blob: b9d909705cf5ca2785f6baa4b1e425102535c437 [file] [log] [blame]
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001/* File object implementation */
2
Martin v. Löwis18e16552006-02-15 17:27:45 +00003#define PY_SSIZE_T_CLEAN
Guido van Rossumc0b618a1997-05-02 03:12:38 +00004#include "Python.h"
Guido van Rossumb6775db1994-08-01 11:34:53 +00005#include "structmember.h"
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00006
Martin v. Löwis0e8bd7e2006-06-10 12:23:46 +00007#ifdef HAVE_SYS_TYPES_H
Guido van Rossum41498431999-01-07 22:09:51 +00008#include <sys/types.h>
Martin v. Löwis0e8bd7e2006-06-10 12:23:46 +00009#endif /* HAVE_SYS_TYPES_H */
Guido van Rossum41498431999-01-07 22:09:51 +000010
Martin v. Löwis6238d2b2002-06-30 15:26:10 +000011#ifdef MS_WINDOWS
Guido van Rossumb8199141997-05-06 15:23:24 +000012#define fileno _fileno
Tim Petersfb05db22002-03-11 00:24:00 +000013/* can simulate truncate with Win32 API functions; see file_truncate */
Guido van Rossumb8199141997-05-06 15:23:24 +000014#define HAVE_FTRUNCATE
Tim Peters7a1f9172002-07-14 22:14:19 +000015#define WIN32_LEAN_AND_MEAN
Tim Petersfb05db22002-03-11 00:24:00 +000016#include <windows.h>
Guido van Rossumb8199141997-05-06 15:23:24 +000017#endif
18
Mark Hammondc2e85bd2002-10-03 05:10:39 +000019#ifdef _MSC_VER
20/* Need GetVersion to see if on NT so safe to use _wfopen */
21#define WIN32_LEAN_AND_MEAN
22#include <windows.h>
23#endif /* _MSC_VER */
24
Andrew MacIntyrec4874392002-02-26 11:36:35 +000025#if defined(PYOS_OS2) && defined(PYCC_GCC)
26#include <io.h>
27#endif
28
Gregory P. Smithdd96db62008-06-09 04:58:54 +000029#define BUF(v) PyString_AS_STRING((PyStringObject *)v)
Guido van Rossumce5ba841991-03-06 13:06:18 +000030
Guido van Rossumff7e83d1999-08-27 20:39:37 +000031#ifndef DONT_HAVE_ERRNO_H
Guido van Rossumf1dc5661993-07-05 10:31:29 +000032#include <errno.h>
Guido van Rossumff7e83d1999-08-27 20:39:37 +000033#endif
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000034
Jack Jansen7b8c7542002-04-14 20:12:41 +000035#ifdef HAVE_GETC_UNLOCKED
36#define GETC(f) getc_unlocked(f)
37#define FLOCKFILE(f) flockfile(f)
38#define FUNLOCKFILE(f) funlockfile(f)
39#else
40#define GETC(f) getc(f)
41#define FLOCKFILE(f)
42#define FUNLOCKFILE(f)
43#endif
44
Jack Jansen7b8c7542002-04-14 20:12:41 +000045/* Bits in f_newlinetypes */
Antoine Pitrouc7c96a92010-05-09 15:15:40 +000046#define NEWLINE_UNKNOWN 0 /* No newline seen, yet */
47#define NEWLINE_CR 1 /* \r newline seen */
48#define NEWLINE_LF 2 /* \n newline seen */
49#define NEWLINE_CRLF 4 /* \r\n newline seen */
Trent Mickf29f47b2000-08-11 19:02:59 +000050
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +000051/*
52 * These macros release the GIL while preventing the f_close() function being
53 * called in the interval between them. For that purpose, a running total of
54 * the number of currently running unlocked code sections is kept in
55 * the unlocked_count field of the PyFileObject. The close() method raises
56 * an IOError if that field is non-zero. See issue #815646, #595601.
57 */
58
59#define FILE_BEGIN_ALLOW_THREADS(fobj) \
60{ \
Antoine Pitrouc7c96a92010-05-09 15:15:40 +000061 fobj->unlocked_count++; \
62 Py_BEGIN_ALLOW_THREADS
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +000063
64#define FILE_END_ALLOW_THREADS(fobj) \
Antoine Pitrouc7c96a92010-05-09 15:15:40 +000065 Py_END_ALLOW_THREADS \
66 fobj->unlocked_count--; \
67 assert(fobj->unlocked_count >= 0); \
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +000068}
69
70#define FILE_ABORT_ALLOW_THREADS(fobj) \
Antoine Pitrouc7c96a92010-05-09 15:15:40 +000071 Py_BLOCK_THREADS \
72 fobj->unlocked_count--; \
73 assert(fobj->unlocked_count >= 0);
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +000074
Anthony Baxterac6bd462006-04-13 02:06:09 +000075#ifdef __cplusplus
76extern "C" {
77#endif
78
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000079FILE *
Fred Drakefd99de62000-07-09 05:02:18 +000080PyFile_AsFile(PyObject *f)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000081{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +000082 if (f == NULL || !PyFile_Check(f))
83 return NULL;
84 else
85 return ((PyFileObject *)f)->f_fp;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000086}
87
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +000088void PyFile_IncUseCount(PyFileObject *fobj)
89{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +000090 fobj->unlocked_count++;
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +000091}
92
93void PyFile_DecUseCount(PyFileObject *fobj)
94{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +000095 fobj->unlocked_count--;
96 assert(fobj->unlocked_count >= 0);
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +000097}
98
Guido van Rossumc0b618a1997-05-02 03:12:38 +000099PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000100PyFile_Name(PyObject *f)
Guido van Rossumdb3165e1993-10-18 17:06:59 +0000101{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000102 if (f == NULL || !PyFile_Check(f))
103 return NULL;
104 else
105 return ((PyFileObject *)f)->f_name;
Guido van Rossumdb3165e1993-10-18 17:06:59 +0000106}
107
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000108/* This is a safe wrapper around PyObject_Print to print to the FILE
109 of a PyFileObject. PyObject_Print releases the GIL but knows nothing
110 about PyFileObject. */
111static int
112file_PyObject_Print(PyObject *op, PyFileObject *f, int flags)
113{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000114 int result;
115 PyFile_IncUseCount(f);
116 result = PyObject_Print(op, f->f_fp, flags);
117 PyFile_DecUseCount(f);
118 return result;
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000119}
120
Neil Schemenauered19b882002-03-23 02:06:50 +0000121/* On Unix, fopen will succeed for directories.
122 In Python, there should be no file objects referring to
123 directories, so we need a check. */
124
125static PyFileObject*
126dircheck(PyFileObject* f)
127{
128#if defined(HAVE_FSTAT) && defined(S_IFDIR) && defined(EISDIR)
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000129 struct stat buf;
130 if (f->f_fp == NULL)
131 return f;
132 if (fstat(fileno(f->f_fp), &buf) == 0 &&
133 S_ISDIR(buf.st_mode)) {
134 char *msg = strerror(EISDIR);
135 PyObject *exc = PyObject_CallFunction(PyExc_IOError, "(isO)",
136 EISDIR, msg, f->f_name);
137 PyErr_SetObject(PyExc_IOError, exc);
138 Py_XDECREF(exc);
139 return NULL;
140 }
Neil Schemenauered19b882002-03-23 02:06:50 +0000141#endif
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000142 return f;
Neil Schemenauered19b882002-03-23 02:06:50 +0000143}
144
Tim Peters59c9a642001-09-13 05:38:56 +0000145
146static PyObject *
Nicholas Bastinabce8a62004-03-21 20:24:07 +0000147fill_file_fields(PyFileObject *f, FILE *fp, PyObject *name, char *mode,
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000148 int (*close)(FILE *))
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000149{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000150 assert(name != NULL);
151 assert(f != NULL);
152 assert(PyFile_Check(f));
153 assert(f->f_fp == NULL);
Tim Peters44410012001-09-14 03:26:08 +0000154
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000155 Py_DECREF(f->f_name);
156 Py_DECREF(f->f_mode);
157 Py_DECREF(f->f_encoding);
158 Py_DECREF(f->f_errors);
Nicholas Bastinabce8a62004-03-21 20:24:07 +0000159
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000160 Py_INCREF(name);
161 f->f_name = name;
Nicholas Bastinabce8a62004-03-21 20:24:07 +0000162
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000163 f->f_mode = PyString_FromString(mode);
Tim Peters44410012001-09-14 03:26:08 +0000164
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000165 f->f_close = close;
166 f->f_softspace = 0;
167 f->f_binary = strchr(mode,'b') != NULL;
168 f->f_buf = NULL;
169 f->f_univ_newline = (strchr(mode, 'U') != NULL);
170 f->f_newlinetypes = NEWLINE_UNKNOWN;
171 f->f_skipnextlf = 0;
172 Py_INCREF(Py_None);
173 f->f_encoding = Py_None;
174 Py_INCREF(Py_None);
175 f->f_errors = Py_None;
176 f->readable = f->writable = 0;
177 if (strchr(mode, 'r') != NULL || f->f_univ_newline)
178 f->readable = 1;
179 if (strchr(mode, 'w') != NULL || strchr(mode, 'a') != NULL)
180 f->writable = 1;
181 if (strchr(mode, '+') != NULL)
182 f->readable = f->writable = 1;
Tim Petersf1827cf2003-09-07 03:30:18 +0000183
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000184 if (f->f_mode == NULL)
185 return NULL;
186 f->f_fp = fp;
187 f = dircheck(f);
188 return (PyObject *) f;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000189}
190
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000191/* check for known incorrect mode strings - problem is, platforms are
192 free to accept any mode characters they like and are supposed to
193 ignore stuff they don't understand... write or append mode with
Georg Brandl7b90e162006-05-18 07:01:27 +0000194 universal newline support is expressly forbidden by PEP 278.
195 Additionally, remove the 'U' from the mode string as platforms
Kristján Valur Jónsson0a440d42007-04-26 09:15:08 +0000196 won't know what it is. Non-zero return signals an exception */
197int
198_PyFile_SanitizeMode(char *mode)
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000199{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000200 char *upos;
201 size_t len = strlen(mode);
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000202
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000203 if (!len) {
204 PyErr_SetString(PyExc_ValueError, "empty mode string");
205 return -1;
206 }
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000207
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000208 upos = strchr(mode, 'U');
209 if (upos) {
210 memmove(upos, upos+1, len-(upos-mode)); /* incl null char */
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000211
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000212 if (mode[0] == 'w' || mode[0] == 'a') {
213 PyErr_Format(PyExc_ValueError, "universal newline "
214 "mode can only be used with modes "
215 "starting with 'r'");
216 return -1;
217 }
Georg Brandl7b90e162006-05-18 07:01:27 +0000218
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000219 if (mode[0] != 'r') {
220 memmove(mode+1, mode, strlen(mode)+1);
221 mode[0] = 'r';
222 }
Georg Brandl7b90e162006-05-18 07:01:27 +0000223
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000224 if (!strchr(mode, 'b')) {
225 memmove(mode+2, mode+1, strlen(mode));
226 mode[1] = 'b';
227 }
228 } else if (mode[0] != 'r' && mode[0] != 'w' && mode[0] != 'a') {
229 PyErr_Format(PyExc_ValueError, "mode string must begin with "
230 "one of 'r', 'w', 'a' or 'U', not '%.200s'", mode);
231 return -1;
232 }
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000233
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000234 return 0;
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000235}
236
Tim Peters59c9a642001-09-13 05:38:56 +0000237static PyObject *
238open_the_file(PyFileObject *f, char *name, char *mode)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000239{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000240 char *newmode;
241 assert(f != NULL);
242 assert(PyFile_Check(f));
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000243#ifdef MS_WINDOWS
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000244 /* windows ignores the passed name in order to support Unicode */
245 assert(f->f_name != NULL);
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000246#else
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000247 assert(name != NULL);
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000248#endif
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000249 assert(mode != NULL);
250 assert(f->f_fp == NULL);
Tim Peters59c9a642001-09-13 05:38:56 +0000251
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000252 /* probably need to replace 'U' by 'rb' */
253 newmode = PyMem_MALLOC(strlen(mode) + 3);
254 if (!newmode) {
255 PyErr_NoMemory();
256 return NULL;
257 }
258 strcpy(newmode, mode);
Georg Brandl7b90e162006-05-18 07:01:27 +0000259
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000260 if (_PyFile_SanitizeMode(newmode)) {
261 f = NULL;
262 goto cleanup;
263 }
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000264
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000265 /* rexec.py can't stop a user from getting the file() constructor --
266 all they have to do is get *any* file object f, and then do
267 type(f). Here we prevent them from doing damage with it. */
268 if (PyEval_GetRestricted()) {
269 PyErr_SetString(PyExc_IOError,
270 "file() constructor not accessible in restricted mode");
271 f = NULL;
272 goto cleanup;
273 }
274 errno = 0;
Skip Montanaro51ffac62004-06-11 04:49:03 +0000275
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000276#ifdef MS_WINDOWS
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000277 if (PyUnicode_Check(f->f_name)) {
278 PyObject *wmode;
279 wmode = PyUnicode_DecodeASCII(newmode, strlen(newmode), NULL);
280 if (f->f_name && wmode) {
281 FILE_BEGIN_ALLOW_THREADS(f)
282 /* PyUnicode_AS_UNICODE OK without thread
283 lock as it is a simple dereference. */
284 f->f_fp = _wfopen(PyUnicode_AS_UNICODE(f->f_name),
285 PyUnicode_AS_UNICODE(wmode));
286 FILE_END_ALLOW_THREADS(f)
287 }
288 Py_XDECREF(wmode);
289 }
Skip Montanaro51ffac62004-06-11 04:49:03 +0000290#endif
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000291 if (NULL == f->f_fp && NULL != name) {
292 FILE_BEGIN_ALLOW_THREADS(f)
293 f->f_fp = fopen(name, newmode);
294 FILE_END_ALLOW_THREADS(f)
295 }
Skip Montanaro51ffac62004-06-11 04:49:03 +0000296
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000297 if (f->f_fp == NULL) {
Kristján Valur Jónsson74c3ea02006-07-03 14:59:05 +0000298#if defined _MSC_VER && (_MSC_VER < 1400 || !defined(__STDC_SECURE_LIB__))
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000299 /* MSVC 6 (Microsoft) leaves errno at 0 for bad mode strings,
300 * across all Windows flavors. When it sets EINVAL varies
301 * across Windows flavors, the exact conditions aren't
302 * documented, and the answer lies in the OS's implementation
303 * of Win32's CreateFile function (whose source is secret).
304 * Seems the best we can do is map EINVAL to ENOENT.
305 * Starting with Visual Studio .NET 2005, EINVAL is correctly
306 * set by our CRT error handler (set in exceptions.c.)
307 */
308 if (errno == 0) /* bad mode string */
309 errno = EINVAL;
310 else if (errno == EINVAL) /* unknown, but not a mode string */
311 errno = ENOENT;
Tim Peters2ea91112002-04-08 04:13:12 +0000312#endif
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000313 /* EINVAL is returned when an invalid filename or
314 * an invalid mode is supplied. */
315 if (errno == EINVAL) {
316 PyObject *v;
317 char message[100];
318 PyOS_snprintf(message, 100,
319 "invalid mode ('%.50s') or filename", mode);
320 v = Py_BuildValue("(isO)", errno, message, f->f_name);
321 if (v != NULL) {
322 PyErr_SetObject(PyExc_IOError, v);
323 Py_DECREF(v);
324 }
325 }
326 else
327 PyErr_SetFromErrnoWithFilenameObject(PyExc_IOError, f->f_name);
328 f = NULL;
329 }
330 if (f != NULL)
331 f = dircheck(f);
Georg Brandl7b90e162006-05-18 07:01:27 +0000332
333cleanup:
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000334 PyMem_FREE(newmode);
Georg Brandl7b90e162006-05-18 07:01:27 +0000335
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000336 return (PyObject *)f;
Tim Peters59c9a642001-09-13 05:38:56 +0000337}
338
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000339static PyObject *
340close_the_file(PyFileObject *f)
341{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000342 int sts = 0;
343 int (*local_close)(FILE *);
344 FILE *local_fp = f->f_fp;
345 if (local_fp != NULL) {
346 local_close = f->f_close;
347 if (local_close != NULL && f->unlocked_count > 0) {
348 if (f->ob_refcnt > 0) {
349 PyErr_SetString(PyExc_IOError,
350 "close() called during concurrent "
351 "operation on the same file object.");
352 } else {
353 /* This should not happen unless someone is
354 * carelessly playing with the PyFileObject
355 * struct fields and/or its associated FILE
356 * pointer. */
357 PyErr_SetString(PyExc_SystemError,
358 "PyFileObject locking error in "
359 "destructor (refcnt <= 0 at close).");
360 }
361 return NULL;
362 }
363 /* NULL out the FILE pointer before releasing the GIL, because
364 * it will not be valid anymore after the close() function is
365 * called. */
366 f->f_fp = NULL;
367 if (local_close != NULL) {
368 Py_BEGIN_ALLOW_THREADS
369 errno = 0;
370 sts = (*local_close)(local_fp);
371 Py_END_ALLOW_THREADS
372 if (sts == EOF)
373 return PyErr_SetFromErrno(PyExc_IOError);
374 if (sts != 0)
375 return PyInt_FromLong((long)sts);
376 }
377 }
378 Py_RETURN_NONE;
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000379}
380
Tim Peters59c9a642001-09-13 05:38:56 +0000381PyObject *
382PyFile_FromFile(FILE *fp, char *name, char *mode, int (*close)(FILE *))
383{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000384 PyFileObject *f = (PyFileObject *)PyFile_Type.tp_new(&PyFile_Type,
385 NULL, NULL);
386 if (f != NULL) {
387 PyObject *o_name = PyString_FromString(name);
388 if (o_name == NULL)
389 return NULL;
390 if (fill_file_fields(f, fp, o_name, mode, close) == NULL) {
391 Py_DECREF(f);
392 f = NULL;
393 }
394 Py_DECREF(o_name);
395 }
396 return (PyObject *) f;
Tim Peters59c9a642001-09-13 05:38:56 +0000397}
398
399PyObject *
400PyFile_FromString(char *name, char *mode)
401{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000402 extern int fclose(FILE *);
403 PyFileObject *f;
Tim Peters59c9a642001-09-13 05:38:56 +0000404
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000405 f = (PyFileObject *)PyFile_FromFile((FILE *)NULL, name, mode, fclose);
406 if (f != NULL) {
407 if (open_the_file(f, name, mode) == NULL) {
408 Py_DECREF(f);
409 f = NULL;
410 }
411 }
412 return (PyObject *)f;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000413}
414
Guido van Rossumb6775db1994-08-01 11:34:53 +0000415void
Fred Drakefd99de62000-07-09 05:02:18 +0000416PyFile_SetBufSize(PyObject *f, int bufsize)
Guido van Rossumb6775db1994-08-01 11:34:53 +0000417{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000418 PyFileObject *file = (PyFileObject *)f;
419 if (bufsize >= 0) {
420 int type;
421 switch (bufsize) {
422 case 0:
423 type = _IONBF;
424 break;
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000425#ifdef HAVE_SETVBUF
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000426 case 1:
427 type = _IOLBF;
428 bufsize = BUFSIZ;
429 break;
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000430#endif
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000431 default:
432 type = _IOFBF;
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000433#ifndef HAVE_SETVBUF
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000434 bufsize = BUFSIZ;
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000435#endif
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000436 break;
437 }
438 fflush(file->f_fp);
439 if (type == _IONBF) {
440 PyMem_Free(file->f_setbuf);
441 file->f_setbuf = NULL;
442 } else {
443 file->f_setbuf = (char *)PyMem_Realloc(file->f_setbuf,
444 bufsize);
445 }
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000446#ifdef HAVE_SETVBUF
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000447 setvbuf(file->f_fp, file->f_setbuf, type, bufsize);
Guido van Rossumf8b4de01998-03-06 15:32:40 +0000448#else /* !HAVE_SETVBUF */
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000449 setbuf(file->f_fp, file->f_setbuf);
Guido van Rossumf8b4de01998-03-06 15:32:40 +0000450#endif /* !HAVE_SETVBUF */
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000451 }
Guido van Rossumb6775db1994-08-01 11:34:53 +0000452}
453
Martin v. Löwis5467d4c2003-05-10 07:10:12 +0000454/* Set the encoding used to output Unicode strings.
Martin v. Löwis99815892008-06-01 07:20:46 +0000455 Return 1 on success, 0 on failure. */
Martin v. Löwis5467d4c2003-05-10 07:10:12 +0000456
457int
458PyFile_SetEncoding(PyObject *f, const char *enc)
459{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000460 return PyFile_SetEncodingAndErrors(f, enc, NULL);
Martin v. Löwis99815892008-06-01 07:20:46 +0000461}
462
463int
464PyFile_SetEncodingAndErrors(PyObject *f, const char *enc, char* errors)
465{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000466 PyFileObject *file = (PyFileObject*)f;
467 PyObject *str, *oerrors;
Thomas Woutersafea5292007-01-23 13:42:00 +0000468
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000469 assert(PyFile_Check(f));
470 str = PyString_FromString(enc);
471 if (!str)
472 return 0;
473 if (errors) {
474 oerrors = PyString_FromString(errors);
475 if (!oerrors) {
476 Py_DECREF(str);
477 return 0;
478 }
479 } else {
480 oerrors = Py_None;
481 Py_INCREF(Py_None);
482 }
483 Py_DECREF(file->f_encoding);
484 file->f_encoding = str;
485 Py_DECREF(file->f_errors);
486 file->f_errors = oerrors;
487 return 1;
Martin v. Löwis5467d4c2003-05-10 07:10:12 +0000488}
489
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000490static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000491err_closed(void)
Guido van Rossumd7297e61992-07-06 14:19:26 +0000492{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000493 PyErr_SetString(PyExc_ValueError, "I/O operation on closed file");
494 return NULL;
Guido van Rossumd7297e61992-07-06 14:19:26 +0000495}
496
Antoine Pitrou24837282010-02-05 17:11:32 +0000497static PyObject *
498err_mode(char *action)
499{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000500 PyErr_Format(PyExc_IOError, "File not open for %s", action);
501 return NULL;
Antoine Pitrou24837282010-02-05 17:11:32 +0000502}
503
Thomas Woutersc45251a2006-02-12 11:53:32 +0000504/* Refuse regular file I/O if there's data in the iteration-buffer.
505 * Mixing them would cause data to arrive out of order, as the read*
506 * methods don't use the iteration buffer. */
507static PyObject *
508err_iterbuffered(void)
509{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000510 PyErr_SetString(PyExc_ValueError,
511 "Mixing iteration and read methods would lose data");
512 return NULL;
Thomas Woutersc45251a2006-02-12 11:53:32 +0000513}
514
Neal Norwitzd8b995f2002-08-06 21:50:54 +0000515static void drop_readahead(PyFileObject *);
Guido van Rossum7a6e9592002-08-06 15:55:28 +0000516
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000517/* Methods */
518
519static void
Fred Drakefd99de62000-07-09 05:02:18 +0000520file_dealloc(PyFileObject *f)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000521{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000522 PyObject *ret;
523 if (f->weakreflist != NULL)
524 PyObject_ClearWeakRefs((PyObject *) f);
525 ret = close_the_file(f);
526 if (!ret) {
527 PySys_WriteStderr("close failed in file object destructor:\n");
528 PyErr_Print();
529 }
530 else {
531 Py_DECREF(ret);
532 }
533 PyMem_Free(f->f_setbuf);
534 Py_XDECREF(f->f_name);
535 Py_XDECREF(f->f_mode);
536 Py_XDECREF(f->f_encoding);
537 Py_XDECREF(f->f_errors);
538 drop_readahead(f);
539 Py_TYPE(f)->tp_free((PyObject *)f);
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000540}
541
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000542static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000543file_repr(PyFileObject *f)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000544{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000545 if (PyUnicode_Check(f->f_name)) {
Martin v. Löwis0073f2e2002-11-21 23:52:35 +0000546#ifdef Py_USING_UNICODE
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000547 PyObject *ret = NULL;
548 PyObject *name = PyUnicode_AsUnicodeEscapeString(f->f_name);
549 const char *name_str = name ? PyString_AsString(name) : "?";
550 ret = PyString_FromFormat("<%s file u'%s', mode '%s' at %p>",
551 f->f_fp == NULL ? "closed" : "open",
552 name_str,
553 PyString_AsString(f->f_mode),
554 f);
555 Py_XDECREF(name);
556 return ret;
Martin v. Löwis0073f2e2002-11-21 23:52:35 +0000557#endif
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000558 } else {
559 return PyString_FromFormat("<%s file '%s', mode '%s' at %p>",
560 f->f_fp == NULL ? "closed" : "open",
561 PyString_AsString(f->f_name),
562 PyString_AsString(f->f_mode),
563 f);
564 }
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000565}
566
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000567static PyObject *
Martin v. Löwise3eb1f22001-08-16 13:15:00 +0000568file_close(PyFileObject *f)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000569{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000570 PyObject *sts = close_the_file(f);
571 PyMem_Free(f->f_setbuf);
572 f->f_setbuf = NULL;
573 return sts;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000574}
575
Trent Mickf29f47b2000-08-11 19:02:59 +0000576
Guido van Rossumb8552162001-09-05 14:58:11 +0000577/* Our very own off_t-like type, 64-bit if possible */
578#if !defined(HAVE_LARGEFILE_SUPPORT)
579typedef off_t Py_off_t;
580#elif SIZEOF_OFF_T >= 8
581typedef off_t Py_off_t;
582#elif SIZEOF_FPOS_T >= 8
Guido van Rossum4f53da02001-03-01 18:26:53 +0000583typedef fpos_t Py_off_t;
584#else
Guido van Rossumb8552162001-09-05 14:58:11 +0000585#error "Large file support, but neither off_t nor fpos_t is large enough."
Guido van Rossum4f53da02001-03-01 18:26:53 +0000586#endif
587
588
Trent Mickf29f47b2000-08-11 19:02:59 +0000589/* a portable fseek() function
590 return 0 on success, non-zero on failure (with errno set) */
Guido van Rossumf68d8e52001-04-14 17:55:09 +0000591static int
Guido van Rossum4f53da02001-03-01 18:26:53 +0000592_portable_fseek(FILE *fp, Py_off_t offset, int whence)
Trent Mickf29f47b2000-08-11 19:02:59 +0000593{
Guido van Rossumb8552162001-09-05 14:58:11 +0000594#if !defined(HAVE_LARGEFILE_SUPPORT)
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000595 return fseek(fp, offset, whence);
Guido van Rossumb8552162001-09-05 14:58:11 +0000596#elif defined(HAVE_FSEEKO) && SIZEOF_OFF_T >= 8
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000597 return fseeko(fp, offset, whence);
Trent Mickf29f47b2000-08-11 19:02:59 +0000598#elif defined(HAVE_FSEEK64)
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000599 return fseek64(fp, offset, whence);
Fred Drakedb810ac2000-10-06 20:42:33 +0000600#elif defined(__BEOS__)
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000601 return _fseek(fp, offset, whence);
Guido van Rossumb8552162001-09-05 14:58:11 +0000602#elif SIZEOF_FPOS_T >= 8
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000603 /* lacking a 64-bit capable fseek(), use a 64-bit capable fsetpos()
604 and fgetpos() to implement fseek()*/
605 fpos_t pos;
606 switch (whence) {
607 case SEEK_END:
Guido van Rossum8b4e43e2001-09-10 20:43:35 +0000608#ifdef MS_WINDOWS
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000609 fflush(fp);
610 if (_lseeki64(fileno(fp), 0, 2) == -1)
611 return -1;
Guido van Rossum8b4e43e2001-09-10 20:43:35 +0000612#else
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000613 if (fseek(fp, 0, SEEK_END) != 0)
614 return -1;
Guido van Rossum8b4e43e2001-09-10 20:43:35 +0000615#endif
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000616 /* fall through */
617 case SEEK_CUR:
618 if (fgetpos(fp, &pos) != 0)
619 return -1;
620 offset += pos;
621 break;
622 /* case SEEK_SET: break; */
623 }
624 return fsetpos(fp, &offset);
Trent Mickf29f47b2000-08-11 19:02:59 +0000625#else
Guido van Rossumb8552162001-09-05 14:58:11 +0000626#error "Large file support, but no way to fseek."
Trent Mickf29f47b2000-08-11 19:02:59 +0000627#endif
628}
629
630
631/* a portable ftell() function
632 Return -1 on failure with errno set appropriately, current file
633 position on success */
Guido van Rossumf68d8e52001-04-14 17:55:09 +0000634static Py_off_t
Fred Drake8ce159a2000-08-31 05:18:54 +0000635_portable_ftell(FILE* fp)
Trent Mickf29f47b2000-08-11 19:02:59 +0000636{
Guido van Rossumb8552162001-09-05 14:58:11 +0000637#if !defined(HAVE_LARGEFILE_SUPPORT)
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000638 return ftell(fp);
Guido van Rossumb8552162001-09-05 14:58:11 +0000639#elif defined(HAVE_FTELLO) && SIZEOF_OFF_T >= 8
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000640 return ftello(fp);
Guido van Rossumb8552162001-09-05 14:58:11 +0000641#elif defined(HAVE_FTELL64)
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000642 return ftell64(fp);
Guido van Rossumb8552162001-09-05 14:58:11 +0000643#elif SIZEOF_FPOS_T >= 8
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000644 fpos_t pos;
645 if (fgetpos(fp, &pos) != 0)
646 return -1;
647 return pos;
Trent Mickf29f47b2000-08-11 19:02:59 +0000648#else
Guido van Rossumb8552162001-09-05 14:58:11 +0000649#error "Large file support, but no way to ftell."
Trent Mickf29f47b2000-08-11 19:02:59 +0000650#endif
651}
652
653
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000654static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000655file_seek(PyFileObject *f, PyObject *args)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000656{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000657 int whence;
658 int ret;
659 Py_off_t offset;
660 PyObject *offobj, *off_index;
Tim Peters86821b22001-01-07 21:19:34 +0000661
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000662 if (f->f_fp == NULL)
663 return err_closed();
664 drop_readahead(f);
665 whence = 0;
666 if (!PyArg_ParseTuple(args, "O|i:seek", &offobj, &whence))
667 return NULL;
668 off_index = PyNumber_Index(offobj);
669 if (!off_index) {
670 if (!PyFloat_Check(offobj))
671 return NULL;
672 /* Deprecated in 2.6 */
673 PyErr_Clear();
674 if (PyErr_WarnEx(PyExc_DeprecationWarning,
675 "integer argument expected, got float",
676 1) < 0)
677 return NULL;
678 off_index = offobj;
679 Py_INCREF(offobj);
680 }
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000681#if !defined(HAVE_LARGEFILE_SUPPORT)
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000682 offset = PyInt_AsLong(off_index);
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000683#else
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000684 offset = PyLong_Check(off_index) ?
685 PyLong_AsLongLong(off_index) : PyInt_AsLong(off_index);
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000686#endif
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000687 Py_DECREF(off_index);
688 if (PyErr_Occurred())
689 return NULL;
Tim Peters86821b22001-01-07 21:19:34 +0000690
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000691 FILE_BEGIN_ALLOW_THREADS(f)
692 errno = 0;
693 ret = _portable_fseek(f->f_fp, offset, whence);
694 FILE_END_ALLOW_THREADS(f)
Trent Mickf29f47b2000-08-11 19:02:59 +0000695
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000696 if (ret != 0) {
697 PyErr_SetFromErrno(PyExc_IOError);
698 clearerr(f->f_fp);
699 return NULL;
700 }
701 f->f_skipnextlf = 0;
702 Py_INCREF(Py_None);
703 return Py_None;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000704}
705
Trent Mickf29f47b2000-08-11 19:02:59 +0000706
Guido van Rossumd7047b31995-01-02 19:07:15 +0000707#ifdef HAVE_FTRUNCATE
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000708static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000709file_truncate(PyFileObject *f, PyObject *args)
Guido van Rossumd7047b31995-01-02 19:07:15 +0000710{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000711 Py_off_t newsize;
712 PyObject *newsizeobj = NULL;
713 Py_off_t initialpos;
714 int ret;
Tim Peters86821b22001-01-07 21:19:34 +0000715
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000716 if (f->f_fp == NULL)
717 return err_closed();
718 if (!f->writable)
719 return err_mode("writing");
720 if (!PyArg_UnpackTuple(args, "truncate", 0, 1, &newsizeobj))
721 return NULL;
Tim Petersfb05db22002-03-11 00:24:00 +0000722
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000723 /* Get current file position. If the file happens to be open for
724 * update and the last operation was an input operation, C doesn't
725 * define what the later fflush() will do, but we promise truncate()
726 * won't change the current position (and fflush() *does* change it
727 * then at least on Windows). The easiest thing is to capture
728 * current pos now and seek back to it at the end.
729 */
730 FILE_BEGIN_ALLOW_THREADS(f)
731 errno = 0;
732 initialpos = _portable_ftell(f->f_fp);
733 FILE_END_ALLOW_THREADS(f)
734 if (initialpos == -1)
735 goto onioerror;
Tim Petersf1827cf2003-09-07 03:30:18 +0000736
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000737 /* Set newsize to current postion if newsizeobj NULL, else to the
738 * specified value.
739 */
740 if (newsizeobj != NULL) {
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000741#if !defined(HAVE_LARGEFILE_SUPPORT)
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000742 newsize = PyInt_AsLong(newsizeobj);
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000743#else
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000744 newsize = PyLong_Check(newsizeobj) ?
745 PyLong_AsLongLong(newsizeobj) :
746 PyInt_AsLong(newsizeobj);
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000747#endif
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000748 if (PyErr_Occurred())
749 return NULL;
750 }
751 else /* default to current position */
752 newsize = initialpos;
Tim Petersfb05db22002-03-11 00:24:00 +0000753
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000754 /* Flush the stream. We're mixing stream-level I/O with lower-level
755 * I/O, and a flush may be necessary to synch both platform views
756 * of the current file state.
757 */
758 FILE_BEGIN_ALLOW_THREADS(f)
759 errno = 0;
760 ret = fflush(f->f_fp);
761 FILE_END_ALLOW_THREADS(f)
762 if (ret != 0)
763 goto onioerror;
Trent Mickf29f47b2000-08-11 19:02:59 +0000764
Martin v. Löwis6238d2b2002-06-30 15:26:10 +0000765#ifdef MS_WINDOWS
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000766 /* MS _chsize doesn't work if newsize doesn't fit in 32 bits,
767 so don't even try using it. */
768 {
769 HANDLE hFile;
Tim Petersfb05db22002-03-11 00:24:00 +0000770
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000771 /* Have to move current pos to desired endpoint on Windows. */
772 FILE_BEGIN_ALLOW_THREADS(f)
773 errno = 0;
774 ret = _portable_fseek(f->f_fp, newsize, SEEK_SET) != 0;
775 FILE_END_ALLOW_THREADS(f)
776 if (ret)
777 goto onioerror;
Tim Petersfb05db22002-03-11 00:24:00 +0000778
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000779 /* Truncate. Note that this may grow the file! */
780 FILE_BEGIN_ALLOW_THREADS(f)
781 errno = 0;
782 hFile = (HANDLE)_get_osfhandle(fileno(f->f_fp));
783 ret = hFile == (HANDLE)-1;
784 if (ret == 0) {
785 ret = SetEndOfFile(hFile) == 0;
786 if (ret)
787 errno = EACCES;
788 }
789 FILE_END_ALLOW_THREADS(f)
790 if (ret)
791 goto onioerror;
792 }
Trent Mickf29f47b2000-08-11 19:02:59 +0000793#else
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000794 FILE_BEGIN_ALLOW_THREADS(f)
795 errno = 0;
796 ret = ftruncate(fileno(f->f_fp), newsize);
797 FILE_END_ALLOW_THREADS(f)
798 if (ret != 0)
799 goto onioerror;
Martin v. Löwis6238d2b2002-06-30 15:26:10 +0000800#endif /* !MS_WINDOWS */
Tim Peters86821b22001-01-07 21:19:34 +0000801
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000802 /* Restore original file position. */
803 FILE_BEGIN_ALLOW_THREADS(f)
804 errno = 0;
805 ret = _portable_fseek(f->f_fp, initialpos, SEEK_SET) != 0;
806 FILE_END_ALLOW_THREADS(f)
807 if (ret)
808 goto onioerror;
Tim Petersf1827cf2003-09-07 03:30:18 +0000809
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000810 Py_INCREF(Py_None);
811 return Py_None;
Trent Mickf29f47b2000-08-11 19:02:59 +0000812
813onioerror:
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000814 PyErr_SetFromErrno(PyExc_IOError);
815 clearerr(f->f_fp);
816 return NULL;
Guido van Rossumd7047b31995-01-02 19:07:15 +0000817}
818#endif /* HAVE_FTRUNCATE */
819
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000820static PyObject *
Martin v. Löwise3eb1f22001-08-16 13:15:00 +0000821file_tell(PyFileObject *f)
Guido van Rossumce5ba841991-03-06 13:06:18 +0000822{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000823 Py_off_t pos;
Trent Mickf29f47b2000-08-11 19:02:59 +0000824
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000825 if (f->f_fp == NULL)
826 return err_closed();
827 FILE_BEGIN_ALLOW_THREADS(f)
828 errno = 0;
829 pos = _portable_ftell(f->f_fp);
830 FILE_END_ALLOW_THREADS(f)
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000831
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000832 if (pos == -1) {
833 PyErr_SetFromErrno(PyExc_IOError);
834 clearerr(f->f_fp);
835 return NULL;
836 }
837 if (f->f_skipnextlf) {
838 int c;
839 c = GETC(f->f_fp);
840 if (c == '\n') {
841 f->f_newlinetypes |= NEWLINE_CRLF;
842 pos++;
843 f->f_skipnextlf = 0;
844 } else if (c != EOF) ungetc(c, f->f_fp);
845 }
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000846#if !defined(HAVE_LARGEFILE_SUPPORT)
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000847 return PyInt_FromLong(pos);
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000848#else
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000849 return PyLong_FromLongLong(pos);
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000850#endif
Guido van Rossumce5ba841991-03-06 13:06:18 +0000851}
852
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000853static PyObject *
Martin v. Löwise3eb1f22001-08-16 13:15:00 +0000854file_fileno(PyFileObject *f)
Guido van Rossumed233a51992-06-23 09:07:03 +0000855{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000856 if (f->f_fp == NULL)
857 return err_closed();
858 return PyInt_FromLong((long) fileno(f->f_fp));
Guido van Rossumed233a51992-06-23 09:07:03 +0000859}
860
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000861static PyObject *
Martin v. Löwise3eb1f22001-08-16 13:15:00 +0000862file_flush(PyFileObject *f)
Guido van Rossumce5ba841991-03-06 13:06:18 +0000863{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000864 int res;
Tim Peters86821b22001-01-07 21:19:34 +0000865
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000866 if (f->f_fp == NULL)
867 return err_closed();
868 FILE_BEGIN_ALLOW_THREADS(f)
869 errno = 0;
870 res = fflush(f->f_fp);
871 FILE_END_ALLOW_THREADS(f)
872 if (res != 0) {
873 PyErr_SetFromErrno(PyExc_IOError);
874 clearerr(f->f_fp);
875 return NULL;
876 }
877 Py_INCREF(Py_None);
878 return Py_None;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000879}
880
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000881static PyObject *
Martin v. Löwise3eb1f22001-08-16 13:15:00 +0000882file_isatty(PyFileObject *f)
Guido van Rossuma1ab7fa1991-06-04 19:37:39 +0000883{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000884 long res;
885 if (f->f_fp == NULL)
886 return err_closed();
887 FILE_BEGIN_ALLOW_THREADS(f)
888 res = isatty((int)fileno(f->f_fp));
889 FILE_END_ALLOW_THREADS(f)
890 return PyBool_FromLong(res);
Guido van Rossuma1ab7fa1991-06-04 19:37:39 +0000891}
892
Guido van Rossumff7e83d1999-08-27 20:39:37 +0000893
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000894#if BUFSIZ < 8192
895#define SMALLCHUNK 8192
896#else
897#define SMALLCHUNK BUFSIZ
898#endif
899
Guido van Rossum3c259041999-01-14 19:00:14 +0000900#if SIZEOF_INT < 4
901#define BIGCHUNK (512 * 32)
902#else
903#define BIGCHUNK (512 * 1024)
904#endif
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000905
906static size_t
Fred Drakefd99de62000-07-09 05:02:18 +0000907new_buffersize(PyFileObject *f, size_t currentsize)
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000908{
909#ifdef HAVE_FSTAT
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000910 off_t pos, end;
911 struct stat st;
912 if (fstat(fileno(f->f_fp), &st) == 0) {
913 end = st.st_size;
914 /* The following is not a bug: we really need to call lseek()
915 *and* ftell(). The reason is that some stdio libraries
916 mistakenly flush their buffer when ftell() is called and
917 the lseek() call it makes fails, thereby throwing away
918 data that cannot be recovered in any way. To avoid this,
919 we first test lseek(), and only call ftell() if lseek()
920 works. We can't use the lseek() value either, because we
921 need to take the amount of buffered data into account.
922 (Yet another reason why stdio stinks. :-) */
923 pos = lseek(fileno(f->f_fp), 0L, SEEK_CUR);
924 if (pos >= 0) {
925 pos = ftell(f->f_fp);
926 }
927 if (pos < 0)
928 clearerr(f->f_fp);
929 if (end > pos && pos >= 0)
930 return currentsize + end - pos + 1;
931 /* Add 1 so if the file were to grow we'd notice. */
932 }
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000933#endif
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000934 if (currentsize > SMALLCHUNK) {
935 /* Keep doubling until we reach BIGCHUNK;
936 then keep adding BIGCHUNK. */
937 if (currentsize <= BIGCHUNK)
938 return currentsize + currentsize;
939 else
940 return currentsize + BIGCHUNK;
941 }
942 return currentsize + SMALLCHUNK;
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000943}
944
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +0000945#if defined(EWOULDBLOCK) && defined(EAGAIN) && EWOULDBLOCK != EAGAIN
946#define BLOCKED_ERRNO(x) ((x) == EWOULDBLOCK || (x) == EAGAIN)
947#else
948#ifdef EWOULDBLOCK
949#define BLOCKED_ERRNO(x) ((x) == EWOULDBLOCK)
950#else
951#ifdef EAGAIN
952#define BLOCKED_ERRNO(x) ((x) == EAGAIN)
953#else
954#define BLOCKED_ERRNO(x) 0
955#endif
956#endif
957#endif
958
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000959static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000960file_read(PyFileObject *f, PyObject *args)
Guido van Rossumce5ba841991-03-06 13:06:18 +0000961{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000962 long bytesrequested = -1;
963 size_t bytesread, buffersize, chunksize;
964 PyObject *v;
Tim Peters86821b22001-01-07 21:19:34 +0000965
Antoine Pitrouc7c96a92010-05-09 15:15:40 +0000966 if (f->f_fp == NULL)
967 return err_closed();
968 if (!f->readable)
969 return err_mode("reading");
970 /* refuse to mix with f.next() */
971 if (f->f_buf != NULL &&
972 (f->f_bufend - f->f_bufptr) > 0 &&
973 f->f_buf[0] != '\0')
974 return err_iterbuffered();
975 if (!PyArg_ParseTuple(args, "|l:read", &bytesrequested))
976 return NULL;
977 if (bytesrequested < 0)
978 buffersize = new_buffersize(f, (size_t)0);
979 else
980 buffersize = bytesrequested;
981 if (buffersize > PY_SSIZE_T_MAX) {
982 PyErr_SetString(PyExc_OverflowError,
983 "requested number of bytes is more than a Python string can hold");
984 return NULL;
985 }
986 v = PyString_FromStringAndSize((char *)NULL, buffersize);
987 if (v == NULL)
988 return NULL;
989 bytesread = 0;
990 for (;;) {
991 FILE_BEGIN_ALLOW_THREADS(f)
992 errno = 0;
993 chunksize = Py_UniversalNewlineFread(BUF(v) + bytesread,
994 buffersize - bytesread, f->f_fp, (PyObject *)f);
995 FILE_END_ALLOW_THREADS(f)
996 if (chunksize == 0) {
997 if (!ferror(f->f_fp))
998 break;
999 clearerr(f->f_fp);
1000 /* When in non-blocking mode, data shouldn't
1001 * be discarded if a blocking signal was
1002 * received. That will also happen if
1003 * chunksize != 0, but bytesread < buffersize. */
1004 if (bytesread > 0 && BLOCKED_ERRNO(errno))
1005 break;
1006 PyErr_SetFromErrno(PyExc_IOError);
1007 Py_DECREF(v);
1008 return NULL;
1009 }
1010 bytesread += chunksize;
1011 if (bytesread < buffersize) {
1012 clearerr(f->f_fp);
1013 break;
1014 }
1015 if (bytesrequested < 0) {
1016 buffersize = new_buffersize(f, buffersize);
1017 if (_PyString_Resize(&v, buffersize) < 0)
1018 return NULL;
1019 } else {
1020 /* Got what was requested. */
1021 break;
1022 }
1023 }
1024 if (bytesread != buffersize)
1025 _PyString_Resize(&v, bytesread);
1026 return v;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001027}
1028
Guido van Rossumfdf95dd1997-05-05 22:15:02 +00001029static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001030file_readinto(PyFileObject *f, PyObject *args)
Guido van Rossumfdf95dd1997-05-05 22:15:02 +00001031{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001032 char *ptr;
1033 Py_ssize_t ntodo;
1034 Py_ssize_t ndone, nnow;
1035 Py_buffer pbuf;
Tim Peters86821b22001-01-07 21:19:34 +00001036
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001037 if (f->f_fp == NULL)
1038 return err_closed();
1039 if (!f->readable)
1040 return err_mode("reading");
1041 /* refuse to mix with f.next() */
1042 if (f->f_buf != NULL &&
1043 (f->f_bufend - f->f_bufptr) > 0 &&
1044 f->f_buf[0] != '\0')
1045 return err_iterbuffered();
1046 if (!PyArg_ParseTuple(args, "w*", &pbuf))
1047 return NULL;
1048 ptr = pbuf.buf;
1049 ntodo = pbuf.len;
1050 ndone = 0;
1051 while (ntodo > 0) {
1052 FILE_BEGIN_ALLOW_THREADS(f)
1053 errno = 0;
1054 nnow = Py_UniversalNewlineFread(ptr+ndone, ntodo, f->f_fp,
1055 (PyObject *)f);
1056 FILE_END_ALLOW_THREADS(f)
1057 if (nnow == 0) {
1058 if (!ferror(f->f_fp))
1059 break;
1060 PyErr_SetFromErrno(PyExc_IOError);
1061 clearerr(f->f_fp);
1062 PyBuffer_Release(&pbuf);
1063 return NULL;
1064 }
1065 ndone += nnow;
1066 ntodo -= nnow;
1067 }
1068 PyBuffer_Release(&pbuf);
1069 return PyInt_FromSsize_t(ndone);
Guido van Rossumfdf95dd1997-05-05 22:15:02 +00001070}
1071
Tim Peters86821b22001-01-07 21:19:34 +00001072/**************************************************************************
Tim Petersf29b64d2001-01-15 06:33:19 +00001073Routine to get next line using platform fgets().
Tim Peters86821b22001-01-07 21:19:34 +00001074
1075Under MSVC 6:
1076
Tim Peters1c733232001-01-08 04:02:07 +00001077+ MS threadsafe getc is very slow (multiple layers of function calls before+
1078 after each character, to lock+unlock the stream).
1079+ The stream-locking functions are MS-internal -- can't access them from user
1080 code.
1081+ There's nothing Tim could find in the MS C or platform SDK libraries that
1082 can worm around this.
Tim Peters86821b22001-01-07 21:19:34 +00001083+ MS fgets locks/unlocks only once per line; it's the only hook we have.
1084
1085So we use fgets for speed(!), despite that it's painful.
1086
1087MS realloc is also slow.
1088
Tim Petersf29b64d2001-01-15 06:33:19 +00001089Reports from other platforms on this method vs getc_unlocked (which MS doesn't
1090have):
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001091 Linux a wash
1092 Solaris a wash
1093 Tru64 Unix getline_via_fgets significantly faster
Tim Peters86821b22001-01-07 21:19:34 +00001094
Tim Petersf29b64d2001-01-15 06:33:19 +00001095CAUTION: The C std isn't clear about this: in those cases where fgets
1096writes something into the buffer, can it write into any position beyond the
1097required trailing null byte? MSVC 6 fgets does not, and no platform is (yet)
1098known on which it does; and it would be a strange way to code fgets. Still,
1099getline_via_fgets may not work correctly if it does. The std test
1100test_bufio.py should fail if platform fgets() routinely writes beyond the
1101trailing null byte. #define DONT_USE_FGETS_IN_GETLINE to disable this code.
Tim Peters86821b22001-01-07 21:19:34 +00001102**************************************************************************/
1103
Tim Petersf29b64d2001-01-15 06:33:19 +00001104/* Use this routine if told to, or by default on non-get_unlocked()
1105 * platforms unless told not to. Yikes! Let's spell that out:
1106 * On a platform with getc_unlocked():
1107 * By default, use getc_unlocked().
1108 * If you want to use fgets() instead, #define USE_FGETS_IN_GETLINE.
1109 * On a platform without getc_unlocked():
1110 * By default, use fgets().
1111 * If you don't want to use fgets(), #define DONT_USE_FGETS_IN_GETLINE.
1112 */
1113#if !defined(USE_FGETS_IN_GETLINE) && !defined(HAVE_GETC_UNLOCKED)
1114#define USE_FGETS_IN_GETLINE
Tim Peters86821b22001-01-07 21:19:34 +00001115#endif
1116
Tim Petersf29b64d2001-01-15 06:33:19 +00001117#if defined(DONT_USE_FGETS_IN_GETLINE) && defined(USE_FGETS_IN_GETLINE)
1118#undef USE_FGETS_IN_GETLINE
1119#endif
1120
1121#ifdef USE_FGETS_IN_GETLINE
Tim Peters86821b22001-01-07 21:19:34 +00001122static PyObject*
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00001123getline_via_fgets(PyFileObject *f, FILE *fp)
Tim Peters86821b22001-01-07 21:19:34 +00001124{
Tim Peters15b83852001-01-08 00:53:12 +00001125/* INITBUFSIZE is the maximum line length that lets us get away with the fast
Tim Peters142297a2001-01-15 10:36:56 +00001126 * no-realloc, one-fgets()-call path. Boosting it isn't free, because we have
1127 * to fill this much of the buffer with a known value in order to figure out
1128 * how much of the buffer fgets() overwrites. So if INITBUFSIZE is larger
1129 * than "most" lines, we waste time filling unused buffer slots. 100 is
1130 * surely adequate for most peoples' email archives, chewing over source code,
1131 * etc -- "regular old text files".
1132 * MAXBUFSIZE is the maximum line length that lets us get away with the less
1133 * fast (but still zippy) no-realloc, two-fgets()-call path. See above for
1134 * cautions about boosting that. 300 was chosen because the worst real-life
1135 * text-crunching job reported on Python-Dev was a mail-log crawler where over
1136 * half the lines were 254 chars.
Tim Peters15b83852001-01-08 00:53:12 +00001137 */
Tim Peters142297a2001-01-15 10:36:56 +00001138#define INITBUFSIZE 100
1139#define MAXBUFSIZE 300
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001140 char* p; /* temp */
1141 char buf[MAXBUFSIZE];
1142 PyObject* v; /* the string object result */
1143 char* pvfree; /* address of next free slot */
1144 char* pvend; /* address one beyond last free slot */
1145 size_t nfree; /* # of free buffer slots; pvend-pvfree */
1146 size_t total_v_size; /* total # of slots in buffer */
1147 size_t increment; /* amount to increment the buffer */
1148 size_t prev_v_size;
Tim Peters86821b22001-01-07 21:19:34 +00001149
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001150 /* Optimize for normal case: avoid _PyString_Resize if at all
1151 * possible via first reading into stack buffer "buf".
1152 */
1153 total_v_size = INITBUFSIZE; /* start small and pray */
1154 pvfree = buf;
1155 for (;;) {
1156 FILE_BEGIN_ALLOW_THREADS(f)
1157 pvend = buf + total_v_size;
1158 nfree = pvend - pvfree;
1159 memset(pvfree, '\n', nfree);
1160 assert(nfree < INT_MAX); /* Should be atmost MAXBUFSIZE */
1161 p = fgets(pvfree, (int)nfree, fp);
1162 FILE_END_ALLOW_THREADS(f)
Tim Peters15b83852001-01-08 00:53:12 +00001163
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001164 if (p == NULL) {
1165 clearerr(fp);
1166 if (PyErr_CheckSignals())
1167 return NULL;
1168 v = PyString_FromStringAndSize(buf, pvfree - buf);
1169 return v;
1170 }
1171 /* fgets read *something* */
1172 p = memchr(pvfree, '\n', nfree);
1173 if (p != NULL) {
1174 /* Did the \n come from fgets or from us?
1175 * Since fgets stops at the first \n, and then writes
1176 * \0, if it's from fgets a \0 must be next. But if
1177 * that's so, it could not have come from us, since
1178 * the \n's we filled the buffer with have only more
1179 * \n's to the right.
1180 */
1181 if (p+1 < pvend && *(p+1) == '\0') {
1182 /* It's from fgets: we win! In particular,
1183 * we haven't done any mallocs yet, and can
1184 * build the final result on the first try.
1185 */
1186 ++p; /* include \n from fgets */
1187 }
1188 else {
1189 /* Must be from us: fgets didn't fill the
1190 * buffer and didn't find a newline, so it
1191 * must be the last and newline-free line of
1192 * the file.
1193 */
1194 assert(p > pvfree && *(p-1) == '\0');
1195 --p; /* don't include \0 from fgets */
1196 }
1197 v = PyString_FromStringAndSize(buf, p - buf);
1198 return v;
1199 }
1200 /* yuck: fgets overwrote all the newlines, i.e. the entire
1201 * buffer. So this line isn't over yet, or maybe it is but
1202 * we're exactly at EOF. If we haven't already, try using the
1203 * rest of the stack buffer.
1204 */
1205 assert(*(pvend-1) == '\0');
1206 if (pvfree == buf) {
1207 pvfree = pvend - 1; /* overwrite trailing null */
1208 total_v_size = MAXBUFSIZE;
1209 }
1210 else
1211 break;
1212 }
Tim Peters142297a2001-01-15 10:36:56 +00001213
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001214 /* The stack buffer isn't big enough; malloc a string object and read
1215 * into its buffer.
1216 */
1217 total_v_size = MAXBUFSIZE << 1;
1218 v = PyString_FromStringAndSize((char*)NULL, (int)total_v_size);
1219 if (v == NULL)
1220 return v;
1221 /* copy over everything except the last null byte */
1222 memcpy(BUF(v), buf, MAXBUFSIZE-1);
1223 pvfree = BUF(v) + MAXBUFSIZE - 1;
Tim Peters86821b22001-01-07 21:19:34 +00001224
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001225 /* Keep reading stuff into v; if it ever ends successfully, break
1226 * after setting p one beyond the end of the line. The code here is
1227 * very much like the code above, except reads into v's buffer; see
1228 * the code above for detailed comments about the logic.
1229 */
1230 for (;;) {
1231 FILE_BEGIN_ALLOW_THREADS(f)
1232 pvend = BUF(v) + total_v_size;
1233 nfree = pvend - pvfree;
1234 memset(pvfree, '\n', nfree);
1235 assert(nfree < INT_MAX);
1236 p = fgets(pvfree, (int)nfree, fp);
1237 FILE_END_ALLOW_THREADS(f)
Tim Peters86821b22001-01-07 21:19:34 +00001238
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001239 if (p == NULL) {
1240 clearerr(fp);
1241 if (PyErr_CheckSignals()) {
1242 Py_DECREF(v);
1243 return NULL;
1244 }
1245 p = pvfree;
1246 break;
1247 }
1248 p = memchr(pvfree, '\n', nfree);
1249 if (p != NULL) {
1250 if (p+1 < pvend && *(p+1) == '\0') {
1251 /* \n came from fgets */
1252 ++p;
1253 break;
1254 }
1255 /* \n came from us; last line of file, no newline */
1256 assert(p > pvfree && *(p-1) == '\0');
1257 --p;
1258 break;
1259 }
1260 /* expand buffer and try again */
1261 assert(*(pvend-1) == '\0');
1262 increment = total_v_size >> 2; /* mild exponential growth */
1263 prev_v_size = total_v_size;
1264 total_v_size += increment;
1265 /* check for overflow */
1266 if (total_v_size <= prev_v_size ||
1267 total_v_size > PY_SSIZE_T_MAX) {
1268 PyErr_SetString(PyExc_OverflowError,
1269 "line is longer than a Python string can hold");
1270 Py_DECREF(v);
1271 return NULL;
1272 }
1273 if (_PyString_Resize(&v, (int)total_v_size) < 0)
1274 return NULL;
1275 /* overwrite the trailing null byte */
1276 pvfree = BUF(v) + (prev_v_size - 1);
1277 }
1278 if (BUF(v) + total_v_size != p)
1279 _PyString_Resize(&v, p - BUF(v));
1280 return v;
Tim Peters86821b22001-01-07 21:19:34 +00001281#undef INITBUFSIZE
Tim Peters142297a2001-01-15 10:36:56 +00001282#undef MAXBUFSIZE
Tim Peters86821b22001-01-07 21:19:34 +00001283}
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001284#endif /* ifdef USE_FGETS_IN_GETLINE */
Guido van Rossumfdf95dd1997-05-05 22:15:02 +00001285
Guido van Rossum0bd24411991-04-04 15:21:57 +00001286/* Internal routine to get a line.
1287 Size argument interpretation:
1288 > 0: max length;
Guido van Rossum86282062001-01-08 01:26:47 +00001289 <= 0: read arbitrary line
Guido van Rossumce5ba841991-03-06 13:06:18 +00001290*/
1291
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001292static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001293get_line(PyFileObject *f, int n)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001294{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001295 FILE *fp = f->f_fp;
1296 int c;
1297 char *buf, *end;
1298 size_t total_v_size; /* total # of slots in buffer */
1299 size_t used_v_size; /* # used slots in buffer */
1300 size_t increment; /* amount to increment the buffer */
1301 PyObject *v;
1302 int newlinetypes = f->f_newlinetypes;
1303 int skipnextlf = f->f_skipnextlf;
1304 int univ_newline = f->f_univ_newline;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001305
Jack Jansen7b8c7542002-04-14 20:12:41 +00001306#if defined(USE_FGETS_IN_GETLINE)
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001307 if (n <= 0 && !univ_newline )
1308 return getline_via_fgets(f, fp);
Tim Peters86821b22001-01-07 21:19:34 +00001309#endif
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001310 total_v_size = n > 0 ? n : 100;
1311 v = PyString_FromStringAndSize((char *)NULL, total_v_size);
1312 if (v == NULL)
1313 return NULL;
1314 buf = BUF(v);
1315 end = buf + total_v_size;
Guido van Rossum1984f1e1992-08-04 12:41:02 +00001316
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001317 for (;;) {
1318 FILE_BEGIN_ALLOW_THREADS(f)
1319 FLOCKFILE(fp);
1320 if (univ_newline) {
1321 c = 'x'; /* Shut up gcc warning */
1322 while ( buf != end && (c = GETC(fp)) != EOF ) {
1323 if (skipnextlf ) {
1324 skipnextlf = 0;
1325 if (c == '\n') {
1326 /* Seeing a \n here with
1327 * skipnextlf true means we
1328 * saw a \r before.
1329 */
1330 newlinetypes |= NEWLINE_CRLF;
1331 c = GETC(fp);
1332 if (c == EOF) break;
1333 } else {
1334 newlinetypes |= NEWLINE_CR;
1335 }
1336 }
1337 if (c == '\r') {
1338 skipnextlf = 1;
1339 c = '\n';
1340 } else if ( c == '\n')
1341 newlinetypes |= NEWLINE_LF;
1342 *buf++ = c;
1343 if (c == '\n') break;
1344 }
1345 if ( c == EOF && skipnextlf )
1346 newlinetypes |= NEWLINE_CR;
1347 } else /* If not universal newlines use the normal loop */
1348 while ((c = GETC(fp)) != EOF &&
1349 (*buf++ = c) != '\n' &&
1350 buf != end)
1351 ;
1352 FUNLOCKFILE(fp);
1353 FILE_END_ALLOW_THREADS(f)
1354 f->f_newlinetypes = newlinetypes;
1355 f->f_skipnextlf = skipnextlf;
1356 if (c == '\n')
1357 break;
1358 if (c == EOF) {
1359 if (ferror(fp)) {
1360 PyErr_SetFromErrno(PyExc_IOError);
1361 clearerr(fp);
1362 Py_DECREF(v);
1363 return NULL;
1364 }
1365 clearerr(fp);
1366 if (PyErr_CheckSignals()) {
1367 Py_DECREF(v);
1368 return NULL;
1369 }
1370 break;
1371 }
1372 /* Must be because buf == end */
1373 if (n > 0)
1374 break;
1375 used_v_size = total_v_size;
1376 increment = total_v_size >> 2; /* mild exponential growth */
1377 total_v_size += increment;
1378 if (total_v_size > PY_SSIZE_T_MAX) {
1379 PyErr_SetString(PyExc_OverflowError,
1380 "line is longer than a Python string can hold");
1381 Py_DECREF(v);
1382 return NULL;
1383 }
1384 if (_PyString_Resize(&v, total_v_size) < 0)
1385 return NULL;
1386 buf = BUF(v) + used_v_size;
1387 end = BUF(v) + total_v_size;
1388 }
Guido van Rossum1984f1e1992-08-04 12:41:02 +00001389
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001390 used_v_size = buf - BUF(v);
1391 if (used_v_size != total_v_size)
1392 _PyString_Resize(&v, used_v_size);
1393 return v;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001394}
1395
Guido van Rossum0bd24411991-04-04 15:21:57 +00001396/* External C interface */
1397
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001398PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001399PyFile_GetLine(PyObject *f, int n)
Guido van Rossum0bd24411991-04-04 15:21:57 +00001400{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001401 PyObject *result;
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001402
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001403 if (f == NULL) {
1404 PyErr_BadInternalCall();
1405 return NULL;
1406 }
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001407
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001408 if (PyFile_Check(f)) {
1409 PyFileObject *fo = (PyFileObject *)f;
1410 if (fo->f_fp == NULL)
1411 return err_closed();
1412 if (!fo->readable)
1413 return err_mode("reading");
1414 /* refuse to mix with f.next() */
1415 if (fo->f_buf != NULL &&
1416 (fo->f_bufend - fo->f_bufptr) > 0 &&
1417 fo->f_buf[0] != '\0')
1418 return err_iterbuffered();
1419 result = get_line(fo, n);
1420 }
1421 else {
1422 PyObject *reader;
1423 PyObject *args;
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001424
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001425 reader = PyObject_GetAttrString(f, "readline");
1426 if (reader == NULL)
1427 return NULL;
1428 if (n <= 0)
1429 args = PyTuple_New(0);
1430 else
1431 args = Py_BuildValue("(i)", n);
1432 if (args == NULL) {
1433 Py_DECREF(reader);
1434 return NULL;
1435 }
1436 result = PyEval_CallObject(reader, args);
1437 Py_DECREF(reader);
1438 Py_DECREF(args);
1439 if (result != NULL && !PyString_Check(result) &&
1440 !PyUnicode_Check(result)) {
1441 Py_DECREF(result);
1442 result = NULL;
1443 PyErr_SetString(PyExc_TypeError,
1444 "object.readline() returned non-string");
1445 }
1446 }
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001447
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001448 if (n < 0 && result != NULL && PyString_Check(result)) {
1449 char *s = PyString_AS_STRING(result);
1450 Py_ssize_t len = PyString_GET_SIZE(result);
1451 if (len == 0) {
1452 Py_DECREF(result);
1453 result = NULL;
1454 PyErr_SetString(PyExc_EOFError,
1455 "EOF when reading a line");
1456 }
1457 else if (s[len-1] == '\n') {
1458 if (result->ob_refcnt == 1)
1459 _PyString_Resize(&result, len-1);
1460 else {
1461 PyObject *v;
1462 v = PyString_FromStringAndSize(s, len-1);
1463 Py_DECREF(result);
1464 result = v;
1465 }
1466 }
1467 }
Martin v. Löwisaf6a27a2003-01-03 19:16:14 +00001468#ifdef Py_USING_UNICODE
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001469 if (n < 0 && result != NULL && PyUnicode_Check(result)) {
1470 Py_UNICODE *s = PyUnicode_AS_UNICODE(result);
1471 Py_ssize_t len = PyUnicode_GET_SIZE(result);
1472 if (len == 0) {
1473 Py_DECREF(result);
1474 result = NULL;
1475 PyErr_SetString(PyExc_EOFError,
1476 "EOF when reading a line");
1477 }
1478 else if (s[len-1] == '\n') {
1479 if (result->ob_refcnt == 1)
1480 PyUnicode_Resize(&result, len-1);
1481 else {
1482 PyObject *v;
1483 v = PyUnicode_FromUnicode(s, len-1);
1484 Py_DECREF(result);
1485 result = v;
1486 }
1487 }
1488 }
Martin v. Löwisaf6a27a2003-01-03 19:16:14 +00001489#endif
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001490 return result;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001491}
1492
1493/* Python method */
1494
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001495static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001496file_readline(PyFileObject *f, PyObject *args)
Guido van Rossum0bd24411991-04-04 15:21:57 +00001497{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001498 int n = -1;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001499
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001500 if (f->f_fp == NULL)
1501 return err_closed();
1502 if (!f->readable)
1503 return err_mode("reading");
1504 /* refuse to mix with f.next() */
1505 if (f->f_buf != NULL &&
1506 (f->f_bufend - f->f_bufptr) > 0 &&
1507 f->f_buf[0] != '\0')
1508 return err_iterbuffered();
1509 if (!PyArg_ParseTuple(args, "|i:readline", &n))
1510 return NULL;
1511 if (n == 0)
1512 return PyString_FromString("");
1513 if (n < 0)
1514 n = 0;
1515 return get_line(f, n);
Guido van Rossum0bd24411991-04-04 15:21:57 +00001516}
1517
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001518static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001519file_readlines(PyFileObject *f, PyObject *args)
Guido van Rossumce5ba841991-03-06 13:06:18 +00001520{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001521 long sizehint = 0;
1522 PyObject *list = NULL;
1523 PyObject *line;
1524 char small_buffer[SMALLCHUNK];
1525 char *buffer = small_buffer;
1526 size_t buffersize = SMALLCHUNK;
1527 PyObject *big_buffer = NULL;
1528 size_t nfilled = 0;
1529 size_t nread;
1530 size_t totalread = 0;
1531 char *p, *q, *end;
1532 int err;
1533 int shortread = 0;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001534
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001535 if (f->f_fp == NULL)
1536 return err_closed();
1537 if (!f->readable)
1538 return err_mode("reading");
1539 /* refuse to mix with f.next() */
1540 if (f->f_buf != NULL &&
1541 (f->f_bufend - f->f_bufptr) > 0 &&
1542 f->f_buf[0] != '\0')
1543 return err_iterbuffered();
1544 if (!PyArg_ParseTuple(args, "|l:readlines", &sizehint))
1545 return NULL;
1546 if ((list = PyList_New(0)) == NULL)
1547 return NULL;
1548 for (;;) {
1549 if (shortread)
1550 nread = 0;
1551 else {
1552 FILE_BEGIN_ALLOW_THREADS(f)
1553 errno = 0;
1554 nread = Py_UniversalNewlineFread(buffer+nfilled,
1555 buffersize-nfilled, f->f_fp, (PyObject *)f);
1556 FILE_END_ALLOW_THREADS(f)
1557 shortread = (nread < buffersize-nfilled);
1558 }
1559 if (nread == 0) {
1560 sizehint = 0;
1561 if (!ferror(f->f_fp))
1562 break;
1563 PyErr_SetFromErrno(PyExc_IOError);
1564 clearerr(f->f_fp);
1565 goto error;
1566 }
1567 totalread += nread;
1568 p = (char *)memchr(buffer+nfilled, '\n', nread);
1569 if (p == NULL) {
1570 /* Need a larger buffer to fit this line */
1571 nfilled += nread;
1572 buffersize *= 2;
1573 if (buffersize > PY_SSIZE_T_MAX) {
1574 PyErr_SetString(PyExc_OverflowError,
1575 "line is longer than a Python string can hold");
1576 goto error;
1577 }
1578 if (big_buffer == NULL) {
1579 /* Create the big buffer */
1580 big_buffer = PyString_FromStringAndSize(
1581 NULL, buffersize);
1582 if (big_buffer == NULL)
1583 goto error;
1584 buffer = PyString_AS_STRING(big_buffer);
1585 memcpy(buffer, small_buffer, nfilled);
1586 }
1587 else {
1588 /* Grow the big buffer */
1589 if ( _PyString_Resize(&big_buffer, buffersize) < 0 )
1590 goto error;
1591 buffer = PyString_AS_STRING(big_buffer);
1592 }
1593 continue;
1594 }
1595 end = buffer+nfilled+nread;
1596 q = buffer;
1597 do {
1598 /* Process complete lines */
1599 p++;
1600 line = PyString_FromStringAndSize(q, p-q);
1601 if (line == NULL)
1602 goto error;
1603 err = PyList_Append(list, line);
1604 Py_DECREF(line);
1605 if (err != 0)
1606 goto error;
1607 q = p;
1608 p = (char *)memchr(q, '\n', end-q);
1609 } while (p != NULL);
1610 /* Move the remaining incomplete line to the start */
1611 nfilled = end-q;
1612 memmove(buffer, q, nfilled);
1613 if (sizehint > 0)
1614 if (totalread >= (size_t)sizehint)
1615 break;
1616 }
1617 if (nfilled != 0) {
1618 /* Partial last line */
1619 line = PyString_FromStringAndSize(buffer, nfilled);
1620 if (line == NULL)
1621 goto error;
1622 if (sizehint > 0) {
1623 /* Need to complete the last line */
1624 PyObject *rest = get_line(f, 0);
1625 if (rest == NULL) {
1626 Py_DECREF(line);
1627 goto error;
1628 }
1629 PyString_Concat(&line, rest);
1630 Py_DECREF(rest);
1631 if (line == NULL)
1632 goto error;
1633 }
1634 err = PyList_Append(list, line);
1635 Py_DECREF(line);
1636 if (err != 0)
1637 goto error;
1638 }
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00001639
1640cleanup:
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001641 Py_XDECREF(big_buffer);
1642 return list;
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00001643
1644error:
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001645 Py_CLEAR(list);
1646 goto cleanup;
Guido van Rossumce5ba841991-03-06 13:06:18 +00001647}
1648
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001649static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001650file_write(PyFileObject *f, PyObject *args)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001651{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001652 Py_buffer pbuf;
1653 char *s;
1654 Py_ssize_t n, n2;
1655 if (f->f_fp == NULL)
1656 return err_closed();
1657 if (!f->writable)
1658 return err_mode("writing");
1659 if (f->f_binary) {
1660 if (!PyArg_ParseTuple(args, "s*", &pbuf))
1661 return NULL;
1662 s = pbuf.buf;
1663 n = pbuf.len;
1664 } else
1665 if (!PyArg_ParseTuple(args, "t#", &s, &n))
1666 return NULL;
1667 f->f_softspace = 0;
1668 FILE_BEGIN_ALLOW_THREADS(f)
1669 errno = 0;
1670 n2 = fwrite(s, 1, n, f->f_fp);
1671 FILE_END_ALLOW_THREADS(f)
1672 if (f->f_binary)
1673 PyBuffer_Release(&pbuf);
1674 if (n2 != n) {
1675 PyErr_SetFromErrno(PyExc_IOError);
1676 clearerr(f->f_fp);
1677 return NULL;
1678 }
1679 Py_INCREF(Py_None);
1680 return Py_None;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001681}
1682
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001683static PyObject *
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001684file_writelines(PyFileObject *f, PyObject *seq)
Guido van Rossum5a2a6831993-10-25 09:59:04 +00001685{
Guido van Rossumee70ad12000-03-13 16:27:06 +00001686#define CHUNKSIZE 1000
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001687 PyObject *list, *line;
1688 PyObject *it; /* iter(seq) */
1689 PyObject *result;
1690 int index, islist;
1691 Py_ssize_t i, j, nwritten, len;
Guido van Rossumee70ad12000-03-13 16:27:06 +00001692
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001693 assert(seq != NULL);
1694 if (f->f_fp == NULL)
1695 return err_closed();
1696 if (!f->writable)
1697 return err_mode("writing");
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001698
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001699 result = NULL;
1700 list = NULL;
1701 islist = PyList_Check(seq);
1702 if (islist)
1703 it = NULL;
1704 else {
1705 it = PyObject_GetIter(seq);
1706 if (it == NULL) {
1707 PyErr_SetString(PyExc_TypeError,
1708 "writelines() requires an iterable argument");
1709 return NULL;
1710 }
1711 /* From here on, fail by going to error, to reclaim "it". */
1712 list = PyList_New(CHUNKSIZE);
1713 if (list == NULL)
1714 goto error;
1715 }
Guido van Rossumee70ad12000-03-13 16:27:06 +00001716
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001717 /* Strategy: slurp CHUNKSIZE lines into a private list,
1718 checking that they are all strings, then write that list
1719 without holding the interpreter lock, then come back for more. */
1720 for (index = 0; ; index += CHUNKSIZE) {
1721 if (islist) {
1722 Py_XDECREF(list);
1723 list = PyList_GetSlice(seq, index, index+CHUNKSIZE);
1724 if (list == NULL)
1725 goto error;
1726 j = PyList_GET_SIZE(list);
1727 }
1728 else {
1729 for (j = 0; j < CHUNKSIZE; j++) {
1730 line = PyIter_Next(it);
1731 if (line == NULL) {
1732 if (PyErr_Occurred())
1733 goto error;
1734 break;
1735 }
1736 PyList_SetItem(list, j, line);
1737 }
1738 }
1739 if (j == 0)
1740 break;
Guido van Rossumee70ad12000-03-13 16:27:06 +00001741
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001742 /* Check that all entries are indeed strings. If not,
1743 apply the same rules as for file.write() and
1744 convert the results to strings. This is slow, but
1745 seems to be the only way since all conversion APIs
1746 could potentially execute Python code. */
1747 for (i = 0; i < j; i++) {
1748 PyObject *v = PyList_GET_ITEM(list, i);
1749 if (!PyString_Check(v)) {
1750 const char *buffer;
1751 if (((f->f_binary &&
1752 PyObject_AsReadBuffer(v,
1753 (const void**)&buffer,
1754 &len)) ||
1755 PyObject_AsCharBuffer(v,
1756 &buffer,
1757 &len))) {
1758 PyErr_SetString(PyExc_TypeError,
1759 "writelines() argument must be a sequence of strings");
1760 goto error;
1761 }
1762 line = PyString_FromStringAndSize(buffer,
1763 len);
1764 if (line == NULL)
1765 goto error;
1766 Py_DECREF(v);
1767 PyList_SET_ITEM(list, i, line);
1768 }
1769 }
Marc-André Lemburg6ef68b52000-08-25 22:39:50 +00001770
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001771 /* Since we are releasing the global lock, the
1772 following code may *not* execute Python code. */
1773 f->f_softspace = 0;
1774 FILE_BEGIN_ALLOW_THREADS(f)
1775 errno = 0;
1776 for (i = 0; i < j; i++) {
1777 line = PyList_GET_ITEM(list, i);
1778 len = PyString_GET_SIZE(line);
1779 nwritten = fwrite(PyString_AS_STRING(line),
1780 1, len, f->f_fp);
1781 if (nwritten != len) {
1782 FILE_ABORT_ALLOW_THREADS(f)
1783 PyErr_SetFromErrno(PyExc_IOError);
1784 clearerr(f->f_fp);
1785 goto error;
1786 }
1787 }
1788 FILE_END_ALLOW_THREADS(f)
Guido van Rossumee70ad12000-03-13 16:27:06 +00001789
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001790 if (j < CHUNKSIZE)
1791 break;
1792 }
Guido van Rossumee70ad12000-03-13 16:27:06 +00001793
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001794 Py_INCREF(Py_None);
1795 result = Py_None;
Guido van Rossumee70ad12000-03-13 16:27:06 +00001796 error:
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001797 Py_XDECREF(list);
1798 Py_XDECREF(it);
1799 return result;
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001800#undef CHUNKSIZE
Guido van Rossum5a2a6831993-10-25 09:59:04 +00001801}
1802
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001803static PyObject *
Guido van Rossum1a5e21e2006-02-28 21:57:43 +00001804file_self(PyFileObject *f)
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001805{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001806 if (f->f_fp == NULL)
1807 return err_closed();
1808 Py_INCREF(f);
1809 return (PyObject *)f;
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001810}
1811
Georg Brandl98b40ad2006-06-08 14:50:21 +00001812static PyObject *
Georg Brandla9916b52008-05-17 22:11:54 +00001813file_xreadlines(PyFileObject *f)
1814{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001815 if (PyErr_WarnPy3k("f.xreadlines() not supported in 3.x, "
1816 "try 'for line in f' instead", 1) < 0)
1817 return NULL;
1818 return file_self(f);
Georg Brandla9916b52008-05-17 22:11:54 +00001819}
1820
1821static PyObject *
Georg Brandlad61bc82008-02-23 15:11:18 +00001822file_exit(PyObject *f, PyObject *args)
Georg Brandl98b40ad2006-06-08 14:50:21 +00001823{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001824 PyObject *ret = PyObject_CallMethod(f, "close", NULL);
1825 if (!ret)
1826 /* If error occurred, pass through */
1827 return NULL;
1828 Py_DECREF(ret);
1829 /* We cannot return the result of close since a true
1830 * value will be interpreted as "yes, swallow the
1831 * exception if one was raised inside the with block". */
1832 Py_RETURN_NONE;
Georg Brandl98b40ad2006-06-08 14:50:21 +00001833}
1834
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001835PyDoc_STRVAR(readline_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001836"readline([size]) -> next line from the file, as a string.\n"
1837"\n"
1838"Retain newline. A non-negative size argument limits the maximum\n"
1839"number of bytes to return (an incomplete line may be returned then).\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001840"Return an empty string at EOF.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001841
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001842PyDoc_STRVAR(read_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001843"read([size]) -> read at most size bytes, returned as a string.\n"
1844"\n"
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +00001845"If the size argument is negative or omitted, read until EOF is reached.\n"
1846"Notice that when in non-blocking mode, less data than what was requested\n"
1847"may be returned, even if no size parameter was given.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001848
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001849PyDoc_STRVAR(write_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001850"write(str) -> None. Write string str to file.\n"
1851"\n"
1852"Note that due to buffering, flush() or close() may be needed before\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001853"the file on disk reflects the data written.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001854
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001855PyDoc_STRVAR(fileno_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001856"fileno() -> integer \"file descriptor\".\n"
1857"\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001858"This is needed for lower-level file interfaces, such os.read().");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001859
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001860PyDoc_STRVAR(seek_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001861"seek(offset[, whence]) -> None. Move to new file position.\n"
1862"\n"
1863"Argument offset is a byte count. Optional argument whence defaults to\n"
1864"0 (offset from start of file, offset should be >= 0); other values are 1\n"
1865"(move relative to current position, positive or negative), and 2 (move\n"
1866"relative to end of file, usually negative, although many platforms allow\n"
Martin v. Löwis849a9722003-10-18 09:38:01 +00001867"seeking beyond the end of a file). If the file is opened in text mode,\n"
1868"only offsets returned by tell() are legal. Use of other offsets causes\n"
1869"undefined behavior."
Tim Petersefc3a3a2001-09-20 07:55:22 +00001870"\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001871"Note that not all file objects are seekable.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001872
Guido van Rossumd7047b31995-01-02 19:07:15 +00001873#ifdef HAVE_FTRUNCATE
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001874PyDoc_STRVAR(truncate_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001875"truncate([size]) -> None. Truncate the file to at most size bytes.\n"
1876"\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001877"Size defaults to the current file position, as returned by tell().");
Guido van Rossumd7047b31995-01-02 19:07:15 +00001878#endif
Tim Petersefc3a3a2001-09-20 07:55:22 +00001879
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001880PyDoc_STRVAR(tell_doc,
1881"tell() -> current file position, an integer (may be a long integer).");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001882
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001883PyDoc_STRVAR(readinto_doc,
1884"readinto() -> Undocumented. Don't use this; it may go away.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001885
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001886PyDoc_STRVAR(readlines_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001887"readlines([size]) -> list of strings, each a line from the file.\n"
1888"\n"
1889"Call readline() repeatedly and return a list of the lines so read.\n"
1890"The optional size argument, if given, is an approximate bound on the\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001891"total number of bytes in the lines returned.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001892
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001893PyDoc_STRVAR(xreadlines_doc,
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001894"xreadlines() -> returns self.\n"
Tim Petersefc3a3a2001-09-20 07:55:22 +00001895"\n"
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001896"For backward compatibility. File objects now include the performance\n"
1897"optimizations previously implemented in the xreadlines module.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001898
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001899PyDoc_STRVAR(writelines_doc,
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001900"writelines(sequence_of_strings) -> None. Write the strings to the file.\n"
Tim Petersefc3a3a2001-09-20 07:55:22 +00001901"\n"
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001902"Note that newlines are not added. The sequence can be any iterable object\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001903"producing strings. This is equivalent to calling write() for each string.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001904
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001905PyDoc_STRVAR(flush_doc,
1906"flush() -> None. Flush the internal I/O buffer.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001907
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001908PyDoc_STRVAR(close_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001909"close() -> None or (perhaps) an integer. Close the file.\n"
1910"\n"
Guido van Rossum77f6a652002-04-03 22:41:51 +00001911"Sets data attribute .closed to True. A closed file cannot be used for\n"
Tim Petersefc3a3a2001-09-20 07:55:22 +00001912"further I/O operations. close() may be called more than once without\n"
1913"error. Some kinds of file objects (for example, opened by popen())\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001914"may return an exit status upon closing.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001915
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001916PyDoc_STRVAR(isatty_doc,
1917"isatty() -> true or false. True if the file is connected to a tty device.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001918
Guido van Rossum1a5e21e2006-02-28 21:57:43 +00001919PyDoc_STRVAR(enter_doc,
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001920 "__enter__() -> self.");
Guido van Rossum1a5e21e2006-02-28 21:57:43 +00001921
Georg Brandl98b40ad2006-06-08 14:50:21 +00001922PyDoc_STRVAR(exit_doc,
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001923 "__exit__(*excinfo) -> None. Closes the file.");
Georg Brandl98b40ad2006-06-08 14:50:21 +00001924
Tim Petersefc3a3a2001-09-20 07:55:22 +00001925static PyMethodDef file_methods[] = {
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001926 {"readline", (PyCFunction)file_readline, METH_VARARGS, readline_doc},
1927 {"read", (PyCFunction)file_read, METH_VARARGS, read_doc},
1928 {"write", (PyCFunction)file_write, METH_VARARGS, write_doc},
1929 {"fileno", (PyCFunction)file_fileno, METH_NOARGS, fileno_doc},
1930 {"seek", (PyCFunction)file_seek, METH_VARARGS, seek_doc},
Tim Petersefc3a3a2001-09-20 07:55:22 +00001931#ifdef HAVE_FTRUNCATE
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001932 {"truncate", (PyCFunction)file_truncate, METH_VARARGS, truncate_doc},
Tim Petersefc3a3a2001-09-20 07:55:22 +00001933#endif
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001934 {"tell", (PyCFunction)file_tell, METH_NOARGS, tell_doc},
1935 {"readinto", (PyCFunction)file_readinto, METH_VARARGS, readinto_doc},
1936 {"readlines", (PyCFunction)file_readlines, METH_VARARGS, readlines_doc},
1937 {"xreadlines",(PyCFunction)file_xreadlines, METH_NOARGS, xreadlines_doc},
1938 {"writelines",(PyCFunction)file_writelines, METH_O, writelines_doc},
1939 {"flush", (PyCFunction)file_flush, METH_NOARGS, flush_doc},
1940 {"close", (PyCFunction)file_close, METH_NOARGS, close_doc},
1941 {"isatty", (PyCFunction)file_isatty, METH_NOARGS, isatty_doc},
1942 {"__enter__", (PyCFunction)file_self, METH_NOARGS, enter_doc},
1943 {"__exit__", (PyCFunction)file_exit, METH_VARARGS, exit_doc},
1944 {NULL, NULL} /* sentinel */
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001945};
1946
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001947#define OFF(x) offsetof(PyFileObject, x)
Guido van Rossumb6775db1994-08-01 11:34:53 +00001948
Guido van Rossum6f799372001-09-20 20:46:19 +00001949static PyMemberDef file_memberlist[] = {
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001950 {"mode", T_OBJECT, OFF(f_mode), RO,
1951 "file mode ('r', 'U', 'w', 'a', possibly with 'b' or '+' added)"},
1952 {"name", T_OBJECT, OFF(f_name), RO,
1953 "file name"},
1954 {"encoding", T_OBJECT, OFF(f_encoding), RO,
1955 "file encoding"},
1956 {"errors", T_OBJECT, OFF(f_errors), RO,
1957 "Unicode error handler"},
1958 /* getattr(f, "closed") is implemented without this table */
1959 {NULL} /* Sentinel */
Guido van Rossumb6775db1994-08-01 11:34:53 +00001960};
1961
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001962static PyObject *
Tim Peters6d6c1a32001-08-02 04:15:00 +00001963get_closed(PyFileObject *f, void *closure)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001964{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001965 return PyBool_FromLong((long)(f->f_fp == 0));
Guido van Rossumb6775db1994-08-01 11:34:53 +00001966}
Jack Jansen7b8c7542002-04-14 20:12:41 +00001967static PyObject *
1968get_newlines(PyFileObject *f, void *closure)
1969{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001970 switch (f->f_newlinetypes) {
1971 case NEWLINE_UNKNOWN:
1972 Py_INCREF(Py_None);
1973 return Py_None;
1974 case NEWLINE_CR:
1975 return PyString_FromString("\r");
1976 case NEWLINE_LF:
1977 return PyString_FromString("\n");
1978 case NEWLINE_CR|NEWLINE_LF:
1979 return Py_BuildValue("(ss)", "\r", "\n");
1980 case NEWLINE_CRLF:
1981 return PyString_FromString("\r\n");
1982 case NEWLINE_CR|NEWLINE_CRLF:
1983 return Py_BuildValue("(ss)", "\r", "\r\n");
1984 case NEWLINE_LF|NEWLINE_CRLF:
1985 return Py_BuildValue("(ss)", "\n", "\r\n");
1986 case NEWLINE_CR|NEWLINE_LF|NEWLINE_CRLF:
1987 return Py_BuildValue("(sss)", "\r", "\n", "\r\n");
1988 default:
1989 PyErr_Format(PyExc_SystemError,
1990 "Unknown newlines value 0x%x\n",
1991 f->f_newlinetypes);
1992 return NULL;
1993 }
Jack Jansen7b8c7542002-04-14 20:12:41 +00001994}
Guido van Rossumb6775db1994-08-01 11:34:53 +00001995
Georg Brandl65bb42d2008-03-21 20:38:24 +00001996static PyObject *
1997get_softspace(PyFileObject *f, void *closure)
1998{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00001999 if (PyErr_WarnPy3k("file.softspace not supported in 3.x", 1) < 0)
2000 return NULL;
2001 return PyInt_FromLong(f->f_softspace);
Georg Brandl65bb42d2008-03-21 20:38:24 +00002002}
2003
2004static int
2005set_softspace(PyFileObject *f, PyObject *value)
2006{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002007 int new;
2008 if (PyErr_WarnPy3k("file.softspace not supported in 3.x", 1) < 0)
2009 return -1;
Georg Brandl65bb42d2008-03-21 20:38:24 +00002010
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002011 if (value == NULL) {
2012 PyErr_SetString(PyExc_TypeError,
2013 "can't delete softspace attribute");
2014 return -1;
2015 }
Georg Brandl65bb42d2008-03-21 20:38:24 +00002016
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002017 new = PyInt_AsLong(value);
2018 if (new == -1 && PyErr_Occurred())
2019 return -1;
2020 f->f_softspace = new;
2021 return 0;
Georg Brandl65bb42d2008-03-21 20:38:24 +00002022}
2023
Guido van Rossum32d34c82001-09-20 21:45:26 +00002024static PyGetSetDef file_getsetlist[] = {
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002025 {"closed", (getter)get_closed, NULL, "True if the file is closed"},
2026 {"newlines", (getter)get_newlines, NULL,
2027 "end-of-line convention used in this file"},
2028 {"softspace", (getter)get_softspace, (setter)set_softspace,
2029 "flag indicating that a space needs to be printed; used by print"},
2030 {0},
Tim Peters6d6c1a32001-08-02 04:15:00 +00002031};
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00002032
Neal Norwitzd8b995f2002-08-06 21:50:54 +00002033static void
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002034drop_readahead(PyFileObject *f)
Guido van Rossum65967252001-04-21 13:20:18 +00002035{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002036 if (f->f_buf != NULL) {
2037 PyMem_Free(f->f_buf);
2038 f->f_buf = NULL;
2039 }
Guido van Rossum65967252001-04-21 13:20:18 +00002040}
2041
Tim Petersf1827cf2003-09-07 03:30:18 +00002042/* Make sure that file has a readahead buffer with at least one byte
2043 (unless at EOF) and no more than bufsize. Returns negative value on
Georg Brandled02eb62006-03-31 20:31:02 +00002044 error, will set MemoryError if bufsize bytes cannot be allocated. */
Neal Norwitzd8b995f2002-08-06 21:50:54 +00002045static int
2046readahead(PyFileObject *f, int bufsize)
2047{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002048 Py_ssize_t chunksize;
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002049
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002050 if (f->f_buf != NULL) {
2051 if( (f->f_bufend - f->f_bufptr) >= 1)
2052 return 0;
2053 else
2054 drop_readahead(f);
2055 }
2056 if ((f->f_buf = (char *)PyMem_Malloc(bufsize)) == NULL) {
2057 PyErr_NoMemory();
2058 return -1;
2059 }
2060 FILE_BEGIN_ALLOW_THREADS(f)
2061 errno = 0;
2062 chunksize = Py_UniversalNewlineFread(
2063 f->f_buf, bufsize, f->f_fp, (PyObject *)f);
2064 FILE_END_ALLOW_THREADS(f)
2065 if (chunksize == 0) {
2066 if (ferror(f->f_fp)) {
2067 PyErr_SetFromErrno(PyExc_IOError);
2068 clearerr(f->f_fp);
2069 drop_readahead(f);
2070 return -1;
2071 }
2072 }
2073 f->f_bufptr = f->f_buf;
2074 f->f_bufend = f->f_buf + chunksize;
2075 return 0;
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002076}
2077
2078/* Used by file_iternext. The returned string will start with 'skip'
Tim Petersf1827cf2003-09-07 03:30:18 +00002079 uninitialized bytes followed by the remainder of the line. Don't be
2080 horrified by the recursive call: maximum recursion depth is limited by
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002081 logarithmic buffer growth to about 50 even when reading a 1gb line. */
2082
Gregory P. Smithdd96db62008-06-09 04:58:54 +00002083static PyStringObject *
Neal Norwitzd8b995f2002-08-06 21:50:54 +00002084readahead_get_line_skip(PyFileObject *f, int skip, int bufsize)
2085{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002086 PyStringObject* s;
2087 char *bufptr;
2088 char *buf;
2089 Py_ssize_t len;
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002090
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002091 if (f->f_buf == NULL)
2092 if (readahead(f, bufsize) < 0)
2093 return NULL;
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002094
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002095 len = f->f_bufend - f->f_bufptr;
2096 if (len == 0)
2097 return (PyStringObject *)
2098 PyString_FromStringAndSize(NULL, skip);
2099 bufptr = (char *)memchr(f->f_bufptr, '\n', len);
2100 if (bufptr != NULL) {
2101 bufptr++; /* Count the '\n' */
2102 len = bufptr - f->f_bufptr;
2103 s = (PyStringObject *)
2104 PyString_FromStringAndSize(NULL, skip+len);
2105 if (s == NULL)
2106 return NULL;
2107 memcpy(PyString_AS_STRING(s)+skip, f->f_bufptr, len);
2108 f->f_bufptr = bufptr;
2109 if (bufptr == f->f_bufend)
2110 drop_readahead(f);
2111 } else {
2112 bufptr = f->f_bufptr;
2113 buf = f->f_buf;
2114 f->f_buf = NULL; /* Force new readahead buffer */
2115 assert(skip+len < INT_MAX);
2116 s = readahead_get_line_skip(
2117 f, (int)(skip+len), bufsize + (bufsize>>2) );
2118 if (s == NULL) {
2119 PyMem_Free(buf);
2120 return NULL;
2121 }
2122 memcpy(PyString_AS_STRING(s)+skip, bufptr, len);
2123 PyMem_Free(buf);
2124 }
2125 return s;
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002126}
2127
2128/* A larger buffer size may actually decrease performance. */
2129#define READAHEAD_BUFSIZE 8192
2130
2131static PyObject *
2132file_iternext(PyFileObject *f)
2133{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002134 PyStringObject* l;
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002135
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002136 if (f->f_fp == NULL)
2137 return err_closed();
2138 if (!f->readable)
2139 return err_mode("reading");
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002140
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002141 l = readahead_get_line_skip(f, 0, READAHEAD_BUFSIZE);
2142 if (l == NULL || PyString_GET_SIZE(l) == 0) {
2143 Py_XDECREF(l);
2144 return NULL;
2145 }
2146 return (PyObject *)l;
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002147}
2148
2149
Tim Peters59c9a642001-09-13 05:38:56 +00002150static PyObject *
2151file_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
2152{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002153 PyObject *self;
2154 static PyObject *not_yet_string;
Tim Peters44410012001-09-14 03:26:08 +00002155
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002156 assert(type != NULL && type->tp_alloc != NULL);
Tim Peters44410012001-09-14 03:26:08 +00002157
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002158 if (not_yet_string == NULL) {
2159 not_yet_string = PyString_InternFromString("<uninitialized file>");
2160 if (not_yet_string == NULL)
2161 return NULL;
2162 }
Tim Peters44410012001-09-14 03:26:08 +00002163
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002164 self = type->tp_alloc(type, 0);
2165 if (self != NULL) {
2166 /* Always fill in the name and mode, so that nobody else
2167 needs to special-case NULLs there. */
2168 Py_INCREF(not_yet_string);
2169 ((PyFileObject *)self)->f_name = not_yet_string;
2170 Py_INCREF(not_yet_string);
2171 ((PyFileObject *)self)->f_mode = not_yet_string;
2172 Py_INCREF(Py_None);
2173 ((PyFileObject *)self)->f_encoding = Py_None;
2174 Py_INCREF(Py_None);
2175 ((PyFileObject *)self)->f_errors = Py_None;
2176 ((PyFileObject *)self)->weakreflist = NULL;
2177 ((PyFileObject *)self)->unlocked_count = 0;
2178 }
2179 return self;
Tim Peters44410012001-09-14 03:26:08 +00002180}
2181
2182static int
2183file_init(PyObject *self, PyObject *args, PyObject *kwds)
2184{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002185 PyFileObject *foself = (PyFileObject *)self;
2186 int ret = 0;
2187 static char *kwlist[] = {"name", "mode", "buffering", 0};
2188 char *name = NULL;
2189 char *mode = "r";
2190 int bufsize = -1;
2191 int wideargument = 0;
Tim Peters44410012001-09-14 03:26:08 +00002192
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002193 assert(PyFile_Check(self));
2194 if (foself->f_fp != NULL) {
2195 /* Have to close the existing file first. */
2196 PyObject *closeresult = file_close(foself);
2197 if (closeresult == NULL)
2198 return -1;
2199 Py_DECREF(closeresult);
2200 }
Tim Peters59c9a642001-09-13 05:38:56 +00002201
Mark Hammondc2e85bd2002-10-03 05:10:39 +00002202#ifdef Py_WIN_WIDE_FILENAMES
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002203 if (GetVersion() < 0x80000000) { /* On NT, so wide API available */
2204 PyObject *po;
2205 if (PyArg_ParseTupleAndKeywords(args, kwds, "U|si:file",
2206 kwlist, &po, &mode, &bufsize)) {
2207 wideargument = 1;
2208 if (fill_file_fields(foself, NULL, po, mode,
2209 fclose) == NULL)
2210 goto Error;
2211 } else {
2212 /* Drop the argument parsing error as narrow
2213 strings are also valid. */
2214 PyErr_Clear();
2215 }
2216 }
Mark Hammondc2e85bd2002-10-03 05:10:39 +00002217#endif
2218
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002219 if (!wideargument) {
2220 PyObject *o_name;
Nicholas Bastinabce8a62004-03-21 20:24:07 +00002221
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002222 if (!PyArg_ParseTupleAndKeywords(args, kwds, "et|si:file", kwlist,
2223 Py_FileSystemDefaultEncoding,
2224 &name,
2225 &mode, &bufsize))
2226 return -1;
Nicholas Bastinabce8a62004-03-21 20:24:07 +00002227
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002228 /* We parse again to get the name as a PyObject */
2229 if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|si:file",
2230 kwlist, &o_name, &mode,
2231 &bufsize))
2232 goto Error;
Nicholas Bastinabce8a62004-03-21 20:24:07 +00002233
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002234 if (fill_file_fields(foself, NULL, o_name, mode,
2235 fclose) == NULL)
2236 goto Error;
2237 }
2238 if (open_the_file(foself, name, mode) == NULL)
2239 goto Error;
2240 foself->f_setbuf = NULL;
2241 PyFile_SetBufSize(self, bufsize);
2242 goto Done;
Tim Peters44410012001-09-14 03:26:08 +00002243
2244Error:
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002245 ret = -1;
2246 /* fall through */
Tim Peters44410012001-09-14 03:26:08 +00002247Done:
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002248 PyMem_Free(name); /* free the encoded string */
2249 return ret;
Tim Peters59c9a642001-09-13 05:38:56 +00002250}
2251
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00002252PyDoc_VAR(file_doc) =
2253PyDoc_STR(
Tim Peters59c9a642001-09-13 05:38:56 +00002254"file(name[, mode[, buffering]]) -> file object\n"
2255"\n"
2256"Open a file. The mode can be 'r', 'w' or 'a' for reading (default),\n"
2257"writing or appending. The file will be created if it doesn't exist\n"
2258"when opened for writing or appending; it will be truncated when\n"
2259"opened for writing. Add a 'b' to the mode for binary files.\n"
2260"Add a '+' to the mode to allow simultaneous reading and writing.\n"
2261"If the buffering argument is given, 0 means unbuffered, 1 means line\n"
Skip Montanaro4e3ebe02007-12-08 14:37:43 +00002262"buffered, and larger numbers specify the buffer size. The preferred way\n"
2263"to open a file is with the builtin open() function.\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00002264)
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00002265PyDoc_STR(
Barry Warsaw4be55b52002-05-22 20:37:53 +00002266"Add a 'U' to mode to open the file for input with universal newline\n"
2267"support. Any line ending in the input file will be seen as a '\\n'\n"
2268"in Python. Also, a file so opened gains the attribute 'newlines';\n"
2269"the value for this attribute is one of None (no newline read yet),\n"
2270"'\\r', '\\n', '\\r\\n' or a tuple containing all the newline types seen.\n"
2271"\n"
2272"'U' cannot be combined with 'w' or '+' mode.\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00002273);
Tim Peters59c9a642001-09-13 05:38:56 +00002274
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002275PyTypeObject PyFile_Type = {
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002276 PyVarObject_HEAD_INIT(&PyType_Type, 0)
2277 "file",
2278 sizeof(PyFileObject),
2279 0,
2280 (destructor)file_dealloc, /* tp_dealloc */
2281 0, /* tp_print */
2282 0, /* tp_getattr */
2283 0, /* tp_setattr */
2284 0, /* tp_compare */
2285 (reprfunc)file_repr, /* tp_repr */
2286 0, /* tp_as_number */
2287 0, /* tp_as_sequence */
2288 0, /* tp_as_mapping */
2289 0, /* tp_hash */
2290 0, /* tp_call */
2291 0, /* tp_str */
2292 PyObject_GenericGetAttr, /* tp_getattro */
2293 /* softspace is writable: we must supply tp_setattro */
2294 PyObject_GenericSetAttr, /* tp_setattro */
2295 0, /* tp_as_buffer */
2296 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | Py_TPFLAGS_HAVE_WEAKREFS, /* tp_flags */
2297 file_doc, /* tp_doc */
2298 0, /* tp_traverse */
2299 0, /* tp_clear */
2300 0, /* tp_richcompare */
2301 offsetof(PyFileObject, weakreflist), /* tp_weaklistoffset */
2302 (getiterfunc)file_self, /* tp_iter */
2303 (iternextfunc)file_iternext, /* tp_iternext */
2304 file_methods, /* tp_methods */
2305 file_memberlist, /* tp_members */
2306 file_getsetlist, /* tp_getset */
2307 0, /* tp_base */
2308 0, /* tp_dict */
2309 0, /* tp_descr_get */
2310 0, /* tp_descr_set */
2311 0, /* tp_dictoffset */
2312 file_init, /* tp_init */
2313 PyType_GenericAlloc, /* tp_alloc */
2314 file_new, /* tp_new */
2315 PyObject_Del, /* tp_free */
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00002316};
Guido van Rossumeb183da1991-04-04 10:44:06 +00002317
2318/* Interface for the 'soft space' between print items. */
2319
2320int
Fred Drakefd99de62000-07-09 05:02:18 +00002321PyFile_SoftSpace(PyObject *f, int newflag)
Guido van Rossumeb183da1991-04-04 10:44:06 +00002322{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002323 long oldflag = 0;
2324 if (f == NULL) {
2325 /* Do nothing */
2326 }
2327 else if (PyFile_Check(f)) {
2328 oldflag = ((PyFileObject *)f)->f_softspace;
2329 ((PyFileObject *)f)->f_softspace = newflag;
2330 }
2331 else {
2332 PyObject *v;
2333 v = PyObject_GetAttrString(f, "softspace");
2334 if (v == NULL)
2335 PyErr_Clear();
2336 else {
2337 if (PyInt_Check(v))
2338 oldflag = PyInt_AsLong(v);
2339 assert(oldflag < INT_MAX);
2340 Py_DECREF(v);
2341 }
2342 v = PyInt_FromLong((long)newflag);
2343 if (v == NULL)
2344 PyErr_Clear();
2345 else {
2346 if (PyObject_SetAttrString(f, "softspace", v) != 0)
2347 PyErr_Clear();
2348 Py_DECREF(v);
2349 }
2350 }
2351 return (int)oldflag;
Guido van Rossumeb183da1991-04-04 10:44:06 +00002352}
Guido van Rossum3165fe61992-09-25 21:59:05 +00002353
2354/* Interfaces to write objects/strings to file-like objects */
2355
2356int
Fred Drakefd99de62000-07-09 05:02:18 +00002357PyFile_WriteObject(PyObject *v, PyObject *f, int flags)
Guido van Rossum3165fe61992-09-25 21:59:05 +00002358{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002359 PyObject *writer, *value, *args, *result;
2360 if (f == NULL) {
2361 PyErr_SetString(PyExc_TypeError, "writeobject with NULL file");
2362 return -1;
2363 }
2364 else if (PyFile_Check(f)) {
2365 PyFileObject *fobj = (PyFileObject *) f;
Fred Drake086a0f72004-03-19 15:22:36 +00002366#ifdef Py_USING_UNICODE
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002367 PyObject *enc = fobj->f_encoding;
2368 int result;
Fred Drake086a0f72004-03-19 15:22:36 +00002369#endif
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002370 if (fobj->f_fp == NULL) {
2371 err_closed();
2372 return -1;
2373 }
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00002374#ifdef Py_USING_UNICODE
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002375 if ((flags & Py_PRINT_RAW) &&
2376 PyUnicode_Check(v) && enc != Py_None) {
2377 char *cenc = PyString_AS_STRING(enc);
2378 char *errors = fobj->f_errors == Py_None ?
2379 "strict" : PyString_AS_STRING(fobj->f_errors);
2380 value = PyUnicode_AsEncodedString(v, cenc, errors);
2381 if (value == NULL)
2382 return -1;
2383 } else {
2384 value = v;
2385 Py_INCREF(value);
2386 }
2387 result = file_PyObject_Print(value, fobj, flags);
2388 Py_DECREF(value);
2389 return result;
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00002390#else
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002391 return file_PyObject_Print(v, fobj, flags);
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00002392#endif
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002393 }
2394 writer = PyObject_GetAttrString(f, "write");
2395 if (writer == NULL)
2396 return -1;
2397 if (flags & Py_PRINT_RAW) {
2398 if (PyUnicode_Check(v)) {
2399 value = v;
2400 Py_INCREF(value);
2401 } else
2402 value = PyObject_Str(v);
2403 }
2404 else
2405 value = PyObject_Repr(v);
2406 if (value == NULL) {
2407 Py_DECREF(writer);
2408 return -1;
2409 }
2410 args = PyTuple_Pack(1, value);
2411 if (args == NULL) {
2412 Py_DECREF(value);
2413 Py_DECREF(writer);
2414 return -1;
2415 }
2416 result = PyEval_CallObject(writer, args);
2417 Py_DECREF(args);
2418 Py_DECREF(value);
2419 Py_DECREF(writer);
2420 if (result == NULL)
2421 return -1;
2422 Py_DECREF(result);
2423 return 0;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002424}
2425
Guido van Rossum27a60b11997-05-22 22:25:11 +00002426int
Tim Petersc1bbcb82001-11-28 22:13:25 +00002427PyFile_WriteString(const char *s, PyObject *f)
Guido van Rossum3165fe61992-09-25 21:59:05 +00002428{
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00002429
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002430 if (f == NULL) {
2431 /* Should be caused by a pre-existing error */
2432 if (!PyErr_Occurred())
2433 PyErr_SetString(PyExc_SystemError,
2434 "null file for PyFile_WriteString");
2435 return -1;
2436 }
2437 else if (PyFile_Check(f)) {
2438 PyFileObject *fobj = (PyFileObject *) f;
2439 FILE *fp = PyFile_AsFile(f);
2440 if (fp == NULL) {
2441 err_closed();
2442 return -1;
2443 }
2444 FILE_BEGIN_ALLOW_THREADS(fobj)
2445 fputs(s, fp);
2446 FILE_END_ALLOW_THREADS(fobj)
2447 return 0;
2448 }
2449 else if (!PyErr_Occurred()) {
2450 PyObject *v = PyString_FromString(s);
2451 int err;
2452 if (v == NULL)
2453 return -1;
2454 err = PyFile_WriteObject(v, f, Py_PRINT_RAW);
2455 Py_DECREF(v);
2456 return err;
2457 }
2458 else
2459 return -1;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002460}
Andrew M. Kuchling06051ed2000-07-13 23:56:54 +00002461
2462/* Try to get a file-descriptor from a Python object. If the object
2463 is an integer or long integer, its value is returned. If not, the
2464 object's fileno() method is called if it exists; the method must return
2465 an integer or long integer, which is returned as the file descriptor value.
2466 -1 is returned on failure.
2467*/
2468
2469int PyObject_AsFileDescriptor(PyObject *o)
2470{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002471 int fd;
2472 PyObject *meth;
Andrew M. Kuchling06051ed2000-07-13 23:56:54 +00002473
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002474 if (PyInt_Check(o)) {
2475 fd = PyInt_AsLong(o);
2476 }
2477 else if (PyLong_Check(o)) {
2478 fd = PyLong_AsLong(o);
2479 }
2480 else if ((meth = PyObject_GetAttrString(o, "fileno")) != NULL)
2481 {
2482 PyObject *fno = PyEval_CallObject(meth, NULL);
2483 Py_DECREF(meth);
2484 if (fno == NULL)
2485 return -1;
Tim Peters86821b22001-01-07 21:19:34 +00002486
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002487 if (PyInt_Check(fno)) {
2488 fd = PyInt_AsLong(fno);
2489 Py_DECREF(fno);
2490 }
2491 else if (PyLong_Check(fno)) {
2492 fd = PyLong_AsLong(fno);
2493 Py_DECREF(fno);
2494 }
2495 else {
2496 PyErr_SetString(PyExc_TypeError,
2497 "fileno() returned a non-integer");
2498 Py_DECREF(fno);
2499 return -1;
2500 }
2501 }
2502 else {
2503 PyErr_SetString(PyExc_TypeError,
2504 "argument must be an int, or have a fileno() method.");
2505 return -1;
2506 }
Andrew M. Kuchling06051ed2000-07-13 23:56:54 +00002507
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002508 if (fd < 0) {
2509 PyErr_Format(PyExc_ValueError,
2510 "file descriptor cannot be a negative integer (%i)",
2511 fd);
2512 return -1;
2513 }
2514 return fd;
Andrew M. Kuchling06051ed2000-07-13 23:56:54 +00002515}
Jack Jansen7b8c7542002-04-14 20:12:41 +00002516
Jack Jansen7b8c7542002-04-14 20:12:41 +00002517/* From here on we need access to the real fgets and fread */
2518#undef fgets
2519#undef fread
2520
2521/*
2522** Py_UniversalNewlineFgets is an fgets variation that understands
2523** all of \r, \n and \r\n conventions.
2524** The stream should be opened in binary mode.
2525** If fobj is NULL the routine always does newline conversion, and
2526** it may peek one char ahead to gobble the second char in \r\n.
2527** If fobj is non-NULL it must be a PyFileObject. In this case there
2528** is no readahead but in stead a flag is used to skip a following
2529** \n on the next read. Also, if the file is open in binary mode
2530** the whole conversion is skipped. Finally, the routine keeps track of
2531** the different types of newlines seen.
2532** Note that we need no error handling: fgets() treats error and eof
2533** identically.
2534*/
2535char *
2536Py_UniversalNewlineFgets(char *buf, int n, FILE *stream, PyObject *fobj)
2537{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002538 char *p = buf;
2539 int c;
2540 int newlinetypes = 0;
2541 int skipnextlf = 0;
2542 int univ_newline = 1;
Tim Peters058b1412002-04-21 07:29:14 +00002543
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002544 if (fobj) {
2545 if (!PyFile_Check(fobj)) {
2546 errno = ENXIO; /* What can you do... */
2547 return NULL;
2548 }
2549 univ_newline = ((PyFileObject *)fobj)->f_univ_newline;
2550 if ( !univ_newline )
2551 return fgets(buf, n, stream);
2552 newlinetypes = ((PyFileObject *)fobj)->f_newlinetypes;
2553 skipnextlf = ((PyFileObject *)fobj)->f_skipnextlf;
2554 }
2555 FLOCKFILE(stream);
2556 c = 'x'; /* Shut up gcc warning */
2557 while (--n > 0 && (c = GETC(stream)) != EOF ) {
2558 if (skipnextlf ) {
2559 skipnextlf = 0;
2560 if (c == '\n') {
2561 /* Seeing a \n here with skipnextlf true
2562 ** means we saw a \r before.
2563 */
2564 newlinetypes |= NEWLINE_CRLF;
2565 c = GETC(stream);
2566 if (c == EOF) break;
2567 } else {
2568 /*
2569 ** Note that c == EOF also brings us here,
2570 ** so we're okay if the last char in the file
2571 ** is a CR.
2572 */
2573 newlinetypes |= NEWLINE_CR;
2574 }
2575 }
2576 if (c == '\r') {
2577 /* A \r is translated into a \n, and we skip
2578 ** an adjacent \n, if any. We don't set the
2579 ** newlinetypes flag until we've seen the next char.
2580 */
2581 skipnextlf = 1;
2582 c = '\n';
2583 } else if ( c == '\n') {
2584 newlinetypes |= NEWLINE_LF;
2585 }
2586 *p++ = c;
2587 if (c == '\n') break;
2588 }
2589 if ( c == EOF && skipnextlf )
2590 newlinetypes |= NEWLINE_CR;
2591 FUNLOCKFILE(stream);
2592 *p = '\0';
2593 if (fobj) {
2594 ((PyFileObject *)fobj)->f_newlinetypes = newlinetypes;
2595 ((PyFileObject *)fobj)->f_skipnextlf = skipnextlf;
2596 } else if ( skipnextlf ) {
2597 /* If we have no file object we cannot save the
2598 ** skipnextlf flag. We have to readahead, which
2599 ** will cause a pause if we're reading from an
2600 ** interactive stream, but that is very unlikely
2601 ** unless we're doing something silly like
2602 ** execfile("/dev/tty").
2603 */
2604 c = GETC(stream);
2605 if ( c != '\n' )
2606 ungetc(c, stream);
2607 }
2608 if (p == buf)
2609 return NULL;
2610 return buf;
Jack Jansen7b8c7542002-04-14 20:12:41 +00002611}
2612
2613/*
2614** Py_UniversalNewlineFread is an fread variation that understands
2615** all of \r, \n and \r\n conventions.
2616** The stream should be opened in binary mode.
2617** fobj must be a PyFileObject. In this case there
2618** is no readahead but in stead a flag is used to skip a following
2619** \n on the next read. Also, if the file is open in binary mode
2620** the whole conversion is skipped. Finally, the routine keeps track of
2621** the different types of newlines seen.
2622*/
2623size_t
Tim Peters058b1412002-04-21 07:29:14 +00002624Py_UniversalNewlineFread(char *buf, size_t n,
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002625 FILE *stream, PyObject *fobj)
Jack Jansen7b8c7542002-04-14 20:12:41 +00002626{
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002627 char *dst = buf;
2628 PyFileObject *f = (PyFileObject *)fobj;
2629 int newlinetypes, skipnextlf;
Tim Peters058b1412002-04-21 07:29:14 +00002630
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002631 assert(buf != NULL);
2632 assert(stream != NULL);
Tim Peters058b1412002-04-21 07:29:14 +00002633
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002634 if (!fobj || !PyFile_Check(fobj)) {
2635 errno = ENXIO; /* What can you do... */
2636 return 0;
2637 }
2638 if (!f->f_univ_newline)
2639 return fread(buf, 1, n, stream);
2640 newlinetypes = f->f_newlinetypes;
2641 skipnextlf = f->f_skipnextlf;
2642 /* Invariant: n is the number of bytes remaining to be filled
2643 * in the buffer.
2644 */
2645 while (n) {
2646 size_t nread;
2647 int shortread;
2648 char *src = dst;
Tim Peters058b1412002-04-21 07:29:14 +00002649
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002650 nread = fread(dst, 1, n, stream);
2651 assert(nread <= n);
2652 if (nread == 0)
2653 break;
Neal Norwitzcb3319f2003-02-09 01:10:02 +00002654
Antoine Pitrouc7c96a92010-05-09 15:15:40 +00002655 n -= nread; /* assuming 1 byte out for each in; will adjust */
2656 shortread = n != 0; /* true iff EOF or error */
2657 while (nread--) {
2658 char c = *src++;
2659 if (c == '\r') {
2660 /* Save as LF and set flag to skip next LF. */
2661 *dst++ = '\n';
2662 skipnextlf = 1;
2663 }
2664 else if (skipnextlf && c == '\n') {
2665 /* Skip LF, and remember we saw CR LF. */
2666 skipnextlf = 0;
2667 newlinetypes |= NEWLINE_CRLF;
2668 ++n;
2669 }
2670 else {
2671 /* Normal char to be stored in buffer. Also
2672 * update the newlinetypes flag if either this
2673 * is an LF or the previous char was a CR.
2674 */
2675 if (c == '\n')
2676 newlinetypes |= NEWLINE_LF;
2677 else if (skipnextlf)
2678 newlinetypes |= NEWLINE_CR;
2679 *dst++ = c;
2680 skipnextlf = 0;
2681 }
2682 }
2683 if (shortread) {
2684 /* If this is EOF, update type flags. */
2685 if (skipnextlf && feof(stream))
2686 newlinetypes |= NEWLINE_CR;
2687 break;
2688 }
2689 }
2690 f->f_newlinetypes = newlinetypes;
2691 f->f_skipnextlf = skipnextlf;
2692 return dst - buf;
Jack Jansen7b8c7542002-04-14 20:12:41 +00002693}
Anthony Baxterac6bd462006-04-13 02:06:09 +00002694
2695#ifdef __cplusplus
2696}
2697#endif