blob: b34dd5259269f6f9539c150f13041dcc45b5736e [file] [log] [blame]
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001/* File object implementation */
2
Guido van Rossumc0b618a1997-05-02 03:12:38 +00003#include "Python.h"
Guido van Rossumb6775db1994-08-01 11:34:53 +00004#include "structmember.h"
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00005
Guido van Rossumff7e83d1999-08-27 20:39:37 +00006#ifndef DONT_HAVE_SYS_TYPES_H
Guido van Rossum41498431999-01-07 22:09:51 +00007#include <sys/types.h>
Guido van Rossumff7e83d1999-08-27 20:39:37 +00008#endif /* DONT_HAVE_SYS_TYPES_H */
Guido van Rossum41498431999-01-07 22:09:51 +00009
Martin v. Löwis6238d2b2002-06-30 15:26:10 +000010#ifdef MS_WINDOWS
Guido van Rossumb8199141997-05-06 15:23:24 +000011#define fileno _fileno
Tim Petersfb05db22002-03-11 00:24:00 +000012/* can simulate truncate with Win32 API functions; see file_truncate */
Guido van Rossumb8199141997-05-06 15:23:24 +000013#define HAVE_FTRUNCATE
Tim Peters7a1f9172002-07-14 22:14:19 +000014#define WIN32_LEAN_AND_MEAN
Tim Petersfb05db22002-03-11 00:24:00 +000015#include <windows.h>
Guido van Rossumb8199141997-05-06 15:23:24 +000016#endif
17
Mark Hammondc2e85bd2002-10-03 05:10:39 +000018#ifdef _MSC_VER
19/* Need GetVersion to see if on NT so safe to use _wfopen */
20#define WIN32_LEAN_AND_MEAN
21#include <windows.h>
22#endif /* _MSC_VER */
23
Andrew MacIntyrec4874392002-02-26 11:36:35 +000024#if defined(PYOS_OS2) && defined(PYCC_GCC)
25#include <io.h>
26#endif
27
Guido van Rossumc0b618a1997-05-02 03:12:38 +000028#define BUF(v) PyString_AS_STRING((PyStringObject *)v)
Guido van Rossumce5ba841991-03-06 13:06:18 +000029
Guido van Rossumff7e83d1999-08-27 20:39:37 +000030#ifndef DONT_HAVE_ERRNO_H
Guido van Rossumf1dc5661993-07-05 10:31:29 +000031#include <errno.h>
Guido van Rossumff7e83d1999-08-27 20:39:37 +000032#endif
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000033
Jack Jansen7b8c7542002-04-14 20:12:41 +000034#ifdef HAVE_GETC_UNLOCKED
35#define GETC(f) getc_unlocked(f)
36#define FLOCKFILE(f) flockfile(f)
37#define FUNLOCKFILE(f) funlockfile(f)
38#else
39#define GETC(f) getc(f)
40#define FLOCKFILE(f)
41#define FUNLOCKFILE(f)
42#endif
43
Jack Jansen7b8c7542002-04-14 20:12:41 +000044/* Bits in f_newlinetypes */
45#define NEWLINE_UNKNOWN 0 /* No newline seen, yet */
46#define NEWLINE_CR 1 /* \r newline seen */
47#define NEWLINE_LF 2 /* \n newline seen */
48#define NEWLINE_CRLF 4 /* \r\n newline seen */
Trent Mickf29f47b2000-08-11 19:02:59 +000049
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000050FILE *
Fred Drakefd99de62000-07-09 05:02:18 +000051PyFile_AsFile(PyObject *f)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000052{
Guido van Rossumc0b618a1997-05-02 03:12:38 +000053 if (f == NULL || !PyFile_Check(f))
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000054 return NULL;
Guido van Rossum3165fe61992-09-25 21:59:05 +000055 else
Guido van Rossumc0b618a1997-05-02 03:12:38 +000056 return ((PyFileObject *)f)->f_fp;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000057}
58
Guido van Rossumc0b618a1997-05-02 03:12:38 +000059PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +000060PyFile_Name(PyObject *f)
Guido van Rossumdb3165e1993-10-18 17:06:59 +000061{
Guido van Rossumc0b618a1997-05-02 03:12:38 +000062 if (f == NULL || !PyFile_Check(f))
Guido van Rossumdb3165e1993-10-18 17:06:59 +000063 return NULL;
64 else
Guido van Rossumc0b618a1997-05-02 03:12:38 +000065 return ((PyFileObject *)f)->f_name;
Guido van Rossumdb3165e1993-10-18 17:06:59 +000066}
67
Neil Schemenauered19b882002-03-23 02:06:50 +000068/* On Unix, fopen will succeed for directories.
69 In Python, there should be no file objects referring to
70 directories, so we need a check. */
71
72static PyFileObject*
73dircheck(PyFileObject* f)
74{
75#if defined(HAVE_FSTAT) && defined(S_IFDIR) && defined(EISDIR)
76 struct stat buf;
77 if (f->f_fp == NULL)
78 return f;
79 if (fstat(fileno(f->f_fp), &buf) == 0 &&
80 S_ISDIR(buf.st_mode)) {
81#ifdef HAVE_STRERROR
82 char *msg = strerror(EISDIR);
83#else
84 char *msg = "Is a directory";
85#endif
Tim Petersf1827cf2003-09-07 03:30:18 +000086 PyObject *exc = PyObject_CallFunction(PyExc_IOError, "(is)",
Jeremy Hylton8b735422002-08-14 21:01:41 +000087 EISDIR, msg);
Neil Schemenauered19b882002-03-23 02:06:50 +000088 PyErr_SetObject(PyExc_IOError, exc);
Neal Norwitz98cad482003-08-15 20:05:45 +000089 Py_XDECREF(exc);
Neil Schemenauered19b882002-03-23 02:06:50 +000090 return NULL;
91 }
92#endif
93 return f;
94}
95
Tim Peters59c9a642001-09-13 05:38:56 +000096
97static PyObject *
Nicholas Bastinabce8a62004-03-21 20:24:07 +000098fill_file_fields(PyFileObject *f, FILE *fp, PyObject *name, char *mode,
99 int (*close)(FILE *))
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000100{
Tim Peters59c9a642001-09-13 05:38:56 +0000101 assert(f != NULL);
102 assert(PyFile_Check(f));
Tim Peters44410012001-09-14 03:26:08 +0000103 assert(f->f_fp == NULL);
104
105 Py_DECREF(f->f_name);
106 Py_DECREF(f->f_mode);
Martin v. Löwis5467d4c2003-05-10 07:10:12 +0000107 Py_DECREF(f->f_encoding);
Nicholas Bastinabce8a62004-03-21 20:24:07 +0000108
109 Py_INCREF (name);
110 f->f_name = name;
111
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000112 f->f_mode = PyString_FromString(mode);
Tim Peters44410012001-09-14 03:26:08 +0000113
Guido van Rossuma1ab7fa1991-06-04 19:37:39 +0000114 f->f_close = close;
Guido van Rossumeb183da1991-04-04 10:44:06 +0000115 f->f_softspace = 0;
Tim Peters59c9a642001-09-13 05:38:56 +0000116 f->f_binary = strchr(mode,'b') != NULL;
Guido van Rossum7a6e9592002-08-06 15:55:28 +0000117 f->f_buf = NULL;
Jack Jansen7b8c7542002-04-14 20:12:41 +0000118 f->f_univ_newline = (strchr(mode, 'U') != NULL);
119 f->f_newlinetypes = NEWLINE_UNKNOWN;
120 f->f_skipnextlf = 0;
Martin v. Löwis5467d4c2003-05-10 07:10:12 +0000121 Py_INCREF(Py_None);
122 f->f_encoding = Py_None;
Tim Petersf1827cf2003-09-07 03:30:18 +0000123
Tim Peters59c9a642001-09-13 05:38:56 +0000124 if (f->f_name == NULL || f->f_mode == NULL)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000125 return NULL;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000126 f->f_fp = fp;
Neil Schemenauered19b882002-03-23 02:06:50 +0000127 f = dircheck(f);
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000128 return (PyObject *) f;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000129}
130
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000131/* check for known incorrect mode strings - problem is, platforms are
132 free to accept any mode characters they like and are supposed to
133 ignore stuff they don't understand... write or append mode with
134 universal newline support is expressly forbidden by PEP 278. */
135/* zero return is kewl - one is un-kewl */
136static int
137check_the_mode(char *mode)
138{
Neal Norwitz76dc0812006-01-08 06:13:13 +0000139 size_t len = strlen(mode);
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000140
141 switch (len) {
142 case 0:
143 PyErr_SetString(PyExc_ValueError, "empty mode string");
144 return 1;
145
146 /* reject wU, aU */
147 case 2:
148 switch (mode[0]) {
149 case 'w':
150 case 'a':
151 if (mode[1] == 'U') {
152 PyErr_SetString(PyExc_ValueError,
153 "invalid mode string");
154 return 1;
155 }
156 break;
157 }
158 break;
159
160 /* reject w+U, a+U, wU+, aU+ */
161 case 3:
162 switch (mode[0]) {
163 case 'w':
164 case 'a':
165 if ((mode[1] == '+' && mode[2] == 'U') ||
166 (mode[1] == 'U' && mode[2] == '+')) {
167 PyErr_SetString(PyExc_ValueError,
168 "invalid mode string");
169 return 1;
170 }
171 break;
172 }
173 break;
174 }
175
176 return 0;
177}
178
Tim Peters59c9a642001-09-13 05:38:56 +0000179static PyObject *
180open_the_file(PyFileObject *f, char *name, char *mode)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000181{
Tim Peters59c9a642001-09-13 05:38:56 +0000182 assert(f != NULL);
183 assert(PyFile_Check(f));
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000184#ifdef MS_WINDOWS
185 /* windows ignores the passed name in order to support Unicode */
186 assert(f->f_name != NULL);
187#else
Tim Peters59c9a642001-09-13 05:38:56 +0000188 assert(name != NULL);
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000189#endif
Tim Peters59c9a642001-09-13 05:38:56 +0000190 assert(mode != NULL);
Tim Peters44410012001-09-14 03:26:08 +0000191 assert(f->f_fp == NULL);
Tim Peters59c9a642001-09-13 05:38:56 +0000192
Skip Montanarobbf12ba2005-05-20 03:07:06 +0000193 if (check_the_mode(mode))
194 return NULL;
195
Tim Peters8fa45672001-09-13 21:01:29 +0000196 /* rexec.py can't stop a user from getting the file() constructor --
197 all they have to do is get *any* file object f, and then do
198 type(f). Here we prevent them from doing damage with it. */
199 if (PyEval_GetRestricted()) {
200 PyErr_SetString(PyExc_IOError,
Jeremy Hylton8b735422002-08-14 21:01:41 +0000201 "file() constructor not accessible in restricted mode");
Tim Peters8fa45672001-09-13 21:01:29 +0000202 return NULL;
203 }
Tim Petersa27a1502001-11-09 20:59:14 +0000204 errno = 0;
Skip Montanaro51ffac62004-06-11 04:49:03 +0000205
206 if (strcmp(mode, "U") == 0 || strcmp(mode, "rU") == 0)
207 mode = "rb";
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000208#ifdef MS_WINDOWS
Skip Montanaro51ffac62004-06-11 04:49:03 +0000209 if (PyUnicode_Check(f->f_name)) {
210 PyObject *wmode;
211 wmode = PyUnicode_DecodeASCII(mode, strlen(mode), NULL);
212 if (f->f_name && wmode) {
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000213 Py_BEGIN_ALLOW_THREADS
Skip Montanaro51ffac62004-06-11 04:49:03 +0000214 /* PyUnicode_AS_UNICODE OK without thread
215 lock as it is a simple dereference. */
216 f->f_fp = _wfopen(PyUnicode_AS_UNICODE(f->f_name),
217 PyUnicode_AS_UNICODE(wmode));
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000218 Py_END_ALLOW_THREADS
219 }
Skip Montanaro51ffac62004-06-11 04:49:03 +0000220 Py_XDECREF(wmode);
Guido van Rossumff4949e1992-08-05 19:58:53 +0000221 }
Skip Montanaro51ffac62004-06-11 04:49:03 +0000222#endif
223 if (NULL == f->f_fp && NULL != name) {
224 Py_BEGIN_ALLOW_THREADS
225 f->f_fp = fopen(name, mode);
226 Py_END_ALLOW_THREADS
227 }
228
Guido van Rossuma08095a1991-02-13 23:25:27 +0000229 if (f->f_fp == NULL) {
Tim Peters2ea91112002-04-08 04:13:12 +0000230#ifdef _MSC_VER
231 /* MSVC 6 (Microsoft) leaves errno at 0 for bad mode strings,
232 * across all Windows flavors. When it sets EINVAL varies
233 * across Windows flavors, the exact conditions aren't
234 * documented, and the answer lies in the OS's implementation
235 * of Win32's CreateFile function (whose source is secret).
236 * Seems the best we can do is map EINVAL to ENOENT.
237 */
238 if (errno == 0) /* bad mode string */
239 errno = EINVAL;
240 else if (errno == EINVAL) /* unknown, but not a mode string */
241 errno = ENOENT;
242#endif
Jeremy Hylton41c83212001-11-09 16:17:24 +0000243 if (errno == EINVAL)
Tim Peters2ea91112002-04-08 04:13:12 +0000244 PyErr_Format(PyExc_IOError, "invalid mode: %s",
Jeremy Hylton41c83212001-11-09 16:17:24 +0000245 mode);
246 else
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000247 PyErr_SetFromErrnoWithFilenameObject(PyExc_IOError, f->f_name);
Tim Peters59c9a642001-09-13 05:38:56 +0000248 f = NULL;
249 }
Tim Peters2ea91112002-04-08 04:13:12 +0000250 if (f != NULL)
Neil Schemenauered19b882002-03-23 02:06:50 +0000251 f = dircheck(f);
Tim Peters59c9a642001-09-13 05:38:56 +0000252 return (PyObject *)f;
253}
254
255PyObject *
256PyFile_FromFile(FILE *fp, char *name, char *mode, int (*close)(FILE *))
257{
Tim Peters44410012001-09-14 03:26:08 +0000258 PyFileObject *f = (PyFileObject *)PyFile_Type.tp_new(&PyFile_Type,
259 NULL, NULL);
Tim Peters59c9a642001-09-13 05:38:56 +0000260 if (f != NULL) {
Nicholas Bastinabce8a62004-03-21 20:24:07 +0000261 PyObject *o_name = PyString_FromString(name);
262 if (fill_file_fields(f, fp, o_name, mode, close) == NULL) {
Tim Peters59c9a642001-09-13 05:38:56 +0000263 Py_DECREF(f);
264 f = NULL;
265 }
Nicholas Bastinabce8a62004-03-21 20:24:07 +0000266 Py_DECREF(o_name);
Tim Peters59c9a642001-09-13 05:38:56 +0000267 }
268 return (PyObject *) f;
269}
270
271PyObject *
272PyFile_FromString(char *name, char *mode)
273{
274 extern int fclose(FILE *);
275 PyFileObject *f;
276
277 f = (PyFileObject *)PyFile_FromFile((FILE *)NULL, name, mode, fclose);
278 if (f != NULL) {
279 if (open_the_file(f, name, mode) == NULL) {
280 Py_DECREF(f);
281 f = NULL;
282 }
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000283 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000284 return (PyObject *)f;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000285}
286
Guido van Rossumb6775db1994-08-01 11:34:53 +0000287void
Fred Drakefd99de62000-07-09 05:02:18 +0000288PyFile_SetBufSize(PyObject *f, int bufsize)
Guido van Rossumb6775db1994-08-01 11:34:53 +0000289{
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000290 PyFileObject *file = (PyFileObject *)f;
Guido van Rossumb6775db1994-08-01 11:34:53 +0000291 if (bufsize >= 0) {
Guido van Rossumb6775db1994-08-01 11:34:53 +0000292 int type;
293 switch (bufsize) {
294 case 0:
295 type = _IONBF;
296 break;
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000297#ifdef HAVE_SETVBUF
Guido van Rossumb6775db1994-08-01 11:34:53 +0000298 case 1:
299 type = _IOLBF;
300 bufsize = BUFSIZ;
301 break;
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000302#endif
Guido van Rossumb6775db1994-08-01 11:34:53 +0000303 default:
304 type = _IOFBF;
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000305#ifndef HAVE_SETVBUF
306 bufsize = BUFSIZ;
307#endif
308 break;
Guido van Rossumb6775db1994-08-01 11:34:53 +0000309 }
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000310 fflush(file->f_fp);
311 if (type == _IONBF) {
312 PyMem_Free(file->f_setbuf);
313 file->f_setbuf = NULL;
314 } else {
315 file->f_setbuf = PyMem_Realloc(file->f_setbuf, bufsize);
316 }
317#ifdef HAVE_SETVBUF
318 setvbuf(file->f_fp, file->f_setbuf, type, bufsize);
Guido van Rossumf8b4de01998-03-06 15:32:40 +0000319#else /* !HAVE_SETVBUF */
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +0000320 setbuf(file->f_fp, file->f_setbuf);
Guido van Rossumf8b4de01998-03-06 15:32:40 +0000321#endif /* !HAVE_SETVBUF */
Guido van Rossumb6775db1994-08-01 11:34:53 +0000322 }
323}
324
Martin v. Löwis5467d4c2003-05-10 07:10:12 +0000325/* Set the encoding used to output Unicode strings.
326 Returh 1 on success, 0 on failure. */
327
328int
329PyFile_SetEncoding(PyObject *f, const char *enc)
330{
331 PyFileObject *file = (PyFileObject*)f;
332 PyObject *str = PyString_FromString(enc);
333 if (!str)
334 return 0;
335 Py_DECREF(file->f_encoding);
336 file->f_encoding = str;
337 return 1;
338}
339
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000340static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000341err_closed(void)
Guido van Rossumd7297e61992-07-06 14:19:26 +0000342{
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000343 PyErr_SetString(PyExc_ValueError, "I/O operation on closed file");
Guido van Rossumd7297e61992-07-06 14:19:26 +0000344 return NULL;
345}
346
Neal Norwitzd8b995f2002-08-06 21:50:54 +0000347static void drop_readahead(PyFileObject *);
Guido van Rossum7a6e9592002-08-06 15:55:28 +0000348
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000349/* Methods */
350
351static void
Fred Drakefd99de62000-07-09 05:02:18 +0000352file_dealloc(PyFileObject *f)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000353{
Peter Astrandf8e74b12004-11-07 14:15:28 +0000354 int sts = 0;
Raymond Hettingercb87bc82004-05-31 00:35:52 +0000355 if (f->weakreflist != NULL)
356 PyObject_ClearWeakRefs((PyObject *) f);
Guido van Rossumff4949e1992-08-05 19:58:53 +0000357 if (f->f_fp != NULL && f->f_close != NULL) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000358 Py_BEGIN_ALLOW_THREADS
Peter Astrandf8e74b12004-11-07 14:15:28 +0000359 sts = (*f->f_close)(f->f_fp);
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000360 Py_END_ALLOW_THREADS
Peter Astrandf8e74b12004-11-07 14:15:28 +0000361 if (sts == EOF)
362#ifdef HAVE_STRERROR
363 PySys_WriteStderr("close failed: [Errno %d] %s\n", errno, strerror(errno));
364#else
365 PySys_WriteStderr("close failed: [Errno %d]\n", errno);
366#endif
Guido van Rossumff4949e1992-08-05 19:58:53 +0000367 }
Andrew MacIntyre4e10ed32004-04-04 07:01:35 +0000368 PyMem_Free(f->f_setbuf);
Tim Peters44410012001-09-14 03:26:08 +0000369 Py_XDECREF(f->f_name);
370 Py_XDECREF(f->f_mode);
Martin v. Löwis5467d4c2003-05-10 07:10:12 +0000371 Py_XDECREF(f->f_encoding);
Guido van Rossum7a6e9592002-08-06 15:55:28 +0000372 drop_readahead(f);
Guido van Rossum9475a232001-10-05 20:51:39 +0000373 f->ob_type->tp_free((PyObject *)f);
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000374}
375
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000376static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000377file_repr(PyFileObject *f)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000378{
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000379 if (PyUnicode_Check(f->f_name)) {
Martin v. Löwis0073f2e2002-11-21 23:52:35 +0000380#ifdef Py_USING_UNICODE
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000381 PyObject *ret = NULL;
382 PyObject *name;
383 name = PyUnicode_AsUnicodeEscapeString(f->f_name);
384 ret = PyString_FromFormat("<%s file u'%s', mode '%s' at %p>",
385 f->f_fp == NULL ? "closed" : "open",
386 PyString_AsString(name),
387 PyString_AsString(f->f_mode),
388 f);
389 Py_XDECREF(name);
390 return ret;
Martin v. Löwis0073f2e2002-11-21 23:52:35 +0000391#endif
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000392 } else {
393 return PyString_FromFormat("<%s file '%s', mode '%s' at %p>",
Barry Warsaw7ce36942001-08-24 18:34:26 +0000394 f->f_fp == NULL ? "closed" : "open",
395 PyString_AsString(f->f_name),
396 PyString_AsString(f->f_mode),
397 f);
Mark Hammondc2e85bd2002-10-03 05:10:39 +0000398 }
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000399}
400
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000401static PyObject *
Martin v. Löwise3eb1f22001-08-16 13:15:00 +0000402file_close(PyFileObject *f)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000403{
Guido van Rossuma1ab7fa1991-06-04 19:37:39 +0000404 int sts = 0;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000405 if (f->f_fp != NULL) {
Guido van Rossumff4949e1992-08-05 19:58:53 +0000406 if (f->f_close != NULL) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000407 Py_BEGIN_ALLOW_THREADS
Guido van Rossumff4949e1992-08-05 19:58:53 +0000408 errno = 0;
Guido van Rossuma1ab7fa1991-06-04 19:37:39 +0000409 sts = (*f->f_close)(f->f_fp);
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000410 Py_END_ALLOW_THREADS
Guido van Rossumff4949e1992-08-05 19:58:53 +0000411 }
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000412 f->f_fp = NULL;
413 }
Martin v. Löwis7bbcde72003-09-07 20:42:29 +0000414 PyMem_Free(f->f_setbuf);
Andrew MacIntyre4e10ed32004-04-04 07:01:35 +0000415 f->f_setbuf = NULL;
Guido van Rossumfebd5511992-03-04 16:39:24 +0000416 if (sts == EOF)
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000417 return PyErr_SetFromErrno(PyExc_IOError);
Guido van Rossuma1ab7fa1991-06-04 19:37:39 +0000418 if (sts != 0)
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000419 return PyInt_FromLong((long)sts);
420 Py_INCREF(Py_None);
421 return Py_None;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000422}
423
Trent Mickf29f47b2000-08-11 19:02:59 +0000424
Guido van Rossumb8552162001-09-05 14:58:11 +0000425/* Our very own off_t-like type, 64-bit if possible */
426#if !defined(HAVE_LARGEFILE_SUPPORT)
427typedef off_t Py_off_t;
428#elif SIZEOF_OFF_T >= 8
429typedef off_t Py_off_t;
430#elif SIZEOF_FPOS_T >= 8
Guido van Rossum4f53da02001-03-01 18:26:53 +0000431typedef fpos_t Py_off_t;
432#else
Guido van Rossumb8552162001-09-05 14:58:11 +0000433#error "Large file support, but neither off_t nor fpos_t is large enough."
Guido van Rossum4f53da02001-03-01 18:26:53 +0000434#endif
435
436
Trent Mickf29f47b2000-08-11 19:02:59 +0000437/* a portable fseek() function
438 return 0 on success, non-zero on failure (with errno set) */
Guido van Rossumf68d8e52001-04-14 17:55:09 +0000439static int
Guido van Rossum4f53da02001-03-01 18:26:53 +0000440_portable_fseek(FILE *fp, Py_off_t offset, int whence)
Trent Mickf29f47b2000-08-11 19:02:59 +0000441{
Guido van Rossumb8552162001-09-05 14:58:11 +0000442#if !defined(HAVE_LARGEFILE_SUPPORT)
443 return fseek(fp, offset, whence);
444#elif defined(HAVE_FSEEKO) && SIZEOF_OFF_T >= 8
Trent Mickf29f47b2000-08-11 19:02:59 +0000445 return fseeko(fp, offset, whence);
446#elif defined(HAVE_FSEEK64)
447 return fseek64(fp, offset, whence);
Fred Drakedb810ac2000-10-06 20:42:33 +0000448#elif defined(__BEOS__)
449 return _fseek(fp, offset, whence);
Guido van Rossumb8552162001-09-05 14:58:11 +0000450#elif SIZEOF_FPOS_T >= 8
Guido van Rossume54e0be2001-01-16 20:53:31 +0000451 /* lacking a 64-bit capable fseek(), use a 64-bit capable fsetpos()
452 and fgetpos() to implement fseek()*/
Trent Mickf29f47b2000-08-11 19:02:59 +0000453 fpos_t pos;
454 switch (whence) {
Guido van Rossume54e0be2001-01-16 20:53:31 +0000455 case SEEK_END:
Guido van Rossum8b4e43e2001-09-10 20:43:35 +0000456#ifdef MS_WINDOWS
457 fflush(fp);
458 if (_lseeki64(fileno(fp), 0, 2) == -1)
459 return -1;
460#else
Guido van Rossume54e0be2001-01-16 20:53:31 +0000461 if (fseek(fp, 0, SEEK_END) != 0)
462 return -1;
Guido van Rossum8b4e43e2001-09-10 20:43:35 +0000463#endif
Guido van Rossume54e0be2001-01-16 20:53:31 +0000464 /* fall through */
465 case SEEK_CUR:
466 if (fgetpos(fp, &pos) != 0)
467 return -1;
468 offset += pos;
469 break;
470 /* case SEEK_SET: break; */
Trent Mickf29f47b2000-08-11 19:02:59 +0000471 }
472 return fsetpos(fp, &offset);
473#else
Guido van Rossumb8552162001-09-05 14:58:11 +0000474#error "Large file support, but no way to fseek."
Trent Mickf29f47b2000-08-11 19:02:59 +0000475#endif
476}
477
478
479/* a portable ftell() function
480 Return -1 on failure with errno set appropriately, current file
481 position on success */
Guido van Rossumf68d8e52001-04-14 17:55:09 +0000482static Py_off_t
Fred Drake8ce159a2000-08-31 05:18:54 +0000483_portable_ftell(FILE* fp)
Trent Mickf29f47b2000-08-11 19:02:59 +0000484{
Guido van Rossumb8552162001-09-05 14:58:11 +0000485#if !defined(HAVE_LARGEFILE_SUPPORT)
486 return ftell(fp);
487#elif defined(HAVE_FTELLO) && SIZEOF_OFF_T >= 8
488 return ftello(fp);
489#elif defined(HAVE_FTELL64)
490 return ftell64(fp);
491#elif SIZEOF_FPOS_T >= 8
Trent Mickf29f47b2000-08-11 19:02:59 +0000492 fpos_t pos;
493 if (fgetpos(fp, &pos) != 0)
494 return -1;
495 return pos;
496#else
Guido van Rossumb8552162001-09-05 14:58:11 +0000497#error "Large file support, but no way to ftell."
Trent Mickf29f47b2000-08-11 19:02:59 +0000498#endif
499}
500
501
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000502static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000503file_seek(PyFileObject *f, PyObject *args)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000504{
Guido van Rossumd7297e61992-07-06 14:19:26 +0000505 int whence;
Guido van Rossumff4949e1992-08-05 19:58:53 +0000506 int ret;
Guido van Rossum4f53da02001-03-01 18:26:53 +0000507 Py_off_t offset;
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000508 PyObject *offobj;
Tim Peters86821b22001-01-07 21:19:34 +0000509
Guido van Rossumd7297e61992-07-06 14:19:26 +0000510 if (f->f_fp == NULL)
511 return err_closed();
Guido van Rossum7a6e9592002-08-06 15:55:28 +0000512 drop_readahead(f);
Guido van Rossumd7297e61992-07-06 14:19:26 +0000513 whence = 0;
Guido van Rossum43713e52000-02-29 13:59:29 +0000514 if (!PyArg_ParseTuple(args, "O|i:seek", &offobj, &whence))
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000515 return NULL;
516#if !defined(HAVE_LARGEFILE_SUPPORT)
517 offset = PyInt_AsLong(offobj);
518#else
519 offset = PyLong_Check(offobj) ?
520 PyLong_AsLongLong(offobj) : PyInt_AsLong(offobj);
521#endif
522 if (PyErr_Occurred())
Guido van Rossum88303191999-01-04 17:22:18 +0000523 return NULL;
Tim Peters86821b22001-01-07 21:19:34 +0000524
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000525 Py_BEGIN_ALLOW_THREADS
Guido van Rossumce5ba841991-03-06 13:06:18 +0000526 errno = 0;
Trent Mickf29f47b2000-08-11 19:02:59 +0000527 ret = _portable_fseek(f->f_fp, offset, whence);
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000528 Py_END_ALLOW_THREADS
Trent Mickf29f47b2000-08-11 19:02:59 +0000529
Guido van Rossumff4949e1992-08-05 19:58:53 +0000530 if (ret != 0) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000531 PyErr_SetFromErrno(PyExc_IOError);
Guido van Rossumfebd5511992-03-04 16:39:24 +0000532 clearerr(f->f_fp);
533 return NULL;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000534 }
Jack Jansen7b8c7542002-04-14 20:12:41 +0000535 f->f_skipnextlf = 0;
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000536 Py_INCREF(Py_None);
537 return Py_None;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000538}
539
Trent Mickf29f47b2000-08-11 19:02:59 +0000540
Guido van Rossumd7047b31995-01-02 19:07:15 +0000541#ifdef HAVE_FTRUNCATE
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000542static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000543file_truncate(PyFileObject *f, PyObject *args)
Guido van Rossumd7047b31995-01-02 19:07:15 +0000544{
Guido van Rossum4f53da02001-03-01 18:26:53 +0000545 Py_off_t newsize;
Tim Petersf1827cf2003-09-07 03:30:18 +0000546 PyObject *newsizeobj = NULL;
547 Py_off_t initialpos;
548 int ret;
Tim Peters86821b22001-01-07 21:19:34 +0000549
Guido van Rossumd7047b31995-01-02 19:07:15 +0000550 if (f->f_fp == NULL)
551 return err_closed();
Raymond Hettingerea3fdf42002-12-29 16:33:45 +0000552 if (!PyArg_UnpackTuple(args, "truncate", 0, 1, &newsizeobj))
Guido van Rossum88303191999-01-04 17:22:18 +0000553 return NULL;
Tim Petersfb05db22002-03-11 00:24:00 +0000554
Tim Petersf1827cf2003-09-07 03:30:18 +0000555 /* Get current file position. If the file happens to be open for
556 * update and the last operation was an input operation, C doesn't
557 * define what the later fflush() will do, but we promise truncate()
558 * won't change the current position (and fflush() *does* change it
559 * then at least on Windows). The easiest thing is to capture
560 * current pos now and seek back to it at the end.
561 */
562 Py_BEGIN_ALLOW_THREADS
563 errno = 0;
564 initialpos = _portable_ftell(f->f_fp);
565 Py_END_ALLOW_THREADS
566 if (initialpos == -1)
567 goto onioerror;
568
Tim Petersfb05db22002-03-11 00:24:00 +0000569 /* Set newsize to current postion if newsizeobj NULL, else to the
Tim Petersf1827cf2003-09-07 03:30:18 +0000570 * specified value.
571 */
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000572 if (newsizeobj != NULL) {
573#if !defined(HAVE_LARGEFILE_SUPPORT)
574 newsize = PyInt_AsLong(newsizeobj);
575#else
576 newsize = PyLong_Check(newsizeobj) ?
577 PyLong_AsLongLong(newsizeobj) :
578 PyInt_AsLong(newsizeobj);
579#endif
580 if (PyErr_Occurred())
581 return NULL;
Tim Petersfb05db22002-03-11 00:24:00 +0000582 }
Tim Petersf1827cf2003-09-07 03:30:18 +0000583 else /* default to current position */
584 newsize = initialpos;
Tim Petersfb05db22002-03-11 00:24:00 +0000585
Tim Petersf1827cf2003-09-07 03:30:18 +0000586 /* Flush the stream. We're mixing stream-level I/O with lower-level
587 * I/O, and a flush may be necessary to synch both platform views
588 * of the current file state.
589 */
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000590 Py_BEGIN_ALLOW_THREADS
Guido van Rossumd7047b31995-01-02 19:07:15 +0000591 errno = 0;
592 ret = fflush(f->f_fp);
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000593 Py_END_ALLOW_THREADS
Tim Petersfb05db22002-03-11 00:24:00 +0000594 if (ret != 0)
595 goto onioerror;
Trent Mickf29f47b2000-08-11 19:02:59 +0000596
Martin v. Löwis6238d2b2002-06-30 15:26:10 +0000597#ifdef MS_WINDOWS
Tim Petersfb05db22002-03-11 00:24:00 +0000598 /* MS _chsize doesn't work if newsize doesn't fit in 32 bits,
Tim Peters8f01b682002-03-12 03:04:44 +0000599 so don't even try using it. */
Tim Petersfb05db22002-03-11 00:24:00 +0000600 {
Tim Petersfb05db22002-03-11 00:24:00 +0000601 HANDLE hFile;
Tim Petersfb05db22002-03-11 00:24:00 +0000602
Tim Petersf1827cf2003-09-07 03:30:18 +0000603 /* Have to move current pos to desired endpoint on Windows. */
604 Py_BEGIN_ALLOW_THREADS
605 errno = 0;
606 ret = _portable_fseek(f->f_fp, newsize, SEEK_SET) != 0;
607 Py_END_ALLOW_THREADS
608 if (ret)
609 goto onioerror;
Tim Petersfb05db22002-03-11 00:24:00 +0000610
Tim Peters8f01b682002-03-12 03:04:44 +0000611 /* Truncate. Note that this may grow the file! */
612 Py_BEGIN_ALLOW_THREADS
613 errno = 0;
614 hFile = (HANDLE)_get_osfhandle(fileno(f->f_fp));
Tim Petersf1827cf2003-09-07 03:30:18 +0000615 ret = hFile == (HANDLE)-1;
616 if (ret == 0) {
617 ret = SetEndOfFile(hFile) == 0;
618 if (ret)
Tim Peters8f01b682002-03-12 03:04:44 +0000619 errno = EACCES;
620 }
621 Py_END_ALLOW_THREADS
Tim Petersf1827cf2003-09-07 03:30:18 +0000622 if (ret)
Tim Peters8f01b682002-03-12 03:04:44 +0000623 goto onioerror;
Guido van Rossumd7047b31995-01-02 19:07:15 +0000624 }
Trent Mickf29f47b2000-08-11 19:02:59 +0000625#else
626 Py_BEGIN_ALLOW_THREADS
627 errno = 0;
628 ret = ftruncate(fileno(f->f_fp), newsize);
629 Py_END_ALLOW_THREADS
Tim Petersf1827cf2003-09-07 03:30:18 +0000630 if (ret != 0)
631 goto onioerror;
Martin v. Löwis6238d2b2002-06-30 15:26:10 +0000632#endif /* !MS_WINDOWS */
Tim Peters86821b22001-01-07 21:19:34 +0000633
Tim Petersf1827cf2003-09-07 03:30:18 +0000634 /* Restore original file position. */
635 Py_BEGIN_ALLOW_THREADS
636 errno = 0;
637 ret = _portable_fseek(f->f_fp, initialpos, SEEK_SET) != 0;
638 Py_END_ALLOW_THREADS
639 if (ret)
640 goto onioerror;
641
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000642 Py_INCREF(Py_None);
643 return Py_None;
Trent Mickf29f47b2000-08-11 19:02:59 +0000644
645onioerror:
646 PyErr_SetFromErrno(PyExc_IOError);
647 clearerr(f->f_fp);
648 return NULL;
Guido van Rossumd7047b31995-01-02 19:07:15 +0000649}
650#endif /* HAVE_FTRUNCATE */
651
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000652static PyObject *
Martin v. Löwise3eb1f22001-08-16 13:15:00 +0000653file_tell(PyFileObject *f)
Guido van Rossumce5ba841991-03-06 13:06:18 +0000654{
Guido van Rossum4f53da02001-03-01 18:26:53 +0000655 Py_off_t pos;
Trent Mickf29f47b2000-08-11 19:02:59 +0000656
Guido van Rossumd7297e61992-07-06 14:19:26 +0000657 if (f->f_fp == NULL)
658 return err_closed();
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000659 Py_BEGIN_ALLOW_THREADS
Guido van Rossumce5ba841991-03-06 13:06:18 +0000660 errno = 0;
Trent Mickf29f47b2000-08-11 19:02:59 +0000661 pos = _portable_ftell(f->f_fp);
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000662 Py_END_ALLOW_THREADS
Trent Mickf29f47b2000-08-11 19:02:59 +0000663 if (pos == -1) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000664 PyErr_SetFromErrno(PyExc_IOError);
Guido van Rossumfebd5511992-03-04 16:39:24 +0000665 clearerr(f->f_fp);
666 return NULL;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000667 }
Jack Jansen7b8c7542002-04-14 20:12:41 +0000668 if (f->f_skipnextlf) {
669 int c;
670 c = GETC(f->f_fp);
671 if (c == '\n') {
672 pos++;
673 f->f_skipnextlf = 0;
674 } else if (c != EOF) ungetc(c, f->f_fp);
675 }
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000676#if !defined(HAVE_LARGEFILE_SUPPORT)
Trent Mickf29f47b2000-08-11 19:02:59 +0000677 return PyInt_FromLong(pos);
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000678#else
Trent Mickf29f47b2000-08-11 19:02:59 +0000679 return PyLong_FromLongLong(pos);
Guido van Rossum3c9fe0c1999-01-06 18:51:17 +0000680#endif
Guido van Rossumce5ba841991-03-06 13:06:18 +0000681}
682
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000683static PyObject *
Martin v. Löwise3eb1f22001-08-16 13:15:00 +0000684file_fileno(PyFileObject *f)
Guido van Rossumed233a51992-06-23 09:07:03 +0000685{
Guido van Rossumd7297e61992-07-06 14:19:26 +0000686 if (f->f_fp == NULL)
687 return err_closed();
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000688 return PyInt_FromLong((long) fileno(f->f_fp));
Guido van Rossumed233a51992-06-23 09:07:03 +0000689}
690
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000691static PyObject *
Martin v. Löwise3eb1f22001-08-16 13:15:00 +0000692file_flush(PyFileObject *f)
Guido van Rossumce5ba841991-03-06 13:06:18 +0000693{
Guido van Rossumff4949e1992-08-05 19:58:53 +0000694 int res;
Tim Peters86821b22001-01-07 21:19:34 +0000695
Guido van Rossumd7297e61992-07-06 14:19:26 +0000696 if (f->f_fp == NULL)
697 return err_closed();
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000698 Py_BEGIN_ALLOW_THREADS
Guido van Rossumce5ba841991-03-06 13:06:18 +0000699 errno = 0;
Guido van Rossumff4949e1992-08-05 19:58:53 +0000700 res = fflush(f->f_fp);
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000701 Py_END_ALLOW_THREADS
Guido van Rossumff4949e1992-08-05 19:58:53 +0000702 if (res != 0) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000703 PyErr_SetFromErrno(PyExc_IOError);
Guido van Rossumfebd5511992-03-04 16:39:24 +0000704 clearerr(f->f_fp);
705 return NULL;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000706 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000707 Py_INCREF(Py_None);
708 return Py_None;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000709}
710
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000711static PyObject *
Martin v. Löwise3eb1f22001-08-16 13:15:00 +0000712file_isatty(PyFileObject *f)
Guido van Rossuma1ab7fa1991-06-04 19:37:39 +0000713{
Guido van Rossumff4949e1992-08-05 19:58:53 +0000714 long res;
Guido van Rossumd7297e61992-07-06 14:19:26 +0000715 if (f->f_fp == NULL)
716 return err_closed();
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000717 Py_BEGIN_ALLOW_THREADS
Guido van Rossumff4949e1992-08-05 19:58:53 +0000718 res = isatty((int)fileno(f->f_fp));
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000719 Py_END_ALLOW_THREADS
Guido van Rossum7f7666f2002-04-07 06:28:00 +0000720 return PyBool_FromLong(res);
Guido van Rossuma1ab7fa1991-06-04 19:37:39 +0000721}
722
Guido van Rossumff7e83d1999-08-27 20:39:37 +0000723
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000724#if BUFSIZ < 8192
725#define SMALLCHUNK 8192
726#else
727#define SMALLCHUNK BUFSIZ
728#endif
729
Guido van Rossum3c259041999-01-14 19:00:14 +0000730#if SIZEOF_INT < 4
731#define BIGCHUNK (512 * 32)
732#else
733#define BIGCHUNK (512 * 1024)
734#endif
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000735
736static size_t
Fred Drakefd99de62000-07-09 05:02:18 +0000737new_buffersize(PyFileObject *f, size_t currentsize)
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000738{
739#ifdef HAVE_FSTAT
Fred Drake1bc8fab2001-07-19 21:49:38 +0000740 off_t pos, end;
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000741 struct stat st;
742 if (fstat(fileno(f->f_fp), &st) == 0) {
743 end = st.st_size;
Guido van Rossumcada2931998-12-11 20:44:56 +0000744 /* The following is not a bug: we really need to call lseek()
745 *and* ftell(). The reason is that some stdio libraries
746 mistakenly flush their buffer when ftell() is called and
747 the lseek() call it makes fails, thereby throwing away
748 data that cannot be recovered in any way. To avoid this,
749 we first test lseek(), and only call ftell() if lseek()
750 works. We can't use the lseek() value either, because we
751 need to take the amount of buffered data into account.
752 (Yet another reason why stdio stinks. :-) */
Guido van Rossum91aaa921998-05-05 22:21:35 +0000753 pos = lseek(fileno(f->f_fp), 0L, SEEK_CUR);
Jack Jansen2771b5b2001-10-10 22:03:27 +0000754 if (pos >= 0) {
Guido van Rossum91aaa921998-05-05 22:21:35 +0000755 pos = ftell(f->f_fp);
Jack Jansen2771b5b2001-10-10 22:03:27 +0000756 }
Guido van Rossumd30dc0a1998-04-27 19:01:08 +0000757 if (pos < 0)
758 clearerr(f->f_fp);
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000759 if (end > pos && pos >= 0)
Guido van Rossumcada2931998-12-11 20:44:56 +0000760 return currentsize + end - pos + 1;
Guido van Rossumdcb5e7f1998-03-03 22:36:10 +0000761 /* Add 1 so if the file were to grow we'd notice. */
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000762 }
763#endif
764 if (currentsize > SMALLCHUNK) {
765 /* Keep doubling until we reach BIGCHUNK;
766 then keep adding BIGCHUNK. */
767 if (currentsize <= BIGCHUNK)
768 return currentsize + currentsize;
769 else
770 return currentsize + BIGCHUNK;
771 }
772 return currentsize + SMALLCHUNK;
773}
774
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +0000775#if defined(EWOULDBLOCK) && defined(EAGAIN) && EWOULDBLOCK != EAGAIN
776#define BLOCKED_ERRNO(x) ((x) == EWOULDBLOCK || (x) == EAGAIN)
777#else
778#ifdef EWOULDBLOCK
779#define BLOCKED_ERRNO(x) ((x) == EWOULDBLOCK)
780#else
781#ifdef EAGAIN
782#define BLOCKED_ERRNO(x) ((x) == EAGAIN)
783#else
784#define BLOCKED_ERRNO(x) 0
785#endif
786#endif
787#endif
788
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000789static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000790file_read(PyFileObject *f, PyObject *args)
Guido van Rossumce5ba841991-03-06 13:06:18 +0000791{
Guido van Rossum789a1611997-05-10 22:33:55 +0000792 long bytesrequested = -1;
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000793 size_t bytesread, buffersize, chunksize;
Guido van Rossumc0b618a1997-05-02 03:12:38 +0000794 PyObject *v;
Tim Peters86821b22001-01-07 21:19:34 +0000795
Guido van Rossumd7297e61992-07-06 14:19:26 +0000796 if (f->f_fp == NULL)
797 return err_closed();
Guido van Rossum43713e52000-02-29 13:59:29 +0000798 if (!PyArg_ParseTuple(args, "|l:read", &bytesrequested))
Guido van Rossum789a1611997-05-10 22:33:55 +0000799 return NULL;
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000800 if (bytesrequested < 0)
Guido van Rossumff1ccbf1999-04-10 15:48:23 +0000801 buffersize = new_buffersize(f, (size_t)0);
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000802 else
803 buffersize = bytesrequested;
Trent Mickf29f47b2000-08-11 19:02:59 +0000804 if (buffersize > INT_MAX) {
805 PyErr_SetString(PyExc_OverflowError,
Jeremy Hylton8b735422002-08-14 21:01:41 +0000806 "requested number of bytes is more than a Python string can hold");
Trent Mickf29f47b2000-08-11 19:02:59 +0000807 return NULL;
808 }
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000809 v = PyString_FromStringAndSize((char *)NULL, buffersize);
Guido van Rossum3f5da241990-12-20 15:06:42 +0000810 if (v == NULL)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000811 return NULL;
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000812 bytesread = 0;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000813 for (;;) {
Guido van Rossum6263d541997-05-10 22:07:25 +0000814 Py_BEGIN_ALLOW_THREADS
815 errno = 0;
Jack Jansen7b8c7542002-04-14 20:12:41 +0000816 chunksize = Py_UniversalNewlineFread(BUF(v) + bytesread,
Jeremy Hylton8b735422002-08-14 21:01:41 +0000817 buffersize - bytesread, f->f_fp, (PyObject *)f);
Guido van Rossum6263d541997-05-10 22:07:25 +0000818 Py_END_ALLOW_THREADS
819 if (chunksize == 0) {
820 if (!ferror(f->f_fp))
821 break;
Guido van Rossum6263d541997-05-10 22:07:25 +0000822 clearerr(f->f_fp);
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +0000823 /* When in non-blocking mode, data shouldn't
824 * be discarded if a blocking signal was
825 * received. That will also happen if
826 * chunksize != 0, but bytesread < buffersize. */
827 if (bytesread > 0 && BLOCKED_ERRNO(errno))
828 break;
829 PyErr_SetFromErrno(PyExc_IOError);
Guido van Rossum6263d541997-05-10 22:07:25 +0000830 Py_DECREF(v);
831 return NULL;
832 }
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000833 bytesread += chunksize;
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +0000834 if (bytesread < buffersize) {
835 clearerr(f->f_fp);
Guido van Rossumce5ba841991-03-06 13:06:18 +0000836 break;
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +0000837 }
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000838 if (bytesrequested < 0) {
Guido van Rossumcada2931998-12-11 20:44:56 +0000839 buffersize = new_buffersize(f, buffersize);
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000840 if (_PyString_Resize(&v, buffersize) < 0)
Guido van Rossumce5ba841991-03-06 13:06:18 +0000841 return NULL;
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +0000842 } else {
Gustavo Niemeyera080be82002-12-17 17:48:00 +0000843 /* Got what was requested. */
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +0000844 break;
Guido van Rossumce5ba841991-03-06 13:06:18 +0000845 }
846 }
Guido van Rossum5449b6e1997-05-09 22:27:31 +0000847 if (bytesread != buffersize)
848 _PyString_Resize(&v, bytesread);
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000849 return v;
850}
851
Guido van Rossumfdf95dd1997-05-05 22:15:02 +0000852static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +0000853file_readinto(PyFileObject *f, PyObject *args)
Guido van Rossumfdf95dd1997-05-05 22:15:02 +0000854{
855 char *ptr;
Guido van Rossum00ebd462001-10-23 21:25:24 +0000856 int ntodo;
857 size_t ndone, nnow;
Tim Peters86821b22001-01-07 21:19:34 +0000858
Guido van Rossumfdf95dd1997-05-05 22:15:02 +0000859 if (f->f_fp == NULL)
860 return err_closed();
Neal Norwitz62f5a9d2002-04-01 00:09:00 +0000861 if (!PyArg_ParseTuple(args, "w#", &ptr, &ntodo))
Guido van Rossumfdf95dd1997-05-05 22:15:02 +0000862 return NULL;
863 ndone = 0;
Guido van Rossum6263d541997-05-10 22:07:25 +0000864 while (ntodo > 0) {
865 Py_BEGIN_ALLOW_THREADS
866 errno = 0;
Tim Petersf1827cf2003-09-07 03:30:18 +0000867 nnow = Py_UniversalNewlineFread(ptr+ndone, ntodo, f->f_fp,
Jeremy Hylton8b735422002-08-14 21:01:41 +0000868 (PyObject *)f);
Guido van Rossum6263d541997-05-10 22:07:25 +0000869 Py_END_ALLOW_THREADS
870 if (nnow == 0) {
871 if (!ferror(f->f_fp))
872 break;
Guido van Rossumfdf95dd1997-05-05 22:15:02 +0000873 PyErr_SetFromErrno(PyExc_IOError);
874 clearerr(f->f_fp);
875 return NULL;
876 }
Guido van Rossumfdf95dd1997-05-05 22:15:02 +0000877 ndone += nnow;
878 ntodo -= nnow;
879 }
Trent Mickf29f47b2000-08-11 19:02:59 +0000880 return PyInt_FromLong((long)ndone);
Guido van Rossumfdf95dd1997-05-05 22:15:02 +0000881}
882
Tim Peters86821b22001-01-07 21:19:34 +0000883/**************************************************************************
Tim Petersf29b64d2001-01-15 06:33:19 +0000884Routine to get next line using platform fgets().
Tim Peters86821b22001-01-07 21:19:34 +0000885
886Under MSVC 6:
887
Tim Peters1c733232001-01-08 04:02:07 +0000888+ MS threadsafe getc is very slow (multiple layers of function calls before+
889 after each character, to lock+unlock the stream).
890+ The stream-locking functions are MS-internal -- can't access them from user
891 code.
892+ There's nothing Tim could find in the MS C or platform SDK libraries that
893 can worm around this.
Tim Peters86821b22001-01-07 21:19:34 +0000894+ MS fgets locks/unlocks only once per line; it's the only hook we have.
895
896So we use fgets for speed(!), despite that it's painful.
897
898MS realloc is also slow.
899
Tim Petersf29b64d2001-01-15 06:33:19 +0000900Reports from other platforms on this method vs getc_unlocked (which MS doesn't
901have):
902 Linux a wash
903 Solaris a wash
904 Tru64 Unix getline_via_fgets significantly faster
Tim Peters86821b22001-01-07 21:19:34 +0000905
Tim Petersf29b64d2001-01-15 06:33:19 +0000906CAUTION: The C std isn't clear about this: in those cases where fgets
907writes something into the buffer, can it write into any position beyond the
908required trailing null byte? MSVC 6 fgets does not, and no platform is (yet)
909known on which it does; and it would be a strange way to code fgets. Still,
910getline_via_fgets may not work correctly if it does. The std test
911test_bufio.py should fail if platform fgets() routinely writes beyond the
912trailing null byte. #define DONT_USE_FGETS_IN_GETLINE to disable this code.
Tim Peters86821b22001-01-07 21:19:34 +0000913**************************************************************************/
914
Tim Petersf29b64d2001-01-15 06:33:19 +0000915/* Use this routine if told to, or by default on non-get_unlocked()
916 * platforms unless told not to. Yikes! Let's spell that out:
917 * On a platform with getc_unlocked():
918 * By default, use getc_unlocked().
919 * If you want to use fgets() instead, #define USE_FGETS_IN_GETLINE.
920 * On a platform without getc_unlocked():
921 * By default, use fgets().
922 * If you don't want to use fgets(), #define DONT_USE_FGETS_IN_GETLINE.
923 */
924#if !defined(USE_FGETS_IN_GETLINE) && !defined(HAVE_GETC_UNLOCKED)
925#define USE_FGETS_IN_GETLINE
Tim Peters86821b22001-01-07 21:19:34 +0000926#endif
927
Tim Petersf29b64d2001-01-15 06:33:19 +0000928#if defined(DONT_USE_FGETS_IN_GETLINE) && defined(USE_FGETS_IN_GETLINE)
929#undef USE_FGETS_IN_GETLINE
930#endif
931
932#ifdef USE_FGETS_IN_GETLINE
Tim Peters86821b22001-01-07 21:19:34 +0000933static PyObject*
Tim Petersf29b64d2001-01-15 06:33:19 +0000934getline_via_fgets(FILE *fp)
Tim Peters86821b22001-01-07 21:19:34 +0000935{
Tim Peters15b83852001-01-08 00:53:12 +0000936/* INITBUFSIZE is the maximum line length that lets us get away with the fast
Tim Peters142297a2001-01-15 10:36:56 +0000937 * no-realloc, one-fgets()-call path. Boosting it isn't free, because we have
938 * to fill this much of the buffer with a known value in order to figure out
939 * how much of the buffer fgets() overwrites. So if INITBUFSIZE is larger
940 * than "most" lines, we waste time filling unused buffer slots. 100 is
941 * surely adequate for most peoples' email archives, chewing over source code,
942 * etc -- "regular old text files".
943 * MAXBUFSIZE is the maximum line length that lets us get away with the less
944 * fast (but still zippy) no-realloc, two-fgets()-call path. See above for
945 * cautions about boosting that. 300 was chosen because the worst real-life
946 * text-crunching job reported on Python-Dev was a mail-log crawler where over
947 * half the lines were 254 chars.
Tim Peters15b83852001-01-08 00:53:12 +0000948 */
Tim Peters142297a2001-01-15 10:36:56 +0000949#define INITBUFSIZE 100
950#define MAXBUFSIZE 300
Tim Peters142297a2001-01-15 10:36:56 +0000951 char* p; /* temp */
952 char buf[MAXBUFSIZE];
Tim Peters86821b22001-01-07 21:19:34 +0000953 PyObject* v; /* the string object result */
Tim Peters86821b22001-01-07 21:19:34 +0000954 char* pvfree; /* address of next free slot */
955 char* pvend; /* address one beyond last free slot */
Tim Peters142297a2001-01-15 10:36:56 +0000956 size_t nfree; /* # of free buffer slots; pvend-pvfree */
957 size_t total_v_size; /* total # of slots in buffer */
Tim Petersddea2082002-03-23 10:03:50 +0000958 size_t increment; /* amount to increment the buffer */
Tim Peters86821b22001-01-07 21:19:34 +0000959
Tim Peters15b83852001-01-08 00:53:12 +0000960 /* Optimize for normal case: avoid _PyString_Resize if at all
Tim Peters142297a2001-01-15 10:36:56 +0000961 * possible via first reading into stack buffer "buf".
Tim Peters15b83852001-01-08 00:53:12 +0000962 */
Tim Peters142297a2001-01-15 10:36:56 +0000963 total_v_size = INITBUFSIZE; /* start small and pray */
964 pvfree = buf;
965 for (;;) {
966 Py_BEGIN_ALLOW_THREADS
967 pvend = buf + total_v_size;
968 nfree = pvend - pvfree;
969 memset(pvfree, '\n', nfree);
970 p = fgets(pvfree, nfree, fp);
971 Py_END_ALLOW_THREADS
Tim Peters15b83852001-01-08 00:53:12 +0000972
Tim Peters142297a2001-01-15 10:36:56 +0000973 if (p == NULL) {
974 clearerr(fp);
975 if (PyErr_CheckSignals())
976 return NULL;
977 v = PyString_FromStringAndSize(buf, pvfree - buf);
Tim Peters86821b22001-01-07 21:19:34 +0000978 return v;
979 }
Tim Peters142297a2001-01-15 10:36:56 +0000980 /* fgets read *something* */
981 p = memchr(pvfree, '\n', nfree);
982 if (p != NULL) {
983 /* Did the \n come from fgets or from us?
984 * Since fgets stops at the first \n, and then writes
985 * \0, if it's from fgets a \0 must be next. But if
986 * that's so, it could not have come from us, since
987 * the \n's we filled the buffer with have only more
988 * \n's to the right.
989 */
990 if (p+1 < pvend && *(p+1) == '\0') {
991 /* It's from fgets: we win! In particular,
992 * we haven't done any mallocs yet, and can
993 * build the final result on the first try.
994 */
995 ++p; /* include \n from fgets */
996 }
997 else {
998 /* Must be from us: fgets didn't fill the
999 * buffer and didn't find a newline, so it
1000 * must be the last and newline-free line of
1001 * the file.
1002 */
1003 assert(p > pvfree && *(p-1) == '\0');
1004 --p; /* don't include \0 from fgets */
1005 }
1006 v = PyString_FromStringAndSize(buf, p - buf);
1007 return v;
1008 }
1009 /* yuck: fgets overwrote all the newlines, i.e. the entire
1010 * buffer. So this line isn't over yet, or maybe it is but
1011 * we're exactly at EOF. If we haven't already, try using the
1012 * rest of the stack buffer.
Tim Peters86821b22001-01-07 21:19:34 +00001013 */
Tim Peters142297a2001-01-15 10:36:56 +00001014 assert(*(pvend-1) == '\0');
1015 if (pvfree == buf) {
1016 pvfree = pvend - 1; /* overwrite trailing null */
1017 total_v_size = MAXBUFSIZE;
1018 }
1019 else
1020 break;
Tim Peters86821b22001-01-07 21:19:34 +00001021 }
Tim Peters142297a2001-01-15 10:36:56 +00001022
1023 /* The stack buffer isn't big enough; malloc a string object and read
1024 * into its buffer.
Tim Peters15b83852001-01-08 00:53:12 +00001025 */
Tim Petersddea2082002-03-23 10:03:50 +00001026 total_v_size = MAXBUFSIZE << 1;
Tim Peters1c733232001-01-08 04:02:07 +00001027 v = PyString_FromStringAndSize((char*)NULL, (int)total_v_size);
Tim Peters15b83852001-01-08 00:53:12 +00001028 if (v == NULL)
1029 return v;
1030 /* copy over everything except the last null byte */
Tim Peters142297a2001-01-15 10:36:56 +00001031 memcpy(BUF(v), buf, MAXBUFSIZE-1);
1032 pvfree = BUF(v) + MAXBUFSIZE - 1;
Tim Peters86821b22001-01-07 21:19:34 +00001033
1034 /* Keep reading stuff into v; if it ever ends successfully, break
Tim Peters15b83852001-01-08 00:53:12 +00001035 * after setting p one beyond the end of the line. The code here is
1036 * very much like the code above, except reads into v's buffer; see
1037 * the code above for detailed comments about the logic.
Tim Peters86821b22001-01-07 21:19:34 +00001038 */
1039 for (;;) {
Tim Peters86821b22001-01-07 21:19:34 +00001040 Py_BEGIN_ALLOW_THREADS
1041 pvend = BUF(v) + total_v_size;
1042 nfree = pvend - pvfree;
1043 memset(pvfree, '\n', nfree);
1044 p = fgets(pvfree, nfree, fp);
1045 Py_END_ALLOW_THREADS
1046
1047 if (p == NULL) {
1048 clearerr(fp);
1049 if (PyErr_CheckSignals()) {
1050 Py_DECREF(v);
1051 return NULL;
1052 }
1053 p = pvfree;
1054 break;
1055 }
Tim Peters86821b22001-01-07 21:19:34 +00001056 p = memchr(pvfree, '\n', nfree);
1057 if (p != NULL) {
1058 if (p+1 < pvend && *(p+1) == '\0') {
1059 /* \n came from fgets */
1060 ++p;
1061 break;
1062 }
1063 /* \n came from us; last line of file, no newline */
1064 assert(p > pvfree && *(p-1) == '\0');
1065 --p;
1066 break;
1067 }
1068 /* expand buffer and try again */
1069 assert(*(pvend-1) == '\0');
Tim Petersddea2082002-03-23 10:03:50 +00001070 increment = total_v_size >> 2; /* mild exponential growth */
1071 total_v_size += increment;
Tim Peters86821b22001-01-07 21:19:34 +00001072 if (total_v_size > INT_MAX) {
1073 PyErr_SetString(PyExc_OverflowError,
1074 "line is longer than a Python string can hold");
1075 Py_DECREF(v);
1076 return NULL;
1077 }
1078 if (_PyString_Resize(&v, (int)total_v_size) < 0)
1079 return NULL;
1080 /* overwrite the trailing null byte */
Tim Petersddea2082002-03-23 10:03:50 +00001081 pvfree = BUF(v) + (total_v_size - increment - 1);
Tim Peters86821b22001-01-07 21:19:34 +00001082 }
1083 if (BUF(v) + total_v_size != p)
1084 _PyString_Resize(&v, p - BUF(v));
1085 return v;
1086#undef INITBUFSIZE
Tim Peters142297a2001-01-15 10:36:56 +00001087#undef MAXBUFSIZE
Tim Peters86821b22001-01-07 21:19:34 +00001088}
Tim Petersf29b64d2001-01-15 06:33:19 +00001089#endif /* ifdef USE_FGETS_IN_GETLINE */
Guido van Rossumfdf95dd1997-05-05 22:15:02 +00001090
Guido van Rossum0bd24411991-04-04 15:21:57 +00001091/* Internal routine to get a line.
1092 Size argument interpretation:
1093 > 0: max length;
Guido van Rossum86282062001-01-08 01:26:47 +00001094 <= 0: read arbitrary line
Guido van Rossumce5ba841991-03-06 13:06:18 +00001095*/
1096
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001097static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001098get_line(PyFileObject *f, int n)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001099{
Guido van Rossum1187aa42001-01-05 14:43:05 +00001100 FILE *fp = f->f_fp;
1101 int c;
Andrew M. Kuchling4b2b4452000-11-29 02:53:22 +00001102 char *buf, *end;
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001103 size_t total_v_size; /* total # of slots in buffer */
1104 size_t used_v_size; /* # used slots in buffer */
1105 size_t increment; /* amount to increment the buffer */
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001106 PyObject *v;
Jack Jansen7b8c7542002-04-14 20:12:41 +00001107 int newlinetypes = f->f_newlinetypes;
1108 int skipnextlf = f->f_skipnextlf;
1109 int univ_newline = f->f_univ_newline;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001110
Jack Jansen7b8c7542002-04-14 20:12:41 +00001111#if defined(USE_FGETS_IN_GETLINE)
Jack Jansen7b8c7542002-04-14 20:12:41 +00001112 if (n <= 0 && !univ_newline )
Tim Petersf29b64d2001-01-15 06:33:19 +00001113 return getline_via_fgets(fp);
Tim Peters86821b22001-01-07 21:19:34 +00001114#endif
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001115 total_v_size = n > 0 ? n : 100;
1116 v = PyString_FromStringAndSize((char *)NULL, total_v_size);
Guido van Rossum3f5da241990-12-20 15:06:42 +00001117 if (v == NULL)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001118 return NULL;
Guido van Rossumce5ba841991-03-06 13:06:18 +00001119 buf = BUF(v);
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001120 end = buf + total_v_size;
Guido van Rossum1984f1e1992-08-04 12:41:02 +00001121
Guido van Rossumce5ba841991-03-06 13:06:18 +00001122 for (;;) {
Guido van Rossum1187aa42001-01-05 14:43:05 +00001123 Py_BEGIN_ALLOW_THREADS
1124 FLOCKFILE(fp);
Jack Jansen7b8c7542002-04-14 20:12:41 +00001125 if (univ_newline) {
1126 c = 'x'; /* Shut up gcc warning */
1127 while ( buf != end && (c = GETC(fp)) != EOF ) {
1128 if (skipnextlf ) {
1129 skipnextlf = 0;
1130 if (c == '\n') {
Tim Petersf1827cf2003-09-07 03:30:18 +00001131 /* Seeing a \n here with
1132 * skipnextlf true means we
Jeremy Hylton8b735422002-08-14 21:01:41 +00001133 * saw a \r before.
1134 */
Jack Jansen7b8c7542002-04-14 20:12:41 +00001135 newlinetypes |= NEWLINE_CRLF;
1136 c = GETC(fp);
1137 if (c == EOF) break;
1138 } else {
1139 newlinetypes |= NEWLINE_CR;
1140 }
1141 }
1142 if (c == '\r') {
1143 skipnextlf = 1;
1144 c = '\n';
1145 } else if ( c == '\n')
1146 newlinetypes |= NEWLINE_LF;
1147 *buf++ = c;
1148 if (c == '\n') break;
1149 }
1150 if ( c == EOF && skipnextlf )
1151 newlinetypes |= NEWLINE_CR;
1152 } else /* If not universal newlines use the normal loop */
Guido van Rossum1187aa42001-01-05 14:43:05 +00001153 while ((c = GETC(fp)) != EOF &&
1154 (*buf++ = c) != '\n' &&
1155 buf != end)
1156 ;
1157 FUNLOCKFILE(fp);
1158 Py_END_ALLOW_THREADS
Jack Jansen7b8c7542002-04-14 20:12:41 +00001159 f->f_newlinetypes = newlinetypes;
1160 f->f_skipnextlf = skipnextlf;
Guido van Rossum1187aa42001-01-05 14:43:05 +00001161 if (c == '\n')
1162 break;
1163 if (c == EOF) {
Guido van Rossum29206bc2001-08-09 18:14:59 +00001164 if (ferror(fp)) {
1165 PyErr_SetFromErrno(PyExc_IOError);
1166 clearerr(fp);
1167 Py_DECREF(v);
1168 return NULL;
1169 }
Guido van Rossum76ad8ed1991-06-03 10:54:55 +00001170 clearerr(fp);
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001171 if (PyErr_CheckSignals()) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001172 Py_DECREF(v);
Guido van Rossum0bd24411991-04-04 15:21:57 +00001173 return NULL;
1174 }
Guido van Rossumce5ba841991-03-06 13:06:18 +00001175 break;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001176 }
Guido van Rossum1187aa42001-01-05 14:43:05 +00001177 /* Must be because buf == end */
1178 if (n > 0)
Guido van Rossum0bd24411991-04-04 15:21:57 +00001179 break;
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001180 used_v_size = total_v_size;
1181 increment = total_v_size >> 2; /* mild exponential growth */
1182 total_v_size += increment;
1183 if (total_v_size > INT_MAX) {
Guido van Rossum1187aa42001-01-05 14:43:05 +00001184 PyErr_SetString(PyExc_OverflowError,
1185 "line is longer than a Python string can hold");
Tim Peters86821b22001-01-07 21:19:34 +00001186 Py_DECREF(v);
Guido van Rossum1187aa42001-01-05 14:43:05 +00001187 return NULL;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001188 }
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001189 if (_PyString_Resize(&v, total_v_size) < 0)
Guido van Rossum1187aa42001-01-05 14:43:05 +00001190 return NULL;
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001191 buf = BUF(v) + used_v_size;
1192 end = BUF(v) + total_v_size;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001193 }
Guido van Rossum1984f1e1992-08-04 12:41:02 +00001194
Neil Schemenauer3a204a72002-03-23 19:41:34 +00001195 used_v_size = buf - BUF(v);
1196 if (used_v_size != total_v_size)
1197 _PyString_Resize(&v, used_v_size);
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001198 return v;
1199}
1200
Guido van Rossum0bd24411991-04-04 15:21:57 +00001201/* External C interface */
1202
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001203PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001204PyFile_GetLine(PyObject *f, int n)
Guido van Rossum0bd24411991-04-04 15:21:57 +00001205{
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001206 PyObject *result;
1207
Guido van Rossum3165fe61992-09-25 21:59:05 +00001208 if (f == NULL) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001209 PyErr_BadInternalCall();
Guido van Rossum0bd24411991-04-04 15:21:57 +00001210 return NULL;
1211 }
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001212
1213 if (PyFile_Check(f)) {
1214 if (((PyFileObject*)f)->f_fp == NULL)
1215 return err_closed();
1216 result = get_line((PyFileObject *)f, n);
1217 }
1218 else {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001219 PyObject *reader;
1220 PyObject *args;
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001221
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001222 reader = PyObject_GetAttrString(f, "readline");
Guido van Rossum3165fe61992-09-25 21:59:05 +00001223 if (reader == NULL)
1224 return NULL;
1225 if (n <= 0)
Raymond Hettinger8ae46892003-10-12 19:09:37 +00001226 args = PyTuple_New(0);
Guido van Rossum3165fe61992-09-25 21:59:05 +00001227 else
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001228 args = Py_BuildValue("(i)", n);
Guido van Rossum3165fe61992-09-25 21:59:05 +00001229 if (args == NULL) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001230 Py_DECREF(reader);
Guido van Rossum3165fe61992-09-25 21:59:05 +00001231 return NULL;
1232 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001233 result = PyEval_CallObject(reader, args);
1234 Py_DECREF(reader);
1235 Py_DECREF(args);
Martin v. Löwisaf6a27a2003-01-03 19:16:14 +00001236 if (result != NULL && !PyString_Check(result) &&
1237 !PyUnicode_Check(result)) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001238 Py_DECREF(result);
Guido van Rossum3165fe61992-09-25 21:59:05 +00001239 result = NULL;
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001240 PyErr_SetString(PyExc_TypeError,
Guido van Rossum3165fe61992-09-25 21:59:05 +00001241 "object.readline() returned non-string");
1242 }
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001243 }
1244
1245 if (n < 0 && result != NULL && PyString_Check(result)) {
1246 char *s = PyString_AS_STRING(result);
1247 int len = PyString_GET_SIZE(result);
1248 if (len == 0) {
1249 Py_DECREF(result);
1250 result = NULL;
1251 PyErr_SetString(PyExc_EOFError,
1252 "EOF when reading a line");
1253 }
1254 else if (s[len-1] == '\n') {
1255 if (result->ob_refcnt == 1)
1256 _PyString_Resize(&result, len-1);
1257 else {
1258 PyObject *v;
1259 v = PyString_FromStringAndSize(s, len-1);
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001260 Py_DECREF(result);
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001261 result = v;
Guido van Rossum3165fe61992-09-25 21:59:05 +00001262 }
1263 }
Guido van Rossum3165fe61992-09-25 21:59:05 +00001264 }
Martin v. Löwisaf6a27a2003-01-03 19:16:14 +00001265#ifdef Py_USING_UNICODE
1266 if (n < 0 && result != NULL && PyUnicode_Check(result)) {
1267 Py_UNICODE *s = PyUnicode_AS_UNICODE(result);
1268 int len = PyUnicode_GET_SIZE(result);
1269 if (len == 0) {
1270 Py_DECREF(result);
1271 result = NULL;
1272 PyErr_SetString(PyExc_EOFError,
1273 "EOF when reading a line");
1274 }
1275 else if (s[len-1] == '\n') {
1276 if (result->ob_refcnt == 1)
1277 PyUnicode_Resize(&result, len-1);
1278 else {
1279 PyObject *v;
1280 v = PyUnicode_FromUnicode(s, len-1);
1281 Py_DECREF(result);
1282 result = v;
1283 }
1284 }
1285 }
1286#endif
Guido van Rossum4ddf0a02001-01-07 20:51:39 +00001287 return result;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001288}
1289
1290/* Python method */
1291
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001292static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001293file_readline(PyFileObject *f, PyObject *args)
Guido van Rossum0bd24411991-04-04 15:21:57 +00001294{
Guido van Rossum789a1611997-05-10 22:33:55 +00001295 int n = -1;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001296
Guido van Rossumd7297e61992-07-06 14:19:26 +00001297 if (f->f_fp == NULL)
1298 return err_closed();
Guido van Rossum43713e52000-02-29 13:59:29 +00001299 if (!PyArg_ParseTuple(args, "|i:readline", &n))
Guido van Rossum789a1611997-05-10 22:33:55 +00001300 return NULL;
1301 if (n == 0)
1302 return PyString_FromString("");
1303 if (n < 0)
1304 n = 0;
Marc-André Lemburg1f468602000-07-05 15:32:40 +00001305 return get_line(f, n);
Guido van Rossum0bd24411991-04-04 15:21:57 +00001306}
1307
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001308static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001309file_readlines(PyFileObject *f, PyObject *args)
Guido van Rossumce5ba841991-03-06 13:06:18 +00001310{
Guido van Rossum789a1611997-05-10 22:33:55 +00001311 long sizehint = 0;
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001312 PyObject *list;
1313 PyObject *line;
Guido van Rossum6263d541997-05-10 22:07:25 +00001314 char small_buffer[SMALLCHUNK];
1315 char *buffer = small_buffer;
1316 size_t buffersize = SMALLCHUNK;
1317 PyObject *big_buffer = NULL;
1318 size_t nfilled = 0;
1319 size_t nread;
Guido van Rossum789a1611997-05-10 22:33:55 +00001320 size_t totalread = 0;
Guido van Rossum6263d541997-05-10 22:07:25 +00001321 char *p, *q, *end;
1322 int err;
Guido van Rossum79fd0fc2001-10-12 20:01:53 +00001323 int shortread = 0;
Guido van Rossum0bd24411991-04-04 15:21:57 +00001324
Guido van Rossumd7297e61992-07-06 14:19:26 +00001325 if (f->f_fp == NULL)
1326 return err_closed();
Guido van Rossum43713e52000-02-29 13:59:29 +00001327 if (!PyArg_ParseTuple(args, "|l:readlines", &sizehint))
Guido van Rossum0bd24411991-04-04 15:21:57 +00001328 return NULL;
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001329 if ((list = PyList_New(0)) == NULL)
Guido van Rossumce5ba841991-03-06 13:06:18 +00001330 return NULL;
1331 for (;;) {
Guido van Rossum79fd0fc2001-10-12 20:01:53 +00001332 if (shortread)
1333 nread = 0;
1334 else {
1335 Py_BEGIN_ALLOW_THREADS
1336 errno = 0;
Tim Peters058b1412002-04-21 07:29:14 +00001337 nread = Py_UniversalNewlineFread(buffer+nfilled,
Jack Jansen7b8c7542002-04-14 20:12:41 +00001338 buffersize-nfilled, f->f_fp, (PyObject *)f);
Guido van Rossum79fd0fc2001-10-12 20:01:53 +00001339 Py_END_ALLOW_THREADS
1340 shortread = (nread < buffersize-nfilled);
1341 }
Guido van Rossum6263d541997-05-10 22:07:25 +00001342 if (nread == 0) {
Guido van Rossum789a1611997-05-10 22:33:55 +00001343 sizehint = 0;
Guido van Rossum3da3fce1998-02-19 20:46:48 +00001344 if (!ferror(f->f_fp))
Guido van Rossum6263d541997-05-10 22:07:25 +00001345 break;
1346 PyErr_SetFromErrno(PyExc_IOError);
1347 clearerr(f->f_fp);
1348 error:
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001349 Py_DECREF(list);
Guido van Rossum6263d541997-05-10 22:07:25 +00001350 list = NULL;
1351 goto cleanup;
Guido van Rossumce5ba841991-03-06 13:06:18 +00001352 }
Guido van Rossum789a1611997-05-10 22:33:55 +00001353 totalread += nread;
Guido van Rossum6263d541997-05-10 22:07:25 +00001354 p = memchr(buffer+nfilled, '\n', nread);
1355 if (p == NULL) {
1356 /* Need a larger buffer to fit this line */
1357 nfilled += nread;
1358 buffersize *= 2;
Trent Mickf29f47b2000-08-11 19:02:59 +00001359 if (buffersize > INT_MAX) {
1360 PyErr_SetString(PyExc_OverflowError,
Guido van Rossume07d5cf2001-01-09 21:50:24 +00001361 "line is longer than a Python string can hold");
Trent Mickf29f47b2000-08-11 19:02:59 +00001362 goto error;
1363 }
Guido van Rossum6263d541997-05-10 22:07:25 +00001364 if (big_buffer == NULL) {
1365 /* Create the big buffer */
1366 big_buffer = PyString_FromStringAndSize(
1367 NULL, buffersize);
1368 if (big_buffer == NULL)
1369 goto error;
1370 buffer = PyString_AS_STRING(big_buffer);
1371 memcpy(buffer, small_buffer, nfilled);
1372 }
1373 else {
1374 /* Grow the big buffer */
Jack Jansen7b8c7542002-04-14 20:12:41 +00001375 if ( _PyString_Resize(&big_buffer, buffersize) < 0 )
1376 goto error;
Guido van Rossum6263d541997-05-10 22:07:25 +00001377 buffer = PyString_AS_STRING(big_buffer);
1378 }
1379 continue;
1380 }
1381 end = buffer+nfilled+nread;
1382 q = buffer;
1383 do {
1384 /* Process complete lines */
1385 p++;
1386 line = PyString_FromStringAndSize(q, p-q);
1387 if (line == NULL)
1388 goto error;
1389 err = PyList_Append(list, line);
1390 Py_DECREF(line);
1391 if (err != 0)
1392 goto error;
1393 q = p;
1394 p = memchr(q, '\n', end-q);
1395 } while (p != NULL);
1396 /* Move the remaining incomplete line to the start */
1397 nfilled = end-q;
1398 memmove(buffer, q, nfilled);
Guido van Rossum789a1611997-05-10 22:33:55 +00001399 if (sizehint > 0)
1400 if (totalread >= (size_t)sizehint)
1401 break;
Guido van Rossumce5ba841991-03-06 13:06:18 +00001402 }
Guido van Rossum6263d541997-05-10 22:07:25 +00001403 if (nfilled != 0) {
1404 /* Partial last line */
1405 line = PyString_FromStringAndSize(buffer, nfilled);
1406 if (line == NULL)
1407 goto error;
Guido van Rossum789a1611997-05-10 22:33:55 +00001408 if (sizehint > 0) {
1409 /* Need to complete the last line */
Marc-André Lemburg1f468602000-07-05 15:32:40 +00001410 PyObject *rest = get_line(f, 0);
Guido van Rossum789a1611997-05-10 22:33:55 +00001411 if (rest == NULL) {
1412 Py_DECREF(line);
1413 goto error;
1414 }
1415 PyString_Concat(&line, rest);
1416 Py_DECREF(rest);
1417 if (line == NULL)
1418 goto error;
1419 }
Guido van Rossum6263d541997-05-10 22:07:25 +00001420 err = PyList_Append(list, line);
1421 Py_DECREF(line);
1422 if (err != 0)
1423 goto error;
1424 }
1425 cleanup:
Tim Peters5de98422002-04-27 18:44:32 +00001426 Py_XDECREF(big_buffer);
Guido van Rossumce5ba841991-03-06 13:06:18 +00001427 return list;
1428}
1429
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001430static PyObject *
Fred Drakefd99de62000-07-09 05:02:18 +00001431file_write(PyFileObject *f, PyObject *args)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001432{
Guido van Rossumd7297e61992-07-06 14:19:26 +00001433 char *s;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001434 int n, n2;
Guido van Rossumd7297e61992-07-06 14:19:26 +00001435 if (f->f_fp == NULL)
1436 return err_closed();
Michael W. Hudsone2ec3eb2001-10-31 18:51:01 +00001437 if (!PyArg_ParseTuple(args, f->f_binary ? "s#" : "t#", &s, &n))
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001438 return NULL;
Guido van Rossumeb183da1991-04-04 10:44:06 +00001439 f->f_softspace = 0;
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001440 Py_BEGIN_ALLOW_THREADS
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001441 errno = 0;
Guido van Rossumd7297e61992-07-06 14:19:26 +00001442 n2 = fwrite(s, 1, n, f->f_fp);
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001443 Py_END_ALLOW_THREADS
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001444 if (n2 != n) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001445 PyErr_SetFromErrno(PyExc_IOError);
Guido van Rossumfebd5511992-03-04 16:39:24 +00001446 clearerr(f->f_fp);
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001447 return NULL;
1448 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001449 Py_INCREF(Py_None);
1450 return Py_None;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001451}
1452
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001453static PyObject *
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001454file_writelines(PyFileObject *f, PyObject *seq)
Guido van Rossum5a2a6831993-10-25 09:59:04 +00001455{
Guido van Rossumee70ad12000-03-13 16:27:06 +00001456#define CHUNKSIZE 1000
1457 PyObject *list, *line;
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001458 PyObject *it; /* iter(seq) */
Guido van Rossumee70ad12000-03-13 16:27:06 +00001459 PyObject *result;
1460 int i, j, index, len, nwritten, islist;
1461
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001462 assert(seq != NULL);
Guido van Rossum5a2a6831993-10-25 09:59:04 +00001463 if (f->f_fp == NULL)
1464 return err_closed();
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001465
1466 result = NULL;
1467 list = NULL;
1468 islist = PyList_Check(seq);
1469 if (islist)
1470 it = NULL;
1471 else {
1472 it = PyObject_GetIter(seq);
1473 if (it == NULL) {
1474 PyErr_SetString(PyExc_TypeError,
1475 "writelines() requires an iterable argument");
1476 return NULL;
1477 }
1478 /* From here on, fail by going to error, to reclaim "it". */
1479 list = PyList_New(CHUNKSIZE);
1480 if (list == NULL)
1481 goto error;
Guido van Rossum5a2a6831993-10-25 09:59:04 +00001482 }
Guido van Rossumee70ad12000-03-13 16:27:06 +00001483
1484 /* Strategy: slurp CHUNKSIZE lines into a private list,
1485 checking that they are all strings, then write that list
1486 without holding the interpreter lock, then come back for more. */
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001487 for (index = 0; ; index += CHUNKSIZE) {
Guido van Rossumee70ad12000-03-13 16:27:06 +00001488 if (islist) {
1489 Py_XDECREF(list);
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001490 list = PyList_GetSlice(seq, index, index+CHUNKSIZE);
Guido van Rossumee70ad12000-03-13 16:27:06 +00001491 if (list == NULL)
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001492 goto error;
Guido van Rossumee70ad12000-03-13 16:27:06 +00001493 j = PyList_GET_SIZE(list);
1494 }
1495 else {
1496 for (j = 0; j < CHUNKSIZE; j++) {
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001497 line = PyIter_Next(it);
Guido van Rossumee70ad12000-03-13 16:27:06 +00001498 if (line == NULL) {
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001499 if (PyErr_Occurred())
1500 goto error;
1501 break;
Guido van Rossumee70ad12000-03-13 16:27:06 +00001502 }
Guido van Rossumee70ad12000-03-13 16:27:06 +00001503 PyList_SetItem(list, j, line);
1504 }
1505 }
1506 if (j == 0)
1507 break;
1508
Marc-André Lemburg6ef68b52000-08-25 22:39:50 +00001509 /* Check that all entries are indeed strings. If not,
1510 apply the same rules as for file.write() and
1511 convert the results to strings. This is slow, but
1512 seems to be the only way since all conversion APIs
1513 could potentially execute Python code. */
1514 for (i = 0; i < j; i++) {
1515 PyObject *v = PyList_GET_ITEM(list, i);
1516 if (!PyString_Check(v)) {
1517 const char *buffer;
1518 int len;
Tim Peters86821b22001-01-07 21:19:34 +00001519 if (((f->f_binary &&
Marc-André Lemburg6ef68b52000-08-25 22:39:50 +00001520 PyObject_AsReadBuffer(v,
1521 (const void**)&buffer,
1522 &len)) ||
1523 PyObject_AsCharBuffer(v,
1524 &buffer,
1525 &len))) {
1526 PyErr_SetString(PyExc_TypeError,
Jeremy Hylton8b735422002-08-14 21:01:41 +00001527 "writelines() argument must be a sequence of strings");
Marc-André Lemburg6ef68b52000-08-25 22:39:50 +00001528 goto error;
1529 }
1530 line = PyString_FromStringAndSize(buffer,
1531 len);
1532 if (line == NULL)
1533 goto error;
1534 Py_DECREF(v);
Marc-André Lemburgf5e96fa2000-08-25 22:49:05 +00001535 PyList_SET_ITEM(list, i, line);
Marc-André Lemburg6ef68b52000-08-25 22:39:50 +00001536 }
1537 }
1538
1539 /* Since we are releasing the global lock, the
1540 following code may *not* execute Python code. */
Guido van Rossumee70ad12000-03-13 16:27:06 +00001541 Py_BEGIN_ALLOW_THREADS
1542 f->f_softspace = 0;
1543 errno = 0;
1544 for (i = 0; i < j; i++) {
Marc-André Lemburg6ef68b52000-08-25 22:39:50 +00001545 line = PyList_GET_ITEM(list, i);
Guido van Rossumee70ad12000-03-13 16:27:06 +00001546 len = PyString_GET_SIZE(line);
1547 nwritten = fwrite(PyString_AS_STRING(line),
1548 1, len, f->f_fp);
1549 if (nwritten != len) {
1550 Py_BLOCK_THREADS
1551 PyErr_SetFromErrno(PyExc_IOError);
1552 clearerr(f->f_fp);
1553 goto error;
1554 }
1555 }
1556 Py_END_ALLOW_THREADS
1557
1558 if (j < CHUNKSIZE)
1559 break;
Guido van Rossumee70ad12000-03-13 16:27:06 +00001560 }
1561
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001562 Py_INCREF(Py_None);
Guido van Rossumee70ad12000-03-13 16:27:06 +00001563 result = Py_None;
1564 error:
1565 Py_XDECREF(list);
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001566 Py_XDECREF(it);
Guido van Rossumee70ad12000-03-13 16:27:06 +00001567 return result;
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001568#undef CHUNKSIZE
Guido van Rossum5a2a6831993-10-25 09:59:04 +00001569}
1570
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001571static PyObject *
1572file_getiter(PyFileObject *f)
1573{
1574 if (f->f_fp == NULL)
1575 return err_closed();
1576 Py_INCREF(f);
1577 return (PyObject *)f;
1578}
1579
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001580PyDoc_STRVAR(readline_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001581"readline([size]) -> next line from the file, as a string.\n"
1582"\n"
1583"Retain newline. A non-negative size argument limits the maximum\n"
1584"number of bytes to return (an incomplete line may be returned then).\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001585"Return an empty string at EOF.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001586
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001587PyDoc_STRVAR(read_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001588"read([size]) -> read at most size bytes, returned as a string.\n"
1589"\n"
Gustavo Niemeyer786ddb22002-12-16 18:12:53 +00001590"If the size argument is negative or omitted, read until EOF is reached.\n"
1591"Notice that when in non-blocking mode, less data than what was requested\n"
1592"may be returned, even if no size parameter was given.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001593
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001594PyDoc_STRVAR(write_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001595"write(str) -> None. Write string str to file.\n"
1596"\n"
1597"Note that due to buffering, flush() or close() may be needed before\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001598"the file on disk reflects the data written.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001599
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001600PyDoc_STRVAR(fileno_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001601"fileno() -> integer \"file descriptor\".\n"
1602"\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001603"This is needed for lower-level file interfaces, such os.read().");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001604
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001605PyDoc_STRVAR(seek_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001606"seek(offset[, whence]) -> None. Move to new file position.\n"
1607"\n"
1608"Argument offset is a byte count. Optional argument whence defaults to\n"
1609"0 (offset from start of file, offset should be >= 0); other values are 1\n"
1610"(move relative to current position, positive or negative), and 2 (move\n"
1611"relative to end of file, usually negative, although many platforms allow\n"
Martin v. Löwis849a9722003-10-18 09:38:01 +00001612"seeking beyond the end of a file). If the file is opened in text mode,\n"
1613"only offsets returned by tell() are legal. Use of other offsets causes\n"
1614"undefined behavior."
Tim Petersefc3a3a2001-09-20 07:55:22 +00001615"\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001616"Note that not all file objects are seekable.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001617
Guido van Rossumd7047b31995-01-02 19:07:15 +00001618#ifdef HAVE_FTRUNCATE
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001619PyDoc_STRVAR(truncate_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001620"truncate([size]) -> None. Truncate the file to at most size bytes.\n"
1621"\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001622"Size defaults to the current file position, as returned by tell().");
Guido van Rossumd7047b31995-01-02 19:07:15 +00001623#endif
Tim Petersefc3a3a2001-09-20 07:55:22 +00001624
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001625PyDoc_STRVAR(tell_doc,
1626"tell() -> current file position, an integer (may be a long integer).");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001627
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001628PyDoc_STRVAR(readinto_doc,
1629"readinto() -> Undocumented. Don't use this; it may go away.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001630
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001631PyDoc_STRVAR(readlines_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001632"readlines([size]) -> list of strings, each a line from the file.\n"
1633"\n"
1634"Call readline() repeatedly and return a list of the lines so read.\n"
1635"The optional size argument, if given, is an approximate bound on the\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001636"total number of bytes in the lines returned.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001637
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001638PyDoc_STRVAR(xreadlines_doc,
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001639"xreadlines() -> returns self.\n"
Tim Petersefc3a3a2001-09-20 07:55:22 +00001640"\n"
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001641"For backward compatibility. File objects now include the performance\n"
1642"optimizations previously implemented in the xreadlines module.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001643
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001644PyDoc_STRVAR(writelines_doc,
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001645"writelines(sequence_of_strings) -> None. Write the strings to the file.\n"
Tim Petersefc3a3a2001-09-20 07:55:22 +00001646"\n"
Tim Peters2c9aa5e2001-09-23 04:06:05 +00001647"Note that newlines are not added. The sequence can be any iterable object\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001648"producing strings. This is equivalent to calling write() for each string.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001649
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001650PyDoc_STRVAR(flush_doc,
1651"flush() -> None. Flush the internal I/O buffer.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001652
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001653PyDoc_STRVAR(close_doc,
Tim Petersefc3a3a2001-09-20 07:55:22 +00001654"close() -> None or (perhaps) an integer. Close the file.\n"
1655"\n"
Guido van Rossum77f6a652002-04-03 22:41:51 +00001656"Sets data attribute .closed to True. A closed file cannot be used for\n"
Tim Petersefc3a3a2001-09-20 07:55:22 +00001657"further I/O operations. close() may be called more than once without\n"
1658"error. Some kinds of file objects (for example, opened by popen())\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001659"may return an exit status upon closing.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001660
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001661PyDoc_STRVAR(isatty_doc,
1662"isatty() -> true or false. True if the file is connected to a tty device.");
Tim Petersefc3a3a2001-09-20 07:55:22 +00001663
1664static PyMethodDef file_methods[] = {
Jeremy Hylton8b735422002-08-14 21:01:41 +00001665 {"readline", (PyCFunction)file_readline, METH_VARARGS, readline_doc},
1666 {"read", (PyCFunction)file_read, METH_VARARGS, read_doc},
1667 {"write", (PyCFunction)file_write, METH_VARARGS, write_doc},
1668 {"fileno", (PyCFunction)file_fileno, METH_NOARGS, fileno_doc},
1669 {"seek", (PyCFunction)file_seek, METH_VARARGS, seek_doc},
Tim Petersefc3a3a2001-09-20 07:55:22 +00001670#ifdef HAVE_FTRUNCATE
Jeremy Hylton8b735422002-08-14 21:01:41 +00001671 {"truncate", (PyCFunction)file_truncate, METH_VARARGS, truncate_doc},
Tim Petersefc3a3a2001-09-20 07:55:22 +00001672#endif
Jeremy Hylton8b735422002-08-14 21:01:41 +00001673 {"tell", (PyCFunction)file_tell, METH_NOARGS, tell_doc},
1674 {"readinto", (PyCFunction)file_readinto, METH_VARARGS, readinto_doc},
1675 {"readlines", (PyCFunction)file_readlines,METH_VARARGS, readlines_doc},
1676 {"xreadlines",(PyCFunction)file_getiter, METH_NOARGS, xreadlines_doc},
1677 {"writelines",(PyCFunction)file_writelines, METH_O, writelines_doc},
1678 {"flush", (PyCFunction)file_flush, METH_NOARGS, flush_doc},
1679 {"close", (PyCFunction)file_close, METH_NOARGS, close_doc},
1680 {"isatty", (PyCFunction)file_isatty, METH_NOARGS, isatty_doc},
1681 {NULL, NULL} /* sentinel */
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001682};
1683
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001684#define OFF(x) offsetof(PyFileObject, x)
Guido van Rossumb6775db1994-08-01 11:34:53 +00001685
Guido van Rossum6f799372001-09-20 20:46:19 +00001686static PyMemberDef file_memberlist[] = {
1687 {"softspace", T_INT, OFF(f_softspace), 0,
1688 "flag indicating that a space needs to be printed; used by print"},
1689 {"mode", T_OBJECT, OFF(f_mode), RO,
Martin v. Löwis6233c9b2002-12-11 13:06:53 +00001690 "file mode ('r', 'U', 'w', 'a', possibly with 'b' or '+' added)"},
Guido van Rossum6f799372001-09-20 20:46:19 +00001691 {"name", T_OBJECT, OFF(f_name), RO,
1692 "file name"},
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00001693 {"encoding", T_OBJECT, OFF(f_encoding), RO,
1694 "file encoding"},
Guido van Rossumb6775db1994-08-01 11:34:53 +00001695 /* getattr(f, "closed") is implemented without this table */
Guido van Rossumb6775db1994-08-01 11:34:53 +00001696 {NULL} /* Sentinel */
1697};
1698
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001699static PyObject *
Tim Peters6d6c1a32001-08-02 04:15:00 +00001700get_closed(PyFileObject *f, void *closure)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001701{
Guido van Rossum77f6a652002-04-03 22:41:51 +00001702 return PyBool_FromLong((long)(f->f_fp == 0));
Guido van Rossumb6775db1994-08-01 11:34:53 +00001703}
Jack Jansen7b8c7542002-04-14 20:12:41 +00001704static PyObject *
1705get_newlines(PyFileObject *f, void *closure)
1706{
1707 switch (f->f_newlinetypes) {
1708 case NEWLINE_UNKNOWN:
1709 Py_INCREF(Py_None);
1710 return Py_None;
1711 case NEWLINE_CR:
1712 return PyString_FromString("\r");
1713 case NEWLINE_LF:
1714 return PyString_FromString("\n");
1715 case NEWLINE_CR|NEWLINE_LF:
1716 return Py_BuildValue("(ss)", "\r", "\n");
1717 case NEWLINE_CRLF:
1718 return PyString_FromString("\r\n");
1719 case NEWLINE_CR|NEWLINE_CRLF:
1720 return Py_BuildValue("(ss)", "\r", "\r\n");
1721 case NEWLINE_LF|NEWLINE_CRLF:
1722 return Py_BuildValue("(ss)", "\n", "\r\n");
1723 case NEWLINE_CR|NEWLINE_LF|NEWLINE_CRLF:
1724 return Py_BuildValue("(sss)", "\r", "\n", "\r\n");
1725 default:
Tim Petersf1827cf2003-09-07 03:30:18 +00001726 PyErr_Format(PyExc_SystemError,
1727 "Unknown newlines value 0x%x\n",
Jeremy Hylton8b735422002-08-14 21:01:41 +00001728 f->f_newlinetypes);
Jack Jansen7b8c7542002-04-14 20:12:41 +00001729 return NULL;
1730 }
1731}
Guido van Rossumb6775db1994-08-01 11:34:53 +00001732
Guido van Rossum32d34c82001-09-20 21:45:26 +00001733static PyGetSetDef file_getsetlist[] = {
Guido van Rossum77f6a652002-04-03 22:41:51 +00001734 {"closed", (getter)get_closed, NULL, "True if the file is closed"},
Tim Petersf1827cf2003-09-07 03:30:18 +00001735 {"newlines", (getter)get_newlines, NULL,
Jeremy Hylton8b735422002-08-14 21:01:41 +00001736 "end-of-line convention used in this file"},
Tim Peters6d6c1a32001-08-02 04:15:00 +00001737 {0},
1738};
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001739
Neal Norwitzd8b995f2002-08-06 21:50:54 +00001740static void
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001741drop_readahead(PyFileObject *f)
Guido van Rossum65967252001-04-21 13:20:18 +00001742{
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001743 if (f->f_buf != NULL) {
1744 PyMem_Free(f->f_buf);
1745 f->f_buf = NULL;
1746 }
Guido van Rossum65967252001-04-21 13:20:18 +00001747}
1748
Tim Petersf1827cf2003-09-07 03:30:18 +00001749/* Make sure that file has a readahead buffer with at least one byte
1750 (unless at EOF) and no more than bufsize. Returns negative value on
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001751 error */
Neal Norwitzd8b995f2002-08-06 21:50:54 +00001752static int
1753readahead(PyFileObject *f, int bufsize)
1754{
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001755 int chunksize;
1756
1757 if (f->f_buf != NULL) {
Tim Petersf1827cf2003-09-07 03:30:18 +00001758 if( (f->f_bufend - f->f_bufptr) >= 1)
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001759 return 0;
1760 else
1761 drop_readahead(f);
1762 }
1763 if ((f->f_buf = PyMem_Malloc(bufsize)) == NULL) {
1764 return -1;
1765 }
1766 Py_BEGIN_ALLOW_THREADS
1767 errno = 0;
1768 chunksize = Py_UniversalNewlineFread(
1769 f->f_buf, bufsize, f->f_fp, (PyObject *)f);
1770 Py_END_ALLOW_THREADS
1771 if (chunksize == 0) {
1772 if (ferror(f->f_fp)) {
1773 PyErr_SetFromErrno(PyExc_IOError);
1774 clearerr(f->f_fp);
1775 drop_readahead(f);
1776 return -1;
1777 }
1778 }
1779 f->f_bufptr = f->f_buf;
1780 f->f_bufend = f->f_buf + chunksize;
1781 return 0;
1782}
1783
1784/* Used by file_iternext. The returned string will start with 'skip'
Tim Petersf1827cf2003-09-07 03:30:18 +00001785 uninitialized bytes followed by the remainder of the line. Don't be
1786 horrified by the recursive call: maximum recursion depth is limited by
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001787 logarithmic buffer growth to about 50 even when reading a 1gb line. */
1788
Neal Norwitzd8b995f2002-08-06 21:50:54 +00001789static PyStringObject *
1790readahead_get_line_skip(PyFileObject *f, int skip, int bufsize)
1791{
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001792 PyStringObject* s;
1793 char *bufptr;
1794 char *buf;
1795 int len;
1796
1797 if (f->f_buf == NULL)
Tim Petersf1827cf2003-09-07 03:30:18 +00001798 if (readahead(f, bufsize) < 0)
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001799 return NULL;
1800
1801 len = f->f_bufend - f->f_bufptr;
Tim Petersf1827cf2003-09-07 03:30:18 +00001802 if (len == 0)
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001803 return (PyStringObject *)
1804 PyString_FromStringAndSize(NULL, skip);
1805 bufptr = memchr(f->f_bufptr, '\n', len);
1806 if (bufptr != NULL) {
1807 bufptr++; /* Count the '\n' */
1808 len = bufptr - f->f_bufptr;
1809 s = (PyStringObject *)
1810 PyString_FromStringAndSize(NULL, skip+len);
Tim Petersf1827cf2003-09-07 03:30:18 +00001811 if (s == NULL)
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001812 return NULL;
1813 memcpy(PyString_AS_STRING(s)+skip, f->f_bufptr, len);
1814 f->f_bufptr = bufptr;
1815 if (bufptr == f->f_bufend)
1816 drop_readahead(f);
1817 } else {
1818 bufptr = f->f_bufptr;
1819 buf = f->f_buf;
1820 f->f_buf = NULL; /* Force new readahead buffer */
1821 s = readahead_get_line_skip(
1822 f, skip+len, bufsize + (bufsize>>2) );
1823 if (s == NULL) {
1824 PyMem_Free(buf);
1825 return NULL;
1826 }
1827 memcpy(PyString_AS_STRING(s)+skip, bufptr, len);
1828 PyMem_Free(buf);
1829 }
1830 return s;
1831}
1832
1833/* A larger buffer size may actually decrease performance. */
1834#define READAHEAD_BUFSIZE 8192
1835
1836static PyObject *
1837file_iternext(PyFileObject *f)
1838{
1839 PyStringObject* l;
1840
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001841 if (f->f_fp == NULL)
1842 return err_closed();
1843
Guido van Rossum7a6e9592002-08-06 15:55:28 +00001844 l = readahead_get_line_skip(f, 0, READAHEAD_BUFSIZE);
1845 if (l == NULL || PyString_GET_SIZE(l) == 0) {
1846 Py_XDECREF(l);
1847 return NULL;
1848 }
1849 return (PyObject *)l;
1850}
1851
1852
Tim Peters59c9a642001-09-13 05:38:56 +00001853static PyObject *
1854file_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1855{
Tim Peters44410012001-09-14 03:26:08 +00001856 PyObject *self;
1857 static PyObject *not_yet_string;
1858
1859 assert(type != NULL && type->tp_alloc != NULL);
1860
1861 if (not_yet_string == NULL) {
1862 not_yet_string = PyString_FromString("<uninitialized file>");
1863 if (not_yet_string == NULL)
1864 return NULL;
1865 }
1866
1867 self = type->tp_alloc(type, 0);
1868 if (self != NULL) {
1869 /* Always fill in the name and mode, so that nobody else
1870 needs to special-case NULLs there. */
1871 Py_INCREF(not_yet_string);
1872 ((PyFileObject *)self)->f_name = not_yet_string;
1873 Py_INCREF(not_yet_string);
1874 ((PyFileObject *)self)->f_mode = not_yet_string;
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00001875 Py_INCREF(Py_None);
1876 ((PyFileObject *)self)->f_encoding = Py_None;
Raymond Hettingercb87bc82004-05-31 00:35:52 +00001877 ((PyFileObject *)self)->weakreflist = NULL;
Tim Peters44410012001-09-14 03:26:08 +00001878 }
1879 return self;
1880}
1881
1882static int
1883file_init(PyObject *self, PyObject *args, PyObject *kwds)
1884{
1885 PyFileObject *foself = (PyFileObject *)self;
1886 int ret = 0;
Jeremy Hyltonaf68c872005-12-10 18:50:16 +00001887 static const char *kwlist[] = {"name", "mode", "buffering", 0};
Tim Peters59c9a642001-09-13 05:38:56 +00001888 char *name = NULL;
1889 char *mode = "r";
1890 int bufsize = -1;
Mark Hammondc2e85bd2002-10-03 05:10:39 +00001891 int wideargument = 0;
Tim Peters44410012001-09-14 03:26:08 +00001892
1893 assert(PyFile_Check(self));
1894 if (foself->f_fp != NULL) {
1895 /* Have to close the existing file first. */
1896 PyObject *closeresult = file_close(foself);
1897 if (closeresult == NULL)
1898 return -1;
1899 Py_DECREF(closeresult);
1900 }
Tim Peters59c9a642001-09-13 05:38:56 +00001901
Mark Hammondc2e85bd2002-10-03 05:10:39 +00001902#ifdef Py_WIN_WIDE_FILENAMES
1903 if (GetVersion() < 0x80000000) { /* On NT, so wide API available */
1904 PyObject *po;
1905 if (PyArg_ParseTupleAndKeywords(args, kwds, "U|si:file",
1906 kwlist, &po, &mode, &bufsize)) {
1907 wideargument = 1;
Nicholas Bastinabce8a62004-03-21 20:24:07 +00001908 if (fill_file_fields(foself, NULL, po, mode,
1909 fclose) == NULL)
Mark Hammondc2e85bd2002-10-03 05:10:39 +00001910 goto Error;
1911 } else {
1912 /* Drop the argument parsing error as narrow
1913 strings are also valid. */
1914 PyErr_Clear();
1915 }
1916 }
1917#endif
1918
1919 if (!wideargument) {
Nicholas Bastinabce8a62004-03-21 20:24:07 +00001920 PyObject *o_name;
1921
Mark Hammondc2e85bd2002-10-03 05:10:39 +00001922 if (!PyArg_ParseTupleAndKeywords(args, kwds, "et|si:file", kwlist,
1923 Py_FileSystemDefaultEncoding,
1924 &name,
1925 &mode, &bufsize))
1926 return -1;
Nicholas Bastinabce8a62004-03-21 20:24:07 +00001927
1928 /* We parse again to get the name as a PyObject */
Jeremy Hyltonaf68c872005-12-10 18:50:16 +00001929 if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|si:file",
1930 kwlist, &o_name, &mode,
1931 &bufsize))
Nicholas Bastinabce8a62004-03-21 20:24:07 +00001932 return -1;
1933
1934 if (fill_file_fields(foself, NULL, o_name, mode,
1935 fclose) == NULL)
Mark Hammondc2e85bd2002-10-03 05:10:39 +00001936 goto Error;
1937 }
Tim Peters44410012001-09-14 03:26:08 +00001938 if (open_the_file(foself, name, mode) == NULL)
1939 goto Error;
Martin v. Löwis1e3bdf62003-09-04 19:01:46 +00001940 foself->f_setbuf = NULL;
Tim Peters44410012001-09-14 03:26:08 +00001941 PyFile_SetBufSize(self, bufsize);
1942 goto Done;
1943
1944Error:
1945 ret = -1;
1946 /* fall through */
1947Done:
Tim Peters59c9a642001-09-13 05:38:56 +00001948 PyMem_Free(name); /* free the encoded string */
Tim Peters44410012001-09-14 03:26:08 +00001949 return ret;
Tim Peters59c9a642001-09-13 05:38:56 +00001950}
1951
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001952PyDoc_VAR(file_doc) =
1953PyDoc_STR(
Tim Peters59c9a642001-09-13 05:38:56 +00001954"file(name[, mode[, buffering]]) -> file object\n"
1955"\n"
1956"Open a file. The mode can be 'r', 'w' or 'a' for reading (default),\n"
1957"writing or appending. The file will be created if it doesn't exist\n"
1958"when opened for writing or appending; it will be truncated when\n"
1959"opened for writing. Add a 'b' to the mode for binary files.\n"
1960"Add a '+' to the mode to allow simultaneous reading and writing.\n"
1961"If the buffering argument is given, 0 means unbuffered, 1 means line\n"
Tim Peters742dfd62001-09-13 21:49:44 +00001962"buffered, and larger numbers specify the buffer size.\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001963)
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001964PyDoc_STR(
Barry Warsaw4be55b52002-05-22 20:37:53 +00001965"Add a 'U' to mode to open the file for input with universal newline\n"
1966"support. Any line ending in the input file will be seen as a '\\n'\n"
1967"in Python. Also, a file so opened gains the attribute 'newlines';\n"
1968"the value for this attribute is one of None (no newline read yet),\n"
1969"'\\r', '\\n', '\\r\\n' or a tuple containing all the newline types seen.\n"
1970"\n"
1971"'U' cannot be combined with 'w' or '+' mode.\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001972)
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001973PyDoc_STR(
Barry Warsaw4be55b52002-05-22 20:37:53 +00001974"\n"
Martin v. Löwis14f8b4c2002-06-13 20:33:02 +00001975"Note: open() is an alias for file()."
1976);
Tim Peters59c9a642001-09-13 05:38:56 +00001977
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001978PyTypeObject PyFile_Type = {
1979 PyObject_HEAD_INIT(&PyType_Type)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001980 0,
1981 "file",
Guido van Rossumc0b618a1997-05-02 03:12:38 +00001982 sizeof(PyFileObject),
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00001983 0,
Guido van Rossum65967252001-04-21 13:20:18 +00001984 (destructor)file_dealloc, /* tp_dealloc */
1985 0, /* tp_print */
Tim Peters6d6c1a32001-08-02 04:15:00 +00001986 0, /* tp_getattr */
1987 0, /* tp_setattr */
Guido van Rossum65967252001-04-21 13:20:18 +00001988 0, /* tp_compare */
Tim Peters6d6c1a32001-08-02 04:15:00 +00001989 (reprfunc)file_repr, /* tp_repr */
Guido van Rossum65967252001-04-21 13:20:18 +00001990 0, /* tp_as_number */
1991 0, /* tp_as_sequence */
1992 0, /* tp_as_mapping */
1993 0, /* tp_hash */
1994 0, /* tp_call */
1995 0, /* tp_str */
Tim Peters6d6c1a32001-08-02 04:15:00 +00001996 PyObject_GenericGetAttr, /* tp_getattro */
Tim Peters015dd822003-05-04 04:16:52 +00001997 /* softspace is writable: we must supply tp_setattro */
1998 PyObject_GenericSetAttr, /* tp_setattro */
Guido van Rossum65967252001-04-21 13:20:18 +00001999 0, /* tp_as_buffer */
Raymond Hettingercb87bc82004-05-31 00:35:52 +00002000 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | Py_TPFLAGS_HAVE_WEAKREFS, /* tp_flags */
Tim Peters59c9a642001-09-13 05:38:56 +00002001 file_doc, /* tp_doc */
Tim Peters6d6c1a32001-08-02 04:15:00 +00002002 0, /* tp_traverse */
2003 0, /* tp_clear */
Guido van Rossum65967252001-04-21 13:20:18 +00002004 0, /* tp_richcompare */
Raymond Hettingercb87bc82004-05-31 00:35:52 +00002005 offsetof(PyFileObject, weakreflist), /* tp_weaklistoffset */
Guido van Rossum7a6e9592002-08-06 15:55:28 +00002006 (getiterfunc)file_getiter, /* tp_iter */
2007 (iternextfunc)file_iternext, /* tp_iternext */
Tim Peters6d6c1a32001-08-02 04:15:00 +00002008 file_methods, /* tp_methods */
2009 file_memberlist, /* tp_members */
2010 file_getsetlist, /* tp_getset */
2011 0, /* tp_base */
2012 0, /* tp_dict */
Tim Peters59c9a642001-09-13 05:38:56 +00002013 0, /* tp_descr_get */
2014 0, /* tp_descr_set */
2015 0, /* tp_dictoffset */
Tim Peters44410012001-09-14 03:26:08 +00002016 (initproc)file_init, /* tp_init */
2017 PyType_GenericAlloc, /* tp_alloc */
Tim Peters59c9a642001-09-13 05:38:56 +00002018 file_new, /* tp_new */
Neil Schemenaueraa769ae2002-04-12 02:44:10 +00002019 PyObject_Del, /* tp_free */
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00002020};
Guido van Rossumeb183da1991-04-04 10:44:06 +00002021
2022/* Interface for the 'soft space' between print items. */
2023
2024int
Fred Drakefd99de62000-07-09 05:02:18 +00002025PyFile_SoftSpace(PyObject *f, int newflag)
Guido van Rossumeb183da1991-04-04 10:44:06 +00002026{
2027 int oldflag = 0;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002028 if (f == NULL) {
2029 /* Do nothing */
2030 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002031 else if (PyFile_Check(f)) {
2032 oldflag = ((PyFileObject *)f)->f_softspace;
2033 ((PyFileObject *)f)->f_softspace = newflag;
Guido van Rossumeb183da1991-04-04 10:44:06 +00002034 }
Guido van Rossum3165fe61992-09-25 21:59:05 +00002035 else {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002036 PyObject *v;
2037 v = PyObject_GetAttrString(f, "softspace");
Guido van Rossum3165fe61992-09-25 21:59:05 +00002038 if (v == NULL)
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002039 PyErr_Clear();
Guido van Rossum3165fe61992-09-25 21:59:05 +00002040 else {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002041 if (PyInt_Check(v))
2042 oldflag = PyInt_AsLong(v);
2043 Py_DECREF(v);
Guido van Rossum3165fe61992-09-25 21:59:05 +00002044 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002045 v = PyInt_FromLong((long)newflag);
Guido van Rossum3165fe61992-09-25 21:59:05 +00002046 if (v == NULL)
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002047 PyErr_Clear();
Guido van Rossum3165fe61992-09-25 21:59:05 +00002048 else {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002049 if (PyObject_SetAttrString(f, "softspace", v) != 0)
2050 PyErr_Clear();
2051 Py_DECREF(v);
Guido van Rossum3165fe61992-09-25 21:59:05 +00002052 }
2053 }
Guido van Rossumeb183da1991-04-04 10:44:06 +00002054 return oldflag;
2055}
Guido van Rossum3165fe61992-09-25 21:59:05 +00002056
2057/* Interfaces to write objects/strings to file-like objects */
2058
2059int
Fred Drakefd99de62000-07-09 05:02:18 +00002060PyFile_WriteObject(PyObject *v, PyObject *f, int flags)
Guido van Rossum3165fe61992-09-25 21:59:05 +00002061{
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002062 PyObject *writer, *value, *args, *result;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002063 if (f == NULL) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002064 PyErr_SetString(PyExc_TypeError, "writeobject with NULL file");
Guido van Rossum3165fe61992-09-25 21:59:05 +00002065 return -1;
2066 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002067 else if (PyFile_Check(f)) {
2068 FILE *fp = PyFile_AsFile(f);
Fred Drake086a0f72004-03-19 15:22:36 +00002069#ifdef Py_USING_UNICODE
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00002070 PyObject *enc = ((PyFileObject*)f)->f_encoding;
2071 int result;
Fred Drake086a0f72004-03-19 15:22:36 +00002072#endif
Guido van Rossum3165fe61992-09-25 21:59:05 +00002073 if (fp == NULL) {
2074 err_closed();
2075 return -1;
2076 }
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00002077#ifdef Py_USING_UNICODE
Tim Petersf1827cf2003-09-07 03:30:18 +00002078 if ((flags & Py_PRINT_RAW) &&
Martin v. Löwis415da6e2003-05-18 12:56:25 +00002079 PyUnicode_Check(v) && enc != Py_None) {
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00002080 char *cenc = PyString_AS_STRING(enc);
2081 value = PyUnicode_AsEncodedString(v, cenc, "strict");
2082 if (value == NULL)
2083 return -1;
2084 } else {
2085 value = v;
2086 Py_INCREF(value);
2087 }
2088 result = PyObject_Print(value, fp, flags);
2089 Py_DECREF(value);
2090 return result;
2091#else
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002092 return PyObject_Print(v, fp, flags);
Martin v. Löwis5467d4c2003-05-10 07:10:12 +00002093#endif
Guido van Rossum3165fe61992-09-25 21:59:05 +00002094 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002095 writer = PyObject_GetAttrString(f, "write");
Guido van Rossum3165fe61992-09-25 21:59:05 +00002096 if (writer == NULL)
2097 return -1;
Martin v. Löwis2777c022001-09-19 13:47:32 +00002098 if (flags & Py_PRINT_RAW) {
2099 if (PyUnicode_Check(v)) {
2100 value = v;
2101 Py_INCREF(value);
2102 } else
2103 value = PyObject_Str(v);
2104 }
2105 else
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002106 value = PyObject_Repr(v);
Guido van Rossumc6004111993-11-05 10:22:19 +00002107 if (value == NULL) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002108 Py_DECREF(writer);
Guido van Rossumc6004111993-11-05 10:22:19 +00002109 return -1;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002110 }
Raymond Hettinger8ae46892003-10-12 19:09:37 +00002111 args = PyTuple_Pack(1, value);
Guido van Rossume9eec541997-05-22 14:02:25 +00002112 if (args == NULL) {
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002113 Py_DECREF(value);
2114 Py_DECREF(writer);
Guido van Rossumd3f9a1a1995-07-10 23:32:26 +00002115 return -1;
2116 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002117 result = PyEval_CallObject(writer, args);
2118 Py_DECREF(args);
2119 Py_DECREF(value);
2120 Py_DECREF(writer);
Guido van Rossum3165fe61992-09-25 21:59:05 +00002121 if (result == NULL)
2122 return -1;
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002123 Py_DECREF(result);
Guido van Rossum3165fe61992-09-25 21:59:05 +00002124 return 0;
2125}
2126
Guido van Rossum27a60b11997-05-22 22:25:11 +00002127int
Tim Petersc1bbcb82001-11-28 22:13:25 +00002128PyFile_WriteString(const char *s, PyObject *f)
Guido van Rossum3165fe61992-09-25 21:59:05 +00002129{
2130 if (f == NULL) {
Guido van Rossum27a60b11997-05-22 22:25:11 +00002131 /* Should be caused by a pre-existing error */
Fred Drakefd99de62000-07-09 05:02:18 +00002132 if (!PyErr_Occurred())
Guido van Rossum27a60b11997-05-22 22:25:11 +00002133 PyErr_SetString(PyExc_SystemError,
2134 "null file for PyFile_WriteString");
2135 return -1;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002136 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002137 else if (PyFile_Check(f)) {
2138 FILE *fp = PyFile_AsFile(f);
Guido van Rossum27a60b11997-05-22 22:25:11 +00002139 if (fp == NULL) {
2140 err_closed();
2141 return -1;
2142 }
2143 fputs(s, fp);
2144 return 0;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002145 }
Guido van Rossumc0b618a1997-05-02 03:12:38 +00002146 else if (!PyErr_Occurred()) {
2147 PyObject *v = PyString_FromString(s);
Guido van Rossum27a60b11997-05-22 22:25:11 +00002148 int err;
2149 if (v == NULL)
2150 return -1;
2151 err = PyFile_WriteObject(v, f, Py_PRINT_RAW);
2152 Py_DECREF(v);
2153 return err;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002154 }
Guido van Rossum74ba2471997-07-13 03:56:50 +00002155 else
2156 return -1;
Guido van Rossum3165fe61992-09-25 21:59:05 +00002157}
Andrew M. Kuchling06051ed2000-07-13 23:56:54 +00002158
2159/* Try to get a file-descriptor from a Python object. If the object
2160 is an integer or long integer, its value is returned. If not, the
2161 object's fileno() method is called if it exists; the method must return
2162 an integer or long integer, which is returned as the file descriptor value.
2163 -1 is returned on failure.
2164*/
2165
2166int PyObject_AsFileDescriptor(PyObject *o)
2167{
2168 int fd;
2169 PyObject *meth;
2170
2171 if (PyInt_Check(o)) {
2172 fd = PyInt_AsLong(o);
2173 }
2174 else if (PyLong_Check(o)) {
2175 fd = PyLong_AsLong(o);
2176 }
2177 else if ((meth = PyObject_GetAttrString(o, "fileno")) != NULL)
2178 {
2179 PyObject *fno = PyEval_CallObject(meth, NULL);
2180 Py_DECREF(meth);
2181 if (fno == NULL)
2182 return -1;
Tim Peters86821b22001-01-07 21:19:34 +00002183
Andrew M. Kuchling06051ed2000-07-13 23:56:54 +00002184 if (PyInt_Check(fno)) {
2185 fd = PyInt_AsLong(fno);
2186 Py_DECREF(fno);
2187 }
2188 else if (PyLong_Check(fno)) {
2189 fd = PyLong_AsLong(fno);
2190 Py_DECREF(fno);
2191 }
2192 else {
2193 PyErr_SetString(PyExc_TypeError,
2194 "fileno() returned a non-integer");
2195 Py_DECREF(fno);
2196 return -1;
2197 }
2198 }
2199 else {
2200 PyErr_SetString(PyExc_TypeError,
2201 "argument must be an int, or have a fileno() method.");
2202 return -1;
2203 }
2204
2205 if (fd < 0) {
2206 PyErr_Format(PyExc_ValueError,
2207 "file descriptor cannot be a negative integer (%i)",
2208 fd);
2209 return -1;
2210 }
2211 return fd;
2212}
Jack Jansen7b8c7542002-04-14 20:12:41 +00002213
Jack Jansen7b8c7542002-04-14 20:12:41 +00002214/* From here on we need access to the real fgets and fread */
2215#undef fgets
2216#undef fread
2217
2218/*
2219** Py_UniversalNewlineFgets is an fgets variation that understands
2220** all of \r, \n and \r\n conventions.
2221** The stream should be opened in binary mode.
2222** If fobj is NULL the routine always does newline conversion, and
2223** it may peek one char ahead to gobble the second char in \r\n.
2224** If fobj is non-NULL it must be a PyFileObject. In this case there
2225** is no readahead but in stead a flag is used to skip a following
2226** \n on the next read. Also, if the file is open in binary mode
2227** the whole conversion is skipped. Finally, the routine keeps track of
2228** the different types of newlines seen.
2229** Note that we need no error handling: fgets() treats error and eof
2230** identically.
2231*/
2232char *
2233Py_UniversalNewlineFgets(char *buf, int n, FILE *stream, PyObject *fobj)
2234{
2235 char *p = buf;
2236 int c;
2237 int newlinetypes = 0;
2238 int skipnextlf = 0;
2239 int univ_newline = 1;
Tim Peters058b1412002-04-21 07:29:14 +00002240
Jack Jansen7b8c7542002-04-14 20:12:41 +00002241 if (fobj) {
2242 if (!PyFile_Check(fobj)) {
2243 errno = ENXIO; /* What can you do... */
2244 return NULL;
2245 }
2246 univ_newline = ((PyFileObject *)fobj)->f_univ_newline;
2247 if ( !univ_newline )
2248 return fgets(buf, n, stream);
2249 newlinetypes = ((PyFileObject *)fobj)->f_newlinetypes;
2250 skipnextlf = ((PyFileObject *)fobj)->f_skipnextlf;
2251 }
2252 FLOCKFILE(stream);
2253 c = 'x'; /* Shut up gcc warning */
2254 while (--n > 0 && (c = GETC(stream)) != EOF ) {
2255 if (skipnextlf ) {
2256 skipnextlf = 0;
2257 if (c == '\n') {
2258 /* Seeing a \n here with skipnextlf true
2259 ** means we saw a \r before.
2260 */
2261 newlinetypes |= NEWLINE_CRLF;
2262 c = GETC(stream);
2263 if (c == EOF) break;
2264 } else {
2265 /*
2266 ** Note that c == EOF also brings us here,
2267 ** so we're okay if the last char in the file
2268 ** is a CR.
2269 */
2270 newlinetypes |= NEWLINE_CR;
2271 }
2272 }
2273 if (c == '\r') {
2274 /* A \r is translated into a \n, and we skip
2275 ** an adjacent \n, if any. We don't set the
2276 ** newlinetypes flag until we've seen the next char.
2277 */
2278 skipnextlf = 1;
2279 c = '\n';
2280 } else if ( c == '\n') {
2281 newlinetypes |= NEWLINE_LF;
2282 }
2283 *p++ = c;
2284 if (c == '\n') break;
2285 }
2286 if ( c == EOF && skipnextlf )
2287 newlinetypes |= NEWLINE_CR;
2288 FUNLOCKFILE(stream);
2289 *p = '\0';
2290 if (fobj) {
2291 ((PyFileObject *)fobj)->f_newlinetypes = newlinetypes;
2292 ((PyFileObject *)fobj)->f_skipnextlf = skipnextlf;
2293 } else if ( skipnextlf ) {
2294 /* If we have no file object we cannot save the
2295 ** skipnextlf flag. We have to readahead, which
2296 ** will cause a pause if we're reading from an
2297 ** interactive stream, but that is very unlikely
2298 ** unless we're doing something silly like
2299 ** execfile("/dev/tty").
2300 */
2301 c = GETC(stream);
2302 if ( c != '\n' )
2303 ungetc(c, stream);
2304 }
2305 if (p == buf)
2306 return NULL;
2307 return buf;
2308}
2309
2310/*
2311** Py_UniversalNewlineFread is an fread variation that understands
2312** all of \r, \n and \r\n conventions.
2313** The stream should be opened in binary mode.
2314** fobj must be a PyFileObject. In this case there
2315** is no readahead but in stead a flag is used to skip a following
2316** \n on the next read. Also, if the file is open in binary mode
2317** the whole conversion is skipped. Finally, the routine keeps track of
2318** the different types of newlines seen.
2319*/
2320size_t
Tim Peters058b1412002-04-21 07:29:14 +00002321Py_UniversalNewlineFread(char *buf, size_t n,
Jack Jansen7b8c7542002-04-14 20:12:41 +00002322 FILE *stream, PyObject *fobj)
2323{
Tim Peters058b1412002-04-21 07:29:14 +00002324 char *dst = buf;
2325 PyFileObject *f = (PyFileObject *)fobj;
2326 int newlinetypes, skipnextlf;
2327
2328 assert(buf != NULL);
2329 assert(stream != NULL);
2330
Jack Jansen7b8c7542002-04-14 20:12:41 +00002331 if (!fobj || !PyFile_Check(fobj)) {
2332 errno = ENXIO; /* What can you do... */
Neal Norwitzcb3319f2003-02-09 01:10:02 +00002333 return 0;
Jack Jansen7b8c7542002-04-14 20:12:41 +00002334 }
Tim Peters058b1412002-04-21 07:29:14 +00002335 if (!f->f_univ_newline)
Jack Jansen7b8c7542002-04-14 20:12:41 +00002336 return fread(buf, 1, n, stream);
Tim Peters058b1412002-04-21 07:29:14 +00002337 newlinetypes = f->f_newlinetypes;
2338 skipnextlf = f->f_skipnextlf;
2339 /* Invariant: n is the number of bytes remaining to be filled
2340 * in the buffer.
2341 */
2342 while (n) {
2343 size_t nread;
2344 int shortread;
2345 char *src = dst;
2346
2347 nread = fread(dst, 1, n, stream);
2348 assert(nread <= n);
Neal Norwitzcb3319f2003-02-09 01:10:02 +00002349 if (nread == 0)
2350 break;
2351
Tim Peterse1682a82002-04-21 18:15:20 +00002352 n -= nread; /* assuming 1 byte out for each in; will adjust */
2353 shortread = n != 0; /* true iff EOF or error */
Tim Peters058b1412002-04-21 07:29:14 +00002354 while (nread--) {
2355 char c = *src++;
Jack Jansen7b8c7542002-04-14 20:12:41 +00002356 if (c == '\r') {
Tim Peters058b1412002-04-21 07:29:14 +00002357 /* Save as LF and set flag to skip next LF. */
Jack Jansen7b8c7542002-04-14 20:12:41 +00002358 *dst++ = '\n';
2359 skipnextlf = 1;
Tim Peters058b1412002-04-21 07:29:14 +00002360 }
2361 else if (skipnextlf && c == '\n') {
2362 /* Skip LF, and remember we saw CR LF. */
Jack Jansen7b8c7542002-04-14 20:12:41 +00002363 skipnextlf = 0;
2364 newlinetypes |= NEWLINE_CRLF;
Tim Peterse1682a82002-04-21 18:15:20 +00002365 ++n;
Tim Peters058b1412002-04-21 07:29:14 +00002366 }
2367 else {
2368 /* Normal char to be stored in buffer. Also
2369 * update the newlinetypes flag if either this
2370 * is an LF or the previous char was a CR.
2371 */
Jack Jansen7b8c7542002-04-14 20:12:41 +00002372 if (c == '\n')
2373 newlinetypes |= NEWLINE_LF;
2374 else if (skipnextlf)
2375 newlinetypes |= NEWLINE_CR;
2376 *dst++ = c;
2377 skipnextlf = 0;
2378 }
2379 }
Tim Peters058b1412002-04-21 07:29:14 +00002380 if (shortread) {
2381 /* If this is EOF, update type flags. */
2382 if (skipnextlf && feof(stream))
2383 newlinetypes |= NEWLINE_CR;
2384 break;
2385 }
Jack Jansen7b8c7542002-04-14 20:12:41 +00002386 }
Tim Peters058b1412002-04-21 07:29:14 +00002387 f->f_newlinetypes = newlinetypes;
2388 f->f_skipnextlf = skipnextlf;
2389 return dst - buf;
Jack Jansen7b8c7542002-04-14 20:12:41 +00002390}