001238 if (!_PyVerify_fd(fd)) {
Victor Stinner82c3e452015-04-01 18:34:45 +02001239 if (gil_held) {
1240 /* save/restore errno because PyErr_SetFromErrno() can modify it */
1241 err = errno;
1242 PyErr_SetFromErrno(PyExc_OSError);
1243 errno = err;
1244 }
Victor Stinner66aab0c2015-03-19 22:53:20 +01001245 return -1;
1246 }
1247
Steve Dower8fc89802015-04-12 00:26:27 -04001248 _Py_BEGIN_SUPPRESS_IPH
Victor Stinner66aab0c2015-03-19 22:53:20 +01001249#ifdef MS_WINDOWS
1250 if (count > 32767 && isatty(fd)) {
1251 /* Issue #11395: the Windows console returns an error (12: not
1252 enough space error) on writing into stdout if stdout mode is
1253 binary and the length is greater than 66,000 bytes (or less,
1254 depending on heap usage). */
1255 count = 32767;
1256 }
1257 else if (count > INT_MAX)
1258 count = INT_MAX;
1259#else
1260 if (count > PY_SSIZE_T_MAX) {
1261 /* write() should truncate count to PY_SSIZE_T_MAX, but it's safer
1262 * to do it ourself to have a portable behaviour. */
1263 count = PY_SSIZE_T_MAX;
1264 }
1265#endif
1266
Victor Stinner82c3e452015-04-01 18:34:45 +02001267 if (gil_held) {
1268 do {
1269 Py_BEGIN_ALLOW_THREADS
1270 errno = 0;
Victor Stinner66aab0c2015-03-19 22:53:20 +01001271#ifdef MS_WINDOWS
Victor Stinner82c3e452015-04-01 18:34:45 +02001272 n = write(fd, buf, (int)count);
Victor Stinner66aab0c2015-03-19 22:53:20 +01001273#else
Victor Stinner82c3e452015-04-01 18:34:45 +02001274 n = write(fd, buf, count);
Victor Stinner66aab0c2015-03-19 22:53:20 +01001275#endif
Victor Stinner82c3e452015-04-01 18:34:45 +02001276 /* save/restore errno because PyErr_CheckSignals()
1277 * and PyErr_SetFromErrno() can modify it */
1278 err = errno;
1279 Py_END_ALLOW_THREADS
1280 } while (n < 0 && err == EINTR &&
1281 !(async_err = PyErr_CheckSignals()));
1282 }
1283 else {
1284 do {
1285 errno = 0;
1286#ifdef MS_WINDOWS
1287 n = write(fd, buf, (int)count);
1288#else
1289 n = write(fd, buf, count);
1290#endif
1291 err = errno;
1292 } while (n < 0 && err == EINTR);
1293 }
Steve Dower8fc89802015-04-12 00:26:27 -04001294 _Py_END_SUPPRESS_IPH
Victor Stinner66aab0c2015-03-19 22:53:20 +01001295
1296 if (async_err) {
1297 /* write() was interrupted by a signal (failed with EINTR)
Victor Stinner82c3e452015-04-01 18:34:45 +02001298 and the Python signal handler raised an exception (if gil_held is
1299 nonzero). */
Victor Stinnera3c02022015-03-20 11:58:18 +01001300 errno = err;
Victor Stinner82c3e452015-04-01 18:34:45 +02001301 assert(errno == EINTR && (!gil_held || PyErr_Occurred()));
Victor Stinner66aab0c2015-03-19 22:53:20 +01001302 return -1;
1303 }
1304 if (n < 0) {
Victor Stinner82c3e452015-04-01 18:34:45 +02001305 if (gil_held)
1306 PyErr_SetFromErrno(PyExc_OSError);
Victor Stinnera3c02022015-03-20 11:58:18 +01001307 errno = err;
Victor Stinner66aab0c2015-03-19 22:53:20 +01001308 return -1;
1309 }
1310
1311 return n;
1312}
1313
Victor Stinner82c3e452015-04-01 18:34:45 +02001314/* Write count bytes of buf into fd.
1315
1316 On success, return the number of written bytes, it can be lower than count
1317 including 0. On error, raise an exception, set errno and return -1.
1318
1319 When interrupted by a signal (write() fails with EINTR), retry the syscall.
1320 If the Python signal handler raises an exception, the function returns -1
1321 (the syscall is not retried).
1322
1323 Release the GIL to call write(). The caller must hold the GIL. */
1324Py_ssize_t
1325_Py_write(int fd, const void *buf, size_t count)
1326{
1327 /* _Py_write() must not be called with an exception set, otherwise the
1328 * caller may think that write() was interrupted by a signal and the signal
1329 * handler raised an exception. */
1330 assert(!PyErr_Occurred());
1331
1332 return _Py_write_impl(fd, buf, count, 1);
1333}
1334
1335/* Write count bytes of buf into fd.
1336 *
1337 * On success, return the number of written bytes, it can be lower than count
1338 * including 0. On error, set errno and return -1.
1339 *
1340 * When interrupted by a signal (write() fails with EINTR), retry the syscall
1341 * without calling the Python signal handler. */
1342Py_ssize_t
1343_Py_write_noraise(int fd, const void *buf, size_t count)
1344{
1345 return _Py_write_impl(fd, buf, count, 0);
1346}
1347
Victor Stinner4e314432010-10-07 21:45:39 +00001348#ifdef HAVE_READLINK
Victor Stinner6672d0c2010-10-07 22:53:43 +00001349
1350/* Read value of symbolic link. Encode the path to the locale encoding, decode
Victor Stinneraf02e1c2011-12-16 23:56:01 +01001351 the result from the locale encoding. Return -1 on error. */
Victor Stinner6672d0c2010-10-07 22:53:43 +00001352
Victor Stinner4e314432010-10-07 21:45:39 +00001353int
1354_Py_wreadlink(const wchar_t *path, wchar_t *buf, size_t bufsiz)
1355{
1356 char *cpath;
Victor Stinnerb11d6cb2013-11-15 18:14:11 +01001357 char cbuf[MAXPATHLEN];
Victor Stinner3f711f42010-10-16 22:47:37 +00001358 wchar_t *wbuf;
Victor Stinner4e314432010-10-07 21:45:39 +00001359 int res;
1360 size_t r1;
1361
Victor Stinnerf6a271a2014-08-01 12:28:48 +02001362 cpath = Py_EncodeLocale(path, NULL);
Victor Stinner4e314432010-10-07 21:45:39 +00001363 if (cpath == NULL) {
1364 errno = EINVAL;
1365 return -1;
1366 }
Victor Stinnerb11d6cb2013-11-15 18:14:11 +01001367 res = (int)readlink(cpath, cbuf, Py_ARRAY_LENGTH(cbuf));
Victor Stinner4e314432010-10-07 21:45:39 +00001368 PyMem_Free(cpath);
1369 if (res == -1)
1370 return -1;
Victor Stinnerb11d6cb2013-11-15 18:14:11 +01001371 if (res == Py_ARRAY_LENGTH(cbuf)) {
Victor Stinner4e314432010-10-07 21:45:39 +00001372 errno = EINVAL;
1373 return -1;
1374 }
1375 cbuf[res] = '\0'; /* buf will be null terminated */
Victor Stinnerf6a271a2014-08-01 12:28:48 +02001376 wbuf = Py_DecodeLocale(cbuf, &r1);
Victor Stinner350147b2010-10-16 22:52:09 +00001377 if (wbuf == NULL) {
1378 errno = EINVAL;
1379 return -1;
1380 }
Victor Stinner3f711f42010-10-16 22:47:37 +00001381 if (bufsiz <= r1) {
Victor Stinner1a7425f2013-07-07 16:25:15 +02001382 PyMem_RawFree(wbuf);
Victor Stinner4e314432010-10-07 21:45:39 +00001383 errno = EINVAL;
1384 return -1;
1385 }
Victor Stinner3f711f42010-10-16 22:47:37 +00001386 wcsncpy(buf, wbuf, bufsiz);
Victor Stinner1a7425f2013-07-07 16:25:15 +02001387 PyMem_RawFree(wbuf);
Victor Stinner4e314432010-10-07 21:45:39 +00001388 return (int)r1;
1389}
1390#endif
1391
1392#ifdef HAVE_REALPATH
Victor Stinner6672d0c2010-10-07 22:53:43 +00001393
1394/* Return the canonicalized absolute pathname. Encode path to the locale
Victor Stinneraf02e1c2011-12-16 23:56:01 +01001395 encoding, decode the result from the locale encoding.
1396 Return NULL on error. */
Victor Stinner6672d0c2010-10-07 22:53:43 +00001397
Victor Stinner4e314432010-10-07 21:45:39 +00001398wchar_t*
Victor Stinner015f4d82010-10-07 22:29:53 +00001399_Py_wrealpath(const wchar_t *path,
1400 wchar_t *resolved_path, size_t resolved_path_size)
Victor Stinner4e314432010-10-07 21:45:39 +00001401{
1402 char *cpath;
Victor Stinnerb11d6cb2013-11-15 18:14:11 +01001403 char cresolved_path[MAXPATHLEN];
Victor Stinner0a1b8cb2010-10-16 22:55:47 +00001404 wchar_t *wresolved_path;
Victor Stinner4e314432010-10-07 21:45:39 +00001405 char *res;
1406 size_t r;
Victor Stinnerf6a271a2014-08-01 12:28:48 +02001407 cpath = Py_EncodeLocale(path, NULL);
Victor Stinner4e314432010-10-07 21:45:39 +00001408 if (cpath == NULL) {
1409 errno = EINVAL;
1410 return NULL;
1411 }
1412 res = realpath(cpath, cresolved_path);
1413 PyMem_Free(cpath);
1414 if (res == NULL)
1415 return NULL;
Victor Stinner0a1b8cb2010-10-16 22:55:47 +00001416
Victor Stinnerf6a271a2014-08-01 12:28:48 +02001417 wresolved_path = Py_DecodeLocale(cresolved_path, &r);
Victor Stinner0a1b8cb2010-10-16 22:55:47 +00001418 if (wresolved_path == NULL) {
Victor Stinner4e314432010-10-07 21:45:39 +00001419 errno = EINVAL;
1420 return NULL;
1421 }
Victor Stinner0a1b8cb2010-10-16 22:55:47 +00001422 if (resolved_path_size <= r) {
Victor Stinner1a7425f2013-07-07 16:25:15 +02001423 PyMem_RawFree(wresolved_path);
Victor Stinner0a1b8cb2010-10-16 22:55:47 +00001424 errno = EINVAL;
1425 return NULL;
1426 }
1427 wcsncpy(resolved_path, wresolved_path, resolved_path_size);
Victor Stinner1a7425f2013-07-07 16:25:15 +02001428 PyMem_RawFree(wresolved_path);
Victor Stinner4e314432010-10-07 21:45:39 +00001429 return resolved_path;
1430}
1431#endif
1432
Victor Stinnerf4061da2010-10-14 12:37:19 +00001433/* Get the current directory. size is the buffer size in wide characters
Victor Stinneraf02e1c2011-12-16 23:56:01 +01001434 including the null character. Decode the path from the locale encoding.
1435 Return NULL on error. */
Victor Stinner6672d0c2010-10-07 22:53:43 +00001436
Victor Stinner4e314432010-10-07 21:45:39 +00001437wchar_t*
1438_Py_wgetcwd(wchar_t *buf, size_t size)
1439{
1440#ifdef MS_WINDOWS
Victor Stinner56785ea2013-06-05 00:46:29 +02001441 int isize = (int)Py_MIN(size, INT_MAX);
1442 return _wgetcwd(buf, isize);
Victor Stinner4e314432010-10-07 21:45:39 +00001443#else
Victor Stinnerb11d6cb2013-11-15 18:14:11 +01001444 char fname[MAXPATHLEN];
Victor Stinnerf4061da2010-10-14 12:37:19 +00001445 wchar_t *wname;
Victor Stinner168e1172010-10-16 23:16:16 +00001446 size_t len;
Victor Stinnerf4061da2010-10-14 12:37:19 +00001447
Victor Stinnerb11d6cb2013-11-15 18:14:11 +01001448 if (getcwd(fname, Py_ARRAY_LENGTH(fname)) == NULL)
Victor Stinner4e314432010-10-07 21:45:39 +00001449 return NULL;
Victor Stinnerf6a271a2014-08-01 12:28:48 +02001450 wname = Py_DecodeLocale(fname, &len);
Victor Stinnerf4061da2010-10-14 12:37:19 +00001451 if (wname == NULL)
1452 return NULL;
Victor Stinner168e1172010-10-16 23:16:16 +00001453 if (size <= len) {
Victor Stinner1a7425f2013-07-07 16:25:15 +02001454 PyMem_RawFree(wname);
Victor Stinner4e314432010-10-07 21:45:39 +00001455 return NULL;
1456 }
Victor Stinnerf4061da2010-10-14 12:37:19 +00001457 wcsncpy(buf, wname, size);
Victor Stinner1a7425f2013-07-07 16:25:15 +02001458 PyMem_RawFree(wname);
Victor Stinner4e314432010-10-07 21:45:39 +00001459 return buf;
1460#endif
1461}
1462
Victor Stinnerdaf45552013-08-28 00:53:59 +02001463/* Duplicate a file descriptor. The new file descriptor is created as
1464 non-inheritable. Return a new file descriptor on success, raise an OSError
1465 exception and return -1 on error.
1466
1467 The GIL is released to call dup(). The caller must hold the GIL. */
1468int
1469_Py_dup(int fd)
1470{
1471#ifdef MS_WINDOWS
1472 HANDLE handle;
1473 DWORD ftype;
1474#endif
1475
1476 if (!_PyVerify_fd(fd)) {
1477 PyErr_SetFromErrno(PyExc_OSError);
1478 return -1;
1479 }
1480
1481#ifdef MS_WINDOWS
Steve Dower8fc89802015-04-12 00:26:27 -04001482 _Py_BEGIN_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +02001483 handle = (HANDLE)_get_osfhandle(fd);
Steve Dower8fc89802015-04-12 00:26:27 -04001484 _Py_END_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +02001485 if (handle == INVALID_HANDLE_VALUE) {
Steve Dower41e72442015-03-14 11:38:27 -07001486 PyErr_SetFromErrno(PyExc_OSError);
Victor Stinnerdaf45552013-08-28 00:53:59 +02001487 return -1;
1488 }
1489
1490 /* get the file type, ignore the error if it failed */
1491 ftype = GetFileType(handle);
1492
1493 Py_BEGIN_ALLOW_THREADS
Steve Dower8fc89802015-04-12 00:26:27 -04001494 _Py_BEGIN_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +02001495 fd = dup(fd);
Steve Dower8fc89802015-04-12 00:26:27 -04001496 _Py_END_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +02001497 Py_END_ALLOW_THREADS
1498 if (fd < 0) {
1499 PyErr_SetFromErrno(PyExc_OSError);
1500 return -1;
1501 }
1502
1503 /* Character files like console cannot be make non-inheritable */
1504 if (ftype != FILE_TYPE_CHAR) {
1505 if (_Py_set_inheritable(fd, 0, NULL) < 0) {
Steve Dower8fc89802015-04-12 00:26:27 -04001506 _Py_BEGIN_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +02001507 close(fd);
Steve Dower8fc89802015-04-12 00:26:27 -04001508 _Py_END_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +02001509 return -1;
1510 }
1511 }
1512#elif defined(HAVE_FCNTL_H) && defined(F_DUPFD_CLOEXEC)
1513 Py_BEGIN_ALLOW_THREADS
Steve Dower8fc89802015-04-12 00:26:27 -04001514 _Py_BEGIN_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +02001515 fd = fcntl(fd, F_DUPFD_CLOEXEC, 0);
Steve Dower8fc89802015-04-12 00:26:27 -04001516 _Py_END_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +02001517 Py_END_ALLOW_THREADS
1518 if (fd < 0) {
1519 PyErr_SetFromErrno(PyExc_OSError);
1520 return -1;
1521 }
1522
1523#else
1524 Py_BEGIN_ALLOW_THREADS
Steve Dower8fc89802015-04-12 00:26:27 -04001525 _Py_BEGIN_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +02001526 fd = dup(fd);
Steve Dower8fc89802015-04-12 00:26:27 -04001527 _Py_END_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +02001528 Py_END_ALLOW_THREADS
1529 if (fd < 0) {
1530 PyErr_SetFromErrno(PyExc_OSError);
1531 return -1;
1532 }
1533
1534 if (_Py_set_inheritable(fd, 0, NULL) < 0) {
Steve Dower8fc89802015-04-12 00:26:27 -04001535 _Py_BEGIN_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +02001536 close(fd);
Steve Dower8fc89802015-04-12 00:26:27 -04001537 _Py_END_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +02001538 return -1;
1539 }
1540#endif
1541 return fd;
1542}
1543
Victor Stinner1db9e7b2014-07-29 22:32:47 +02001544#ifndef MS_WINDOWS
1545/* Get the blocking mode of the file descriptor.
1546 Return 0 if the O_NONBLOCK flag is set, 1 if the flag is cleared,
1547 raise an exception and return -1 on error. */
1548int
1549_Py_get_blocking(int fd)
1550{
Steve Dower8fc89802015-04-12 00:26:27 -04001551 int flags;
1552 _Py_BEGIN_SUPPRESS_IPH
1553 flags = fcntl(fd, F_GETFL, 0);
1554 _Py_END_SUPPRESS_IPH
Victor Stinner1db9e7b2014-07-29 22:32:47 +02001555 if (flags < 0) {
1556 PyErr_SetFromErrno(PyExc_OSError);
1557 return -1;
1558 }
1559
1560 return !(flags & O_NONBLOCK);
1561}
1562
1563/* Set the blocking mode of the specified file descriptor.
1564
1565 Set the O_NONBLOCK flag if blocking is False, clear the O_NONBLOCK flag
1566 otherwise.
1567
1568 Return 0 on success, raise an exception and return -1 on error. */
1569int
1570_Py_set_blocking(int fd, int blocking)
1571{
1572#if defined(HAVE_SYS_IOCTL_H) && defined(FIONBIO)
1573 int arg = !blocking;
1574 if (ioctl(fd, FIONBIO, &arg) < 0)
1575 goto error;
1576#else
1577 int flags, res;
1578
Steve Dower8fc89802015-04-12 00:26:27 -04001579 _Py_BEGIN_SUPPRESS_IPH
Victor Stinner1db9e7b2014-07-29 22:32:47 +02001580 flags = fcntl(fd, F_GETFL, 0);
Steve Dower8fc89802015-04-12 00:26:27 -04001581 if (flags >= 0) {
1582 if (blocking)
1583 flags = flags & (~O_NONBLOCK);
1584 else
1585 flags = flags | O_NONBLOCK;
Victor Stinner1db9e7b2014-07-29 22:32:47 +02001586
Steve Dower8fc89802015-04-12 00:26:27 -04001587 res = fcntl(fd, F_SETFL, flags);
1588 } else {
1589 res = -1;
1590 }
1591 _Py_END_SUPPRESS_IPH
Victor Stinner1db9e7b2014-07-29 22:32:47 +02001592
Victor Stinner1db9e7b2014-07-29 22:32:47 +02001593 if (res < 0)
1594 goto error;
1595#endif
1596 return 0;
1597
1598error:
1599 PyErr_SetFromErrno(PyExc_OSError);
1600 return -1;
1601}
1602#endif
1603
Steve Dower8fc89802015-04-12 00:26:27 -04001604#if defined _MSC_VER && _MSC_VER >= 1400 && _MSC_VER < 1900
Steve Dowerd81431f2015-03-06 14:47:02 -08001605/* Legacy implementation of _PyVerify_fd while transitioning to
1606 * MSVC 14.0. This should eventually be removed. (issue23524)
1607 */
1608
1609/* Microsoft CRT in VS2005 and higher will verify that a filehandle is
1610 * valid and raise an assertion if it isn't.
1611 * Normally, an invalid fd is likely to be a C program error and therefore
1612 * an assertion can be useful, but it does contradict the POSIX standard
1613 * which for write(2) states:
1614 * "Otherwise, -1 shall be returned and errno set to indicate the error."
1615 * "[EBADF] The fildes argument is not a valid file descriptor open for
1616 * writing."
1617 * Furthermore, python allows the user to enter any old integer
1618 * as a fd and should merely raise a python exception on error.
1619 * The Microsoft CRT doesn't provide an official way to check for the
1620 * validity of a file descriptor, but we can emulate its internal behaviour
1621 * by using the exported __pinfo data member and knowledge of the
1622 * internal structures involved.
1623 * The structures below must be updated for each version of visual studio
1624 * according to the file internal.h in the CRT source, until MS comes
1625 * up with a less hacky way to do this.
1626 * (all of this is to avoid globally modifying the CRT behaviour using
1627 * _set_invalid_parameter_handler() and _CrtSetReportMode())
1628 */
1629/* The actual size of the structure is determined at runtime.
1630 * Only the first items must be present.
1631 */
1632typedef struct {
1633 intptr_t osfhnd;
1634 char osfile;
1635} my_ioinfo;
1636
1637extern __declspec(dllimport) char * __pioinfo[];
1638#define IOINFO_L2E 5
1639#define IOINFO_ARRAYS 64
1640#define IOINFO_ARRAY_ELTS (1 << IOINFO_L2E)
1641#define _NHANDLE_ (IOINFO_ARRAYS * IOINFO_ARRAY_ELTS)
1642#define FOPEN 0x01
1643#define _NO_CONSOLE_FILENO (intptr_t)-2
1644
1645/* This function emulates what the windows CRT does to validate file handles */
1646int
1647_PyVerify_fd(int fd)
1648{
1649 const int i1 = fd >> IOINFO_L2E;
1650 const int i2 = fd & ((1 << IOINFO_L2E) - 1);
1651
1652 static size_t sizeof_ioinfo = 0;
1653
1654 /* Determine the actual size of the ioinfo structure,
1655 * as used by the CRT loaded in memory
1656 */
1657 if (sizeof_ioinfo == 0 && __pioinfo[0] != NULL) {
1658 sizeof_ioinfo = _msize(__pioinfo[0]) / IOINFO_ARRAY_ELTS;
1659 }
1660 if (sizeof_ioinfo == 0) {
1661 /* This should not happen... */
1662 goto fail;
1663 }
1664
1665 /* See that it isn't a special CLEAR fileno */
1666 if (fd != _NO_CONSOLE_FILENO) {
1667 /* Microsoft CRT would check that 0<=fd<_nhandle but we can't do that. Instead
1668 * we check pointer validity and other info
1669 */
1670 if (0 <= i1 && i1 < IOINFO_ARRAYS && __pioinfo[i1] != NULL) {
1671 /* finally, check that the file is open */
1672 my_ioinfo* info = (my_ioinfo*)(__pioinfo[i1] + i2 * sizeof_ioinfo);
1673 if (info->osfile & FOPEN) {
1674 return 1;
1675 }
1676 }
1677 }
1678 fail:
1679 errno = EBADF;
1680 return 0;
1681}
1682
Steve Dower8fc89802015-04-12 00:26:27 -04001683#endif /* defined _MSC_VER && _MSC_VER >= 1400 && _MSC_VER < 1900 */