blob: b2051d79fe45737dce9d19e5b7d63ab179e8aed6 [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 */
46#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{ \
61 fobj->unlocked_count++; \
62 Py_BEGIN_ALLOW_THREADS
63
64#define FILE_END_ALLOW_THREADS(fobj) \
65 Py_END_ALLOW_THREADS \
66 fobj->unlocked_count--; \
67 assert(fobj->unlocked_count >= 0); \
68}
69
70#define FILE_ABORT_ALLOW_THREADS(fobj) \
71 Py_BLOCK_THREADS \
72 fobj->unlocked_count--; \
73 assert(fobj->unlocked_count >= 0);
74
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{
Guido van Rossumc0b618a1997-05-02 03:12:38 +000082 if (f == NULL || !PyFile_Check(f))
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000083 return NULL;
Guido van Rossum3165fe61992-09-25 21:59:05 +000084 else
Guido van Rossumc0b618a1997-05-02 03:12:38 +000085 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{
90 fobj->unlocked_count++;
91}
92
93void PyFile_DecUseCount(PyFileObject *fobj)
94{
95 fobj->unlocked_count--;
96 assert(fobj->unlocked_count >= 0);
97}
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{
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000102 if (f == NULL || !PyFile_Check(f))
Guido van Rossumdb3165e1993-10-18 17:06:59 +0000103 return NULL;
104 else
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000105 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{
114 int result;
115 PyFile_IncUseCount(f);
116 result = PyObject_Print(op, f->f_fp, flags);
117 PyFile_DecUseCount(f);
118 return result;
119}
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)
129 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)) {
Neil Schemenauered19b882002-03-23 02:06:50 +0000134 char *msg = strerror(EISDIR);
Tim Petersf1827cf2003-09-07 03:30:18 +0000135 PyObject *exc = PyObject_CallFunction(PyExc_IOError, "(is)",
Jeremy Hylton8b735422002-08-14 21:01:41 +0000136 EISDIR, msg);
Neil Schemenauered19b882002-03-23 02:06:50 +0000137 PyErr_SetObject(PyExc_IOError, exc);
Neal Norwitz98cad482003-08-15 20:05:45 +0000138 Py_XDECREF(exc);
Neil Schemenauered19b882002-03-23 02:06:50 +0000139 return NULL;
140 }
141#endif
142 return f;
143}
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,
148 int (*close)(FILE *))
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000149{
Neal Norwitzb337bb52006-07-17 00:55:45 +0000150 assert(name != NULL);
Tim Peters59c9a642001-09-13 05:38:56 +0000151 assert(f != NULL);
152 assert(PyFile_Check(f));
Tim Peters44410012001-09-14 03:26:08 +0000153 assert(f->f_fp == NULL);
154
155 Py_DECREF(f->f_name);
156 Py_DECREF(f->f_mode);
Martin v. Löwis5467d4c2003-05-10 07:10:12 +0000157 Py_DECREF(f->f_encoding);
Martin v. Löwis99815892008-06-01 07:20:46 +0000158 Py_DECREF(f->f_errors);
Nicholas Bastinabce8a62004-03-21 20:24:07 +0000159
Neal Norwitzb337bb52006-07-17 00:55:45 +0000160 Py_INCREF(name);
Nicholas Bastinabce8a62004-03-21 20:24:07 +0000161 f->f_name = name;
162
Gregory P. Smithdd96db62008-06-09 04:58:54 +0000163 f->f_mode = PyString_FromString(mode);
Tim Peters44410012001-09-14 03:26:08 +0000164
Guido van Rossuma1ab7fa1991-06-04 19:37:39 +0000165 f->f_close = close;
Guido van Rossumeb183da1991-04-04 10:44:06 +0000166 f->f_softspace = 0;
Tim Peters59c9a642001-09-13 05:38:56 +0000167 f->f_binary = strchr(mode,'b') != NULL;
Guido van Rossum7a6e9592002-08-06 15:55:28 +0000168 f->f_buf = NULL;
Jack Jansen7b8c7542002-04-14 20:12:41 +0000169 f->f_univ_newline = (strchr(mode, 'U') != NULL);
170 f->f_newlinetypes = NEWLINE_UNKNOWN;
171 f->f_skipnextlf = 0;
Martin v. Löwis5467d4c2003-05-10 07:10:12 +0000172 Py_INCREF(Py_None);
173 f->f_encoding = Py_None;
Martin v. Löwis99815892008-06-01 07:20:46 +0000174 Py_INCREF(Py_None);
175 f->f_errors = Py_None;
Tim Petersf1827cf2003-09-07 03:30:18 +0000176
Neal Norwitzb337bb52006-07-17 00:55:45 +0000177 if (f->f_mode == NULL)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000178 return NULL;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000179 f->f_fp = fp;
Neil Schemenauered19b882002-03-23 02:06:50 +0000180 f = dircheck(f);
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000181 return (PyObject *) f;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000182}
183
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000184/* check for known incorrect mode strings - problem is, platforms are
185 free to accept any mode characters they like and are supposed to
186 ignore stuff they don't understand... write or append mode with
Georg Brandl7b90e162006-05-18 07:01:27 +0000187 universal newline support is expressly forbidden by PEP 278.
188 Additionally, remove the 'U' from the mode string as platforms
Kristján Valur Jónsson0a440d42007-04-26 09:15:08 +0000189 won't know what it is. Non-zero return signals an exception */
190int
191_PyFile_SanitizeMode(char *mode)
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000192{
Georg Brandl7b90e162006-05-18 07:01:27 +0000193 char *upos;
Neal Norwitz76dc0812006-01-08 06:13:13 +0000194 size_t len = strlen(mode);
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000195
Georg Brandl7b90e162006-05-18 07:01:27 +0000196 if (!len) {
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000197 PyErr_SetString(PyExc_ValueError, "empty mode string");
Kristján Valur Jónsson0a440d42007-04-26 09:15:08 +0000198 return -1;
Georg Brandl7b90e162006-05-18 07:01:27 +0000199 }
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000200
Georg Brandl7b90e162006-05-18 07:01:27 +0000201 upos = strchr(mode, 'U');
202 if (upos) {
203 memmove(upos, upos+1, len-(upos-mode)); /* incl null char */
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000204
Georg Brandl7b90e162006-05-18 07:01:27 +0000205 if (mode[0] == 'w' || mode[0] == 'a') {
206 PyErr_Format(PyExc_ValueError, "universal newline "
207 "mode can only be used with modes "
208 "starting with 'r'");
Kristján Valur Jónsson0a440d42007-04-26 09:15:08 +0000209 return -1;
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000210 }
Georg Brandl7b90e162006-05-18 07:01:27 +0000211
212 if (mode[0] != 'r') {
213 memmove(mode+1, mode, strlen(mode)+1);
214 mode[0] = 'r';
215 }
216
217 if (!strchr(mode, 'b')) {
218 memmove(mode+2, mode+1, strlen(mode));
219 mode[1] = 'b';
220 }
221 } else if (mode[0] != 'r' && mode[0] != 'w' && mode[0] != 'a') {
222 PyErr_Format(PyExc_ValueError, "mode string must begin with "
223 "one of 'r', 'w', 'a' or 'U', not '%.200s'", mode);
Kristján Valur Jónsson0a440d42007-04-26 09:15:08 +0000224 return -1;
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000225 }
226
227 return 0;
228}
229
Tim Peters59c9a642001-09-13 05:38:56 +0000230static PyObject *
231open_the_file(PyFileObject *f, char *name, char *mode)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000232{
Georg Brandl7b90e162006-05-18 07:01:27 +0000233 char *newmode;
Tim Peters59c9a642001-09-13 05:38:56 +0000234 assert(f != NULL);
235 assert(PyFile_Check(f));
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000236#ifdef MS_WINDOWS
237 /* windows ignores the passed name in order to support Unicode */
238 assert(f->f_name != NULL);
239#else
Tim Peters59c9a642001-09-13 05:38:56 +0000240 assert(name != NULL);
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000241#endif
Tim Peters59c9a642001-09-13 05:38:56 +0000242 assert(mode != NULL);
Tim Peters44410012001-09-14 03:26:08 +0000243 assert(f->f_fp == NULL);
Tim Peters59c9a642001-09-13 05:38:56 +0000244
Georg Brandl7b90e162006-05-18 07:01:27 +0000245 /* probably need to replace 'U' by 'rb' */
246 newmode = PyMem_MALLOC(strlen(mode) + 3);
247 if (!newmode) {
248 PyErr_NoMemory();
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000249 return NULL;
Georg Brandl7b90e162006-05-18 07:01:27 +0000250 }
251 strcpy(newmode, mode);
252
Kristján Valur Jónsson0a440d42007-04-26 09:15:08 +0000253 if (_PyFile_SanitizeMode(newmode)) {
Georg Brandl7b90e162006-05-18 07:01:27 +0000254 f = NULL;
255 goto cleanup;
256 }
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000257
Tim Peters8fa45672001-09-13 21:01:29 +0000258 /* rexec.py can't stop a user from getting the file() constructor --
259 all they have to do is get *any* file object f, and then do
260 type(f). Here we prevent them from doing damage with it. */
261 if (PyEval_GetRestricted()) {
262 PyErr_SetString(PyExc_IOError,
Jeremy Hylton8b735422002-08-14 21:01:41 +0000263 "file() constructor not accessible in restricted mode");
Georg Brandl7b90e162006-05-18 07:01:27 +0000264 f = NULL;
265 goto cleanup;
Tim Peters8fa45672001-09-13 21:01:29 +0000266 }
Tim Petersa27a1502001-11-09 20:59:14 +0000267 errno = 0;
Skip Montanaro51ffac62004-06-11 04:49:03 +0000268
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000269#ifdef MS_WINDOWS
Skip Montanaro51ffac62004-06-11 04:49:03 +0000270 if (PyUnicode_Check(f->f_name)) {
271 PyObject *wmode;
Georg Brandl7b90e162006-05-18 07:01:27 +0000272 wmode = PyUnicode_DecodeASCII(newmode, strlen(newmode), NULL);
Skip Montanaro51ffac62004-06-11 04:49:03 +0000273 if (f->f_name && wmode) {
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000274 FILE_BEGIN_ALLOW_THREADS(f)
Skip Montanaro51ffac62004-06-11 04:49:03 +0000275 /* PyUnicode_AS_UNICODE OK without thread
276 lock as it is a simple dereference. */
277 f->f_fp = _wfopen(PyUnicode_AS_UNICODE(f->f_name),
278 PyUnicode_AS_UNICODE(wmode));
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000279 FILE_END_ALLOW_THREADS(f)
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000280 }
Skip Montanaro51ffac62004-06-11 04:49:03 +0000281 Py_XDECREF(wmode);
Guido van Rossumff4949e1992-08-05 19:58:53 +0000282 }
Skip Montanaro51ffac62004-06-11 04:49:03 +0000283#endif
284 if (NULL == f->f_fp && NULL != name) {
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000285 FILE_BEGIN_ALLOW_THREADS(f)
Georg Brandl7b90e162006-05-18 07:01:27 +0000286 f->f_fp = fopen(name, newmode);
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000287 FILE_END_ALLOW_THREADS(f)
Skip Montanaro51ffac62004-06-11 04:49:03 +0000288 }
289
Guido van Rossuma08095a1991-02-13 23:25:27 +0000290 if (f->f_fp == NULL) {
Kristján Valur Jónsson74c3ea02006-07-03 14:59:05 +0000291#if defined _MSC_VER && (_MSC_VER < 1400 || !defined(__STDC_SECURE_LIB__))
Tim Peters2ea91112002-04-08 04:13:12 +0000292 /* MSVC 6 (Microsoft) leaves errno at 0 for bad mode strings,
293 * across all Windows flavors. When it sets EINVAL varies
294 * across Windows flavors, the exact conditions aren't
295 * documented, and the answer lies in the OS's implementation
296 * of Win32's CreateFile function (whose source is secret).
297 * Seems the best we can do is map EINVAL to ENOENT.
Kristján Valur Jónssonf6083172006-06-12 15:45:12 +0000298 * Starting with Visual Studio .NET 2005, EINVAL is correctly
299 * set by our CRT error handler (set in exceptions.c.)
Tim Peters2ea91112002-04-08 04:13:12 +0000300 */
301 if (errno == 0) /* bad mode string */
302 errno = EINVAL;
303 else if (errno == EINVAL) /* unknown, but not a mode string */
304 errno = ENOENT;
305#endif
Gregory P. Smith887290d2008-03-18 00:20:01 +0000306 /* EINVAL is returned when an invalid filename or
307 * an invalid mode is supplied. */
Amaury Forgeot d'Arc17617a02008-09-25 20:52:56 +0000308 if (errno == EINVAL) {
309 PyObject *v;
310 char message[100];
311 PyOS_snprintf(message, 100,
312 "invalid mode ('%.50s') or filename", mode);
313 v = Py_BuildValue("(isO)", errno, message, f->f_name);
314 if (v != NULL) {
315 PyErr_SetObject(PyExc_IOError, v);
316 Py_DECREF(v);
317 }
318 }
Jeremy Hylton41c83212001-11-09 16:17:24 +0000319 else
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000320 PyErr_SetFromErrnoWithFilenameObject(PyExc_IOError, f->f_name);
Tim Peters59c9a642001-09-13 05:38:56 +0000321 f = NULL;
322 }
Tim Peters2ea91112002-04-08 04:13:12 +0000323 if (f != NULL)
Neil Schemenauered19b882002-03-23 02:06:50 +0000324 f = dircheck(f);
Georg Brandl7b90e162006-05-18 07:01:27 +0000325
326cleanup:
327 PyMem_FREE(newmode);
328
Tim Peters59c9a642001-09-13 05:38:56 +0000329 return (PyObject *)f;
330}
331
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000332static PyObject *
333close_the_file(PyFileObject *f)
334{
335 int sts = 0;
336 int (*local_close)(FILE *);
337 FILE *local_fp = f->f_fp;
338 if (local_fp != NULL) {
339 local_close = f->f_close;
340 if (local_close != NULL && f->unlocked_count > 0) {
341 if (f->ob_refcnt > 0) {
342 PyErr_SetString(PyExc_IOError,
343 "close() called during concurrent "
344 "operation on the same file object.");
345 } else {
346 /* This should not happen unless someone is
347 * carelessly playing with the PyFileObject
348 * struct fields and/or its associated FILE
349 * pointer. */
350 PyErr_SetString(PyExc_SystemError,
351 "PyFileObject locking error in "
352 "destructor (refcnt <= 0 at close).");
353 }
354 return NULL;
355 }
356 /* NULL out the FILE pointer before releasing the GIL, because
357 * it will not be valid anymore after the close() function is
358 * called. */
359 f->f_fp = NULL;
360 if (local_close != NULL) {
361 Py_BEGIN_ALLOW_THREADS
362 errno = 0;
363 sts = (*local_close)(local_fp);
364 Py_END_ALLOW_THREADS
365 if (sts == EOF)
366 return PyErr_SetFromErrno(PyExc_IOError);
367 if (sts != 0)
368 return PyInt_FromLong((long)sts);
369 }
370 }
371 Py_RETURN_NONE;
372}
373
Tim Peters59c9a642001-09-13 05:38:56 +0000374PyObject *
375PyFile_FromFile(FILE *fp, char *name, char *mode, int (*close)(FILE *))
376{
Tim Peters44410012001-09-14 03:26:08 +0000377 PyFileObject *f = (PyFileObject *)PyFile_Type.tp_new(&PyFile_Type,
378 NULL, NULL);
Tim Peters59c9a642001-09-13 05:38:56 +0000379 if (f != NULL) {
Gregory P. Smithdd96db62008-06-09 04:58:54 +0000380 PyObject *o_name = PyString_FromString(name);
Neal Norwitzb337bb52006-07-17 00:55:45 +0000381 if (o_name == NULL)
382 return NULL;
Nicholas Bastinabce8a62004-03-21 20:24:07 +0000383 if (fill_file_fields(f, fp, o_name, mode, close) == NULL) {
Tim Peters59c9a642001-09-13 05:38:56 +0000384 Py_DECREF(f);
385 f = NULL;
386 }
Nicholas Bastinabce8a62004-03-21 20:24:07 +0000387 Py_DECREF(o_name);
Tim Peters59c9a642001-09-13 05:38:56 +0000388 }
389 return (PyObject *) f;
390}
391
392PyObject *
393PyFile_FromString(char *name, char *mode)
394{
395 extern int fclose(FILE *);
396 PyFileObject *f;
397
398 f = (PyFileObject *)PyFile_FromFile((FILE *)NULL, name, mode, fclose);
399 if (f != NULL) {
400 if (open_the_file(f, name, mode) == NULL) {
401 Py_DECREF(f);
402 f = NULL;
403 }
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000404 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000405 return (PyObject *)f;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000406}
407
Guido van Rossumb6775db1994-08-01 11:34:53 +0000408void
Fred Drakefd99de62000-07-09 05:02:18 +0000409PyFile_SetBufSize(PyObject *f, int bufsize)
Guido van Rossumb6775db1994-08-01 11:34:53 +0000410{
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000411 PyFileObject *file = (PyFileObject *)f;
Guido van Rossumb6775db1994-08-01 11:34:53 +0000412 if (bufsize >= 0) {
Guido van Rossumb6775db1994-08-01 11:34:53 +0000413 int type;
414 switch (bufsize) {
415 case 0:
416 type = _IONBF;
417 break;
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000418#ifdef HAVE_SETVBUF
Guido van Rossumb6775db1994-08-01 11:34:53 +0000419 case 1:
420 type = _IOLBF;
421 bufsize = BUFSIZ;
422 break;
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000423#endif
Guido van Rossumb6775db1994-08-01 11:34:53 +0000424 default:
425 type = _IOFBF;
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000426#ifndef HAVE_SETVBUF
427 bufsize = BUFSIZ;
428#endif
429 break;
Guido van Rossumb6775db1994-08-01 11:34:53 +0000430 }
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000431 fflush(file->f_fp);
432 if (type == _IONBF) {
433 PyMem_Free(file->f_setbuf);
434 file->f_setbuf = NULL;
435 } else {
Anthony Baxter377be112006-04-11 06:54:30 +0000436 file->f_setbuf = (char *)PyMem_Realloc(file->f_setbuf,
437 bufsize);
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000438 }
439#ifdef HAVE_SETVBUF
440 setvbuf(file->f_fp, file->f_setbuf, type, bufsize);
Guido van Rossumf8b4de01998-03-06 15:32:40 +0000441#else /* !HAVE_SETVBUF */
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000442 setbuf(file->f_fp, file->f_setbuf);
Guido van Rossumf8b4de01998-03-06 15:32:40 +0000443#endif /* !HAVE_SETVBUF */
Guido van Rossumb6775db1994-08-01 11:34:53 +0000444 }
445}
446
Martin v. Löwis5467d4c2003-05-10 07:10:12 +0000447/* Set the encoding used to output Unicode strings.
Martin v. Löwis99815892008-06-01 07:20:46 +0000448 Return 1 on success, 0 on failure. */
Martin v. Löwis5467d4c2003-05-10 07:10:12 +0000449
450int
451PyFile_SetEncoding(PyObject *f, const char *enc)
452{
Martin v. Löwis99815892008-06-01 07:20:46 +0000453 return PyFile_SetEncodingAndErrors(f, enc, NULL);
454}
455
456int
457PyFile_SetEncodingAndErrors(PyObject *f, const char *enc, char* errors)
458{
Martin v. Löwis5467d4c2003-05-10 07:10:12 +0000459 PyFileObject *file = (PyFileObject*)f;
Martin v. Löwis99815892008-06-01 07:20:46 +0000460 PyObject *str, *oerrors;
Thomas Woutersafea5292007-01-23 13:42:00 +0000461
462 assert(PyFile_Check(f));
Gregory P. Smith99a3dce2008-06-10 17:42:36 +0000463 str = PyString_FromString(enc);
Martin v. Löwis5467d4c2003-05-10 07:10:12 +0000464 if (!str)
465 return 0;
Martin v. Löwis99815892008-06-01 07:20:46 +0000466 if (errors) {
467 oerrors = PyString_FromString(errors);
468 if (!oerrors) {
469 Py_DECREF(str);
470 return 0;
471 }
472 } else {
473 oerrors = Py_None;
474 Py_INCREF(Py_None);
475 }
Martin v. Löwis5467d4c2003-05-10 07:10:12 +0000476 Py_DECREF(file->f_encoding);
477 file->f_encoding = str;
Martin v. Löwis99815892008-06-01 07:20:46 +0000478 Py_DECREF(file->f_errors);
479 file->f_errors = oerrors;
Martin v. Löwis5467d4c2003-05-10 07:10:12 +0000480 return 1;
481}
482
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000483static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000484err_closed(void)
Guido van Rossumd7297e61992-07-06 14:19:26 +0000485{
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000486 PyErr_SetString(PyExc_ValueError, "I/O operation on closed file");
Guido van Rossumd7297e61992-07-06 14:19:26 +0000487 return NULL;
488}
489
Thomas Woutersc45251a2006-02-12 11:53:32 +0000490/* Refuse regular file I/O if there's data in the iteration-buffer.
491 * Mixing them would cause data to arrive out of order, as the read*
492 * methods don't use the iteration buffer. */
493static PyObject *
494err_iterbuffered(void)
495{
496 PyErr_SetString(PyExc_ValueError,
497 "Mixing iteration and read methods would lose data");
498 return NULL;
499}
500
Neal Norwitzd8b995f2002-08-06 21:50:54 +0000501static void drop_readahead(PyFileObject *);
Guido van Rossum7a6e9592002-08-06 15:55:28 +0000502
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000503/* Methods */
504
505static void
Fred Drakefd99de62000-07-09 05:02:18 +0000506file_dealloc(PyFileObject *f)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000507{
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000508 PyObject *ret;
Raymond Hettingercb87bc82004-05-31 00:35:52 +0000509 if (f->weakreflist != NULL)
510 PyObject_ClearWeakRefs((PyObject *) f);
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000511 ret = close_the_file(f);
512 if (!ret) {
513 PySys_WriteStderr("close failed in file object destructor:\n");
514 PyErr_Print();
515 }
516 else {
517 Py_DECREF(ret);
Guido van Rossumff4949e1992-08-05 19:58:53 +0000518 }
Andrew MacIntyre4e10ed32004-04-04 07:01:35 +0000519 PyMem_Free(f->f_setbuf);
Tim Peters44410012001-09-14 03:26:08 +0000520 Py_XDECREF(f->f_name);
521 Py_XDECREF(f->f_mode);
Martin v. Löwis5467d4c2003-05-10 07:10:12 +0000522 Py_XDECREF(f->f_encoding);
Martin v. Löwis99815892008-06-01 07:20:46 +0000523 Py_XDECREF(f->f_errors);
Guido van Rossum7a6e9592002-08-06 15:55:28 +0000524 drop_readahead(f);
Christian Heimese93237d2007-12-19 02:37:44 +0000525 Py_TYPE(f)->tp_free((PyObject *)f);
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000526}
527
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000528static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000529file_repr(PyFileObject *f)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000530{
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000531 if (PyUnicode_Check(f->f_name)) {
Martin v. Löwis0073f2e2002-11-21 23:52:35 +0000532#ifdef Py_USING_UNICODE
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000533 PyObject *ret = NULL;
Neal Norwitzfc28e0d2006-07-16 02:32:03 +0000534 PyObject *name = PyUnicode_AsUnicodeEscapeString(f->f_name);
Gregory P. Smithdd96db62008-06-09 04:58:54 +0000535 const char *name_str = name ? PyString_AsString(name) : "?";
536 ret = PyString_FromFormat("<%s file u'%s', mode '%s' at %p>",
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000537 f->f_fp == NULL ? "closed" : "open",
Neal Norwitzfc28e0d2006-07-16 02:32:03 +0000538 name_str,
Gregory P. Smithdd96db62008-06-09 04:58:54 +0000539 PyString_AsString(f->f_mode),
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000540 f);
541 Py_XDECREF(name);
542 return ret;
Martin v. Löwis0073f2e2002-11-21 23:52:35 +0000543#endif
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000544 } else {
Gregory P. Smithdd96db62008-06-09 04:58:54 +0000545 return PyString_FromFormat("<%s file '%s', mode '%s' at %p>",
Barry Warsaw7ce36942001-08-24 18:34:26 +0000546 f->f_fp == NULL ? "closed" : "open",
Gregory P. Smithdd96db62008-06-09 04:58:54 +0000547 PyString_AsString(f->f_name),
548 PyString_AsString(f->f_mode),
Barry Warsaw7ce36942001-08-24 18:34:26 +0000549 f);
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000550 }
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000551}
552
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000553static PyObject *
Martin v. Löwise3eb1f22001-08-16 13:15:00 +0000554file_close(PyFileObject *f)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000555{
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000556 PyObject *sts = close_the_file(f);
Martin v. Löwis7bbcde72003-09-07 20:42:29 +0000557 PyMem_Free(f->f_setbuf);
Andrew MacIntyre4e10ed32004-04-04 07:01:35 +0000558 f->f_setbuf = NULL;
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000559 return sts;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000560}
561
Trent Mickf29f47b2000-08-11 19:02:59 +0000562
Guido van Rossumb8552162001-09-05 14:58:11 +0000563/* Our very own off_t-like type, 64-bit if possible */
564#if !defined(HAVE_LARGEFILE_SUPPORT)
565typedef off_t Py_off_t;
566#elif SIZEOF_OFF_T >= 8
567typedef off_t Py_off_t;
568#elif SIZEOF_FPOS_T >= 8
Guido van Rossum4f53da02001-03-01 18:26:53 +0000569typedef fpos_t Py_off_t;
570#else
Guido van Rossumb8552162001-09-05 14:58:11 +0000571#error "Large file support, but neither off_t nor fpos_t is large enough."
Guido van Rossum4f53da02001-03-01 18:26:53 +0000572#endif
573
574
Trent Mickf29f47b2000-08-11 19:02:59 +0000575/* a portable fseek() function
576 return 0 on success, non-zero on failure (with errno set) */
Guido van Rossumf68d8e52001-04-14 17:55:09 +0000577static int
Guido van Rossum4f53da02001-03-01 18:26:53 +0000578_portable_fseek(FILE *fp, Py_off_t offset, int whence)
Trent Mickf29f47b2000-08-11 19:02:59 +0000579{
Guido van Rossumb8552162001-09-05 14:58:11 +0000580#if !defined(HAVE_LARGEFILE_SUPPORT)
581 return fseek(fp, offset, whence);
582#elif defined(HAVE_FSEEKO) && SIZEOF_OFF_T >= 8
Trent Mickf29f47b2000-08-11 19:02:59 +0000583 return fseeko(fp, offset, whence);
584#elif defined(HAVE_FSEEK64)
585 return fseek64(fp, offset, whence);
Fred Drakedb810ac2000-10-06 20:42:33 +0000586#elif defined(__BEOS__)
587 return _fseek(fp, offset, whence);
Guido van Rossumb8552162001-09-05 14:58:11 +0000588#elif SIZEOF_FPOS_T >= 8
Guido van Rossume54e0be2001-01-16 20:53:31 +0000589 /* lacking a 64-bit capable fseek(), use a 64-bit capable fsetpos()
590 and fgetpos() to implement fseek()*/
Trent Mickf29f47b2000-08-11 19:02:59 +0000591 fpos_t pos;
592 switch (whence) {
Guido van Rossume54e0be2001-01-16 20:53:31 +0000593 case SEEK_END:
Guido van Rossum8b4e43e2001-09-10 20:43:35 +0000594#ifdef MS_WINDOWS
595 fflush(fp);
596 if (_lseeki64(fileno(fp), 0, 2) == -1)
597 return -1;
598#else
Guido van Rossume54e0be2001-01-16 20:53:31 +0000599 if (fseek(fp, 0, SEEK_END) != 0)
600 return -1;
Guido van Rossum8b4e43e2001-09-10 20:43:35 +0000601#endif
Guido van Rossume54e0be2001-01-16 20:53:31 +0000602 /* fall through */
603 case SEEK_CUR:
604 if (fgetpos(fp, &pos) != 0)
605 return -1;
606 offset += pos;
607 break;
608 /* case SEEK_SET: break; */
Trent Mickf29f47b2000-08-11 19:02:59 +0000609 }
610 return fsetpos(fp, &offset);
611#else
Guido van Rossumb8552162001-09-05 14:58:11 +0000612#error "Large file support, but no way to fseek."
Trent Mickf29f47b2000-08-11 19:02:59 +0000613#endif
614}
615
616
617/* a portable ftell() function
618 Return -1 on failure with errno set appropriately, current file
619 position on success */
Guido van Rossumf68d8e52001-04-14 17:55:09 +0000620static Py_off_t
Fred Drake8ce159a2000-08-31 05:18:54 +0000621_portable_ftell(FILE* fp)
Trent Mickf29f47b2000-08-11 19:02:59 +0000622{
Guido van Rossumb8552162001-09-05 14:58:11 +0000623#if !defined(HAVE_LARGEFILE_SUPPORT)
624 return ftell(fp);
625#elif defined(HAVE_FTELLO) && SIZEOF_OFF_T >= 8
626 return ftello(fp);
627#elif defined(HAVE_FTELL64)
628 return ftell64(fp);
629#elif SIZEOF_FPOS_T >= 8
Trent Mickf29f47b2000-08-11 19:02:59 +0000630 fpos_t pos;
631 if (fgetpos(fp, &pos) != 0)
632 return -1;
633 return pos;
634#else
Guido van Rossumb8552162001-09-05 14:58:11 +0000635#error "Large file support, but no way to ftell."
Trent Mickf29f47b2000-08-11 19:02:59 +0000636#endif
637}
638
639
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000640static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000641file_seek(PyFileObject *f, PyObject *args)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000642{
Guido van Rossumd7297e61992-07-06 14:19:26 +0000643 int whence;
Guido van Rossumff4949e1992-08-05 19:58:53 +0000644 int ret;
Guido van Rossum4f53da02001-03-01 18:26:53 +0000645 Py_off_t offset;
Martin v. Löwis056dac12006-11-12 18:24:26 +0000646 PyObject *offobj, *off_index;
Tim Peters86821b22001-01-07 21:19:34 +0000647
Guido van Rossumd7297e61992-07-06 14:19:26 +0000648 if (f->f_fp == NULL)
649 return err_closed();
Guido van Rossum7a6e9592002-08-06 15:55:28 +0000650 drop_readahead(f);
Guido van Rossumd7297e61992-07-06 14:19:26 +0000651 whence = 0;
Guido van Rossum43713e52000-02-29 13:59:29 +0000652 if (!PyArg_ParseTuple(args, "O|i:seek", &offobj, &whence))
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000653 return NULL;
Martin v. Löwis056dac12006-11-12 18:24:26 +0000654 off_index = PyNumber_Index(offobj);
655 if (!off_index) {
656 if (!PyFloat_Check(offobj))
657 return NULL;
658 /* Deprecated in 2.6 */
659 PyErr_Clear();
Benjamin Petersonf19a7b92008-04-27 18:40:21 +0000660 if (PyErr_WarnEx(PyExc_DeprecationWarning,
661 "integer argument expected, got float",
662 1) < 0)
Martin v. Löwis056dac12006-11-12 18:24:26 +0000663 return NULL;
664 off_index = offobj;
665 Py_INCREF(offobj);
666 }
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000667#if !defined(HAVE_LARGEFILE_SUPPORT)
Martin v. Löwis056dac12006-11-12 18:24:26 +0000668 offset = PyInt_AsLong(off_index);
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000669#else
Martin v. Löwis056dac12006-11-12 18:24:26 +0000670 offset = PyLong_Check(off_index) ?
671 PyLong_AsLongLong(off_index) : PyInt_AsLong(off_index);
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000672#endif
Martin v. Löwis056dac12006-11-12 18:24:26 +0000673 Py_DECREF(off_index);
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000674 if (PyErr_Occurred())
Guido van Rossum88303191999-01-04 17:22:18 +0000675 return NULL;
Tim Peters86821b22001-01-07 21:19:34 +0000676
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000677 FILE_BEGIN_ALLOW_THREADS(f)
Guido van Rossumce5ba841991-03-06 13:06:18 +0000678 errno = 0;
Trent Mickf29f47b2000-08-11 19:02:59 +0000679 ret = _portable_fseek(f->f_fp, offset, whence);
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000680 FILE_END_ALLOW_THREADS(f)
Trent Mickf29f47b2000-08-11 19:02:59 +0000681
Guido van Rossumff4949e1992-08-05 19:58:53 +0000682 if (ret != 0) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000683 PyErr_SetFromErrno(PyExc_IOError);
Guido van Rossumfebd5511992-03-04 16:39:24 +0000684 clearerr(f->f_fp);
685 return NULL;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000686 }
Jack Jansen7b8c7542002-04-14 20:12:41 +0000687 f->f_skipnextlf = 0;
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000688 Py_INCREF(Py_None);
689 return Py_None;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000690}
691
Trent Mickf29f47b2000-08-11 19:02:59 +0000692
Guido van Rossumd7047b31995-01-02 19:07:15 +0000693#ifdef HAVE_FTRUNCATE
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000694static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000695file_truncate(PyFileObject *f, PyObject *args)
Guido van Rossumd7047b31995-01-02 19:07:15 +0000696{
Guido van Rossum4f53da02001-03-01 18:26:53 +0000697 Py_off_t newsize;
Tim Petersf1827cf2003-09-07 03:30:18 +0000698 PyObject *newsizeobj = NULL;
699 Py_off_t initialpos;
700 int ret;
Tim Peters86821b22001-01-07 21:19:34 +0000701
Guido van Rossumd7047b31995-01-02 19:07:15 +0000702 if (f->f_fp == NULL)
703 return err_closed();
Raymond Hettingerea3fdf42002-12-29 16:33:45 +0000704 if (!PyArg_UnpackTuple(args, "truncate", 0, 1, &newsizeobj))
Guido van Rossum88303191999-01-04 17:22:18 +0000705 return NULL;
Tim Petersfb05db22002-03-11 00:24:00 +0000706
Tim Petersf1827cf2003-09-07 03:30:18 +0000707 /* Get current file position. If the file happens to be open for
708 * update and the last operation was an input operation, C doesn't
709 * define what the later fflush() will do, but we promise truncate()
710 * won't change the current position (and fflush() *does* change it
711 * then at least on Windows). The easiest thing is to capture
712 * current pos now and seek back to it at the end.
713 */
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000714 FILE_BEGIN_ALLOW_THREADS(f)
Tim Petersf1827cf2003-09-07 03:30:18 +0000715 errno = 0;
716 initialpos = _portable_ftell(f->f_fp);
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000717 FILE_END_ALLOW_THREADS(f)
Tim Petersf1827cf2003-09-07 03:30:18 +0000718 if (initialpos == -1)
719 goto onioerror;
720
Tim Petersfb05db22002-03-11 00:24:00 +0000721 /* Set newsize to current postion if newsizeobj NULL, else to the
Tim Petersf1827cf2003-09-07 03:30:18 +0000722 * specified value.
723 */
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000724 if (newsizeobj != NULL) {
725#if !defined(HAVE_LARGEFILE_SUPPORT)
726 newsize = PyInt_AsLong(newsizeobj);
727#else
728 newsize = PyLong_Check(newsizeobj) ?
729 PyLong_AsLongLong(newsizeobj) :
730 PyInt_AsLong(newsizeobj);
731#endif
732 if (PyErr_Occurred())
733 return NULL;
Tim Petersfb05db22002-03-11 00:24:00 +0000734 }
Tim Petersf1827cf2003-09-07 03:30:18 +0000735 else /* default to current position */
736 newsize = initialpos;
Tim Petersfb05db22002-03-11 00:24:00 +0000737
Tim Petersf1827cf2003-09-07 03:30:18 +0000738 /* Flush the stream. We're mixing stream-level I/O with lower-level
739 * I/O, and a flush may be necessary to synch both platform views
740 * of the current file state.
741 */
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000742 FILE_BEGIN_ALLOW_THREADS(f)
Guido van Rossumd7047b31995-01-02 19:07:15 +0000743 errno = 0;
744 ret = fflush(f->f_fp);
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000745 FILE_END_ALLOW_THREADS(f)
Tim Petersfb05db22002-03-11 00:24:00 +0000746 if (ret != 0)
747 goto onioerror;
Trent Mickf29f47b2000-08-11 19:02:59 +0000748
Martin v. Löwis6238d2b2002-06-30 15:26:10 +0000749#ifdef MS_WINDOWS
Tim Petersfb05db22002-03-11 00:24:00 +0000750 /* MS _chsize doesn't work if newsize doesn't fit in 32 bits,
Tim Peters8f01b682002-03-12 03:04:44 +0000751 so don't even try using it. */
Tim Petersfb05db22002-03-11 00:24:00 +0000752 {
Tim Petersfb05db22002-03-11 00:24:00 +0000753 HANDLE hFile;
Tim Petersfb05db22002-03-11 00:24:00 +0000754
Tim Petersf1827cf2003-09-07 03:30:18 +0000755 /* Have to move current pos to desired endpoint on Windows. */
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000756 FILE_BEGIN_ALLOW_THREADS(f)
Tim Petersf1827cf2003-09-07 03:30:18 +0000757 errno = 0;
758 ret = _portable_fseek(f->f_fp, newsize, SEEK_SET) != 0;
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000759 FILE_END_ALLOW_THREADS(f)
Tim Petersf1827cf2003-09-07 03:30:18 +0000760 if (ret)
761 goto onioerror;
Tim Petersfb05db22002-03-11 00:24:00 +0000762
Tim Peters8f01b682002-03-12 03:04:44 +0000763 /* Truncate. Note that this may grow the file! */
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000764 FILE_BEGIN_ALLOW_THREADS(f)
Tim Peters8f01b682002-03-12 03:04:44 +0000765 errno = 0;
766 hFile = (HANDLE)_get_osfhandle(fileno(f->f_fp));
Tim Petersf1827cf2003-09-07 03:30:18 +0000767 ret = hFile == (HANDLE)-1;
768 if (ret == 0) {
769 ret = SetEndOfFile(hFile) == 0;
770 if (ret)
Tim Peters8f01b682002-03-12 03:04:44 +0000771 errno = EACCES;
772 }
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000773 FILE_END_ALLOW_THREADS(f)
Tim Petersf1827cf2003-09-07 03:30:18 +0000774 if (ret)
Tim Peters8f01b682002-03-12 03:04:44 +0000775 goto onioerror;
Guido van Rossumd7047b31995-01-02 19:07:15 +0000776 }
Trent Mickf29f47b2000-08-11 19:02:59 +0000777#else
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000778 FILE_BEGIN_ALLOW_THREADS(f)
Trent Mickf29f47b2000-08-11 19:02:59 +0000779 errno = 0;
780 ret = ftruncate(fileno(f->f_fp), newsize);
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000781 FILE_END_ALLOW_THREADS(f)
Tim Petersf1827cf2003-09-07 03:30:18 +0000782 if (ret != 0)
783 goto onioerror;
Martin v. Löwis6238d2b2002-06-30 15:26:10 +0000784#endif /* !MS_WINDOWS */
Tim Peters86821b22001-01-07 21:19:34 +0000785
Tim Petersf1827cf2003-09-07 03:30:18 +0000786 /* Restore original file position. */
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000787 FILE_BEGIN_ALLOW_THREADS(f)
Tim Petersf1827cf2003-09-07 03:30:18 +0000788 errno = 0;
789 ret = _portable_fseek(f->f_fp, initialpos, SEEK_SET) != 0;
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000790 FILE_END_ALLOW_THREADS(f)
Tim Petersf1827cf2003-09-07 03:30:18 +0000791 if (ret)
792 goto onioerror;
793
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000794 Py_INCREF(Py_None);
795 return Py_None;
Trent Mickf29f47b2000-08-11 19:02:59 +0000796
797onioerror:
798 PyErr_SetFromErrno(PyExc_IOError);
799 clearerr(f->f_fp);
800 return NULL;
Guido van Rossumd7047b31995-01-02 19:07:15 +0000801}
802#endif /* HAVE_FTRUNCATE */
803
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000804static PyObject *
Martin v. Löwise3eb1f22001-08-16 13:15:00 +0000805file_tell(PyFileObject *f)
Guido van Rossumce5ba841991-03-06 13:06:18 +0000806{
Guido van Rossum4f53da02001-03-01 18:26:53 +0000807 Py_off_t pos;
Trent Mickf29f47b2000-08-11 19:02:59 +0000808
Guido van Rossumd7297e61992-07-06 14:19:26 +0000809 if (f->f_fp == NULL)
810 return err_closed();
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000811 FILE_BEGIN_ALLOW_THREADS(f)
Guido van Rossumce5ba841991-03-06 13:06:18 +0000812 errno = 0;
Trent Mickf29f47b2000-08-11 19:02:59 +0000813 pos = _portable_ftell(f->f_fp);
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000814 FILE_END_ALLOW_THREADS(f)
815
Trent Mickf29f47b2000-08-11 19:02:59 +0000816 if (pos == -1) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000817 PyErr_SetFromErrno(PyExc_IOError);
Guido van Rossumfebd5511992-03-04 16:39:24 +0000818 clearerr(f->f_fp);
819 return NULL;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000820 }
Jack Jansen7b8c7542002-04-14 20:12:41 +0000821 if (f->f_skipnextlf) {
822 int c;
823 c = GETC(f->f_fp);
824 if (c == '\n') {
Guido van Rossumad8fb0d2007-09-22 20:18:03 +0000825 f->f_newlinetypes |= NEWLINE_CRLF;
Jack Jansen7b8c7542002-04-14 20:12:41 +0000826 pos++;
827 f->f_skipnextlf = 0;
828 } else if (c != EOF) ungetc(c, f->f_fp);
829 }
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000830#if !defined(HAVE_LARGEFILE_SUPPORT)
Trent Mickf29f47b2000-08-11 19:02:59 +0000831 return PyInt_FromLong(pos);
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000832#else
Trent Mickf29f47b2000-08-11 19:02:59 +0000833 return PyLong_FromLongLong(pos);
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000834#endif
Guido van Rossumce5ba841991-03-06 13:06:18 +0000835}
836
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000837static PyObject *
Martin v. Löwise3eb1f22001-08-16 13:15:00 +0000838file_fileno(PyFileObject *f)
Guido van Rossumed233a51992-06-23 09:07:03 +0000839{
Guido van Rossumd7297e61992-07-06 14:19:26 +0000840 if (f->f_fp == NULL)
841 return err_closed();
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000842 return PyInt_FromLong((long) fileno(f->f_fp));
Guido van Rossumed233a51992-06-23 09:07:03 +0000843}
844
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000845static PyObject *
Martin v. Löwise3eb1f22001-08-16 13:15:00 +0000846file_flush(PyFileObject *f)
Guido van Rossumce5ba841991-03-06 13:06:18 +0000847{
Guido van Rossumff4949e1992-08-05 19:58:53 +0000848 int res;
Tim Peters86821b22001-01-07 21:19:34 +0000849
Guido van Rossumd7297e61992-07-06 14:19:26 +0000850 if (f->f_fp == NULL)
851 return err_closed();
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000852 FILE_BEGIN_ALLOW_THREADS(f)
Guido van Rossumce5ba841991-03-06 13:06:18 +0000853 errno = 0;
Guido van Rossumff4949e1992-08-05 19:58:53 +0000854 res = fflush(f->f_fp);
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000855 FILE_END_ALLOW_THREADS(f)
Guido van Rossumff4949e1992-08-05 19:58:53 +0000856 if (res != 0) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000857 PyErr_SetFromErrno(PyExc_IOError);
Guido van Rossumfebd5511992-03-04 16:39:24 +0000858 clearerr(f->f_fp);
859 return NULL;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000860 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000861 Py_INCREF(Py_None);
862 return Py_None;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000863}
864
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000865static PyObject *
Martin v. Löwise3eb1f22001-08-16 13:15:00 +0000866file_isatty(PyFileObject *f)
Guido van Rossuma1ab7fa1991-06-04 19:37:39 +0000867{
Guido van Rossumff4949e1992-08-05 19:58:53 +0000868 long res;
Guido van Rossumd7297e61992-07-06 14:19:26 +0000869 if (f->f_fp == NULL)
870 return err_closed();
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000871 FILE_BEGIN_ALLOW_THREADS(f)
Guido van Rossumff4949e1992-08-05 19:58:53 +0000872 res = isatty((int)fileno(f->f_fp));
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000873 FILE_END_ALLOW_THREADS(f)
Guido van Rossum7f7666f2002-04-07 06:28:00 +0000874 return PyBool_FromLong(res);
Guido van Rossuma1ab7fa1991-06-04 19:37:39 +0000875}
876
Guido van Rossumff7e83d1999-08-27 20:39:37 +0000877
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000878#if BUFSIZ < 8192
879#define SMALLCHUNK 8192
880#else
881#define SMALLCHUNK BUFSIZ
882#endif
883
Guido van Rossum3c259041999-01-14 19:00:14 +0000884#if SIZEOF_INT < 4
885#define BIGCHUNK (512 * 32)
886#else
887#define BIGCHUNK (512 * 1024)
888#endif
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000889
890static size_t
Fred Drakefd99de62000-07-09 05:02:18 +0000891new_buffersize(PyFileObject *f, size_t currentsize)
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000892{
893#ifdef HAVE_FSTAT
Fred Drake1bc8fab2001-07-19 21:49:38 +0000894 off_t pos, end;
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000895 struct stat st;
896 if (fstat(fileno(f->f_fp), &st) == 0) {
897 end = st.st_size;
Guido van Rossumcada2931998-12-11 20:44:56 +0000898 /* The following is not a bug: we really need to call lseek()
899 *and* ftell(). The reason is that some stdio libraries
900 mistakenly flush their buffer when ftell() is called and
901 the lseek() call it makes fails, thereby throwing away
902 data that cannot be recovered in any way. To avoid this,
903 we first test lseek(), and only call ftell() if lseek()
904 works. We can't use the lseek() value either, because we
905 need to take the amount of buffered data into account.
906 (Yet another reason why stdio stinks. :-) */
Guido van Rossum91aaa921998-05-05 22:21:35 +0000907 pos = lseek(fileno(f->f_fp), 0L, SEEK_CUR);
Jack Jansen2771b5b2001-10-10 22:03:27 +0000908 if (pos >= 0) {
Guido van Rossum91aaa921998-05-05 22:21:35 +0000909 pos = ftell(f->f_fp);
Jack Jansen2771b5b2001-10-10 22:03:27 +0000910 }
Guido van Rossumd30dc0a1998-04-27 19:01:08 +0000911 if (pos < 0)
912 clearerr(f->f_fp);
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000913 if (end > pos && pos >= 0)
Guido van Rossumcada2931998-12-11 20:44:56 +0000914 return currentsize + end - pos + 1;
Guido van Rossumdcb5e7f1998-03-03 22:36:10 +0000915 /* Add 1 so if the file were to grow we'd notice. */
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000916 }
917#endif
918 if (currentsize > SMALLCHUNK) {
919 /* Keep doubling until we reach BIGCHUNK;
920 then keep adding BIGCHUNK. */
921 if (currentsize <= BIGCHUNK)
922 return currentsize + currentsize;
923 else
924 return currentsize + BIGCHUNK;
925 }
926 return currentsize + SMALLCHUNK;
927}
928
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +0000929#if defined(EWOULDBLOCK) && defined(EAGAIN) && EWOULDBLOCK != EAGAIN
930#define BLOCKED_ERRNO(x) ((x) == EWOULDBLOCK || (x) == EAGAIN)
931#else
932#ifdef EWOULDBLOCK
933#define BLOCKED_ERRNO(x) ((x) == EWOULDBLOCK)
934#else
935#ifdef EAGAIN
936#define BLOCKED_ERRNO(x) ((x) == EAGAIN)
937#else
938#define BLOCKED_ERRNO(x) 0
939#endif
940#endif
941#endif
942
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000943static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000944file_read(PyFileObject *f, PyObject *args)
Guido van Rossumce5ba841991-03-06 13:06:18 +0000945{
Guido van Rossum789a1611997-05-10 22:33:55 +0000946 long bytesrequested = -1;
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000947 size_t bytesread, buffersize, chunksize;
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000948 PyObject *v;
Tim Peters86821b22001-01-07 21:19:34 +0000949
Guido van Rossumd7297e61992-07-06 14:19:26 +0000950 if (f->f_fp == NULL)
951 return err_closed();
Thomas Woutersc45251a2006-02-12 11:53:32 +0000952 /* refuse to mix with f.next() */
953 if (f->f_buf != NULL &&
954 (f->f_bufend - f->f_bufptr) > 0 &&
955 f->f_buf[0] != '\0')
956 return err_iterbuffered();
Guido van Rossum43713e52000-02-29 13:59:29 +0000957 if (!PyArg_ParseTuple(args, "|l:read", &bytesrequested))
Guido van Rossum789a1611997-05-10 22:33:55 +0000958 return NULL;
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000959 if (bytesrequested < 0)
Guido van Rossumff1ccbf1999-04-10 15:48:23 +0000960 buffersize = new_buffersize(f, (size_t)0);
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000961 else
962 buffersize = bytesrequested;
Martin v. Löwis2a190742006-04-13 07:37:25 +0000963 if (buffersize > PY_SSIZE_T_MAX) {
Trent Mickf29f47b2000-08-11 19:02:59 +0000964 PyErr_SetString(PyExc_OverflowError,
Jeremy Hylton8b735422002-08-14 21:01:41 +0000965 "requested number of bytes is more than a Python string can hold");
Trent Mickf29f47b2000-08-11 19:02:59 +0000966 return NULL;
967 }
Gregory P. Smithdd96db62008-06-09 04:58:54 +0000968 v = PyString_FromStringAndSize((char *)NULL, buffersize);
Guido van Rossum3f5da241990-12-20 15:06:42 +0000969 if (v == NULL)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000970 return NULL;
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000971 bytesread = 0;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000972 for (;;) {
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000973 FILE_BEGIN_ALLOW_THREADS(f)
Guido van Rossum6263d541997-05-10 22:07:25 +0000974 errno = 0;
Jack Jansen7b8c7542002-04-14 20:12:41 +0000975 chunksize = Py_UniversalNewlineFread(BUF(v) + bytesread,
Jeremy Hylton8b735422002-08-14 21:01:41 +0000976 buffersize - bytesread, f->f_fp, (PyObject *)f);
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +0000977 FILE_END_ALLOW_THREADS(f)
Guido van Rossum6263d541997-05-10 22:07:25 +0000978 if (chunksize == 0) {
979 if (!ferror(f->f_fp))
980 break;
Guido van Rossum6263d541997-05-10 22:07:25 +0000981 clearerr(f->f_fp);
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +0000982 /* When in non-blocking mode, data shouldn't
983 * be discarded if a blocking signal was
984 * received. That will also happen if
985 * chunksize != 0, but bytesread < buffersize. */
986 if (bytesread > 0 && BLOCKED_ERRNO(errno))
987 break;
988 PyErr_SetFromErrno(PyExc_IOError);
Guido van Rossum6263d541997-05-10 22:07:25 +0000989 Py_DECREF(v);
990 return NULL;
991 }
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000992 bytesread += chunksize;
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +0000993 if (bytesread < buffersize) {
994 clearerr(f->f_fp);
Guido van Rossumce5ba841991-03-06 13:06:18 +0000995 break;
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +0000996 }
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000997 if (bytesrequested < 0) {
Guido van Rossumcada2931998-12-11 20:44:56 +0000998 buffersize = new_buffersize(f, buffersize);
Gregory P. Smithdd96db62008-06-09 04:58:54 +0000999 if (_PyString_Resize(&v, buffersize) < 0)
Guido van Rossumce5ba841991-03-06 13:06:18 +00001000 return NULL;
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +00001001 } else {
Gustavo Niemeyera080be82002-12-17 17:48:00 +00001002 /* Got what was requested. */
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +00001003 break;
Guido van Rossumce5ba841991-03-06 13:06:18 +00001004 }
1005 }
Guido van Rossum5449b6e1997-05-09 22:27:31 +00001006 if (bytesread != buffersize)
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001007 _PyString_Resize(&v, bytesread);
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001008 return v;
1009}
1010
Guido van Rossumfdf95dd1997-05-05 22:15:02 +00001011static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001012file_readinto(PyFileObject *f, PyObject *args)
Guido van Rossumfdf95dd1997-05-05 22:15:02 +00001013{
1014 char *ptr;
Martin v. Löwis18e16552006-02-15 17:27:45 +00001015 Py_ssize_t ntodo;
1016 Py_ssize_t ndone, nnow;
Martin v. Löwisf91d46a2008-08-12 14:49:50 +00001017 Py_buffer pbuf;
Tim Peters86821b22001-01-07 21:19:34 +00001018
Guido van Rossumfdf95dd1997-05-05 22:15:02 +00001019 if (f->f_fp == NULL)
1020 return err_closed();
Thomas Woutersc45251a2006-02-12 11:53:32 +00001021 /* refuse to mix with f.next() */
1022 if (f->f_buf != NULL &&
1023 (f->f_bufend - f->f_bufptr) > 0 &&
1024 f->f_buf[0] != '\0')
1025 return err_iterbuffered();
Martin v. Löwisf91d46a2008-08-12 14:49:50 +00001026 if (!PyArg_ParseTuple(args, "w*", &pbuf))
Guido van Rossumfdf95dd1997-05-05 22:15:02 +00001027 return NULL;
Martin v. Löwisf91d46a2008-08-12 14:49:50 +00001028 ptr = pbuf.buf;
1029 ntodo = pbuf.len;
Guido van Rossumfdf95dd1997-05-05 22:15:02 +00001030 ndone = 0;
Guido van Rossum6263d541997-05-10 22:07:25 +00001031 while (ntodo > 0) {
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00001032 FILE_BEGIN_ALLOW_THREADS(f)
Guido van Rossum6263d541997-05-10 22:07:25 +00001033 errno = 0;
Tim Petersf1827cf2003-09-07 03:30:18 +00001034 nnow = Py_UniversalNewlineFread(ptr+ndone, ntodo, f->f_fp,
Jeremy Hylton8b735422002-08-14 21:01:41 +00001035 (PyObject *)f);
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00001036 FILE_END_ALLOW_THREADS(f)
Guido van Rossum6263d541997-05-10 22:07:25 +00001037 if (nnow == 0) {
1038 if (!ferror(f->f_fp))
1039 break;
Guido van Rossumfdf95dd1997-05-05 22:15:02 +00001040 PyErr_SetFromErrno(PyExc_IOError);
1041 clearerr(f->f_fp);
Martin v. Löwisf91d46a2008-08-12 14:49:50 +00001042 PyBuffer_Release(&pbuf);
Guido van Rossumfdf95dd1997-05-05 22:15:02 +00001043 return NULL;
1044 }
Guido van Rossumfdf95dd1997-05-05 22:15:02 +00001045 ndone += nnow;
1046 ntodo -= nnow;
1047 }
Martin v. Löwisf91d46a2008-08-12 14:49:50 +00001048 PyBuffer_Release(&pbuf);
Neal Norwitz076d1e02006-08-21 18:20:10 +00001049 return PyInt_FromSsize_t(ndone);
Guido van Rossumfdf95dd1997-05-05 22:15:02 +00001050}
1051
Tim Peters86821b22001-01-07 21:19:34 +00001052/**************************************************************************
Tim Petersf29b64d2001-01-15 06:33:19 +00001053Routine to get next line using platform fgets().
Tim Peters86821b22001-01-07 21:19:34 +00001054
1055Under MSVC 6:
1056
Tim Peters1c733232001-01-08 04:02:07 +00001057+ MS threadsafe getc is very slow (multiple layers of function calls before+
1058 after each character, to lock+unlock the stream).
1059+ The stream-locking functions are MS-internal -- can't access them from user
1060 code.
1061+ There's nothing Tim could find in the MS C or platform SDK libraries that
1062 can worm around this.
Tim Peters86821b22001-01-07 21:19:34 +00001063+ MS fgets locks/unlocks only once per line; it's the only hook we have.
1064
1065So we use fgets for speed(!), despite that it's painful.
1066
1067MS realloc is also slow.
1068
Tim Petersf29b64d2001-01-15 06:33:19 +00001069Reports from other platforms on this method vs getc_unlocked (which MS doesn't
1070have):
1071 Linux a wash
1072 Solaris a wash
1073 Tru64 Unix getline_via_fgets significantly faster
Tim Peters86821b22001-01-07 21:19:34 +00001074
Tim Petersf29b64d2001-01-15 06:33:19 +00001075CAUTION: The C std isn't clear about this: in those cases where fgets
1076writes something into the buffer, can it write into any position beyond the
1077required trailing null byte? MSVC 6 fgets does not, and no platform is (yet)
1078known on which it does; and it would be a strange way to code fgets. Still,
1079getline_via_fgets may not work correctly if it does. The std test
1080test_bufio.py should fail if platform fgets() routinely writes beyond the
1081trailing null byte. #define DONT_USE_FGETS_IN_GETLINE to disable this code.
Tim Peters86821b22001-01-07 21:19:34 +00001082**************************************************************************/
1083
Tim Petersf29b64d2001-01-15 06:33:19 +00001084/* Use this routine if told to, or by default on non-get_unlocked()
1085 * platforms unless told not to. Yikes! Let's spell that out:
1086 * On a platform with getc_unlocked():
1087 * By default, use getc_unlocked().
1088 * If you want to use fgets() instead, #define USE_FGETS_IN_GETLINE.
1089 * On a platform without getc_unlocked():
1090 * By default, use fgets().
1091 * If you don't want to use fgets(), #define DONT_USE_FGETS_IN_GETLINE.
1092 */
1093#if !defined(USE_FGETS_IN_GETLINE) && !defined(HAVE_GETC_UNLOCKED)
1094#define USE_FGETS_IN_GETLINE
Tim Peters86821b22001-01-07 21:19:34 +00001095#endif
1096
Tim Petersf29b64d2001-01-15 06:33:19 +00001097#if defined(DONT_USE_FGETS_IN_GETLINE) && defined(USE_FGETS_IN_GETLINE)
1098#undef USE_FGETS_IN_GETLINE
1099#endif
1100
1101#ifdef USE_FGETS_IN_GETLINE
Tim Peters86821b22001-01-07 21:19:34 +00001102static PyObject*
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00001103getline_via_fgets(PyFileObject *f, FILE *fp)
Tim Peters86821b22001-01-07 21:19:34 +00001104{
Tim Peters15b83852001-01-08 00:53:12 +00001105/* INITBUFSIZE is the maximum line length that lets us get away with the fast
Tim Peters142297a2001-01-15 10:36:56 +00001106 * no-realloc, one-fgets()-call path. Boosting it isn't free, because we have
1107 * to fill this much of the buffer with a known value in order to figure out
1108 * how much of the buffer fgets() overwrites. So if INITBUFSIZE is larger
1109 * than "most" lines, we waste time filling unused buffer slots. 100 is
1110 * surely adequate for most peoples' email archives, chewing over source code,
1111 * etc -- "regular old text files".
1112 * MAXBUFSIZE is the maximum line length that lets us get away with the less
1113 * fast (but still zippy) no-realloc, two-fgets()-call path. See above for
1114 * cautions about boosting that. 300 was chosen because the worst real-life
1115 * text-crunching job reported on Python-Dev was a mail-log crawler where over
1116 * half the lines were 254 chars.
Tim Peters15b83852001-01-08 00:53:12 +00001117 */
Tim Peters142297a2001-01-15 10:36:56 +00001118#define INITBUFSIZE 100
1119#define MAXBUFSIZE 300
Tim Peters142297a2001-01-15 10:36:56 +00001120 char* p; /* temp */
1121 char buf[MAXBUFSIZE];
Tim Peters86821b22001-01-07 21:19:34 +00001122 PyObject* v; /* the string object result */
Tim Peters86821b22001-01-07 21:19:34 +00001123 char* pvfree; /* address of next free slot */
1124 char* pvend; /* address one beyond last free slot */
Tim Peters142297a2001-01-15 10:36:56 +00001125 size_t nfree; /* # of free buffer slots; pvend-pvfree */
1126 size_t total_v_size; /* total # of slots in buffer */
Tim Petersddea2082002-03-23 10:03:50 +00001127 size_t increment; /* amount to increment the buffer */
Armin Rigo7ccbca92006-10-04 12:17:45 +00001128 size_t prev_v_size;
Tim Peters86821b22001-01-07 21:19:34 +00001129
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001130 /* Optimize for normal case: avoid _PyString_Resize if at all
Tim Peters142297a2001-01-15 10:36:56 +00001131 * possible via first reading into stack buffer "buf".
Tim Peters15b83852001-01-08 00:53:12 +00001132 */
Tim Peters142297a2001-01-15 10:36:56 +00001133 total_v_size = INITBUFSIZE; /* start small and pray */
1134 pvfree = buf;
1135 for (;;) {
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00001136 FILE_BEGIN_ALLOW_THREADS(f)
Tim Peters142297a2001-01-15 10:36:56 +00001137 pvend = buf + total_v_size;
1138 nfree = pvend - pvfree;
1139 memset(pvfree, '\n', nfree);
Martin v. Löwis18e16552006-02-15 17:27:45 +00001140 assert(nfree < INT_MAX); /* Should be atmost MAXBUFSIZE */
1141 p = fgets(pvfree, (int)nfree, fp);
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00001142 FILE_END_ALLOW_THREADS(f)
Tim Peters15b83852001-01-08 00:53:12 +00001143
Tim Peters142297a2001-01-15 10:36:56 +00001144 if (p == NULL) {
1145 clearerr(fp);
1146 if (PyErr_CheckSignals())
1147 return NULL;
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001148 v = PyString_FromStringAndSize(buf, pvfree - buf);
Tim Peters86821b22001-01-07 21:19:34 +00001149 return v;
1150 }
Tim Peters142297a2001-01-15 10:36:56 +00001151 /* fgets read *something* */
1152 p = memchr(pvfree, '\n', nfree);
1153 if (p != NULL) {
1154 /* Did the \n come from fgets or from us?
1155 * Since fgets stops at the first \n, and then writes
1156 * \0, if it's from fgets a \0 must be next. But if
1157 * that's so, it could not have come from us, since
1158 * the \n's we filled the buffer with have only more
1159 * \n's to the right.
1160 */
1161 if (p+1 < pvend && *(p+1) == '\0') {
1162 /* It's from fgets: we win! In particular,
1163 * we haven't done any mallocs yet, and can
1164 * build the final result on the first try.
1165 */
1166 ++p; /* include \n from fgets */
1167 }
1168 else {
1169 /* Must be from us: fgets didn't fill the
1170 * buffer and didn't find a newline, so it
1171 * must be the last and newline-free line of
1172 * the file.
1173 */
1174 assert(p > pvfree && *(p-1) == '\0');
1175 --p; /* don't include \0 from fgets */
1176 }
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001177 v = PyString_FromStringAndSize(buf, p - buf);
Tim Peters142297a2001-01-15 10:36:56 +00001178 return v;
1179 }
1180 /* yuck: fgets overwrote all the newlines, i.e. the entire
1181 * buffer. So this line isn't over yet, or maybe it is but
1182 * we're exactly at EOF. If we haven't already, try using the
1183 * rest of the stack buffer.
Tim Peters86821b22001-01-07 21:19:34 +00001184 */
Tim Peters142297a2001-01-15 10:36:56 +00001185 assert(*(pvend-1) == '\0');
1186 if (pvfree == buf) {
1187 pvfree = pvend - 1; /* overwrite trailing null */
1188 total_v_size = MAXBUFSIZE;
1189 }
1190 else
1191 break;
Tim Peters86821b22001-01-07 21:19:34 +00001192 }
Tim Peters142297a2001-01-15 10:36:56 +00001193
1194 /* The stack buffer isn't big enough; malloc a string object and read
1195 * into its buffer.
Tim Peters15b83852001-01-08 00:53:12 +00001196 */
Tim Petersddea2082002-03-23 10:03:50 +00001197 total_v_size = MAXBUFSIZE << 1;
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001198 v = PyString_FromStringAndSize((char*)NULL, (int)total_v_size);
Tim Peters15b83852001-01-08 00:53:12 +00001199 if (v == NULL)
1200 return v;
1201 /* copy over everything except the last null byte */
Tim Peters142297a2001-01-15 10:36:56 +00001202 memcpy(BUF(v), buf, MAXBUFSIZE-1);
1203 pvfree = BUF(v) + MAXBUFSIZE - 1;
Tim Peters86821b22001-01-07 21:19:34 +00001204
1205 /* Keep reading stuff into v; if it ever ends successfully, break
Tim Peters15b83852001-01-08 00:53:12 +00001206 * after setting p one beyond the end of the line. The code here is
1207 * very much like the code above, except reads into v's buffer; see
1208 * the code above for detailed comments about the logic.
Tim Peters86821b22001-01-07 21:19:34 +00001209 */
1210 for (;;) {
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00001211 FILE_BEGIN_ALLOW_THREADS(f)
Tim Peters86821b22001-01-07 21:19:34 +00001212 pvend = BUF(v) + total_v_size;
1213 nfree = pvend - pvfree;
1214 memset(pvfree, '\n', nfree);
Martin v. Löwis18e16552006-02-15 17:27:45 +00001215 assert(nfree < INT_MAX);
1216 p = fgets(pvfree, (int)nfree, fp);
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00001217 FILE_END_ALLOW_THREADS(f)
Tim Peters86821b22001-01-07 21:19:34 +00001218
1219 if (p == NULL) {
1220 clearerr(fp);
1221 if (PyErr_CheckSignals()) {
1222 Py_DECREF(v);
1223 return NULL;
1224 }
1225 p = pvfree;
1226 break;
1227 }
Tim Peters86821b22001-01-07 21:19:34 +00001228 p = memchr(pvfree, '\n', nfree);
1229 if (p != NULL) {
1230 if (p+1 < pvend && *(p+1) == '\0') {
1231 /* \n came from fgets */
1232 ++p;
1233 break;
1234 }
1235 /* \n came from us; last line of file, no newline */
1236 assert(p > pvfree && *(p-1) == '\0');
1237 --p;
1238 break;
1239 }
1240 /* expand buffer and try again */
1241 assert(*(pvend-1) == '\0');
Tim Petersddea2082002-03-23 10:03:50 +00001242 increment = total_v_size >> 2; /* mild exponential growth */
Armin Rigo7ccbca92006-10-04 12:17:45 +00001243 prev_v_size = total_v_size;
Tim Petersddea2082002-03-23 10:03:50 +00001244 total_v_size += increment;
Armin Rigo7ccbca92006-10-04 12:17:45 +00001245 /* check for overflow */
1246 if (total_v_size <= prev_v_size ||
1247 total_v_size > PY_SSIZE_T_MAX) {
Tim Peters86821b22001-01-07 21:19:34 +00001248 PyErr_SetString(PyExc_OverflowError,
1249 "line is longer than a Python string can hold");
1250 Py_DECREF(v);
1251 return NULL;
1252 }
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001253 if (_PyString_Resize(&v, (int)total_v_size) < 0)
Tim Peters86821b22001-01-07 21:19:34 +00001254 return NULL;
1255 /* overwrite the trailing null byte */
Armin Rigo7ccbca92006-10-04 12:17:45 +00001256 pvfree = BUF(v) + (prev_v_size - 1);
Tim Peters86821b22001-01-07 21:19:34 +00001257 }
1258 if (BUF(v) + total_v_size != p)
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001259 _PyString_Resize(&v, p - BUF(v));
Tim Peters86821b22001-01-07 21:19:34 +00001260 return v;
1261#undef INITBUFSIZE
Tim Peters142297a2001-01-15 10:36:56 +00001262#undef MAXBUFSIZE
Tim Peters86821b22001-01-07 21:19:34 +00001263}
Tim Petersf29b64d2001-01-15 06:33:19 +00001264#endif /* ifdef USE_FGETS_IN_GETLINE */
Guido van Rossumfdf95dd1997-05-05 22:15:02 +00001265
Guido van Rossum0bd24411991-04-04 15:21:57 +00001266/* Internal routine to get a line.
1267 Size argument interpretation:
1268 > 0: max length;
Guido van Rossum86282062001-01-08 01:26:47 +00001269 <= 0: read arbitrary line
Guido van Rossumce5ba841991-03-06 13:06:18 +00001270*/
1271
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001272static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001273get_line(PyFileObject *f, int n)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001274{
Guido van Rossum1187aa42001-01-05 14:43:05 +00001275 FILE *fp = f->f_fp;
1276 int c;
Andrew M. Kuchling4b2b4452000-11-29 02:53:22 +00001277 char *buf, *end;
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001278 size_t total_v_size; /* total # of slots in buffer */
1279 size_t used_v_size; /* # used slots in buffer */
1280 size_t increment; /* amount to increment the buffer */
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001281 PyObject *v;
Jack Jansen7b8c7542002-04-14 20:12:41 +00001282 int newlinetypes = f->f_newlinetypes;
1283 int skipnextlf = f->f_skipnextlf;
1284 int univ_newline = f->f_univ_newline;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001285
Jack Jansen7b8c7542002-04-14 20:12:41 +00001286#if defined(USE_FGETS_IN_GETLINE)
Jack Jansen7b8c7542002-04-14 20:12:41 +00001287 if (n <= 0 && !univ_newline )
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00001288 return getline_via_fgets(f, fp);
Tim Peters86821b22001-01-07 21:19:34 +00001289#endif
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001290 total_v_size = n > 0 ? n : 100;
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001291 v = PyString_FromStringAndSize((char *)NULL, total_v_size);
Guido van Rossum3f5da241990-12-20 15:06:42 +00001292 if (v == NULL)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001293 return NULL;
Guido van Rossumce5ba841991-03-06 13:06:18 +00001294 buf = BUF(v);
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001295 end = buf + total_v_size;
Guido van Rossum1984f1e1992-08-04 12:41:02 +00001296
Guido van Rossumce5ba841991-03-06 13:06:18 +00001297 for (;;) {
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00001298 FILE_BEGIN_ALLOW_THREADS(f)
Guido van Rossum1187aa42001-01-05 14:43:05 +00001299 FLOCKFILE(fp);
Jack Jansen7b8c7542002-04-14 20:12:41 +00001300 if (univ_newline) {
1301 c = 'x'; /* Shut up gcc warning */
1302 while ( buf != end && (c = GETC(fp)) != EOF ) {
1303 if (skipnextlf ) {
1304 skipnextlf = 0;
1305 if (c == '\n') {
Tim Petersf1827cf2003-09-07 03:30:18 +00001306 /* Seeing a \n here with
1307 * skipnextlf true means we
Jeremy Hylton8b735422002-08-14 21:01:41 +00001308 * saw a \r before.
1309 */
Jack Jansen7b8c7542002-04-14 20:12:41 +00001310 newlinetypes |= NEWLINE_CRLF;
1311 c = GETC(fp);
1312 if (c == EOF) break;
1313 } else {
1314 newlinetypes |= NEWLINE_CR;
1315 }
1316 }
1317 if (c == '\r') {
1318 skipnextlf = 1;
1319 c = '\n';
1320 } else if ( c == '\n')
1321 newlinetypes |= NEWLINE_LF;
1322 *buf++ = c;
1323 if (c == '\n') break;
1324 }
1325 if ( c == EOF && skipnextlf )
1326 newlinetypes |= NEWLINE_CR;
1327 } else /* If not universal newlines use the normal loop */
Guido van Rossum1187aa42001-01-05 14:43:05 +00001328 while ((c = GETC(fp)) != EOF &&
1329 (*buf++ = c) != '\n' &&
1330 buf != end)
1331 ;
1332 FUNLOCKFILE(fp);
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00001333 FILE_END_ALLOW_THREADS(f)
Jack Jansen7b8c7542002-04-14 20:12:41 +00001334 f->f_newlinetypes = newlinetypes;
1335 f->f_skipnextlf = skipnextlf;
Guido van Rossum1187aa42001-01-05 14:43:05 +00001336 if (c == '\n')
1337 break;
1338 if (c == EOF) {
Guido van Rossum29206bc2001-08-09 18:14:59 +00001339 if (ferror(fp)) {
1340 PyErr_SetFromErrno(PyExc_IOError);
1341 clearerr(fp);
1342 Py_DECREF(v);
1343 return NULL;
1344 }
Guido van Rossum76ad8ed1991-06-03 10:54:55 +00001345 clearerr(fp);
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001346 if (PyErr_CheckSignals()) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001347 Py_DECREF(v);
Guido van Rossum0bd24411991-04-04 15:21:57 +00001348 return NULL;
1349 }
Guido van Rossumce5ba841991-03-06 13:06:18 +00001350 break;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001351 }
Guido van Rossum1187aa42001-01-05 14:43:05 +00001352 /* Must be because buf == end */
1353 if (n > 0)
Guido van Rossum0bd24411991-04-04 15:21:57 +00001354 break;
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001355 used_v_size = total_v_size;
1356 increment = total_v_size >> 2; /* mild exponential growth */
1357 total_v_size += increment;
Martin v. Löwis2a190742006-04-13 07:37:25 +00001358 if (total_v_size > PY_SSIZE_T_MAX) {
Guido van Rossum1187aa42001-01-05 14:43:05 +00001359 PyErr_SetString(PyExc_OverflowError,
1360 "line is longer than a Python string can hold");
Tim Peters86821b22001-01-07 21:19:34 +00001361 Py_DECREF(v);
Guido van Rossum1187aa42001-01-05 14:43:05 +00001362 return NULL;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001363 }
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001364 if (_PyString_Resize(&v, total_v_size) < 0)
Guido van Rossum1187aa42001-01-05 14:43:05 +00001365 return NULL;
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001366 buf = BUF(v) + used_v_size;
1367 end = BUF(v) + total_v_size;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001368 }
Guido van Rossum1984f1e1992-08-04 12:41:02 +00001369
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001370 used_v_size = buf - BUF(v);
1371 if (used_v_size != total_v_size)
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001372 _PyString_Resize(&v, used_v_size);
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001373 return v;
1374}
1375
Guido van Rossum0bd24411991-04-04 15:21:57 +00001376/* External C interface */
1377
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001378PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001379PyFile_GetLine(PyObject *f, int n)
Guido van Rossum0bd24411991-04-04 15:21:57 +00001380{
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001381 PyObject *result;
1382
Guido van Rossum3165fe61992-09-25 21:59:05 +00001383 if (f == NULL) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001384 PyErr_BadInternalCall();
Guido van Rossum0bd24411991-04-04 15:21:57 +00001385 return NULL;
1386 }
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001387
1388 if (PyFile_Check(f)) {
Thomas Woutersc45251a2006-02-12 11:53:32 +00001389 PyFileObject *fo = (PyFileObject *)f;
1390 if (fo->f_fp == NULL)
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001391 return err_closed();
Thomas Woutersc45251a2006-02-12 11:53:32 +00001392 /* refuse to mix with f.next() */
1393 if (fo->f_buf != NULL &&
1394 (fo->f_bufend - fo->f_bufptr) > 0 &&
1395 fo->f_buf[0] != '\0')
1396 return err_iterbuffered();
1397 result = get_line(fo, n);
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001398 }
1399 else {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001400 PyObject *reader;
1401 PyObject *args;
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001402
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001403 reader = PyObject_GetAttrString(f, "readline");
Guido van Rossum3165fe61992-09-25 21:59:05 +00001404 if (reader == NULL)
1405 return NULL;
1406 if (n <= 0)
Raymond Hettinger8ae46892003-10-12 19:09:37 +00001407 args = PyTuple_New(0);
Guido van Rossum3165fe61992-09-25 21:59:05 +00001408 else
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001409 args = Py_BuildValue("(i)", n);
Guido van Rossum3165fe61992-09-25 21:59:05 +00001410 if (args == NULL) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001411 Py_DECREF(reader);
Guido van Rossum3165fe61992-09-25 21:59:05 +00001412 return NULL;
1413 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001414 result = PyEval_CallObject(reader, args);
1415 Py_DECREF(reader);
1416 Py_DECREF(args);
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001417 if (result != NULL && !PyString_Check(result) &&
Martin v. Löwisaf6a27a2003-01-03 19:16:14 +00001418 !PyUnicode_Check(result)) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001419 Py_DECREF(result);
Guido van Rossum3165fe61992-09-25 21:59:05 +00001420 result = NULL;
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001421 PyErr_SetString(PyExc_TypeError,
Guido van Rossum3165fe61992-09-25 21:59:05 +00001422 "object.readline() returned non-string");
1423 }
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001424 }
1425
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001426 if (n < 0 && result != NULL && PyString_Check(result)) {
1427 char *s = PyString_AS_STRING(result);
1428 Py_ssize_t len = PyString_GET_SIZE(result);
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001429 if (len == 0) {
1430 Py_DECREF(result);
1431 result = NULL;
1432 PyErr_SetString(PyExc_EOFError,
1433 "EOF when reading a line");
1434 }
1435 else if (s[len-1] == '\n') {
1436 if (result->ob_refcnt == 1)
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001437 _PyString_Resize(&result, len-1);
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001438 else {
1439 PyObject *v;
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001440 v = PyString_FromStringAndSize(s, len-1);
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001441 Py_DECREF(result);
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001442 result = v;
Guido van Rossum3165fe61992-09-25 21:59:05 +00001443 }
1444 }
Guido van Rossum3165fe61992-09-25 21:59:05 +00001445 }
Martin v. Löwisaf6a27a2003-01-03 19:16:14 +00001446#ifdef Py_USING_UNICODE
1447 if (n < 0 && result != NULL && PyUnicode_Check(result)) {
1448 Py_UNICODE *s = PyUnicode_AS_UNICODE(result);
Martin v. Löwis18e16552006-02-15 17:27:45 +00001449 Py_ssize_t len = PyUnicode_GET_SIZE(result);
Martin v. Löwisaf6a27a2003-01-03 19:16:14 +00001450 if (len == 0) {
1451 Py_DECREF(result);
1452 result = NULL;
1453 PyErr_SetString(PyExc_EOFError,
1454 "EOF when reading a line");
1455 }
1456 else if (s[len-1] == '\n') {
1457 if (result->ob_refcnt == 1)
1458 PyUnicode_Resize(&result, len-1);
1459 else {
1460 PyObject *v;
1461 v = PyUnicode_FromUnicode(s, len-1);
1462 Py_DECREF(result);
1463 result = v;
1464 }
1465 }
1466 }
1467#endif
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001468 return result;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001469}
1470
1471/* Python method */
1472
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001473static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001474file_readline(PyFileObject *f, PyObject *args)
Guido van Rossum0bd24411991-04-04 15:21:57 +00001475{
Guido van Rossum789a1611997-05-10 22:33:55 +00001476 int n = -1;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001477
Guido van Rossumd7297e61992-07-06 14:19:26 +00001478 if (f->f_fp == NULL)
1479 return err_closed();
Thomas Woutersc45251a2006-02-12 11:53:32 +00001480 /* refuse to mix with f.next() */
1481 if (f->f_buf != NULL &&
1482 (f->f_bufend - f->f_bufptr) > 0 &&
1483 f->f_buf[0] != '\0')
1484 return err_iterbuffered();
Guido van Rossum43713e52000-02-29 13:59:29 +00001485 if (!PyArg_ParseTuple(args, "|i:readline", &n))
Guido van Rossum789a1611997-05-10 22:33:55 +00001486 return NULL;
1487 if (n == 0)
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001488 return PyString_FromString("");
Guido van Rossum789a1611997-05-10 22:33:55 +00001489 if (n < 0)
1490 n = 0;
Marc-André Lemburg1f468602000-07-05 15:32:40 +00001491 return get_line(f, n);
Guido van Rossum0bd24411991-04-04 15:21:57 +00001492}
1493
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001494static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001495file_readlines(PyFileObject *f, PyObject *args)
Guido van Rossumce5ba841991-03-06 13:06:18 +00001496{
Guido van Rossum789a1611997-05-10 22:33:55 +00001497 long sizehint = 0;
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00001498 PyObject *list = NULL;
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001499 PyObject *line;
Guido van Rossum6263d541997-05-10 22:07:25 +00001500 char small_buffer[SMALLCHUNK];
1501 char *buffer = small_buffer;
1502 size_t buffersize = SMALLCHUNK;
1503 PyObject *big_buffer = NULL;
1504 size_t nfilled = 0;
1505 size_t nread;
Guido van Rossum789a1611997-05-10 22:33:55 +00001506 size_t totalread = 0;
Guido van Rossum6263d541997-05-10 22:07:25 +00001507 char *p, *q, *end;
1508 int err;
Guido van Rossum79fd0fc2001-10-12 20:01:53 +00001509 int shortread = 0;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001510
Guido van Rossumd7297e61992-07-06 14:19:26 +00001511 if (f->f_fp == NULL)
1512 return err_closed();
Thomas Woutersc45251a2006-02-12 11:53:32 +00001513 /* refuse to mix with f.next() */
1514 if (f->f_buf != NULL &&
1515 (f->f_bufend - f->f_bufptr) > 0 &&
1516 f->f_buf[0] != '\0')
1517 return err_iterbuffered();
Guido van Rossum43713e52000-02-29 13:59:29 +00001518 if (!PyArg_ParseTuple(args, "|l:readlines", &sizehint))
Guido van Rossum0bd24411991-04-04 15:21:57 +00001519 return NULL;
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001520 if ((list = PyList_New(0)) == NULL)
Guido van Rossumce5ba841991-03-06 13:06:18 +00001521 return NULL;
1522 for (;;) {
Guido van Rossum79fd0fc2001-10-12 20:01:53 +00001523 if (shortread)
1524 nread = 0;
1525 else {
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00001526 FILE_BEGIN_ALLOW_THREADS(f)
Guido van Rossum79fd0fc2001-10-12 20:01:53 +00001527 errno = 0;
Tim Peters058b1412002-04-21 07:29:14 +00001528 nread = Py_UniversalNewlineFread(buffer+nfilled,
Jack Jansen7b8c7542002-04-14 20:12:41 +00001529 buffersize-nfilled, f->f_fp, (PyObject *)f);
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00001530 FILE_END_ALLOW_THREADS(f)
Guido van Rossum79fd0fc2001-10-12 20:01:53 +00001531 shortread = (nread < buffersize-nfilled);
1532 }
Guido van Rossum6263d541997-05-10 22:07:25 +00001533 if (nread == 0) {
Guido van Rossum789a1611997-05-10 22:33:55 +00001534 sizehint = 0;
Guido van Rossum3da3fce1998-02-19 20:46:48 +00001535 if (!ferror(f->f_fp))
Guido van Rossum6263d541997-05-10 22:07:25 +00001536 break;
1537 PyErr_SetFromErrno(PyExc_IOError);
1538 clearerr(f->f_fp);
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00001539 goto error;
Guido van Rossumce5ba841991-03-06 13:06:18 +00001540 }
Guido van Rossum789a1611997-05-10 22:33:55 +00001541 totalread += nread;
Anthony Baxter377be112006-04-11 06:54:30 +00001542 p = (char *)memchr(buffer+nfilled, '\n', nread);
Guido van Rossum6263d541997-05-10 22:07:25 +00001543 if (p == NULL) {
1544 /* Need a larger buffer to fit this line */
1545 nfilled += nread;
1546 buffersize *= 2;
Martin v. Löwis2a190742006-04-13 07:37:25 +00001547 if (buffersize > PY_SSIZE_T_MAX) {
Trent Mickf29f47b2000-08-11 19:02:59 +00001548 PyErr_SetString(PyExc_OverflowError,
Guido van Rossume07d5cf2001-01-09 21:50:24 +00001549 "line is longer than a Python string can hold");
Trent Mickf29f47b2000-08-11 19:02:59 +00001550 goto error;
1551 }
Guido van Rossum6263d541997-05-10 22:07:25 +00001552 if (big_buffer == NULL) {
1553 /* Create the big buffer */
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001554 big_buffer = PyString_FromStringAndSize(
Guido van Rossum6263d541997-05-10 22:07:25 +00001555 NULL, buffersize);
1556 if (big_buffer == NULL)
1557 goto error;
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001558 buffer = PyString_AS_STRING(big_buffer);
Guido van Rossum6263d541997-05-10 22:07:25 +00001559 memcpy(buffer, small_buffer, nfilled);
1560 }
1561 else {
1562 /* Grow the big buffer */
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001563 if ( _PyString_Resize(&big_buffer, buffersize) < 0 )
Jack Jansen7b8c7542002-04-14 20:12:41 +00001564 goto error;
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001565 buffer = PyString_AS_STRING(big_buffer);
Guido van Rossum6263d541997-05-10 22:07:25 +00001566 }
1567 continue;
1568 }
1569 end = buffer+nfilled+nread;
1570 q = buffer;
1571 do {
1572 /* Process complete lines */
1573 p++;
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001574 line = PyString_FromStringAndSize(q, p-q);
Guido van Rossum6263d541997-05-10 22:07:25 +00001575 if (line == NULL)
1576 goto error;
1577 err = PyList_Append(list, line);
1578 Py_DECREF(line);
1579 if (err != 0)
1580 goto error;
1581 q = p;
Anthony Baxter377be112006-04-11 06:54:30 +00001582 p = (char *)memchr(q, '\n', end-q);
Guido van Rossum6263d541997-05-10 22:07:25 +00001583 } while (p != NULL);
1584 /* Move the remaining incomplete line to the start */
1585 nfilled = end-q;
1586 memmove(buffer, q, nfilled);
Guido van Rossum789a1611997-05-10 22:33:55 +00001587 if (sizehint > 0)
1588 if (totalread >= (size_t)sizehint)
1589 break;
Guido van Rossumce5ba841991-03-06 13:06:18 +00001590 }
Guido van Rossum6263d541997-05-10 22:07:25 +00001591 if (nfilled != 0) {
1592 /* Partial last line */
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001593 line = PyString_FromStringAndSize(buffer, nfilled);
Guido van Rossum6263d541997-05-10 22:07:25 +00001594 if (line == NULL)
1595 goto error;
Guido van Rossum789a1611997-05-10 22:33:55 +00001596 if (sizehint > 0) {
1597 /* Need to complete the last line */
Marc-André Lemburg1f468602000-07-05 15:32:40 +00001598 PyObject *rest = get_line(f, 0);
Guido van Rossum789a1611997-05-10 22:33:55 +00001599 if (rest == NULL) {
1600 Py_DECREF(line);
1601 goto error;
1602 }
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001603 PyString_Concat(&line, rest);
Guido van Rossum789a1611997-05-10 22:33:55 +00001604 Py_DECREF(rest);
1605 if (line == NULL)
1606 goto error;
1607 }
Guido van Rossum6263d541997-05-10 22:07:25 +00001608 err = PyList_Append(list, line);
1609 Py_DECREF(line);
1610 if (err != 0)
1611 goto error;
1612 }
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00001613
1614cleanup:
Tim Peters5de98422002-04-27 18:44:32 +00001615 Py_XDECREF(big_buffer);
Guido van Rossumce5ba841991-03-06 13:06:18 +00001616 return list;
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00001617
1618error:
1619 Py_CLEAR(list);
1620 goto cleanup;
Guido van Rossumce5ba841991-03-06 13:06:18 +00001621}
1622
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001623static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001624file_write(PyFileObject *f, PyObject *args)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001625{
Martin v. Löwisf91d46a2008-08-12 14:49:50 +00001626 Py_buffer pbuf;
Guido van Rossumd7297e61992-07-06 14:19:26 +00001627 char *s;
Martin v. Löwis18e16552006-02-15 17:27:45 +00001628 Py_ssize_t n, n2;
Guido van Rossumd7297e61992-07-06 14:19:26 +00001629 if (f->f_fp == NULL)
1630 return err_closed();
Martin v. Löwisf91d46a2008-08-12 14:49:50 +00001631 if (f->f_binary) {
1632 if (!PyArg_ParseTuple(args, "s*", &pbuf))
1633 return NULL;
1634 s = pbuf.buf;
1635 n = pbuf.len;
1636 } else
1637 if (!PyArg_ParseTuple(args, "t#", &s, &n))
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001638 return NULL;
Guido van Rossumeb183da1991-04-04 10:44:06 +00001639 f->f_softspace = 0;
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00001640 FILE_BEGIN_ALLOW_THREADS(f)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001641 errno = 0;
Guido van Rossumd7297e61992-07-06 14:19:26 +00001642 n2 = fwrite(s, 1, n, f->f_fp);
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00001643 FILE_END_ALLOW_THREADS(f)
Martin v. Löwisf91d46a2008-08-12 14:49:50 +00001644 if (f->f_binary)
1645 PyBuffer_Release(&pbuf);
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001646 if (n2 != n) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001647 PyErr_SetFromErrno(PyExc_IOError);
Guido van Rossumfebd5511992-03-04 16:39:24 +00001648 clearerr(f->f_fp);
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001649 return NULL;
1650 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001651 Py_INCREF(Py_None);
1652 return Py_None;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001653}
1654
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001655static PyObject *
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001656file_writelines(PyFileObject *f, PyObject *seq)
Guido van Rossum5a2a6831993-10-25 09:59:04 +00001657{
Guido van Rossumee70ad12000-03-13 16:27:06 +00001658#define CHUNKSIZE 1000
1659 PyObject *list, *line;
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001660 PyObject *it; /* iter(seq) */
Guido van Rossumee70ad12000-03-13 16:27:06 +00001661 PyObject *result;
Martin v. Löwis18e16552006-02-15 17:27:45 +00001662 int index, islist;
1663 Py_ssize_t i, j, nwritten, len;
Guido van Rossumee70ad12000-03-13 16:27:06 +00001664
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001665 assert(seq != NULL);
Guido van Rossum5a2a6831993-10-25 09:59:04 +00001666 if (f->f_fp == NULL)
1667 return err_closed();
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001668
1669 result = NULL;
1670 list = NULL;
1671 islist = PyList_Check(seq);
1672 if (islist)
1673 it = NULL;
1674 else {
1675 it = PyObject_GetIter(seq);
1676 if (it == NULL) {
1677 PyErr_SetString(PyExc_TypeError,
1678 "writelines() requires an iterable argument");
1679 return NULL;
1680 }
1681 /* From here on, fail by going to error, to reclaim "it". */
1682 list = PyList_New(CHUNKSIZE);
1683 if (list == NULL)
1684 goto error;
Guido van Rossum5a2a6831993-10-25 09:59:04 +00001685 }
Guido van Rossumee70ad12000-03-13 16:27:06 +00001686
1687 /* Strategy: slurp CHUNKSIZE lines into a private list,
1688 checking that they are all strings, then write that list
1689 without holding the interpreter lock, then come back for more. */
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001690 for (index = 0; ; index += CHUNKSIZE) {
Guido van Rossumee70ad12000-03-13 16:27:06 +00001691 if (islist) {
1692 Py_XDECREF(list);
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001693 list = PyList_GetSlice(seq, index, index+CHUNKSIZE);
Guido van Rossumee70ad12000-03-13 16:27:06 +00001694 if (list == NULL)
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001695 goto error;
Guido van Rossumee70ad12000-03-13 16:27:06 +00001696 j = PyList_GET_SIZE(list);
1697 }
1698 else {
1699 for (j = 0; j < CHUNKSIZE; j++) {
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001700 line = PyIter_Next(it);
Guido van Rossumee70ad12000-03-13 16:27:06 +00001701 if (line == NULL) {
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001702 if (PyErr_Occurred())
1703 goto error;
1704 break;
Guido van Rossumee70ad12000-03-13 16:27:06 +00001705 }
Guido van Rossumee70ad12000-03-13 16:27:06 +00001706 PyList_SetItem(list, j, line);
1707 }
1708 }
1709 if (j == 0)
1710 break;
1711
Marc-André Lemburg6ef68b52000-08-25 22:39:50 +00001712 /* Check that all entries are indeed strings. If not,
1713 apply the same rules as for file.write() and
1714 convert the results to strings. This is slow, but
1715 seems to be the only way since all conversion APIs
1716 could potentially execute Python code. */
1717 for (i = 0; i < j; i++) {
1718 PyObject *v = PyList_GET_ITEM(list, i);
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001719 if (!PyString_Check(v)) {
Marc-André Lemburg6ef68b52000-08-25 22:39:50 +00001720 const char *buffer;
Tim Peters86821b22001-01-07 21:19:34 +00001721 if (((f->f_binary &&
Marc-André Lemburg6ef68b52000-08-25 22:39:50 +00001722 PyObject_AsReadBuffer(v,
1723 (const void**)&buffer,
1724 &len)) ||
1725 PyObject_AsCharBuffer(v,
1726 &buffer,
1727 &len))) {
1728 PyErr_SetString(PyExc_TypeError,
Jeremy Hylton8b735422002-08-14 21:01:41 +00001729 "writelines() argument must be a sequence of strings");
Marc-André Lemburg6ef68b52000-08-25 22:39:50 +00001730 goto error;
1731 }
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001732 line = PyString_FromStringAndSize(buffer,
Marc-André Lemburg6ef68b52000-08-25 22:39:50 +00001733 len);
1734 if (line == NULL)
1735 goto error;
1736 Py_DECREF(v);
Marc-André Lemburgf5e96fa2000-08-25 22:49:05 +00001737 PyList_SET_ITEM(list, i, line);
Marc-André Lemburg6ef68b52000-08-25 22:39:50 +00001738 }
1739 }
1740
1741 /* Since we are releasing the global lock, the
1742 following code may *not* execute Python code. */
Guido van Rossumee70ad12000-03-13 16:27:06 +00001743 f->f_softspace = 0;
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00001744 FILE_BEGIN_ALLOW_THREADS(f)
Guido van Rossumee70ad12000-03-13 16:27:06 +00001745 errno = 0;
1746 for (i = 0; i < j; i++) {
Marc-André Lemburg6ef68b52000-08-25 22:39:50 +00001747 line = PyList_GET_ITEM(list, i);
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001748 len = PyString_GET_SIZE(line);
1749 nwritten = fwrite(PyString_AS_STRING(line),
Guido van Rossumee70ad12000-03-13 16:27:06 +00001750 1, len, f->f_fp);
1751 if (nwritten != len) {
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00001752 FILE_ABORT_ALLOW_THREADS(f)
Guido van Rossumee70ad12000-03-13 16:27:06 +00001753 PyErr_SetFromErrno(PyExc_IOError);
1754 clearerr(f->f_fp);
1755 goto error;
1756 }
1757 }
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00001758 FILE_END_ALLOW_THREADS(f)
Guido van Rossumee70ad12000-03-13 16:27:06 +00001759
1760 if (j < CHUNKSIZE)
1761 break;
Guido van Rossumee70ad12000-03-13 16:27:06 +00001762 }
1763
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001764 Py_INCREF(Py_None);
Guido van Rossumee70ad12000-03-13 16:27:06 +00001765 result = Py_None;
1766 error:
1767 Py_XDECREF(list);
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001768 Py_XDECREF(it);
Guido van Rossumee70ad12000-03-13 16:27:06 +00001769 return result;
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001770#undef CHUNKSIZE
Guido van Rossum5a2a6831993-10-25 09:59:04 +00001771}
1772
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001773static PyObject *
Guido van Rossum1a5e21e2006-02-28 21:57:43 +00001774file_self(PyFileObject *f)
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001775{
1776 if (f->f_fp == NULL)
1777 return err_closed();
1778 Py_INCREF(f);
1779 return (PyObject *)f;
1780}
1781
Georg Brandl98b40ad2006-06-08 14:50:21 +00001782static PyObject *
Georg Brandla9916b52008-05-17 22:11:54 +00001783file_xreadlines(PyFileObject *f)
1784{
1785 if (PyErr_WarnPy3k("f.xreadlines() not supported in 3.x, "
1786 "try 'for line in f' instead", 1) < 0)
1787 return NULL;
1788 return file_self(f);
1789}
1790
1791static PyObject *
Georg Brandlad61bc82008-02-23 15:11:18 +00001792file_exit(PyObject *f, PyObject *args)
Georg Brandl98b40ad2006-06-08 14:50:21 +00001793{
Georg Brandlad61bc82008-02-23 15:11:18 +00001794 PyObject *ret = PyObject_CallMethod(f, "close", NULL);
Georg Brandl98b40ad2006-06-08 14:50:21 +00001795 if (!ret)
1796 /* If error occurred, pass through */
1797 return NULL;
1798 Py_DECREF(ret);
1799 /* We cannot return the result of close since a true
1800 * value will be interpreted as "yes, swallow the
1801 * exception if one was raised inside the with block". */
1802 Py_RETURN_NONE;
1803}
1804
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001805PyDoc_STRVAR(readline_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001806"readline([size]) -> next line from the file, as a string.\n"
1807"\n"
1808"Retain newline. A non-negative size argument limits the maximum\n"
1809"number of bytes to return (an incomplete line may be returned then).\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001810"Return an empty string at EOF.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001811
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001812PyDoc_STRVAR(read_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001813"read([size]) -> read at most size bytes, returned as a string.\n"
1814"\n"
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +00001815"If the size argument is negative or omitted, read until EOF is reached.\n"
1816"Notice that when in non-blocking mode, less data than what was requested\n"
1817"may be returned, even if no size parameter was given.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001818
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001819PyDoc_STRVAR(write_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001820"write(str) -> None. Write string str to file.\n"
1821"\n"
1822"Note that due to buffering, flush() or close() may be needed before\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001823"the file on disk reflects the data written.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001824
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001825PyDoc_STRVAR(fileno_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001826"fileno() -> integer \"file descriptor\".\n"
1827"\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001828"This is needed for lower-level file interfaces, such os.read().");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001829
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001830PyDoc_STRVAR(seek_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001831"seek(offset[, whence]) -> None. Move to new file position.\n"
1832"\n"
1833"Argument offset is a byte count. Optional argument whence defaults to\n"
1834"0 (offset from start of file, offset should be >= 0); other values are 1\n"
1835"(move relative to current position, positive or negative), and 2 (move\n"
1836"relative to end of file, usually negative, although many platforms allow\n"
Martin v. Löwis849a9722003-10-18 09:38:01 +00001837"seeking beyond the end of a file). If the file is opened in text mode,\n"
1838"only offsets returned by tell() are legal. Use of other offsets causes\n"
1839"undefined behavior."
Tim Petersefc3a3a2001-09-20 07:55:22 +00001840"\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001841"Note that not all file objects are seekable.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001842
Guido van Rossumd7047b31995-01-02 19:07:15 +00001843#ifdef HAVE_FTRUNCATE
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001844PyDoc_STRVAR(truncate_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001845"truncate([size]) -> None. Truncate the file to at most size bytes.\n"
1846"\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001847"Size defaults to the current file position, as returned by tell().");
Guido van Rossumd7047b31995-01-02 19:07:15 +00001848#endif
Tim Petersefc3a3a2001-09-20 07:55:22 +00001849
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001850PyDoc_STRVAR(tell_doc,
1851"tell() -> current file position, an integer (may be a long integer).");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001852
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001853PyDoc_STRVAR(readinto_doc,
1854"readinto() -> Undocumented. Don't use this; it may go away.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001855
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001856PyDoc_STRVAR(readlines_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001857"readlines([size]) -> list of strings, each a line from the file.\n"
1858"\n"
1859"Call readline() repeatedly and return a list of the lines so read.\n"
1860"The optional size argument, if given, is an approximate bound on the\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001861"total number of bytes in the lines returned.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001862
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001863PyDoc_STRVAR(xreadlines_doc,
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001864"xreadlines() -> returns self.\n"
Tim Petersefc3a3a2001-09-20 07:55:22 +00001865"\n"
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001866"For backward compatibility. File objects now include the performance\n"
1867"optimizations previously implemented in the xreadlines module.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001868
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001869PyDoc_STRVAR(writelines_doc,
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001870"writelines(sequence_of_strings) -> None. Write the strings to the file.\n"
Tim Petersefc3a3a2001-09-20 07:55:22 +00001871"\n"
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001872"Note that newlines are not added. The sequence can be any iterable object\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001873"producing strings. This is equivalent to calling write() for each string.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001874
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001875PyDoc_STRVAR(flush_doc,
1876"flush() -> None. Flush the internal I/O buffer.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001877
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001878PyDoc_STRVAR(close_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001879"close() -> None or (perhaps) an integer. Close the file.\n"
1880"\n"
Guido van Rossum77f6a652002-04-03 22:41:51 +00001881"Sets data attribute .closed to True. A closed file cannot be used for\n"
Tim Petersefc3a3a2001-09-20 07:55:22 +00001882"further I/O operations. close() may be called more than once without\n"
1883"error. Some kinds of file objects (for example, opened by popen())\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001884"may return an exit status upon closing.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001885
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001886PyDoc_STRVAR(isatty_doc,
1887"isatty() -> true or false. True if the file is connected to a tty device.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001888
Guido van Rossum1a5e21e2006-02-28 21:57:43 +00001889PyDoc_STRVAR(enter_doc,
1890 "__enter__() -> self.");
1891
Georg Brandl98b40ad2006-06-08 14:50:21 +00001892PyDoc_STRVAR(exit_doc,
1893 "__exit__(*excinfo) -> None. Closes the file.");
1894
Tim Petersefc3a3a2001-09-20 07:55:22 +00001895static PyMethodDef file_methods[] = {
Jeremy Hylton8b735422002-08-14 21:01:41 +00001896 {"readline", (PyCFunction)file_readline, METH_VARARGS, readline_doc},
1897 {"read", (PyCFunction)file_read, METH_VARARGS, read_doc},
1898 {"write", (PyCFunction)file_write, METH_VARARGS, write_doc},
1899 {"fileno", (PyCFunction)file_fileno, METH_NOARGS, fileno_doc},
1900 {"seek", (PyCFunction)file_seek, METH_VARARGS, seek_doc},
Tim Petersefc3a3a2001-09-20 07:55:22 +00001901#ifdef HAVE_FTRUNCATE
Jeremy Hylton8b735422002-08-14 21:01:41 +00001902 {"truncate", (PyCFunction)file_truncate, METH_VARARGS, truncate_doc},
Tim Petersefc3a3a2001-09-20 07:55:22 +00001903#endif
Jeremy Hylton8b735422002-08-14 21:01:41 +00001904 {"tell", (PyCFunction)file_tell, METH_NOARGS, tell_doc},
1905 {"readinto", (PyCFunction)file_readinto, METH_VARARGS, readinto_doc},
Georg Brandla9916b52008-05-17 22:11:54 +00001906 {"readlines", (PyCFunction)file_readlines, METH_VARARGS, readlines_doc},
1907 {"xreadlines",(PyCFunction)file_xreadlines, METH_NOARGS, xreadlines_doc},
1908 {"writelines",(PyCFunction)file_writelines, METH_O, writelines_doc},
Jeremy Hylton8b735422002-08-14 21:01:41 +00001909 {"flush", (PyCFunction)file_flush, METH_NOARGS, flush_doc},
1910 {"close", (PyCFunction)file_close, METH_NOARGS, close_doc},
1911 {"isatty", (PyCFunction)file_isatty, METH_NOARGS, isatty_doc},
Guido van Rossum1a5e21e2006-02-28 21:57:43 +00001912 {"__enter__", (PyCFunction)file_self, METH_NOARGS, enter_doc},
Georg Brandl98b40ad2006-06-08 14:50:21 +00001913 {"__exit__", (PyCFunction)file_exit, METH_VARARGS, exit_doc},
Jeremy Hylton8b735422002-08-14 21:01:41 +00001914 {NULL, NULL} /* sentinel */
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001915};
1916
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001917#define OFF(x) offsetof(PyFileObject, x)
Guido van Rossumb6775db1994-08-01 11:34:53 +00001918
Guido van Rossum6f799372001-09-20 20:46:19 +00001919static PyMemberDef file_memberlist[] = {
Guido van Rossum6f799372001-09-20 20:46:19 +00001920 {"mode", T_OBJECT, OFF(f_mode), RO,
Martin v. Löwis6233c9b2002-12-11 13:06:53 +00001921 "file mode ('r', 'U', 'w', 'a', possibly with 'b' or '+' added)"},
Guido van Rossum6f799372001-09-20 20:46:19 +00001922 {"name", T_OBJECT, OFF(f_name), RO,
1923 "file name"},
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00001924 {"encoding", T_OBJECT, OFF(f_encoding), RO,
1925 "file encoding"},
Martin v. Löwis99815892008-06-01 07:20:46 +00001926 {"errors", T_OBJECT, OFF(f_errors), RO,
1927 "Unicode error handler"},
Guido van Rossumb6775db1994-08-01 11:34:53 +00001928 /* getattr(f, "closed") is implemented without this table */
Guido van Rossumb6775db1994-08-01 11:34:53 +00001929 {NULL} /* Sentinel */
1930};
1931
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001932static PyObject *
Tim Peters6d6c1a32001-08-02 04:15:00 +00001933get_closed(PyFileObject *f, void *closure)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001934{
Guido van Rossum77f6a652002-04-03 22:41:51 +00001935 return PyBool_FromLong((long)(f->f_fp == 0));
Guido van Rossumb6775db1994-08-01 11:34:53 +00001936}
Jack Jansen7b8c7542002-04-14 20:12:41 +00001937static PyObject *
1938get_newlines(PyFileObject *f, void *closure)
1939{
1940 switch (f->f_newlinetypes) {
1941 case NEWLINE_UNKNOWN:
1942 Py_INCREF(Py_None);
1943 return Py_None;
1944 case NEWLINE_CR:
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001945 return PyString_FromString("\r");
Jack Jansen7b8c7542002-04-14 20:12:41 +00001946 case NEWLINE_LF:
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001947 return PyString_FromString("\n");
Jack Jansen7b8c7542002-04-14 20:12:41 +00001948 case NEWLINE_CR|NEWLINE_LF:
1949 return Py_BuildValue("(ss)", "\r", "\n");
1950 case NEWLINE_CRLF:
Gregory P. Smithdd96db62008-06-09 04:58:54 +00001951 return PyString_FromString("\r\n");
Jack Jansen7b8c7542002-04-14 20:12:41 +00001952 case NEWLINE_CR|NEWLINE_CRLF:
1953 return Py_BuildValue("(ss)", "\r", "\r\n");
1954 case NEWLINE_LF|NEWLINE_CRLF:
1955 return Py_BuildValue("(ss)", "\n", "\r\n");
1956 case NEWLINE_CR|NEWLINE_LF|NEWLINE_CRLF:
1957 return Py_BuildValue("(sss)", "\r", "\n", "\r\n");
1958 default:
Tim Petersf1827cf2003-09-07 03:30:18 +00001959 PyErr_Format(PyExc_SystemError,
1960 "Unknown newlines value 0x%x\n",
Jeremy Hylton8b735422002-08-14 21:01:41 +00001961 f->f_newlinetypes);
Jack Jansen7b8c7542002-04-14 20:12:41 +00001962 return NULL;
1963 }
1964}
Guido van Rossumb6775db1994-08-01 11:34:53 +00001965
Georg Brandl65bb42d2008-03-21 20:38:24 +00001966static PyObject *
1967get_softspace(PyFileObject *f, void *closure)
1968{
Benjamin Peterson9f4f4812008-04-27 03:01:45 +00001969 if (PyErr_WarnPy3k("file.softspace not supported in 3.x", 1) < 0)
Georg Brandl65bb42d2008-03-21 20:38:24 +00001970 return NULL;
1971 return PyInt_FromLong(f->f_softspace);
1972}
1973
1974static int
1975set_softspace(PyFileObject *f, PyObject *value)
1976{
1977 int new;
Benjamin Peterson9f4f4812008-04-27 03:01:45 +00001978 if (PyErr_WarnPy3k("file.softspace not supported in 3.x", 1) < 0)
Georg Brandl65bb42d2008-03-21 20:38:24 +00001979 return -1;
1980
1981 if (value == NULL) {
1982 PyErr_SetString(PyExc_TypeError,
1983 "can't delete softspace attribute");
1984 return -1;
1985 }
1986
1987 new = PyInt_AsLong(value);
1988 if (new == -1 && PyErr_Occurred())
1989 return -1;
1990 f->f_softspace = new;
1991 return 0;
1992}
1993
Guido van Rossum32d34c82001-09-20 21:45:26 +00001994static PyGetSetDef file_getsetlist[] = {
Guido van Rossum77f6a652002-04-03 22:41:51 +00001995 {"closed", (getter)get_closed, NULL, "True if the file is closed"},
Tim Petersf1827cf2003-09-07 03:30:18 +00001996 {"newlines", (getter)get_newlines, NULL,
Jeremy Hylton8b735422002-08-14 21:01:41 +00001997 "end-of-line convention used in this file"},
Georg Brandl65bb42d2008-03-21 20:38:24 +00001998 {"softspace", (getter)get_softspace, (setter)set_softspace,
1999 "flag indicating that a space needs to be printed; used by print"},
Tim Peters6d6c1a32001-08-02 04:15:00 +00002000 {0},
2001};
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00002002
Neal Norwitzd8b995f2002-08-06 21:50:54 +00002003static void
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002004drop_readahead(PyFileObject *f)
Guido van Rossum65967252001-04-21 13:20:18 +00002005{
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002006 if (f->f_buf != NULL) {
2007 PyMem_Free(f->f_buf);
2008 f->f_buf = NULL;
2009 }
Guido van Rossum65967252001-04-21 13:20:18 +00002010}
2011
Tim Petersf1827cf2003-09-07 03:30:18 +00002012/* Make sure that file has a readahead buffer with at least one byte
2013 (unless at EOF) and no more than bufsize. Returns negative value on
Georg Brandled02eb62006-03-31 20:31:02 +00002014 error, will set MemoryError if bufsize bytes cannot be allocated. */
Neal Norwitzd8b995f2002-08-06 21:50:54 +00002015static int
2016readahead(PyFileObject *f, int bufsize)
2017{
Martin v. Löwis18e16552006-02-15 17:27:45 +00002018 Py_ssize_t chunksize;
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002019
2020 if (f->f_buf != NULL) {
Tim Petersf1827cf2003-09-07 03:30:18 +00002021 if( (f->f_bufend - f->f_bufptr) >= 1)
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002022 return 0;
2023 else
2024 drop_readahead(f);
2025 }
Anthony Baxter377be112006-04-11 06:54:30 +00002026 if ((f->f_buf = (char *)PyMem_Malloc(bufsize)) == NULL) {
Georg Brandled02eb62006-03-31 20:31:02 +00002027 PyErr_NoMemory();
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002028 return -1;
2029 }
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00002030 FILE_BEGIN_ALLOW_THREADS(f)
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002031 errno = 0;
2032 chunksize = Py_UniversalNewlineFread(
2033 f->f_buf, bufsize, f->f_fp, (PyObject *)f);
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00002034 FILE_END_ALLOW_THREADS(f)
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002035 if (chunksize == 0) {
2036 if (ferror(f->f_fp)) {
2037 PyErr_SetFromErrno(PyExc_IOError);
2038 clearerr(f->f_fp);
2039 drop_readahead(f);
2040 return -1;
2041 }
2042 }
2043 f->f_bufptr = f->f_buf;
2044 f->f_bufend = f->f_buf + chunksize;
2045 return 0;
2046}
2047
2048/* Used by file_iternext. The returned string will start with 'skip'
Tim Petersf1827cf2003-09-07 03:30:18 +00002049 uninitialized bytes followed by the remainder of the line. Don't be
2050 horrified by the recursive call: maximum recursion depth is limited by
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002051 logarithmic buffer growth to about 50 even when reading a 1gb line. */
2052
Gregory P. Smithdd96db62008-06-09 04:58:54 +00002053static PyStringObject *
Neal Norwitzd8b995f2002-08-06 21:50:54 +00002054readahead_get_line_skip(PyFileObject *f, int skip, int bufsize)
2055{
Gregory P. Smithdd96db62008-06-09 04:58:54 +00002056 PyStringObject* s;
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002057 char *bufptr;
2058 char *buf;
Martin v. Löwis18e16552006-02-15 17:27:45 +00002059 Py_ssize_t len;
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002060
2061 if (f->f_buf == NULL)
Tim Petersf1827cf2003-09-07 03:30:18 +00002062 if (readahead(f, bufsize) < 0)
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002063 return NULL;
2064
2065 len = f->f_bufend - f->f_bufptr;
Tim Petersf1827cf2003-09-07 03:30:18 +00002066 if (len == 0)
Gregory P. Smithdd96db62008-06-09 04:58:54 +00002067 return (PyStringObject *)
2068 PyString_FromStringAndSize(NULL, skip);
Anthony Baxter377be112006-04-11 06:54:30 +00002069 bufptr = (char *)memchr(f->f_bufptr, '\n', len);
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002070 if (bufptr != NULL) {
2071 bufptr++; /* Count the '\n' */
2072 len = bufptr - f->f_bufptr;
Gregory P. Smithdd96db62008-06-09 04:58:54 +00002073 s = (PyStringObject *)
2074 PyString_FromStringAndSize(NULL, skip+len);
Tim Petersf1827cf2003-09-07 03:30:18 +00002075 if (s == NULL)
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002076 return NULL;
Gregory P. Smithdd96db62008-06-09 04:58:54 +00002077 memcpy(PyString_AS_STRING(s)+skip, f->f_bufptr, len);
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002078 f->f_bufptr = bufptr;
2079 if (bufptr == f->f_bufend)
2080 drop_readahead(f);
2081 } else {
2082 bufptr = f->f_bufptr;
2083 buf = f->f_buf;
2084 f->f_buf = NULL; /* Force new readahead buffer */
Martin v. Löwis18e16552006-02-15 17:27:45 +00002085 assert(skip+len < INT_MAX);
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002086 s = readahead_get_line_skip(
Martin v. Löwis18e16552006-02-15 17:27:45 +00002087 f, (int)(skip+len), bufsize + (bufsize>>2) );
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002088 if (s == NULL) {
2089 PyMem_Free(buf);
2090 return NULL;
2091 }
Gregory P. Smithdd96db62008-06-09 04:58:54 +00002092 memcpy(PyString_AS_STRING(s)+skip, bufptr, len);
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002093 PyMem_Free(buf);
2094 }
2095 return s;
2096}
2097
2098/* A larger buffer size may actually decrease performance. */
2099#define READAHEAD_BUFSIZE 8192
2100
2101static PyObject *
2102file_iternext(PyFileObject *f)
2103{
Gregory P. Smithdd96db62008-06-09 04:58:54 +00002104 PyStringObject* l;
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002105
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002106 if (f->f_fp == NULL)
2107 return err_closed();
2108
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002109 l = readahead_get_line_skip(f, 0, READAHEAD_BUFSIZE);
Gregory P. Smithdd96db62008-06-09 04:58:54 +00002110 if (l == NULL || PyString_GET_SIZE(l) == 0) {
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002111 Py_XDECREF(l);
2112 return NULL;
2113 }
2114 return (PyObject *)l;
2115}
2116
2117
Tim Peters59c9a642001-09-13 05:38:56 +00002118static PyObject *
2119file_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
2120{
Tim Peters44410012001-09-14 03:26:08 +00002121 PyObject *self;
2122 static PyObject *not_yet_string;
2123
2124 assert(type != NULL && type->tp_alloc != NULL);
2125
2126 if (not_yet_string == NULL) {
Gregory P. Smithdd96db62008-06-09 04:58:54 +00002127 not_yet_string = PyString_InternFromString("<uninitialized file>");
Tim Peters44410012001-09-14 03:26:08 +00002128 if (not_yet_string == NULL)
2129 return NULL;
2130 }
2131
2132 self = type->tp_alloc(type, 0);
2133 if (self != NULL) {
2134 /* Always fill in the name and mode, so that nobody else
2135 needs to special-case NULLs there. */
2136 Py_INCREF(not_yet_string);
2137 ((PyFileObject *)self)->f_name = not_yet_string;
2138 Py_INCREF(not_yet_string);
2139 ((PyFileObject *)self)->f_mode = not_yet_string;
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00002140 Py_INCREF(Py_None);
2141 ((PyFileObject *)self)->f_encoding = Py_None;
Martin v. Löwis99815892008-06-01 07:20:46 +00002142 Py_INCREF(Py_None);
2143 ((PyFileObject *)self)->f_errors = Py_None;
Raymond Hettingercb87bc82004-05-31 00:35:52 +00002144 ((PyFileObject *)self)->weakreflist = NULL;
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00002145 ((PyFileObject *)self)->unlocked_count = 0;
Tim Peters44410012001-09-14 03:26:08 +00002146 }
2147 return self;
2148}
2149
2150static int
2151file_init(PyObject *self, PyObject *args, PyObject *kwds)
2152{
2153 PyFileObject *foself = (PyFileObject *)self;
2154 int ret = 0;
Martin v. Löwis15e62742006-02-27 16:46:16 +00002155 static char *kwlist[] = {"name", "mode", "buffering", 0};
Tim Peters59c9a642001-09-13 05:38:56 +00002156 char *name = NULL;
2157 char *mode = "r";
2158 int bufsize = -1;
Mark Hammondc2e85bd2002-10-03 05:10:39 +00002159 int wideargument = 0;
Tim Peters44410012001-09-14 03:26:08 +00002160
2161 assert(PyFile_Check(self));
2162 if (foself->f_fp != NULL) {
2163 /* Have to close the existing file first. */
2164 PyObject *closeresult = file_close(foself);
2165 if (closeresult == NULL)
2166 return -1;
2167 Py_DECREF(closeresult);
2168 }
Tim Peters59c9a642001-09-13 05:38:56 +00002169
Mark Hammondc2e85bd2002-10-03 05:10:39 +00002170#ifdef Py_WIN_WIDE_FILENAMES
2171 if (GetVersion() < 0x80000000) { /* On NT, so wide API available */
2172 PyObject *po;
2173 if (PyArg_ParseTupleAndKeywords(args, kwds, "U|si:file",
2174 kwlist, &po, &mode, &bufsize)) {
2175 wideargument = 1;
Nicholas Bastinabce8a62004-03-21 20:24:07 +00002176 if (fill_file_fields(foself, NULL, po, mode,
2177 fclose) == NULL)
Mark Hammondc2e85bd2002-10-03 05:10:39 +00002178 goto Error;
2179 } else {
2180 /* Drop the argument parsing error as narrow
2181 strings are also valid. */
2182 PyErr_Clear();
2183 }
2184 }
2185#endif
2186
2187 if (!wideargument) {
Nicholas Bastinabce8a62004-03-21 20:24:07 +00002188 PyObject *o_name;
2189
Mark Hammondc2e85bd2002-10-03 05:10:39 +00002190 if (!PyArg_ParseTupleAndKeywords(args, kwds, "et|si:file", kwlist,
2191 Py_FileSystemDefaultEncoding,
2192 &name,
2193 &mode, &bufsize))
2194 return -1;
Nicholas Bastinabce8a62004-03-21 20:24:07 +00002195
2196 /* We parse again to get the name as a PyObject */
Jeremy Hyltonaf68c872005-12-10 18:50:16 +00002197 if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|si:file",
2198 kwlist, &o_name, &mode,
2199 &bufsize))
Brett Cannon2b3666f2006-08-31 18:54:26 +00002200 goto Error;
Nicholas Bastinabce8a62004-03-21 20:24:07 +00002201
2202 if (fill_file_fields(foself, NULL, o_name, mode,
2203 fclose) == NULL)
Mark Hammondc2e85bd2002-10-03 05:10:39 +00002204 goto Error;
2205 }
Tim Peters44410012001-09-14 03:26:08 +00002206 if (open_the_file(foself, name, mode) == NULL)
2207 goto Error;
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +00002208 foself->f_setbuf = NULL;
Tim Peters44410012001-09-14 03:26:08 +00002209 PyFile_SetBufSize(self, bufsize);
2210 goto Done;
2211
2212Error:
2213 ret = -1;
2214 /* fall through */
2215Done:
Tim Peters59c9a642001-09-13 05:38:56 +00002216 PyMem_Free(name); /* free the encoded string */
Tim Peters44410012001-09-14 03:26:08 +00002217 return ret;
Tim Peters59c9a642001-09-13 05:38:56 +00002218}
2219
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00002220PyDoc_VAR(file_doc) =
2221PyDoc_STR(
Tim Peters59c9a642001-09-13 05:38:56 +00002222"file(name[, mode[, buffering]]) -> file object\n"
2223"\n"
2224"Open a file. The mode can be 'r', 'w' or 'a' for reading (default),\n"
2225"writing or appending. The file will be created if it doesn't exist\n"
2226"when opened for writing or appending; it will be truncated when\n"
2227"opened for writing. Add a 'b' to the mode for binary files.\n"
2228"Add a '+' to the mode to allow simultaneous reading and writing.\n"
2229"If the buffering argument is given, 0 means unbuffered, 1 means line\n"
Skip Montanaro4e3ebe02007-12-08 14:37:43 +00002230"buffered, and larger numbers specify the buffer size. The preferred way\n"
2231"to open a file is with the builtin open() function.\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00002232)
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00002233PyDoc_STR(
Barry Warsaw4be55b52002-05-22 20:37:53 +00002234"Add a 'U' to mode to open the file for input with universal newline\n"
2235"support. Any line ending in the input file will be seen as a '\\n'\n"
2236"in Python. Also, a file so opened gains the attribute 'newlines';\n"
2237"the value for this attribute is one of None (no newline read yet),\n"
2238"'\\r', '\\n', '\\r\\n' or a tuple containing all the newline types seen.\n"
2239"\n"
2240"'U' cannot be combined with 'w' or '+' mode.\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00002241);
Tim Peters59c9a642001-09-13 05:38:56 +00002242
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002243PyTypeObject PyFile_Type = {
Martin v. Löwis68192102007-07-21 06:55:02 +00002244 PyVarObject_HEAD_INIT(&PyType_Type, 0)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00002245 "file",
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002246 sizeof(PyFileObject),
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00002247 0,
Guido van Rossum65967252001-04-21 13:20:18 +00002248 (destructor)file_dealloc, /* tp_dealloc */
2249 0, /* tp_print */
Tim Peters6d6c1a32001-08-02 04:15:00 +00002250 0, /* tp_getattr */
2251 0, /* tp_setattr */
Guido van Rossum65967252001-04-21 13:20:18 +00002252 0, /* tp_compare */
Tim Peters6d6c1a32001-08-02 04:15:00 +00002253 (reprfunc)file_repr, /* tp_repr */
Guido van Rossum65967252001-04-21 13:20:18 +00002254 0, /* tp_as_number */
2255 0, /* tp_as_sequence */
2256 0, /* tp_as_mapping */
2257 0, /* tp_hash */
2258 0, /* tp_call */
2259 0, /* tp_str */
Tim Peters6d6c1a32001-08-02 04:15:00 +00002260 PyObject_GenericGetAttr, /* tp_getattro */
Tim Peters015dd822003-05-04 04:16:52 +00002261 /* softspace is writable: we must supply tp_setattro */
2262 PyObject_GenericSetAttr, /* tp_setattro */
Guido van Rossum65967252001-04-21 13:20:18 +00002263 0, /* tp_as_buffer */
Raymond Hettingercb87bc82004-05-31 00:35:52 +00002264 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | Py_TPFLAGS_HAVE_WEAKREFS, /* tp_flags */
Tim Peters59c9a642001-09-13 05:38:56 +00002265 file_doc, /* tp_doc */
Tim Peters6d6c1a32001-08-02 04:15:00 +00002266 0, /* tp_traverse */
2267 0, /* tp_clear */
Guido van Rossum65967252001-04-21 13:20:18 +00002268 0, /* tp_richcompare */
Raymond Hettingercb87bc82004-05-31 00:35:52 +00002269 offsetof(PyFileObject, weakreflist), /* tp_weaklistoffset */
Guido van Rossum1a5e21e2006-02-28 21:57:43 +00002270 (getiterfunc)file_self, /* tp_iter */
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002271 (iternextfunc)file_iternext, /* tp_iternext */
Tim Peters6d6c1a32001-08-02 04:15:00 +00002272 file_methods, /* tp_methods */
2273 file_memberlist, /* tp_members */
2274 file_getsetlist, /* tp_getset */
2275 0, /* tp_base */
2276 0, /* tp_dict */
Tim Peters59c9a642001-09-13 05:38:56 +00002277 0, /* tp_descr_get */
2278 0, /* tp_descr_set */
2279 0, /* tp_dictoffset */
Georg Brandl347b3002006-03-30 11:57:00 +00002280 file_init, /* tp_init */
Tim Peters44410012001-09-14 03:26:08 +00002281 PyType_GenericAlloc, /* tp_alloc */
Tim Peters59c9a642001-09-13 05:38:56 +00002282 file_new, /* tp_new */
Neil Schemenaueraa769ae2002-04-12 02:44:10 +00002283 PyObject_Del, /* tp_free */
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00002284};
Guido van Rossumeb183da1991-04-04 10:44:06 +00002285
2286/* Interface for the 'soft space' between print items. */
2287
2288int
Fred Drakefd99de62000-07-09 05:02:18 +00002289PyFile_SoftSpace(PyObject *f, int newflag)
Guido van Rossumeb183da1991-04-04 10:44:06 +00002290{
Martin v. Löwis18e16552006-02-15 17:27:45 +00002291 long oldflag = 0;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002292 if (f == NULL) {
2293 /* Do nothing */
2294 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002295 else if (PyFile_Check(f)) {
2296 oldflag = ((PyFileObject *)f)->f_softspace;
2297 ((PyFileObject *)f)->f_softspace = newflag;
Guido van Rossumeb183da1991-04-04 10:44:06 +00002298 }
Guido van Rossum3165fe61992-09-25 21:59:05 +00002299 else {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002300 PyObject *v;
2301 v = PyObject_GetAttrString(f, "softspace");
Guido van Rossum3165fe61992-09-25 21:59:05 +00002302 if (v == NULL)
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002303 PyErr_Clear();
Guido van Rossum3165fe61992-09-25 21:59:05 +00002304 else {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002305 if (PyInt_Check(v))
2306 oldflag = PyInt_AsLong(v);
Martin v. Löwis18e16552006-02-15 17:27:45 +00002307 assert(oldflag < INT_MAX);
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002308 Py_DECREF(v);
Guido van Rossum3165fe61992-09-25 21:59:05 +00002309 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002310 v = PyInt_FromLong((long)newflag);
Guido van Rossum3165fe61992-09-25 21:59:05 +00002311 if (v == NULL)
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002312 PyErr_Clear();
Guido van Rossum3165fe61992-09-25 21:59:05 +00002313 else {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002314 if (PyObject_SetAttrString(f, "softspace", v) != 0)
2315 PyErr_Clear();
2316 Py_DECREF(v);
Guido van Rossum3165fe61992-09-25 21:59:05 +00002317 }
2318 }
Martin v. Löwis18e16552006-02-15 17:27:45 +00002319 return (int)oldflag;
Guido van Rossumeb183da1991-04-04 10:44:06 +00002320}
Guido van Rossum3165fe61992-09-25 21:59:05 +00002321
2322/* Interfaces to write objects/strings to file-like objects */
2323
2324int
Fred Drakefd99de62000-07-09 05:02:18 +00002325PyFile_WriteObject(PyObject *v, PyObject *f, int flags)
Guido van Rossum3165fe61992-09-25 21:59:05 +00002326{
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002327 PyObject *writer, *value, *args, *result;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002328 if (f == NULL) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002329 PyErr_SetString(PyExc_TypeError, "writeobject with NULL file");
Guido van Rossum3165fe61992-09-25 21:59:05 +00002330 return -1;
2331 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002332 else if (PyFile_Check(f)) {
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00002333 PyFileObject *fobj = (PyFileObject *) f;
Fred Drake086a0f72004-03-19 15:22:36 +00002334#ifdef Py_USING_UNICODE
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00002335 PyObject *enc = fobj->f_encoding;
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00002336 int result;
Fred Drake086a0f72004-03-19 15:22:36 +00002337#endif
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00002338 if (fobj->f_fp == NULL) {
Guido van Rossum3165fe61992-09-25 21:59:05 +00002339 err_closed();
2340 return -1;
2341 }
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00002342#ifdef Py_USING_UNICODE
Tim Petersf1827cf2003-09-07 03:30:18 +00002343 if ((flags & Py_PRINT_RAW) &&
Martin v. Löwis415da6e2003-05-18 12:56:25 +00002344 PyUnicode_Check(v) && enc != Py_None) {
Gregory P. Smith99a3dce2008-06-10 17:42:36 +00002345 char *cenc = PyString_AS_STRING(enc);
Martin v. Löwis99815892008-06-01 07:20:46 +00002346 char *errors = fobj->f_errors == Py_None ?
Gregory P. Smith99a3dce2008-06-10 17:42:36 +00002347 "strict" : PyString_AS_STRING(fobj->f_errors);
Martin v. Löwis99815892008-06-01 07:20:46 +00002348 value = PyUnicode_AsEncodedString(v, cenc, errors);
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00002349 if (value == NULL)
2350 return -1;
2351 } else {
2352 value = v;
2353 Py_INCREF(value);
2354 }
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00002355 result = file_PyObject_Print(value, fobj, flags);
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00002356 Py_DECREF(value);
2357 return result;
2358#else
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00002359 return file_PyObject_Print(v, fobj, flags);
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00002360#endif
Guido van Rossum3165fe61992-09-25 21:59:05 +00002361 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002362 writer = PyObject_GetAttrString(f, "write");
Guido van Rossum3165fe61992-09-25 21:59:05 +00002363 if (writer == NULL)
2364 return -1;
Martin v. Löwis2777c022001-09-19 13:47:32 +00002365 if (flags & Py_PRINT_RAW) {
2366 if (PyUnicode_Check(v)) {
2367 value = v;
2368 Py_INCREF(value);
2369 } else
2370 value = PyObject_Str(v);
2371 }
2372 else
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002373 value = PyObject_Repr(v);
Guido van Rossumc6004111993-11-05 10:22:19 +00002374 if (value == NULL) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002375 Py_DECREF(writer);
Guido van Rossumc6004111993-11-05 10:22:19 +00002376 return -1;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002377 }
Raymond Hettinger8ae46892003-10-12 19:09:37 +00002378 args = PyTuple_Pack(1, value);
Guido van Rossume9eec541997-05-22 14:02:25 +00002379 if (args == NULL) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002380 Py_DECREF(value);
2381 Py_DECREF(writer);
Guido van Rossumd3f9a1a1995-07-10 23:32:26 +00002382 return -1;
2383 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002384 result = PyEval_CallObject(writer, args);
2385 Py_DECREF(args);
2386 Py_DECREF(value);
2387 Py_DECREF(writer);
Guido van Rossum3165fe61992-09-25 21:59:05 +00002388 if (result == NULL)
2389 return -1;
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002390 Py_DECREF(result);
Guido van Rossum3165fe61992-09-25 21:59:05 +00002391 return 0;
2392}
2393
Guido van Rossum27a60b11997-05-22 22:25:11 +00002394int
Tim Petersc1bbcb82001-11-28 22:13:25 +00002395PyFile_WriteString(const char *s, PyObject *f)
Guido van Rossum3165fe61992-09-25 21:59:05 +00002396{
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00002397
Guido van Rossum3165fe61992-09-25 21:59:05 +00002398 if (f == NULL) {
Guido van Rossum27a60b11997-05-22 22:25:11 +00002399 /* Should be caused by a pre-existing error */
Fred Drakefd99de62000-07-09 05:02:18 +00002400 if (!PyErr_Occurred())
Guido van Rossum27a60b11997-05-22 22:25:11 +00002401 PyErr_SetString(PyExc_SystemError,
2402 "null file for PyFile_WriteString");
2403 return -1;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002404 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002405 else if (PyFile_Check(f)) {
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00002406 PyFileObject *fobj = (PyFileObject *) f;
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002407 FILE *fp = PyFile_AsFile(f);
Guido van Rossum27a60b11997-05-22 22:25:11 +00002408 if (fp == NULL) {
2409 err_closed();
2410 return -1;
2411 }
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00002412 FILE_BEGIN_ALLOW_THREADS(fobj)
Guido van Rossum27a60b11997-05-22 22:25:11 +00002413 fputs(s, fp);
Gregory P. Smithaa63d0d2008-04-06 23:11:17 +00002414 FILE_END_ALLOW_THREADS(fobj)
Guido van Rossum27a60b11997-05-22 22:25:11 +00002415 return 0;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002416 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002417 else if (!PyErr_Occurred()) {
Gregory P. Smithdd96db62008-06-09 04:58:54 +00002418 PyObject *v = PyString_FromString(s);
Guido van Rossum27a60b11997-05-22 22:25:11 +00002419 int err;
2420 if (v == NULL)
2421 return -1;
2422 err = PyFile_WriteObject(v, f, Py_PRINT_RAW);
2423 Py_DECREF(v);
2424 return err;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002425 }
Guido van Rossum74ba2471997-07-13 03:56:50 +00002426 else
2427 return -1;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002428}
Andrew M. Kuchling06051ed2000-07-13 23:56:54 +00002429
2430/* Try to get a file-descriptor from a Python object. If the object
2431 is an integer or long integer, its value is returned. If not, the
2432 object's fileno() method is called if it exists; the method must return
2433 an integer or long integer, which is returned as the file descriptor value.
2434 -1 is returned on failure.
2435*/
2436
2437int PyObject_AsFileDescriptor(PyObject *o)
2438{
2439 int fd;
2440 PyObject *meth;
2441
2442 if (PyInt_Check(o)) {
2443 fd = PyInt_AsLong(o);
2444 }
2445 else if (PyLong_Check(o)) {
2446 fd = PyLong_AsLong(o);
2447 }
2448 else if ((meth = PyObject_GetAttrString(o, "fileno")) != NULL)
2449 {
2450 PyObject *fno = PyEval_CallObject(meth, NULL);
2451 Py_DECREF(meth);
2452 if (fno == NULL)
2453 return -1;
Tim Peters86821b22001-01-07 21:19:34 +00002454
Andrew M. Kuchling06051ed2000-07-13 23:56:54 +00002455 if (PyInt_Check(fno)) {
2456 fd = PyInt_AsLong(fno);
2457 Py_DECREF(fno);
2458 }
2459 else if (PyLong_Check(fno)) {
2460 fd = PyLong_AsLong(fno);
2461 Py_DECREF(fno);
2462 }
2463 else {
2464 PyErr_SetString(PyExc_TypeError,
2465 "fileno() returned a non-integer");
2466 Py_DECREF(fno);
2467 return -1;
2468 }
2469 }
2470 else {
2471 PyErr_SetString(PyExc_TypeError,
2472 "argument must be an int, or have a fileno() method.");
2473 return -1;
2474 }
2475
2476 if (fd < 0) {
2477 PyErr_Format(PyExc_ValueError,
2478 "file descriptor cannot be a negative integer (%i)",
2479 fd);
2480 return -1;
2481 }
2482 return fd;
2483}
Jack Jansen7b8c7542002-04-14 20:12:41 +00002484
Jack Jansen7b8c7542002-04-14 20:12:41 +00002485/* From here on we need access to the real fgets and fread */
2486#undef fgets
2487#undef fread
2488
2489/*
2490** Py_UniversalNewlineFgets is an fgets variation that understands
2491** all of \r, \n and \r\n conventions.
2492** The stream should be opened in binary mode.
2493** If fobj is NULL the routine always does newline conversion, and
2494** it may peek one char ahead to gobble the second char in \r\n.
2495** If fobj is non-NULL it must be a PyFileObject. In this case there
2496** is no readahead but in stead a flag is used to skip a following
2497** \n on the next read. Also, if the file is open in binary mode
2498** the whole conversion is skipped. Finally, the routine keeps track of
2499** the different types of newlines seen.
2500** Note that we need no error handling: fgets() treats error and eof
2501** identically.
2502*/
2503char *
2504Py_UniversalNewlineFgets(char *buf, int n, FILE *stream, PyObject *fobj)
2505{
2506 char *p = buf;
2507 int c;
2508 int newlinetypes = 0;
2509 int skipnextlf = 0;
2510 int univ_newline = 1;
Tim Peters058b1412002-04-21 07:29:14 +00002511
Jack Jansen7b8c7542002-04-14 20:12:41 +00002512 if (fobj) {
2513 if (!PyFile_Check(fobj)) {
2514 errno = ENXIO; /* What can you do... */
2515 return NULL;
2516 }
2517 univ_newline = ((PyFileObject *)fobj)->f_univ_newline;
2518 if ( !univ_newline )
2519 return fgets(buf, n, stream);
2520 newlinetypes = ((PyFileObject *)fobj)->f_newlinetypes;
2521 skipnextlf = ((PyFileObject *)fobj)->f_skipnextlf;
2522 }
2523 FLOCKFILE(stream);
2524 c = 'x'; /* Shut up gcc warning */
2525 while (--n > 0 && (c = GETC(stream)) != EOF ) {
2526 if (skipnextlf ) {
2527 skipnextlf = 0;
2528 if (c == '\n') {
2529 /* Seeing a \n here with skipnextlf true
2530 ** means we saw a \r before.
2531 */
2532 newlinetypes |= NEWLINE_CRLF;
2533 c = GETC(stream);
2534 if (c == EOF) break;
2535 } else {
2536 /*
2537 ** Note that c == EOF also brings us here,
2538 ** so we're okay if the last char in the file
2539 ** is a CR.
2540 */
2541 newlinetypes |= NEWLINE_CR;
2542 }
2543 }
2544 if (c == '\r') {
2545 /* A \r is translated into a \n, and we skip
2546 ** an adjacent \n, if any. We don't set the
2547 ** newlinetypes flag until we've seen the next char.
2548 */
2549 skipnextlf = 1;
2550 c = '\n';
2551 } else if ( c == '\n') {
2552 newlinetypes |= NEWLINE_LF;
2553 }
2554 *p++ = c;
2555 if (c == '\n') break;
2556 }
2557 if ( c == EOF && skipnextlf )
2558 newlinetypes |= NEWLINE_CR;
2559 FUNLOCKFILE(stream);
2560 *p = '\0';
2561 if (fobj) {
2562 ((PyFileObject *)fobj)->f_newlinetypes = newlinetypes;
2563 ((PyFileObject *)fobj)->f_skipnextlf = skipnextlf;
2564 } else if ( skipnextlf ) {
2565 /* If we have no file object we cannot save the
2566 ** skipnextlf flag. We have to readahead, which
2567 ** will cause a pause if we're reading from an
2568 ** interactive stream, but that is very unlikely
2569 ** unless we're doing something silly like
2570 ** execfile("/dev/tty").
2571 */
2572 c = GETC(stream);
2573 if ( c != '\n' )
2574 ungetc(c, stream);
2575 }
2576 if (p == buf)
2577 return NULL;
2578 return buf;
2579}
2580
2581/*
2582** Py_UniversalNewlineFread is an fread variation that understands
2583** all of \r, \n and \r\n conventions.
2584** The stream should be opened in binary mode.
2585** fobj must be a PyFileObject. In this case there
2586** is no readahead but in stead a flag is used to skip a following
2587** \n on the next read. Also, if the file is open in binary mode
2588** the whole conversion is skipped. Finally, the routine keeps track of
2589** the different types of newlines seen.
2590*/
2591size_t
Tim Peters058b1412002-04-21 07:29:14 +00002592Py_UniversalNewlineFread(char *buf, size_t n,
Jack Jansen7b8c7542002-04-14 20:12:41 +00002593 FILE *stream, PyObject *fobj)
2594{
Tim Peters058b1412002-04-21 07:29:14 +00002595 char *dst = buf;
2596 PyFileObject *f = (PyFileObject *)fobj;
2597 int newlinetypes, skipnextlf;
2598
2599 assert(buf != NULL);
2600 assert(stream != NULL);
2601
Jack Jansen7b8c7542002-04-14 20:12:41 +00002602 if (!fobj || !PyFile_Check(fobj)) {
2603 errno = ENXIO; /* What can you do... */
Neal Norwitzcb3319f2003-02-09 01:10:02 +00002604 return 0;
Jack Jansen7b8c7542002-04-14 20:12:41 +00002605 }
Tim Peters058b1412002-04-21 07:29:14 +00002606 if (!f->f_univ_newline)
Jack Jansen7b8c7542002-04-14 20:12:41 +00002607 return fread(buf, 1, n, stream);
Tim Peters058b1412002-04-21 07:29:14 +00002608 newlinetypes = f->f_newlinetypes;
2609 skipnextlf = f->f_skipnextlf;
2610 /* Invariant: n is the number of bytes remaining to be filled
2611 * in the buffer.
2612 */
2613 while (n) {
2614 size_t nread;
2615 int shortread;
2616 char *src = dst;
2617
2618 nread = fread(dst, 1, n, stream);
2619 assert(nread <= n);
Neal Norwitzcb3319f2003-02-09 01:10:02 +00002620 if (nread == 0)
2621 break;
2622
Tim Peterse1682a82002-04-21 18:15:20 +00002623 n -= nread; /* assuming 1 byte out for each in; will adjust */
2624 shortread = n != 0; /* true iff EOF or error */
Tim Peters058b1412002-04-21 07:29:14 +00002625 while (nread--) {
2626 char c = *src++;
Jack Jansen7b8c7542002-04-14 20:12:41 +00002627 if (c == '\r') {
Tim Peters058b1412002-04-21 07:29:14 +00002628 /* Save as LF and set flag to skip next LF. */
Jack Jansen7b8c7542002-04-14 20:12:41 +00002629 *dst++ = '\n';
2630 skipnextlf = 1;
Tim Peters058b1412002-04-21 07:29:14 +00002631 }
2632 else if (skipnextlf && c == '\n') {
2633 /* Skip LF, and remember we saw CR LF. */
Jack Jansen7b8c7542002-04-14 20:12:41 +00002634 skipnextlf = 0;
2635 newlinetypes |= NEWLINE_CRLF;
Tim Peterse1682a82002-04-21 18:15:20 +00002636 ++n;
Tim Peters058b1412002-04-21 07:29:14 +00002637 }
2638 else {
2639 /* Normal char to be stored in buffer. Also
2640 * update the newlinetypes flag if either this
2641 * is an LF or the previous char was a CR.
2642 */
Jack Jansen7b8c7542002-04-14 20:12:41 +00002643 if (c == '\n')
2644 newlinetypes |= NEWLINE_LF;
2645 else if (skipnextlf)
2646 newlinetypes |= NEWLINE_CR;
2647 *dst++ = c;
2648 skipnextlf = 0;
2649 }
2650 }
Tim Peters058b1412002-04-21 07:29:14 +00002651 if (shortread) {
2652 /* If this is EOF, update type flags. */
2653 if (skipnextlf && feof(stream))
2654 newlinetypes |= NEWLINE_CR;
2655 break;
2656 }
Jack Jansen7b8c7542002-04-14 20:12:41 +00002657 }
Tim Peters058b1412002-04-21 07:29:14 +00002658 f->f_newlinetypes = newlinetypes;
2659 f->f_skipnextlf = skipnextlf;
2660 return dst - buf;
Jack Jansen7b8c7542002-04-14 20:12:41 +00002661}
Anthony Baxterac6bd462006-04-13 02:06:09 +00002662
2663#ifdef __cplusplus
2664}
2665#endif