blob: 71ba01b3ecb58ed5b8abe7c8efa43dd00ae5fa9d [file] [log] [blame]
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001/* File object implementation */
2
Martin v. Löwis18e16552006-02-15 17:27:45 +00003#define PY_SSIZE_T_CLEAN
Guido van Rossumc0b618a1997-05-02 03:12:38 +00004#include "Python.h"
Guido van Rossumb6775db1994-08-01 11:34:53 +00005#include "structmember.h"
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00006
Martin v. Löwis0e8bd7e2006-06-10 12:23:46 +00007#ifdef HAVE_SYS_TYPES_H
Guido van Rossum41498431999-01-07 22:09:51 +00008#include <sys/types.h>
Martin v. Löwis0e8bd7e2006-06-10 12:23:46 +00009#endif /* HAVE_SYS_TYPES_H */
Guido van Rossum41498431999-01-07 22:09:51 +000010
Martin v. Löwis6238d2b2002-06-30 15:26:10 +000011#ifdef MS_WINDOWS
Guido van Rossumb8199141997-05-06 15:23:24 +000012#define fileno _fileno
Tim Petersfb05db22002-03-11 00:24:00 +000013/* can simulate truncate with Win32 API functions; see file_truncate */
Guido van Rossumb8199141997-05-06 15:23:24 +000014#define HAVE_FTRUNCATE
Tim Peters7a1f9172002-07-14 22:14:19 +000015#define WIN32_LEAN_AND_MEAN
Tim Petersfb05db22002-03-11 00:24:00 +000016#include <windows.h>
Guido van Rossumb8199141997-05-06 15:23:24 +000017#endif
18
Mark Hammondc2e85bd2002-10-03 05:10:39 +000019#ifdef _MSC_VER
20/* Need GetVersion to see if on NT so safe to use _wfopen */
21#define WIN32_LEAN_AND_MEAN
22#include <windows.h>
23#endif /* _MSC_VER */
24
Andrew MacIntyrec4874392002-02-26 11:36:35 +000025#if defined(PYOS_OS2) && defined(PYCC_GCC)
26#include <io.h>
27#endif
28
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
Anthony Baxterac6bd462006-04-13 02:06:09 +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{
Neal Norwitzb337bb52006-07-17 00:55:45 +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
Neal Norwitzb337bb52006-07-17 00:55:45 +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;
Guido van Rossumeb183da1991-04-04 10:44:06 +0000121 f->f_softspace = 0;
Tim Peters59c9a642001-09-13 05:38:56 +0000122 f->f_binary = strchr(mode,'b') != NULL;
Guido van Rossum7a6e9592002-08-06 15:55:28 +0000123 f->f_buf = NULL;
Jack Jansen7b8c7542002-04-14 20:12:41 +0000124 f->f_univ_newline = (strchr(mode, 'U') != NULL);
125 f->f_newlinetypes = NEWLINE_UNKNOWN;
126 f->f_skipnextlf = 0;
Martin v. Löwis5467d4c2003-05-10 07:10:12 +0000127 Py_INCREF(Py_None);
128 f->f_encoding = Py_None;
Tim Petersf1827cf2003-09-07 03:30:18 +0000129
Neal Norwitzb337bb52006-07-17 00:55:45 +0000130 if (f->f_mode == NULL)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000131 return NULL;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000132 f->f_fp = fp;
Neil Schemenauered19b882002-03-23 02:06:50 +0000133 f = dircheck(f);
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000134 return (PyObject *) f;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000135}
136
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000137/* check for known incorrect mode strings - problem is, platforms are
138 free to accept any mode characters they like and are supposed to
139 ignore stuff they don't understand... write or append mode with
Georg Brandl7b90e162006-05-18 07:01:27 +0000140 universal newline support is expressly forbidden by PEP 278.
141 Additionally, remove the 'U' from the mode string as platforms
142 won't know what it is. */
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000143/* zero return is kewl - one is un-kewl */
144static int
Georg Brandl7b90e162006-05-18 07:01:27 +0000145sanitize_the_mode(char *mode)
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000146{
Georg Brandl7b90e162006-05-18 07:01:27 +0000147 char *upos;
Neal Norwitz76dc0812006-01-08 06:13:13 +0000148 size_t len = strlen(mode);
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000149
Georg Brandl7b90e162006-05-18 07:01:27 +0000150 if (!len) {
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000151 PyErr_SetString(PyExc_ValueError, "empty mode string");
152 return 1;
Georg Brandl7b90e162006-05-18 07:01:27 +0000153 }
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000154
Georg Brandl7b90e162006-05-18 07:01:27 +0000155 upos = strchr(mode, 'U');
156 if (upos) {
157 memmove(upos, upos+1, len-(upos-mode)); /* incl null char */
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000158
Georg Brandl7b90e162006-05-18 07:01:27 +0000159 if (mode[0] == 'w' || mode[0] == 'a') {
160 PyErr_Format(PyExc_ValueError, "universal newline "
161 "mode can only be used with modes "
162 "starting with 'r'");
163 return 1;
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000164 }
Georg Brandl7b90e162006-05-18 07:01:27 +0000165
166 if (mode[0] != 'r') {
167 memmove(mode+1, mode, strlen(mode)+1);
168 mode[0] = 'r';
169 }
170
171 if (!strchr(mode, 'b')) {
172 memmove(mode+2, mode+1, strlen(mode));
173 mode[1] = 'b';
174 }
175 } else if (mode[0] != 'r' && mode[0] != 'w' && mode[0] != 'a') {
176 PyErr_Format(PyExc_ValueError, "mode string must begin with "
177 "one of 'r', 'w', 'a' or 'U', not '%.200s'", mode);
178 return 1;
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000179 }
180
181 return 0;
182}
183
Tim Peters59c9a642001-09-13 05:38:56 +0000184static PyObject *
185open_the_file(PyFileObject *f, char *name, char *mode)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000186{
Georg Brandl7b90e162006-05-18 07:01:27 +0000187 char *newmode;
Tim Peters59c9a642001-09-13 05:38:56 +0000188 assert(f != NULL);
189 assert(PyFile_Check(f));
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000190#ifdef MS_WINDOWS
191 /* windows ignores the passed name in order to support Unicode */
192 assert(f->f_name != NULL);
193#else
Tim Peters59c9a642001-09-13 05:38:56 +0000194 assert(name != NULL);
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000195#endif
Tim Peters59c9a642001-09-13 05:38:56 +0000196 assert(mode != NULL);
Tim Peters44410012001-09-14 03:26:08 +0000197 assert(f->f_fp == NULL);
Tim Peters59c9a642001-09-13 05:38:56 +0000198
Georg Brandl7b90e162006-05-18 07:01:27 +0000199 /* probably need to replace 'U' by 'rb' */
200 newmode = PyMem_MALLOC(strlen(mode) + 3);
201 if (!newmode) {
202 PyErr_NoMemory();
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000203 return NULL;
Georg Brandl7b90e162006-05-18 07:01:27 +0000204 }
205 strcpy(newmode, mode);
206
207 if (sanitize_the_mode(newmode)) {
208 f = NULL;
209 goto cleanup;
210 }
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000211
Tim Peters8fa45672001-09-13 21:01:29 +0000212 /* rexec.py can't stop a user from getting the file() constructor --
213 all they have to do is get *any* file object f, and then do
214 type(f). Here we prevent them from doing damage with it. */
215 if (PyEval_GetRestricted()) {
216 PyErr_SetString(PyExc_IOError,
Jeremy Hylton8b735422002-08-14 21:01:41 +0000217 "file() constructor not accessible in restricted mode");
Georg Brandl7b90e162006-05-18 07:01:27 +0000218 f = NULL;
219 goto cleanup;
Tim Peters8fa45672001-09-13 21:01:29 +0000220 }
Tim Petersa27a1502001-11-09 20:59:14 +0000221 errno = 0;
Skip Montanaro51ffac62004-06-11 04:49:03 +0000222
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000223#ifdef MS_WINDOWS
Skip Montanaro51ffac62004-06-11 04:49:03 +0000224 if (PyUnicode_Check(f->f_name)) {
225 PyObject *wmode;
Georg Brandl7b90e162006-05-18 07:01:27 +0000226 wmode = PyUnicode_DecodeASCII(newmode, strlen(newmode), NULL);
Skip Montanaro51ffac62004-06-11 04:49:03 +0000227 if (f->f_name && wmode) {
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000228 Py_BEGIN_ALLOW_THREADS
Skip Montanaro51ffac62004-06-11 04:49:03 +0000229 /* PyUnicode_AS_UNICODE OK without thread
230 lock as it is a simple dereference. */
231 f->f_fp = _wfopen(PyUnicode_AS_UNICODE(f->f_name),
232 PyUnicode_AS_UNICODE(wmode));
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000233 Py_END_ALLOW_THREADS
234 }
Skip Montanaro51ffac62004-06-11 04:49:03 +0000235 Py_XDECREF(wmode);
Guido van Rossumff4949e1992-08-05 19:58:53 +0000236 }
Skip Montanaro51ffac62004-06-11 04:49:03 +0000237#endif
238 if (NULL == f->f_fp && NULL != name) {
239 Py_BEGIN_ALLOW_THREADS
Georg Brandl7b90e162006-05-18 07:01:27 +0000240 f->f_fp = fopen(name, newmode);
Skip Montanaro51ffac62004-06-11 04:49:03 +0000241 Py_END_ALLOW_THREADS
242 }
243
Guido van Rossuma08095a1991-02-13 23:25:27 +0000244 if (f->f_fp == NULL) {
Kristján Valur Jónsson74c3ea02006-07-03 14:59:05 +0000245#if defined _MSC_VER && (_MSC_VER < 1400 || !defined(__STDC_SECURE_LIB__))
Tim Peters2ea91112002-04-08 04:13:12 +0000246 /* MSVC 6 (Microsoft) leaves errno at 0 for bad mode strings,
247 * across all Windows flavors. When it sets EINVAL varies
248 * across Windows flavors, the exact conditions aren't
249 * documented, and the answer lies in the OS's implementation
250 * of Win32's CreateFile function (whose source is secret).
251 * Seems the best we can do is map EINVAL to ENOENT.
Kristján Valur Jónssonf6083172006-06-12 15:45:12 +0000252 * Starting with Visual Studio .NET 2005, EINVAL is correctly
253 * set by our CRT error handler (set in exceptions.c.)
Tim Peters2ea91112002-04-08 04:13:12 +0000254 */
255 if (errno == 0) /* bad mode string */
256 errno = EINVAL;
257 else if (errno == EINVAL) /* unknown, but not a mode string */
258 errno = ENOENT;
259#endif
Jeremy Hylton41c83212001-11-09 16:17:24 +0000260 if (errno == EINVAL)
Tim Peters2ea91112002-04-08 04:13:12 +0000261 PyErr_Format(PyExc_IOError, "invalid mode: %s",
Jeremy Hylton41c83212001-11-09 16:17:24 +0000262 mode);
263 else
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000264 PyErr_SetFromErrnoWithFilenameObject(PyExc_IOError, f->f_name);
Tim Peters59c9a642001-09-13 05:38:56 +0000265 f = NULL;
266 }
Tim Peters2ea91112002-04-08 04:13:12 +0000267 if (f != NULL)
Neil Schemenauered19b882002-03-23 02:06:50 +0000268 f = dircheck(f);
Georg Brandl7b90e162006-05-18 07:01:27 +0000269
270cleanup:
271 PyMem_FREE(newmode);
272
Tim Peters59c9a642001-09-13 05:38:56 +0000273 return (PyObject *)f;
274}
275
276PyObject *
277PyFile_FromFile(FILE *fp, char *name, char *mode, int (*close)(FILE *))
278{
Tim Peters44410012001-09-14 03:26:08 +0000279 PyFileObject *f = (PyFileObject *)PyFile_Type.tp_new(&PyFile_Type,
280 NULL, NULL);
Tim Peters59c9a642001-09-13 05:38:56 +0000281 if (f != NULL) {
Neal Norwitzb337bb52006-07-17 00:55:45 +0000282 PyObject *o_name = PyString_FromString(name);
283 if (o_name == NULL)
284 return NULL;
Nicholas Bastinabce8a62004-03-21 20:24:07 +0000285 if (fill_file_fields(f, fp, o_name, mode, close) == NULL) {
Tim Peters59c9a642001-09-13 05:38:56 +0000286 Py_DECREF(f);
287 f = NULL;
288 }
Nicholas Bastinabce8a62004-03-21 20:24:07 +0000289 Py_DECREF(o_name);
Tim Peters59c9a642001-09-13 05:38:56 +0000290 }
291 return (PyObject *) f;
292}
293
294PyObject *
295PyFile_FromString(char *name, char *mode)
296{
297 extern int fclose(FILE *);
298 PyFileObject *f;
299
300 f = (PyFileObject *)PyFile_FromFile((FILE *)NULL, name, mode, fclose);
301 if (f != NULL) {
302 if (open_the_file(f, name, mode) == NULL) {
303 Py_DECREF(f);
304 f = NULL;
305 }
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000306 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000307 return (PyObject *)f;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000308}
309
Guido van Rossumb6775db1994-08-01 11:34:53 +0000310void
Fred Drakefd99de62000-07-09 05:02:18 +0000311PyFile_SetBufSize(PyObject *f, int bufsize)
Guido van Rossumb6775db1994-08-01 11:34:53 +0000312{
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000313 PyFileObject *file = (PyFileObject *)f;
Guido van Rossumb6775db1994-08-01 11:34:53 +0000314 if (bufsize >= 0) {
Guido van Rossumb6775db1994-08-01 11:34:53 +0000315 int type;
316 switch (bufsize) {
317 case 0:
318 type = _IONBF;
319 break;
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000320#ifdef HAVE_SETVBUF
Guido van Rossumb6775db1994-08-01 11:34:53 +0000321 case 1:
322 type = _IOLBF;
323 bufsize = BUFSIZ;
324 break;
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000325#endif
Guido van Rossumb6775db1994-08-01 11:34:53 +0000326 default:
327 type = _IOFBF;
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000328#ifndef HAVE_SETVBUF
329 bufsize = BUFSIZ;
330#endif
331 break;
Guido van Rossumb6775db1994-08-01 11:34:53 +0000332 }
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000333 fflush(file->f_fp);
334 if (type == _IONBF) {
335 PyMem_Free(file->f_setbuf);
336 file->f_setbuf = NULL;
337 } else {
Anthony Baxter377be112006-04-11 06:54:30 +0000338 file->f_setbuf = (char *)PyMem_Realloc(file->f_setbuf,
339 bufsize);
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000340 }
341#ifdef HAVE_SETVBUF
342 setvbuf(file->f_fp, file->f_setbuf, type, bufsize);
Guido van Rossumf8b4de01998-03-06 15:32:40 +0000343#else /* !HAVE_SETVBUF */
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000344 setbuf(file->f_fp, file->f_setbuf);
Guido van Rossumf8b4de01998-03-06 15:32:40 +0000345#endif /* !HAVE_SETVBUF */
Guido van Rossumb6775db1994-08-01 11:34:53 +0000346 }
347}
348
Martin v. Löwis5467d4c2003-05-10 07:10:12 +0000349/* Set the encoding used to output Unicode strings.
350 Returh 1 on success, 0 on failure. */
351
352int
353PyFile_SetEncoding(PyObject *f, const char *enc)
354{
355 PyFileObject *file = (PyFileObject*)f;
356 PyObject *str = PyString_FromString(enc);
357 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;
Neal Norwitzfc28e0d2006-07-16 02:32:03 +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",
Neal Norwitzfc28e0d2006-07-16 02:32:03 +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;
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000543 PyObject *offobj;
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;
551#if !defined(HAVE_LARGEFILE_SUPPORT)
552 offset = PyInt_AsLong(offobj);
553#else
554 offset = PyLong_Check(offobj) ?
555 PyLong_AsLongLong(offobj) : PyInt_AsLong(offobj);
556#endif
557 if (PyErr_Occurred())
Guido van Rossum88303191999-01-04 17:22:18 +0000558 return NULL;
Tim Peters86821b22001-01-07 21:19:34 +0000559
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000560 Py_BEGIN_ALLOW_THREADS
Guido van Rossumce5ba841991-03-06 13:06:18 +0000561 errno = 0;
Trent Mickf29f47b2000-08-11 19:02:59 +0000562 ret = _portable_fseek(f->f_fp, offset, whence);
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000563 Py_END_ALLOW_THREADS
Trent Mickf29f47b2000-08-11 19:02:59 +0000564
Guido van Rossumff4949e1992-08-05 19:58:53 +0000565 if (ret != 0) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000566 PyErr_SetFromErrno(PyExc_IOError);
Guido van Rossumfebd5511992-03-04 16:39:24 +0000567 clearerr(f->f_fp);
568 return NULL;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000569 }
Jack Jansen7b8c7542002-04-14 20:12:41 +0000570 f->f_skipnextlf = 0;
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000571 Py_INCREF(Py_None);
572 return Py_None;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000573}
574
Trent Mickf29f47b2000-08-11 19:02:59 +0000575
Guido van Rossumd7047b31995-01-02 19:07:15 +0000576#ifdef HAVE_FTRUNCATE
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000577static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000578file_truncate(PyFileObject *f, PyObject *args)
Guido van Rossumd7047b31995-01-02 19:07:15 +0000579{
Guido van Rossum4f53da02001-03-01 18:26:53 +0000580 Py_off_t newsize;
Tim Petersf1827cf2003-09-07 03:30:18 +0000581 PyObject *newsizeobj = NULL;
582 Py_off_t initialpos;
583 int ret;
Tim Peters86821b22001-01-07 21:19:34 +0000584
Guido van Rossumd7047b31995-01-02 19:07:15 +0000585 if (f->f_fp == NULL)
586 return err_closed();
Raymond Hettingerea3fdf42002-12-29 16:33:45 +0000587 if (!PyArg_UnpackTuple(args, "truncate", 0, 1, &newsizeobj))
Guido van Rossum88303191999-01-04 17:22:18 +0000588 return NULL;
Tim Petersfb05db22002-03-11 00:24:00 +0000589
Tim Petersf1827cf2003-09-07 03:30:18 +0000590 /* Get current file position. If the file happens to be open for
591 * update and the last operation was an input operation, C doesn't
592 * define what the later fflush() will do, but we promise truncate()
593 * won't change the current position (and fflush() *does* change it
594 * then at least on Windows). The easiest thing is to capture
595 * current pos now and seek back to it at the end.
596 */
597 Py_BEGIN_ALLOW_THREADS
598 errno = 0;
599 initialpos = _portable_ftell(f->f_fp);
600 Py_END_ALLOW_THREADS
601 if (initialpos == -1)
602 goto onioerror;
603
Tim Petersfb05db22002-03-11 00:24:00 +0000604 /* Set newsize to current postion if newsizeobj NULL, else to the
Tim Petersf1827cf2003-09-07 03:30:18 +0000605 * specified value.
606 */
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000607 if (newsizeobj != NULL) {
608#if !defined(HAVE_LARGEFILE_SUPPORT)
609 newsize = PyInt_AsLong(newsizeobj);
610#else
611 newsize = PyLong_Check(newsizeobj) ?
612 PyLong_AsLongLong(newsizeobj) :
613 PyInt_AsLong(newsizeobj);
614#endif
615 if (PyErr_Occurred())
616 return NULL;
Tim Petersfb05db22002-03-11 00:24:00 +0000617 }
Tim Petersf1827cf2003-09-07 03:30:18 +0000618 else /* default to current position */
619 newsize = initialpos;
Tim Petersfb05db22002-03-11 00:24:00 +0000620
Tim Petersf1827cf2003-09-07 03:30:18 +0000621 /* Flush the stream. We're mixing stream-level I/O with lower-level
622 * I/O, and a flush may be necessary to synch both platform views
623 * of the current file state.
624 */
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000625 Py_BEGIN_ALLOW_THREADS
Guido van Rossumd7047b31995-01-02 19:07:15 +0000626 errno = 0;
627 ret = fflush(f->f_fp);
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000628 Py_END_ALLOW_THREADS
Tim Petersfb05db22002-03-11 00:24:00 +0000629 if (ret != 0)
630 goto onioerror;
Trent Mickf29f47b2000-08-11 19:02:59 +0000631
Martin v. Löwis6238d2b2002-06-30 15:26:10 +0000632#ifdef MS_WINDOWS
Tim Petersfb05db22002-03-11 00:24:00 +0000633 /* MS _chsize doesn't work if newsize doesn't fit in 32 bits,
Tim Peters8f01b682002-03-12 03:04:44 +0000634 so don't even try using it. */
Tim Petersfb05db22002-03-11 00:24:00 +0000635 {
Tim Petersfb05db22002-03-11 00:24:00 +0000636 HANDLE hFile;
Tim Petersfb05db22002-03-11 00:24:00 +0000637
Tim Petersf1827cf2003-09-07 03:30:18 +0000638 /* Have to move current pos to desired endpoint on Windows. */
639 Py_BEGIN_ALLOW_THREADS
640 errno = 0;
641 ret = _portable_fseek(f->f_fp, newsize, SEEK_SET) != 0;
642 Py_END_ALLOW_THREADS
643 if (ret)
644 goto onioerror;
Tim Petersfb05db22002-03-11 00:24:00 +0000645
Tim Peters8f01b682002-03-12 03:04:44 +0000646 /* Truncate. Note that this may grow the file! */
647 Py_BEGIN_ALLOW_THREADS
648 errno = 0;
649 hFile = (HANDLE)_get_osfhandle(fileno(f->f_fp));
Tim Petersf1827cf2003-09-07 03:30:18 +0000650 ret = hFile == (HANDLE)-1;
651 if (ret == 0) {
652 ret = SetEndOfFile(hFile) == 0;
653 if (ret)
Tim Peters8f01b682002-03-12 03:04:44 +0000654 errno = EACCES;
655 }
656 Py_END_ALLOW_THREADS
Tim Petersf1827cf2003-09-07 03:30:18 +0000657 if (ret)
Tim Peters8f01b682002-03-12 03:04:44 +0000658 goto onioerror;
Guido van Rossumd7047b31995-01-02 19:07:15 +0000659 }
Trent Mickf29f47b2000-08-11 19:02:59 +0000660#else
661 Py_BEGIN_ALLOW_THREADS
662 errno = 0;
663 ret = ftruncate(fileno(f->f_fp), newsize);
664 Py_END_ALLOW_THREADS
Tim Petersf1827cf2003-09-07 03:30:18 +0000665 if (ret != 0)
666 goto onioerror;
Martin v. Löwis6238d2b2002-06-30 15:26:10 +0000667#endif /* !MS_WINDOWS */
Tim Peters86821b22001-01-07 21:19:34 +0000668
Tim Petersf1827cf2003-09-07 03:30:18 +0000669 /* Restore original file position. */
670 Py_BEGIN_ALLOW_THREADS
671 errno = 0;
672 ret = _portable_fseek(f->f_fp, initialpos, SEEK_SET) != 0;
673 Py_END_ALLOW_THREADS
674 if (ret)
675 goto onioerror;
676
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000677 Py_INCREF(Py_None);
678 return Py_None;
Trent Mickf29f47b2000-08-11 19:02:59 +0000679
680onioerror:
681 PyErr_SetFromErrno(PyExc_IOError);
682 clearerr(f->f_fp);
683 return NULL;
Guido van Rossumd7047b31995-01-02 19:07:15 +0000684}
685#endif /* HAVE_FTRUNCATE */
686
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000687static PyObject *
Martin v. Löwise3eb1f22001-08-16 13:15:00 +0000688file_tell(PyFileObject *f)
Guido van Rossumce5ba841991-03-06 13:06:18 +0000689{
Guido van Rossum4f53da02001-03-01 18:26:53 +0000690 Py_off_t pos;
Trent Mickf29f47b2000-08-11 19:02:59 +0000691
Guido van Rossumd7297e61992-07-06 14:19:26 +0000692 if (f->f_fp == NULL)
693 return err_closed();
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000694 Py_BEGIN_ALLOW_THREADS
Guido van Rossumce5ba841991-03-06 13:06:18 +0000695 errno = 0;
Trent Mickf29f47b2000-08-11 19:02:59 +0000696 pos = _portable_ftell(f->f_fp);
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000697 Py_END_ALLOW_THREADS
Trent Mickf29f47b2000-08-11 19:02:59 +0000698 if (pos == -1) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000699 PyErr_SetFromErrno(PyExc_IOError);
Guido van Rossumfebd5511992-03-04 16:39:24 +0000700 clearerr(f->f_fp);
701 return NULL;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000702 }
Jack Jansen7b8c7542002-04-14 20:12:41 +0000703 if (f->f_skipnextlf) {
704 int c;
705 c = GETC(f->f_fp);
706 if (c == '\n') {
707 pos++;
708 f->f_skipnextlf = 0;
709 } else if (c != EOF) ungetc(c, f->f_fp);
710 }
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000711#if !defined(HAVE_LARGEFILE_SUPPORT)
Trent Mickf29f47b2000-08-11 19:02:59 +0000712 return PyInt_FromLong(pos);
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000713#else
Trent Mickf29f47b2000-08-11 19:02:59 +0000714 return PyLong_FromLongLong(pos);
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000715#endif
Guido van Rossumce5ba841991-03-06 13:06:18 +0000716}
717
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000718static PyObject *
Martin v. Löwise3eb1f22001-08-16 13:15:00 +0000719file_fileno(PyFileObject *f)
Guido van Rossumed233a51992-06-23 09:07:03 +0000720{
Guido van Rossumd7297e61992-07-06 14:19:26 +0000721 if (f->f_fp == NULL)
722 return err_closed();
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000723 return PyInt_FromLong((long) fileno(f->f_fp));
Guido van Rossumed233a51992-06-23 09:07:03 +0000724}
725
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000726static PyObject *
Martin v. Löwise3eb1f22001-08-16 13:15:00 +0000727file_flush(PyFileObject *f)
Guido van Rossumce5ba841991-03-06 13:06:18 +0000728{
Guido van Rossumff4949e1992-08-05 19:58:53 +0000729 int res;
Tim Peters86821b22001-01-07 21:19:34 +0000730
Guido van Rossumd7297e61992-07-06 14:19:26 +0000731 if (f->f_fp == NULL)
732 return err_closed();
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000733 Py_BEGIN_ALLOW_THREADS
Guido van Rossumce5ba841991-03-06 13:06:18 +0000734 errno = 0;
Guido van Rossumff4949e1992-08-05 19:58:53 +0000735 res = fflush(f->f_fp);
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000736 Py_END_ALLOW_THREADS
Guido van Rossumff4949e1992-08-05 19:58:53 +0000737 if (res != 0) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000738 PyErr_SetFromErrno(PyExc_IOError);
Guido van Rossumfebd5511992-03-04 16:39:24 +0000739 clearerr(f->f_fp);
740 return NULL;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000741 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000742 Py_INCREF(Py_None);
743 return Py_None;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000744}
745
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000746static PyObject *
Martin v. Löwise3eb1f22001-08-16 13:15:00 +0000747file_isatty(PyFileObject *f)
Guido van Rossuma1ab7fa1991-06-04 19:37:39 +0000748{
Guido van Rossumff4949e1992-08-05 19:58:53 +0000749 long res;
Guido van Rossumd7297e61992-07-06 14:19:26 +0000750 if (f->f_fp == NULL)
751 return err_closed();
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000752 Py_BEGIN_ALLOW_THREADS
Guido van Rossumff4949e1992-08-05 19:58:53 +0000753 res = isatty((int)fileno(f->f_fp));
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000754 Py_END_ALLOW_THREADS
Guido van Rossum7f7666f2002-04-07 06:28:00 +0000755 return PyBool_FromLong(res);
Guido van Rossuma1ab7fa1991-06-04 19:37:39 +0000756}
757
Guido van Rossumff7e83d1999-08-27 20:39:37 +0000758
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000759#if BUFSIZ < 8192
760#define SMALLCHUNK 8192
761#else
762#define SMALLCHUNK BUFSIZ
763#endif
764
Guido van Rossum3c259041999-01-14 19:00:14 +0000765#if SIZEOF_INT < 4
766#define BIGCHUNK (512 * 32)
767#else
768#define BIGCHUNK (512 * 1024)
769#endif
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000770
771static size_t
Fred Drakefd99de62000-07-09 05:02:18 +0000772new_buffersize(PyFileObject *f, size_t currentsize)
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000773{
774#ifdef HAVE_FSTAT
Fred Drake1bc8fab2001-07-19 21:49:38 +0000775 off_t pos, end;
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000776 struct stat st;
777 if (fstat(fileno(f->f_fp), &st) == 0) {
778 end = st.st_size;
Guido van Rossumcada2931998-12-11 20:44:56 +0000779 /* The following is not a bug: we really need to call lseek()
780 *and* ftell(). The reason is that some stdio libraries
781 mistakenly flush their buffer when ftell() is called and
782 the lseek() call it makes fails, thereby throwing away
783 data that cannot be recovered in any way. To avoid this,
784 we first test lseek(), and only call ftell() if lseek()
785 works. We can't use the lseek() value either, because we
786 need to take the amount of buffered data into account.
787 (Yet another reason why stdio stinks. :-) */
Guido van Rossum91aaa921998-05-05 22:21:35 +0000788 pos = lseek(fileno(f->f_fp), 0L, SEEK_CUR);
Jack Jansen2771b5b2001-10-10 22:03:27 +0000789 if (pos >= 0) {
Guido van Rossum91aaa921998-05-05 22:21:35 +0000790 pos = ftell(f->f_fp);
Jack Jansen2771b5b2001-10-10 22:03:27 +0000791 }
Guido van Rossumd30dc0a1998-04-27 19:01:08 +0000792 if (pos < 0)
793 clearerr(f->f_fp);
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000794 if (end > pos && pos >= 0)
Guido van Rossumcada2931998-12-11 20:44:56 +0000795 return currentsize + end - pos + 1;
Guido van Rossumdcb5e7f1998-03-03 22:36:10 +0000796 /* Add 1 so if the file were to grow we'd notice. */
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000797 }
798#endif
799 if (currentsize > SMALLCHUNK) {
800 /* Keep doubling until we reach BIGCHUNK;
801 then keep adding BIGCHUNK. */
802 if (currentsize <= BIGCHUNK)
803 return currentsize + currentsize;
804 else
805 return currentsize + BIGCHUNK;
806 }
807 return currentsize + SMALLCHUNK;
808}
809
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +0000810#if defined(EWOULDBLOCK) && defined(EAGAIN) && EWOULDBLOCK != EAGAIN
811#define BLOCKED_ERRNO(x) ((x) == EWOULDBLOCK || (x) == EAGAIN)
812#else
813#ifdef EWOULDBLOCK
814#define BLOCKED_ERRNO(x) ((x) == EWOULDBLOCK)
815#else
816#ifdef EAGAIN
817#define BLOCKED_ERRNO(x) ((x) == EAGAIN)
818#else
819#define BLOCKED_ERRNO(x) 0
820#endif
821#endif
822#endif
823
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000824static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000825file_read(PyFileObject *f, PyObject *args)
Guido van Rossumce5ba841991-03-06 13:06:18 +0000826{
Guido van Rossum789a1611997-05-10 22:33:55 +0000827 long bytesrequested = -1;
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000828 size_t bytesread, buffersize, chunksize;
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000829 PyObject *v;
Tim Peters86821b22001-01-07 21:19:34 +0000830
Guido van Rossumd7297e61992-07-06 14:19:26 +0000831 if (f->f_fp == NULL)
832 return err_closed();
Thomas Woutersc45251a2006-02-12 11:53:32 +0000833 /* refuse to mix with f.next() */
834 if (f->f_buf != NULL &&
835 (f->f_bufend - f->f_bufptr) > 0 &&
836 f->f_buf[0] != '\0')
837 return err_iterbuffered();
Guido van Rossum43713e52000-02-29 13:59:29 +0000838 if (!PyArg_ParseTuple(args, "|l:read", &bytesrequested))
Guido van Rossum789a1611997-05-10 22:33:55 +0000839 return NULL;
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000840 if (bytesrequested < 0)
Guido van Rossumff1ccbf1999-04-10 15:48:23 +0000841 buffersize = new_buffersize(f, (size_t)0);
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000842 else
843 buffersize = bytesrequested;
Martin v. Löwis2a190742006-04-13 07:37:25 +0000844 if (buffersize > PY_SSIZE_T_MAX) {
Trent Mickf29f47b2000-08-11 19:02:59 +0000845 PyErr_SetString(PyExc_OverflowError,
Jeremy Hylton8b735422002-08-14 21:01:41 +0000846 "requested number of bytes is more than a Python string can hold");
Trent Mickf29f47b2000-08-11 19:02:59 +0000847 return NULL;
848 }
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000849 v = PyString_FromStringAndSize((char *)NULL, buffersize);
Guido van Rossum3f5da241990-12-20 15:06:42 +0000850 if (v == NULL)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000851 return NULL;
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000852 bytesread = 0;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000853 for (;;) {
Guido van Rossum6263d541997-05-10 22:07:25 +0000854 Py_BEGIN_ALLOW_THREADS
855 errno = 0;
Jack Jansen7b8c7542002-04-14 20:12:41 +0000856 chunksize = Py_UniversalNewlineFread(BUF(v) + bytesread,
Jeremy Hylton8b735422002-08-14 21:01:41 +0000857 buffersize - bytesread, f->f_fp, (PyObject *)f);
Guido van Rossum6263d541997-05-10 22:07:25 +0000858 Py_END_ALLOW_THREADS
859 if (chunksize == 0) {
860 if (!ferror(f->f_fp))
861 break;
Guido van Rossum6263d541997-05-10 22:07:25 +0000862 clearerr(f->f_fp);
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +0000863 /* When in non-blocking mode, data shouldn't
864 * be discarded if a blocking signal was
865 * received. That will also happen if
866 * chunksize != 0, but bytesread < buffersize. */
867 if (bytesread > 0 && BLOCKED_ERRNO(errno))
868 break;
869 PyErr_SetFromErrno(PyExc_IOError);
Guido van Rossum6263d541997-05-10 22:07:25 +0000870 Py_DECREF(v);
871 return NULL;
872 }
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000873 bytesread += chunksize;
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +0000874 if (bytesread < buffersize) {
875 clearerr(f->f_fp);
Guido van Rossumce5ba841991-03-06 13:06:18 +0000876 break;
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +0000877 }
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000878 if (bytesrequested < 0) {
Guido van Rossumcada2931998-12-11 20:44:56 +0000879 buffersize = new_buffersize(f, buffersize);
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000880 if (_PyString_Resize(&v, buffersize) < 0)
Guido van Rossumce5ba841991-03-06 13:06:18 +0000881 return NULL;
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +0000882 } else {
Gustavo Niemeyera080be82002-12-17 17:48:00 +0000883 /* Got what was requested. */
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +0000884 break;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000885 }
886 }
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000887 if (bytesread != buffersize)
888 _PyString_Resize(&v, bytesread);
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000889 return v;
890}
891
Guido van Rossumfdf95dd1997-05-05 22:15:02 +0000892static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000893file_readinto(PyFileObject *f, PyObject *args)
Guido van Rossumfdf95dd1997-05-05 22:15:02 +0000894{
895 char *ptr;
Martin v. Löwis18e16552006-02-15 17:27:45 +0000896 Py_ssize_t ntodo;
897 Py_ssize_t ndone, nnow;
Tim Peters86821b22001-01-07 21:19:34 +0000898
Guido van Rossumfdf95dd1997-05-05 22:15:02 +0000899 if (f->f_fp == NULL)
900 return err_closed();
Thomas Woutersc45251a2006-02-12 11:53:32 +0000901 /* refuse to mix with f.next() */
902 if (f->f_buf != NULL &&
903 (f->f_bufend - f->f_bufptr) > 0 &&
904 f->f_buf[0] != '\0')
905 return err_iterbuffered();
Neal Norwitz62f5a9d2002-04-01 00:09:00 +0000906 if (!PyArg_ParseTuple(args, "w#", &ptr, &ntodo))
Guido van Rossumfdf95dd1997-05-05 22:15:02 +0000907 return NULL;
908 ndone = 0;
Guido van Rossum6263d541997-05-10 22:07:25 +0000909 while (ntodo > 0) {
910 Py_BEGIN_ALLOW_THREADS
911 errno = 0;
Tim Petersf1827cf2003-09-07 03:30:18 +0000912 nnow = Py_UniversalNewlineFread(ptr+ndone, ntodo, f->f_fp,
Jeremy Hylton8b735422002-08-14 21:01:41 +0000913 (PyObject *)f);
Guido van Rossum6263d541997-05-10 22:07:25 +0000914 Py_END_ALLOW_THREADS
915 if (nnow == 0) {
916 if (!ferror(f->f_fp))
917 break;
Guido van Rossumfdf95dd1997-05-05 22:15:02 +0000918 PyErr_SetFromErrno(PyExc_IOError);
919 clearerr(f->f_fp);
920 return NULL;
921 }
Guido van Rossumfdf95dd1997-05-05 22:15:02 +0000922 ndone += nnow;
923 ntodo -= nnow;
924 }
Trent Mickf29f47b2000-08-11 19:02:59 +0000925 return PyInt_FromLong((long)ndone);
Guido van Rossumfdf95dd1997-05-05 22:15:02 +0000926}
927
Tim Peters86821b22001-01-07 21:19:34 +0000928/**************************************************************************
Tim Petersf29b64d2001-01-15 06:33:19 +0000929Routine to get next line using platform fgets().
Tim Peters86821b22001-01-07 21:19:34 +0000930
931Under MSVC 6:
932
Tim Peters1c733232001-01-08 04:02:07 +0000933+ MS threadsafe getc is very slow (multiple layers of function calls before+
934 after each character, to lock+unlock the stream).
935+ The stream-locking functions are MS-internal -- can't access them from user
936 code.
937+ There's nothing Tim could find in the MS C or platform SDK libraries that
938 can worm around this.
Tim Peters86821b22001-01-07 21:19:34 +0000939+ MS fgets locks/unlocks only once per line; it's the only hook we have.
940
941So we use fgets for speed(!), despite that it's painful.
942
943MS realloc is also slow.
944
Tim Petersf29b64d2001-01-15 06:33:19 +0000945Reports from other platforms on this method vs getc_unlocked (which MS doesn't
946have):
947 Linux a wash
948 Solaris a wash
949 Tru64 Unix getline_via_fgets significantly faster
Tim Peters86821b22001-01-07 21:19:34 +0000950
Tim Petersf29b64d2001-01-15 06:33:19 +0000951CAUTION: The C std isn't clear about this: in those cases where fgets
952writes something into the buffer, can it write into any position beyond the
953required trailing null byte? MSVC 6 fgets does not, and no platform is (yet)
954known on which it does; and it would be a strange way to code fgets. Still,
955getline_via_fgets may not work correctly if it does. The std test
956test_bufio.py should fail if platform fgets() routinely writes beyond the
957trailing null byte. #define DONT_USE_FGETS_IN_GETLINE to disable this code.
Tim Peters86821b22001-01-07 21:19:34 +0000958**************************************************************************/
959
Tim Petersf29b64d2001-01-15 06:33:19 +0000960/* Use this routine if told to, or by default on non-get_unlocked()
961 * platforms unless told not to. Yikes! Let's spell that out:
962 * On a platform with getc_unlocked():
963 * By default, use getc_unlocked().
964 * If you want to use fgets() instead, #define USE_FGETS_IN_GETLINE.
965 * On a platform without getc_unlocked():
966 * By default, use fgets().
967 * If you don't want to use fgets(), #define DONT_USE_FGETS_IN_GETLINE.
968 */
969#if !defined(USE_FGETS_IN_GETLINE) && !defined(HAVE_GETC_UNLOCKED)
970#define USE_FGETS_IN_GETLINE
Tim Peters86821b22001-01-07 21:19:34 +0000971#endif
972
Tim Petersf29b64d2001-01-15 06:33:19 +0000973#if defined(DONT_USE_FGETS_IN_GETLINE) && defined(USE_FGETS_IN_GETLINE)
974#undef USE_FGETS_IN_GETLINE
975#endif
976
977#ifdef USE_FGETS_IN_GETLINE
Tim Peters86821b22001-01-07 21:19:34 +0000978static PyObject*
Tim Petersf29b64d2001-01-15 06:33:19 +0000979getline_via_fgets(FILE *fp)
Tim Peters86821b22001-01-07 21:19:34 +0000980{
Tim Peters15b83852001-01-08 00:53:12 +0000981/* INITBUFSIZE is the maximum line length that lets us get away with the fast
Tim Peters142297a2001-01-15 10:36:56 +0000982 * no-realloc, one-fgets()-call path. Boosting it isn't free, because we have
983 * to fill this much of the buffer with a known value in order to figure out
984 * how much of the buffer fgets() overwrites. So if INITBUFSIZE is larger
985 * than "most" lines, we waste time filling unused buffer slots. 100 is
986 * surely adequate for most peoples' email archives, chewing over source code,
987 * etc -- "regular old text files".
988 * MAXBUFSIZE is the maximum line length that lets us get away with the less
989 * fast (but still zippy) no-realloc, two-fgets()-call path. See above for
990 * cautions about boosting that. 300 was chosen because the worst real-life
991 * text-crunching job reported on Python-Dev was a mail-log crawler where over
992 * half the lines were 254 chars.
Tim Peters15b83852001-01-08 00:53:12 +0000993 */
Tim Peters142297a2001-01-15 10:36:56 +0000994#define INITBUFSIZE 100
995#define MAXBUFSIZE 300
Tim Peters142297a2001-01-15 10:36:56 +0000996 char* p; /* temp */
997 char buf[MAXBUFSIZE];
Tim Peters86821b22001-01-07 21:19:34 +0000998 PyObject* v; /* the string object result */
Tim Peters86821b22001-01-07 21:19:34 +0000999 char* pvfree; /* address of next free slot */
1000 char* pvend; /* address one beyond last free slot */
Tim Peters142297a2001-01-15 10:36:56 +00001001 size_t nfree; /* # of free buffer slots; pvend-pvfree */
1002 size_t total_v_size; /* total # of slots in buffer */
Tim Petersddea2082002-03-23 10:03:50 +00001003 size_t increment; /* amount to increment the buffer */
Tim Peters86821b22001-01-07 21:19:34 +00001004
Tim Peters15b83852001-01-08 00:53:12 +00001005 /* Optimize for normal case: avoid _PyString_Resize if at all
Tim Peters142297a2001-01-15 10:36:56 +00001006 * possible via first reading into stack buffer "buf".
Tim Peters15b83852001-01-08 00:53:12 +00001007 */
Tim Peters142297a2001-01-15 10:36:56 +00001008 total_v_size = INITBUFSIZE; /* start small and pray */
1009 pvfree = buf;
1010 for (;;) {
1011 Py_BEGIN_ALLOW_THREADS
1012 pvend = buf + total_v_size;
1013 nfree = pvend - pvfree;
1014 memset(pvfree, '\n', nfree);
Martin v. Löwis18e16552006-02-15 17:27:45 +00001015 assert(nfree < INT_MAX); /* Should be atmost MAXBUFSIZE */
1016 p = fgets(pvfree, (int)nfree, fp);
Tim Peters142297a2001-01-15 10:36:56 +00001017 Py_END_ALLOW_THREADS
Tim Peters15b83852001-01-08 00:53:12 +00001018
Tim Peters142297a2001-01-15 10:36:56 +00001019 if (p == NULL) {
1020 clearerr(fp);
1021 if (PyErr_CheckSignals())
1022 return NULL;
1023 v = PyString_FromStringAndSize(buf, pvfree - buf);
Tim Peters86821b22001-01-07 21:19:34 +00001024 return v;
1025 }
Tim Peters142297a2001-01-15 10:36:56 +00001026 /* fgets read *something* */
1027 p = memchr(pvfree, '\n', nfree);
1028 if (p != NULL) {
1029 /* Did the \n come from fgets or from us?
1030 * Since fgets stops at the first \n, and then writes
1031 * \0, if it's from fgets a \0 must be next. But if
1032 * that's so, it could not have come from us, since
1033 * the \n's we filled the buffer with have only more
1034 * \n's to the right.
1035 */
1036 if (p+1 < pvend && *(p+1) == '\0') {
1037 /* It's from fgets: we win! In particular,
1038 * we haven't done any mallocs yet, and can
1039 * build the final result on the first try.
1040 */
1041 ++p; /* include \n from fgets */
1042 }
1043 else {
1044 /* Must be from us: fgets didn't fill the
1045 * buffer and didn't find a newline, so it
1046 * must be the last and newline-free line of
1047 * the file.
1048 */
1049 assert(p > pvfree && *(p-1) == '\0');
1050 --p; /* don't include \0 from fgets */
1051 }
1052 v = PyString_FromStringAndSize(buf, p - buf);
1053 return v;
1054 }
1055 /* yuck: fgets overwrote all the newlines, i.e. the entire
1056 * buffer. So this line isn't over yet, or maybe it is but
1057 * we're exactly at EOF. If we haven't already, try using the
1058 * rest of the stack buffer.
Tim Peters86821b22001-01-07 21:19:34 +00001059 */
Tim Peters142297a2001-01-15 10:36:56 +00001060 assert(*(pvend-1) == '\0');
1061 if (pvfree == buf) {
1062 pvfree = pvend - 1; /* overwrite trailing null */
1063 total_v_size = MAXBUFSIZE;
1064 }
1065 else
1066 break;
Tim Peters86821b22001-01-07 21:19:34 +00001067 }
Tim Peters142297a2001-01-15 10:36:56 +00001068
1069 /* The stack buffer isn't big enough; malloc a string object and read
1070 * into its buffer.
Tim Peters15b83852001-01-08 00:53:12 +00001071 */
Tim Petersddea2082002-03-23 10:03:50 +00001072 total_v_size = MAXBUFSIZE << 1;
Tim Peters1c733232001-01-08 04:02:07 +00001073 v = PyString_FromStringAndSize((char*)NULL, (int)total_v_size);
Tim Peters15b83852001-01-08 00:53:12 +00001074 if (v == NULL)
1075 return v;
1076 /* copy over everything except the last null byte */
Tim Peters142297a2001-01-15 10:36:56 +00001077 memcpy(BUF(v), buf, MAXBUFSIZE-1);
1078 pvfree = BUF(v) + MAXBUFSIZE - 1;
Tim Peters86821b22001-01-07 21:19:34 +00001079
1080 /* Keep reading stuff into v; if it ever ends successfully, break
Tim Peters15b83852001-01-08 00:53:12 +00001081 * after setting p one beyond the end of the line. The code here is
1082 * very much like the code above, except reads into v's buffer; see
1083 * the code above for detailed comments about the logic.
Tim Peters86821b22001-01-07 21:19:34 +00001084 */
1085 for (;;) {
Tim Peters86821b22001-01-07 21:19:34 +00001086 Py_BEGIN_ALLOW_THREADS
1087 pvend = BUF(v) + total_v_size;
1088 nfree = pvend - pvfree;
1089 memset(pvfree, '\n', nfree);
Martin v. Löwis18e16552006-02-15 17:27:45 +00001090 assert(nfree < INT_MAX);
1091 p = fgets(pvfree, (int)nfree, fp);
Tim Peters86821b22001-01-07 21:19:34 +00001092 Py_END_ALLOW_THREADS
1093
1094 if (p == NULL) {
1095 clearerr(fp);
1096 if (PyErr_CheckSignals()) {
1097 Py_DECREF(v);
1098 return NULL;
1099 }
1100 p = pvfree;
1101 break;
1102 }
Tim Peters86821b22001-01-07 21:19:34 +00001103 p = memchr(pvfree, '\n', nfree);
1104 if (p != NULL) {
1105 if (p+1 < pvend && *(p+1) == '\0') {
1106 /* \n came from fgets */
1107 ++p;
1108 break;
1109 }
1110 /* \n came from us; last line of file, no newline */
1111 assert(p > pvfree && *(p-1) == '\0');
1112 --p;
1113 break;
1114 }
1115 /* expand buffer and try again */
1116 assert(*(pvend-1) == '\0');
Tim Petersddea2082002-03-23 10:03:50 +00001117 increment = total_v_size >> 2; /* mild exponential growth */
1118 total_v_size += increment;
Martin v. Löwis2a190742006-04-13 07:37:25 +00001119 if (total_v_size > PY_SSIZE_T_MAX) {
Tim Peters86821b22001-01-07 21:19:34 +00001120 PyErr_SetString(PyExc_OverflowError,
1121 "line is longer than a Python string can hold");
1122 Py_DECREF(v);
1123 return NULL;
1124 }
1125 if (_PyString_Resize(&v, (int)total_v_size) < 0)
1126 return NULL;
1127 /* overwrite the trailing null byte */
Tim Petersddea2082002-03-23 10:03:50 +00001128 pvfree = BUF(v) + (total_v_size - increment - 1);
Tim Peters86821b22001-01-07 21:19:34 +00001129 }
1130 if (BUF(v) + total_v_size != p)
1131 _PyString_Resize(&v, p - BUF(v));
1132 return v;
1133#undef INITBUFSIZE
Tim Peters142297a2001-01-15 10:36:56 +00001134#undef MAXBUFSIZE
Tim Peters86821b22001-01-07 21:19:34 +00001135}
Tim Petersf29b64d2001-01-15 06:33:19 +00001136#endif /* ifdef USE_FGETS_IN_GETLINE */
Guido van Rossumfdf95dd1997-05-05 22:15:02 +00001137
Guido van Rossum0bd24411991-04-04 15:21:57 +00001138/* Internal routine to get a line.
1139 Size argument interpretation:
1140 > 0: max length;
Guido van Rossum86282062001-01-08 01:26:47 +00001141 <= 0: read arbitrary line
Guido van Rossumce5ba841991-03-06 13:06:18 +00001142*/
1143
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001144static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001145get_line(PyFileObject *f, int n)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001146{
Guido van Rossum1187aa42001-01-05 14:43:05 +00001147 FILE *fp = f->f_fp;
1148 int c;
Andrew M. Kuchling4b2b4452000-11-29 02:53:22 +00001149 char *buf, *end;
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001150 size_t total_v_size; /* total # of slots in buffer */
1151 size_t used_v_size; /* # used slots in buffer */
1152 size_t increment; /* amount to increment the buffer */
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001153 PyObject *v;
Jack Jansen7b8c7542002-04-14 20:12:41 +00001154 int newlinetypes = f->f_newlinetypes;
1155 int skipnextlf = f->f_skipnextlf;
1156 int univ_newline = f->f_univ_newline;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001157
Jack Jansen7b8c7542002-04-14 20:12:41 +00001158#if defined(USE_FGETS_IN_GETLINE)
Jack Jansen7b8c7542002-04-14 20:12:41 +00001159 if (n <= 0 && !univ_newline )
Tim Petersf29b64d2001-01-15 06:33:19 +00001160 return getline_via_fgets(fp);
Tim Peters86821b22001-01-07 21:19:34 +00001161#endif
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001162 total_v_size = n > 0 ? n : 100;
1163 v = PyString_FromStringAndSize((char *)NULL, total_v_size);
Guido van Rossum3f5da241990-12-20 15:06:42 +00001164 if (v == NULL)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001165 return NULL;
Guido van Rossumce5ba841991-03-06 13:06:18 +00001166 buf = BUF(v);
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001167 end = buf + total_v_size;
Guido van Rossum1984f1e1992-08-04 12:41:02 +00001168
Guido van Rossumce5ba841991-03-06 13:06:18 +00001169 for (;;) {
Guido van Rossum1187aa42001-01-05 14:43:05 +00001170 Py_BEGIN_ALLOW_THREADS
1171 FLOCKFILE(fp);
Jack Jansen7b8c7542002-04-14 20:12:41 +00001172 if (univ_newline) {
1173 c = 'x'; /* Shut up gcc warning */
1174 while ( buf != end && (c = GETC(fp)) != EOF ) {
1175 if (skipnextlf ) {
1176 skipnextlf = 0;
1177 if (c == '\n') {
Tim Petersf1827cf2003-09-07 03:30:18 +00001178 /* Seeing a \n here with
1179 * skipnextlf true means we
Jeremy Hylton8b735422002-08-14 21:01:41 +00001180 * saw a \r before.
1181 */
Jack Jansen7b8c7542002-04-14 20:12:41 +00001182 newlinetypes |= NEWLINE_CRLF;
1183 c = GETC(fp);
1184 if (c == EOF) break;
1185 } else {
1186 newlinetypes |= NEWLINE_CR;
1187 }
1188 }
1189 if (c == '\r') {
1190 skipnextlf = 1;
1191 c = '\n';
1192 } else if ( c == '\n')
1193 newlinetypes |= NEWLINE_LF;
1194 *buf++ = c;
1195 if (c == '\n') break;
1196 }
1197 if ( c == EOF && skipnextlf )
1198 newlinetypes |= NEWLINE_CR;
1199 } else /* If not universal newlines use the normal loop */
Guido van Rossum1187aa42001-01-05 14:43:05 +00001200 while ((c = GETC(fp)) != EOF &&
1201 (*buf++ = c) != '\n' &&
1202 buf != end)
1203 ;
1204 FUNLOCKFILE(fp);
1205 Py_END_ALLOW_THREADS
Jack Jansen7b8c7542002-04-14 20:12:41 +00001206 f->f_newlinetypes = newlinetypes;
1207 f->f_skipnextlf = skipnextlf;
Guido van Rossum1187aa42001-01-05 14:43:05 +00001208 if (c == '\n')
1209 break;
1210 if (c == EOF) {
Guido van Rossum29206bc2001-08-09 18:14:59 +00001211 if (ferror(fp)) {
1212 PyErr_SetFromErrno(PyExc_IOError);
1213 clearerr(fp);
1214 Py_DECREF(v);
1215 return NULL;
1216 }
Guido van Rossum76ad8ed1991-06-03 10:54:55 +00001217 clearerr(fp);
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001218 if (PyErr_CheckSignals()) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001219 Py_DECREF(v);
Guido van Rossum0bd24411991-04-04 15:21:57 +00001220 return NULL;
1221 }
Guido van Rossumce5ba841991-03-06 13:06:18 +00001222 break;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001223 }
Guido van Rossum1187aa42001-01-05 14:43:05 +00001224 /* Must be because buf == end */
1225 if (n > 0)
Guido van Rossum0bd24411991-04-04 15:21:57 +00001226 break;
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001227 used_v_size = total_v_size;
1228 increment = total_v_size >> 2; /* mild exponential growth */
1229 total_v_size += increment;
Martin v. Löwis2a190742006-04-13 07:37:25 +00001230 if (total_v_size > PY_SSIZE_T_MAX) {
Guido van Rossum1187aa42001-01-05 14:43:05 +00001231 PyErr_SetString(PyExc_OverflowError,
1232 "line is longer than a Python string can hold");
Tim Peters86821b22001-01-07 21:19:34 +00001233 Py_DECREF(v);
Guido van Rossum1187aa42001-01-05 14:43:05 +00001234 return NULL;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001235 }
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001236 if (_PyString_Resize(&v, total_v_size) < 0)
Guido van Rossum1187aa42001-01-05 14:43:05 +00001237 return NULL;
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001238 buf = BUF(v) + used_v_size;
1239 end = BUF(v) + total_v_size;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001240 }
Guido van Rossum1984f1e1992-08-04 12:41:02 +00001241
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001242 used_v_size = buf - BUF(v);
1243 if (used_v_size != total_v_size)
1244 _PyString_Resize(&v, used_v_size);
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001245 return v;
1246}
1247
Guido van Rossum0bd24411991-04-04 15:21:57 +00001248/* External C interface */
1249
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001250PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001251PyFile_GetLine(PyObject *f, int n)
Guido van Rossum0bd24411991-04-04 15:21:57 +00001252{
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001253 PyObject *result;
1254
Guido van Rossum3165fe61992-09-25 21:59:05 +00001255 if (f == NULL) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001256 PyErr_BadInternalCall();
Guido van Rossum0bd24411991-04-04 15:21:57 +00001257 return NULL;
1258 }
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001259
1260 if (PyFile_Check(f)) {
Thomas Woutersc45251a2006-02-12 11:53:32 +00001261 PyFileObject *fo = (PyFileObject *)f;
1262 if (fo->f_fp == NULL)
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001263 return err_closed();
Thomas Woutersc45251a2006-02-12 11:53:32 +00001264 /* refuse to mix with f.next() */
1265 if (fo->f_buf != NULL &&
1266 (fo->f_bufend - fo->f_bufptr) > 0 &&
1267 fo->f_buf[0] != '\0')
1268 return err_iterbuffered();
1269 result = get_line(fo, n);
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001270 }
1271 else {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001272 PyObject *reader;
1273 PyObject *args;
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001274
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001275 reader = PyObject_GetAttrString(f, "readline");
Guido van Rossum3165fe61992-09-25 21:59:05 +00001276 if (reader == NULL)
1277 return NULL;
1278 if (n <= 0)
Raymond Hettinger8ae46892003-10-12 19:09:37 +00001279 args = PyTuple_New(0);
Guido van Rossum3165fe61992-09-25 21:59:05 +00001280 else
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001281 args = Py_BuildValue("(i)", n);
Guido van Rossum3165fe61992-09-25 21:59:05 +00001282 if (args == NULL) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001283 Py_DECREF(reader);
Guido van Rossum3165fe61992-09-25 21:59:05 +00001284 return NULL;
1285 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001286 result = PyEval_CallObject(reader, args);
1287 Py_DECREF(reader);
1288 Py_DECREF(args);
Martin v. Löwisaf6a27a2003-01-03 19:16:14 +00001289 if (result != NULL && !PyString_Check(result) &&
1290 !PyUnicode_Check(result)) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001291 Py_DECREF(result);
Guido van Rossum3165fe61992-09-25 21:59:05 +00001292 result = NULL;
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001293 PyErr_SetString(PyExc_TypeError,
Guido van Rossum3165fe61992-09-25 21:59:05 +00001294 "object.readline() returned non-string");
1295 }
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001296 }
1297
1298 if (n < 0 && result != NULL && PyString_Check(result)) {
1299 char *s = PyString_AS_STRING(result);
Martin v. Löwis18e16552006-02-15 17:27:45 +00001300 Py_ssize_t len = PyString_GET_SIZE(result);
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001301 if (len == 0) {
1302 Py_DECREF(result);
1303 result = NULL;
1304 PyErr_SetString(PyExc_EOFError,
1305 "EOF when reading a line");
1306 }
1307 else if (s[len-1] == '\n') {
1308 if (result->ob_refcnt == 1)
1309 _PyString_Resize(&result, len-1);
1310 else {
1311 PyObject *v;
1312 v = PyString_FromStringAndSize(s, len-1);
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001313 Py_DECREF(result);
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001314 result = v;
Guido van Rossum3165fe61992-09-25 21:59:05 +00001315 }
1316 }
Guido van Rossum3165fe61992-09-25 21:59:05 +00001317 }
Martin v. Löwisaf6a27a2003-01-03 19:16:14 +00001318#ifdef Py_USING_UNICODE
1319 if (n < 0 && result != NULL && PyUnicode_Check(result)) {
1320 Py_UNICODE *s = PyUnicode_AS_UNICODE(result);
Martin v. Löwis18e16552006-02-15 17:27:45 +00001321 Py_ssize_t len = PyUnicode_GET_SIZE(result);
Martin v. Löwisaf6a27a2003-01-03 19:16:14 +00001322 if (len == 0) {
1323 Py_DECREF(result);
1324 result = NULL;
1325 PyErr_SetString(PyExc_EOFError,
1326 "EOF when reading a line");
1327 }
1328 else if (s[len-1] == '\n') {
1329 if (result->ob_refcnt == 1)
1330 PyUnicode_Resize(&result, len-1);
1331 else {
1332 PyObject *v;
1333 v = PyUnicode_FromUnicode(s, len-1);
1334 Py_DECREF(result);
1335 result = v;
1336 }
1337 }
1338 }
1339#endif
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001340 return result;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001341}
1342
1343/* Python method */
1344
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001345static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001346file_readline(PyFileObject *f, PyObject *args)
Guido van Rossum0bd24411991-04-04 15:21:57 +00001347{
Guido van Rossum789a1611997-05-10 22:33:55 +00001348 int n = -1;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001349
Guido van Rossumd7297e61992-07-06 14:19:26 +00001350 if (f->f_fp == NULL)
1351 return err_closed();
Thomas Woutersc45251a2006-02-12 11:53:32 +00001352 /* refuse to mix with f.next() */
1353 if (f->f_buf != NULL &&
1354 (f->f_bufend - f->f_bufptr) > 0 &&
1355 f->f_buf[0] != '\0')
1356 return err_iterbuffered();
Guido van Rossum43713e52000-02-29 13:59:29 +00001357 if (!PyArg_ParseTuple(args, "|i:readline", &n))
Guido van Rossum789a1611997-05-10 22:33:55 +00001358 return NULL;
1359 if (n == 0)
1360 return PyString_FromString("");
1361 if (n < 0)
1362 n = 0;
Marc-André Lemburg1f468602000-07-05 15:32:40 +00001363 return get_line(f, n);
Guido van Rossum0bd24411991-04-04 15:21:57 +00001364}
1365
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001366static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001367file_readlines(PyFileObject *f, PyObject *args)
Guido van Rossumce5ba841991-03-06 13:06:18 +00001368{
Guido van Rossum789a1611997-05-10 22:33:55 +00001369 long sizehint = 0;
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001370 PyObject *list;
1371 PyObject *line;
Guido van Rossum6263d541997-05-10 22:07:25 +00001372 char small_buffer[SMALLCHUNK];
1373 char *buffer = small_buffer;
1374 size_t buffersize = SMALLCHUNK;
1375 PyObject *big_buffer = NULL;
1376 size_t nfilled = 0;
1377 size_t nread;
Guido van Rossum789a1611997-05-10 22:33:55 +00001378 size_t totalread = 0;
Guido van Rossum6263d541997-05-10 22:07:25 +00001379 char *p, *q, *end;
1380 int err;
Guido van Rossum79fd0fc2001-10-12 20:01:53 +00001381 int shortread = 0;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001382
Guido van Rossumd7297e61992-07-06 14:19:26 +00001383 if (f->f_fp == NULL)
1384 return err_closed();
Thomas Woutersc45251a2006-02-12 11:53:32 +00001385 /* refuse to mix with f.next() */
1386 if (f->f_buf != NULL &&
1387 (f->f_bufend - f->f_bufptr) > 0 &&
1388 f->f_buf[0] != '\0')
1389 return err_iterbuffered();
Guido van Rossum43713e52000-02-29 13:59:29 +00001390 if (!PyArg_ParseTuple(args, "|l:readlines", &sizehint))
Guido van Rossum0bd24411991-04-04 15:21:57 +00001391 return NULL;
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001392 if ((list = PyList_New(0)) == NULL)
Guido van Rossumce5ba841991-03-06 13:06:18 +00001393 return NULL;
1394 for (;;) {
Guido van Rossum79fd0fc2001-10-12 20:01:53 +00001395 if (shortread)
1396 nread = 0;
1397 else {
1398 Py_BEGIN_ALLOW_THREADS
1399 errno = 0;
Tim Peters058b1412002-04-21 07:29:14 +00001400 nread = Py_UniversalNewlineFread(buffer+nfilled,
Jack Jansen7b8c7542002-04-14 20:12:41 +00001401 buffersize-nfilled, f->f_fp, (PyObject *)f);
Guido van Rossum79fd0fc2001-10-12 20:01:53 +00001402 Py_END_ALLOW_THREADS
1403 shortread = (nread < buffersize-nfilled);
1404 }
Guido van Rossum6263d541997-05-10 22:07:25 +00001405 if (nread == 0) {
Guido van Rossum789a1611997-05-10 22:33:55 +00001406 sizehint = 0;
Guido van Rossum3da3fce1998-02-19 20:46:48 +00001407 if (!ferror(f->f_fp))
Guido van Rossum6263d541997-05-10 22:07:25 +00001408 break;
1409 PyErr_SetFromErrno(PyExc_IOError);
1410 clearerr(f->f_fp);
1411 error:
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001412 Py_DECREF(list);
Guido van Rossum6263d541997-05-10 22:07:25 +00001413 list = NULL;
1414 goto cleanup;
Guido van Rossumce5ba841991-03-06 13:06:18 +00001415 }
Guido van Rossum789a1611997-05-10 22:33:55 +00001416 totalread += nread;
Anthony Baxter377be112006-04-11 06:54:30 +00001417 p = (char *)memchr(buffer+nfilled, '\n', nread);
Guido van Rossum6263d541997-05-10 22:07:25 +00001418 if (p == NULL) {
1419 /* Need a larger buffer to fit this line */
1420 nfilled += nread;
1421 buffersize *= 2;
Martin v. Löwis2a190742006-04-13 07:37:25 +00001422 if (buffersize > PY_SSIZE_T_MAX) {
Trent Mickf29f47b2000-08-11 19:02:59 +00001423 PyErr_SetString(PyExc_OverflowError,
Guido van Rossume07d5cf2001-01-09 21:50:24 +00001424 "line is longer than a Python string can hold");
Trent Mickf29f47b2000-08-11 19:02:59 +00001425 goto error;
1426 }
Guido van Rossum6263d541997-05-10 22:07:25 +00001427 if (big_buffer == NULL) {
1428 /* Create the big buffer */
1429 big_buffer = PyString_FromStringAndSize(
1430 NULL, buffersize);
1431 if (big_buffer == NULL)
1432 goto error;
1433 buffer = PyString_AS_STRING(big_buffer);
1434 memcpy(buffer, small_buffer, nfilled);
1435 }
1436 else {
1437 /* Grow the big buffer */
Jack Jansen7b8c7542002-04-14 20:12:41 +00001438 if ( _PyString_Resize(&big_buffer, buffersize) < 0 )
1439 goto error;
Guido van Rossum6263d541997-05-10 22:07:25 +00001440 buffer = PyString_AS_STRING(big_buffer);
1441 }
1442 continue;
1443 }
1444 end = buffer+nfilled+nread;
1445 q = buffer;
1446 do {
1447 /* Process complete lines */
1448 p++;
1449 line = PyString_FromStringAndSize(q, p-q);
1450 if (line == NULL)
1451 goto error;
1452 err = PyList_Append(list, line);
1453 Py_DECREF(line);
1454 if (err != 0)
1455 goto error;
1456 q = p;
Anthony Baxter377be112006-04-11 06:54:30 +00001457 p = (char *)memchr(q, '\n', end-q);
Guido van Rossum6263d541997-05-10 22:07:25 +00001458 } while (p != NULL);
1459 /* Move the remaining incomplete line to the start */
1460 nfilled = end-q;
1461 memmove(buffer, q, nfilled);
Guido van Rossum789a1611997-05-10 22:33:55 +00001462 if (sizehint > 0)
1463 if (totalread >= (size_t)sizehint)
1464 break;
Guido van Rossumce5ba841991-03-06 13:06:18 +00001465 }
Guido van Rossum6263d541997-05-10 22:07:25 +00001466 if (nfilled != 0) {
1467 /* Partial last line */
1468 line = PyString_FromStringAndSize(buffer, nfilled);
1469 if (line == NULL)
1470 goto error;
Guido van Rossum789a1611997-05-10 22:33:55 +00001471 if (sizehint > 0) {
1472 /* Need to complete the last line */
Marc-André Lemburg1f468602000-07-05 15:32:40 +00001473 PyObject *rest = get_line(f, 0);
Guido van Rossum789a1611997-05-10 22:33:55 +00001474 if (rest == NULL) {
1475 Py_DECREF(line);
1476 goto error;
1477 }
1478 PyString_Concat(&line, rest);
1479 Py_DECREF(rest);
1480 if (line == NULL)
1481 goto error;
1482 }
Guido van Rossum6263d541997-05-10 22:07:25 +00001483 err = PyList_Append(list, line);
1484 Py_DECREF(line);
1485 if (err != 0)
1486 goto error;
1487 }
1488 cleanup:
Tim Peters5de98422002-04-27 18:44:32 +00001489 Py_XDECREF(big_buffer);
Guido van Rossumce5ba841991-03-06 13:06:18 +00001490 return list;
1491}
1492
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001493static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001494file_write(PyFileObject *f, PyObject *args)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001495{
Guido van Rossumd7297e61992-07-06 14:19:26 +00001496 char *s;
Martin v. Löwis18e16552006-02-15 17:27:45 +00001497 Py_ssize_t n, n2;
Guido van Rossumd7297e61992-07-06 14:19:26 +00001498 if (f->f_fp == NULL)
1499 return err_closed();
Michael W. Hudsone2ec3eb2001-10-31 18:51:01 +00001500 if (!PyArg_ParseTuple(args, f->f_binary ? "s#" : "t#", &s, &n))
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001501 return NULL;
Guido van Rossumeb183da1991-04-04 10:44:06 +00001502 f->f_softspace = 0;
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001503 Py_BEGIN_ALLOW_THREADS
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001504 errno = 0;
Guido van Rossumd7297e61992-07-06 14:19:26 +00001505 n2 = fwrite(s, 1, n, f->f_fp);
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001506 Py_END_ALLOW_THREADS
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001507 if (n2 != n) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001508 PyErr_SetFromErrno(PyExc_IOError);
Guido van Rossumfebd5511992-03-04 16:39:24 +00001509 clearerr(f->f_fp);
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001510 return NULL;
1511 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001512 Py_INCREF(Py_None);
1513 return Py_None;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001514}
1515
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001516static PyObject *
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001517file_writelines(PyFileObject *f, PyObject *seq)
Guido van Rossum5a2a6831993-10-25 09:59:04 +00001518{
Guido van Rossumee70ad12000-03-13 16:27:06 +00001519#define CHUNKSIZE 1000
1520 PyObject *list, *line;
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001521 PyObject *it; /* iter(seq) */
Guido van Rossumee70ad12000-03-13 16:27:06 +00001522 PyObject *result;
Martin v. Löwis18e16552006-02-15 17:27:45 +00001523 int index, islist;
1524 Py_ssize_t i, j, nwritten, len;
Guido van Rossumee70ad12000-03-13 16:27:06 +00001525
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001526 assert(seq != NULL);
Guido van Rossum5a2a6831993-10-25 09:59:04 +00001527 if (f->f_fp == NULL)
1528 return err_closed();
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001529
1530 result = NULL;
1531 list = NULL;
1532 islist = PyList_Check(seq);
1533 if (islist)
1534 it = NULL;
1535 else {
1536 it = PyObject_GetIter(seq);
1537 if (it == NULL) {
1538 PyErr_SetString(PyExc_TypeError,
1539 "writelines() requires an iterable argument");
1540 return NULL;
1541 }
1542 /* From here on, fail by going to error, to reclaim "it". */
1543 list = PyList_New(CHUNKSIZE);
1544 if (list == NULL)
1545 goto error;
Guido van Rossum5a2a6831993-10-25 09:59:04 +00001546 }
Guido van Rossumee70ad12000-03-13 16:27:06 +00001547
1548 /* Strategy: slurp CHUNKSIZE lines into a private list,
1549 checking that they are all strings, then write that list
1550 without holding the interpreter lock, then come back for more. */
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001551 for (index = 0; ; index += CHUNKSIZE) {
Guido van Rossumee70ad12000-03-13 16:27:06 +00001552 if (islist) {
1553 Py_XDECREF(list);
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001554 list = PyList_GetSlice(seq, index, index+CHUNKSIZE);
Guido van Rossumee70ad12000-03-13 16:27:06 +00001555 if (list == NULL)
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001556 goto error;
Guido van Rossumee70ad12000-03-13 16:27:06 +00001557 j = PyList_GET_SIZE(list);
1558 }
1559 else {
1560 for (j = 0; j < CHUNKSIZE; j++) {
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001561 line = PyIter_Next(it);
Guido van Rossumee70ad12000-03-13 16:27:06 +00001562 if (line == NULL) {
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001563 if (PyErr_Occurred())
1564 goto error;
1565 break;
Guido van Rossumee70ad12000-03-13 16:27:06 +00001566 }
Guido van Rossumee70ad12000-03-13 16:27:06 +00001567 PyList_SetItem(list, j, line);
1568 }
1569 }
1570 if (j == 0)
1571 break;
1572
Marc-André Lemburg6ef68b52000-08-25 22:39:50 +00001573 /* Check that all entries are indeed strings. If not,
1574 apply the same rules as for file.write() and
1575 convert the results to strings. This is slow, but
1576 seems to be the only way since all conversion APIs
1577 could potentially execute Python code. */
1578 for (i = 0; i < j; i++) {
1579 PyObject *v = PyList_GET_ITEM(list, i);
1580 if (!PyString_Check(v)) {
1581 const char *buffer;
Tim Peters86821b22001-01-07 21:19:34 +00001582 if (((f->f_binary &&
Marc-André Lemburg6ef68b52000-08-25 22:39:50 +00001583 PyObject_AsReadBuffer(v,
1584 (const void**)&buffer,
1585 &len)) ||
1586 PyObject_AsCharBuffer(v,
1587 &buffer,
1588 &len))) {
1589 PyErr_SetString(PyExc_TypeError,
Jeremy Hylton8b735422002-08-14 21:01:41 +00001590 "writelines() argument must be a sequence of strings");
Marc-André Lemburg6ef68b52000-08-25 22:39:50 +00001591 goto error;
1592 }
1593 line = PyString_FromStringAndSize(buffer,
1594 len);
1595 if (line == NULL)
1596 goto error;
1597 Py_DECREF(v);
Marc-André Lemburgf5e96fa2000-08-25 22:49:05 +00001598 PyList_SET_ITEM(list, i, line);
Marc-André Lemburg6ef68b52000-08-25 22:39:50 +00001599 }
1600 }
1601
1602 /* Since we are releasing the global lock, the
1603 following code may *not* execute Python code. */
Guido van Rossumee70ad12000-03-13 16:27:06 +00001604 Py_BEGIN_ALLOW_THREADS
1605 f->f_softspace = 0;
1606 errno = 0;
1607 for (i = 0; i < j; i++) {
Marc-André Lemburg6ef68b52000-08-25 22:39:50 +00001608 line = PyList_GET_ITEM(list, i);
Guido van Rossumee70ad12000-03-13 16:27:06 +00001609 len = PyString_GET_SIZE(line);
1610 nwritten = fwrite(PyString_AS_STRING(line),
1611 1, len, f->f_fp);
1612 if (nwritten != len) {
1613 Py_BLOCK_THREADS
1614 PyErr_SetFromErrno(PyExc_IOError);
1615 clearerr(f->f_fp);
1616 goto error;
1617 }
1618 }
1619 Py_END_ALLOW_THREADS
1620
1621 if (j < CHUNKSIZE)
1622 break;
Guido van Rossumee70ad12000-03-13 16:27:06 +00001623 }
1624
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001625 Py_INCREF(Py_None);
Guido van Rossumee70ad12000-03-13 16:27:06 +00001626 result = Py_None;
1627 error:
1628 Py_XDECREF(list);
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001629 Py_XDECREF(it);
Guido van Rossumee70ad12000-03-13 16:27:06 +00001630 return result;
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001631#undef CHUNKSIZE
Guido van Rossum5a2a6831993-10-25 09:59:04 +00001632}
1633
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001634static PyObject *
Guido van Rossum1a5e21e2006-02-28 21:57:43 +00001635file_self(PyFileObject *f)
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001636{
1637 if (f->f_fp == NULL)
1638 return err_closed();
1639 Py_INCREF(f);
1640 return (PyObject *)f;
1641}
1642
Georg Brandl98b40ad2006-06-08 14:50:21 +00001643static PyObject *
1644file_exit(PyFileObject *f, PyObject *args)
1645{
1646 PyObject *ret = file_close(f);
1647 if (!ret)
1648 /* If error occurred, pass through */
1649 return NULL;
1650 Py_DECREF(ret);
1651 /* We cannot return the result of close since a true
1652 * value will be interpreted as "yes, swallow the
1653 * exception if one was raised inside the with block". */
1654 Py_RETURN_NONE;
1655}
1656
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001657PyDoc_STRVAR(readline_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001658"readline([size]) -> next line from the file, as a string.\n"
1659"\n"
1660"Retain newline. A non-negative size argument limits the maximum\n"
1661"number of bytes to return (an incomplete line may be returned then).\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001662"Return an empty string at EOF.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001663
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001664PyDoc_STRVAR(read_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001665"read([size]) -> read at most size bytes, returned as a string.\n"
1666"\n"
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +00001667"If the size argument is negative or omitted, read until EOF is reached.\n"
1668"Notice that when in non-blocking mode, less data than what was requested\n"
1669"may be returned, even if no size parameter was given.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001670
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001671PyDoc_STRVAR(write_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001672"write(str) -> None. Write string str to file.\n"
1673"\n"
1674"Note that due to buffering, flush() or close() may be needed before\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001675"the file on disk reflects the data written.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001676
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001677PyDoc_STRVAR(fileno_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001678"fileno() -> integer \"file descriptor\".\n"
1679"\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001680"This is needed for lower-level file interfaces, such os.read().");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001681
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001682PyDoc_STRVAR(seek_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001683"seek(offset[, whence]) -> None. Move to new file position.\n"
1684"\n"
1685"Argument offset is a byte count. Optional argument whence defaults to\n"
1686"0 (offset from start of file, offset should be >= 0); other values are 1\n"
1687"(move relative to current position, positive or negative), and 2 (move\n"
1688"relative to end of file, usually negative, although many platforms allow\n"
Martin v. Löwis849a9722003-10-18 09:38:01 +00001689"seeking beyond the end of a file). If the file is opened in text mode,\n"
1690"only offsets returned by tell() are legal. Use of other offsets causes\n"
1691"undefined behavior."
Tim Petersefc3a3a2001-09-20 07:55:22 +00001692"\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001693"Note that not all file objects are seekable.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001694
Guido van Rossumd7047b31995-01-02 19:07:15 +00001695#ifdef HAVE_FTRUNCATE
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001696PyDoc_STRVAR(truncate_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001697"truncate([size]) -> None. Truncate the file to at most size bytes.\n"
1698"\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001699"Size defaults to the current file position, as returned by tell().");
Guido van Rossumd7047b31995-01-02 19:07:15 +00001700#endif
Tim Petersefc3a3a2001-09-20 07:55:22 +00001701
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001702PyDoc_STRVAR(tell_doc,
1703"tell() -> current file position, an integer (may be a long integer).");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001704
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001705PyDoc_STRVAR(readinto_doc,
1706"readinto() -> Undocumented. Don't use this; it may go away.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001707
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001708PyDoc_STRVAR(readlines_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001709"readlines([size]) -> list of strings, each a line from the file.\n"
1710"\n"
1711"Call readline() repeatedly and return a list of the lines so read.\n"
1712"The optional size argument, if given, is an approximate bound on the\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001713"total number of bytes in the lines returned.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001714
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001715PyDoc_STRVAR(xreadlines_doc,
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001716"xreadlines() -> returns self.\n"
Tim Petersefc3a3a2001-09-20 07:55:22 +00001717"\n"
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001718"For backward compatibility. File objects now include the performance\n"
1719"optimizations previously implemented in the xreadlines module.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001720
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001721PyDoc_STRVAR(writelines_doc,
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001722"writelines(sequence_of_strings) -> None. Write the strings to the file.\n"
Tim Petersefc3a3a2001-09-20 07:55:22 +00001723"\n"
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001724"Note that newlines are not added. The sequence can be any iterable object\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001725"producing strings. This is equivalent to calling write() for each string.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001726
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001727PyDoc_STRVAR(flush_doc,
1728"flush() -> None. Flush the internal I/O buffer.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001729
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001730PyDoc_STRVAR(close_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001731"close() -> None or (perhaps) an integer. Close the file.\n"
1732"\n"
Guido van Rossum77f6a652002-04-03 22:41:51 +00001733"Sets data attribute .closed to True. A closed file cannot be used for\n"
Tim Petersefc3a3a2001-09-20 07:55:22 +00001734"further I/O operations. close() may be called more than once without\n"
1735"error. Some kinds of file objects (for example, opened by popen())\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001736"may return an exit status upon closing.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001737
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001738PyDoc_STRVAR(isatty_doc,
1739"isatty() -> true or false. True if the file is connected to a tty device.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001740
Guido van Rossum1a5e21e2006-02-28 21:57:43 +00001741PyDoc_STRVAR(enter_doc,
1742 "__enter__() -> self.");
1743
Georg Brandl98b40ad2006-06-08 14:50:21 +00001744PyDoc_STRVAR(exit_doc,
1745 "__exit__(*excinfo) -> None. Closes the file.");
1746
Tim Petersefc3a3a2001-09-20 07:55:22 +00001747static PyMethodDef file_methods[] = {
Jeremy Hylton8b735422002-08-14 21:01:41 +00001748 {"readline", (PyCFunction)file_readline, METH_VARARGS, readline_doc},
1749 {"read", (PyCFunction)file_read, METH_VARARGS, read_doc},
1750 {"write", (PyCFunction)file_write, METH_VARARGS, write_doc},
1751 {"fileno", (PyCFunction)file_fileno, METH_NOARGS, fileno_doc},
1752 {"seek", (PyCFunction)file_seek, METH_VARARGS, seek_doc},
Tim Petersefc3a3a2001-09-20 07:55:22 +00001753#ifdef HAVE_FTRUNCATE
Jeremy Hylton8b735422002-08-14 21:01:41 +00001754 {"truncate", (PyCFunction)file_truncate, METH_VARARGS, truncate_doc},
Tim Petersefc3a3a2001-09-20 07:55:22 +00001755#endif
Jeremy Hylton8b735422002-08-14 21:01:41 +00001756 {"tell", (PyCFunction)file_tell, METH_NOARGS, tell_doc},
1757 {"readinto", (PyCFunction)file_readinto, METH_VARARGS, readinto_doc},
1758 {"readlines", (PyCFunction)file_readlines,METH_VARARGS, readlines_doc},
Guido van Rossum1a5e21e2006-02-28 21:57:43 +00001759 {"xreadlines",(PyCFunction)file_self, METH_NOARGS, xreadlines_doc},
Jeremy Hylton8b735422002-08-14 21:01:41 +00001760 {"writelines",(PyCFunction)file_writelines, METH_O, writelines_doc},
1761 {"flush", (PyCFunction)file_flush, METH_NOARGS, flush_doc},
1762 {"close", (PyCFunction)file_close, METH_NOARGS, close_doc},
1763 {"isatty", (PyCFunction)file_isatty, METH_NOARGS, isatty_doc},
Guido van Rossum1a5e21e2006-02-28 21:57:43 +00001764 {"__enter__", (PyCFunction)file_self, METH_NOARGS, enter_doc},
Georg Brandl98b40ad2006-06-08 14:50:21 +00001765 {"__exit__", (PyCFunction)file_exit, METH_VARARGS, exit_doc},
Jeremy Hylton8b735422002-08-14 21:01:41 +00001766 {NULL, NULL} /* sentinel */
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001767};
1768
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001769#define OFF(x) offsetof(PyFileObject, x)
Guido van Rossumb6775db1994-08-01 11:34:53 +00001770
Guido van Rossum6f799372001-09-20 20:46:19 +00001771static PyMemberDef file_memberlist[] = {
1772 {"softspace", T_INT, OFF(f_softspace), 0,
1773 "flag indicating that a space needs to be printed; used by print"},
1774 {"mode", T_OBJECT, OFF(f_mode), RO,
Martin v. Löwis6233c9b2002-12-11 13:06:53 +00001775 "file mode ('r', 'U', 'w', 'a', possibly with 'b' or '+' added)"},
Guido van Rossum6f799372001-09-20 20:46:19 +00001776 {"name", T_OBJECT, OFF(f_name), RO,
1777 "file name"},
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00001778 {"encoding", T_OBJECT, OFF(f_encoding), RO,
1779 "file encoding"},
Guido van Rossumb6775db1994-08-01 11:34:53 +00001780 /* getattr(f, "closed") is implemented without this table */
Guido van Rossumb6775db1994-08-01 11:34:53 +00001781 {NULL} /* Sentinel */
1782};
1783
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001784static PyObject *
Tim Peters6d6c1a32001-08-02 04:15:00 +00001785get_closed(PyFileObject *f, void *closure)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001786{
Guido van Rossum77f6a652002-04-03 22:41:51 +00001787 return PyBool_FromLong((long)(f->f_fp == 0));
Guido van Rossumb6775db1994-08-01 11:34:53 +00001788}
Jack Jansen7b8c7542002-04-14 20:12:41 +00001789static PyObject *
1790get_newlines(PyFileObject *f, void *closure)
1791{
1792 switch (f->f_newlinetypes) {
1793 case NEWLINE_UNKNOWN:
1794 Py_INCREF(Py_None);
1795 return Py_None;
1796 case NEWLINE_CR:
1797 return PyString_FromString("\r");
1798 case NEWLINE_LF:
1799 return PyString_FromString("\n");
1800 case NEWLINE_CR|NEWLINE_LF:
1801 return Py_BuildValue("(ss)", "\r", "\n");
1802 case NEWLINE_CRLF:
1803 return PyString_FromString("\r\n");
1804 case NEWLINE_CR|NEWLINE_CRLF:
1805 return Py_BuildValue("(ss)", "\r", "\r\n");
1806 case NEWLINE_LF|NEWLINE_CRLF:
1807 return Py_BuildValue("(ss)", "\n", "\r\n");
1808 case NEWLINE_CR|NEWLINE_LF|NEWLINE_CRLF:
1809 return Py_BuildValue("(sss)", "\r", "\n", "\r\n");
1810 default:
Tim Petersf1827cf2003-09-07 03:30:18 +00001811 PyErr_Format(PyExc_SystemError,
1812 "Unknown newlines value 0x%x\n",
Jeremy Hylton8b735422002-08-14 21:01:41 +00001813 f->f_newlinetypes);
Jack Jansen7b8c7542002-04-14 20:12:41 +00001814 return NULL;
1815 }
1816}
Guido van Rossumb6775db1994-08-01 11:34:53 +00001817
Guido van Rossum32d34c82001-09-20 21:45:26 +00001818static PyGetSetDef file_getsetlist[] = {
Guido van Rossum77f6a652002-04-03 22:41:51 +00001819 {"closed", (getter)get_closed, NULL, "True if the file is closed"},
Tim Petersf1827cf2003-09-07 03:30:18 +00001820 {"newlines", (getter)get_newlines, NULL,
Jeremy Hylton8b735422002-08-14 21:01:41 +00001821 "end-of-line convention used in this file"},
Tim Peters6d6c1a32001-08-02 04:15:00 +00001822 {0},
1823};
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001824
Neal Norwitzd8b995f2002-08-06 21:50:54 +00001825static void
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001826drop_readahead(PyFileObject *f)
Guido van Rossum65967252001-04-21 13:20:18 +00001827{
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001828 if (f->f_buf != NULL) {
1829 PyMem_Free(f->f_buf);
1830 f->f_buf = NULL;
1831 }
Guido van Rossum65967252001-04-21 13:20:18 +00001832}
1833
Tim Petersf1827cf2003-09-07 03:30:18 +00001834/* Make sure that file has a readahead buffer with at least one byte
1835 (unless at EOF) and no more than bufsize. Returns negative value on
Georg Brandled02eb62006-03-31 20:31:02 +00001836 error, will set MemoryError if bufsize bytes cannot be allocated. */
Neal Norwitzd8b995f2002-08-06 21:50:54 +00001837static int
1838readahead(PyFileObject *f, int bufsize)
1839{
Martin v. Löwis18e16552006-02-15 17:27:45 +00001840 Py_ssize_t chunksize;
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001841
1842 if (f->f_buf != NULL) {
Tim Petersf1827cf2003-09-07 03:30:18 +00001843 if( (f->f_bufend - f->f_bufptr) >= 1)
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001844 return 0;
1845 else
1846 drop_readahead(f);
1847 }
Anthony Baxter377be112006-04-11 06:54:30 +00001848 if ((f->f_buf = (char *)PyMem_Malloc(bufsize)) == NULL) {
Georg Brandled02eb62006-03-31 20:31:02 +00001849 PyErr_NoMemory();
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001850 return -1;
1851 }
1852 Py_BEGIN_ALLOW_THREADS
1853 errno = 0;
1854 chunksize = Py_UniversalNewlineFread(
1855 f->f_buf, bufsize, f->f_fp, (PyObject *)f);
1856 Py_END_ALLOW_THREADS
1857 if (chunksize == 0) {
1858 if (ferror(f->f_fp)) {
1859 PyErr_SetFromErrno(PyExc_IOError);
1860 clearerr(f->f_fp);
1861 drop_readahead(f);
1862 return -1;
1863 }
1864 }
1865 f->f_bufptr = f->f_buf;
1866 f->f_bufend = f->f_buf + chunksize;
1867 return 0;
1868}
1869
1870/* Used by file_iternext. The returned string will start with 'skip'
Tim Petersf1827cf2003-09-07 03:30:18 +00001871 uninitialized bytes followed by the remainder of the line. Don't be
1872 horrified by the recursive call: maximum recursion depth is limited by
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001873 logarithmic buffer growth to about 50 even when reading a 1gb line. */
1874
Neal Norwitzd8b995f2002-08-06 21:50:54 +00001875static PyStringObject *
1876readahead_get_line_skip(PyFileObject *f, int skip, int bufsize)
1877{
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001878 PyStringObject* s;
1879 char *bufptr;
1880 char *buf;
Martin v. Löwis18e16552006-02-15 17:27:45 +00001881 Py_ssize_t len;
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001882
1883 if (f->f_buf == NULL)
Tim Petersf1827cf2003-09-07 03:30:18 +00001884 if (readahead(f, bufsize) < 0)
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001885 return NULL;
1886
1887 len = f->f_bufend - f->f_bufptr;
Tim Petersf1827cf2003-09-07 03:30:18 +00001888 if (len == 0)
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001889 return (PyStringObject *)
1890 PyString_FromStringAndSize(NULL, skip);
Anthony Baxter377be112006-04-11 06:54:30 +00001891 bufptr = (char *)memchr(f->f_bufptr, '\n', len);
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001892 if (bufptr != NULL) {
1893 bufptr++; /* Count the '\n' */
1894 len = bufptr - f->f_bufptr;
1895 s = (PyStringObject *)
1896 PyString_FromStringAndSize(NULL, skip+len);
Tim Petersf1827cf2003-09-07 03:30:18 +00001897 if (s == NULL)
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001898 return NULL;
1899 memcpy(PyString_AS_STRING(s)+skip, f->f_bufptr, len);
1900 f->f_bufptr = bufptr;
1901 if (bufptr == f->f_bufend)
1902 drop_readahead(f);
1903 } else {
1904 bufptr = f->f_bufptr;
1905 buf = f->f_buf;
1906 f->f_buf = NULL; /* Force new readahead buffer */
Martin v. Löwis18e16552006-02-15 17:27:45 +00001907 assert(skip+len < INT_MAX);
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001908 s = readahead_get_line_skip(
Martin v. Löwis18e16552006-02-15 17:27:45 +00001909 f, (int)(skip+len), bufsize + (bufsize>>2) );
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001910 if (s == NULL) {
1911 PyMem_Free(buf);
1912 return NULL;
1913 }
1914 memcpy(PyString_AS_STRING(s)+skip, bufptr, len);
1915 PyMem_Free(buf);
1916 }
1917 return s;
1918}
1919
1920/* A larger buffer size may actually decrease performance. */
1921#define READAHEAD_BUFSIZE 8192
1922
1923static PyObject *
1924file_iternext(PyFileObject *f)
1925{
1926 PyStringObject* l;
1927
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001928 if (f->f_fp == NULL)
1929 return err_closed();
1930
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001931 l = readahead_get_line_skip(f, 0, READAHEAD_BUFSIZE);
1932 if (l == NULL || PyString_GET_SIZE(l) == 0) {
1933 Py_XDECREF(l);
1934 return NULL;
1935 }
1936 return (PyObject *)l;
1937}
1938
1939
Tim Peters59c9a642001-09-13 05:38:56 +00001940static PyObject *
1941file_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1942{
Tim Peters44410012001-09-14 03:26:08 +00001943 PyObject *self;
1944 static PyObject *not_yet_string;
1945
1946 assert(type != NULL && type->tp_alloc != NULL);
1947
1948 if (not_yet_string == NULL) {
1949 not_yet_string = PyString_FromString("<uninitialized file>");
1950 if (not_yet_string == NULL)
1951 return NULL;
1952 }
1953
1954 self = type->tp_alloc(type, 0);
1955 if (self != NULL) {
1956 /* Always fill in the name and mode, so that nobody else
1957 needs to special-case NULLs there. */
1958 Py_INCREF(not_yet_string);
1959 ((PyFileObject *)self)->f_name = not_yet_string;
1960 Py_INCREF(not_yet_string);
1961 ((PyFileObject *)self)->f_mode = not_yet_string;
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00001962 Py_INCREF(Py_None);
1963 ((PyFileObject *)self)->f_encoding = Py_None;
Raymond Hettingercb87bc82004-05-31 00:35:52 +00001964 ((PyFileObject *)self)->weakreflist = NULL;
Tim Peters44410012001-09-14 03:26:08 +00001965 }
1966 return self;
1967}
1968
1969static int
1970file_init(PyObject *self, PyObject *args, PyObject *kwds)
1971{
1972 PyFileObject *foself = (PyFileObject *)self;
1973 int ret = 0;
Martin v. Löwis15e62742006-02-27 16:46:16 +00001974 static char *kwlist[] = {"name", "mode", "buffering", 0};
Tim Peters59c9a642001-09-13 05:38:56 +00001975 char *name = NULL;
1976 char *mode = "r";
1977 int bufsize = -1;
Mark Hammondc2e85bd2002-10-03 05:10:39 +00001978 int wideargument = 0;
Tim Peters44410012001-09-14 03:26:08 +00001979
1980 assert(PyFile_Check(self));
1981 if (foself->f_fp != NULL) {
1982 /* Have to close the existing file first. */
1983 PyObject *closeresult = file_close(foself);
1984 if (closeresult == NULL)
1985 return -1;
1986 Py_DECREF(closeresult);
1987 }
Tim Peters59c9a642001-09-13 05:38:56 +00001988
Mark Hammondc2e85bd2002-10-03 05:10:39 +00001989#ifdef Py_WIN_WIDE_FILENAMES
1990 if (GetVersion() < 0x80000000) { /* On NT, so wide API available */
1991 PyObject *po;
1992 if (PyArg_ParseTupleAndKeywords(args, kwds, "U|si:file",
1993 kwlist, &po, &mode, &bufsize)) {
1994 wideargument = 1;
Nicholas Bastinabce8a62004-03-21 20:24:07 +00001995 if (fill_file_fields(foself, NULL, po, mode,
1996 fclose) == NULL)
Mark Hammondc2e85bd2002-10-03 05:10:39 +00001997 goto Error;
1998 } else {
1999 /* Drop the argument parsing error as narrow
2000 strings are also valid. */
2001 PyErr_Clear();
2002 }
2003 }
2004#endif
2005
2006 if (!wideargument) {
Nicholas Bastinabce8a62004-03-21 20:24:07 +00002007 PyObject *o_name;
2008
Mark Hammondc2e85bd2002-10-03 05:10:39 +00002009 if (!PyArg_ParseTupleAndKeywords(args, kwds, "et|si:file", kwlist,
2010 Py_FileSystemDefaultEncoding,
2011 &name,
2012 &mode, &bufsize))
2013 return -1;
Nicholas Bastinabce8a62004-03-21 20:24:07 +00002014
2015 /* We parse again to get the name as a PyObject */
Jeremy Hyltonaf68c872005-12-10 18:50:16 +00002016 if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|si:file",
2017 kwlist, &o_name, &mode,
2018 &bufsize))
Nicholas Bastinabce8a62004-03-21 20:24:07 +00002019 return -1;
2020
2021 if (fill_file_fields(foself, NULL, o_name, mode,
2022 fclose) == NULL)
Mark Hammondc2e85bd2002-10-03 05:10:39 +00002023 goto Error;
2024 }
Tim Peters44410012001-09-14 03:26:08 +00002025 if (open_the_file(foself, name, mode) == NULL)
2026 goto Error;
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +00002027 foself->f_setbuf = NULL;
Tim Peters44410012001-09-14 03:26:08 +00002028 PyFile_SetBufSize(self, bufsize);
2029 goto Done;
2030
2031Error:
2032 ret = -1;
2033 /* fall through */
2034Done:
Tim Peters59c9a642001-09-13 05:38:56 +00002035 PyMem_Free(name); /* free the encoded string */
Tim Peters44410012001-09-14 03:26:08 +00002036 return ret;
Tim Peters59c9a642001-09-13 05:38:56 +00002037}
2038
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00002039PyDoc_VAR(file_doc) =
2040PyDoc_STR(
Tim Peters59c9a642001-09-13 05:38:56 +00002041"file(name[, mode[, buffering]]) -> file object\n"
2042"\n"
2043"Open a file. The mode can be 'r', 'w' or 'a' for reading (default),\n"
2044"writing or appending. The file will be created if it doesn't exist\n"
2045"when opened for writing or appending; it will be truncated when\n"
2046"opened for writing. Add a 'b' to the mode for binary files.\n"
2047"Add a '+' to the mode to allow simultaneous reading and writing.\n"
2048"If the buffering argument is given, 0 means unbuffered, 1 means line\n"
Tim Peters742dfd62001-09-13 21:49:44 +00002049"buffered, and larger numbers specify the buffer size.\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00002050)
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00002051PyDoc_STR(
Barry Warsaw4be55b52002-05-22 20:37:53 +00002052"Add a 'U' to mode to open the file for input with universal newline\n"
2053"support. Any line ending in the input file will be seen as a '\\n'\n"
2054"in Python. Also, a file so opened gains the attribute 'newlines';\n"
2055"the value for this attribute is one of None (no newline read yet),\n"
2056"'\\r', '\\n', '\\r\\n' or a tuple containing all the newline types seen.\n"
2057"\n"
2058"'U' cannot be combined with 'w' or '+' mode.\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00002059);
Tim Peters59c9a642001-09-13 05:38:56 +00002060
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002061PyTypeObject PyFile_Type = {
2062 PyObject_HEAD_INIT(&PyType_Type)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00002063 0,
2064 "file",
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002065 sizeof(PyFileObject),
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00002066 0,
Guido van Rossum65967252001-04-21 13:20:18 +00002067 (destructor)file_dealloc, /* tp_dealloc */
2068 0, /* tp_print */
Tim Peters6d6c1a32001-08-02 04:15:00 +00002069 0, /* tp_getattr */
2070 0, /* tp_setattr */
Guido van Rossum65967252001-04-21 13:20:18 +00002071 0, /* tp_compare */
Tim Peters6d6c1a32001-08-02 04:15:00 +00002072 (reprfunc)file_repr, /* tp_repr */
Guido van Rossum65967252001-04-21 13:20:18 +00002073 0, /* tp_as_number */
2074 0, /* tp_as_sequence */
2075 0, /* tp_as_mapping */
2076 0, /* tp_hash */
2077 0, /* tp_call */
2078 0, /* tp_str */
Tim Peters6d6c1a32001-08-02 04:15:00 +00002079 PyObject_GenericGetAttr, /* tp_getattro */
Tim Peters015dd822003-05-04 04:16:52 +00002080 /* softspace is writable: we must supply tp_setattro */
2081 PyObject_GenericSetAttr, /* tp_setattro */
Guido van Rossum65967252001-04-21 13:20:18 +00002082 0, /* tp_as_buffer */
Raymond Hettingercb87bc82004-05-31 00:35:52 +00002083 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | Py_TPFLAGS_HAVE_WEAKREFS, /* tp_flags */
Tim Peters59c9a642001-09-13 05:38:56 +00002084 file_doc, /* tp_doc */
Tim Peters6d6c1a32001-08-02 04:15:00 +00002085 0, /* tp_traverse */
2086 0, /* tp_clear */
Guido van Rossum65967252001-04-21 13:20:18 +00002087 0, /* tp_richcompare */
Raymond Hettingercb87bc82004-05-31 00:35:52 +00002088 offsetof(PyFileObject, weakreflist), /* tp_weaklistoffset */
Guido van Rossum1a5e21e2006-02-28 21:57:43 +00002089 (getiterfunc)file_self, /* tp_iter */
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002090 (iternextfunc)file_iternext, /* tp_iternext */
Tim Peters6d6c1a32001-08-02 04:15:00 +00002091 file_methods, /* tp_methods */
2092 file_memberlist, /* tp_members */
2093 file_getsetlist, /* tp_getset */
2094 0, /* tp_base */
2095 0, /* tp_dict */
Tim Peters59c9a642001-09-13 05:38:56 +00002096 0, /* tp_descr_get */
2097 0, /* tp_descr_set */
2098 0, /* tp_dictoffset */
Georg Brandl347b3002006-03-30 11:57:00 +00002099 file_init, /* tp_init */
Tim Peters44410012001-09-14 03:26:08 +00002100 PyType_GenericAlloc, /* tp_alloc */
Tim Peters59c9a642001-09-13 05:38:56 +00002101 file_new, /* tp_new */
Neil Schemenaueraa769ae2002-04-12 02:44:10 +00002102 PyObject_Del, /* tp_free */
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00002103};
Guido van Rossumeb183da1991-04-04 10:44:06 +00002104
2105/* Interface for the 'soft space' between print items. */
2106
2107int
Fred Drakefd99de62000-07-09 05:02:18 +00002108PyFile_SoftSpace(PyObject *f, int newflag)
Guido van Rossumeb183da1991-04-04 10:44:06 +00002109{
Martin v. Löwis18e16552006-02-15 17:27:45 +00002110 long oldflag = 0;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002111 if (f == NULL) {
2112 /* Do nothing */
2113 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002114 else if (PyFile_Check(f)) {
2115 oldflag = ((PyFileObject *)f)->f_softspace;
2116 ((PyFileObject *)f)->f_softspace = newflag;
Guido van Rossumeb183da1991-04-04 10:44:06 +00002117 }
Guido van Rossum3165fe61992-09-25 21:59:05 +00002118 else {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002119 PyObject *v;
2120 v = PyObject_GetAttrString(f, "softspace");
Guido van Rossum3165fe61992-09-25 21:59:05 +00002121 if (v == NULL)
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002122 PyErr_Clear();
Guido van Rossum3165fe61992-09-25 21:59:05 +00002123 else {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002124 if (PyInt_Check(v))
2125 oldflag = PyInt_AsLong(v);
Martin v. Löwis18e16552006-02-15 17:27:45 +00002126 assert(oldflag < INT_MAX);
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002127 Py_DECREF(v);
Guido van Rossum3165fe61992-09-25 21:59:05 +00002128 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002129 v = PyInt_FromLong((long)newflag);
Guido van Rossum3165fe61992-09-25 21:59:05 +00002130 if (v == NULL)
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002131 PyErr_Clear();
Guido van Rossum3165fe61992-09-25 21:59:05 +00002132 else {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002133 if (PyObject_SetAttrString(f, "softspace", v) != 0)
2134 PyErr_Clear();
2135 Py_DECREF(v);
Guido van Rossum3165fe61992-09-25 21:59:05 +00002136 }
2137 }
Martin v. Löwis18e16552006-02-15 17:27:45 +00002138 return (int)oldflag;
Guido van Rossumeb183da1991-04-04 10:44:06 +00002139}
Guido van Rossum3165fe61992-09-25 21:59:05 +00002140
2141/* Interfaces to write objects/strings to file-like objects */
2142
2143int
Fred Drakefd99de62000-07-09 05:02:18 +00002144PyFile_WriteObject(PyObject *v, PyObject *f, int flags)
Guido van Rossum3165fe61992-09-25 21:59:05 +00002145{
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002146 PyObject *writer, *value, *args, *result;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002147 if (f == NULL) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002148 PyErr_SetString(PyExc_TypeError, "writeobject with NULL file");
Guido van Rossum3165fe61992-09-25 21:59:05 +00002149 return -1;
2150 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002151 else if (PyFile_Check(f)) {
2152 FILE *fp = PyFile_AsFile(f);
Fred Drake086a0f72004-03-19 15:22:36 +00002153#ifdef Py_USING_UNICODE
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00002154 PyObject *enc = ((PyFileObject*)f)->f_encoding;
2155 int result;
Fred Drake086a0f72004-03-19 15:22:36 +00002156#endif
Guido van Rossum3165fe61992-09-25 21:59:05 +00002157 if (fp == NULL) {
2158 err_closed();
2159 return -1;
2160 }
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00002161#ifdef Py_USING_UNICODE
Tim Petersf1827cf2003-09-07 03:30:18 +00002162 if ((flags & Py_PRINT_RAW) &&
Martin v. Löwis415da6e2003-05-18 12:56:25 +00002163 PyUnicode_Check(v) && enc != Py_None) {
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00002164 char *cenc = PyString_AS_STRING(enc);
2165 value = PyUnicode_AsEncodedString(v, cenc, "strict");
2166 if (value == NULL)
2167 return -1;
2168 } else {
2169 value = v;
2170 Py_INCREF(value);
2171 }
2172 result = PyObject_Print(value, fp, flags);
2173 Py_DECREF(value);
2174 return result;
2175#else
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002176 return PyObject_Print(v, fp, flags);
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00002177#endif
Guido van Rossum3165fe61992-09-25 21:59:05 +00002178 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002179 writer = PyObject_GetAttrString(f, "write");
Guido van Rossum3165fe61992-09-25 21:59:05 +00002180 if (writer == NULL)
2181 return -1;
Martin v. Löwis2777c022001-09-19 13:47:32 +00002182 if (flags & Py_PRINT_RAW) {
2183 if (PyUnicode_Check(v)) {
2184 value = v;
2185 Py_INCREF(value);
2186 } else
2187 value = PyObject_Str(v);
2188 }
2189 else
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002190 value = PyObject_Repr(v);
Guido van Rossumc6004111993-11-05 10:22:19 +00002191 if (value == NULL) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002192 Py_DECREF(writer);
Guido van Rossumc6004111993-11-05 10:22:19 +00002193 return -1;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002194 }
Raymond Hettinger8ae46892003-10-12 19:09:37 +00002195 args = PyTuple_Pack(1, value);
Guido van Rossume9eec541997-05-22 14:02:25 +00002196 if (args == NULL) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002197 Py_DECREF(value);
2198 Py_DECREF(writer);
Guido van Rossumd3f9a1a1995-07-10 23:32:26 +00002199 return -1;
2200 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002201 result = PyEval_CallObject(writer, args);
2202 Py_DECREF(args);
2203 Py_DECREF(value);
2204 Py_DECREF(writer);
Guido van Rossum3165fe61992-09-25 21:59:05 +00002205 if (result == NULL)
2206 return -1;
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002207 Py_DECREF(result);
Guido van Rossum3165fe61992-09-25 21:59:05 +00002208 return 0;
2209}
2210
Guido van Rossum27a60b11997-05-22 22:25:11 +00002211int
Tim Petersc1bbcb82001-11-28 22:13:25 +00002212PyFile_WriteString(const char *s, PyObject *f)
Guido van Rossum3165fe61992-09-25 21:59:05 +00002213{
2214 if (f == NULL) {
Guido van Rossum27a60b11997-05-22 22:25:11 +00002215 /* Should be caused by a pre-existing error */
Fred Drakefd99de62000-07-09 05:02:18 +00002216 if (!PyErr_Occurred())
Guido van Rossum27a60b11997-05-22 22:25:11 +00002217 PyErr_SetString(PyExc_SystemError,
2218 "null file for PyFile_WriteString");
2219 return -1;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002220 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002221 else if (PyFile_Check(f)) {
2222 FILE *fp = PyFile_AsFile(f);
Guido van Rossum27a60b11997-05-22 22:25:11 +00002223 if (fp == NULL) {
2224 err_closed();
2225 return -1;
2226 }
2227 fputs(s, fp);
2228 return 0;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002229 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002230 else if (!PyErr_Occurred()) {
2231 PyObject *v = PyString_FromString(s);
Guido van Rossum27a60b11997-05-22 22:25:11 +00002232 int err;
2233 if (v == NULL)
2234 return -1;
2235 err = PyFile_WriteObject(v, f, Py_PRINT_RAW);
2236 Py_DECREF(v);
2237 return err;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002238 }
Guido van Rossum74ba2471997-07-13 03:56:50 +00002239 else
2240 return -1;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002241}
Andrew M. Kuchling06051ed2000-07-13 23:56:54 +00002242
2243/* Try to get a file-descriptor from a Python object. If the object
2244 is an integer or long integer, its value is returned. If not, the
2245 object's fileno() method is called if it exists; the method must return
2246 an integer or long integer, which is returned as the file descriptor value.
2247 -1 is returned on failure.
2248*/
2249
2250int PyObject_AsFileDescriptor(PyObject *o)
2251{
2252 int fd;
2253 PyObject *meth;
2254
2255 if (PyInt_Check(o)) {
2256 fd = PyInt_AsLong(o);
2257 }
2258 else if (PyLong_Check(o)) {
2259 fd = PyLong_AsLong(o);
2260 }
2261 else if ((meth = PyObject_GetAttrString(o, "fileno")) != NULL)
2262 {
2263 PyObject *fno = PyEval_CallObject(meth, NULL);
2264 Py_DECREF(meth);
2265 if (fno == NULL)
2266 return -1;
Tim Peters86821b22001-01-07 21:19:34 +00002267
Andrew M. Kuchling06051ed2000-07-13 23:56:54 +00002268 if (PyInt_Check(fno)) {
2269 fd = PyInt_AsLong(fno);
2270 Py_DECREF(fno);
2271 }
2272 else if (PyLong_Check(fno)) {
2273 fd = PyLong_AsLong(fno);
2274 Py_DECREF(fno);
2275 }
2276 else {
2277 PyErr_SetString(PyExc_TypeError,
2278 "fileno() returned a non-integer");
2279 Py_DECREF(fno);
2280 return -1;
2281 }
2282 }
2283 else {
2284 PyErr_SetString(PyExc_TypeError,
2285 "argument must be an int, or have a fileno() method.");
2286 return -1;
2287 }
2288
2289 if (fd < 0) {
2290 PyErr_Format(PyExc_ValueError,
2291 "file descriptor cannot be a negative integer (%i)",
2292 fd);
2293 return -1;
2294 }
2295 return fd;
2296}
Jack Jansen7b8c7542002-04-14 20:12:41 +00002297
Jack Jansen7b8c7542002-04-14 20:12:41 +00002298/* From here on we need access to the real fgets and fread */
2299#undef fgets
2300#undef fread
2301
2302/*
2303** Py_UniversalNewlineFgets is an fgets variation that understands
2304** all of \r, \n and \r\n conventions.
2305** The stream should be opened in binary mode.
2306** If fobj is NULL the routine always does newline conversion, and
2307** it may peek one char ahead to gobble the second char in \r\n.
2308** If fobj is non-NULL it must be a PyFileObject. In this case there
2309** is no readahead but in stead a flag is used to skip a following
2310** \n on the next read. Also, if the file is open in binary mode
2311** the whole conversion is skipped. Finally, the routine keeps track of
2312** the different types of newlines seen.
2313** Note that we need no error handling: fgets() treats error and eof
2314** identically.
2315*/
2316char *
2317Py_UniversalNewlineFgets(char *buf, int n, FILE *stream, PyObject *fobj)
2318{
2319 char *p = buf;
2320 int c;
2321 int newlinetypes = 0;
2322 int skipnextlf = 0;
2323 int univ_newline = 1;
Tim Peters058b1412002-04-21 07:29:14 +00002324
Jack Jansen7b8c7542002-04-14 20:12:41 +00002325 if (fobj) {
2326 if (!PyFile_Check(fobj)) {
2327 errno = ENXIO; /* What can you do... */
2328 return NULL;
2329 }
2330 univ_newline = ((PyFileObject *)fobj)->f_univ_newline;
2331 if ( !univ_newline )
2332 return fgets(buf, n, stream);
2333 newlinetypes = ((PyFileObject *)fobj)->f_newlinetypes;
2334 skipnextlf = ((PyFileObject *)fobj)->f_skipnextlf;
2335 }
2336 FLOCKFILE(stream);
2337 c = 'x'; /* Shut up gcc warning */
2338 while (--n > 0 && (c = GETC(stream)) != EOF ) {
2339 if (skipnextlf ) {
2340 skipnextlf = 0;
2341 if (c == '\n') {
2342 /* Seeing a \n here with skipnextlf true
2343 ** means we saw a \r before.
2344 */
2345 newlinetypes |= NEWLINE_CRLF;
2346 c = GETC(stream);
2347 if (c == EOF) break;
2348 } else {
2349 /*
2350 ** Note that c == EOF also brings us here,
2351 ** so we're okay if the last char in the file
2352 ** is a CR.
2353 */
2354 newlinetypes |= NEWLINE_CR;
2355 }
2356 }
2357 if (c == '\r') {
2358 /* A \r is translated into a \n, and we skip
2359 ** an adjacent \n, if any. We don't set the
2360 ** newlinetypes flag until we've seen the next char.
2361 */
2362 skipnextlf = 1;
2363 c = '\n';
2364 } else if ( c == '\n') {
2365 newlinetypes |= NEWLINE_LF;
2366 }
2367 *p++ = c;
2368 if (c == '\n') break;
2369 }
2370 if ( c == EOF && skipnextlf )
2371 newlinetypes |= NEWLINE_CR;
2372 FUNLOCKFILE(stream);
2373 *p = '\0';
2374 if (fobj) {
2375 ((PyFileObject *)fobj)->f_newlinetypes = newlinetypes;
2376 ((PyFileObject *)fobj)->f_skipnextlf = skipnextlf;
2377 } else if ( skipnextlf ) {
2378 /* If we have no file object we cannot save the
2379 ** skipnextlf flag. We have to readahead, which
2380 ** will cause a pause if we're reading from an
2381 ** interactive stream, but that is very unlikely
2382 ** unless we're doing something silly like
2383 ** execfile("/dev/tty").
2384 */
2385 c = GETC(stream);
2386 if ( c != '\n' )
2387 ungetc(c, stream);
2388 }
2389 if (p == buf)
2390 return NULL;
2391 return buf;
2392}
2393
2394/*
2395** Py_UniversalNewlineFread is an fread variation that understands
2396** all of \r, \n and \r\n conventions.
2397** The stream should be opened in binary mode.
2398** fobj must be a PyFileObject. In this case there
2399** is no readahead but in stead a flag is used to skip a following
2400** \n on the next read. Also, if the file is open in binary mode
2401** the whole conversion is skipped. Finally, the routine keeps track of
2402** the different types of newlines seen.
2403*/
2404size_t
Tim Peters058b1412002-04-21 07:29:14 +00002405Py_UniversalNewlineFread(char *buf, size_t n,
Jack Jansen7b8c7542002-04-14 20:12:41 +00002406 FILE *stream, PyObject *fobj)
2407{
Tim Peters058b1412002-04-21 07:29:14 +00002408 char *dst = buf;
2409 PyFileObject *f = (PyFileObject *)fobj;
2410 int newlinetypes, skipnextlf;
2411
2412 assert(buf != NULL);
2413 assert(stream != NULL);
2414
Jack Jansen7b8c7542002-04-14 20:12:41 +00002415 if (!fobj || !PyFile_Check(fobj)) {
2416 errno = ENXIO; /* What can you do... */
Neal Norwitzcb3319f2003-02-09 01:10:02 +00002417 return 0;
Jack Jansen7b8c7542002-04-14 20:12:41 +00002418 }
Tim Peters058b1412002-04-21 07:29:14 +00002419 if (!f->f_univ_newline)
Jack Jansen7b8c7542002-04-14 20:12:41 +00002420 return fread(buf, 1, n, stream);
Tim Peters058b1412002-04-21 07:29:14 +00002421 newlinetypes = f->f_newlinetypes;
2422 skipnextlf = f->f_skipnextlf;
2423 /* Invariant: n is the number of bytes remaining to be filled
2424 * in the buffer.
2425 */
2426 while (n) {
2427 size_t nread;
2428 int shortread;
2429 char *src = dst;
2430
2431 nread = fread(dst, 1, n, stream);
2432 assert(nread <= n);
Neal Norwitzcb3319f2003-02-09 01:10:02 +00002433 if (nread == 0)
2434 break;
2435
Tim Peterse1682a82002-04-21 18:15:20 +00002436 n -= nread; /* assuming 1 byte out for each in; will adjust */
2437 shortread = n != 0; /* true iff EOF or error */
Tim Peters058b1412002-04-21 07:29:14 +00002438 while (nread--) {
2439 char c = *src++;
Jack Jansen7b8c7542002-04-14 20:12:41 +00002440 if (c == '\r') {
Tim Peters058b1412002-04-21 07:29:14 +00002441 /* Save as LF and set flag to skip next LF. */
Jack Jansen7b8c7542002-04-14 20:12:41 +00002442 *dst++ = '\n';
2443 skipnextlf = 1;
Tim Peters058b1412002-04-21 07:29:14 +00002444 }
2445 else if (skipnextlf && c == '\n') {
2446 /* Skip LF, and remember we saw CR LF. */
Jack Jansen7b8c7542002-04-14 20:12:41 +00002447 skipnextlf = 0;
2448 newlinetypes |= NEWLINE_CRLF;
Tim Peterse1682a82002-04-21 18:15:20 +00002449 ++n;
Tim Peters058b1412002-04-21 07:29:14 +00002450 }
2451 else {
2452 /* Normal char to be stored in buffer. Also
2453 * update the newlinetypes flag if either this
2454 * is an LF or the previous char was a CR.
2455 */
Jack Jansen7b8c7542002-04-14 20:12:41 +00002456 if (c == '\n')
2457 newlinetypes |= NEWLINE_LF;
2458 else if (skipnextlf)
2459 newlinetypes |= NEWLINE_CR;
2460 *dst++ = c;
2461 skipnextlf = 0;
2462 }
2463 }
Tim Peters058b1412002-04-21 07:29:14 +00002464 if (shortread) {
2465 /* If this is EOF, update type flags. */
2466 if (skipnextlf && feof(stream))
2467 newlinetypes |= NEWLINE_CR;
2468 break;
2469 }
Jack Jansen7b8c7542002-04-14 20:12:41 +00002470 }
Tim Peters058b1412002-04-21 07:29:14 +00002471 f->f_newlinetypes = newlinetypes;
2472 f->f_skipnextlf = skipnextlf;
2473 return dst - buf;
Jack Jansen7b8c7542002-04-14 20:12:41 +00002474}
Anthony Baxterac6bd462006-04-13 02:06:09 +00002475
2476#ifdef __cplusplus
2477}
2478#endif