blob: d12e1325b01c9327158efb92cca44ebbfbd8678e [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
Thomas Wouters0e3f5912006-08-11 14:57:12 +00007#ifdef HAVE_SYS_TYPES_H
Guido van Rossum41498431999-01-07 22:09:51 +00008#include <sys/types.h>
Thomas Wouters0e3f5912006-08-11 14:57:12 +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
Guido van Rossumc0b618a1997-05-02 03:12:38 +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
Thomas Wouters49fd7fa2006-04-21 10:40:58 +000051#ifdef __cplusplus
52extern "C" {
53#endif
54
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000055FILE *
Fred Drakefd99de62000-07-09 05:02:18 +000056PyFile_AsFile(PyObject *f)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000057{
Guido van Rossumc0b618a1997-05-02 03:12:38 +000058 if (f == NULL || !PyFile_Check(f))
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000059 return NULL;
Guido van Rossum3165fe61992-09-25 21:59:05 +000060 else
Guido van Rossumc0b618a1997-05-02 03:12:38 +000061 return ((PyFileObject *)f)->f_fp;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000062}
63
Guido van Rossumc0b618a1997-05-02 03:12:38 +000064PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +000065PyFile_Name(PyObject *f)
Guido van Rossumdb3165e1993-10-18 17:06:59 +000066{
Guido van Rossumc0b618a1997-05-02 03:12:38 +000067 if (f == NULL || !PyFile_Check(f))
Guido van Rossumdb3165e1993-10-18 17:06:59 +000068 return NULL;
69 else
Guido van Rossumc0b618a1997-05-02 03:12:38 +000070 return ((PyFileObject *)f)->f_name;
Guido van Rossumdb3165e1993-10-18 17:06:59 +000071}
72
Neil Schemenauered19b882002-03-23 02:06:50 +000073/* On Unix, fopen will succeed for directories.
74 In Python, there should be no file objects referring to
75 directories, so we need a check. */
76
77static PyFileObject*
78dircheck(PyFileObject* f)
79{
80#if defined(HAVE_FSTAT) && defined(S_IFDIR) && defined(EISDIR)
81 struct stat buf;
82 if (f->f_fp == NULL)
83 return f;
84 if (fstat(fileno(f->f_fp), &buf) == 0 &&
85 S_ISDIR(buf.st_mode)) {
86#ifdef HAVE_STRERROR
87 char *msg = strerror(EISDIR);
88#else
89 char *msg = "Is a directory";
90#endif
Tim Petersf1827cf2003-09-07 03:30:18 +000091 PyObject *exc = PyObject_CallFunction(PyExc_IOError, "(is)",
Jeremy Hylton8b735422002-08-14 21:01:41 +000092 EISDIR, msg);
Neil Schemenauered19b882002-03-23 02:06:50 +000093 PyErr_SetObject(PyExc_IOError, exc);
Neal Norwitz98cad482003-08-15 20:05:45 +000094 Py_XDECREF(exc);
Neil Schemenauered19b882002-03-23 02:06:50 +000095 return NULL;
96 }
97#endif
98 return f;
99}
100
Tim Peters59c9a642001-09-13 05:38:56 +0000101
102static PyObject *
Nicholas Bastinabce8a62004-03-21 20:24:07 +0000103fill_file_fields(PyFileObject *f, FILE *fp, PyObject *name, char *mode,
104 int (*close)(FILE *))
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000105{
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000106 assert(name != NULL);
Tim Peters59c9a642001-09-13 05:38:56 +0000107 assert(f != NULL);
108 assert(PyFile_Check(f));
Tim Peters44410012001-09-14 03:26:08 +0000109 assert(f->f_fp == NULL);
110
111 Py_DECREF(f->f_name);
112 Py_DECREF(f->f_mode);
Martin v. Löwis5467d4c2003-05-10 07:10:12 +0000113 Py_DECREF(f->f_encoding);
Nicholas Bastinabce8a62004-03-21 20:24:07 +0000114
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000115 Py_INCREF(name);
Nicholas Bastinabce8a62004-03-21 20:24:07 +0000116 f->f_name = name;
117
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000118 f->f_mode = PyString_FromString(mode);
Tim Peters44410012001-09-14 03:26:08 +0000119
Guido van Rossuma1ab7fa1991-06-04 19:37:39 +0000120 f->f_close = close;
Tim Peters59c9a642001-09-13 05:38:56 +0000121 f->f_binary = strchr(mode,'b') != NULL;
Guido van Rossum7a6e9592002-08-06 15:55:28 +0000122 f->f_buf = NULL;
Jack Jansen7b8c7542002-04-14 20:12:41 +0000123 f->f_univ_newline = (strchr(mode, 'U') != NULL);
124 f->f_newlinetypes = NEWLINE_UNKNOWN;
125 f->f_skipnextlf = 0;
Martin v. Löwis5467d4c2003-05-10 07:10:12 +0000126 Py_INCREF(Py_None);
127 f->f_encoding = Py_None;
Tim Petersf1827cf2003-09-07 03:30:18 +0000128
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000129 if (f->f_mode == NULL)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000130 return NULL;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000131 f->f_fp = fp;
Neil Schemenauered19b882002-03-23 02:06:50 +0000132 f = dircheck(f);
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000133 return (PyObject *) f;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000134}
135
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000136/* check for known incorrect mode strings - problem is, platforms are
137 free to accept any mode characters they like and are supposed to
138 ignore stuff they don't understand... write or append mode with
Thomas Wouters477c8d52006-05-27 19:21:47 +0000139 universal newline support is expressly forbidden by PEP 278.
140 Additionally, remove the 'U' from the mode string as platforms
Guido van Rossumd8faa362007-04-27 19:54:29 +0000141 won't know what it is. Non-zero return signals an exception */
142int
143_PyFile_SanitizeMode(char *mode)
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000144{
Thomas Wouters477c8d52006-05-27 19:21:47 +0000145 char *upos;
Neal Norwitz76dc0812006-01-08 06:13:13 +0000146 size_t len = strlen(mode);
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000147
Thomas Wouters477c8d52006-05-27 19:21:47 +0000148 if (!len) {
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000149 PyErr_SetString(PyExc_ValueError, "empty mode string");
Guido van Rossumd8faa362007-04-27 19:54:29 +0000150 return -1;
Thomas Wouters477c8d52006-05-27 19:21:47 +0000151 }
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000152
Thomas Wouters477c8d52006-05-27 19:21:47 +0000153 upos = strchr(mode, 'U');
154 if (upos) {
155 memmove(upos, upos+1, len-(upos-mode)); /* incl null char */
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000156
Thomas Wouters477c8d52006-05-27 19:21:47 +0000157 if (mode[0] == 'w' || mode[0] == 'a') {
158 PyErr_Format(PyExc_ValueError, "universal newline "
159 "mode can only be used with modes "
160 "starting with 'r'");
Guido van Rossumd8faa362007-04-27 19:54:29 +0000161 return -1;
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000162 }
Thomas Wouters477c8d52006-05-27 19:21:47 +0000163
164 if (mode[0] != 'r') {
165 memmove(mode+1, mode, strlen(mode)+1);
166 mode[0] = 'r';
167 }
168
169 if (!strchr(mode, 'b')) {
170 memmove(mode+2, mode+1, strlen(mode));
171 mode[1] = 'b';
172 }
173 } else if (mode[0] != 'r' && mode[0] != 'w' && mode[0] != 'a') {
174 PyErr_Format(PyExc_ValueError, "mode string must begin with "
175 "one of 'r', 'w', 'a' or 'U', not '%.200s'", mode);
Guido van Rossumd8faa362007-04-27 19:54:29 +0000176 return -1;
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000177 }
178
179 return 0;
180}
181
Tim Peters59c9a642001-09-13 05:38:56 +0000182static PyObject *
183open_the_file(PyFileObject *f, char *name, char *mode)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000184{
Thomas Wouters477c8d52006-05-27 19:21:47 +0000185 char *newmode;
Tim Peters59c9a642001-09-13 05:38:56 +0000186 assert(f != NULL);
187 assert(PyFile_Check(f));
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000188#ifdef MS_WINDOWS
189 /* windows ignores the passed name in order to support Unicode */
190 assert(f->f_name != NULL);
191#else
Tim Peters59c9a642001-09-13 05:38:56 +0000192 assert(name != NULL);
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000193#endif
Tim Peters59c9a642001-09-13 05:38:56 +0000194 assert(mode != NULL);
Tim Peters44410012001-09-14 03:26:08 +0000195 assert(f->f_fp == NULL);
Tim Peters59c9a642001-09-13 05:38:56 +0000196
Thomas Wouters477c8d52006-05-27 19:21:47 +0000197 /* probably need to replace 'U' by 'rb' */
198 newmode = PyMem_MALLOC(strlen(mode) + 3);
199 if (!newmode) {
200 PyErr_NoMemory();
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000201 return NULL;
Thomas Wouters477c8d52006-05-27 19:21:47 +0000202 }
203 strcpy(newmode, mode);
204
Guido van Rossumd8faa362007-04-27 19:54:29 +0000205 if (_PyFile_SanitizeMode(newmode)) {
Thomas Wouters477c8d52006-05-27 19:21:47 +0000206 f = NULL;
207 goto cleanup;
208 }
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000209
Tim Peters8fa45672001-09-13 21:01:29 +0000210 /* rexec.py can't stop a user from getting the file() constructor --
211 all they have to do is get *any* file object f, and then do
212 type(f). Here we prevent them from doing damage with it. */
213 if (PyEval_GetRestricted()) {
214 PyErr_SetString(PyExc_IOError,
Jeremy Hylton8b735422002-08-14 21:01:41 +0000215 "file() constructor not accessible in restricted mode");
Thomas Wouters477c8d52006-05-27 19:21:47 +0000216 f = NULL;
217 goto cleanup;
Tim Peters8fa45672001-09-13 21:01:29 +0000218 }
Tim Petersa27a1502001-11-09 20:59:14 +0000219 errno = 0;
Skip Montanaro51ffac62004-06-11 04:49:03 +0000220
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000221#ifdef MS_WINDOWS
Skip Montanaro51ffac62004-06-11 04:49:03 +0000222 if (PyUnicode_Check(f->f_name)) {
223 PyObject *wmode;
Thomas Wouters477c8d52006-05-27 19:21:47 +0000224 wmode = PyUnicode_DecodeASCII(newmode, strlen(newmode), NULL);
Skip Montanaro51ffac62004-06-11 04:49:03 +0000225 if (f->f_name && wmode) {
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000226 Py_BEGIN_ALLOW_THREADS
Skip Montanaro51ffac62004-06-11 04:49:03 +0000227 /* PyUnicode_AS_UNICODE OK without thread
228 lock as it is a simple dereference. */
229 f->f_fp = _wfopen(PyUnicode_AS_UNICODE(f->f_name),
230 PyUnicode_AS_UNICODE(wmode));
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000231 Py_END_ALLOW_THREADS
232 }
Skip Montanaro51ffac62004-06-11 04:49:03 +0000233 Py_XDECREF(wmode);
Guido van Rossumff4949e1992-08-05 19:58:53 +0000234 }
Skip Montanaro51ffac62004-06-11 04:49:03 +0000235#endif
236 if (NULL == f->f_fp && NULL != name) {
237 Py_BEGIN_ALLOW_THREADS
Thomas Wouters477c8d52006-05-27 19:21:47 +0000238 f->f_fp = fopen(name, newmode);
Skip Montanaro51ffac62004-06-11 04:49:03 +0000239 Py_END_ALLOW_THREADS
240 }
241
Guido van Rossuma08095a1991-02-13 23:25:27 +0000242 if (f->f_fp == NULL) {
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000243#if defined _MSC_VER && (_MSC_VER < 1400 || !defined(__STDC_SECURE_LIB__))
Tim Peters2ea91112002-04-08 04:13:12 +0000244 /* MSVC 6 (Microsoft) leaves errno at 0 for bad mode strings,
245 * across all Windows flavors. When it sets EINVAL varies
246 * across Windows flavors, the exact conditions aren't
247 * documented, and the answer lies in the OS's implementation
248 * of Win32's CreateFile function (whose source is secret).
249 * Seems the best we can do is map EINVAL to ENOENT.
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000250 * Starting with Visual Studio .NET 2005, EINVAL is correctly
251 * set by our CRT error handler (set in exceptions.c.)
Tim Peters2ea91112002-04-08 04:13:12 +0000252 */
253 if (errno == 0) /* bad mode string */
254 errno = EINVAL;
255 else if (errno == EINVAL) /* unknown, but not a mode string */
256 errno = ENOENT;
257#endif
Jeremy Hylton41c83212001-11-09 16:17:24 +0000258 if (errno == EINVAL)
Tim Peters2ea91112002-04-08 04:13:12 +0000259 PyErr_Format(PyExc_IOError, "invalid mode: %s",
Jeremy Hylton41c83212001-11-09 16:17:24 +0000260 mode);
261 else
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000262 PyErr_SetFromErrnoWithFilenameObject(PyExc_IOError, f->f_name);
Tim Peters59c9a642001-09-13 05:38:56 +0000263 f = NULL;
264 }
Tim Peters2ea91112002-04-08 04:13:12 +0000265 if (f != NULL)
Neil Schemenauered19b882002-03-23 02:06:50 +0000266 f = dircheck(f);
Thomas Wouters477c8d52006-05-27 19:21:47 +0000267
268cleanup:
269 PyMem_FREE(newmode);
270
Tim Peters59c9a642001-09-13 05:38:56 +0000271 return (PyObject *)f;
272}
273
274PyObject *
275PyFile_FromFile(FILE *fp, char *name, char *mode, int (*close)(FILE *))
276{
Tim Peters44410012001-09-14 03:26:08 +0000277 PyFileObject *f = (PyFileObject *)PyFile_Type.tp_new(&PyFile_Type,
278 NULL, NULL);
Tim Peters59c9a642001-09-13 05:38:56 +0000279 if (f != NULL) {
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000280 PyObject *o_name = PyString_FromString(name);
281 if (o_name == NULL)
282 return NULL;
Nicholas Bastinabce8a62004-03-21 20:24:07 +0000283 if (fill_file_fields(f, fp, o_name, mode, close) == NULL) {
Tim Peters59c9a642001-09-13 05:38:56 +0000284 Py_DECREF(f);
285 f = NULL;
286 }
Nicholas Bastinabce8a62004-03-21 20:24:07 +0000287 Py_DECREF(o_name);
Tim Peters59c9a642001-09-13 05:38:56 +0000288 }
289 return (PyObject *) f;
290}
291
292PyObject *
293PyFile_FromString(char *name, char *mode)
294{
295 extern int fclose(FILE *);
296 PyFileObject *f;
297
298 f = (PyFileObject *)PyFile_FromFile((FILE *)NULL, name, mode, fclose);
299 if (f != NULL) {
300 if (open_the_file(f, name, mode) == NULL) {
301 Py_DECREF(f);
302 f = NULL;
303 }
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000304 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000305 return (PyObject *)f;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000306}
307
Guido van Rossumb6775db1994-08-01 11:34:53 +0000308void
Fred Drakefd99de62000-07-09 05:02:18 +0000309PyFile_SetBufSize(PyObject *f, int bufsize)
Guido van Rossumb6775db1994-08-01 11:34:53 +0000310{
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000311 PyFileObject *file = (PyFileObject *)f;
Guido van Rossumb6775db1994-08-01 11:34:53 +0000312 if (bufsize >= 0) {
Guido van Rossumb6775db1994-08-01 11:34:53 +0000313 int type;
314 switch (bufsize) {
315 case 0:
316 type = _IONBF;
317 break;
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000318#ifdef HAVE_SETVBUF
Guido van Rossumb6775db1994-08-01 11:34:53 +0000319 case 1:
320 type = _IOLBF;
321 bufsize = BUFSIZ;
322 break;
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000323#endif
Guido van Rossumb6775db1994-08-01 11:34:53 +0000324 default:
325 type = _IOFBF;
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000326#ifndef HAVE_SETVBUF
327 bufsize = BUFSIZ;
328#endif
329 break;
Guido van Rossumb6775db1994-08-01 11:34:53 +0000330 }
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000331 fflush(file->f_fp);
332 if (type == _IONBF) {
333 PyMem_Free(file->f_setbuf);
334 file->f_setbuf = NULL;
335 } else {
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000336 file->f_setbuf = (char *)PyMem_Realloc(file->f_setbuf,
337 bufsize);
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000338 }
339#ifdef HAVE_SETVBUF
340 setvbuf(file->f_fp, file->f_setbuf, type, bufsize);
Guido van Rossumf8b4de01998-03-06 15:32:40 +0000341#else /* !HAVE_SETVBUF */
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000342 setbuf(file->f_fp, file->f_setbuf);
Guido van Rossumf8b4de01998-03-06 15:32:40 +0000343#endif /* !HAVE_SETVBUF */
Guido van Rossumb6775db1994-08-01 11:34:53 +0000344 }
345}
346
Martin v. Löwis5467d4c2003-05-10 07:10:12 +0000347/* Set the encoding used to output Unicode strings.
348 Returh 1 on success, 0 on failure. */
349
350int
351PyFile_SetEncoding(PyObject *f, const char *enc)
352{
353 PyFileObject *file = (PyFileObject*)f;
354 PyObject *str = PyString_FromString(enc);
Thomas Woutersb2137042007-02-01 18:02:27 +0000355
356 assert(PyFile_Check(f));
Martin v. Löwis5467d4c2003-05-10 07:10:12 +0000357 if (!str)
358 return 0;
359 Py_DECREF(file->f_encoding);
360 file->f_encoding = str;
361 return 1;
362}
363
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000364static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000365err_closed(void)
Guido van Rossumd7297e61992-07-06 14:19:26 +0000366{
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000367 PyErr_SetString(PyExc_ValueError, "I/O operation on closed file");
Guido van Rossumd7297e61992-07-06 14:19:26 +0000368 return NULL;
369}
370
Thomas Woutersc45251a2006-02-12 11:53:32 +0000371/* Refuse regular file I/O if there's data in the iteration-buffer.
372 * Mixing them would cause data to arrive out of order, as the read*
373 * methods don't use the iteration buffer. */
374static PyObject *
375err_iterbuffered(void)
376{
377 PyErr_SetString(PyExc_ValueError,
378 "Mixing iteration and read methods would lose data");
379 return NULL;
380}
381
Neal Norwitzd8b995f2002-08-06 21:50:54 +0000382static void drop_readahead(PyFileObject *);
Guido van Rossum7a6e9592002-08-06 15:55:28 +0000383
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000384/* Methods */
385
386static void
Fred Drakefd99de62000-07-09 05:02:18 +0000387file_dealloc(PyFileObject *f)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000388{
Peter Astrandf8e74b12004-11-07 14:15:28 +0000389 int sts = 0;
Raymond Hettingercb87bc82004-05-31 00:35:52 +0000390 if (f->weakreflist != NULL)
391 PyObject_ClearWeakRefs((PyObject *) f);
Guido van Rossumff4949e1992-08-05 19:58:53 +0000392 if (f->f_fp != NULL && f->f_close != NULL) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000393 Py_BEGIN_ALLOW_THREADS
Peter Astrandf8e74b12004-11-07 14:15:28 +0000394 sts = (*f->f_close)(f->f_fp);
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000395 Py_END_ALLOW_THREADS
Peter Astrandf8e74b12004-11-07 14:15:28 +0000396 if (sts == EOF)
397#ifdef HAVE_STRERROR
398 PySys_WriteStderr("close failed: [Errno %d] %s\n", errno, strerror(errno));
399#else
400 PySys_WriteStderr("close failed: [Errno %d]\n", errno);
401#endif
Guido van Rossumff4949e1992-08-05 19:58:53 +0000402 }
Andrew MacIntyre4e10ed32004-04-04 07:01:35 +0000403 PyMem_Free(f->f_setbuf);
Tim Peters44410012001-09-14 03:26:08 +0000404 Py_XDECREF(f->f_name);
405 Py_XDECREF(f->f_mode);
Martin v. Löwis5467d4c2003-05-10 07:10:12 +0000406 Py_XDECREF(f->f_encoding);
Guido van Rossum7a6e9592002-08-06 15:55:28 +0000407 drop_readahead(f);
Guido van Rossum9475a232001-10-05 20:51:39 +0000408 f->ob_type->tp_free((PyObject *)f);
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000409}
410
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000411static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000412file_repr(PyFileObject *f)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000413{
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000414 if (PyUnicode_Check(f->f_name)) {
Martin v. Löwis0073f2e2002-11-21 23:52:35 +0000415#ifdef Py_USING_UNICODE
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000416 PyObject *ret = NULL;
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000417 PyObject *name = PyUnicode_AsUnicodeEscapeString(f->f_name);
418 const char *name_str = name ? PyString_AsString(name) : "?";
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000419 ret = PyString_FromFormat("<%s file u'%s', mode '%s' at %p>",
420 f->f_fp == NULL ? "closed" : "open",
Thomas Wouters0e3f5912006-08-11 14:57:12 +0000421 name_str,
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000422 PyString_AsString(f->f_mode),
423 f);
424 Py_XDECREF(name);
425 return ret;
Martin v. Löwis0073f2e2002-11-21 23:52:35 +0000426#endif
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000427 } else {
428 return PyString_FromFormat("<%s file '%s', mode '%s' at %p>",
Barry Warsaw7ce36942001-08-24 18:34:26 +0000429 f->f_fp == NULL ? "closed" : "open",
430 PyString_AsString(f->f_name),
431 PyString_AsString(f->f_mode),
432 f);
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000433 }
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000434}
435
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000436static PyObject *
Martin v. Löwise3eb1f22001-08-16 13:15:00 +0000437file_close(PyFileObject *f)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000438{
Guido van Rossuma1ab7fa1991-06-04 19:37:39 +0000439 int sts = 0;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000440 if (f->f_fp != NULL) {
Guido van Rossumff4949e1992-08-05 19:58:53 +0000441 if (f->f_close != NULL) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000442 Py_BEGIN_ALLOW_THREADS
Guido van Rossumff4949e1992-08-05 19:58:53 +0000443 errno = 0;
Guido van Rossuma1ab7fa1991-06-04 19:37:39 +0000444 sts = (*f->f_close)(f->f_fp);
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000445 Py_END_ALLOW_THREADS
Guido van Rossumff4949e1992-08-05 19:58:53 +0000446 }
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000447 f->f_fp = NULL;
448 }
Martin v. Löwis7bbcde72003-09-07 20:42:29 +0000449 PyMem_Free(f->f_setbuf);
Andrew MacIntyre4e10ed32004-04-04 07:01:35 +0000450 f->f_setbuf = NULL;
Guido van Rossumfebd5511992-03-04 16:39:24 +0000451 if (sts == EOF)
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000452 return PyErr_SetFromErrno(PyExc_IOError);
Guido van Rossuma1ab7fa1991-06-04 19:37:39 +0000453 if (sts != 0)
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000454 return PyInt_FromLong((long)sts);
455 Py_INCREF(Py_None);
456 return Py_None;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000457}
458
Trent Mickf29f47b2000-08-11 19:02:59 +0000459
Guido van Rossumb8552162001-09-05 14:58:11 +0000460/* Our very own off_t-like type, 64-bit if possible */
461#if !defined(HAVE_LARGEFILE_SUPPORT)
462typedef off_t Py_off_t;
463#elif SIZEOF_OFF_T >= 8
464typedef off_t Py_off_t;
465#elif SIZEOF_FPOS_T >= 8
Guido van Rossum4f53da02001-03-01 18:26:53 +0000466typedef fpos_t Py_off_t;
467#else
Guido van Rossumb8552162001-09-05 14:58:11 +0000468#error "Large file support, but neither off_t nor fpos_t is large enough."
Guido van Rossum4f53da02001-03-01 18:26:53 +0000469#endif
470
471
Trent Mickf29f47b2000-08-11 19:02:59 +0000472/* a portable fseek() function
473 return 0 on success, non-zero on failure (with errno set) */
Guido van Rossumf68d8e52001-04-14 17:55:09 +0000474static int
Guido van Rossum4f53da02001-03-01 18:26:53 +0000475_portable_fseek(FILE *fp, Py_off_t offset, int whence)
Trent Mickf29f47b2000-08-11 19:02:59 +0000476{
Guido van Rossumb8552162001-09-05 14:58:11 +0000477#if !defined(HAVE_LARGEFILE_SUPPORT)
478 return fseek(fp, offset, whence);
479#elif defined(HAVE_FSEEKO) && SIZEOF_OFF_T >= 8
Trent Mickf29f47b2000-08-11 19:02:59 +0000480 return fseeko(fp, offset, whence);
481#elif defined(HAVE_FSEEK64)
482 return fseek64(fp, offset, whence);
Fred Drakedb810ac2000-10-06 20:42:33 +0000483#elif defined(__BEOS__)
484 return _fseek(fp, offset, whence);
Guido van Rossumb8552162001-09-05 14:58:11 +0000485#elif SIZEOF_FPOS_T >= 8
Guido van Rossume54e0be2001-01-16 20:53:31 +0000486 /* lacking a 64-bit capable fseek(), use a 64-bit capable fsetpos()
487 and fgetpos() to implement fseek()*/
Trent Mickf29f47b2000-08-11 19:02:59 +0000488 fpos_t pos;
489 switch (whence) {
Guido van Rossume54e0be2001-01-16 20:53:31 +0000490 case SEEK_END:
Guido van Rossum8b4e43e2001-09-10 20:43:35 +0000491#ifdef MS_WINDOWS
492 fflush(fp);
493 if (_lseeki64(fileno(fp), 0, 2) == -1)
494 return -1;
495#else
Guido van Rossume54e0be2001-01-16 20:53:31 +0000496 if (fseek(fp, 0, SEEK_END) != 0)
497 return -1;
Guido van Rossum8b4e43e2001-09-10 20:43:35 +0000498#endif
Guido van Rossume54e0be2001-01-16 20:53:31 +0000499 /* fall through */
500 case SEEK_CUR:
501 if (fgetpos(fp, &pos) != 0)
502 return -1;
503 offset += pos;
504 break;
505 /* case SEEK_SET: break; */
Trent Mickf29f47b2000-08-11 19:02:59 +0000506 }
507 return fsetpos(fp, &offset);
508#else
Guido van Rossumb8552162001-09-05 14:58:11 +0000509#error "Large file support, but no way to fseek."
Trent Mickf29f47b2000-08-11 19:02:59 +0000510#endif
511}
512
513
514/* a portable ftell() function
515 Return -1 on failure with errno set appropriately, current file
516 position on success */
Guido van Rossumf68d8e52001-04-14 17:55:09 +0000517static Py_off_t
Fred Drake8ce159a2000-08-31 05:18:54 +0000518_portable_ftell(FILE* fp)
Trent Mickf29f47b2000-08-11 19:02:59 +0000519{
Guido van Rossumb8552162001-09-05 14:58:11 +0000520#if !defined(HAVE_LARGEFILE_SUPPORT)
521 return ftell(fp);
522#elif defined(HAVE_FTELLO) && SIZEOF_OFF_T >= 8
523 return ftello(fp);
524#elif defined(HAVE_FTELL64)
525 return ftell64(fp);
526#elif SIZEOF_FPOS_T >= 8
Trent Mickf29f47b2000-08-11 19:02:59 +0000527 fpos_t pos;
528 if (fgetpos(fp, &pos) != 0)
529 return -1;
530 return pos;
531#else
Guido van Rossumb8552162001-09-05 14:58:11 +0000532#error "Large file support, but no way to ftell."
Trent Mickf29f47b2000-08-11 19:02:59 +0000533#endif
534}
535
536
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000537static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000538file_seek(PyFileObject *f, PyObject *args)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000539{
Guido van Rossumd7297e61992-07-06 14:19:26 +0000540 int whence;
Guido van Rossumff4949e1992-08-05 19:58:53 +0000541 int ret;
Guido van Rossum4f53da02001-03-01 18:26:53 +0000542 Py_off_t offset;
Thomas Wouters89f507f2006-12-13 04:49:30 +0000543 PyObject *offobj, *off_index;
Tim Peters86821b22001-01-07 21:19:34 +0000544
Guido van Rossumd7297e61992-07-06 14:19:26 +0000545 if (f->f_fp == NULL)
546 return err_closed();
Guido van Rossum7a6e9592002-08-06 15:55:28 +0000547 drop_readahead(f);
Guido van Rossumd7297e61992-07-06 14:19:26 +0000548 whence = 0;
Guido van Rossum43713e52000-02-29 13:59:29 +0000549 if (!PyArg_ParseTuple(args, "O|i:seek", &offobj, &whence))
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000550 return NULL;
Thomas Wouters89f507f2006-12-13 04:49:30 +0000551 off_index = PyNumber_Index(offobj);
552 if (!off_index) {
553 if (!PyFloat_Check(offobj))
554 return NULL;
555 /* Deprecated in 2.6 */
556 PyErr_Clear();
557 if (PyErr_Warn(PyExc_DeprecationWarning,
558 "integer argument expected, got float"))
559 return NULL;
560 off_index = offobj;
561 Py_INCREF(offobj);
562 }
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000563#if !defined(HAVE_LARGEFILE_SUPPORT)
Thomas Wouters89f507f2006-12-13 04:49:30 +0000564 offset = PyInt_AsLong(off_index);
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000565#else
Thomas Wouters89f507f2006-12-13 04:49:30 +0000566 offset = PyLong_Check(off_index) ?
567 PyLong_AsLongLong(off_index) : PyInt_AsLong(off_index);
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000568#endif
Thomas Wouters89f507f2006-12-13 04:49:30 +0000569 Py_DECREF(off_index);
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000570 if (PyErr_Occurred())
Guido van Rossum88303191999-01-04 17:22:18 +0000571 return NULL;
Tim Peters86821b22001-01-07 21:19:34 +0000572
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000573 Py_BEGIN_ALLOW_THREADS
Guido van Rossumce5ba841991-03-06 13:06:18 +0000574 errno = 0;
Trent Mickf29f47b2000-08-11 19:02:59 +0000575 ret = _portable_fseek(f->f_fp, offset, whence);
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000576 Py_END_ALLOW_THREADS
Trent Mickf29f47b2000-08-11 19:02:59 +0000577
Guido van Rossumff4949e1992-08-05 19:58:53 +0000578 if (ret != 0) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000579 PyErr_SetFromErrno(PyExc_IOError);
Guido van Rossumfebd5511992-03-04 16:39:24 +0000580 clearerr(f->f_fp);
581 return NULL;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000582 }
Jack Jansen7b8c7542002-04-14 20:12:41 +0000583 f->f_skipnextlf = 0;
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000584 Py_INCREF(Py_None);
585 return Py_None;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000586}
587
Trent Mickf29f47b2000-08-11 19:02:59 +0000588
Guido van Rossumd7047b31995-01-02 19:07:15 +0000589#ifdef HAVE_FTRUNCATE
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000590static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000591file_truncate(PyFileObject *f, PyObject *args)
Guido van Rossumd7047b31995-01-02 19:07:15 +0000592{
Guido van Rossum4f53da02001-03-01 18:26:53 +0000593 Py_off_t newsize;
Tim Petersf1827cf2003-09-07 03:30:18 +0000594 PyObject *newsizeobj = NULL;
595 Py_off_t initialpos;
596 int ret;
Tim Peters86821b22001-01-07 21:19:34 +0000597
Guido van Rossumd7047b31995-01-02 19:07:15 +0000598 if (f->f_fp == NULL)
599 return err_closed();
Raymond Hettingerea3fdf42002-12-29 16:33:45 +0000600 if (!PyArg_UnpackTuple(args, "truncate", 0, 1, &newsizeobj))
Guido van Rossum88303191999-01-04 17:22:18 +0000601 return NULL;
Tim Petersfb05db22002-03-11 00:24:00 +0000602
Tim Petersf1827cf2003-09-07 03:30:18 +0000603 /* Get current file position. If the file happens to be open for
604 * update and the last operation was an input operation, C doesn't
605 * define what the later fflush() will do, but we promise truncate()
606 * won't change the current position (and fflush() *does* change it
607 * then at least on Windows). The easiest thing is to capture
608 * current pos now and seek back to it at the end.
609 */
610 Py_BEGIN_ALLOW_THREADS
611 errno = 0;
612 initialpos = _portable_ftell(f->f_fp);
613 Py_END_ALLOW_THREADS
614 if (initialpos == -1)
615 goto onioerror;
616
Tim Petersfb05db22002-03-11 00:24:00 +0000617 /* Set newsize to current postion if newsizeobj NULL, else to the
Tim Petersf1827cf2003-09-07 03:30:18 +0000618 * specified value.
619 */
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000620 if (newsizeobj != NULL) {
621#if !defined(HAVE_LARGEFILE_SUPPORT)
622 newsize = PyInt_AsLong(newsizeobj);
623#else
624 newsize = PyLong_Check(newsizeobj) ?
625 PyLong_AsLongLong(newsizeobj) :
626 PyInt_AsLong(newsizeobj);
627#endif
628 if (PyErr_Occurred())
629 return NULL;
Tim Petersfb05db22002-03-11 00:24:00 +0000630 }
Tim Petersf1827cf2003-09-07 03:30:18 +0000631 else /* default to current position */
632 newsize = initialpos;
Tim Petersfb05db22002-03-11 00:24:00 +0000633
Tim Petersf1827cf2003-09-07 03:30:18 +0000634 /* Flush the stream. We're mixing stream-level I/O with lower-level
635 * I/O, and a flush may be necessary to synch both platform views
636 * of the current file state.
637 */
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000638 Py_BEGIN_ALLOW_THREADS
Guido van Rossumd7047b31995-01-02 19:07:15 +0000639 errno = 0;
640 ret = fflush(f->f_fp);
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000641 Py_END_ALLOW_THREADS
Tim Petersfb05db22002-03-11 00:24:00 +0000642 if (ret != 0)
643 goto onioerror;
Trent Mickf29f47b2000-08-11 19:02:59 +0000644
Martin v. Löwis6238d2b2002-06-30 15:26:10 +0000645#ifdef MS_WINDOWS
Tim Petersfb05db22002-03-11 00:24:00 +0000646 /* MS _chsize doesn't work if newsize doesn't fit in 32 bits,
Tim Peters8f01b682002-03-12 03:04:44 +0000647 so don't even try using it. */
Tim Petersfb05db22002-03-11 00:24:00 +0000648 {
Tim Petersfb05db22002-03-11 00:24:00 +0000649 HANDLE hFile;
Tim Petersfb05db22002-03-11 00:24:00 +0000650
Tim Petersf1827cf2003-09-07 03:30:18 +0000651 /* Have to move current pos to desired endpoint on Windows. */
652 Py_BEGIN_ALLOW_THREADS
653 errno = 0;
654 ret = _portable_fseek(f->f_fp, newsize, SEEK_SET) != 0;
655 Py_END_ALLOW_THREADS
656 if (ret)
657 goto onioerror;
Tim Petersfb05db22002-03-11 00:24:00 +0000658
Tim Peters8f01b682002-03-12 03:04:44 +0000659 /* Truncate. Note that this may grow the file! */
660 Py_BEGIN_ALLOW_THREADS
661 errno = 0;
662 hFile = (HANDLE)_get_osfhandle(fileno(f->f_fp));
Tim Petersf1827cf2003-09-07 03:30:18 +0000663 ret = hFile == (HANDLE)-1;
664 if (ret == 0) {
665 ret = SetEndOfFile(hFile) == 0;
666 if (ret)
Tim Peters8f01b682002-03-12 03:04:44 +0000667 errno = EACCES;
668 }
669 Py_END_ALLOW_THREADS
Tim Petersf1827cf2003-09-07 03:30:18 +0000670 if (ret)
Tim Peters8f01b682002-03-12 03:04:44 +0000671 goto onioerror;
Guido van Rossumd7047b31995-01-02 19:07:15 +0000672 }
Trent Mickf29f47b2000-08-11 19:02:59 +0000673#else
674 Py_BEGIN_ALLOW_THREADS
675 errno = 0;
676 ret = ftruncate(fileno(f->f_fp), newsize);
677 Py_END_ALLOW_THREADS
Tim Petersf1827cf2003-09-07 03:30:18 +0000678 if (ret != 0)
679 goto onioerror;
Martin v. Löwis6238d2b2002-06-30 15:26:10 +0000680#endif /* !MS_WINDOWS */
Tim Peters86821b22001-01-07 21:19:34 +0000681
Tim Petersf1827cf2003-09-07 03:30:18 +0000682 /* Restore original file position. */
683 Py_BEGIN_ALLOW_THREADS
684 errno = 0;
685 ret = _portable_fseek(f->f_fp, initialpos, SEEK_SET) != 0;
686 Py_END_ALLOW_THREADS
687 if (ret)
688 goto onioerror;
689
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000690 Py_INCREF(Py_None);
691 return Py_None;
Trent Mickf29f47b2000-08-11 19:02:59 +0000692
693onioerror:
694 PyErr_SetFromErrno(PyExc_IOError);
695 clearerr(f->f_fp);
696 return NULL;
Guido van Rossumd7047b31995-01-02 19:07:15 +0000697}
698#endif /* HAVE_FTRUNCATE */
699
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000700static PyObject *
Martin v. Löwise3eb1f22001-08-16 13:15:00 +0000701file_tell(PyFileObject *f)
Guido van Rossumce5ba841991-03-06 13:06:18 +0000702{
Guido van Rossum4f53da02001-03-01 18:26:53 +0000703 Py_off_t pos;
Trent Mickf29f47b2000-08-11 19:02:59 +0000704
Guido van Rossumd7297e61992-07-06 14:19:26 +0000705 if (f->f_fp == NULL)
706 return err_closed();
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000707 Py_BEGIN_ALLOW_THREADS
Guido van Rossumce5ba841991-03-06 13:06:18 +0000708 errno = 0;
Trent Mickf29f47b2000-08-11 19:02:59 +0000709 pos = _portable_ftell(f->f_fp);
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000710 Py_END_ALLOW_THREADS
Trent Mickf29f47b2000-08-11 19:02:59 +0000711 if (pos == -1) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000712 PyErr_SetFromErrno(PyExc_IOError);
Guido van Rossumfebd5511992-03-04 16:39:24 +0000713 clearerr(f->f_fp);
714 return NULL;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000715 }
Jack Jansen7b8c7542002-04-14 20:12:41 +0000716 if (f->f_skipnextlf) {
717 int c;
718 c = GETC(f->f_fp);
719 if (c == '\n') {
720 pos++;
721 f->f_skipnextlf = 0;
722 } else if (c != EOF) ungetc(c, f->f_fp);
723 }
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000724#if !defined(HAVE_LARGEFILE_SUPPORT)
Trent Mickf29f47b2000-08-11 19:02:59 +0000725 return PyInt_FromLong(pos);
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000726#else
Trent Mickf29f47b2000-08-11 19:02:59 +0000727 return PyLong_FromLongLong(pos);
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000728#endif
Guido van Rossumce5ba841991-03-06 13:06:18 +0000729}
730
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000731static PyObject *
Martin v. Löwise3eb1f22001-08-16 13:15:00 +0000732file_fileno(PyFileObject *f)
Guido van Rossumed233a51992-06-23 09:07:03 +0000733{
Guido van Rossumd7297e61992-07-06 14:19:26 +0000734 if (f->f_fp == NULL)
735 return err_closed();
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000736 return PyInt_FromLong((long) fileno(f->f_fp));
Guido van Rossumed233a51992-06-23 09:07:03 +0000737}
738
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000739static PyObject *
Martin v. Löwise3eb1f22001-08-16 13:15:00 +0000740file_flush(PyFileObject *f)
Guido van Rossumce5ba841991-03-06 13:06:18 +0000741{
Guido van Rossumff4949e1992-08-05 19:58:53 +0000742 int res;
Tim Peters86821b22001-01-07 21:19:34 +0000743
Guido van Rossumd7297e61992-07-06 14:19:26 +0000744 if (f->f_fp == NULL)
745 return err_closed();
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000746 Py_BEGIN_ALLOW_THREADS
Guido van Rossumce5ba841991-03-06 13:06:18 +0000747 errno = 0;
Guido van Rossumff4949e1992-08-05 19:58:53 +0000748 res = fflush(f->f_fp);
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000749 Py_END_ALLOW_THREADS
Guido van Rossumff4949e1992-08-05 19:58:53 +0000750 if (res != 0) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000751 PyErr_SetFromErrno(PyExc_IOError);
Guido van Rossumfebd5511992-03-04 16:39:24 +0000752 clearerr(f->f_fp);
753 return NULL;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000754 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000755 Py_INCREF(Py_None);
756 return Py_None;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000757}
758
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000759static PyObject *
Martin v. Löwise3eb1f22001-08-16 13:15:00 +0000760file_isatty(PyFileObject *f)
Guido van Rossuma1ab7fa1991-06-04 19:37:39 +0000761{
Guido van Rossumff4949e1992-08-05 19:58:53 +0000762 long res;
Guido van Rossumd7297e61992-07-06 14:19:26 +0000763 if (f->f_fp == NULL)
764 return err_closed();
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000765 Py_BEGIN_ALLOW_THREADS
Guido van Rossumff4949e1992-08-05 19:58:53 +0000766 res = isatty((int)fileno(f->f_fp));
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000767 Py_END_ALLOW_THREADS
Guido van Rossum7f7666f2002-04-07 06:28:00 +0000768 return PyBool_FromLong(res);
Guido van Rossuma1ab7fa1991-06-04 19:37:39 +0000769}
770
Guido van Rossumff7e83d1999-08-27 20:39:37 +0000771
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000772#if BUFSIZ < 8192
773#define SMALLCHUNK 8192
774#else
775#define SMALLCHUNK BUFSIZ
776#endif
777
Guido van Rossum3c259041999-01-14 19:00:14 +0000778#if SIZEOF_INT < 4
779#define BIGCHUNK (512 * 32)
780#else
781#define BIGCHUNK (512 * 1024)
782#endif
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000783
784static size_t
Fred Drakefd99de62000-07-09 05:02:18 +0000785new_buffersize(PyFileObject *f, size_t currentsize)
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000786{
787#ifdef HAVE_FSTAT
Fred Drake1bc8fab2001-07-19 21:49:38 +0000788 off_t pos, end;
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000789 struct stat st;
790 if (fstat(fileno(f->f_fp), &st) == 0) {
791 end = st.st_size;
Guido van Rossumcada2931998-12-11 20:44:56 +0000792 /* The following is not a bug: we really need to call lseek()
793 *and* ftell(). The reason is that some stdio libraries
794 mistakenly flush their buffer when ftell() is called and
795 the lseek() call it makes fails, thereby throwing away
796 data that cannot be recovered in any way. To avoid this,
797 we first test lseek(), and only call ftell() if lseek()
798 works. We can't use the lseek() value either, because we
799 need to take the amount of buffered data into account.
800 (Yet another reason why stdio stinks. :-) */
Guido van Rossum91aaa921998-05-05 22:21:35 +0000801 pos = lseek(fileno(f->f_fp), 0L, SEEK_CUR);
Jack Jansen2771b5b2001-10-10 22:03:27 +0000802 if (pos >= 0) {
Guido van Rossum91aaa921998-05-05 22:21:35 +0000803 pos = ftell(f->f_fp);
Jack Jansen2771b5b2001-10-10 22:03:27 +0000804 }
Guido van Rossumd30dc0a1998-04-27 19:01:08 +0000805 if (pos < 0)
806 clearerr(f->f_fp);
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000807 if (end > pos && pos >= 0)
Guido van Rossumcada2931998-12-11 20:44:56 +0000808 return currentsize + end - pos + 1;
Guido van Rossumdcb5e7f1998-03-03 22:36:10 +0000809 /* Add 1 so if the file were to grow we'd notice. */
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000810 }
811#endif
812 if (currentsize > SMALLCHUNK) {
813 /* Keep doubling until we reach BIGCHUNK;
814 then keep adding BIGCHUNK. */
815 if (currentsize <= BIGCHUNK)
816 return currentsize + currentsize;
817 else
818 return currentsize + BIGCHUNK;
819 }
820 return currentsize + SMALLCHUNK;
821}
822
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +0000823#if defined(EWOULDBLOCK) && defined(EAGAIN) && EWOULDBLOCK != EAGAIN
824#define BLOCKED_ERRNO(x) ((x) == EWOULDBLOCK || (x) == EAGAIN)
825#else
826#ifdef EWOULDBLOCK
827#define BLOCKED_ERRNO(x) ((x) == EWOULDBLOCK)
828#else
829#ifdef EAGAIN
830#define BLOCKED_ERRNO(x) ((x) == EAGAIN)
831#else
832#define BLOCKED_ERRNO(x) 0
833#endif
834#endif
835#endif
836
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000837static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000838file_read(PyFileObject *f, PyObject *args)
Guido van Rossumce5ba841991-03-06 13:06:18 +0000839{
Guido van Rossum789a1611997-05-10 22:33:55 +0000840 long bytesrequested = -1;
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000841 size_t bytesread, buffersize, chunksize;
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000842 PyObject *v;
Tim Peters86821b22001-01-07 21:19:34 +0000843
Guido van Rossumd7297e61992-07-06 14:19:26 +0000844 if (f->f_fp == NULL)
845 return err_closed();
Thomas Woutersc45251a2006-02-12 11:53:32 +0000846 /* refuse to mix with f.next() */
847 if (f->f_buf != NULL &&
848 (f->f_bufend - f->f_bufptr) > 0 &&
849 f->f_buf[0] != '\0')
850 return err_iterbuffered();
Guido van Rossum43713e52000-02-29 13:59:29 +0000851 if (!PyArg_ParseTuple(args, "|l:read", &bytesrequested))
Guido van Rossum789a1611997-05-10 22:33:55 +0000852 return NULL;
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000853 if (bytesrequested < 0)
Guido van Rossumff1ccbf1999-04-10 15:48:23 +0000854 buffersize = new_buffersize(f, (size_t)0);
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000855 else
856 buffersize = bytesrequested;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000857 if (buffersize > PY_SSIZE_T_MAX) {
Trent Mickf29f47b2000-08-11 19:02:59 +0000858 PyErr_SetString(PyExc_OverflowError,
Jeremy Hylton8b735422002-08-14 21:01:41 +0000859 "requested number of bytes is more than a Python string can hold");
Trent Mickf29f47b2000-08-11 19:02:59 +0000860 return NULL;
861 }
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000862 v = PyString_FromStringAndSize((char *)NULL, buffersize);
Guido van Rossum3f5da241990-12-20 15:06:42 +0000863 if (v == NULL)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000864 return NULL;
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000865 bytesread = 0;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000866 for (;;) {
Guido van Rossum6263d541997-05-10 22:07:25 +0000867 Py_BEGIN_ALLOW_THREADS
868 errno = 0;
Jack Jansen7b8c7542002-04-14 20:12:41 +0000869 chunksize = Py_UniversalNewlineFread(BUF(v) + bytesread,
Jeremy Hylton8b735422002-08-14 21:01:41 +0000870 buffersize - bytesread, f->f_fp, (PyObject *)f);
Guido van Rossum6263d541997-05-10 22:07:25 +0000871 Py_END_ALLOW_THREADS
872 if (chunksize == 0) {
873 if (!ferror(f->f_fp))
874 break;
Guido van Rossum6263d541997-05-10 22:07:25 +0000875 clearerr(f->f_fp);
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +0000876 /* When in non-blocking mode, data shouldn't
877 * be discarded if a blocking signal was
878 * received. That will also happen if
879 * chunksize != 0, but bytesread < buffersize. */
880 if (bytesread > 0 && BLOCKED_ERRNO(errno))
881 break;
882 PyErr_SetFromErrno(PyExc_IOError);
Guido van Rossum6263d541997-05-10 22:07:25 +0000883 Py_DECREF(v);
884 return NULL;
885 }
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000886 bytesread += chunksize;
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +0000887 if (bytesread < buffersize) {
888 clearerr(f->f_fp);
Guido van Rossumce5ba841991-03-06 13:06:18 +0000889 break;
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +0000890 }
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000891 if (bytesrequested < 0) {
Guido van Rossumcada2931998-12-11 20:44:56 +0000892 buffersize = new_buffersize(f, buffersize);
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000893 if (_PyString_Resize(&v, buffersize) < 0)
Guido van Rossumce5ba841991-03-06 13:06:18 +0000894 return NULL;
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +0000895 } else {
Gustavo Niemeyera080be82002-12-17 17:48:00 +0000896 /* Got what was requested. */
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +0000897 break;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000898 }
899 }
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000900 if (bytesread != buffersize)
901 _PyString_Resize(&v, bytesread);
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000902 return v;
903}
904
Guido van Rossumfdf95dd1997-05-05 22:15:02 +0000905static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000906file_readinto(PyFileObject *f, PyObject *args)
Guido van Rossumfdf95dd1997-05-05 22:15:02 +0000907{
908 char *ptr;
Martin v. Löwis18e16552006-02-15 17:27:45 +0000909 Py_ssize_t ntodo;
910 Py_ssize_t ndone, nnow;
Tim Peters86821b22001-01-07 21:19:34 +0000911
Guido van Rossumfdf95dd1997-05-05 22:15:02 +0000912 if (f->f_fp == NULL)
913 return err_closed();
Guido van Rossumd624f182006-04-24 13:47:05 +0000914 if (!f->f_binary) {
915 PyErr_SetString(PyExc_TypeError,
916 "readinto() requires binary mode");
917 return NULL;
918 }
Thomas Woutersc45251a2006-02-12 11:53:32 +0000919 /* refuse to mix with f.next() */
920 if (f->f_buf != NULL &&
921 (f->f_bufend - f->f_bufptr) > 0 &&
922 f->f_buf[0] != '\0')
923 return err_iterbuffered();
Neal Norwitz62f5a9d2002-04-01 00:09:00 +0000924 if (!PyArg_ParseTuple(args, "w#", &ptr, &ntodo))
Guido van Rossumfdf95dd1997-05-05 22:15:02 +0000925 return NULL;
926 ndone = 0;
Guido van Rossum6263d541997-05-10 22:07:25 +0000927 while (ntodo > 0) {
928 Py_BEGIN_ALLOW_THREADS
929 errno = 0;
Tim Petersf1827cf2003-09-07 03:30:18 +0000930 nnow = Py_UniversalNewlineFread(ptr+ndone, ntodo, f->f_fp,
Jeremy Hylton8b735422002-08-14 21:01:41 +0000931 (PyObject *)f);
Guido van Rossum6263d541997-05-10 22:07:25 +0000932 Py_END_ALLOW_THREADS
933 if (nnow == 0) {
934 if (!ferror(f->f_fp))
935 break;
Guido van Rossumfdf95dd1997-05-05 22:15:02 +0000936 PyErr_SetFromErrno(PyExc_IOError);
937 clearerr(f->f_fp);
938 return NULL;
939 }
Guido van Rossumfdf95dd1997-05-05 22:15:02 +0000940 ndone += nnow;
941 ntodo -= nnow;
942 }
Thomas Wouters89f507f2006-12-13 04:49:30 +0000943 return PyInt_FromSsize_t(ndone);
Guido van Rossumfdf95dd1997-05-05 22:15:02 +0000944}
945
Tim Peters86821b22001-01-07 21:19:34 +0000946/**************************************************************************
Tim Petersf29b64d2001-01-15 06:33:19 +0000947Routine to get next line using platform fgets().
Tim Peters86821b22001-01-07 21:19:34 +0000948
949Under MSVC 6:
950
Tim Peters1c733232001-01-08 04:02:07 +0000951+ MS threadsafe getc is very slow (multiple layers of function calls before+
952 after each character, to lock+unlock the stream).
953+ The stream-locking functions are MS-internal -- can't access them from user
954 code.
955+ There's nothing Tim could find in the MS C or platform SDK libraries that
956 can worm around this.
Tim Peters86821b22001-01-07 21:19:34 +0000957+ MS fgets locks/unlocks only once per line; it's the only hook we have.
958
959So we use fgets for speed(!), despite that it's painful.
960
961MS realloc is also slow.
962
Tim Petersf29b64d2001-01-15 06:33:19 +0000963Reports from other platforms on this method vs getc_unlocked (which MS doesn't
964have):
965 Linux a wash
966 Solaris a wash
967 Tru64 Unix getline_via_fgets significantly faster
Tim Peters86821b22001-01-07 21:19:34 +0000968
Tim Petersf29b64d2001-01-15 06:33:19 +0000969CAUTION: The C std isn't clear about this: in those cases where fgets
970writes something into the buffer, can it write into any position beyond the
971required trailing null byte? MSVC 6 fgets does not, and no platform is (yet)
972known on which it does; and it would be a strange way to code fgets. Still,
973getline_via_fgets may not work correctly if it does. The std test
974test_bufio.py should fail if platform fgets() routinely writes beyond the
975trailing null byte. #define DONT_USE_FGETS_IN_GETLINE to disable this code.
Tim Peters86821b22001-01-07 21:19:34 +0000976**************************************************************************/
977
Tim Petersf29b64d2001-01-15 06:33:19 +0000978/* Use this routine if told to, or by default on non-get_unlocked()
979 * platforms unless told not to. Yikes! Let's spell that out:
980 * On a platform with getc_unlocked():
981 * By default, use getc_unlocked().
982 * If you want to use fgets() instead, #define USE_FGETS_IN_GETLINE.
983 * On a platform without getc_unlocked():
984 * By default, use fgets().
985 * If you don't want to use fgets(), #define DONT_USE_FGETS_IN_GETLINE.
986 */
987#if !defined(USE_FGETS_IN_GETLINE) && !defined(HAVE_GETC_UNLOCKED)
988#define USE_FGETS_IN_GETLINE
Tim Peters86821b22001-01-07 21:19:34 +0000989#endif
990
Tim Petersf29b64d2001-01-15 06:33:19 +0000991#if defined(DONT_USE_FGETS_IN_GETLINE) && defined(USE_FGETS_IN_GETLINE)
992#undef USE_FGETS_IN_GETLINE
993#endif
994
995#ifdef USE_FGETS_IN_GETLINE
Tim Peters86821b22001-01-07 21:19:34 +0000996static PyObject*
Tim Petersf29b64d2001-01-15 06:33:19 +0000997getline_via_fgets(FILE *fp)
Tim Peters86821b22001-01-07 21:19:34 +0000998{
Tim Peters15b83852001-01-08 00:53:12 +0000999/* INITBUFSIZE is the maximum line length that lets us get away with the fast
Tim Peters142297a2001-01-15 10:36:56 +00001000 * no-realloc, one-fgets()-call path. Boosting it isn't free, because we have
1001 * to fill this much of the buffer with a known value in order to figure out
1002 * how much of the buffer fgets() overwrites. So if INITBUFSIZE is larger
1003 * than "most" lines, we waste time filling unused buffer slots. 100 is
1004 * surely adequate for most peoples' email archives, chewing over source code,
1005 * etc -- "regular old text files".
1006 * MAXBUFSIZE is the maximum line length that lets us get away with the less
1007 * fast (but still zippy) no-realloc, two-fgets()-call path. See above for
1008 * cautions about boosting that. 300 was chosen because the worst real-life
1009 * text-crunching job reported on Python-Dev was a mail-log crawler where over
1010 * half the lines were 254 chars.
Tim Peters15b83852001-01-08 00:53:12 +00001011 */
Tim Peters142297a2001-01-15 10:36:56 +00001012#define INITBUFSIZE 100
1013#define MAXBUFSIZE 300
Tim Peters142297a2001-01-15 10:36:56 +00001014 char* p; /* temp */
1015 char buf[MAXBUFSIZE];
Tim Peters86821b22001-01-07 21:19:34 +00001016 PyObject* v; /* the string object result */
Tim Peters86821b22001-01-07 21:19:34 +00001017 char* pvfree; /* address of next free slot */
1018 char* pvend; /* address one beyond last free slot */
Tim Peters142297a2001-01-15 10:36:56 +00001019 size_t nfree; /* # of free buffer slots; pvend-pvfree */
1020 size_t total_v_size; /* total # of slots in buffer */
Tim Petersddea2082002-03-23 10:03:50 +00001021 size_t increment; /* amount to increment the buffer */
Thomas Wouters89f507f2006-12-13 04:49:30 +00001022 size_t prev_v_size;
Tim Peters86821b22001-01-07 21:19:34 +00001023
Tim Peters15b83852001-01-08 00:53:12 +00001024 /* Optimize for normal case: avoid _PyString_Resize if at all
Tim Peters142297a2001-01-15 10:36:56 +00001025 * possible via first reading into stack buffer "buf".
Tim Peters15b83852001-01-08 00:53:12 +00001026 */
Tim Peters142297a2001-01-15 10:36:56 +00001027 total_v_size = INITBUFSIZE; /* start small and pray */
1028 pvfree = buf;
1029 for (;;) {
1030 Py_BEGIN_ALLOW_THREADS
1031 pvend = buf + total_v_size;
1032 nfree = pvend - pvfree;
1033 memset(pvfree, '\n', nfree);
Martin v. Löwis18e16552006-02-15 17:27:45 +00001034 assert(nfree < INT_MAX); /* Should be atmost MAXBUFSIZE */
1035 p = fgets(pvfree, (int)nfree, fp);
Tim Peters142297a2001-01-15 10:36:56 +00001036 Py_END_ALLOW_THREADS
Tim Peters15b83852001-01-08 00:53:12 +00001037
Tim Peters142297a2001-01-15 10:36:56 +00001038 if (p == NULL) {
1039 clearerr(fp);
1040 if (PyErr_CheckSignals())
1041 return NULL;
1042 v = PyString_FromStringAndSize(buf, pvfree - buf);
Tim Peters86821b22001-01-07 21:19:34 +00001043 return v;
1044 }
Tim Peters142297a2001-01-15 10:36:56 +00001045 /* fgets read *something* */
1046 p = memchr(pvfree, '\n', nfree);
1047 if (p != NULL) {
1048 /* Did the \n come from fgets or from us?
1049 * Since fgets stops at the first \n, and then writes
1050 * \0, if it's from fgets a \0 must be next. But if
1051 * that's so, it could not have come from us, since
1052 * the \n's we filled the buffer with have only more
1053 * \n's to the right.
1054 */
1055 if (p+1 < pvend && *(p+1) == '\0') {
1056 /* It's from fgets: we win! In particular,
1057 * we haven't done any mallocs yet, and can
1058 * build the final result on the first try.
1059 */
1060 ++p; /* include \n from fgets */
1061 }
1062 else {
1063 /* Must be from us: fgets didn't fill the
1064 * buffer and didn't find a newline, so it
1065 * must be the last and newline-free line of
1066 * the file.
1067 */
1068 assert(p > pvfree && *(p-1) == '\0');
1069 --p; /* don't include \0 from fgets */
1070 }
1071 v = PyString_FromStringAndSize(buf, p - buf);
1072 return v;
1073 }
1074 /* yuck: fgets overwrote all the newlines, i.e. the entire
1075 * buffer. So this line isn't over yet, or maybe it is but
1076 * we're exactly at EOF. If we haven't already, try using the
1077 * rest of the stack buffer.
Tim Peters86821b22001-01-07 21:19:34 +00001078 */
Tim Peters142297a2001-01-15 10:36:56 +00001079 assert(*(pvend-1) == '\0');
1080 if (pvfree == buf) {
1081 pvfree = pvend - 1; /* overwrite trailing null */
1082 total_v_size = MAXBUFSIZE;
1083 }
1084 else
1085 break;
Tim Peters86821b22001-01-07 21:19:34 +00001086 }
Tim Peters142297a2001-01-15 10:36:56 +00001087
1088 /* The stack buffer isn't big enough; malloc a string object and read
1089 * into its buffer.
Tim Peters15b83852001-01-08 00:53:12 +00001090 */
Tim Petersddea2082002-03-23 10:03:50 +00001091 total_v_size = MAXBUFSIZE << 1;
Tim Peters1c733232001-01-08 04:02:07 +00001092 v = PyString_FromStringAndSize((char*)NULL, (int)total_v_size);
Tim Peters15b83852001-01-08 00:53:12 +00001093 if (v == NULL)
1094 return v;
1095 /* copy over everything except the last null byte */
Tim Peters142297a2001-01-15 10:36:56 +00001096 memcpy(BUF(v), buf, MAXBUFSIZE-1);
1097 pvfree = BUF(v) + MAXBUFSIZE - 1;
Tim Peters86821b22001-01-07 21:19:34 +00001098
1099 /* Keep reading stuff into v; if it ever ends successfully, break
Tim Peters15b83852001-01-08 00:53:12 +00001100 * after setting p one beyond the end of the line. The code here is
1101 * very much like the code above, except reads into v's buffer; see
1102 * the code above for detailed comments about the logic.
Tim Peters86821b22001-01-07 21:19:34 +00001103 */
1104 for (;;) {
Tim Peters86821b22001-01-07 21:19:34 +00001105 Py_BEGIN_ALLOW_THREADS
1106 pvend = BUF(v) + total_v_size;
1107 nfree = pvend - pvfree;
1108 memset(pvfree, '\n', nfree);
Martin v. Löwis18e16552006-02-15 17:27:45 +00001109 assert(nfree < INT_MAX);
1110 p = fgets(pvfree, (int)nfree, fp);
Tim Peters86821b22001-01-07 21:19:34 +00001111 Py_END_ALLOW_THREADS
1112
1113 if (p == NULL) {
1114 clearerr(fp);
1115 if (PyErr_CheckSignals()) {
1116 Py_DECREF(v);
1117 return NULL;
1118 }
1119 p = pvfree;
1120 break;
1121 }
Tim Peters86821b22001-01-07 21:19:34 +00001122 p = memchr(pvfree, '\n', nfree);
1123 if (p != NULL) {
1124 if (p+1 < pvend && *(p+1) == '\0') {
1125 /* \n came from fgets */
1126 ++p;
1127 break;
1128 }
1129 /* \n came from us; last line of file, no newline */
1130 assert(p > pvfree && *(p-1) == '\0');
1131 --p;
1132 break;
1133 }
1134 /* expand buffer and try again */
1135 assert(*(pvend-1) == '\0');
Tim Petersddea2082002-03-23 10:03:50 +00001136 increment = total_v_size >> 2; /* mild exponential growth */
Thomas Wouters89f507f2006-12-13 04:49:30 +00001137 prev_v_size = total_v_size;
Tim Petersddea2082002-03-23 10:03:50 +00001138 total_v_size += increment;
Thomas Wouters89f507f2006-12-13 04:49:30 +00001139 /* check for overflow */
1140 if (total_v_size <= prev_v_size ||
1141 total_v_size > PY_SSIZE_T_MAX) {
Tim Peters86821b22001-01-07 21:19:34 +00001142 PyErr_SetString(PyExc_OverflowError,
1143 "line is longer than a Python string can hold");
1144 Py_DECREF(v);
1145 return NULL;
1146 }
1147 if (_PyString_Resize(&v, (int)total_v_size) < 0)
1148 return NULL;
1149 /* overwrite the trailing null byte */
Thomas Wouters89f507f2006-12-13 04:49:30 +00001150 pvfree = BUF(v) + (prev_v_size - 1);
Tim Peters86821b22001-01-07 21:19:34 +00001151 }
1152 if (BUF(v) + total_v_size != p)
1153 _PyString_Resize(&v, p - BUF(v));
1154 return v;
1155#undef INITBUFSIZE
Tim Peters142297a2001-01-15 10:36:56 +00001156#undef MAXBUFSIZE
Tim Peters86821b22001-01-07 21:19:34 +00001157}
Tim Petersf29b64d2001-01-15 06:33:19 +00001158#endif /* ifdef USE_FGETS_IN_GETLINE */
Guido van Rossumfdf95dd1997-05-05 22:15:02 +00001159
Guido van Rossum0bd24411991-04-04 15:21:57 +00001160/* Internal routine to get a line.
1161 Size argument interpretation:
1162 > 0: max length;
Guido van Rossum86282062001-01-08 01:26:47 +00001163 <= 0: read arbitrary line
Guido van Rossumce5ba841991-03-06 13:06:18 +00001164*/
1165
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001166static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001167get_line(PyFileObject *f, int n)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001168{
Guido van Rossum1187aa42001-01-05 14:43:05 +00001169 FILE *fp = f->f_fp;
1170 int c;
Andrew M. Kuchling4b2b4452000-11-29 02:53:22 +00001171 char *buf, *end;
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001172 size_t total_v_size; /* total # of slots in buffer */
1173 size_t used_v_size; /* # used slots in buffer */
1174 size_t increment; /* amount to increment the buffer */
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001175 PyObject *v;
Jack Jansen7b8c7542002-04-14 20:12:41 +00001176 int newlinetypes = f->f_newlinetypes;
1177 int skipnextlf = f->f_skipnextlf;
1178 int univ_newline = f->f_univ_newline;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001179
Jack Jansen7b8c7542002-04-14 20:12:41 +00001180#if defined(USE_FGETS_IN_GETLINE)
Jack Jansen7b8c7542002-04-14 20:12:41 +00001181 if (n <= 0 && !univ_newline )
Tim Petersf29b64d2001-01-15 06:33:19 +00001182 return getline_via_fgets(fp);
Tim Peters86821b22001-01-07 21:19:34 +00001183#endif
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001184 total_v_size = n > 0 ? n : 100;
1185 v = PyString_FromStringAndSize((char *)NULL, total_v_size);
Guido van Rossum3f5da241990-12-20 15:06:42 +00001186 if (v == NULL)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001187 return NULL;
Guido van Rossumce5ba841991-03-06 13:06:18 +00001188 buf = BUF(v);
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001189 end = buf + total_v_size;
Guido van Rossum1984f1e1992-08-04 12:41:02 +00001190
Guido van Rossumce5ba841991-03-06 13:06:18 +00001191 for (;;) {
Guido van Rossum1187aa42001-01-05 14:43:05 +00001192 Py_BEGIN_ALLOW_THREADS
1193 FLOCKFILE(fp);
Jack Jansen7b8c7542002-04-14 20:12:41 +00001194 if (univ_newline) {
1195 c = 'x'; /* Shut up gcc warning */
1196 while ( buf != end && (c = GETC(fp)) != EOF ) {
1197 if (skipnextlf ) {
1198 skipnextlf = 0;
1199 if (c == '\n') {
Tim Petersf1827cf2003-09-07 03:30:18 +00001200 /* Seeing a \n here with
1201 * skipnextlf true means we
Jeremy Hylton8b735422002-08-14 21:01:41 +00001202 * saw a \r before.
1203 */
Jack Jansen7b8c7542002-04-14 20:12:41 +00001204 newlinetypes |= NEWLINE_CRLF;
1205 c = GETC(fp);
1206 if (c == EOF) break;
1207 } else {
1208 newlinetypes |= NEWLINE_CR;
1209 }
1210 }
1211 if (c == '\r') {
1212 skipnextlf = 1;
1213 c = '\n';
1214 } else if ( c == '\n')
1215 newlinetypes |= NEWLINE_LF;
1216 *buf++ = c;
1217 if (c == '\n') break;
1218 }
1219 if ( c == EOF && skipnextlf )
1220 newlinetypes |= NEWLINE_CR;
1221 } else /* If not universal newlines use the normal loop */
Guido van Rossum1187aa42001-01-05 14:43:05 +00001222 while ((c = GETC(fp)) != EOF &&
1223 (*buf++ = c) != '\n' &&
1224 buf != end)
1225 ;
1226 FUNLOCKFILE(fp);
1227 Py_END_ALLOW_THREADS
Jack Jansen7b8c7542002-04-14 20:12:41 +00001228 f->f_newlinetypes = newlinetypes;
1229 f->f_skipnextlf = skipnextlf;
Guido van Rossum1187aa42001-01-05 14:43:05 +00001230 if (c == '\n')
1231 break;
1232 if (c == EOF) {
Guido van Rossum29206bc2001-08-09 18:14:59 +00001233 if (ferror(fp)) {
1234 PyErr_SetFromErrno(PyExc_IOError);
1235 clearerr(fp);
1236 Py_DECREF(v);
1237 return NULL;
1238 }
Guido van Rossum76ad8ed1991-06-03 10:54:55 +00001239 clearerr(fp);
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001240 if (PyErr_CheckSignals()) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001241 Py_DECREF(v);
Guido van Rossum0bd24411991-04-04 15:21:57 +00001242 return NULL;
1243 }
Guido van Rossumce5ba841991-03-06 13:06:18 +00001244 break;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001245 }
Guido van Rossum1187aa42001-01-05 14:43:05 +00001246 /* Must be because buf == end */
1247 if (n > 0)
Guido van Rossum0bd24411991-04-04 15:21:57 +00001248 break;
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001249 used_v_size = total_v_size;
1250 increment = total_v_size >> 2; /* mild exponential growth */
1251 total_v_size += increment;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001252 if (total_v_size > PY_SSIZE_T_MAX) {
Guido van Rossum1187aa42001-01-05 14:43:05 +00001253 PyErr_SetString(PyExc_OverflowError,
1254 "line is longer than a Python string can hold");
Tim Peters86821b22001-01-07 21:19:34 +00001255 Py_DECREF(v);
Guido van Rossum1187aa42001-01-05 14:43:05 +00001256 return NULL;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001257 }
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001258 if (_PyString_Resize(&v, total_v_size) < 0)
Guido van Rossum1187aa42001-01-05 14:43:05 +00001259 return NULL;
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001260 buf = BUF(v) + used_v_size;
1261 end = BUF(v) + total_v_size;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001262 }
Guido van Rossum1984f1e1992-08-04 12:41:02 +00001263
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001264 used_v_size = buf - BUF(v);
1265 if (used_v_size != total_v_size)
1266 _PyString_Resize(&v, used_v_size);
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001267 return v;
1268}
1269
Guido van Rossum0bd24411991-04-04 15:21:57 +00001270/* External C interface */
1271
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001272PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001273PyFile_GetLine(PyObject *f, int n)
Guido van Rossum0bd24411991-04-04 15:21:57 +00001274{
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001275 PyObject *result;
1276
Guido van Rossum3165fe61992-09-25 21:59:05 +00001277 if (f == NULL) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001278 PyErr_BadInternalCall();
Guido van Rossum0bd24411991-04-04 15:21:57 +00001279 return NULL;
1280 }
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001281
1282 if (PyFile_Check(f)) {
Thomas Woutersc45251a2006-02-12 11:53:32 +00001283 PyFileObject *fo = (PyFileObject *)f;
1284 if (fo->f_fp == NULL)
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001285 return err_closed();
Thomas Woutersc45251a2006-02-12 11:53:32 +00001286 /* refuse to mix with f.next() */
1287 if (fo->f_buf != NULL &&
1288 (fo->f_bufend - fo->f_bufptr) > 0 &&
1289 fo->f_buf[0] != '\0')
1290 return err_iterbuffered();
1291 result = get_line(fo, n);
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001292 }
1293 else {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001294 PyObject *reader;
1295 PyObject *args;
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001296
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001297 reader = PyObject_GetAttrString(f, "readline");
Guido van Rossum3165fe61992-09-25 21:59:05 +00001298 if (reader == NULL)
1299 return NULL;
1300 if (n <= 0)
Raymond Hettinger8ae46892003-10-12 19:09:37 +00001301 args = PyTuple_New(0);
Guido van Rossum3165fe61992-09-25 21:59:05 +00001302 else
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001303 args = Py_BuildValue("(i)", n);
Guido van Rossum3165fe61992-09-25 21:59:05 +00001304 if (args == NULL) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001305 Py_DECREF(reader);
Guido van Rossum3165fe61992-09-25 21:59:05 +00001306 return NULL;
1307 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001308 result = PyEval_CallObject(reader, args);
1309 Py_DECREF(reader);
1310 Py_DECREF(args);
Martin v. Löwisaf6a27a2003-01-03 19:16:14 +00001311 if (result != NULL && !PyString_Check(result) &&
1312 !PyUnicode_Check(result)) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001313 Py_DECREF(result);
Guido van Rossum3165fe61992-09-25 21:59:05 +00001314 result = NULL;
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001315 PyErr_SetString(PyExc_TypeError,
Guido van Rossum3165fe61992-09-25 21:59:05 +00001316 "object.readline() returned non-string");
1317 }
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001318 }
1319
1320 if (n < 0 && result != NULL && PyString_Check(result)) {
1321 char *s = PyString_AS_STRING(result);
Martin v. Löwis18e16552006-02-15 17:27:45 +00001322 Py_ssize_t len = PyString_GET_SIZE(result);
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001323 if (len == 0) {
1324 Py_DECREF(result);
1325 result = NULL;
1326 PyErr_SetString(PyExc_EOFError,
1327 "EOF when reading a line");
1328 }
1329 else if (s[len-1] == '\n') {
1330 if (result->ob_refcnt == 1)
1331 _PyString_Resize(&result, len-1);
1332 else {
1333 PyObject *v;
1334 v = PyString_FromStringAndSize(s, len-1);
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001335 Py_DECREF(result);
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001336 result = v;
Guido van Rossum3165fe61992-09-25 21:59:05 +00001337 }
1338 }
Guido van Rossum3165fe61992-09-25 21:59:05 +00001339 }
Martin v. Löwisaf6a27a2003-01-03 19:16:14 +00001340#ifdef Py_USING_UNICODE
1341 if (n < 0 && result != NULL && PyUnicode_Check(result)) {
1342 Py_UNICODE *s = PyUnicode_AS_UNICODE(result);
Martin v. Löwis18e16552006-02-15 17:27:45 +00001343 Py_ssize_t len = PyUnicode_GET_SIZE(result);
Martin v. Löwisaf6a27a2003-01-03 19:16:14 +00001344 if (len == 0) {
1345 Py_DECREF(result);
1346 result = NULL;
1347 PyErr_SetString(PyExc_EOFError,
1348 "EOF when reading a line");
1349 }
1350 else if (s[len-1] == '\n') {
1351 if (result->ob_refcnt == 1)
1352 PyUnicode_Resize(&result, len-1);
1353 else {
1354 PyObject *v;
1355 v = PyUnicode_FromUnicode(s, len-1);
1356 Py_DECREF(result);
1357 result = v;
1358 }
1359 }
1360 }
1361#endif
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001362 return result;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001363}
1364
1365/* Python method */
1366
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001367static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001368file_readline(PyFileObject *f, PyObject *args)
Guido van Rossum0bd24411991-04-04 15:21:57 +00001369{
Guido van Rossum789a1611997-05-10 22:33:55 +00001370 int n = -1;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001371
Guido van Rossumd7297e61992-07-06 14:19:26 +00001372 if (f->f_fp == NULL)
1373 return err_closed();
Thomas Woutersc45251a2006-02-12 11:53:32 +00001374 /* refuse to mix with f.next() */
1375 if (f->f_buf != NULL &&
1376 (f->f_bufend - f->f_bufptr) > 0 &&
1377 f->f_buf[0] != '\0')
1378 return err_iterbuffered();
Guido van Rossum43713e52000-02-29 13:59:29 +00001379 if (!PyArg_ParseTuple(args, "|i:readline", &n))
Guido van Rossum789a1611997-05-10 22:33:55 +00001380 return NULL;
1381 if (n == 0)
1382 return PyString_FromString("");
1383 if (n < 0)
1384 n = 0;
Marc-André Lemburg1f468602000-07-05 15:32:40 +00001385 return get_line(f, n);
Guido van Rossum0bd24411991-04-04 15:21:57 +00001386}
1387
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001388static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001389file_readlines(PyFileObject *f, PyObject *args)
Guido van Rossumce5ba841991-03-06 13:06:18 +00001390{
Guido van Rossum789a1611997-05-10 22:33:55 +00001391 long sizehint = 0;
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001392 PyObject *list;
1393 PyObject *line;
Guido van Rossum6263d541997-05-10 22:07:25 +00001394 char small_buffer[SMALLCHUNK];
1395 char *buffer = small_buffer;
1396 size_t buffersize = SMALLCHUNK;
1397 PyObject *big_buffer = NULL;
1398 size_t nfilled = 0;
1399 size_t nread;
Guido van Rossum789a1611997-05-10 22:33:55 +00001400 size_t totalread = 0;
Guido van Rossum6263d541997-05-10 22:07:25 +00001401 char *p, *q, *end;
1402 int err;
Guido van Rossum79fd0fc2001-10-12 20:01:53 +00001403 int shortread = 0;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001404
Guido van Rossumd7297e61992-07-06 14:19:26 +00001405 if (f->f_fp == NULL)
1406 return err_closed();
Thomas Woutersc45251a2006-02-12 11:53:32 +00001407 /* refuse to mix with f.next() */
1408 if (f->f_buf != NULL &&
1409 (f->f_bufend - f->f_bufptr) > 0 &&
1410 f->f_buf[0] != '\0')
1411 return err_iterbuffered();
Guido van Rossum43713e52000-02-29 13:59:29 +00001412 if (!PyArg_ParseTuple(args, "|l:readlines", &sizehint))
Guido van Rossum0bd24411991-04-04 15:21:57 +00001413 return NULL;
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001414 if ((list = PyList_New(0)) == NULL)
Guido van Rossumce5ba841991-03-06 13:06:18 +00001415 return NULL;
1416 for (;;) {
Guido van Rossum79fd0fc2001-10-12 20:01:53 +00001417 if (shortread)
1418 nread = 0;
1419 else {
1420 Py_BEGIN_ALLOW_THREADS
1421 errno = 0;
Tim Peters058b1412002-04-21 07:29:14 +00001422 nread = Py_UniversalNewlineFread(buffer+nfilled,
Jack Jansen7b8c7542002-04-14 20:12:41 +00001423 buffersize-nfilled, f->f_fp, (PyObject *)f);
Guido van Rossum79fd0fc2001-10-12 20:01:53 +00001424 Py_END_ALLOW_THREADS
1425 shortread = (nread < buffersize-nfilled);
1426 }
Guido van Rossum6263d541997-05-10 22:07:25 +00001427 if (nread == 0) {
Guido van Rossum789a1611997-05-10 22:33:55 +00001428 sizehint = 0;
Guido van Rossum3da3fce1998-02-19 20:46:48 +00001429 if (!ferror(f->f_fp))
Guido van Rossum6263d541997-05-10 22:07:25 +00001430 break;
1431 PyErr_SetFromErrno(PyExc_IOError);
1432 clearerr(f->f_fp);
1433 error:
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001434 Py_DECREF(list);
Guido van Rossum6263d541997-05-10 22:07:25 +00001435 list = NULL;
1436 goto cleanup;
Guido van Rossumce5ba841991-03-06 13:06:18 +00001437 }
Guido van Rossum789a1611997-05-10 22:33:55 +00001438 totalread += nread;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001439 p = (char *)memchr(buffer+nfilled, '\n', nread);
Guido van Rossum6263d541997-05-10 22:07:25 +00001440 if (p == NULL) {
1441 /* Need a larger buffer to fit this line */
1442 nfilled += nread;
1443 buffersize *= 2;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001444 if (buffersize > PY_SSIZE_T_MAX) {
Trent Mickf29f47b2000-08-11 19:02:59 +00001445 PyErr_SetString(PyExc_OverflowError,
Guido van Rossume07d5cf2001-01-09 21:50:24 +00001446 "line is longer than a Python string can hold");
Trent Mickf29f47b2000-08-11 19:02:59 +00001447 goto error;
1448 }
Guido van Rossum6263d541997-05-10 22:07:25 +00001449 if (big_buffer == NULL) {
1450 /* Create the big buffer */
1451 big_buffer = PyString_FromStringAndSize(
1452 NULL, buffersize);
1453 if (big_buffer == NULL)
1454 goto error;
1455 buffer = PyString_AS_STRING(big_buffer);
1456 memcpy(buffer, small_buffer, nfilled);
1457 }
1458 else {
1459 /* Grow the big buffer */
Jack Jansen7b8c7542002-04-14 20:12:41 +00001460 if ( _PyString_Resize(&big_buffer, buffersize) < 0 )
1461 goto error;
Guido van Rossum6263d541997-05-10 22:07:25 +00001462 buffer = PyString_AS_STRING(big_buffer);
1463 }
1464 continue;
1465 }
1466 end = buffer+nfilled+nread;
1467 q = buffer;
1468 do {
1469 /* Process complete lines */
1470 p++;
1471 line = PyString_FromStringAndSize(q, p-q);
1472 if (line == NULL)
1473 goto error;
1474 err = PyList_Append(list, line);
1475 Py_DECREF(line);
1476 if (err != 0)
1477 goto error;
1478 q = p;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001479 p = (char *)memchr(q, '\n', end-q);
Guido van Rossum6263d541997-05-10 22:07:25 +00001480 } while (p != NULL);
1481 /* Move the remaining incomplete line to the start */
1482 nfilled = end-q;
1483 memmove(buffer, q, nfilled);
Guido van Rossum789a1611997-05-10 22:33:55 +00001484 if (sizehint > 0)
1485 if (totalread >= (size_t)sizehint)
1486 break;
Guido van Rossumce5ba841991-03-06 13:06:18 +00001487 }
Guido van Rossum6263d541997-05-10 22:07:25 +00001488 if (nfilled != 0) {
1489 /* Partial last line */
1490 line = PyString_FromStringAndSize(buffer, nfilled);
1491 if (line == NULL)
1492 goto error;
Guido van Rossum789a1611997-05-10 22:33:55 +00001493 if (sizehint > 0) {
1494 /* Need to complete the last line */
Marc-André Lemburg1f468602000-07-05 15:32:40 +00001495 PyObject *rest = get_line(f, 0);
Guido van Rossum789a1611997-05-10 22:33:55 +00001496 if (rest == NULL) {
1497 Py_DECREF(line);
1498 goto error;
1499 }
1500 PyString_Concat(&line, rest);
1501 Py_DECREF(rest);
1502 if (line == NULL)
1503 goto error;
1504 }
Guido van Rossum6263d541997-05-10 22:07:25 +00001505 err = PyList_Append(list, line);
1506 Py_DECREF(line);
1507 if (err != 0)
1508 goto error;
1509 }
1510 cleanup:
Tim Peters5de98422002-04-27 18:44:32 +00001511 Py_XDECREF(big_buffer);
Guido van Rossumce5ba841991-03-06 13:06:18 +00001512 return list;
1513}
1514
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001515static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001516file_write(PyFileObject *f, PyObject *args)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001517{
Guido van Rossumd7297e61992-07-06 14:19:26 +00001518 char *s;
Martin v. Löwis18e16552006-02-15 17:27:45 +00001519 Py_ssize_t n, n2;
Guido van Rossumd7297e61992-07-06 14:19:26 +00001520 if (f->f_fp == NULL)
1521 return err_closed();
Michael W. Hudsone2ec3eb2001-10-31 18:51:01 +00001522 if (!PyArg_ParseTuple(args, f->f_binary ? "s#" : "t#", &s, &n))
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001523 return NULL;
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001524 Py_BEGIN_ALLOW_THREADS
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001525 errno = 0;
Guido van Rossumd7297e61992-07-06 14:19:26 +00001526 n2 = fwrite(s, 1, n, f->f_fp);
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001527 Py_END_ALLOW_THREADS
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001528 if (n2 != n) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001529 PyErr_SetFromErrno(PyExc_IOError);
Guido van Rossumfebd5511992-03-04 16:39:24 +00001530 clearerr(f->f_fp);
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001531 return NULL;
1532 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001533 Py_INCREF(Py_None);
1534 return Py_None;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001535}
1536
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001537static PyObject *
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001538file_writelines(PyFileObject *f, PyObject *seq)
Guido van Rossum5a2a6831993-10-25 09:59:04 +00001539{
Guido van Rossumee70ad12000-03-13 16:27:06 +00001540#define CHUNKSIZE 1000
1541 PyObject *list, *line;
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001542 PyObject *it; /* iter(seq) */
Guido van Rossumee70ad12000-03-13 16:27:06 +00001543 PyObject *result;
Martin v. Löwis18e16552006-02-15 17:27:45 +00001544 int index, islist;
1545 Py_ssize_t i, j, nwritten, len;
Guido van Rossumee70ad12000-03-13 16:27:06 +00001546
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001547 assert(seq != NULL);
Guido van Rossum5a2a6831993-10-25 09:59:04 +00001548 if (f->f_fp == NULL)
1549 return err_closed();
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001550
1551 result = NULL;
1552 list = NULL;
1553 islist = PyList_Check(seq);
1554 if (islist)
1555 it = NULL;
1556 else {
1557 it = PyObject_GetIter(seq);
1558 if (it == NULL) {
1559 PyErr_SetString(PyExc_TypeError,
1560 "writelines() requires an iterable argument");
1561 return NULL;
1562 }
1563 /* From here on, fail by going to error, to reclaim "it". */
1564 list = PyList_New(CHUNKSIZE);
1565 if (list == NULL)
1566 goto error;
Guido van Rossum5a2a6831993-10-25 09:59:04 +00001567 }
Guido van Rossumee70ad12000-03-13 16:27:06 +00001568
1569 /* Strategy: slurp CHUNKSIZE lines into a private list,
1570 checking that they are all strings, then write that list
1571 without holding the interpreter lock, then come back for more. */
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001572 for (index = 0; ; index += CHUNKSIZE) {
Guido van Rossumee70ad12000-03-13 16:27:06 +00001573 if (islist) {
1574 Py_XDECREF(list);
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001575 list = PyList_GetSlice(seq, index, index+CHUNKSIZE);
Guido van Rossumee70ad12000-03-13 16:27:06 +00001576 if (list == NULL)
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001577 goto error;
Guido van Rossumee70ad12000-03-13 16:27:06 +00001578 j = PyList_GET_SIZE(list);
1579 }
1580 else {
1581 for (j = 0; j < CHUNKSIZE; j++) {
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001582 line = PyIter_Next(it);
Guido van Rossumee70ad12000-03-13 16:27:06 +00001583 if (line == NULL) {
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001584 if (PyErr_Occurred())
1585 goto error;
1586 break;
Guido van Rossumee70ad12000-03-13 16:27:06 +00001587 }
Guido van Rossumee70ad12000-03-13 16:27:06 +00001588 PyList_SetItem(list, j, line);
1589 }
1590 }
1591 if (j == 0)
1592 break;
1593
Marc-André Lemburg6ef68b52000-08-25 22:39:50 +00001594 /* Check that all entries are indeed strings. If not,
1595 apply the same rules as for file.write() and
1596 convert the results to strings. This is slow, but
1597 seems to be the only way since all conversion APIs
1598 could potentially execute Python code. */
1599 for (i = 0; i < j; i++) {
1600 PyObject *v = PyList_GET_ITEM(list, i);
1601 if (!PyString_Check(v)) {
1602 const char *buffer;
Tim Peters86821b22001-01-07 21:19:34 +00001603 if (((f->f_binary &&
Marc-André Lemburg6ef68b52000-08-25 22:39:50 +00001604 PyObject_AsReadBuffer(v,
1605 (const void**)&buffer,
1606 &len)) ||
1607 PyObject_AsCharBuffer(v,
1608 &buffer,
1609 &len))) {
1610 PyErr_SetString(PyExc_TypeError,
Jeremy Hylton8b735422002-08-14 21:01:41 +00001611 "writelines() argument must be a sequence of strings");
Marc-André Lemburg6ef68b52000-08-25 22:39:50 +00001612 goto error;
1613 }
1614 line = PyString_FromStringAndSize(buffer,
1615 len);
1616 if (line == NULL)
1617 goto error;
1618 Py_DECREF(v);
Marc-André Lemburgf5e96fa2000-08-25 22:49:05 +00001619 PyList_SET_ITEM(list, i, line);
Marc-André Lemburg6ef68b52000-08-25 22:39:50 +00001620 }
1621 }
1622
1623 /* Since we are releasing the global lock, the
1624 following code may *not* execute Python code. */
Guido van Rossumee70ad12000-03-13 16:27:06 +00001625 Py_BEGIN_ALLOW_THREADS
Guido van Rossumee70ad12000-03-13 16:27:06 +00001626 errno = 0;
1627 for (i = 0; i < j; i++) {
Marc-André Lemburg6ef68b52000-08-25 22:39:50 +00001628 line = PyList_GET_ITEM(list, i);
Guido van Rossumee70ad12000-03-13 16:27:06 +00001629 len = PyString_GET_SIZE(line);
1630 nwritten = fwrite(PyString_AS_STRING(line),
1631 1, len, f->f_fp);
1632 if (nwritten != len) {
1633 Py_BLOCK_THREADS
1634 PyErr_SetFromErrno(PyExc_IOError);
1635 clearerr(f->f_fp);
1636 goto error;
1637 }
1638 }
1639 Py_END_ALLOW_THREADS
1640
1641 if (j < CHUNKSIZE)
1642 break;
Guido van Rossumee70ad12000-03-13 16:27:06 +00001643 }
1644
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001645 Py_INCREF(Py_None);
Guido van Rossumee70ad12000-03-13 16:27:06 +00001646 result = Py_None;
1647 error:
1648 Py_XDECREF(list);
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001649 Py_XDECREF(it);
Guido van Rossumee70ad12000-03-13 16:27:06 +00001650 return result;
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001651#undef CHUNKSIZE
Guido van Rossum5a2a6831993-10-25 09:59:04 +00001652}
1653
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001654static PyObject *
Guido van Rossum1a5e21e2006-02-28 21:57:43 +00001655file_self(PyFileObject *f)
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001656{
1657 if (f->f_fp == NULL)
1658 return err_closed();
1659 Py_INCREF(f);
1660 return (PyObject *)f;
1661}
1662
Thomas Wouters73e5a5b2006-06-08 15:35:45 +00001663static PyObject *
1664file_exit(PyFileObject *f, PyObject *args)
1665{
1666 PyObject *ret = file_close(f);
1667 if (!ret)
1668 /* If error occurred, pass through */
1669 return NULL;
1670 Py_DECREF(ret);
1671 /* We cannot return the result of close since a true
1672 * value will be interpreted as "yes, swallow the
1673 * exception if one was raised inside the with block". */
1674 Py_RETURN_NONE;
1675}
1676
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001677PyDoc_STRVAR(readline_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001678"readline([size]) -> next line from the file, as a string.\n"
1679"\n"
1680"Retain newline. A non-negative size argument limits the maximum\n"
1681"number of bytes to return (an incomplete line may be returned then).\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001682"Return an empty string at EOF.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001683
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001684PyDoc_STRVAR(read_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001685"read([size]) -> read at most size bytes, returned as a string.\n"
1686"\n"
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +00001687"If the size argument is negative or omitted, read until EOF is reached.\n"
1688"Notice that when in non-blocking mode, less data than what was requested\n"
1689"may be returned, even if no size parameter was given.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001690
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001691PyDoc_STRVAR(write_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001692"write(str) -> None. Write string str to file.\n"
1693"\n"
1694"Note that due to buffering, flush() or close() may be needed before\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001695"the file on disk reflects the data written.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001696
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001697PyDoc_STRVAR(fileno_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001698"fileno() -> integer \"file descriptor\".\n"
1699"\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001700"This is needed for lower-level file interfaces, such os.read().");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001701
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001702PyDoc_STRVAR(seek_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001703"seek(offset[, whence]) -> None. Move to new file position.\n"
1704"\n"
1705"Argument offset is a byte count. Optional argument whence defaults to\n"
1706"0 (offset from start of file, offset should be >= 0); other values are 1\n"
1707"(move relative to current position, positive or negative), and 2 (move\n"
1708"relative to end of file, usually negative, although many platforms allow\n"
Martin v. Löwis849a9722003-10-18 09:38:01 +00001709"seeking beyond the end of a file). If the file is opened in text mode,\n"
1710"only offsets returned by tell() are legal. Use of other offsets causes\n"
1711"undefined behavior."
Tim Petersefc3a3a2001-09-20 07:55:22 +00001712"\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001713"Note that not all file objects are seekable.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001714
Guido van Rossumd7047b31995-01-02 19:07:15 +00001715#ifdef HAVE_FTRUNCATE
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001716PyDoc_STRVAR(truncate_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001717"truncate([size]) -> None. Truncate the file to at most size bytes.\n"
1718"\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001719"Size defaults to the current file position, as returned by tell().");
Guido van Rossumd7047b31995-01-02 19:07:15 +00001720#endif
Tim Petersefc3a3a2001-09-20 07:55:22 +00001721
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001722PyDoc_STRVAR(tell_doc,
1723"tell() -> current file position, an integer (may be a long integer).");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001724
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001725PyDoc_STRVAR(readinto_doc,
1726"readinto() -> Undocumented. Don't use this; it may go away.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001727
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001728PyDoc_STRVAR(readlines_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001729"readlines([size]) -> list of strings, each a line from the file.\n"
1730"\n"
1731"Call readline() repeatedly and return a list of the lines so read.\n"
1732"The optional size argument, if given, is an approximate bound on the\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001733"total number of bytes in the lines returned.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001734
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001735PyDoc_STRVAR(writelines_doc,
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001736"writelines(sequence_of_strings) -> None. Write the strings to the file.\n"
Tim Petersefc3a3a2001-09-20 07:55:22 +00001737"\n"
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001738"Note that newlines are not added. The sequence can be any iterable object\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001739"producing strings. This is equivalent to calling write() for each string.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001740
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001741PyDoc_STRVAR(flush_doc,
1742"flush() -> None. Flush the internal I/O buffer.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001743
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001744PyDoc_STRVAR(close_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001745"close() -> None or (perhaps) an integer. Close the file.\n"
1746"\n"
Guido van Rossum77f6a652002-04-03 22:41:51 +00001747"Sets data attribute .closed to True. A closed file cannot be used for\n"
Tim Petersefc3a3a2001-09-20 07:55:22 +00001748"further I/O operations. close() may be called more than once without\n"
1749"error. Some kinds of file objects (for example, opened by popen())\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001750"may return an exit status upon closing.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001751
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001752PyDoc_STRVAR(isatty_doc,
1753"isatty() -> true or false. True if the file is connected to a tty device.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001754
Guido van Rossum1a5e21e2006-02-28 21:57:43 +00001755PyDoc_STRVAR(enter_doc,
1756 "__enter__() -> self.");
1757
Thomas Wouters73e5a5b2006-06-08 15:35:45 +00001758PyDoc_STRVAR(exit_doc,
1759 "__exit__(*excinfo) -> None. Closes the file.");
1760
Tim Petersefc3a3a2001-09-20 07:55:22 +00001761static PyMethodDef file_methods[] = {
Jeremy Hylton8b735422002-08-14 21:01:41 +00001762 {"readline", (PyCFunction)file_readline, METH_VARARGS, readline_doc},
1763 {"read", (PyCFunction)file_read, METH_VARARGS, read_doc},
1764 {"write", (PyCFunction)file_write, METH_VARARGS, write_doc},
1765 {"fileno", (PyCFunction)file_fileno, METH_NOARGS, fileno_doc},
1766 {"seek", (PyCFunction)file_seek, METH_VARARGS, seek_doc},
Tim Petersefc3a3a2001-09-20 07:55:22 +00001767#ifdef HAVE_FTRUNCATE
Jeremy Hylton8b735422002-08-14 21:01:41 +00001768 {"truncate", (PyCFunction)file_truncate, METH_VARARGS, truncate_doc},
Tim Petersefc3a3a2001-09-20 07:55:22 +00001769#endif
Jeremy Hylton8b735422002-08-14 21:01:41 +00001770 {"tell", (PyCFunction)file_tell, METH_NOARGS, tell_doc},
1771 {"readinto", (PyCFunction)file_readinto, METH_VARARGS, readinto_doc},
1772 {"readlines", (PyCFunction)file_readlines,METH_VARARGS, readlines_doc},
Jeremy Hylton8b735422002-08-14 21:01:41 +00001773 {"writelines",(PyCFunction)file_writelines, METH_O, writelines_doc},
1774 {"flush", (PyCFunction)file_flush, METH_NOARGS, flush_doc},
1775 {"close", (PyCFunction)file_close, METH_NOARGS, close_doc},
1776 {"isatty", (PyCFunction)file_isatty, METH_NOARGS, isatty_doc},
Guido van Rossum1a5e21e2006-02-28 21:57:43 +00001777 {"__enter__", (PyCFunction)file_self, METH_NOARGS, enter_doc},
Thomas Wouters73e5a5b2006-06-08 15:35:45 +00001778 {"__exit__", (PyCFunction)file_exit, METH_VARARGS, exit_doc},
Jeremy Hylton8b735422002-08-14 21:01:41 +00001779 {NULL, NULL} /* sentinel */
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001780};
1781
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001782#define OFF(x) offsetof(PyFileObject, x)
Guido van Rossumb6775db1994-08-01 11:34:53 +00001783
Guido van Rossum6f799372001-09-20 20:46:19 +00001784static PyMemberDef file_memberlist[] = {
Guido van Rossum6f799372001-09-20 20:46:19 +00001785 {"mode", T_OBJECT, OFF(f_mode), RO,
Martin v. Löwis6233c9b2002-12-11 13:06:53 +00001786 "file mode ('r', 'U', 'w', 'a', possibly with 'b' or '+' added)"},
Guido van Rossum6f799372001-09-20 20:46:19 +00001787 {"name", T_OBJECT, OFF(f_name), RO,
1788 "file name"},
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00001789 {"encoding", T_OBJECT, OFF(f_encoding), RO,
1790 "file encoding"},
Guido van Rossumb6775db1994-08-01 11:34:53 +00001791 /* getattr(f, "closed") is implemented without this table */
Guido van Rossumb6775db1994-08-01 11:34:53 +00001792 {NULL} /* Sentinel */
1793};
1794
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001795static PyObject *
Tim Peters6d6c1a32001-08-02 04:15:00 +00001796get_closed(PyFileObject *f, void *closure)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001797{
Guido van Rossum77f6a652002-04-03 22:41:51 +00001798 return PyBool_FromLong((long)(f->f_fp == 0));
Guido van Rossumb6775db1994-08-01 11:34:53 +00001799}
Jack Jansen7b8c7542002-04-14 20:12:41 +00001800static PyObject *
1801get_newlines(PyFileObject *f, void *closure)
1802{
1803 switch (f->f_newlinetypes) {
1804 case NEWLINE_UNKNOWN:
1805 Py_INCREF(Py_None);
1806 return Py_None;
1807 case NEWLINE_CR:
1808 return PyString_FromString("\r");
1809 case NEWLINE_LF:
1810 return PyString_FromString("\n");
1811 case NEWLINE_CR|NEWLINE_LF:
1812 return Py_BuildValue("(ss)", "\r", "\n");
1813 case NEWLINE_CRLF:
1814 return PyString_FromString("\r\n");
1815 case NEWLINE_CR|NEWLINE_CRLF:
1816 return Py_BuildValue("(ss)", "\r", "\r\n");
1817 case NEWLINE_LF|NEWLINE_CRLF:
1818 return Py_BuildValue("(ss)", "\n", "\r\n");
1819 case NEWLINE_CR|NEWLINE_LF|NEWLINE_CRLF:
1820 return Py_BuildValue("(sss)", "\r", "\n", "\r\n");
1821 default:
Tim Petersf1827cf2003-09-07 03:30:18 +00001822 PyErr_Format(PyExc_SystemError,
1823 "Unknown newlines value 0x%x\n",
Jeremy Hylton8b735422002-08-14 21:01:41 +00001824 f->f_newlinetypes);
Jack Jansen7b8c7542002-04-14 20:12:41 +00001825 return NULL;
1826 }
1827}
Guido van Rossumb6775db1994-08-01 11:34:53 +00001828
Guido van Rossum32d34c82001-09-20 21:45:26 +00001829static PyGetSetDef file_getsetlist[] = {
Guido van Rossum77f6a652002-04-03 22:41:51 +00001830 {"closed", (getter)get_closed, NULL, "True if the file is closed"},
Tim Petersf1827cf2003-09-07 03:30:18 +00001831 {"newlines", (getter)get_newlines, NULL,
Jeremy Hylton8b735422002-08-14 21:01:41 +00001832 "end-of-line convention used in this file"},
Tim Peters6d6c1a32001-08-02 04:15:00 +00001833 {0},
1834};
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001835
Neal Norwitzd8b995f2002-08-06 21:50:54 +00001836static void
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001837drop_readahead(PyFileObject *f)
Guido van Rossum65967252001-04-21 13:20:18 +00001838{
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001839 if (f->f_buf != NULL) {
1840 PyMem_Free(f->f_buf);
1841 f->f_buf = NULL;
1842 }
Guido van Rossum65967252001-04-21 13:20:18 +00001843}
1844
Tim Petersf1827cf2003-09-07 03:30:18 +00001845/* Make sure that file has a readahead buffer with at least one byte
1846 (unless at EOF) and no more than bufsize. Returns negative value on
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001847 error, will set MemoryError if bufsize bytes cannot be allocated. */
Neal Norwitzd8b995f2002-08-06 21:50:54 +00001848static int
1849readahead(PyFileObject *f, int bufsize)
1850{
Martin v. Löwis18e16552006-02-15 17:27:45 +00001851 Py_ssize_t chunksize;
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001852
1853 if (f->f_buf != NULL) {
Tim Petersf1827cf2003-09-07 03:30:18 +00001854 if( (f->f_bufend - f->f_bufptr) >= 1)
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001855 return 0;
1856 else
1857 drop_readahead(f);
1858 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001859 if ((f->f_buf = (char *)PyMem_Malloc(bufsize)) == NULL) {
1860 PyErr_NoMemory();
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001861 return -1;
1862 }
1863 Py_BEGIN_ALLOW_THREADS
1864 errno = 0;
1865 chunksize = Py_UniversalNewlineFread(
1866 f->f_buf, bufsize, f->f_fp, (PyObject *)f);
1867 Py_END_ALLOW_THREADS
1868 if (chunksize == 0) {
1869 if (ferror(f->f_fp)) {
1870 PyErr_SetFromErrno(PyExc_IOError);
1871 clearerr(f->f_fp);
1872 drop_readahead(f);
1873 return -1;
1874 }
1875 }
1876 f->f_bufptr = f->f_buf;
1877 f->f_bufend = f->f_buf + chunksize;
1878 return 0;
1879}
1880
1881/* Used by file_iternext. The returned string will start with 'skip'
Tim Petersf1827cf2003-09-07 03:30:18 +00001882 uninitialized bytes followed by the remainder of the line. Don't be
1883 horrified by the recursive call: maximum recursion depth is limited by
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001884 logarithmic buffer growth to about 50 even when reading a 1gb line. */
1885
Neal Norwitzd8b995f2002-08-06 21:50:54 +00001886static PyStringObject *
1887readahead_get_line_skip(PyFileObject *f, int skip, int bufsize)
1888{
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001889 PyStringObject* s;
1890 char *bufptr;
1891 char *buf;
Martin v. Löwis18e16552006-02-15 17:27:45 +00001892 Py_ssize_t len;
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001893
1894 if (f->f_buf == NULL)
Tim Petersf1827cf2003-09-07 03:30:18 +00001895 if (readahead(f, bufsize) < 0)
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001896 return NULL;
1897
1898 len = f->f_bufend - f->f_bufptr;
Tim Petersf1827cf2003-09-07 03:30:18 +00001899 if (len == 0)
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001900 return (PyStringObject *)
1901 PyString_FromStringAndSize(NULL, skip);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001902 bufptr = (char *)memchr(f->f_bufptr, '\n', len);
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001903 if (bufptr != NULL) {
1904 bufptr++; /* Count the '\n' */
1905 len = bufptr - f->f_bufptr;
1906 s = (PyStringObject *)
1907 PyString_FromStringAndSize(NULL, skip+len);
Tim Petersf1827cf2003-09-07 03:30:18 +00001908 if (s == NULL)
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001909 return NULL;
1910 memcpy(PyString_AS_STRING(s)+skip, f->f_bufptr, len);
1911 f->f_bufptr = bufptr;
1912 if (bufptr == f->f_bufend)
1913 drop_readahead(f);
1914 } else {
1915 bufptr = f->f_bufptr;
1916 buf = f->f_buf;
1917 f->f_buf = NULL; /* Force new readahead buffer */
Martin v. Löwis18e16552006-02-15 17:27:45 +00001918 assert(skip+len < INT_MAX);
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001919 s = readahead_get_line_skip(
Martin v. Löwis18e16552006-02-15 17:27:45 +00001920 f, (int)(skip+len), bufsize + (bufsize>>2) );
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001921 if (s == NULL) {
1922 PyMem_Free(buf);
1923 return NULL;
1924 }
1925 memcpy(PyString_AS_STRING(s)+skip, bufptr, len);
1926 PyMem_Free(buf);
1927 }
1928 return s;
1929}
1930
1931/* A larger buffer size may actually decrease performance. */
1932#define READAHEAD_BUFSIZE 8192
1933
1934static PyObject *
1935file_iternext(PyFileObject *f)
1936{
1937 PyStringObject* l;
1938
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001939 if (f->f_fp == NULL)
1940 return err_closed();
1941
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001942 l = readahead_get_line_skip(f, 0, READAHEAD_BUFSIZE);
1943 if (l == NULL || PyString_GET_SIZE(l) == 0) {
1944 Py_XDECREF(l);
1945 return NULL;
1946 }
1947 return (PyObject *)l;
1948}
1949
1950
Tim Peters59c9a642001-09-13 05:38:56 +00001951static PyObject *
1952file_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1953{
Tim Peters44410012001-09-14 03:26:08 +00001954 PyObject *self;
1955 static PyObject *not_yet_string;
1956
1957 assert(type != NULL && type->tp_alloc != NULL);
1958
1959 if (not_yet_string == NULL) {
1960 not_yet_string = PyString_FromString("<uninitialized file>");
1961 if (not_yet_string == NULL)
1962 return NULL;
1963 }
1964
1965 self = type->tp_alloc(type, 0);
1966 if (self != NULL) {
1967 /* Always fill in the name and mode, so that nobody else
1968 needs to special-case NULLs there. */
1969 Py_INCREF(not_yet_string);
1970 ((PyFileObject *)self)->f_name = not_yet_string;
1971 Py_INCREF(not_yet_string);
1972 ((PyFileObject *)self)->f_mode = not_yet_string;
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00001973 Py_INCREF(Py_None);
1974 ((PyFileObject *)self)->f_encoding = Py_None;
Raymond Hettingercb87bc82004-05-31 00:35:52 +00001975 ((PyFileObject *)self)->weakreflist = NULL;
Tim Peters44410012001-09-14 03:26:08 +00001976 }
1977 return self;
1978}
1979
1980static int
1981file_init(PyObject *self, PyObject *args, PyObject *kwds)
1982{
1983 PyFileObject *foself = (PyFileObject *)self;
1984 int ret = 0;
Martin v. Löwis15e62742006-02-27 16:46:16 +00001985 static char *kwlist[] = {"name", "mode", "buffering", 0};
Tim Peters59c9a642001-09-13 05:38:56 +00001986 char *name = NULL;
1987 char *mode = "r";
1988 int bufsize = -1;
Mark Hammondc2e85bd2002-10-03 05:10:39 +00001989 int wideargument = 0;
Tim Peters44410012001-09-14 03:26:08 +00001990
1991 assert(PyFile_Check(self));
1992 if (foself->f_fp != NULL) {
1993 /* Have to close the existing file first. */
1994 PyObject *closeresult = file_close(foself);
1995 if (closeresult == NULL)
1996 return -1;
1997 Py_DECREF(closeresult);
1998 }
Tim Peters59c9a642001-09-13 05:38:56 +00001999
Mark Hammondc2e85bd2002-10-03 05:10:39 +00002000#ifdef Py_WIN_WIDE_FILENAMES
2001 if (GetVersion() < 0x80000000) { /* On NT, so wide API available */
2002 PyObject *po;
2003 if (PyArg_ParseTupleAndKeywords(args, kwds, "U|si:file",
2004 kwlist, &po, &mode, &bufsize)) {
2005 wideargument = 1;
Nicholas Bastinabce8a62004-03-21 20:24:07 +00002006 if (fill_file_fields(foself, NULL, po, mode,
2007 fclose) == NULL)
Mark Hammondc2e85bd2002-10-03 05:10:39 +00002008 goto Error;
2009 } else {
2010 /* Drop the argument parsing error as narrow
2011 strings are also valid. */
2012 PyErr_Clear();
2013 }
2014 }
2015#endif
2016
2017 if (!wideargument) {
Nicholas Bastinabce8a62004-03-21 20:24:07 +00002018 PyObject *o_name;
2019
Mark Hammondc2e85bd2002-10-03 05:10:39 +00002020 if (!PyArg_ParseTupleAndKeywords(args, kwds, "et|si:file", kwlist,
2021 Py_FileSystemDefaultEncoding,
2022 &name,
2023 &mode, &bufsize))
2024 return -1;
Nicholas Bastinabce8a62004-03-21 20:24:07 +00002025
2026 /* We parse again to get the name as a PyObject */
Jeremy Hyltonaf68c872005-12-10 18:50:16 +00002027 if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|si:file",
2028 kwlist, &o_name, &mode,
2029 &bufsize))
Thomas Wouters89f507f2006-12-13 04:49:30 +00002030 goto Error;
Nicholas Bastinabce8a62004-03-21 20:24:07 +00002031
2032 if (fill_file_fields(foself, NULL, o_name, mode,
2033 fclose) == NULL)
Mark Hammondc2e85bd2002-10-03 05:10:39 +00002034 goto Error;
2035 }
Tim Peters44410012001-09-14 03:26:08 +00002036 if (open_the_file(foself, name, mode) == NULL)
2037 goto Error;
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +00002038 foself->f_setbuf = NULL;
Tim Peters44410012001-09-14 03:26:08 +00002039 PyFile_SetBufSize(self, bufsize);
2040 goto Done;
2041
2042Error:
2043 ret = -1;
2044 /* fall through */
2045Done:
Tim Peters59c9a642001-09-13 05:38:56 +00002046 PyMem_Free(name); /* free the encoded string */
Tim Peters44410012001-09-14 03:26:08 +00002047 return ret;
Tim Peters59c9a642001-09-13 05:38:56 +00002048}
2049
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00002050PyDoc_VAR(file_doc) =
2051PyDoc_STR(
Tim Peters59c9a642001-09-13 05:38:56 +00002052"file(name[, mode[, buffering]]) -> file object\n"
2053"\n"
2054"Open a file. The mode can be 'r', 'w' or 'a' for reading (default),\n"
2055"writing or appending. The file will be created if it doesn't exist\n"
2056"when opened for writing or appending; it will be truncated when\n"
2057"opened for writing. Add a 'b' to the mode for binary files.\n"
2058"Add a '+' to the mode to allow simultaneous reading and writing.\n"
2059"If the buffering argument is given, 0 means unbuffered, 1 means line\n"
Tim Peters742dfd62001-09-13 21:49:44 +00002060"buffered, and larger numbers specify the buffer size.\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00002061)
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00002062PyDoc_STR(
Barry Warsaw4be55b52002-05-22 20:37:53 +00002063"Add a 'U' to mode to open the file for input with universal newline\n"
2064"support. Any line ending in the input file will be seen as a '\\n'\n"
2065"in Python. Also, a file so opened gains the attribute 'newlines';\n"
2066"the value for this attribute is one of None (no newline read yet),\n"
2067"'\\r', '\\n', '\\r\\n' or a tuple containing all the newline types seen.\n"
2068"\n"
2069"'U' cannot be combined with 'w' or '+' mode.\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00002070);
Tim Peters59c9a642001-09-13 05:38:56 +00002071
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002072PyTypeObject PyFile_Type = {
2073 PyObject_HEAD_INIT(&PyType_Type)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00002074 0,
2075 "file",
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002076 sizeof(PyFileObject),
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00002077 0,
Guido van Rossum65967252001-04-21 13:20:18 +00002078 (destructor)file_dealloc, /* tp_dealloc */
2079 0, /* tp_print */
Tim Peters6d6c1a32001-08-02 04:15:00 +00002080 0, /* tp_getattr */
2081 0, /* tp_setattr */
Guido van Rossum65967252001-04-21 13:20:18 +00002082 0, /* tp_compare */
Tim Peters6d6c1a32001-08-02 04:15:00 +00002083 (reprfunc)file_repr, /* tp_repr */
Guido van Rossum65967252001-04-21 13:20:18 +00002084 0, /* tp_as_number */
2085 0, /* tp_as_sequence */
2086 0, /* tp_as_mapping */
2087 0, /* tp_hash */
2088 0, /* tp_call */
2089 0, /* tp_str */
Tim Peters6d6c1a32001-08-02 04:15:00 +00002090 PyObject_GenericGetAttr, /* tp_getattro */
Guido van Rossum79139b22007-02-09 23:20:19 +00002091 0, /* tp_setattro */
Guido van Rossum65967252001-04-21 13:20:18 +00002092 0, /* tp_as_buffer */
Guido van Rossum3cf5b1e2006-07-27 21:53:35 +00002093 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE, /* tp_flags */
Tim Peters59c9a642001-09-13 05:38:56 +00002094 file_doc, /* tp_doc */
Tim Peters6d6c1a32001-08-02 04:15:00 +00002095 0, /* tp_traverse */
2096 0, /* tp_clear */
Guido van Rossum65967252001-04-21 13:20:18 +00002097 0, /* tp_richcompare */
Raymond Hettingercb87bc82004-05-31 00:35:52 +00002098 offsetof(PyFileObject, weakreflist), /* tp_weaklistoffset */
Guido van Rossum1a5e21e2006-02-28 21:57:43 +00002099 (getiterfunc)file_self, /* tp_iter */
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002100 (iternextfunc)file_iternext, /* tp_iternext */
Tim Peters6d6c1a32001-08-02 04:15:00 +00002101 file_methods, /* tp_methods */
2102 file_memberlist, /* tp_members */
2103 file_getsetlist, /* tp_getset */
2104 0, /* tp_base */
2105 0, /* tp_dict */
Tim Peters59c9a642001-09-13 05:38:56 +00002106 0, /* tp_descr_get */
2107 0, /* tp_descr_set */
2108 0, /* tp_dictoffset */
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00002109 file_init, /* tp_init */
Tim Peters44410012001-09-14 03:26:08 +00002110 PyType_GenericAlloc, /* tp_alloc */
Tim Peters59c9a642001-09-13 05:38:56 +00002111 file_new, /* tp_new */
Neil Schemenaueraa769ae2002-04-12 02:44:10 +00002112 PyObject_Del, /* tp_free */
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00002113};
Guido van Rossumeb183da1991-04-04 10:44:06 +00002114
Guido van Rossum3165fe61992-09-25 21:59:05 +00002115/* Interfaces to write objects/strings to file-like objects */
2116
2117int
Fred Drakefd99de62000-07-09 05:02:18 +00002118PyFile_WriteObject(PyObject *v, PyObject *f, int flags)
Guido van Rossum3165fe61992-09-25 21:59:05 +00002119{
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002120 PyObject *writer, *value, *args, *result;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002121 if (f == NULL) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002122 PyErr_SetString(PyExc_TypeError, "writeobject with NULL file");
Guido van Rossum3165fe61992-09-25 21:59:05 +00002123 return -1;
2124 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002125 else if (PyFile_Check(f)) {
2126 FILE *fp = PyFile_AsFile(f);
Fred Drake086a0f72004-03-19 15:22:36 +00002127#ifdef Py_USING_UNICODE
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00002128 PyObject *enc = ((PyFileObject*)f)->f_encoding;
2129 int result;
Fred Drake086a0f72004-03-19 15:22:36 +00002130#endif
Guido van Rossum3165fe61992-09-25 21:59:05 +00002131 if (fp == NULL) {
2132 err_closed();
2133 return -1;
2134 }
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00002135#ifdef Py_USING_UNICODE
Tim Petersf1827cf2003-09-07 03:30:18 +00002136 if ((flags & Py_PRINT_RAW) &&
Martin v. Löwis415da6e2003-05-18 12:56:25 +00002137 PyUnicode_Check(v) && enc != Py_None) {
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00002138 char *cenc = PyString_AS_STRING(enc);
2139 value = PyUnicode_AsEncodedString(v, cenc, "strict");
2140 if (value == NULL)
2141 return -1;
2142 } else {
2143 value = v;
2144 Py_INCREF(value);
2145 }
2146 result = PyObject_Print(value, fp, flags);
2147 Py_DECREF(value);
2148 return result;
2149#else
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002150 return PyObject_Print(v, fp, flags);
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00002151#endif
Guido van Rossum3165fe61992-09-25 21:59:05 +00002152 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002153 writer = PyObject_GetAttrString(f, "write");
Guido van Rossum3165fe61992-09-25 21:59:05 +00002154 if (writer == NULL)
2155 return -1;
Martin v. Löwis2777c022001-09-19 13:47:32 +00002156 if (flags & Py_PRINT_RAW) {
2157 if (PyUnicode_Check(v)) {
2158 value = v;
2159 Py_INCREF(value);
2160 } else
2161 value = PyObject_Str(v);
2162 }
2163 else
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002164 value = PyObject_Repr(v);
Guido van Rossumc6004111993-11-05 10:22:19 +00002165 if (value == NULL) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002166 Py_DECREF(writer);
Guido van Rossumc6004111993-11-05 10:22:19 +00002167 return -1;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002168 }
Raymond Hettinger8ae46892003-10-12 19:09:37 +00002169 args = PyTuple_Pack(1, value);
Guido van Rossume9eec541997-05-22 14:02:25 +00002170 if (args == NULL) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002171 Py_DECREF(value);
2172 Py_DECREF(writer);
Guido van Rossumd3f9a1a1995-07-10 23:32:26 +00002173 return -1;
2174 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002175 result = PyEval_CallObject(writer, args);
2176 Py_DECREF(args);
2177 Py_DECREF(value);
2178 Py_DECREF(writer);
Guido van Rossum3165fe61992-09-25 21:59:05 +00002179 if (result == NULL)
2180 return -1;
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002181 Py_DECREF(result);
Guido van Rossum3165fe61992-09-25 21:59:05 +00002182 return 0;
2183}
2184
Guido van Rossum27a60b11997-05-22 22:25:11 +00002185int
Tim Petersc1bbcb82001-11-28 22:13:25 +00002186PyFile_WriteString(const char *s, PyObject *f)
Guido van Rossum3165fe61992-09-25 21:59:05 +00002187{
2188 if (f == NULL) {
Guido van Rossum27a60b11997-05-22 22:25:11 +00002189 /* Should be caused by a pre-existing error */
Fred Drakefd99de62000-07-09 05:02:18 +00002190 if (!PyErr_Occurred())
Guido van Rossum27a60b11997-05-22 22:25:11 +00002191 PyErr_SetString(PyExc_SystemError,
2192 "null file for PyFile_WriteString");
2193 return -1;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002194 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002195 else if (PyFile_Check(f)) {
2196 FILE *fp = PyFile_AsFile(f);
Guido van Rossum27a60b11997-05-22 22:25:11 +00002197 if (fp == NULL) {
2198 err_closed();
2199 return -1;
2200 }
2201 fputs(s, fp);
2202 return 0;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002203 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002204 else if (!PyErr_Occurred()) {
2205 PyObject *v = PyString_FromString(s);
Guido van Rossum27a60b11997-05-22 22:25:11 +00002206 int err;
2207 if (v == NULL)
2208 return -1;
2209 err = PyFile_WriteObject(v, f, Py_PRINT_RAW);
2210 Py_DECREF(v);
2211 return err;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002212 }
Guido van Rossum74ba2471997-07-13 03:56:50 +00002213 else
2214 return -1;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002215}
Andrew M. Kuchling06051ed2000-07-13 23:56:54 +00002216
2217/* Try to get a file-descriptor from a Python object. If the object
2218 is an integer or long integer, its value is returned. If not, the
2219 object's fileno() method is called if it exists; the method must return
2220 an integer or long integer, which is returned as the file descriptor value.
2221 -1 is returned on failure.
2222*/
2223
2224int PyObject_AsFileDescriptor(PyObject *o)
2225{
2226 int fd;
2227 PyObject *meth;
2228
2229 if (PyInt_Check(o)) {
2230 fd = PyInt_AsLong(o);
2231 }
2232 else if (PyLong_Check(o)) {
2233 fd = PyLong_AsLong(o);
2234 }
2235 else if ((meth = PyObject_GetAttrString(o, "fileno")) != NULL)
2236 {
2237 PyObject *fno = PyEval_CallObject(meth, NULL);
2238 Py_DECREF(meth);
2239 if (fno == NULL)
2240 return -1;
Tim Peters86821b22001-01-07 21:19:34 +00002241
Andrew M. Kuchling06051ed2000-07-13 23:56:54 +00002242 if (PyInt_Check(fno)) {
2243 fd = PyInt_AsLong(fno);
2244 Py_DECREF(fno);
2245 }
2246 else if (PyLong_Check(fno)) {
2247 fd = PyLong_AsLong(fno);
2248 Py_DECREF(fno);
2249 }
2250 else {
2251 PyErr_SetString(PyExc_TypeError,
2252 "fileno() returned a non-integer");
2253 Py_DECREF(fno);
2254 return -1;
2255 }
2256 }
2257 else {
2258 PyErr_SetString(PyExc_TypeError,
2259 "argument must be an int, or have a fileno() method.");
2260 return -1;
2261 }
2262
Guido van Rossumddefaf32007-01-14 03:31:43 +00002263 if (fd == -1 && PyErr_Occurred())
2264 return -1;
Andrew M. Kuchling06051ed2000-07-13 23:56:54 +00002265 if (fd < 0) {
2266 PyErr_Format(PyExc_ValueError,
2267 "file descriptor cannot be a negative integer (%i)",
2268 fd);
2269 return -1;
2270 }
2271 return fd;
2272}
Jack Jansen7b8c7542002-04-14 20:12:41 +00002273
Jack Jansen7b8c7542002-04-14 20:12:41 +00002274/* From here on we need access to the real fgets and fread */
2275#undef fgets
2276#undef fread
2277
2278/*
2279** Py_UniversalNewlineFgets is an fgets variation that understands
2280** all of \r, \n and \r\n conventions.
2281** The stream should be opened in binary mode.
2282** If fobj is NULL the routine always does newline conversion, and
2283** it may peek one char ahead to gobble the second char in \r\n.
2284** If fobj is non-NULL it must be a PyFileObject. In this case there
2285** is no readahead but in stead a flag is used to skip a following
2286** \n on the next read. Also, if the file is open in binary mode
2287** the whole conversion is skipped. Finally, the routine keeps track of
2288** the different types of newlines seen.
2289** Note that we need no error handling: fgets() treats error and eof
2290** identically.
2291*/
2292char *
2293Py_UniversalNewlineFgets(char *buf, int n, FILE *stream, PyObject *fobj)
2294{
2295 char *p = buf;
2296 int c;
2297 int newlinetypes = 0;
2298 int skipnextlf = 0;
2299 int univ_newline = 1;
Tim Peters058b1412002-04-21 07:29:14 +00002300
Jack Jansen7b8c7542002-04-14 20:12:41 +00002301 if (fobj) {
2302 if (!PyFile_Check(fobj)) {
2303 errno = ENXIO; /* What can you do... */
2304 return NULL;
2305 }
2306 univ_newline = ((PyFileObject *)fobj)->f_univ_newline;
2307 if ( !univ_newline )
2308 return fgets(buf, n, stream);
2309 newlinetypes = ((PyFileObject *)fobj)->f_newlinetypes;
2310 skipnextlf = ((PyFileObject *)fobj)->f_skipnextlf;
2311 }
2312 FLOCKFILE(stream);
2313 c = 'x'; /* Shut up gcc warning */
2314 while (--n > 0 && (c = GETC(stream)) != EOF ) {
2315 if (skipnextlf ) {
2316 skipnextlf = 0;
2317 if (c == '\n') {
2318 /* Seeing a \n here with skipnextlf true
2319 ** means we saw a \r before.
2320 */
2321 newlinetypes |= NEWLINE_CRLF;
2322 c = GETC(stream);
2323 if (c == EOF) break;
2324 } else {
2325 /*
2326 ** Note that c == EOF also brings us here,
2327 ** so we're okay if the last char in the file
2328 ** is a CR.
2329 */
2330 newlinetypes |= NEWLINE_CR;
2331 }
2332 }
2333 if (c == '\r') {
2334 /* A \r is translated into a \n, and we skip
2335 ** an adjacent \n, if any. We don't set the
2336 ** newlinetypes flag until we've seen the next char.
2337 */
2338 skipnextlf = 1;
2339 c = '\n';
2340 } else if ( c == '\n') {
2341 newlinetypes |= NEWLINE_LF;
2342 }
2343 *p++ = c;
2344 if (c == '\n') break;
2345 }
2346 if ( c == EOF && skipnextlf )
2347 newlinetypes |= NEWLINE_CR;
2348 FUNLOCKFILE(stream);
2349 *p = '\0';
2350 if (fobj) {
2351 ((PyFileObject *)fobj)->f_newlinetypes = newlinetypes;
2352 ((PyFileObject *)fobj)->f_skipnextlf = skipnextlf;
2353 } else if ( skipnextlf ) {
2354 /* If we have no file object we cannot save the
2355 ** skipnextlf flag. We have to readahead, which
2356 ** will cause a pause if we're reading from an
2357 ** interactive stream, but that is very unlikely
2358 ** unless we're doing something silly like
2359 ** execfile("/dev/tty").
2360 */
2361 c = GETC(stream);
2362 if ( c != '\n' )
2363 ungetc(c, stream);
2364 }
2365 if (p == buf)
2366 return NULL;
2367 return buf;
2368}
2369
2370/*
2371** Py_UniversalNewlineFread is an fread variation that understands
2372** all of \r, \n and \r\n conventions.
2373** The stream should be opened in binary mode.
2374** fobj must be a PyFileObject. In this case there
2375** is no readahead but in stead a flag is used to skip a following
2376** \n on the next read. Also, if the file is open in binary mode
2377** the whole conversion is skipped. Finally, the routine keeps track of
2378** the different types of newlines seen.
2379*/
2380size_t
Tim Peters058b1412002-04-21 07:29:14 +00002381Py_UniversalNewlineFread(char *buf, size_t n,
Jack Jansen7b8c7542002-04-14 20:12:41 +00002382 FILE *stream, PyObject *fobj)
2383{
Tim Peters058b1412002-04-21 07:29:14 +00002384 char *dst = buf;
2385 PyFileObject *f = (PyFileObject *)fobj;
2386 int newlinetypes, skipnextlf;
2387
2388 assert(buf != NULL);
2389 assert(stream != NULL);
2390
Jack Jansen7b8c7542002-04-14 20:12:41 +00002391 if (!fobj || !PyFile_Check(fobj)) {
2392 errno = ENXIO; /* What can you do... */
Neal Norwitzcb3319f2003-02-09 01:10:02 +00002393 return 0;
Jack Jansen7b8c7542002-04-14 20:12:41 +00002394 }
Tim Peters058b1412002-04-21 07:29:14 +00002395 if (!f->f_univ_newline)
Jack Jansen7b8c7542002-04-14 20:12:41 +00002396 return fread(buf, 1, n, stream);
Tim Peters058b1412002-04-21 07:29:14 +00002397 newlinetypes = f->f_newlinetypes;
2398 skipnextlf = f->f_skipnextlf;
2399 /* Invariant: n is the number of bytes remaining to be filled
2400 * in the buffer.
2401 */
2402 while (n) {
2403 size_t nread;
2404 int shortread;
2405 char *src = dst;
2406
2407 nread = fread(dst, 1, n, stream);
2408 assert(nread <= n);
Neal Norwitzcb3319f2003-02-09 01:10:02 +00002409 if (nread == 0)
2410 break;
2411
Tim Peterse1682a82002-04-21 18:15:20 +00002412 n -= nread; /* assuming 1 byte out for each in; will adjust */
2413 shortread = n != 0; /* true iff EOF or error */
Tim Peters058b1412002-04-21 07:29:14 +00002414 while (nread--) {
2415 char c = *src++;
Jack Jansen7b8c7542002-04-14 20:12:41 +00002416 if (c == '\r') {
Tim Peters058b1412002-04-21 07:29:14 +00002417 /* Save as LF and set flag to skip next LF. */
Jack Jansen7b8c7542002-04-14 20:12:41 +00002418 *dst++ = '\n';
2419 skipnextlf = 1;
Tim Peters058b1412002-04-21 07:29:14 +00002420 }
2421 else if (skipnextlf && c == '\n') {
2422 /* Skip LF, and remember we saw CR LF. */
Jack Jansen7b8c7542002-04-14 20:12:41 +00002423 skipnextlf = 0;
2424 newlinetypes |= NEWLINE_CRLF;
Tim Peterse1682a82002-04-21 18:15:20 +00002425 ++n;
Tim Peters058b1412002-04-21 07:29:14 +00002426 }
2427 else {
2428 /* Normal char to be stored in buffer. Also
2429 * update the newlinetypes flag if either this
2430 * is an LF or the previous char was a CR.
2431 */
Jack Jansen7b8c7542002-04-14 20:12:41 +00002432 if (c == '\n')
2433 newlinetypes |= NEWLINE_LF;
2434 else if (skipnextlf)
2435 newlinetypes |= NEWLINE_CR;
2436 *dst++ = c;
2437 skipnextlf = 0;
2438 }
2439 }
Tim Peters058b1412002-04-21 07:29:14 +00002440 if (shortread) {
2441 /* If this is EOF, update type flags. */
2442 if (skipnextlf && feof(stream))
2443 newlinetypes |= NEWLINE_CR;
2444 break;
2445 }
Jack Jansen7b8c7542002-04-14 20:12:41 +00002446 }
Tim Peters058b1412002-04-21 07:29:14 +00002447 f->f_newlinetypes = newlinetypes;
2448 f->f_skipnextlf = skipnextlf;
2449 return dst - buf;
Jack Jansen7b8c7542002-04-14 20:12:41 +00002450}
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00002451
2452#ifdef __cplusplus
2453}
2454#endif