blob: 4a0ef48f120d702df1f706fb25b43c184e74be74 [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
Victor Stinner8a1be612016-03-14 22:07:55 +0100686#ifdef WITH_THREAD
687 assert(PyGILState_Check());
688#endif
689
Victor Stinnere134a7f2015-03-30 10:09:31 +0200690 Py_BEGIN_ALLOW_THREADS
691 res = _Py_fstat_noraise(fd, status);
692 Py_END_ALLOW_THREADS
693
694 if (res != 0) {
695#ifdef MS_WINDOWS
696 PyErr_SetFromWindowsErr(0);
697#else
698 PyErr_SetFromErrno(PyExc_OSError);
699#endif
700 return -1;
701 }
702 return 0;
703}
Steve Dowerf2f373f2015-02-21 08:44:05 -0800704
Victor Stinner6672d0c2010-10-07 22:53:43 +0000705/* Call _wstat() on Windows, or encode the path to the filesystem encoding and
706 call stat() otherwise. Only fill st_mode attribute on Windows.
707
Victor Stinnerbd0850b2011-12-18 20:47:30 +0100708 Return 0 on success, -1 on _wstat() / stat() error, -2 if an exception was
709 raised. */
Victor Stinner4e314432010-10-07 21:45:39 +0000710
711int
Victor Stinnera4a75952010-10-07 22:23:10 +0000712_Py_stat(PyObject *path, struct stat *statbuf)
Victor Stinner4e314432010-10-07 21:45:39 +0000713{
714#ifdef MS_WINDOWS
Victor Stinner4e314432010-10-07 21:45:39 +0000715 int err;
716 struct _stat wstatbuf;
Victor Stinneree587ea2011-11-17 00:51:38 +0100717 wchar_t *wpath;
Victor Stinner4e314432010-10-07 21:45:39 +0000718
Victor Stinneree587ea2011-11-17 00:51:38 +0100719 wpath = PyUnicode_AsUnicode(path);
720 if (wpath == NULL)
Victor Stinnerbd0850b2011-12-18 20:47:30 +0100721 return -2;
Victor Stinneree587ea2011-11-17 00:51:38 +0100722 err = _wstat(wpath, &wstatbuf);
Victor Stinner4e314432010-10-07 21:45:39 +0000723 if (!err)
724 statbuf->st_mode = wstatbuf.st_mode;
725 return err;
726#else
727 int ret;
Victor Stinnera4a75952010-10-07 22:23:10 +0000728 PyObject *bytes = PyUnicode_EncodeFSDefault(path);
Victor Stinner4e314432010-10-07 21:45:39 +0000729 if (bytes == NULL)
Victor Stinnerbd0850b2011-12-18 20:47:30 +0100730 return -2;
Victor Stinner4e314432010-10-07 21:45:39 +0000731 ret = stat(PyBytes_AS_STRING(bytes), statbuf);
732 Py_DECREF(bytes);
733 return ret;
734#endif
735}
736
Victor Stinnerd45c7f82012-12-04 01:34:47 +0100737
Antoine Pitrou409b5382013-10-12 22:41:17 +0200738static int
Victor Stinnerdaf45552013-08-28 00:53:59 +0200739get_inheritable(int fd, int raise)
740{
741#ifdef MS_WINDOWS
742 HANDLE handle;
743 DWORD flags;
Victor Stinner6672d0c2010-10-07 22:53:43 +0000744
Victor Stinnerdaf45552013-08-28 00:53:59 +0200745 if (!_PyVerify_fd(fd)) {
746 if (raise)
747 PyErr_SetFromErrno(PyExc_OSError);
748 return -1;
749 }
750
Steve Dower8fc89802015-04-12 00:26:27 -0400751 _Py_BEGIN_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +0200752 handle = (HANDLE)_get_osfhandle(fd);
Steve Dower8fc89802015-04-12 00:26:27 -0400753 _Py_END_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +0200754 if (handle == INVALID_HANDLE_VALUE) {
755 if (raise)
Steve Dower41e72442015-03-14 11:38:27 -0700756 PyErr_SetFromErrno(PyExc_OSError);
Victor Stinnerdaf45552013-08-28 00:53:59 +0200757 return -1;
758 }
759
760 if (!GetHandleInformation(handle, &flags)) {
761 if (raise)
762 PyErr_SetFromWindowsErr(0);
763 return -1;
764 }
765
766 return (flags & HANDLE_FLAG_INHERIT);
767#else
768 int flags;
769
770 flags = fcntl(fd, F_GETFD, 0);
771 if (flags == -1) {
772 if (raise)
773 PyErr_SetFromErrno(PyExc_OSError);
774 return -1;
775 }
776 return !(flags & FD_CLOEXEC);
777#endif
778}
779
780/* Get the inheritable flag of the specified file descriptor.
Victor Stinnerb034eee2013-09-07 10:36:04 +0200781 Return 1 if the file descriptor can be inherited, 0 if it cannot,
Victor Stinnerdaf45552013-08-28 00:53:59 +0200782 raise an exception and return -1 on error. */
783int
784_Py_get_inheritable(int fd)
785{
786 return get_inheritable(fd, 1);
787}
788
789static int
790set_inheritable(int fd, int inheritable, int raise, int *atomic_flag_works)
791{
792#ifdef MS_WINDOWS
793 HANDLE handle;
794 DWORD flags;
Victor Stinner282124b2014-09-02 11:41:04 +0200795#else
796#if defined(HAVE_SYS_IOCTL_H) && defined(FIOCLEX) && defined(FIONCLEX)
797 static int ioctl_works = -1;
Victor Stinnerdaf45552013-08-28 00:53:59 +0200798 int request;
799 int err;
Victor Stinner282124b2014-09-02 11:41:04 +0200800#endif
Victor Stinnera858bbd2016-04-17 16:51:52 +0200801 int flags, new_flags;
Victor Stinnerdaf45552013-08-28 00:53:59 +0200802 int res;
803#endif
804
805 /* atomic_flag_works can only be used to make the file descriptor
806 non-inheritable */
807 assert(!(atomic_flag_works != NULL && inheritable));
808
809 if (atomic_flag_works != NULL && !inheritable) {
810 if (*atomic_flag_works == -1) {
Steve Dower41e72442015-03-14 11:38:27 -0700811 int isInheritable = get_inheritable(fd, raise);
812 if (isInheritable == -1)
Victor Stinnerdaf45552013-08-28 00:53:59 +0200813 return -1;
Steve Dower41e72442015-03-14 11:38:27 -0700814 *atomic_flag_works = !isInheritable;
Victor Stinnerdaf45552013-08-28 00:53:59 +0200815 }
816
817 if (*atomic_flag_works)
818 return 0;
819 }
820
821#ifdef MS_WINDOWS
822 if (!_PyVerify_fd(fd)) {
823 if (raise)
824 PyErr_SetFromErrno(PyExc_OSError);
825 return -1;
826 }
827
Steve Dower8fc89802015-04-12 00:26:27 -0400828 _Py_BEGIN_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +0200829 handle = (HANDLE)_get_osfhandle(fd);
Steve Dower8fc89802015-04-12 00:26:27 -0400830 _Py_END_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +0200831 if (handle == INVALID_HANDLE_VALUE) {
832 if (raise)
Steve Dower41e72442015-03-14 11:38:27 -0700833 PyErr_SetFromErrno(PyExc_OSError);
Victor Stinnerdaf45552013-08-28 00:53:59 +0200834 return -1;
835 }
836
837 if (inheritable)
838 flags = HANDLE_FLAG_INHERIT;
839 else
840 flags = 0;
841 if (!SetHandleInformation(handle, HANDLE_FLAG_INHERIT, flags)) {
842 if (raise)
843 PyErr_SetFromWindowsErr(0);
844 return -1;
845 }
846 return 0;
847
Victor Stinnerdaf45552013-08-28 00:53:59 +0200848#else
Victor Stinner282124b2014-09-02 11:41:04 +0200849
850#if defined(HAVE_SYS_IOCTL_H) && defined(FIOCLEX) && defined(FIONCLEX)
851 if (ioctl_works != 0) {
852 /* fast-path: ioctl() only requires one syscall */
853 if (inheritable)
854 request = FIONCLEX;
855 else
856 request = FIOCLEX;
857 err = ioctl(fd, request, NULL);
858 if (!err) {
859 ioctl_works = 1;
860 return 0;
861 }
862
863 if (errno != ENOTTY) {
864 if (raise)
865 PyErr_SetFromErrno(PyExc_OSError);
866 return -1;
867 }
868 else {
869 /* Issue #22258: Here, ENOTTY means "Inappropriate ioctl for
870 device". The ioctl is declared but not supported by the kernel.
871 Remember that ioctl() doesn't work. It is the case on
872 Illumos-based OS for example. */
873 ioctl_works = 0;
874 }
875 /* fallback to fcntl() if ioctl() does not work */
876 }
877#endif
878
879 /* slow-path: fcntl() requires two syscalls */
Victor Stinnerdaf45552013-08-28 00:53:59 +0200880 flags = fcntl(fd, F_GETFD);
881 if (flags < 0) {
882 if (raise)
883 PyErr_SetFromErrno(PyExc_OSError);
884 return -1;
885 }
886
Victor Stinnera858bbd2016-04-17 16:51:52 +0200887 if (inheritable) {
888 new_flags = flags & ~FD_CLOEXEC;
889 }
890 else {
891 new_flags = flags | FD_CLOEXEC;
892 }
893
894 if (new_flags == flags) {
895 /* FD_CLOEXEC flag already set/cleared: nothing to do */
896 return 0;
897 }
898
Victor Stinnerdaf45552013-08-28 00:53:59 +0200899 res = fcntl(fd, F_SETFD, flags);
900 if (res < 0) {
901 if (raise)
902 PyErr_SetFromErrno(PyExc_OSError);
903 return -1;
904 }
905 return 0;
906#endif
907}
908
909/* Make the file descriptor non-inheritable.
Victor Stinnerb034eee2013-09-07 10:36:04 +0200910 Return 0 on success, set errno and return -1 on error. */
Victor Stinnerdaf45552013-08-28 00:53:59 +0200911static int
912make_non_inheritable(int fd)
913{
914 return set_inheritable(fd, 0, 0, NULL);
915}
916
917/* Set the inheritable flag of the specified file descriptor.
918 On success: return 0, on error: raise an exception if raise is nonzero
919 and return -1.
920
921 If atomic_flag_works is not NULL:
922
923 * if *atomic_flag_works==-1, check if the inheritable is set on the file
924 descriptor: if yes, set *atomic_flag_works to 1, otherwise set to 0 and
925 set the inheritable flag
926 * if *atomic_flag_works==1: do nothing
927 * if *atomic_flag_works==0: set inheritable flag to False
928
929 Set atomic_flag_works to NULL if no atomic flag was used to create the
930 file descriptor.
931
932 atomic_flag_works can only be used to make a file descriptor
933 non-inheritable: atomic_flag_works must be NULL if inheritable=1. */
934int
935_Py_set_inheritable(int fd, int inheritable, int *atomic_flag_works)
936{
937 return set_inheritable(fd, inheritable, 1, atomic_flag_works);
938}
939
Victor Stinnera555cfc2015-03-18 00:22:14 +0100940static int
941_Py_open_impl(const char *pathname, int flags, int gil_held)
Victor Stinnerdaf45552013-08-28 00:53:59 +0200942{
943 int fd;
Victor Stinnera47fc5c2015-03-18 09:52:54 +0100944 int async_err = 0;
Victor Stinnera555cfc2015-03-18 00:22:14 +0100945#ifndef MS_WINDOWS
Victor Stinnerdaf45552013-08-28 00:53:59 +0200946 int *atomic_flag_works;
Victor Stinnera555cfc2015-03-18 00:22:14 +0100947#endif
948
949#ifdef MS_WINDOWS
950 flags |= O_NOINHERIT;
951#elif defined(O_CLOEXEC)
Victor Stinnerdaf45552013-08-28 00:53:59 +0200952 atomic_flag_works = &_Py_open_cloexec_works;
953 flags |= O_CLOEXEC;
954#else
955 atomic_flag_works = NULL;
956#endif
Victor Stinnerdaf45552013-08-28 00:53:59 +0200957
Victor Stinnera555cfc2015-03-18 00:22:14 +0100958 if (gil_held) {
Victor Stinnera47fc5c2015-03-18 09:52:54 +0100959 do {
960 Py_BEGIN_ALLOW_THREADS
961 fd = open(pathname, flags);
962 Py_END_ALLOW_THREADS
963 } while (fd < 0
964 && errno == EINTR && !(async_err = PyErr_CheckSignals()));
965 if (async_err)
966 return -1;
Victor Stinnera555cfc2015-03-18 00:22:14 +0100967 if (fd < 0) {
968 PyErr_SetFromErrnoWithFilename(PyExc_OSError, pathname);
969 return -1;
970 }
971 }
972 else {
973 fd = open(pathname, flags);
974 if (fd < 0)
975 return -1;
976 }
977
978#ifndef MS_WINDOWS
979 if (set_inheritable(fd, 0, gil_held, atomic_flag_works) < 0) {
Victor Stinnerdaf45552013-08-28 00:53:59 +0200980 close(fd);
981 return -1;
982 }
Victor Stinnera555cfc2015-03-18 00:22:14 +0100983#endif
984
Victor Stinnerdaf45552013-08-28 00:53:59 +0200985 return fd;
986}
987
Victor Stinnera555cfc2015-03-18 00:22:14 +0100988/* Open a file with the specified flags (wrapper to open() function).
989 Return a file descriptor on success. Raise an exception and return -1 on
990 error.
991
992 The file descriptor is created non-inheritable.
993
Victor Stinnera47fc5c2015-03-18 09:52:54 +0100994 When interrupted by a signal (open() fails with EINTR), retry the syscall,
995 except if the Python signal handler raises an exception.
996
Victor Stinner6f4fae82015-04-01 18:34:32 +0200997 Release the GIL to call open(). The caller must hold the GIL. */
Victor Stinnera555cfc2015-03-18 00:22:14 +0100998int
999_Py_open(const char *pathname, int flags)
1000{
Victor Stinnerbc5b80b2015-10-11 09:54:42 +02001001#ifdef WITH_THREAD
Victor Stinnera555cfc2015-03-18 00:22:14 +01001002 /* _Py_open() must be called with the GIL held. */
1003 assert(PyGILState_Check());
Victor Stinnerbc5b80b2015-10-11 09:54:42 +02001004#endif
Victor Stinnera555cfc2015-03-18 00:22:14 +01001005 return _Py_open_impl(pathname, flags, 1);
1006}
1007
1008/* Open a file with the specified flags (wrapper to open() function).
1009 Return a file descriptor on success. Set errno and return -1 on error.
1010
Victor Stinnera47fc5c2015-03-18 09:52:54 +01001011 The file descriptor is created non-inheritable.
1012
1013 If interrupted by a signal, fail with EINTR. */
Victor Stinnera555cfc2015-03-18 00:22:14 +01001014int
1015_Py_open_noraise(const char *pathname, int flags)
1016{
1017 return _Py_open_impl(pathname, flags, 0);
1018}
1019
Victor Stinnerdaf45552013-08-28 00:53:59 +02001020/* Open a file. Use _wfopen() on Windows, encode the path to the locale
Victor Stinnere42ccd22015-03-18 01:39:23 +01001021 encoding and use fopen() otherwise.
1022
Victor Stinnera47fc5c2015-03-18 09:52:54 +01001023 The file descriptor is created non-inheritable.
1024
1025 If interrupted by a signal, fail with EINTR. */
Victor Stinner4e314432010-10-07 21:45:39 +00001026FILE *
1027_Py_wfopen(const wchar_t *path, const wchar_t *mode)
1028{
Victor Stinner4e314432010-10-07 21:45:39 +00001029 FILE *f;
Victor Stinnerdaf45552013-08-28 00:53:59 +02001030#ifndef MS_WINDOWS
Victor Stinner4e314432010-10-07 21:45:39 +00001031 char *cpath;
1032 char cmode[10];
1033 size_t r;
1034 r = wcstombs(cmode, mode, 10);
1035 if (r == (size_t)-1 || r >= 10) {
1036 errno = EINVAL;
1037 return NULL;
1038 }
Victor Stinnerf6a271a2014-08-01 12:28:48 +02001039 cpath = Py_EncodeLocale(path, NULL);
Victor Stinner4e314432010-10-07 21:45:39 +00001040 if (cpath == NULL)
1041 return NULL;
1042 f = fopen(cpath, cmode);
1043 PyMem_Free(cpath);
Victor Stinner4e314432010-10-07 21:45:39 +00001044#else
Victor Stinnerdaf45552013-08-28 00:53:59 +02001045 f = _wfopen(path, mode);
Victor Stinner4e314432010-10-07 21:45:39 +00001046#endif
Victor Stinnerdaf45552013-08-28 00:53:59 +02001047 if (f == NULL)
1048 return NULL;
1049 if (make_non_inheritable(fileno(f)) < 0) {
1050 fclose(f);
1051 return NULL;
1052 }
1053 return f;
Victor Stinner4e314432010-10-07 21:45:39 +00001054}
1055
Victor Stinnere42ccd22015-03-18 01:39:23 +01001056/* Wrapper to fopen().
1057
Victor Stinnera47fc5c2015-03-18 09:52:54 +01001058 The file descriptor is created non-inheritable.
1059
1060 If interrupted by a signal, fail with EINTR. */
Victor Stinnerdaf45552013-08-28 00:53:59 +02001061FILE*
1062_Py_fopen(const char *pathname, const char *mode)
1063{
1064 FILE *f = fopen(pathname, mode);
1065 if (f == NULL)
1066 return NULL;
1067 if (make_non_inheritable(fileno(f)) < 0) {
1068 fclose(f);
1069 return NULL;
1070 }
1071 return f;
1072}
1073
1074/* Open a file. Call _wfopen() on Windows, or encode the path to the filesystem
Victor Stinnere42ccd22015-03-18 01:39:23 +01001075 encoding and call fopen() otherwise.
Victor Stinner6672d0c2010-10-07 22:53:43 +00001076
Victor Stinnere42ccd22015-03-18 01:39:23 +01001077 Return the new file object on success. Raise an exception and return NULL
1078 on error.
1079
1080 The file descriptor is created non-inheritable.
1081
Victor Stinnera47fc5c2015-03-18 09:52:54 +01001082 When interrupted by a signal (open() fails with EINTR), retry the syscall,
1083 except if the Python signal handler raises an exception.
1084
Victor Stinner6f4fae82015-04-01 18:34:32 +02001085 Release the GIL to call _wfopen() or fopen(). The caller must hold
1086 the GIL. */
Victor Stinner4e314432010-10-07 21:45:39 +00001087FILE*
Victor Stinnerdaf45552013-08-28 00:53:59 +02001088_Py_fopen_obj(PyObject *path, const char *mode)
Victor Stinner4e314432010-10-07 21:45:39 +00001089{
Victor Stinnerdaf45552013-08-28 00:53:59 +02001090 FILE *f;
Victor Stinnera47fc5c2015-03-18 09:52:54 +01001091 int async_err = 0;
Victor Stinner4e314432010-10-07 21:45:39 +00001092#ifdef MS_WINDOWS
Victor Stinneree587ea2011-11-17 00:51:38 +01001093 wchar_t *wpath;
Victor Stinner4e314432010-10-07 21:45:39 +00001094 wchar_t wmode[10];
1095 int usize;
Victor Stinner4e314432010-10-07 21:45:39 +00001096
Victor Stinnerbc5b80b2015-10-11 09:54:42 +02001097#ifdef WITH_THREAD
Victor Stinnere42ccd22015-03-18 01:39:23 +01001098 assert(PyGILState_Check());
Victor Stinnerbc5b80b2015-10-11 09:54:42 +02001099#endif
Victor Stinnere42ccd22015-03-18 01:39:23 +01001100
Antoine Pitrou0e576f12011-12-22 10:03:38 +01001101 if (!PyUnicode_Check(path)) {
1102 PyErr_Format(PyExc_TypeError,
1103 "str file path expected under Windows, got %R",
1104 Py_TYPE(path));
1105 return NULL;
1106 }
Victor Stinneree587ea2011-11-17 00:51:38 +01001107 wpath = PyUnicode_AsUnicode(path);
1108 if (wpath == NULL)
1109 return NULL;
1110
Victor Stinner4e314432010-10-07 21:45:39 +00001111 usize = MultiByteToWideChar(CP_ACP, 0, mode, -1, wmode, sizeof(wmode));
Victor Stinnere42ccd22015-03-18 01:39:23 +01001112 if (usize == 0) {
1113 PyErr_SetFromWindowsErr(0);
Victor Stinner4e314432010-10-07 21:45:39 +00001114 return NULL;
Victor Stinnere42ccd22015-03-18 01:39:23 +01001115 }
Victor Stinner4e314432010-10-07 21:45:39 +00001116
Victor Stinnera47fc5c2015-03-18 09:52:54 +01001117 do {
1118 Py_BEGIN_ALLOW_THREADS
1119 f = _wfopen(wpath, wmode);
1120 Py_END_ALLOW_THREADS
1121 } while (f == NULL
1122 && errno == EINTR && !(async_err = PyErr_CheckSignals()));
Victor Stinner4e314432010-10-07 21:45:39 +00001123#else
Antoine Pitrou2b1cc892011-12-19 18:19:06 +01001124 PyObject *bytes;
Victor Stinnere42ccd22015-03-18 01:39:23 +01001125 char *path_bytes;
1126
Victor Stinnerbc5b80b2015-10-11 09:54:42 +02001127#ifdef WITH_THREAD
Victor Stinnere42ccd22015-03-18 01:39:23 +01001128 assert(PyGILState_Check());
Victor Stinnerbc5b80b2015-10-11 09:54:42 +02001129#endif
Victor Stinnere42ccd22015-03-18 01:39:23 +01001130
Antoine Pitrou2b1cc892011-12-19 18:19:06 +01001131 if (!PyUnicode_FSConverter(path, &bytes))
Victor Stinner4e314432010-10-07 21:45:39 +00001132 return NULL;
Victor Stinnere42ccd22015-03-18 01:39:23 +01001133 path_bytes = PyBytes_AS_STRING(bytes);
1134
Victor Stinnera47fc5c2015-03-18 09:52:54 +01001135 do {
1136 Py_BEGIN_ALLOW_THREADS
1137 f = fopen(path_bytes, mode);
1138 Py_END_ALLOW_THREADS
1139 } while (f == NULL
1140 && errno == EINTR && !(async_err = PyErr_CheckSignals()));
Victor Stinnere42ccd22015-03-18 01:39:23 +01001141
Victor Stinner4e314432010-10-07 21:45:39 +00001142 Py_DECREF(bytes);
Victor Stinner4e314432010-10-07 21:45:39 +00001143#endif
Victor Stinnera47fc5c2015-03-18 09:52:54 +01001144 if (async_err)
1145 return NULL;
1146
Victor Stinnere42ccd22015-03-18 01:39:23 +01001147 if (f == NULL) {
1148 PyErr_SetFromErrnoWithFilenameObject(PyExc_OSError, path);
Victor Stinnerdaf45552013-08-28 00:53:59 +02001149 return NULL;
Victor Stinnere42ccd22015-03-18 01:39:23 +01001150 }
1151
1152 if (set_inheritable(fileno(f), 0, 1, NULL) < 0) {
Victor Stinnerdaf45552013-08-28 00:53:59 +02001153 fclose(f);
1154 return NULL;
1155 }
1156 return f;
Victor Stinner4e314432010-10-07 21:45:39 +00001157}
1158
Victor Stinner66aab0c2015-03-19 22:53:20 +01001159/* Read count bytes from fd into buf.
Victor Stinner82c3e452015-04-01 18:34:45 +02001160
1161 On success, return the number of read bytes, it can be lower than count.
1162 If the current file offset is at or past the end of file, no bytes are read,
1163 and read() returns zero.
1164
1165 On error, raise an exception, set errno and return -1.
1166
1167 When interrupted by a signal (read() fails with EINTR), retry the syscall.
1168 If the Python signal handler raises an exception, the function returns -1
1169 (the syscall is not retried).
1170
1171 Release the GIL to call read(). The caller must hold the GIL. */
Victor Stinner66aab0c2015-03-19 22:53:20 +01001172Py_ssize_t
1173_Py_read(int fd, void *buf, size_t count)
1174{
1175 Py_ssize_t n;
Victor Stinnera3c02022015-03-20 11:58:18 +01001176 int err;
Victor Stinner66aab0c2015-03-19 22:53:20 +01001177 int async_err = 0;
1178
Victor Stinner8a1be612016-03-14 22:07:55 +01001179#ifdef WITH_THREAD
1180 assert(PyGILState_Check());
1181#endif
1182
Victor Stinner66aab0c2015-03-19 22:53:20 +01001183 /* _Py_read() must not be called with an exception set, otherwise the
1184 * caller may think that read() was interrupted by a signal and the signal
1185 * handler raised an exception. */
1186 assert(!PyErr_Occurred());
1187
Victor Stinnerc1cf4f72015-03-19 23:53:04 +01001188 if (!_PyVerify_fd(fd)) {
Victor Stinnera3c02022015-03-20 11:58:18 +01001189 /* save/restore errno because PyErr_SetFromErrno() can modify it */
1190 err = errno;
Victor Stinner66aab0c2015-03-19 22:53:20 +01001191 PyErr_SetFromErrno(PyExc_OSError);
Victor Stinnera3c02022015-03-20 11:58:18 +01001192 errno = err;
Victor Stinner66aab0c2015-03-19 22:53:20 +01001193 return -1;
1194 }
1195
1196#ifdef MS_WINDOWS
1197 if (count > INT_MAX) {
1198 /* On Windows, the count parameter of read() is an int */
1199 count = INT_MAX;
1200 }
1201#else
1202 if (count > PY_SSIZE_T_MAX) {
1203 /* if count is greater than PY_SSIZE_T_MAX,
1204 * read() result is undefined */
1205 count = PY_SSIZE_T_MAX;
1206 }
1207#endif
1208
Steve Dower8fc89802015-04-12 00:26:27 -04001209 _Py_BEGIN_SUPPRESS_IPH
Victor Stinner66aab0c2015-03-19 22:53:20 +01001210 do {
1211 Py_BEGIN_ALLOW_THREADS
1212 errno = 0;
1213#ifdef MS_WINDOWS
1214 n = read(fd, buf, (int)count);
1215#else
1216 n = read(fd, buf, count);
1217#endif
Victor Stinnera3c02022015-03-20 11:58:18 +01001218 /* save/restore errno because PyErr_CheckSignals()
1219 * and PyErr_SetFromErrno() can modify it */
1220 err = errno;
Victor Stinner66aab0c2015-03-19 22:53:20 +01001221 Py_END_ALLOW_THREADS
Victor Stinnera3c02022015-03-20 11:58:18 +01001222 } while (n < 0 && err == EINTR &&
Victor Stinner66aab0c2015-03-19 22:53:20 +01001223 !(async_err = PyErr_CheckSignals()));
Steve Dower8fc89802015-04-12 00:26:27 -04001224 _Py_END_SUPPRESS_IPH
Victor Stinner66aab0c2015-03-19 22:53:20 +01001225
1226 if (async_err) {
1227 /* read() was interrupted by a signal (failed with EINTR)
1228 * and the Python signal handler raised an exception */
Victor Stinnera3c02022015-03-20 11:58:18 +01001229 errno = err;
1230 assert(errno == EINTR && PyErr_Occurred());
Victor Stinner66aab0c2015-03-19 22:53:20 +01001231 return -1;
1232 }
1233 if (n < 0) {
Victor Stinner66aab0c2015-03-19 22:53:20 +01001234 PyErr_SetFromErrno(PyExc_OSError);
Victor Stinnera3c02022015-03-20 11:58:18 +01001235 errno = err;
Victor Stinner66aab0c2015-03-19 22:53:20 +01001236 return -1;
1237 }
1238
1239 return n;
1240}
1241
Victor Stinner82c3e452015-04-01 18:34:45 +02001242static Py_ssize_t
1243_Py_write_impl(int fd, const void *buf, size_t count, int gil_held)
Victor Stinner66aab0c2015-03-19 22:53:20 +01001244{
1245 Py_ssize_t n;
Victor Stinnera3c02022015-03-20 11:58:18 +01001246 int err;
Victor Stinner66aab0c2015-03-19 22:53:20 +01001247 int async_err = 0;
1248
Victor Stinner66aab0c2015-03-19 22:53:20 +01001249 if (!_PyVerify_fd(fd)) {
Victor Stinner82c3e452015-04-01 18:34:45 +02001250 if (gil_held) {
1251 /* save/restore errno because PyErr_SetFromErrno() can modify it */
1252 err = errno;
1253 PyErr_SetFromErrno(PyExc_OSError);
1254 errno = err;
1255 }
Victor Stinner66aab0c2015-03-19 22:53:20 +01001256 return -1;
1257 }
1258
Steve Dower8fc89802015-04-12 00:26:27 -04001259 _Py_BEGIN_SUPPRESS_IPH
Victor Stinner66aab0c2015-03-19 22:53:20 +01001260#ifdef MS_WINDOWS
1261 if (count > 32767 && isatty(fd)) {
1262 /* Issue #11395: the Windows console returns an error (12: not
1263 enough space error) on writing into stdout if stdout mode is
1264 binary and the length is greater than 66,000 bytes (or less,
1265 depending on heap usage). */
1266 count = 32767;
1267 }
1268 else if (count > INT_MAX)
1269 count = INT_MAX;
1270#else
1271 if (count > PY_SSIZE_T_MAX) {
1272 /* write() should truncate count to PY_SSIZE_T_MAX, but it's safer
1273 * to do it ourself to have a portable behaviour. */
1274 count = PY_SSIZE_T_MAX;
1275 }
1276#endif
1277
Victor Stinner82c3e452015-04-01 18:34:45 +02001278 if (gil_held) {
1279 do {
1280 Py_BEGIN_ALLOW_THREADS
1281 errno = 0;
Victor Stinner66aab0c2015-03-19 22:53:20 +01001282#ifdef MS_WINDOWS
Victor Stinner82c3e452015-04-01 18:34:45 +02001283 n = write(fd, buf, (int)count);
Victor Stinner66aab0c2015-03-19 22:53:20 +01001284#else
Victor Stinner82c3e452015-04-01 18:34:45 +02001285 n = write(fd, buf, count);
Victor Stinner66aab0c2015-03-19 22:53:20 +01001286#endif
Victor Stinner82c3e452015-04-01 18:34:45 +02001287 /* save/restore errno because PyErr_CheckSignals()
1288 * and PyErr_SetFromErrno() can modify it */
1289 err = errno;
1290 Py_END_ALLOW_THREADS
1291 } while (n < 0 && err == EINTR &&
1292 !(async_err = PyErr_CheckSignals()));
1293 }
1294 else {
1295 do {
1296 errno = 0;
1297#ifdef MS_WINDOWS
1298 n = write(fd, buf, (int)count);
1299#else
1300 n = write(fd, buf, count);
1301#endif
1302 err = errno;
1303 } while (n < 0 && err == EINTR);
1304 }
Steve Dower8fc89802015-04-12 00:26:27 -04001305 _Py_END_SUPPRESS_IPH
Victor Stinner66aab0c2015-03-19 22:53:20 +01001306
1307 if (async_err) {
1308 /* write() was interrupted by a signal (failed with EINTR)
Victor Stinner82c3e452015-04-01 18:34:45 +02001309 and the Python signal handler raised an exception (if gil_held is
1310 nonzero). */
Victor Stinnera3c02022015-03-20 11:58:18 +01001311 errno = err;
Victor Stinner82c3e452015-04-01 18:34:45 +02001312 assert(errno == EINTR && (!gil_held || PyErr_Occurred()));
Victor Stinner66aab0c2015-03-19 22:53:20 +01001313 return -1;
1314 }
1315 if (n < 0) {
Victor Stinner82c3e452015-04-01 18:34:45 +02001316 if (gil_held)
1317 PyErr_SetFromErrno(PyExc_OSError);
Victor Stinnera3c02022015-03-20 11:58:18 +01001318 errno = err;
Victor Stinner66aab0c2015-03-19 22:53:20 +01001319 return -1;
1320 }
1321
1322 return n;
1323}
1324
Victor Stinner82c3e452015-04-01 18:34:45 +02001325/* Write count bytes of buf into fd.
1326
1327 On success, return the number of written bytes, it can be lower than count
1328 including 0. On error, raise an exception, set errno and return -1.
1329
1330 When interrupted by a signal (write() fails with EINTR), retry the syscall.
1331 If the Python signal handler raises an exception, the function returns -1
1332 (the syscall is not retried).
1333
1334 Release the GIL to call write(). The caller must hold the GIL. */
1335Py_ssize_t
1336_Py_write(int fd, const void *buf, size_t count)
1337{
Victor Stinner8a1be612016-03-14 22:07:55 +01001338#ifdef WITH_THREAD
1339 assert(PyGILState_Check());
1340#endif
1341
Victor Stinner82c3e452015-04-01 18:34:45 +02001342 /* _Py_write() must not be called with an exception set, otherwise the
1343 * caller may think that write() was interrupted by a signal and the signal
1344 * handler raised an exception. */
1345 assert(!PyErr_Occurred());
1346
1347 return _Py_write_impl(fd, buf, count, 1);
1348}
1349
1350/* Write count bytes of buf into fd.
1351 *
1352 * On success, return the number of written bytes, it can be lower than count
1353 * including 0. On error, set errno and return -1.
1354 *
1355 * When interrupted by a signal (write() fails with EINTR), retry the syscall
1356 * without calling the Python signal handler. */
1357Py_ssize_t
1358_Py_write_noraise(int fd, const void *buf, size_t count)
1359{
1360 return _Py_write_impl(fd, buf, count, 0);
1361}
1362
Victor Stinner4e314432010-10-07 21:45:39 +00001363#ifdef HAVE_READLINK
Victor Stinner6672d0c2010-10-07 22:53:43 +00001364
1365/* Read value of symbolic link. Encode the path to the locale encoding, decode
Victor Stinneraf02e1c2011-12-16 23:56:01 +01001366 the result from the locale encoding. Return -1 on error. */
Victor Stinner6672d0c2010-10-07 22:53:43 +00001367
Victor Stinner4e314432010-10-07 21:45:39 +00001368int
1369_Py_wreadlink(const wchar_t *path, wchar_t *buf, size_t bufsiz)
1370{
1371 char *cpath;
Victor Stinnerb11d6cb2013-11-15 18:14:11 +01001372 char cbuf[MAXPATHLEN];
Victor Stinner3f711f42010-10-16 22:47:37 +00001373 wchar_t *wbuf;
Victor Stinner4e314432010-10-07 21:45:39 +00001374 int res;
1375 size_t r1;
1376
Victor Stinnerf6a271a2014-08-01 12:28:48 +02001377 cpath = Py_EncodeLocale(path, NULL);
Victor Stinner4e314432010-10-07 21:45:39 +00001378 if (cpath == NULL) {
1379 errno = EINVAL;
1380 return -1;
1381 }
Victor Stinnerb11d6cb2013-11-15 18:14:11 +01001382 res = (int)readlink(cpath, cbuf, Py_ARRAY_LENGTH(cbuf));
Victor Stinner4e314432010-10-07 21:45:39 +00001383 PyMem_Free(cpath);
1384 if (res == -1)
1385 return -1;
Victor Stinnerb11d6cb2013-11-15 18:14:11 +01001386 if (res == Py_ARRAY_LENGTH(cbuf)) {
Victor Stinner4e314432010-10-07 21:45:39 +00001387 errno = EINVAL;
1388 return -1;
1389 }
1390 cbuf[res] = '\0'; /* buf will be null terminated */
Victor Stinnerf6a271a2014-08-01 12:28:48 +02001391 wbuf = Py_DecodeLocale(cbuf, &r1);
Victor Stinner350147b2010-10-16 22:52:09 +00001392 if (wbuf == NULL) {
1393 errno = EINVAL;
1394 return -1;
1395 }
Victor Stinner3f711f42010-10-16 22:47:37 +00001396 if (bufsiz <= r1) {
Victor Stinner1a7425f2013-07-07 16:25:15 +02001397 PyMem_RawFree(wbuf);
Victor Stinner4e314432010-10-07 21:45:39 +00001398 errno = EINVAL;
1399 return -1;
1400 }
Victor Stinner3f711f42010-10-16 22:47:37 +00001401 wcsncpy(buf, wbuf, bufsiz);
Victor Stinner1a7425f2013-07-07 16:25:15 +02001402 PyMem_RawFree(wbuf);
Victor Stinner4e314432010-10-07 21:45:39 +00001403 return (int)r1;
1404}
1405#endif
1406
1407#ifdef HAVE_REALPATH
Victor Stinner6672d0c2010-10-07 22:53:43 +00001408
1409/* Return the canonicalized absolute pathname. Encode path to the locale
Victor Stinneraf02e1c2011-12-16 23:56:01 +01001410 encoding, decode the result from the locale encoding.
1411 Return NULL on error. */
Victor Stinner6672d0c2010-10-07 22:53:43 +00001412
Victor Stinner4e314432010-10-07 21:45:39 +00001413wchar_t*
Victor Stinner015f4d82010-10-07 22:29:53 +00001414_Py_wrealpath(const wchar_t *path,
1415 wchar_t *resolved_path, size_t resolved_path_size)
Victor Stinner4e314432010-10-07 21:45:39 +00001416{
1417 char *cpath;
Victor Stinnerb11d6cb2013-11-15 18:14:11 +01001418 char cresolved_path[MAXPATHLEN];
Victor Stinner0a1b8cb2010-10-16 22:55:47 +00001419 wchar_t *wresolved_path;
Victor Stinner4e314432010-10-07 21:45:39 +00001420 char *res;
1421 size_t r;
Victor Stinnerf6a271a2014-08-01 12:28:48 +02001422 cpath = Py_EncodeLocale(path, NULL);
Victor Stinner4e314432010-10-07 21:45:39 +00001423 if (cpath == NULL) {
1424 errno = EINVAL;
1425 return NULL;
1426 }
1427 res = realpath(cpath, cresolved_path);
1428 PyMem_Free(cpath);
1429 if (res == NULL)
1430 return NULL;
Victor Stinner0a1b8cb2010-10-16 22:55:47 +00001431
Victor Stinnerf6a271a2014-08-01 12:28:48 +02001432 wresolved_path = Py_DecodeLocale(cresolved_path, &r);
Victor Stinner0a1b8cb2010-10-16 22:55:47 +00001433 if (wresolved_path == NULL) {
Victor Stinner4e314432010-10-07 21:45:39 +00001434 errno = EINVAL;
1435 return NULL;
1436 }
Victor Stinner0a1b8cb2010-10-16 22:55:47 +00001437 if (resolved_path_size <= r) {
Victor Stinner1a7425f2013-07-07 16:25:15 +02001438 PyMem_RawFree(wresolved_path);
Victor Stinner0a1b8cb2010-10-16 22:55:47 +00001439 errno = EINVAL;
1440 return NULL;
1441 }
1442 wcsncpy(resolved_path, wresolved_path, resolved_path_size);
Victor Stinner1a7425f2013-07-07 16:25:15 +02001443 PyMem_RawFree(wresolved_path);
Victor Stinner4e314432010-10-07 21:45:39 +00001444 return resolved_path;
1445}
1446#endif
1447
Victor Stinnerf4061da2010-10-14 12:37:19 +00001448/* Get the current directory. size is the buffer size in wide characters
Victor Stinneraf02e1c2011-12-16 23:56:01 +01001449 including the null character. Decode the path from the locale encoding.
1450 Return NULL on error. */
Victor Stinner6672d0c2010-10-07 22:53:43 +00001451
Victor Stinner4e314432010-10-07 21:45:39 +00001452wchar_t*
1453_Py_wgetcwd(wchar_t *buf, size_t size)
1454{
1455#ifdef MS_WINDOWS
Victor Stinner56785ea2013-06-05 00:46:29 +02001456 int isize = (int)Py_MIN(size, INT_MAX);
1457 return _wgetcwd(buf, isize);
Victor Stinner4e314432010-10-07 21:45:39 +00001458#else
Victor Stinnerb11d6cb2013-11-15 18:14:11 +01001459 char fname[MAXPATHLEN];
Victor Stinnerf4061da2010-10-14 12:37:19 +00001460 wchar_t *wname;
Victor Stinner168e1172010-10-16 23:16:16 +00001461 size_t len;
Victor Stinnerf4061da2010-10-14 12:37:19 +00001462
Victor Stinnerb11d6cb2013-11-15 18:14:11 +01001463 if (getcwd(fname, Py_ARRAY_LENGTH(fname)) == NULL)
Victor Stinner4e314432010-10-07 21:45:39 +00001464 return NULL;
Victor Stinnerf6a271a2014-08-01 12:28:48 +02001465 wname = Py_DecodeLocale(fname, &len);
Victor Stinnerf4061da2010-10-14 12:37:19 +00001466 if (wname == NULL)
1467 return NULL;
Victor Stinner168e1172010-10-16 23:16:16 +00001468 if (size <= len) {
Victor Stinner1a7425f2013-07-07 16:25:15 +02001469 PyMem_RawFree(wname);
Victor Stinner4e314432010-10-07 21:45:39 +00001470 return NULL;
1471 }
Victor Stinnerf4061da2010-10-14 12:37:19 +00001472 wcsncpy(buf, wname, size);
Victor Stinner1a7425f2013-07-07 16:25:15 +02001473 PyMem_RawFree(wname);
Victor Stinner4e314432010-10-07 21:45:39 +00001474 return buf;
1475#endif
1476}
1477
Victor Stinnerdaf45552013-08-28 00:53:59 +02001478/* Duplicate a file descriptor. The new file descriptor is created as
1479 non-inheritable. Return a new file descriptor on success, raise an OSError
1480 exception and return -1 on error.
1481
1482 The GIL is released to call dup(). The caller must hold the GIL. */
1483int
1484_Py_dup(int fd)
1485{
1486#ifdef MS_WINDOWS
1487 HANDLE handle;
1488 DWORD ftype;
1489#endif
1490
Victor Stinner8a1be612016-03-14 22:07:55 +01001491#ifdef WITH_THREAD
1492 assert(PyGILState_Check());
1493#endif
1494
Victor Stinnerdaf45552013-08-28 00:53:59 +02001495 if (!_PyVerify_fd(fd)) {
1496 PyErr_SetFromErrno(PyExc_OSError);
1497 return -1;
1498 }
1499
1500#ifdef MS_WINDOWS
Steve Dower8fc89802015-04-12 00:26:27 -04001501 _Py_BEGIN_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +02001502 handle = (HANDLE)_get_osfhandle(fd);
Steve Dower8fc89802015-04-12 00:26:27 -04001503 _Py_END_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +02001504 if (handle == INVALID_HANDLE_VALUE) {
Steve Dower41e72442015-03-14 11:38:27 -07001505 PyErr_SetFromErrno(PyExc_OSError);
Victor Stinnerdaf45552013-08-28 00:53:59 +02001506 return -1;
1507 }
1508
1509 /* get the file type, ignore the error if it failed */
1510 ftype = GetFileType(handle);
1511
1512 Py_BEGIN_ALLOW_THREADS
Steve Dower8fc89802015-04-12 00:26:27 -04001513 _Py_BEGIN_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +02001514 fd = dup(fd);
Steve Dower8fc89802015-04-12 00:26:27 -04001515 _Py_END_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +02001516 Py_END_ALLOW_THREADS
1517 if (fd < 0) {
1518 PyErr_SetFromErrno(PyExc_OSError);
1519 return -1;
1520 }
1521
1522 /* Character files like console cannot be make non-inheritable */
1523 if (ftype != FILE_TYPE_CHAR) {
1524 if (_Py_set_inheritable(fd, 0, NULL) < 0) {
Steve Dower8fc89802015-04-12 00:26:27 -04001525 _Py_BEGIN_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +02001526 close(fd);
Steve Dower8fc89802015-04-12 00:26:27 -04001527 _Py_END_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +02001528 return -1;
1529 }
1530 }
1531#elif defined(HAVE_FCNTL_H) && defined(F_DUPFD_CLOEXEC)
1532 Py_BEGIN_ALLOW_THREADS
Steve Dower8fc89802015-04-12 00:26:27 -04001533 _Py_BEGIN_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +02001534 fd = fcntl(fd, F_DUPFD_CLOEXEC, 0);
Steve Dower8fc89802015-04-12 00:26:27 -04001535 _Py_END_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +02001536 Py_END_ALLOW_THREADS
1537 if (fd < 0) {
1538 PyErr_SetFromErrno(PyExc_OSError);
1539 return -1;
1540 }
1541
1542#else
1543 Py_BEGIN_ALLOW_THREADS
Steve Dower8fc89802015-04-12 00:26:27 -04001544 _Py_BEGIN_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +02001545 fd = dup(fd);
Steve Dower8fc89802015-04-12 00:26:27 -04001546 _Py_END_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +02001547 Py_END_ALLOW_THREADS
1548 if (fd < 0) {
1549 PyErr_SetFromErrno(PyExc_OSError);
1550 return -1;
1551 }
1552
1553 if (_Py_set_inheritable(fd, 0, NULL) < 0) {
Steve Dower8fc89802015-04-12 00:26:27 -04001554 _Py_BEGIN_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +02001555 close(fd);
Steve Dower8fc89802015-04-12 00:26:27 -04001556 _Py_END_SUPPRESS_IPH
Victor Stinnerdaf45552013-08-28 00:53:59 +02001557 return -1;
1558 }
1559#endif
1560 return fd;
1561}
1562
Victor Stinner1db9e7b2014-07-29 22:32:47 +02001563#ifndef MS_WINDOWS
1564/* Get the blocking mode of the file descriptor.
1565 Return 0 if the O_NONBLOCK flag is set, 1 if the flag is cleared,
1566 raise an exception and return -1 on error. */
1567int
1568_Py_get_blocking(int fd)
1569{
Steve Dower8fc89802015-04-12 00:26:27 -04001570 int flags;
1571 _Py_BEGIN_SUPPRESS_IPH
1572 flags = fcntl(fd, F_GETFL, 0);
1573 _Py_END_SUPPRESS_IPH
Victor Stinner1db9e7b2014-07-29 22:32:47 +02001574 if (flags < 0) {
1575 PyErr_SetFromErrno(PyExc_OSError);
1576 return -1;
1577 }
1578
1579 return !(flags & O_NONBLOCK);
1580}
1581
1582/* Set the blocking mode of the specified file descriptor.
1583
1584 Set the O_NONBLOCK flag if blocking is False, clear the O_NONBLOCK flag
1585 otherwise.
1586
1587 Return 0 on success, raise an exception and return -1 on error. */
1588int
1589_Py_set_blocking(int fd, int blocking)
1590{
1591#if defined(HAVE_SYS_IOCTL_H) && defined(FIONBIO)
1592 int arg = !blocking;
1593 if (ioctl(fd, FIONBIO, &arg) < 0)
1594 goto error;
1595#else
1596 int flags, res;
1597
Steve Dower8fc89802015-04-12 00:26:27 -04001598 _Py_BEGIN_SUPPRESS_IPH
Victor Stinner1db9e7b2014-07-29 22:32:47 +02001599 flags = fcntl(fd, F_GETFL, 0);
Steve Dower8fc89802015-04-12 00:26:27 -04001600 if (flags >= 0) {
1601 if (blocking)
1602 flags = flags & (~O_NONBLOCK);
1603 else
1604 flags = flags | O_NONBLOCK;
Victor Stinner1db9e7b2014-07-29 22:32:47 +02001605
Steve Dower8fc89802015-04-12 00:26:27 -04001606 res = fcntl(fd, F_SETFL, flags);
1607 } else {
1608 res = -1;
1609 }
1610 _Py_END_SUPPRESS_IPH
Victor Stinner1db9e7b2014-07-29 22:32:47 +02001611
Victor Stinner1db9e7b2014-07-29 22:32:47 +02001612 if (res < 0)
1613 goto error;
1614#endif
1615 return 0;
1616
1617error:
1618 PyErr_SetFromErrno(PyExc_OSError);
1619 return -1;
1620}
1621#endif
1622
Steve Dower8fc89802015-04-12 00:26:27 -04001623#if defined _MSC_VER && _MSC_VER >= 1400 && _MSC_VER < 1900
Steve Dowerd81431f2015-03-06 14:47:02 -08001624/* Legacy implementation of _PyVerify_fd while transitioning to
1625 * MSVC 14.0. This should eventually be removed. (issue23524)
1626 */
1627
1628/* Microsoft CRT in VS2005 and higher will verify that a filehandle is
1629 * valid and raise an assertion if it isn't.
1630 * Normally, an invalid fd is likely to be a C program error and therefore
1631 * an assertion can be useful, but it does contradict the POSIX standard
1632 * which for write(2) states:
1633 * "Otherwise, -1 shall be returned and errno set to indicate the error."
1634 * "[EBADF] The fildes argument is not a valid file descriptor open for
1635 * writing."
1636 * Furthermore, python allows the user to enter any old integer
1637 * as a fd and should merely raise a python exception on error.
1638 * The Microsoft CRT doesn't provide an official way to check for the
1639 * validity of a file descriptor, but we can emulate its internal behaviour
1640 * by using the exported __pinfo data member and knowledge of the
1641 * internal structures involved.
1642 * The structures below must be updated for each version of visual studio
1643 * according to the file internal.h in the CRT source, until MS comes
1644 * up with a less hacky way to do this.
1645 * (all of this is to avoid globally modifying the CRT behaviour using
1646 * _set_invalid_parameter_handler() and _CrtSetReportMode())
1647 */
1648/* The actual size of the structure is determined at runtime.
1649 * Only the first items must be present.
1650 */
1651typedef struct {
1652 intptr_t osfhnd;
1653 char osfile;
1654} my_ioinfo;
1655
1656extern __declspec(dllimport) char * __pioinfo[];
1657#define IOINFO_L2E 5
1658#define IOINFO_ARRAYS 64
1659#define IOINFO_ARRAY_ELTS (1 << IOINFO_L2E)
1660#define _NHANDLE_ (IOINFO_ARRAYS * IOINFO_ARRAY_ELTS)
1661#define FOPEN 0x01
1662#define _NO_CONSOLE_FILENO (intptr_t)-2
1663
1664/* This function emulates what the windows CRT does to validate file handles */
1665int
1666_PyVerify_fd(int fd)
1667{
1668 const int i1 = fd >> IOINFO_L2E;
1669 const int i2 = fd & ((1 << IOINFO_L2E) - 1);
1670
1671 static size_t sizeof_ioinfo = 0;
1672
1673 /* Determine the actual size of the ioinfo structure,
1674 * as used by the CRT loaded in memory
1675 */
1676 if (sizeof_ioinfo == 0 && __pioinfo[0] != NULL) {
1677 sizeof_ioinfo = _msize(__pioinfo[0]) / IOINFO_ARRAY_ELTS;
1678 }
1679 if (sizeof_ioinfo == 0) {
1680 /* This should not happen... */
1681 goto fail;
1682 }
1683
1684 /* See that it isn't a special CLEAR fileno */
1685 if (fd != _NO_CONSOLE_FILENO) {
1686 /* Microsoft CRT would check that 0<=fd<_nhandle but we can't do that. Instead
1687 * we check pointer validity and other info
1688 */
1689 if (0 <= i1 && i1 < IOINFO_ARRAYS && __pioinfo[i1] != NULL) {
1690 /* finally, check that the file is open */
1691 my_ioinfo* info = (my_ioinfo*)(__pioinfo[i1] + i2 * sizeof_ioinfo);
1692 if (info->osfile & FOPEN) {
1693 return 1;
1694 }
1695 }
1696 }
1697 fail:
1698 errno = EBADF;
1699 return 0;
1700}
1701
Steve Dower8fc89802015-04-12 00:26:27 -04001702#endif /* defined _MSC_VER && _MSC_VER >= 1400 && _MSC_VER < 1900 */