blob: ab59ddc2d303b463e98cfb86c8cac851bb576a18 [file] [log] [blame]
Benjamin Peterson9efcc4b2008-04-14 21:30:21 +00001"""The io module provides the Python interfaces to stream handling. The
Benjamin Peterson2c5f8282008-04-13 00:27:46 +00002builtin open function is defined in this module.
3
4At the top of the I/O hierarchy is the abstract base class IOBase. It
5defines the basic interface to a stream. Note, however, that there is no
6seperation between reading and writing to streams; implementations are
7allowed to throw an IOError if they do not support a given operation.
8
9Extending IOBase is RawIOBase which deals simply with the reading and
10writing of raw bytes to a stream. FileIO subclasses RawIOBase to provide
11an interface to OS files.
12
13BufferedIOBase deals with buffering on a raw byte stream (RawIOBase). Its
14subclasses, BufferedWriter, BufferedReader, and BufferedRWPair buffer
15streams that are readable, writable, and both respectively.
16BufferedRandom provides a buffered interface to random access
17streams. BytesIO is a simple stream of in-memory bytes.
18
19Another IOBase subclass, TextIOBase, deals with the encoding and decoding
20of streams into text. TextIOWrapper, which extends it, is a buffered text
21interface to a buffered raw stream (`BufferedIOBase`). Finally, StringIO
22is a in-memory stream for text.
23
24Argument names are not part of the specification, and only the arguments
25of open() are intended to be used as keyword arguments.
26
27data:
28
29DEFAULT_BUFFER_SIZE
30
31 An int containing the default buffer size used by the module's buffered
32 I/O classes. open() uses the file's blksize (as obtained by os.stat) if
33 possible.
34"""
35# New I/O library conforming to PEP 3116.
36
37# This is a prototype; hopefully eventually some of this will be
38# reimplemented in C.
39
40# XXX edge cases when switching between reading/writing
41# XXX need to support 1 meaning line-buffered
42# XXX whenever an argument is None, use the default value
43# XXX read/write ops should check readable/writable
44# XXX buffered readinto should work with arbitrary buffer objects
45# XXX use incremental encoder for text output, at least for UTF-16 and UTF-8-SIG
46# XXX check writable, readable and seekable in appropriate places
47
Guido van Rossum28524c72007-02-27 05:47:44 +000048
Guido van Rossum68bbcd22007-02-27 17:19:33 +000049__author__ = ("Guido van Rossum <guido@python.org>, "
Guido van Rossum78892e42007-04-06 17:31:18 +000050 "Mike Verdone <mike.verdone@gmail.com>, "
51 "Mark Russell <mark.russell@zen.co.uk>")
Guido van Rossum28524c72007-02-27 05:47:44 +000052
Guido van Rossum141f7672007-04-10 00:22:16 +000053__all__ = ["BlockingIOError", "open", "IOBase", "RawIOBase", "FileIO",
Guido van Rossum5abbf752007-08-27 17:39:33 +000054 "BytesIO", "StringIO", "BufferedIOBase",
Guido van Rossum01a27522007-03-07 01:00:12 +000055 "BufferedReader", "BufferedWriter", "BufferedRWPair",
Guido van Rossum141f7672007-04-10 00:22:16 +000056 "BufferedRandom", "TextIOBase", "TextIOWrapper"]
Guido van Rossum28524c72007-02-27 05:47:44 +000057
58import os
Guido van Rossumb7f136e2007-08-22 18:14:10 +000059import abc
Guido van Rossum78892e42007-04-06 17:31:18 +000060import sys
61import codecs
Guido van Rossum141f7672007-04-10 00:22:16 +000062import _fileio
Guido van Rossum78892e42007-04-06 17:31:18 +000063import warnings
Guido van Rossum28524c72007-02-27 05:47:44 +000064
Guido van Rossum5abbf752007-08-27 17:39:33 +000065# open() uses st_blksize whenever we can
Guido van Rossum4f0db6e2007-04-08 23:59:06 +000066DEFAULT_BUFFER_SIZE = 8 * 1024 # bytes
Guido van Rossum01a27522007-03-07 01:00:12 +000067
68
Guido van Rossum141f7672007-04-10 00:22:16 +000069class BlockingIOError(IOError):
Guido van Rossum78892e42007-04-06 17:31:18 +000070
Guido van Rossum141f7672007-04-10 00:22:16 +000071 """Exception raised when I/O would block on a non-blocking I/O stream."""
72
73 def __init__(self, errno, strerror, characters_written=0):
Guido van Rossum01a27522007-03-07 01:00:12 +000074 IOError.__init__(self, errno, strerror)
75 self.characters_written = characters_written
76
Guido van Rossum68bbcd22007-02-27 17:19:33 +000077
Guido van Rossume7fc50f2007-12-03 22:54:21 +000078def open(file, mode="r", buffering=None, encoding=None, errors=None,
79 newline=None, closefd=True):
Christian Heimes5d8da202008-05-06 13:58:24 +000080
81 r"""Open file and return a stream. If the file cannot be opened, an IOError is
82 raised.
Guido van Rossum17e43e52007-02-27 15:45:13 +000083
Benjamin Peterson2c5f8282008-04-13 00:27:46 +000084 file is either a string giving the name (and the path if the file
85 isn't in the current working directory) of the file to be opened or an
86 integer file descriptor of the file to be wrapped. (If a file
87 descriptor is given, it is closed when the returned I/O object is
88 closed, unless closefd is set to False.)
Guido van Rossum8358db22007-08-18 21:39:55 +000089
Benjamin Peterson2c5f8282008-04-13 00:27:46 +000090 mode is an optional string that specifies the mode in which the file
91 is opened. It defaults to 'r' which means open for reading in text
92 mode. Other common values are 'w' for writing (truncating the file if
93 it already exists), and 'a' for appending (which on some Unix systems,
94 means that all writes append to the end of the file regardless of the
95 current seek position). In text mode, if encoding is not specified the
96 encoding used is platform dependent. (For reading and writing raw
97 bytes use binary mode and leave encoding unspecified.) The available
98 modes are:
Guido van Rossum8358db22007-08-18 21:39:55 +000099
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000100 ========= ===============================================================
101 Character Meaning
102 --------- ---------------------------------------------------------------
103 'r' open for reading (default)
104 'w' open for writing, truncating the file first
105 'a' open for writing, appending to the end of the file if it exists
106 'b' binary mode
107 't' text mode (default)
108 '+' open a disk file for updating (reading and writing)
109 'U' universal newline mode (for backwards compatibility; unneeded
110 for new code)
111 ========= ===============================================================
Guido van Rossum17e43e52007-02-27 15:45:13 +0000112
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000113 The default mode is 'rt' (open for reading text). For binary random
114 access, the mode 'w+b' opens and truncates the file to 0 bytes, while
115 'r+b' opens the file without truncation.
Guido van Rossum2dced8b2007-10-30 17:27:30 +0000116
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000117 Python distinguishes between files opened in binary and text modes,
118 even when the underlying operating system doesn't. Files opened in
119 binary mode (appending 'b' to the mode argument) return contents as
120 bytes objects without any decoding. In text mode (the default, or when
121 't' is appended to the mode argument), the contents of the file are
122 returned as strings, the bytes having been first decoded using a
123 platform-dependent encoding or using the specified encoding if given.
Guido van Rossum4f0db6e2007-04-08 23:59:06 +0000124
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000125 buffering is an optional integer used to set the buffering policy. By
126 default full buffering is on. Pass 0 to switch buffering off (only
127 allowed in binary mode), 1 to set line buffering, and an integer > 1
128 for full buffering.
Guido van Rossum17e43e52007-02-27 15:45:13 +0000129
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000130 encoding is the name of the encoding used to decode or encode the
131 file. This should only be used in text mode. The default encoding is
132 platform dependent, but any encoding supported by Python can be
133 passed. See the codecs module for the list of supported encodings.
Guido van Rossum17e43e52007-02-27 15:45:13 +0000134
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000135 errors is an optional string that specifies how encoding errors are to
136 be handled---this argument should not be used in binary mode. Pass
137 'strict' to raise a ValueError exception if there is an encoding error
138 (the default of None has the same effect), or pass 'ignore' to ignore
139 errors. (Note that ignoring encoding errors can lead to data loss.)
140 See the documentation for codecs.register for a list of the permitted
141 encoding error strings.
142
143 newline controls how universal newlines works (it only applies to text
144 mode). It can be None, '', '\n', '\r', and '\r\n'. It works as
145 follows:
146
147 * On input, if newline is None, universal newlines mode is
148 enabled. Lines in the input can end in '\n', '\r', or '\r\n', and
149 these are translated into '\n' before being returned to the
150 caller. If it is '', universal newline mode is enabled, but line
151 endings are returned to the caller untranslated. If it has any of
152 the other legal values, input lines are only terminated by the given
153 string, and the line ending is returned to the caller untranslated.
154
155 * On output, if newline is None, any '\n' characters written are
156 translated to the system default line separator, os.linesep. If
157 newline is '', no translation takes place. If newline is any of the
158 other legal values, any '\n' characters written are translated to
159 the given string.
160
161 If closefd is False, the underlying file descriptor will be kept open
162 when the file is closed. This does not work when a file name is given
163 and must be True in that case.
164
165 open() returns a file object whose type depends on the mode, and
166 through which the standard file operations such as reading and writing
167 are performed. When open() is used to open a file in a text mode ('w',
168 'r', 'wt', 'rt', etc.), it returns a TextIOWrapper. When used to open
169 a file in a binary mode, the returned class varies: in read binary
170 mode, it returns a BufferedReader; in write binary and append binary
171 modes, it returns a BufferedWriter, and in read/write mode, it returns
172 a BufferedRandom.
173
174 It is also possible to use a string or bytearray as a file for both
175 reading and writing. For strings StringIO can be used like a file
176 opened in a text mode, and for bytes a BytesIO can be used like a file
177 opened in a binary mode.
Guido van Rossum17e43e52007-02-27 15:45:13 +0000178 """
Guido van Rossum3172c5d2007-10-16 18:12:55 +0000179 if not isinstance(file, (str, int)):
Guido van Rossum5abbf752007-08-27 17:39:33 +0000180 raise TypeError("invalid file: %r" % file)
Guido van Rossum3172c5d2007-10-16 18:12:55 +0000181 if not isinstance(mode, str):
Guido van Rossum5abbf752007-08-27 17:39:33 +0000182 raise TypeError("invalid mode: %r" % mode)
183 if buffering is not None and not isinstance(buffering, int):
184 raise TypeError("invalid buffering: %r" % buffering)
Guido van Rossum3172c5d2007-10-16 18:12:55 +0000185 if encoding is not None and not isinstance(encoding, str):
Guido van Rossum5abbf752007-08-27 17:39:33 +0000186 raise TypeError("invalid encoding: %r" % encoding)
Guido van Rossume7fc50f2007-12-03 22:54:21 +0000187 if errors is not None and not isinstance(errors, str):
188 raise TypeError("invalid errors: %r" % errors)
Guido van Rossum28524c72007-02-27 05:47:44 +0000189 modes = set(mode)
Guido van Rossum9be55972007-04-07 02:59:27 +0000190 if modes - set("arwb+tU") or len(mode) > len(modes):
Guido van Rossum28524c72007-02-27 05:47:44 +0000191 raise ValueError("invalid mode: %r" % mode)
192 reading = "r" in modes
Guido van Rossum17e43e52007-02-27 15:45:13 +0000193 writing = "w" in modes
Guido van Rossum28524c72007-02-27 05:47:44 +0000194 appending = "a" in modes
195 updating = "+" in modes
Guido van Rossum17e43e52007-02-27 15:45:13 +0000196 text = "t" in modes
197 binary = "b" in modes
Guido van Rossum7165cb12007-07-10 06:54:34 +0000198 if "U" in modes:
199 if writing or appending:
200 raise ValueError("can't use U and writing mode at once")
Guido van Rossum9be55972007-04-07 02:59:27 +0000201 reading = True
Guido van Rossum28524c72007-02-27 05:47:44 +0000202 if text and binary:
203 raise ValueError("can't have text and binary mode at once")
204 if reading + writing + appending > 1:
205 raise ValueError("can't have read/write/append mode at once")
206 if not (reading or writing or appending):
207 raise ValueError("must have exactly one of read/write/append mode")
208 if binary and encoding is not None:
Guido van Rossum9b76da62007-04-11 01:09:03 +0000209 raise ValueError("binary mode doesn't take an encoding argument")
Guido van Rossume7fc50f2007-12-03 22:54:21 +0000210 if binary and errors is not None:
211 raise ValueError("binary mode doesn't take an errors argument")
Guido van Rossum9b76da62007-04-11 01:09:03 +0000212 if binary and newline is not None:
213 raise ValueError("binary mode doesn't take a newline argument")
Guido van Rossum4f0db6e2007-04-08 23:59:06 +0000214 raw = FileIO(file,
Guido van Rossum28524c72007-02-27 05:47:44 +0000215 (reading and "r" or "") +
216 (writing and "w" or "") +
217 (appending and "a" or "") +
Guido van Rossum2dced8b2007-10-30 17:27:30 +0000218 (updating and "+" or ""),
219 closefd)
Guido van Rossum28524c72007-02-27 05:47:44 +0000220 if buffering is None:
Guido van Rossumc2f93dc2007-05-24 00:50:02 +0000221 buffering = -1
Guido van Rossumf64db9f2007-12-06 01:04:26 +0000222 line_buffering = False
223 if buffering == 1 or buffering < 0 and raw.isatty():
224 buffering = -1
225 line_buffering = True
Guido van Rossumc2f93dc2007-05-24 00:50:02 +0000226 if buffering < 0:
Guido van Rossum68bbcd22007-02-27 17:19:33 +0000227 buffering = DEFAULT_BUFFER_SIZE
Guido van Rossum17e43e52007-02-27 15:45:13 +0000228 try:
229 bs = os.fstat(raw.fileno()).st_blksize
230 except (os.error, AttributeError):
Guido van Rossumbb09b212007-03-18 03:36:28 +0000231 pass
232 else:
Guido van Rossum17e43e52007-02-27 15:45:13 +0000233 if bs > 1:
234 buffering = bs
Guido van Rossum28524c72007-02-27 05:47:44 +0000235 if buffering < 0:
236 raise ValueError("invalid buffering size")
237 if buffering == 0:
238 if binary:
Guido van Rossum13633bb2007-04-13 18:42:35 +0000239 raw._name = file
240 raw._mode = mode
Guido van Rossum28524c72007-02-27 05:47:44 +0000241 return raw
242 raise ValueError("can't have unbuffered text I/O")
243 if updating:
244 buffer = BufferedRandom(raw, buffering)
Guido van Rossum17e43e52007-02-27 15:45:13 +0000245 elif writing or appending:
Guido van Rossum28524c72007-02-27 05:47:44 +0000246 buffer = BufferedWriter(raw, buffering)
Guido van Rossum5abbf752007-08-27 17:39:33 +0000247 elif reading:
Guido van Rossum28524c72007-02-27 05:47:44 +0000248 buffer = BufferedReader(raw, buffering)
Guido van Rossum5abbf752007-08-27 17:39:33 +0000249 else:
250 raise ValueError("unknown mode: %r" % mode)
Guido van Rossum28524c72007-02-27 05:47:44 +0000251 if binary:
Guido van Rossum13633bb2007-04-13 18:42:35 +0000252 buffer.name = file
253 buffer.mode = mode
Guido van Rossum28524c72007-02-27 05:47:44 +0000254 return buffer
Guido van Rossumf64db9f2007-12-06 01:04:26 +0000255 text = TextIOWrapper(buffer, encoding, errors, newline, line_buffering)
Guido van Rossum13633bb2007-04-13 18:42:35 +0000256 text.name = file
257 text.mode = mode
258 return text
Guido van Rossum28524c72007-02-27 05:47:44 +0000259
Christian Heimesa33eb062007-12-08 17:47:40 +0000260class _DocDescriptor:
261 """Helper for builtins.open.__doc__
262 """
263 def __get__(self, obj, typ):
264 return (
265 "open(file, mode='r', buffering=None, encoding=None, "
266 "errors=None, newline=None, closefd=True)\n\n" +
267 open.__doc__)
Guido van Rossum28524c72007-02-27 05:47:44 +0000268
Guido van Rossumce3a72a2007-10-19 23:16:50 +0000269class OpenWrapper:
Georg Brandl1a3284e2007-12-02 09:40:06 +0000270 """Wrapper for builtins.open
Guido van Rossumce3a72a2007-10-19 23:16:50 +0000271
272 Trick so that open won't become a bound method when stored
Georg Brandl0a7ac7d2008-05-26 10:29:35 +0000273 as a class variable (as dbm.dumb does).
Guido van Rossumce3a72a2007-10-19 23:16:50 +0000274
275 See initstdio() in Python/pythonrun.c.
276 """
Christian Heimesa33eb062007-12-08 17:47:40 +0000277 __doc__ = _DocDescriptor()
278
Guido van Rossumce3a72a2007-10-19 23:16:50 +0000279 def __new__(cls, *args, **kwargs):
280 return open(*args, **kwargs)
281
282
Guido van Rossum4b5386f2007-07-10 09:12:49 +0000283class UnsupportedOperation(ValueError, IOError):
284 pass
285
286
Guido van Rossumb7f136e2007-08-22 18:14:10 +0000287class IOBase(metaclass=abc.ABCMeta):
Guido van Rossum28524c72007-02-27 05:47:44 +0000288
Benjamin Peterson9efcc4b2008-04-14 21:30:21 +0000289 """The abstract base class for all I/O classes, acting on streams of
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000290 bytes. There is no public constructor.
Guido van Rossum17e43e52007-02-27 15:45:13 +0000291
Guido van Rossum141f7672007-04-10 00:22:16 +0000292 This class provides dummy implementations for many methods that
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000293 derived classes can override selectively; the default implementations
294 represent a file that cannot be read, written or seeked.
Guido van Rossum17e43e52007-02-27 15:45:13 +0000295
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000296 Even though IOBase does not declare read, readinto, or write because
297 their signatures will vary, implementations and clients should
298 consider those methods part of the interface. Also, implementations
299 may raise a IOError when operations they do not support are called.
Guido van Rossum53807da2007-04-10 19:01:47 +0000300
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000301 The basic type used for binary data read from or written to a file is
302 bytes. bytearrays are accepted too, and in some cases (such as
303 readinto) needed. Text I/O classes work with str data.
304
305 Note that calling any method (even inquiries) on a closed stream is
Benjamin Peterson9a89e962008-04-06 16:47:13 +0000306 undefined. Implementations may raise IOError in this case.
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000307
308 IOBase (and its subclasses) support the iterator protocol, meaning
309 that an IOBase object can be iterated over yielding the lines in a
310 stream.
311
312 IOBase also supports the :keyword:`with` statement. In this example,
313 fp is closed after the suite of the with statment is complete:
314
315 with open('spam.txt', 'r') as fp:
316 fp.write('Spam and eggs!')
Guido van Rossum17e43e52007-02-27 15:45:13 +0000317 """
318
Guido van Rossum141f7672007-04-10 00:22:16 +0000319 ### Internal ###
320
321 def _unsupported(self, name: str) -> IOError:
322 """Internal: raise an exception for unsupported operations."""
Guido van Rossum4b5386f2007-07-10 09:12:49 +0000323 raise UnsupportedOperation("%s.%s() not supported" %
324 (self.__class__.__name__, name))
Guido van Rossum4f0db6e2007-04-08 23:59:06 +0000325
Guido van Rossum141f7672007-04-10 00:22:16 +0000326 ### Positioning ###
327
Guido van Rossum53807da2007-04-10 19:01:47 +0000328 def seek(self, pos: int, whence: int = 0) -> int:
Christian Heimes5d8da202008-05-06 13:58:24 +0000329 """Change stream position.
Guido van Rossum141f7672007-04-10 00:22:16 +0000330
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000331 Change the stream position to byte offset offset. offset is
332 interpreted relative to the position indicated by whence. Values
333 for whence are:
334
335 * 0 -- start of stream (the default); offset should be zero or positive
336 * 1 -- current stream position; offset may be negative
337 * 2 -- end of stream; offset is usually negative
338
339 Return the new absolute position.
Guido van Rossum141f7672007-04-10 00:22:16 +0000340 """
341 self._unsupported("seek")
342
343 def tell(self) -> int:
Christian Heimes5d8da202008-05-06 13:58:24 +0000344 """Return current stream position."""
Guido van Rossum53807da2007-04-10 19:01:47 +0000345 return self.seek(0, 1)
Guido van Rossum141f7672007-04-10 00:22:16 +0000346
Guido van Rossum87429772007-04-10 21:06:59 +0000347 def truncate(self, pos: int = None) -> int:
Christian Heimes5d8da202008-05-06 13:58:24 +0000348 """Truncate file to size bytes.
Guido van Rossum141f7672007-04-10 00:22:16 +0000349
Christian Heimes5d8da202008-05-06 13:58:24 +0000350 Size defaults to the current IO position as reported by tell(). Return
351 the new size.
Guido van Rossum141f7672007-04-10 00:22:16 +0000352 """
353 self._unsupported("truncate")
354
355 ### Flush and close ###
356
357 def flush(self) -> None:
Christian Heimes5d8da202008-05-06 13:58:24 +0000358 """Flush write buffers, if applicable.
Guido van Rossum141f7672007-04-10 00:22:16 +0000359
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000360 This is not implemented for read-only and non-blocking streams.
Guido van Rossum141f7672007-04-10 00:22:16 +0000361 """
Guido van Rossumd4103952007-04-12 05:44:49 +0000362 # XXX Should this return the number of bytes written???
Guido van Rossum141f7672007-04-10 00:22:16 +0000363
364 __closed = False
365
366 def close(self) -> None:
Christian Heimes5d8da202008-05-06 13:58:24 +0000367 """Flush and close the IO object.
Guido van Rossum141f7672007-04-10 00:22:16 +0000368
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000369 This method has no effect if the file is already closed.
Guido van Rossum141f7672007-04-10 00:22:16 +0000370 """
371 if not self.__closed:
Guido van Rossum469734b2007-07-10 12:00:45 +0000372 try:
373 self.flush()
Guido van Rossum33e7a8e2007-07-22 20:38:07 +0000374 except IOError:
375 pass # If flush() fails, just give up
376 self.__closed = True
Guido van Rossum141f7672007-04-10 00:22:16 +0000377
378 def __del__(self) -> None:
379 """Destructor. Calls close()."""
380 # The try/except block is in case this is called at program
381 # exit time, when it's possible that globals have already been
382 # deleted, and then the close() call might fail. Since
383 # there's nothing we can do about such failures and they annoy
384 # the end users, we suppress the traceback.
385 try:
386 self.close()
387 except:
388 pass
389
390 ### Inquiries ###
391
392 def seekable(self) -> bool:
Christian Heimes5d8da202008-05-06 13:58:24 +0000393 """Return whether object supports random access.
Guido van Rossum141f7672007-04-10 00:22:16 +0000394
395 If False, seek(), tell() and truncate() will raise IOError.
396 This method may need to do a test seek().
397 """
398 return False
399
Guido van Rossum5abbf752007-08-27 17:39:33 +0000400 def _checkSeekable(self, msg=None):
401 """Internal: raise an IOError if file is not seekable
402 """
403 if not self.seekable():
404 raise IOError("File or stream is not seekable."
405 if msg is None else msg)
406
407
Guido van Rossum141f7672007-04-10 00:22:16 +0000408 def readable(self) -> bool:
Christian Heimes5d8da202008-05-06 13:58:24 +0000409 """Return whether object was opened for reading.
Guido van Rossum141f7672007-04-10 00:22:16 +0000410
411 If False, read() will raise IOError.
412 """
413 return False
414
Guido van Rossum5abbf752007-08-27 17:39:33 +0000415 def _checkReadable(self, msg=None):
416 """Internal: raise an IOError if file is not readable
417 """
418 if not self.readable():
419 raise IOError("File or stream is not readable."
420 if msg is None else msg)
421
Guido van Rossum141f7672007-04-10 00:22:16 +0000422 def writable(self) -> bool:
Christian Heimes5d8da202008-05-06 13:58:24 +0000423 """Return whether object was opened for writing.
Guido van Rossum141f7672007-04-10 00:22:16 +0000424
425 If False, write() and truncate() will raise IOError.
426 """
427 return False
428
Guido van Rossum5abbf752007-08-27 17:39:33 +0000429 def _checkWritable(self, msg=None):
430 """Internal: raise an IOError if file is not writable
431 """
432 if not self.writable():
433 raise IOError("File or stream is not writable."
434 if msg is None else msg)
435
Guido van Rossum141f7672007-04-10 00:22:16 +0000436 @property
437 def closed(self):
438 """closed: bool. True iff the file has been closed.
439
440 For backwards compatibility, this is a property, not a predicate.
441 """
442 return self.__closed
443
Guido van Rossum5abbf752007-08-27 17:39:33 +0000444 def _checkClosed(self, msg=None):
445 """Internal: raise an ValueError if file is closed
446 """
447 if self.closed:
448 raise ValueError("I/O operation on closed file."
449 if msg is None else msg)
450
Guido van Rossum141f7672007-04-10 00:22:16 +0000451 ### Context manager ###
452
453 def __enter__(self) -> "IOBase": # That's a forward reference
454 """Context management protocol. Returns self."""
Christian Heimes3ecfea712008-02-09 20:51:34 +0000455 self._checkClosed()
Guido van Rossum141f7672007-04-10 00:22:16 +0000456 return self
457
458 def __exit__(self, *args) -> None:
459 """Context management protocol. Calls close()"""
460 self.close()
461
462 ### Lower-level APIs ###
463
464 # XXX Should these be present even if unimplemented?
465
466 def fileno(self) -> int:
Christian Heimes5d8da202008-05-06 13:58:24 +0000467 """Returns underlying file descriptor if one exists.
Guido van Rossum141f7672007-04-10 00:22:16 +0000468
Christian Heimes5d8da202008-05-06 13:58:24 +0000469 An IOError is raised if the IO object does not use a file descriptor.
Guido van Rossum141f7672007-04-10 00:22:16 +0000470 """
471 self._unsupported("fileno")
472
473 def isatty(self) -> bool:
Christian Heimes5d8da202008-05-06 13:58:24 +0000474 """Return whether this is an 'interactive' stream.
475
476 Return False if it can't be determined.
Guido van Rossum141f7672007-04-10 00:22:16 +0000477 """
Guido van Rossum5abbf752007-08-27 17:39:33 +0000478 self._checkClosed()
Guido van Rossum141f7672007-04-10 00:22:16 +0000479 return False
480
Guido van Rossum7165cb12007-07-10 06:54:34 +0000481 ### Readline[s] and writelines ###
Guido van Rossum7d0a8262007-05-21 23:13:11 +0000482
Guido van Rossum48fc58a2007-06-07 23:45:37 +0000483 def readline(self, limit: int = -1) -> bytes:
Christian Heimes5d8da202008-05-06 13:58:24 +0000484 r"""Read and return a line from the stream.
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000485
486 If limit is specified, at most limit bytes will be read.
487
488 The line terminator is always b'\n' for binary files; for text
489 files, the newlines argument to open can be used to select the line
490 terminator(s) recognized.
491 """
492 # For backwards compatibility, a (slowish) readline().
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000493 self._checkClosed()
Guido van Rossum2bf71382007-06-08 00:07:57 +0000494 if hasattr(self, "peek"):
495 def nreadahead():
Ka-Ping Yee7a0d3982008-03-17 17:34:48 +0000496 readahead = self.peek(1)
Guido van Rossum2bf71382007-06-08 00:07:57 +0000497 if not readahead:
498 return 1
499 n = (readahead.find(b"\n") + 1) or len(readahead)
500 if limit >= 0:
501 n = min(n, limit)
502 return n
503 else:
504 def nreadahead():
505 return 1
Guido van Rossum48fc58a2007-06-07 23:45:37 +0000506 if limit is None:
507 limit = -1
Guido van Rossum254348e2007-11-21 19:29:53 +0000508 res = bytearray()
Guido van Rossum48fc58a2007-06-07 23:45:37 +0000509 while limit < 0 or len(res) < limit:
Guido van Rossum2bf71382007-06-08 00:07:57 +0000510 b = self.read(nreadahead())
Guido van Rossum7d0a8262007-05-21 23:13:11 +0000511 if not b:
512 break
513 res += b
Guido van Rossum48fc58a2007-06-07 23:45:37 +0000514 if res.endswith(b"\n"):
Guido van Rossum7d0a8262007-05-21 23:13:11 +0000515 break
Guido van Rossum98297ee2007-11-06 21:34:58 +0000516 return bytes(res)
Guido van Rossum7d0a8262007-05-21 23:13:11 +0000517
Guido van Rossum7165cb12007-07-10 06:54:34 +0000518 def __iter__(self):
Guido van Rossum5abbf752007-08-27 17:39:33 +0000519 self._checkClosed()
Guido van Rossum7165cb12007-07-10 06:54:34 +0000520 return self
521
522 def __next__(self):
523 line = self.readline()
524 if not line:
525 raise StopIteration
526 return line
527
528 def readlines(self, hint=None):
Christian Heimes5d8da202008-05-06 13:58:24 +0000529 """Return a list of lines from the stream.
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000530
531 hint can be specified to control the number of lines read: no more
532 lines will be read if the total size (in bytes/characters) of all
533 lines so far exceeds hint.
534 """
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000535 if hint is None or hint <= 0:
Guido van Rossum7165cb12007-07-10 06:54:34 +0000536 return list(self)
537 n = 0
538 lines = []
539 for line in self:
540 lines.append(line)
541 n += len(line)
542 if n >= hint:
543 break
544 return lines
545
546 def writelines(self, lines):
Guido van Rossum5abbf752007-08-27 17:39:33 +0000547 self._checkClosed()
Guido van Rossum7165cb12007-07-10 06:54:34 +0000548 for line in lines:
549 self.write(line)
550
Guido van Rossum141f7672007-04-10 00:22:16 +0000551
552class RawIOBase(IOBase):
553
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000554 """Base class for raw binary I/O."""
Guido van Rossum141f7672007-04-10 00:22:16 +0000555
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000556 # The read() method is implemented by calling readinto(); derived
557 # classes that want to support read() only need to implement
558 # readinto() as a primitive operation. In general, readinto() can be
559 # more efficient than read().
Guido van Rossum141f7672007-04-10 00:22:16 +0000560
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000561 # (It would be tempting to also provide an implementation of
562 # readinto() in terms of read(), in case the latter is a more suitable
563 # primitive operation, but that would lead to nasty recursion in case
564 # a subclass doesn't implement either.)
Guido van Rossum141f7672007-04-10 00:22:16 +0000565
Guido van Rossum7165cb12007-07-10 06:54:34 +0000566 def read(self, n: int = -1) -> bytes:
Christian Heimes5d8da202008-05-06 13:58:24 +0000567 """Read and return up to n bytes.
Guido van Rossum01a27522007-03-07 01:00:12 +0000568
Georg Brandlf91197c2008-04-09 07:33:01 +0000569 Returns an empty bytes object on EOF, or None if the object is
Guido van Rossum01a27522007-03-07 01:00:12 +0000570 set not to block and has no data to read.
571 """
Guido van Rossum7165cb12007-07-10 06:54:34 +0000572 if n is None:
573 n = -1
574 if n < 0:
575 return self.readall()
Guido van Rossum254348e2007-11-21 19:29:53 +0000576 b = bytearray(n.__index__())
Guido van Rossum00efead2007-03-07 05:23:25 +0000577 n = self.readinto(b)
578 del b[n:]
Guido van Rossum98297ee2007-11-06 21:34:58 +0000579 return bytes(b)
Guido van Rossum28524c72007-02-27 05:47:44 +0000580
Guido van Rossum7165cb12007-07-10 06:54:34 +0000581 def readall(self):
Christian Heimes5d8da202008-05-06 13:58:24 +0000582 """Read until EOF, using multiple read() call."""
Guido van Rossum254348e2007-11-21 19:29:53 +0000583 res = bytearray()
Guido van Rossum7165cb12007-07-10 06:54:34 +0000584 while True:
585 data = self.read(DEFAULT_BUFFER_SIZE)
586 if not data:
587 break
588 res += data
Guido van Rossum98297ee2007-11-06 21:34:58 +0000589 return bytes(res)
Guido van Rossum7165cb12007-07-10 06:54:34 +0000590
Benjamin Petersonca2b0152008-04-07 22:27:34 +0000591 def readinto(self, b: bytearray) -> int:
Christian Heimes5d8da202008-05-06 13:58:24 +0000592 """Read up to len(b) bytes into b.
Guido van Rossum78892e42007-04-06 17:31:18 +0000593
594 Returns number of bytes read (0 for EOF), or None if the object
595 is set not to block as has no data to read.
596 """
Guido van Rossum4f0db6e2007-04-08 23:59:06 +0000597 self._unsupported("readinto")
Guido van Rossum28524c72007-02-27 05:47:44 +0000598
Guido van Rossum141f7672007-04-10 00:22:16 +0000599 def write(self, b: bytes) -> int:
Christian Heimes5d8da202008-05-06 13:58:24 +0000600 """Write the given buffer to the IO stream.
Guido van Rossum01a27522007-03-07 01:00:12 +0000601
Guido van Rossum78892e42007-04-06 17:31:18 +0000602 Returns the number of bytes written, which may be less than len(b).
Guido van Rossum01a27522007-03-07 01:00:12 +0000603 """
Guido van Rossum4f0db6e2007-04-08 23:59:06 +0000604 self._unsupported("write")
Guido van Rossum28524c72007-02-27 05:47:44 +0000605
Guido van Rossum78892e42007-04-06 17:31:18 +0000606
Guido van Rossum141f7672007-04-10 00:22:16 +0000607class FileIO(_fileio._FileIO, RawIOBase):
Guido van Rossum28524c72007-02-27 05:47:44 +0000608
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000609 """Raw I/O implementation for OS files."""
Guido van Rossum28524c72007-02-27 05:47:44 +0000610
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000611 # This multiply inherits from _FileIO and RawIOBase to make
612 # isinstance(io.FileIO(), io.RawIOBase) return True without requiring
613 # that _fileio._FileIO inherits from io.RawIOBase (which would be hard
614 # to do since _fileio.c is written in C).
Guido van Rossuma9e20242007-03-08 00:43:48 +0000615
Guido van Rossum87429772007-04-10 21:06:59 +0000616 def close(self):
617 _fileio._FileIO.close(self)
618 RawIOBase.close(self)
619
Guido van Rossum13633bb2007-04-13 18:42:35 +0000620 @property
621 def name(self):
622 return self._name
623
Georg Brandlf91197c2008-04-09 07:33:01 +0000624 # XXX(gb): _FileIO already has a mode property
Guido van Rossum13633bb2007-04-13 18:42:35 +0000625 @property
626 def mode(self):
627 return self._mode
628
Guido van Rossuma9e20242007-03-08 00:43:48 +0000629
Guido van Rossumcce92b22007-04-10 14:41:39 +0000630class BufferedIOBase(IOBase):
Guido van Rossum141f7672007-04-10 00:22:16 +0000631
632 """Base class for buffered IO objects.
633
634 The main difference with RawIOBase is that the read() method
635 supports omitting the size argument, and does not have a default
636 implementation that defers to readinto().
637
638 In addition, read(), readinto() and write() may raise
639 BlockingIOError if the underlying raw stream is in non-blocking
640 mode and not ready; unlike their raw counterparts, they will never
641 return None.
642
643 A typical implementation should not inherit from a RawIOBase
644 implementation, but wrap one.
645 """
646
Guido van Rossumc2f93dc2007-05-24 00:50:02 +0000647 def read(self, n: int = None) -> bytes:
Christian Heimes5d8da202008-05-06 13:58:24 +0000648 """Read and return up to n bytes.
Guido van Rossum141f7672007-04-10 00:22:16 +0000649
Guido van Rossum024da5c2007-05-17 23:59:11 +0000650 If the argument is omitted, None, or negative, reads and
651 returns all data until EOF.
Guido van Rossum141f7672007-04-10 00:22:16 +0000652
653 If the argument is positive, and the underlying raw stream is
654 not 'interactive', multiple raw reads may be issued to satisfy
655 the byte count (unless EOF is reached first). But for
656 interactive raw streams (XXX and for pipes?), at most one raw
657 read will be issued, and a short result does not imply that
658 EOF is imminent.
659
660 Returns an empty bytes array on EOF.
661
662 Raises BlockingIOError if the underlying raw stream has no
663 data at the moment.
664 """
665 self._unsupported("read")
666
Benjamin Petersonca2b0152008-04-07 22:27:34 +0000667 def readinto(self, b: bytearray) -> int:
Christian Heimes5d8da202008-05-06 13:58:24 +0000668 """Read up to len(b) bytes into b.
Guido van Rossum141f7672007-04-10 00:22:16 +0000669
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000670 Like read(), this may issue multiple reads to the underlying raw
671 stream, unless the latter is 'interactive'.
Guido van Rossum141f7672007-04-10 00:22:16 +0000672
673 Returns the number of bytes read (0 for EOF).
674
675 Raises BlockingIOError if the underlying raw stream has no
676 data at the moment.
677 """
Guido van Rossumd4103952007-04-12 05:44:49 +0000678 # XXX This ought to work with anything that supports the buffer API
Guido van Rossum87429772007-04-10 21:06:59 +0000679 data = self.read(len(b))
680 n = len(data)
Guido van Rossum7165cb12007-07-10 06:54:34 +0000681 try:
682 b[:n] = data
683 except TypeError as err:
684 import array
685 if not isinstance(b, array.array):
686 raise err
687 b[:n] = array.array('b', data)
Guido van Rossum87429772007-04-10 21:06:59 +0000688 return n
Guido van Rossum141f7672007-04-10 00:22:16 +0000689
690 def write(self, b: bytes) -> int:
Christian Heimes5d8da202008-05-06 13:58:24 +0000691 """Write the given buffer to the IO stream.
Guido van Rossum141f7672007-04-10 00:22:16 +0000692
Christian Heimes5d8da202008-05-06 13:58:24 +0000693 Return the number of bytes written, which is never less than
Guido van Rossum141f7672007-04-10 00:22:16 +0000694 len(b).
695
696 Raises BlockingIOError if the buffer is full and the
697 underlying raw stream cannot accept more data at the moment.
698 """
699 self._unsupported("write")
700
701
702class _BufferedIOMixin(BufferedIOBase):
703
704 """A mixin implementation of BufferedIOBase with an underlying raw stream.
705
706 This passes most requests on to the underlying raw stream. It
707 does *not* provide implementations of read(), readinto() or
708 write().
709 """
710
711 def __init__(self, raw):
712 self.raw = raw
713
714 ### Positioning ###
715
716 def seek(self, pos, whence=0):
Guido van Rossum53807da2007-04-10 19:01:47 +0000717 return self.raw.seek(pos, whence)
Guido van Rossum141f7672007-04-10 00:22:16 +0000718
719 def tell(self):
720 return self.raw.tell()
721
722 def truncate(self, pos=None):
Guido van Rossum79b79ee2007-10-25 23:21:03 +0000723 # Flush the stream. We're mixing buffered I/O with lower-level I/O,
724 # and a flush may be necessary to synch both views of the current
725 # file state.
726 self.flush()
Guido van Rossum57233cb2007-10-26 17:19:33 +0000727
728 if pos is None:
729 pos = self.tell()
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000730 # XXX: Should seek() be used, instead of passing the position
731 # XXX directly to truncate?
Guido van Rossum57233cb2007-10-26 17:19:33 +0000732 return self.raw.truncate(pos)
Guido van Rossum141f7672007-04-10 00:22:16 +0000733
734 ### Flush and close ###
735
736 def flush(self):
737 self.raw.flush()
738
739 def close(self):
Guido van Rossum4b5386f2007-07-10 09:12:49 +0000740 if not self.closed:
Guido van Rossum33e7a8e2007-07-22 20:38:07 +0000741 try:
742 self.flush()
743 except IOError:
744 pass # If flush() fails, just give up
Guido van Rossum4b5386f2007-07-10 09:12:49 +0000745 self.raw.close()
Guido van Rossum141f7672007-04-10 00:22:16 +0000746
747 ### Inquiries ###
748
749 def seekable(self):
750 return self.raw.seekable()
751
752 def readable(self):
753 return self.raw.readable()
754
755 def writable(self):
756 return self.raw.writable()
757
758 @property
759 def closed(self):
760 return self.raw.closed
761
762 ### Lower-level APIs ###
763
764 def fileno(self):
765 return self.raw.fileno()
766
767 def isatty(self):
768 return self.raw.isatty()
769
770
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000771class _BytesIO(BufferedIOBase):
Guido van Rossum68bbcd22007-02-27 17:19:33 +0000772
Guido van Rossum024da5c2007-05-17 23:59:11 +0000773 """Buffered I/O implementation using an in-memory bytes buffer."""
Guido van Rossum28524c72007-02-27 05:47:44 +0000774
Guido van Rossum024da5c2007-05-17 23:59:11 +0000775 def __init__(self, initial_bytes=None):
Guido van Rossum254348e2007-11-21 19:29:53 +0000776 buf = bytearray()
Guido van Rossum024da5c2007-05-17 23:59:11 +0000777 if initial_bytes is not None:
Guido van Rossum98297ee2007-11-06 21:34:58 +0000778 buf += initial_bytes
779 self._buffer = buf
Guido van Rossum28524c72007-02-27 05:47:44 +0000780 self._pos = 0
Guido van Rossum28524c72007-02-27 05:47:44 +0000781
782 def getvalue(self):
Christian Heimes5d8da202008-05-06 13:58:24 +0000783 """Return the bytes value (contents) of the buffer
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000784 """
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000785 if self.closed:
786 raise ValueError("getvalue on closed file")
Guido van Rossum98297ee2007-11-06 21:34:58 +0000787 return bytes(self._buffer)
Guido van Rossum28524c72007-02-27 05:47:44 +0000788
Guido van Rossum024da5c2007-05-17 23:59:11 +0000789 def read(self, n=None):
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000790 if self.closed:
791 raise ValueError("read from closed file")
Guido van Rossum024da5c2007-05-17 23:59:11 +0000792 if n is None:
793 n = -1
Guido van Rossum141f7672007-04-10 00:22:16 +0000794 if n < 0:
Guido van Rossum68bbcd22007-02-27 17:19:33 +0000795 n = len(self._buffer)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000796 if len(self._buffer) <= self._pos:
Alexandre Vassalotti2e0419d2008-05-07 00:09:04 +0000797 return b""
Guido van Rossum28524c72007-02-27 05:47:44 +0000798 newpos = min(len(self._buffer), self._pos + n)
799 b = self._buffer[self._pos : newpos]
800 self._pos = newpos
Guido van Rossum98297ee2007-11-06 21:34:58 +0000801 return bytes(b)
Guido van Rossum28524c72007-02-27 05:47:44 +0000802
Guido van Rossum024da5c2007-05-17 23:59:11 +0000803 def read1(self, n):
Benjamin Peterson9efcc4b2008-04-14 21:30:21 +0000804 """This is the same as read.
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000805 """
Guido van Rossum024da5c2007-05-17 23:59:11 +0000806 return self.read(n)
807
Guido van Rossum28524c72007-02-27 05:47:44 +0000808 def write(self, b):
Guido van Rossum4b5386f2007-07-10 09:12:49 +0000809 if self.closed:
810 raise ValueError("write to closed file")
Guido van Rossuma74184e2007-08-29 04:05:57 +0000811 if isinstance(b, str):
812 raise TypeError("can't write str to binary stream")
Guido van Rossum28524c72007-02-27 05:47:44 +0000813 n = len(b)
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000814 if n == 0:
815 return 0
Alexandre Vassalotti5f8ced22008-05-16 00:03:33 +0000816 pos = self._pos
817 if pos > len(self._buffer):
Guido van Rossumb972a782007-07-21 00:25:15 +0000818 # Inserts null bytes between the current end of the file
819 # and the new write position.
Alexandre Vassalotti5f8ced22008-05-16 00:03:33 +0000820 padding = b'\x00' * (pos - len(self._buffer))
821 self._buffer += padding
822 self._buffer[pos:pos + n] = b
823 self._pos += n
Guido van Rossum28524c72007-02-27 05:47:44 +0000824 return n
825
826 def seek(self, pos, whence=0):
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000827 if self.closed:
828 raise ValueError("seek on closed file")
Christian Heimes3ab4f652007-11-09 01:27:29 +0000829 try:
830 pos = pos.__index__()
831 except AttributeError as err:
832 raise TypeError("an integer is required") from err
Guido van Rossum28524c72007-02-27 05:47:44 +0000833 if whence == 0:
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000834 if pos < 0:
835 raise ValueError("negative seek position %r" % (pos,))
Alexandre Vassalottif0c0ff62008-05-09 21:21:21 +0000836 self._pos = pos
Guido van Rossum28524c72007-02-27 05:47:44 +0000837 elif whence == 1:
838 self._pos = max(0, self._pos + pos)
839 elif whence == 2:
840 self._pos = max(0, len(self._buffer) + pos)
841 else:
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000842 raise ValueError("invalid whence value")
Guido van Rossum53807da2007-04-10 19:01:47 +0000843 return self._pos
Guido van Rossum28524c72007-02-27 05:47:44 +0000844
845 def tell(self):
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000846 if self.closed:
847 raise ValueError("tell on closed file")
Guido van Rossum28524c72007-02-27 05:47:44 +0000848 return self._pos
849
850 def truncate(self, pos=None):
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000851 if self.closed:
852 raise ValueError("truncate on closed file")
Guido van Rossum28524c72007-02-27 05:47:44 +0000853 if pos is None:
854 pos = self._pos
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000855 elif pos < 0:
856 raise ValueError("negative truncate position %r" % (pos,))
Guido van Rossum28524c72007-02-27 05:47:44 +0000857 del self._buffer[pos:]
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000858 return self.seek(pos)
Guido van Rossum28524c72007-02-27 05:47:44 +0000859
860 def readable(self):
861 return True
862
863 def writable(self):
864 return True
865
866 def seekable(self):
867 return True
Guido van Rossum68bbcd22007-02-27 17:19:33 +0000868
Alexandre Vassalotti77250f42008-05-06 19:48:38 +0000869# Use the faster implementation of BytesIO if available
870try:
871 import _bytesio
872
873 class BytesIO(_bytesio._BytesIO, BufferedIOBase):
874 __doc__ = _bytesio._BytesIO.__doc__
875
876except ImportError:
877 BytesIO = _BytesIO
878
Guido van Rossum68bbcd22007-02-27 17:19:33 +0000879
Guido van Rossum141f7672007-04-10 00:22:16 +0000880class BufferedReader(_BufferedIOMixin):
Guido van Rossum68bbcd22007-02-27 17:19:33 +0000881
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000882 """BufferedReader(raw[, buffer_size])
883
884 A buffer for a readable, sequential BaseRawIO object.
885
886 The constructor creates a BufferedReader for the given readable raw
887 stream and buffer_size. If buffer_size is omitted, DEFAULT_BUFFER_SIZE
888 is used.
889 """
Guido van Rossum68bbcd22007-02-27 17:19:33 +0000890
Guido van Rossum78892e42007-04-06 17:31:18 +0000891 def __init__(self, raw, buffer_size=DEFAULT_BUFFER_SIZE):
Guido van Rossum01a27522007-03-07 01:00:12 +0000892 """Create a new buffered reader using the given readable raw IO object.
Guido van Rossum68bbcd22007-02-27 17:19:33 +0000893 """
Guido van Rossum5abbf752007-08-27 17:39:33 +0000894 raw._checkReadable()
Guido van Rossum141f7672007-04-10 00:22:16 +0000895 _BufferedIOMixin.__init__(self, raw)
Guido van Rossum78892e42007-04-06 17:31:18 +0000896 self.buffer_size = buffer_size
Antoine Pitrouc66f9092008-07-28 19:46:11 +0000897 self._reset_read_buf()
898
899 def _reset_read_buf(self):
900 self._read_buf = b""
901 self._read_pos = 0
Guido van Rossum68bbcd22007-02-27 17:19:33 +0000902
Guido van Rossum024da5c2007-05-17 23:59:11 +0000903 def read(self, n=None):
Guido van Rossum01a27522007-03-07 01:00:12 +0000904 """Read n bytes.
905
906 Returns exactly n bytes of data unless the underlying raw IO
Walter Dörwalda3270002007-05-29 19:13:29 +0000907 stream reaches EOF or if the call would block in non-blocking
Guido van Rossum141f7672007-04-10 00:22:16 +0000908 mode. If n is negative, read until EOF or until read() would
Guido van Rossum01a27522007-03-07 01:00:12 +0000909 block.
Guido van Rossum68bbcd22007-02-27 17:19:33 +0000910 """
Guido van Rossum78892e42007-04-06 17:31:18 +0000911 nodata_val = b""
Antoine Pitrouc66f9092008-07-28 19:46:11 +0000912 empty_values = (b"", None)
913 buf = self._read_buf
914 pos = self._read_pos
915
916 # Special case for when the number of bytes to read is unspecified.
917 if n is None or n == -1:
918 self._reset_read_buf()
919 chunks = [buf[pos:]] # Strip the consumed bytes.
920 current_size = 0
921 while True:
922 # Read until EOF or until read() would block.
923 chunk = self.raw.read()
924 if chunk in empty_values:
925 nodata_val = chunk
926 break
927 current_size += len(chunk)
928 chunks.append(chunk)
929 return b"".join(chunks) or nodata_val
930
931 # The number of bytes to read is specified, return at most n bytes.
932 avail = len(buf) - pos # Length of the available buffered data.
933 if n <= avail:
934 # Fast path: the data to read is fully buffered.
935 self._read_pos += n
936 return buf[pos:pos+n]
937 # Slow path: read from the stream until enough bytes are read,
938 # or until an EOF occurs or until read() would block.
939 chunks = [buf[pos:]]
940 wanted = max(self.buffer_size, n)
941 while avail < n:
942 chunk = self.raw.read(wanted)
943 if chunk in empty_values:
944 nodata_val = chunk
Guido van Rossum68bbcd22007-02-27 17:19:33 +0000945 break
Antoine Pitrouc66f9092008-07-28 19:46:11 +0000946 avail += len(chunk)
947 chunks.append(chunk)
948 # n is more then avail only when an EOF occurred or when
949 # read() would have blocked.
950 n = min(n, avail)
951 out = b"".join(chunks)
952 self._read_buf = out[n:] # Save the extra data in the buffer.
953 self._read_pos = 0
954 return out[:n] if out else nodata_val
Guido van Rossum68bbcd22007-02-27 17:19:33 +0000955
Ka-Ping Yee7a0d3982008-03-17 17:34:48 +0000956 def peek(self, n=0):
Guido van Rossum13633bb2007-04-13 18:42:35 +0000957 """Returns buffered bytes without advancing the position.
958
959 The argument indicates a desired minimal number of bytes; we
960 do at most one raw read to satisfy it. We never return more
961 than self.buffer_size.
Guido van Rossum13633bb2007-04-13 18:42:35 +0000962 """
963 want = min(n, self.buffer_size)
Antoine Pitrouc66f9092008-07-28 19:46:11 +0000964 have = len(self._read_buf) - self._read_pos
Guido van Rossum13633bb2007-04-13 18:42:35 +0000965 if have < want:
966 to_read = self.buffer_size - have
967 current = self.raw.read(to_read)
968 if current:
Antoine Pitrouc66f9092008-07-28 19:46:11 +0000969 self._read_buf = self._read_buf[self._read_pos:] + current
970 self._read_pos = 0
971 return self._read_buf[self._read_pos:]
Guido van Rossum13633bb2007-04-13 18:42:35 +0000972
973 def read1(self, n):
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000974 """Reads up to n bytes, with at most one read() system call."""
975 # Returns up to n bytes. If at least one byte is buffered, we
976 # only return buffered bytes. Otherwise, we do one raw read.
Guido van Rossum13633bb2007-04-13 18:42:35 +0000977 if n <= 0:
978 return b""
Ka-Ping Yee7a0d3982008-03-17 17:34:48 +0000979 self.peek(1)
Antoine Pitrouc66f9092008-07-28 19:46:11 +0000980 return self.read(min(n, len(self._read_buf) - self._read_pos))
Guido van Rossum13633bb2007-04-13 18:42:35 +0000981
Guido van Rossum76c5d4d2007-04-06 19:10:29 +0000982 def tell(self):
Antoine Pitrouc66f9092008-07-28 19:46:11 +0000983 return self.raw.tell() - len(self._read_buf) + self._read_pos
Guido van Rossum76c5d4d2007-04-06 19:10:29 +0000984
985 def seek(self, pos, whence=0):
986 if whence == 1:
Antoine Pitrouc66f9092008-07-28 19:46:11 +0000987 pos -= len(self._read_buf) - self._read_pos
Guido van Rossum53807da2007-04-10 19:01:47 +0000988 pos = self.raw.seek(pos, whence)
Antoine Pitrouc66f9092008-07-28 19:46:11 +0000989 self._reset_read_buf()
Guido van Rossum53807da2007-04-10 19:01:47 +0000990 return pos
Guido van Rossum76c5d4d2007-04-06 19:10:29 +0000991
Guido van Rossum68bbcd22007-02-27 17:19:33 +0000992
Guido van Rossum141f7672007-04-10 00:22:16 +0000993class BufferedWriter(_BufferedIOMixin):
Guido van Rossum68bbcd22007-02-27 17:19:33 +0000994
Christian Heimes5d8da202008-05-06 13:58:24 +0000995 """A buffer for a writeable sequential RawIO object.
Benjamin Peterson2c5f8282008-04-13 00:27:46 +0000996
997 The constructor creates a BufferedWriter for the given writeable raw
998 stream. If the buffer_size is not given, it defaults to
999 DEAFULT_BUFFER_SIZE. If max_buffer_size is omitted, it defaults to
1000 twice the buffer size.
1001 """
Guido van Rossum78892e42007-04-06 17:31:18 +00001002
Guido van Rossum141f7672007-04-10 00:22:16 +00001003 def __init__(self, raw,
1004 buffer_size=DEFAULT_BUFFER_SIZE, max_buffer_size=None):
Guido van Rossum5abbf752007-08-27 17:39:33 +00001005 raw._checkWritable()
Guido van Rossum141f7672007-04-10 00:22:16 +00001006 _BufferedIOMixin.__init__(self, raw)
Guido van Rossum68bbcd22007-02-27 17:19:33 +00001007 self.buffer_size = buffer_size
Guido van Rossum141f7672007-04-10 00:22:16 +00001008 self.max_buffer_size = (2*buffer_size
1009 if max_buffer_size is None
1010 else max_buffer_size)
Guido van Rossum254348e2007-11-21 19:29:53 +00001011 self._write_buf = bytearray()
Guido van Rossum68bbcd22007-02-27 17:19:33 +00001012
1013 def write(self, b):
Guido van Rossum4b5386f2007-07-10 09:12:49 +00001014 if self.closed:
1015 raise ValueError("write to closed file")
Guido van Rossuma74184e2007-08-29 04:05:57 +00001016 if isinstance(b, str):
1017 raise TypeError("can't write str to binary stream")
Guido van Rossum01a27522007-03-07 01:00:12 +00001018 # XXX we can implement some more tricks to try and avoid partial writes
Guido van Rossum01a27522007-03-07 01:00:12 +00001019 if len(self._write_buf) > self.buffer_size:
1020 # We're full, so let's pre-flush the buffer
1021 try:
1022 self.flush()
Guido van Rossum141f7672007-04-10 00:22:16 +00001023 except BlockingIOError as e:
Guido van Rossum01a27522007-03-07 01:00:12 +00001024 # We can't accept anything else.
Guido van Rossum76c5d4d2007-04-06 19:10:29 +00001025 # XXX Why not just let the exception pass through?
Guido van Rossum141f7672007-04-10 00:22:16 +00001026 raise BlockingIOError(e.errno, e.strerror, 0)
Guido van Rossumd4103952007-04-12 05:44:49 +00001027 before = len(self._write_buf)
Guido van Rossum4f0db6e2007-04-08 23:59:06 +00001028 self._write_buf.extend(b)
Guido van Rossumd4103952007-04-12 05:44:49 +00001029 written = len(self._write_buf) - before
Guido van Rossum76c5d4d2007-04-06 19:10:29 +00001030 if len(self._write_buf) > self.buffer_size:
Guido van Rossum01a27522007-03-07 01:00:12 +00001031 try:
1032 self.flush()
Guido van Rossum141f7672007-04-10 00:22:16 +00001033 except BlockingIOError as e:
Guido van Rossum01a27522007-03-07 01:00:12 +00001034 if (len(self._write_buf) > self.max_buffer_size):
1035 # We've hit max_buffer_size. We have to accept a partial
1036 # write and cut back our buffer.
1037 overage = len(self._write_buf) - self.max_buffer_size
1038 self._write_buf = self._write_buf[:self.max_buffer_size]
Guido van Rossum141f7672007-04-10 00:22:16 +00001039 raise BlockingIOError(e.errno, e.strerror, overage)
Guido van Rossumd4103952007-04-12 05:44:49 +00001040 return written
Guido van Rossum68bbcd22007-02-27 17:19:33 +00001041
Alexandre Vassalotti77250f42008-05-06 19:48:38 +00001042 def truncate(self, pos=None):
1043 self.flush()
1044 if pos is None:
1045 pos = self.raw.tell()
1046 return self.raw.truncate(pos)
1047
Guido van Rossum68bbcd22007-02-27 17:19:33 +00001048 def flush(self):
Guido van Rossum4b5386f2007-07-10 09:12:49 +00001049 if self.closed:
1050 raise ValueError("flush of closed file")
Guido van Rossum76c5d4d2007-04-06 19:10:29 +00001051 written = 0
Guido van Rossum01a27522007-03-07 01:00:12 +00001052 try:
Guido van Rossum76c5d4d2007-04-06 19:10:29 +00001053 while self._write_buf:
1054 n = self.raw.write(self._write_buf)
1055 del self._write_buf[:n]
1056 written += n
Guido van Rossum141f7672007-04-10 00:22:16 +00001057 except BlockingIOError as e:
Guido van Rossum76c5d4d2007-04-06 19:10:29 +00001058 n = e.characters_written
1059 del self._write_buf[:n]
1060 written += n
Guido van Rossum141f7672007-04-10 00:22:16 +00001061 raise BlockingIOError(e.errno, e.strerror, written)
Guido van Rossum76c5d4d2007-04-06 19:10:29 +00001062
1063 def tell(self):
1064 return self.raw.tell() + len(self._write_buf)
1065
1066 def seek(self, pos, whence=0):
1067 self.flush()
Guido van Rossum53807da2007-04-10 19:01:47 +00001068 return self.raw.seek(pos, whence)
Guido van Rossum68bbcd22007-02-27 17:19:33 +00001069
Guido van Rossum01a27522007-03-07 01:00:12 +00001070
Guido van Rossum141f7672007-04-10 00:22:16 +00001071class BufferedRWPair(BufferedIOBase):
Guido van Rossum68bbcd22007-02-27 17:19:33 +00001072
Guido van Rossum01a27522007-03-07 01:00:12 +00001073 """A buffered reader and writer object together.
Guido van Rossum68bbcd22007-02-27 17:19:33 +00001074
Benjamin Peterson2c5f8282008-04-13 00:27:46 +00001075 A buffered reader object and buffered writer object put together to
1076 form a sequential IO object that can read and write. This is typically
1077 used with a socket or two-way pipe.
Guido van Rossum78892e42007-04-06 17:31:18 +00001078
Benjamin Peterson2c5f8282008-04-13 00:27:46 +00001079 reader and writer are RawIOBase objects that are readable and
1080 writeable respectively. If the buffer_size is omitted it defaults to
1081 DEFAULT_BUFFER_SIZE. The max_buffer_size (for the buffered writer)
1082 defaults to twice the buffer size.
Guido van Rossum68bbcd22007-02-27 17:19:33 +00001083 """
1084
Benjamin Peterson2c5f8282008-04-13 00:27:46 +00001085 # XXX The usefulness of this (compared to having two separate IO
1086 # objects) is questionable.
1087
Guido van Rossum141f7672007-04-10 00:22:16 +00001088 def __init__(self, reader, writer,
1089 buffer_size=DEFAULT_BUFFER_SIZE, max_buffer_size=None):
1090 """Constructor.
1091
1092 The arguments are two RawIO instances.
1093 """
Guido van Rossum5abbf752007-08-27 17:39:33 +00001094 reader._checkReadable()
1095 writer._checkWritable()
Guido van Rossum141f7672007-04-10 00:22:16 +00001096 self.reader = BufferedReader(reader, buffer_size)
1097 self.writer = BufferedWriter(writer, buffer_size, max_buffer_size)
Guido van Rossum01a27522007-03-07 01:00:12 +00001098
Guido van Rossum024da5c2007-05-17 23:59:11 +00001099 def read(self, n=None):
1100 if n is None:
1101 n = -1
Guido van Rossum01a27522007-03-07 01:00:12 +00001102 return self.reader.read(n)
1103
Guido van Rossum141f7672007-04-10 00:22:16 +00001104 def readinto(self, b):
1105 return self.reader.readinto(b)
1106
Guido van Rossum01a27522007-03-07 01:00:12 +00001107 def write(self, b):
1108 return self.writer.write(b)
1109
Ka-Ping Yee7a0d3982008-03-17 17:34:48 +00001110 def peek(self, n=0):
1111 return self.reader.peek(n)
Guido van Rossum13633bb2007-04-13 18:42:35 +00001112
1113 def read1(self, n):
1114 return self.reader.read1(n)
1115
Guido van Rossum01a27522007-03-07 01:00:12 +00001116 def readable(self):
1117 return self.reader.readable()
1118
1119 def writable(self):
1120 return self.writer.writable()
1121
1122 def flush(self):
1123 return self.writer.flush()
Guido van Rossum68bbcd22007-02-27 17:19:33 +00001124
Guido van Rossum01a27522007-03-07 01:00:12 +00001125 def close(self):
Guido van Rossum01a27522007-03-07 01:00:12 +00001126 self.writer.close()
Guido van Rossum141f7672007-04-10 00:22:16 +00001127 self.reader.close()
1128
1129 def isatty(self):
1130 return self.reader.isatty() or self.writer.isatty()
Guido van Rossum01a27522007-03-07 01:00:12 +00001131
Guido van Rossum4f0db6e2007-04-08 23:59:06 +00001132 @property
1133 def closed(self):
Guido van Rossum141f7672007-04-10 00:22:16 +00001134 return self.writer.closed()
Guido van Rossum01a27522007-03-07 01:00:12 +00001135
Guido van Rossum4f0db6e2007-04-08 23:59:06 +00001136
Guido van Rossum141f7672007-04-10 00:22:16 +00001137class BufferedRandom(BufferedWriter, BufferedReader):
Guido van Rossum01a27522007-03-07 01:00:12 +00001138
Christian Heimes5d8da202008-05-06 13:58:24 +00001139 """A buffered interface to random access streams.
Benjamin Peterson2c5f8282008-04-13 00:27:46 +00001140
1141 The constructor creates a reader and writer for a seekable stream,
1142 raw, given in the first argument. If the buffer_size is omitted it
1143 defaults to DEFAULT_BUFFER_SIZE. The max_buffer_size (for the buffered
1144 writer) defaults to twice the buffer size.
1145 """
Guido van Rossum78892e42007-04-06 17:31:18 +00001146
Guido van Rossum141f7672007-04-10 00:22:16 +00001147 def __init__(self, raw,
1148 buffer_size=DEFAULT_BUFFER_SIZE, max_buffer_size=None):
Guido van Rossum5abbf752007-08-27 17:39:33 +00001149 raw._checkSeekable()
Guido van Rossum4f0db6e2007-04-08 23:59:06 +00001150 BufferedReader.__init__(self, raw, buffer_size)
Guido van Rossum01a27522007-03-07 01:00:12 +00001151 BufferedWriter.__init__(self, raw, buffer_size, max_buffer_size)
1152
Guido van Rossum01a27522007-03-07 01:00:12 +00001153 def seek(self, pos, whence=0):
1154 self.flush()
Guido van Rossum76c5d4d2007-04-06 19:10:29 +00001155 # First do the raw seek, then empty the read buffer, so that
1156 # if the raw seek fails, we don't lose buffered data forever.
Guido van Rossum53807da2007-04-10 19:01:47 +00001157 pos = self.raw.seek(pos, whence)
Antoine Pitrouc66f9092008-07-28 19:46:11 +00001158 self._reset_read_buf()
Guido van Rossum53807da2007-04-10 19:01:47 +00001159 return pos
Guido van Rossum01a27522007-03-07 01:00:12 +00001160
1161 def tell(self):
Antoine Pitrouc66f9092008-07-28 19:46:11 +00001162 if self._write_buf:
Guido van Rossum01a27522007-03-07 01:00:12 +00001163 return self.raw.tell() + len(self._write_buf)
1164 else:
Antoine Pitrouc66f9092008-07-28 19:46:11 +00001165 return BufferedReader.tell(self)
Guido van Rossum01a27522007-03-07 01:00:12 +00001166
Alexandre Vassalotti77250f42008-05-06 19:48:38 +00001167 def truncate(self, pos=None):
1168 if pos is None:
1169 pos = self.tell()
1170 # Use seek to flush the read buffer.
1171 self.seek(pos)
1172 return BufferedWriter.truncate(self)
1173
Guido van Rossum024da5c2007-05-17 23:59:11 +00001174 def read(self, n=None):
1175 if n is None:
1176 n = -1
Guido van Rossum01a27522007-03-07 01:00:12 +00001177 self.flush()
1178 return BufferedReader.read(self, n)
1179
Guido van Rossum141f7672007-04-10 00:22:16 +00001180 def readinto(self, b):
1181 self.flush()
1182 return BufferedReader.readinto(self, b)
1183
Ka-Ping Yee7a0d3982008-03-17 17:34:48 +00001184 def peek(self, n=0):
Guido van Rossum13633bb2007-04-13 18:42:35 +00001185 self.flush()
Ka-Ping Yee7a0d3982008-03-17 17:34:48 +00001186 return BufferedReader.peek(self, n)
Guido van Rossum13633bb2007-04-13 18:42:35 +00001187
1188 def read1(self, n):
1189 self.flush()
1190 return BufferedReader.read1(self, n)
1191
Guido van Rossum01a27522007-03-07 01:00:12 +00001192 def write(self, b):
Guido van Rossum78892e42007-04-06 17:31:18 +00001193 if self._read_buf:
Antoine Pitrouc66f9092008-07-28 19:46:11 +00001194 # Undo readahead
1195 self.raw.seek(self._read_pos - len(self._read_buf), 1)
1196 self._reset_read_buf()
Guido van Rossum01a27522007-03-07 01:00:12 +00001197 return BufferedWriter.write(self, b)
1198
Guido van Rossum78892e42007-04-06 17:31:18 +00001199
Guido van Rossumcce92b22007-04-10 14:41:39 +00001200class TextIOBase(IOBase):
Guido van Rossum78892e42007-04-06 17:31:18 +00001201
1202 """Base class for text I/O.
1203
Benjamin Peterson2c5f8282008-04-13 00:27:46 +00001204 This class provides a character and line based interface to stream
1205 I/O. There is no readinto method because Python's character strings
1206 are immutable. There is no public constructor.
Guido van Rossum78892e42007-04-06 17:31:18 +00001207 """
1208
1209 def read(self, n: int = -1) -> str:
Christian Heimes5d8da202008-05-06 13:58:24 +00001210 """Read at most n characters from stream.
Guido van Rossum78892e42007-04-06 17:31:18 +00001211
1212 Read from underlying buffer until we have n characters or we hit EOF.
1213 If n is negative or omitted, read until EOF.
1214 """
Guido van Rossum4f0db6e2007-04-08 23:59:06 +00001215 self._unsupported("read")
Guido van Rossum78892e42007-04-06 17:31:18 +00001216
Guido van Rossum9b76da62007-04-11 01:09:03 +00001217 def write(self, s: str) -> int:
Christian Heimes5d8da202008-05-06 13:58:24 +00001218 """Write string s to stream."""
Guido van Rossum4f0db6e2007-04-08 23:59:06 +00001219 self._unsupported("write")
Guido van Rossum78892e42007-04-06 17:31:18 +00001220
Guido van Rossum9b76da62007-04-11 01:09:03 +00001221 def truncate(self, pos: int = None) -> int:
Christian Heimes5d8da202008-05-06 13:58:24 +00001222 """Truncate size to pos."""
Alexandre Vassalotti77250f42008-05-06 19:48:38 +00001223 self._unsupported("truncate")
Guido van Rossum9b76da62007-04-11 01:09:03 +00001224
Guido van Rossum78892e42007-04-06 17:31:18 +00001225 def readline(self) -> str:
Christian Heimes5d8da202008-05-06 13:58:24 +00001226 """Read until newline or EOF.
Guido van Rossum78892e42007-04-06 17:31:18 +00001227
1228 Returns an empty string if EOF is hit immediately.
1229 """
Guido van Rossum4f0db6e2007-04-08 23:59:06 +00001230 self._unsupported("readline")
Guido van Rossum78892e42007-04-06 17:31:18 +00001231
Guido van Rossumfc3436b2007-05-24 17:58:06 +00001232 @property
1233 def encoding(self):
1234 """Subclasses should override."""
1235 return None
1236
Guido van Rossum8358db22007-08-18 21:39:55 +00001237 @property
1238 def newlines(self):
Christian Heimes5d8da202008-05-06 13:58:24 +00001239 """Line endings translated so far.
Guido van Rossum8358db22007-08-18 21:39:55 +00001240
1241 Only line endings translated during reading are considered.
1242
1243 Subclasses should override.
1244 """
1245 return None
1246
Guido van Rossum78892e42007-04-06 17:31:18 +00001247
Amaury Forgeot d'Arc1ff99102007-11-19 20:34:10 +00001248class IncrementalNewlineDecoder(codecs.IncrementalDecoder):
Benjamin Peterson2c5f8282008-04-13 00:27:46 +00001249 r"""Codec used when reading a file in universal newlines mode. It wraps
1250 another incremental decoder, translating \r\n and \r into \n. It also
1251 records the types of newlines encountered. When used with
1252 translate=False, it ensures that the newline sequence is returned in
1253 one piece.
Amaury Forgeot d'Arc1ff99102007-11-19 20:34:10 +00001254 """
1255 def __init__(self, decoder, translate, errors='strict'):
1256 codecs.IncrementalDecoder.__init__(self, errors=errors)
1257 self.buffer = b''
1258 self.translate = translate
1259 self.decoder = decoder
1260 self.seennl = 0
1261
1262 def decode(self, input, final=False):
1263 # decode input (with the eventual \r from a previous pass)
1264 if self.buffer:
1265 input = self.buffer + input
1266
1267 output = self.decoder.decode(input, final=final)
1268
1269 # retain last \r even when not translating data:
1270 # then readline() is sure to get \r\n in one pass
1271 if output.endswith("\r") and not final:
1272 output = output[:-1]
1273 self.buffer = b'\r'
1274 else:
1275 self.buffer = b''
1276
1277 # Record which newlines are read
1278 crlf = output.count('\r\n')
1279 cr = output.count('\r') - crlf
1280 lf = output.count('\n') - crlf
1281 self.seennl |= (lf and self._LF) | (cr and self._CR) \
1282 | (crlf and self._CRLF)
1283
1284 if self.translate:
1285 if crlf:
1286 output = output.replace("\r\n", "\n")
1287 if cr:
1288 output = output.replace("\r", "\n")
1289
1290 return output
1291
1292 def getstate(self):
1293 buf, flag = self.decoder.getstate()
1294 return buf + self.buffer, flag
1295
1296 def setstate(self, state):
1297 buf, flag = state
1298 if buf.endswith(b'\r'):
1299 self.buffer = b'\r'
1300 buf = buf[:-1]
1301 else:
1302 self.buffer = b''
1303 self.decoder.setstate((buf, flag))
1304
1305 def reset(self):
Alexandre Vassalottic3d7fe02007-12-28 01:24:22 +00001306 self.seennl = 0
Amaury Forgeot d'Arc1ff99102007-11-19 20:34:10 +00001307 self.buffer = b''
1308 self.decoder.reset()
1309
1310 _LF = 1
1311 _CR = 2
1312 _CRLF = 4
1313
1314 @property
1315 def newlines(self):
1316 return (None,
1317 "\n",
1318 "\r",
1319 ("\r", "\n"),
1320 "\r\n",
1321 ("\n", "\r\n"),
1322 ("\r", "\r\n"),
1323 ("\r", "\n", "\r\n")
1324 )[self.seennl]
1325
1326
Guido van Rossum78892e42007-04-06 17:31:18 +00001327class TextIOWrapper(TextIOBase):
1328
Christian Heimes5d8da202008-05-06 13:58:24 +00001329 r"""Character and line based layer over a BufferedIOBase object, buffer.
Benjamin Peterson2c5f8282008-04-13 00:27:46 +00001330
1331 encoding gives the name of the encoding that the stream will be
1332 decoded or encoded with. It defaults to locale.getpreferredencoding.
1333
1334 errors determines the strictness of encoding and decoding (see the
1335 codecs.register) and defaults to "strict".
1336
1337 newline can be None, '', '\n', '\r', or '\r\n'. It controls the
1338 handling of line endings. If it is None, universal newlines is
1339 enabled. With this enabled, on input, the lines endings '\n', '\r',
1340 or '\r\n' are translated to '\n' before being returned to the
1341 caller. Conversely, on output, '\n' is translated to the system
1342 default line seperator, os.linesep. If newline is any other of its
1343 legal values, that newline becomes the newline when the file is read
1344 and it is returned untranslated. On output, '\n' is converted to the
1345 newline.
1346
1347 If line_buffering is True, a call to flush is implied when a call to
1348 write contains a newline character.
Guido van Rossum78892e42007-04-06 17:31:18 +00001349 """
1350
Guido van Rossumb9c4c3e2007-04-11 16:07:50 +00001351 _CHUNK_SIZE = 128
Guido van Rossum78892e42007-04-06 17:31:18 +00001352
Guido van Rossumf64db9f2007-12-06 01:04:26 +00001353 def __init__(self, buffer, encoding=None, errors=None, newline=None,
1354 line_buffering=False):
Guido van Rossum8358db22007-08-18 21:39:55 +00001355 if newline not in (None, "", "\n", "\r", "\r\n"):
Guido van Rossum9b76da62007-04-11 01:09:03 +00001356 raise ValueError("illegal newline value: %r" % (newline,))
Guido van Rossum78892e42007-04-06 17:31:18 +00001357 if encoding is None:
Martin v. Löwisd1cd4d42007-08-11 14:02:14 +00001358 try:
1359 encoding = os.device_encoding(buffer.fileno())
Brett Cannon041683d2007-10-11 23:08:53 +00001360 except (AttributeError, UnsupportedOperation):
Martin v. Löwisd1cd4d42007-08-11 14:02:14 +00001361 pass
1362 if encoding is None:
Martin v. Löwisd78d3b42007-08-11 15:36:45 +00001363 try:
1364 import locale
1365 except ImportError:
1366 # Importing locale may fail if Python is being built
1367 encoding = "ascii"
1368 else:
1369 encoding = locale.getpreferredencoding()
Guido van Rossum78892e42007-04-06 17:31:18 +00001370
Christian Heimes8bd14fb2007-11-08 16:34:32 +00001371 if not isinstance(encoding, str):
1372 raise ValueError("invalid encoding: %r" % encoding)
1373
Guido van Rossume7fc50f2007-12-03 22:54:21 +00001374 if errors is None:
1375 errors = "strict"
1376 else:
1377 if not isinstance(errors, str):
1378 raise ValueError("invalid errors: %r" % errors)
1379
Guido van Rossum78892e42007-04-06 17:31:18 +00001380 self.buffer = buffer
Guido van Rossumf64db9f2007-12-06 01:04:26 +00001381 self._line_buffering = line_buffering
Guido van Rossum78892e42007-04-06 17:31:18 +00001382 self._encoding = encoding
Guido van Rossume7fc50f2007-12-03 22:54:21 +00001383 self._errors = errors
Guido van Rossum8358db22007-08-18 21:39:55 +00001384 self._readuniversal = not newline
1385 self._readtranslate = newline is None
1386 self._readnl = newline
1387 self._writetranslate = newline != ''
1388 self._writenl = newline or os.linesep
Alexandre Vassalottia38f73b2008-01-07 18:30:48 +00001389 self._encoder = None
Guido van Rossum78892e42007-04-06 17:31:18 +00001390 self._decoder = None
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001391 self._decoded_chars = '' # buffer for text returned from decoder
1392 self._decoded_chars_used = 0 # offset into _decoded_chars for read()
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001393 self._snapshot = None # info for reconstructing decoder state
Guido van Rossumb9c4c3e2007-04-11 16:07:50 +00001394 self._seekable = self._telling = self.buffer.seekable()
Guido van Rossum9b76da62007-04-11 01:09:03 +00001395
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001396 # self._snapshot is either None, or a tuple (dec_flags, next_input)
1397 # where dec_flags is the second (integer) item of the decoder state
1398 # and next_input is the chunk of input bytes that comes next after the
1399 # snapshot point. We use this to reconstruct decoder states in tell().
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001400
1401 # Naming convention:
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001402 # - "bytes_..." for integer variables that count input bytes
1403 # - "chars_..." for integer variables that count decoded characters
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001404
Guido van Rossumfc3436b2007-05-24 17:58:06 +00001405 @property
1406 def encoding(self):
1407 return self._encoding
1408
Guido van Rossume7fc50f2007-12-03 22:54:21 +00001409 @property
1410 def errors(self):
1411 return self._errors
1412
Guido van Rossumf64db9f2007-12-06 01:04:26 +00001413 @property
1414 def line_buffering(self):
1415 return self._line_buffering
1416
Ka-Ping Yeeddaa7062008-03-17 20:35:15 +00001417 def seekable(self):
Guido van Rossum9b76da62007-04-11 01:09:03 +00001418 return self._seekable
Guido van Rossum78892e42007-04-06 17:31:18 +00001419
Alexandre Vassalotti77250f42008-05-06 19:48:38 +00001420 def readable(self):
1421 return self.buffer.readable()
1422
1423 def writable(self):
1424 return self.buffer.writable()
1425
Guido van Rossum4f0db6e2007-04-08 23:59:06 +00001426 def flush(self):
1427 self.buffer.flush()
Guido van Rossumb9c4c3e2007-04-11 16:07:50 +00001428 self._telling = self._seekable
Guido van Rossum4f0db6e2007-04-08 23:59:06 +00001429
1430 def close(self):
Guido van Rossum33e7a8e2007-07-22 20:38:07 +00001431 try:
1432 self.flush()
1433 except:
1434 pass # If flush() fails, just give up
Guido van Rossum4f0db6e2007-04-08 23:59:06 +00001435 self.buffer.close()
1436
1437 @property
1438 def closed(self):
1439 return self.buffer.closed
1440
Guido van Rossum9be55972007-04-07 02:59:27 +00001441 def fileno(self):
1442 return self.buffer.fileno()
1443
Guido van Rossum859b5ec2007-05-27 09:14:51 +00001444 def isatty(self):
1445 return self.buffer.isatty()
1446
Guido van Rossum78892e42007-04-06 17:31:18 +00001447 def write(self, s: str):
Guido van Rossum4b5386f2007-07-10 09:12:49 +00001448 if self.closed:
1449 raise ValueError("write to closed file")
Guido van Rossum3172c5d2007-10-16 18:12:55 +00001450 if not isinstance(s, str):
Guido van Rossumdcce8392007-08-29 18:10:08 +00001451 raise TypeError("can't write %s to text stream" %
1452 s.__class__.__name__)
Amaury Forgeot d'Arc1ff99102007-11-19 20:34:10 +00001453 length = len(s)
Guido van Rossumf64db9f2007-12-06 01:04:26 +00001454 haslf = (self._writetranslate or self._line_buffering) and "\n" in s
Guido van Rossum8358db22007-08-18 21:39:55 +00001455 if haslf and self._writetranslate and self._writenl != "\n":
1456 s = s.replace("\n", self._writenl)
Alexandre Vassalottia38f73b2008-01-07 18:30:48 +00001457 encoder = self._encoder or self._get_encoder()
Guido van Rossum9b76da62007-04-11 01:09:03 +00001458 # XXX What if we were just reading?
Alexandre Vassalottia38f73b2008-01-07 18:30:48 +00001459 b = encoder.encode(s)
Guido van Rossum8358db22007-08-18 21:39:55 +00001460 self.buffer.write(b)
Guido van Rossumf64db9f2007-12-06 01:04:26 +00001461 if self._line_buffering and (haslf or "\r" in s):
Guido van Rossum4f0db6e2007-04-08 23:59:06 +00001462 self.flush()
Amaury Forgeot d'Arc1ff99102007-11-19 20:34:10 +00001463 self._snapshot = None
1464 if self._decoder:
1465 self._decoder.reset()
1466 return length
Guido van Rossum78892e42007-04-06 17:31:18 +00001467
Alexandre Vassalottia38f73b2008-01-07 18:30:48 +00001468 def _get_encoder(self):
1469 make_encoder = codecs.getincrementalencoder(self._encoding)
1470 self._encoder = make_encoder(self._errors)
1471 return self._encoder
1472
Guido van Rossum78892e42007-04-06 17:31:18 +00001473 def _get_decoder(self):
1474 make_decoder = codecs.getincrementaldecoder(self._encoding)
Guido van Rossume7fc50f2007-12-03 22:54:21 +00001475 decoder = make_decoder(self._errors)
Amaury Forgeot d'Arc1ff99102007-11-19 20:34:10 +00001476 if self._readuniversal:
1477 decoder = IncrementalNewlineDecoder(decoder, self._readtranslate)
1478 self._decoder = decoder
Guido van Rossum78892e42007-04-06 17:31:18 +00001479 return decoder
1480
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001481 # The following three methods implement an ADT for _decoded_chars.
1482 # Text returned from the decoder is buffered here until the client
1483 # requests it by calling our read() or readline() method.
1484 def _set_decoded_chars(self, chars):
1485 """Set the _decoded_chars buffer."""
1486 self._decoded_chars = chars
1487 self._decoded_chars_used = 0
1488
1489 def _get_decoded_chars(self, n=None):
1490 """Advance into the _decoded_chars buffer."""
1491 offset = self._decoded_chars_used
1492 if n is None:
1493 chars = self._decoded_chars[offset:]
1494 else:
1495 chars = self._decoded_chars[offset:offset + n]
1496 self._decoded_chars_used += len(chars)
1497 return chars
1498
1499 def _rewind_decoded_chars(self, n):
1500 """Rewind the _decoded_chars buffer."""
1501 if self._decoded_chars_used < n:
1502 raise AssertionError("rewind decoded_chars out of bounds")
1503 self._decoded_chars_used -= n
1504
Guido van Rossum9b76da62007-04-11 01:09:03 +00001505 def _read_chunk(self):
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001506 """
1507 Read and decode the next chunk of data from the BufferedReader.
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001508 """
1509
Benjamin Peterson2c5f8282008-04-13 00:27:46 +00001510 # The return value is True unless EOF was reached. The decoded
1511 # string is placed in self._decoded_chars (replacing its previous
1512 # value). The entire input chunk is sent to the decoder, though
1513 # some of it may remain buffered in the decoder, yet to be
1514 # converted.
1515
Guido van Rossum5abbf752007-08-27 17:39:33 +00001516 if self._decoder is None:
1517 raise ValueError("no decoder")
Guido van Rossum9b76da62007-04-11 01:09:03 +00001518
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001519 if self._telling:
1520 # To prepare for tell(), we need to snapshot a point in the
1521 # file where the decoder's input buffer is empty.
Guido van Rossum9b76da62007-04-11 01:09:03 +00001522
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001523 dec_buffer, dec_flags = self._decoder.getstate()
1524 # Given this, we know there was a valid snapshot point
1525 # len(dec_buffer) bytes ago with decoder state (b'', dec_flags).
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001526
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001527 # Read a chunk, decode it, and put the result in self._decoded_chars.
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001528 input_chunk = self.buffer.read1(self._CHUNK_SIZE)
1529 eof = not input_chunk
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001530 self._set_decoded_chars(self._decoder.decode(input_chunk, eof))
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001531
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001532 if self._telling:
1533 # At the snapshot point, len(dec_buffer) bytes before the read,
1534 # the next input to be decoded is dec_buffer + input_chunk.
1535 self._snapshot = (dec_flags, dec_buffer + input_chunk)
1536
Ka-Ping Yeedbe28e52008-03-20 10:34:07 +00001537 return not eof
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001538
Ka-Ping Yeedbe28e52008-03-20 10:34:07 +00001539 def _pack_cookie(self, position, dec_flags=0,
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001540 bytes_to_feed=0, need_eof=0, chars_to_skip=0):
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001541 # The meaning of a tell() cookie is: seek to position, set the
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001542 # decoder flags to dec_flags, read bytes_to_feed bytes, feed them
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001543 # into the decoder with need_eof as the EOF flag, then skip
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001544 # chars_to_skip characters of the decoded result. For most simple
1545 # decoders, tell() will often just give a byte offset in the file.
1546 return (position | (dec_flags<<64) | (bytes_to_feed<<128) |
1547 (chars_to_skip<<192) | bool(need_eof)<<256)
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001548
Ka-Ping Yeedbe28e52008-03-20 10:34:07 +00001549 def _unpack_cookie(self, bigint):
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001550 rest, position = divmod(bigint, 1<<64)
1551 rest, dec_flags = divmod(rest, 1<<64)
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001552 rest, bytes_to_feed = divmod(rest, 1<<64)
1553 need_eof, chars_to_skip = divmod(rest, 1<<64)
1554 return position, dec_flags, bytes_to_feed, need_eof, chars_to_skip
Guido van Rossum9b76da62007-04-11 01:09:03 +00001555
1556 def tell(self):
1557 if not self._seekable:
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001558 raise IOError("underlying stream is not seekable")
Guido van Rossumb9c4c3e2007-04-11 16:07:50 +00001559 if not self._telling:
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001560 raise IOError("telling position disabled by next() call")
Guido van Rossum9b76da62007-04-11 01:09:03 +00001561 self.flush()
Guido van Rossumcba608c2007-04-11 14:19:59 +00001562 position = self.buffer.tell()
Guido van Rossumd76e7792007-04-17 02:38:04 +00001563 decoder = self._decoder
1564 if decoder is None or self._snapshot is None:
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001565 if self._decoded_chars:
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001566 # This should never happen.
1567 raise AssertionError("pending decoded text")
Guido van Rossumcba608c2007-04-11 14:19:59 +00001568 return position
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001569
1570 # Skip backward to the snapshot point (see _read_chunk).
Ka-Ping Yeedbe28e52008-03-20 10:34:07 +00001571 dec_flags, next_input = self._snapshot
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001572 position -= len(next_input)
1573
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001574 # How many decoded characters have been used up since the snapshot?
1575 chars_to_skip = self._decoded_chars_used
1576 if chars_to_skip == 0:
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001577 # We haven't moved from the snapshot point.
Ka-Ping Yeedbe28e52008-03-20 10:34:07 +00001578 return self._pack_cookie(position, dec_flags)
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001579
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001580 # Starting from the snapshot position, we will walk the decoder
1581 # forward until it gives us enough decoded characters.
Guido van Rossumd76e7792007-04-17 02:38:04 +00001582 saved_state = decoder.getstate()
1583 try:
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001584 # Note our initial start point.
1585 decoder.setstate((b'', dec_flags))
1586 start_pos = position
1587 start_flags, bytes_fed, chars_decoded = dec_flags, 0, 0
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001588 need_eof = 0
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001589
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001590 # Feed the decoder one byte at a time. As we go, note the
1591 # nearest "safe start point" before the current location
1592 # (a point where the decoder has nothing buffered, so seek()
1593 # can safely start from there and advance to this location).
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001594 next_byte = bytearray(1)
1595 for next_byte[0] in next_input:
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001596 bytes_fed += 1
1597 chars_decoded += len(decoder.decode(next_byte))
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001598 dec_buffer, dec_flags = decoder.getstate()
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001599 if not dec_buffer and chars_decoded <= chars_to_skip:
1600 # Decoder buffer is empty, so this is a safe start point.
1601 start_pos += bytes_fed
1602 chars_to_skip -= chars_decoded
1603 start_flags, bytes_fed, chars_decoded = dec_flags, 0, 0
1604 if chars_decoded >= chars_to_skip:
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001605 break
1606 else:
Ka-Ping Yeedbe28e52008-03-20 10:34:07 +00001607 # We didn't get enough decoded data; signal EOF to get more.
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001608 chars_decoded += len(decoder.decode(b'', final=True))
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001609 need_eof = 1
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001610 if chars_decoded < chars_to_skip:
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001611 raise IOError("can't reconstruct logical file position")
1612
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001613 # The returned cookie corresponds to the last safe start point.
Ka-Ping Yeedbe28e52008-03-20 10:34:07 +00001614 return self._pack_cookie(
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001615 start_pos, start_flags, bytes_fed, need_eof, chars_to_skip)
Guido van Rossumd76e7792007-04-17 02:38:04 +00001616 finally:
1617 decoder.setstate(saved_state)
Guido van Rossum9b76da62007-04-11 01:09:03 +00001618
Alexandre Vassalotti77250f42008-05-06 19:48:38 +00001619 def truncate(self, pos=None):
1620 self.flush()
1621 if pos is None:
1622 pos = self.tell()
1623 self.seek(pos)
1624 return self.buffer.truncate()
1625
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001626 def seek(self, cookie, whence=0):
Alexandre Vassalotti77250f42008-05-06 19:48:38 +00001627 if self.closed:
1628 raise ValueError("tell on closed file")
Guido van Rossum9b76da62007-04-11 01:09:03 +00001629 if not self._seekable:
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001630 raise IOError("underlying stream is not seekable")
1631 if whence == 1: # seek relative to current position
1632 if cookie != 0:
1633 raise IOError("can't do nonzero cur-relative seeks")
1634 # Seeking to the current position should attempt to
1635 # sync the underlying buffer with the current position.
Guido van Rossumaa43ed92007-04-12 05:24:24 +00001636 whence = 0
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001637 cookie = self.tell()
1638 if whence == 2: # seek relative to end of file
1639 if cookie != 0:
1640 raise IOError("can't do nonzero end-relative seeks")
Guido van Rossum9b76da62007-04-11 01:09:03 +00001641 self.flush()
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001642 position = self.buffer.seek(0, 2)
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001643 self._set_decoded_chars('')
1644 self._snapshot = None
Amaury Forgeot d'Arc1ff99102007-11-19 20:34:10 +00001645 if self._decoder:
1646 self._decoder.reset()
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001647 return position
Guido van Rossum9b76da62007-04-11 01:09:03 +00001648 if whence != 0:
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001649 raise ValueError("invalid whence (%r, should be 0, 1 or 2)" %
Guido van Rossum9b76da62007-04-11 01:09:03 +00001650 (whence,))
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001651 if cookie < 0:
1652 raise ValueError("negative seek position %r" % (cookie,))
Guido van Rossumb9c4c3e2007-04-11 16:07:50 +00001653 self.flush()
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001654
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001655 # The strategy of seek() is to go back to the safe start point
1656 # and replay the effect of read(chars_to_skip) from there.
1657 start_pos, dec_flags, bytes_to_feed, need_eof, chars_to_skip = \
Ka-Ping Yeedbe28e52008-03-20 10:34:07 +00001658 self._unpack_cookie(cookie)
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001659
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001660 # Seek back to the safe start point.
1661 self.buffer.seek(start_pos)
1662 self._set_decoded_chars('')
Ka-Ping Yeedbe28e52008-03-20 10:34:07 +00001663 self._snapshot = None
1664
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001665 # Restore the decoder to its state from the safe start point.
1666 if self._decoder or dec_flags or chars_to_skip:
1667 self._decoder = self._decoder or self._get_decoder()
1668 self._decoder.setstate((b'', dec_flags))
1669 self._snapshot = (dec_flags, b'')
Ka-Ping Yeedbe28e52008-03-20 10:34:07 +00001670
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001671 if chars_to_skip:
1672 # Just like _read_chunk, feed the decoder and save a snapshot.
1673 input_chunk = self.buffer.read(bytes_to_feed)
1674 self._set_decoded_chars(
1675 self._decoder.decode(input_chunk, need_eof))
1676 self._snapshot = (dec_flags, input_chunk)
1677
1678 # Skip chars_to_skip of the decoded characters.
1679 if len(self._decoded_chars) < chars_to_skip:
1680 raise IOError("can't restore logical file position")
1681 self._decoded_chars_used = chars_to_skip
1682
1683 return cookie
Ka-Ping Yeedbe28e52008-03-20 10:34:07 +00001684
Guido van Rossum024da5c2007-05-17 23:59:11 +00001685 def read(self, n=None):
1686 if n is None:
1687 n = -1
Guido van Rossum78892e42007-04-06 17:31:18 +00001688 decoder = self._decoder or self._get_decoder()
Guido van Rossum78892e42007-04-06 17:31:18 +00001689 if n < 0:
Ka-Ping Yeedbe28e52008-03-20 10:34:07 +00001690 # Read everything.
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001691 result = (self._get_decoded_chars() +
Ka-Ping Yeedbe28e52008-03-20 10:34:07 +00001692 decoder.decode(self.buffer.read(), final=True))
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001693 self._set_decoded_chars('')
1694 self._snapshot = None
Ka-Ping Yeef44c7e82008-03-18 04:51:32 +00001695 return result
Guido van Rossum78892e42007-04-06 17:31:18 +00001696 else:
Ka-Ping Yeedbe28e52008-03-20 10:34:07 +00001697 # Keep reading chunks until we have n characters to return.
1698 eof = False
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001699 result = self._get_decoded_chars(n)
Ka-Ping Yeedbe28e52008-03-20 10:34:07 +00001700 while len(result) < n and not eof:
1701 eof = not self._read_chunk()
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001702 result += self._get_decoded_chars(n - len(result))
Ka-Ping Yeedbe28e52008-03-20 10:34:07 +00001703 return result
Guido van Rossum78892e42007-04-06 17:31:18 +00001704
Guido van Rossum024da5c2007-05-17 23:59:11 +00001705 def __next__(self):
Guido van Rossumb9c4c3e2007-04-11 16:07:50 +00001706 self._telling = False
1707 line = self.readline()
1708 if not line:
1709 self._snapshot = None
1710 self._telling = self._seekable
1711 raise StopIteration
1712 return line
1713
Guido van Rossum4f0db6e2007-04-08 23:59:06 +00001714 def readline(self, limit=None):
Alexandre Vassalotti77250f42008-05-06 19:48:38 +00001715 if self.closed:
1716 raise ValueError("read from closed file")
Guido van Rossum98297ee2007-11-06 21:34:58 +00001717 if limit is None:
1718 limit = -1
Guido van Rossum4f0db6e2007-04-08 23:59:06 +00001719
Ka-Ping Yeedbe28e52008-03-20 10:34:07 +00001720 # Grab all the decoded text (we will rewind any extra bits later).
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001721 line = self._get_decoded_chars()
Ka-Ping Yeedbe28e52008-03-20 10:34:07 +00001722
Guido van Rossum78892e42007-04-06 17:31:18 +00001723 start = 0
1724 decoder = self._decoder or self._get_decoder()
1725
Guido van Rossum8358db22007-08-18 21:39:55 +00001726 pos = endpos = None
Guido van Rossum78892e42007-04-06 17:31:18 +00001727 while True:
Amaury Forgeot d'Arc1ff99102007-11-19 20:34:10 +00001728 if self._readtranslate:
1729 # Newlines are already translated, only search for \n
1730 pos = line.find('\n', start)
1731 if pos >= 0:
1732 endpos = pos + 1
1733 break
1734 else:
1735 start = len(line)
1736
1737 elif self._readuniversal:
Guido van Rossum8358db22007-08-18 21:39:55 +00001738 # Universal newline search. Find any of \r, \r\n, \n
Amaury Forgeot d'Arc1ff99102007-11-19 20:34:10 +00001739 # The decoder ensures that \r\n are not split in two pieces
Guido van Rossum78892e42007-04-06 17:31:18 +00001740
Guido van Rossum8358db22007-08-18 21:39:55 +00001741 # In C we'd look for these in parallel of course.
1742 nlpos = line.find("\n", start)
1743 crpos = line.find("\r", start)
1744 if crpos == -1:
1745 if nlpos == -1:
Amaury Forgeot d'Arc1ff99102007-11-19 20:34:10 +00001746 # Nothing found
Guido van Rossum8358db22007-08-18 21:39:55 +00001747 start = len(line)
Guido van Rossum78892e42007-04-06 17:31:18 +00001748 else:
Guido van Rossum8358db22007-08-18 21:39:55 +00001749 # Found \n
Amaury Forgeot d'Arc1ff99102007-11-19 20:34:10 +00001750 endpos = nlpos + 1
Guido van Rossum8358db22007-08-18 21:39:55 +00001751 break
1752 elif nlpos == -1:
Amaury Forgeot d'Arc1ff99102007-11-19 20:34:10 +00001753 # Found lone \r
1754 endpos = crpos + 1
1755 break
Guido van Rossum8358db22007-08-18 21:39:55 +00001756 elif nlpos < crpos:
1757 # Found \n
Amaury Forgeot d'Arc1ff99102007-11-19 20:34:10 +00001758 endpos = nlpos + 1
Guido van Rossum78892e42007-04-06 17:31:18 +00001759 break
Guido van Rossum8358db22007-08-18 21:39:55 +00001760 elif nlpos == crpos + 1:
1761 # Found \r\n
Amaury Forgeot d'Arc1ff99102007-11-19 20:34:10 +00001762 endpos = crpos + 2
Guido van Rossum8358db22007-08-18 21:39:55 +00001763 break
1764 else:
1765 # Found \r
Amaury Forgeot d'Arc1ff99102007-11-19 20:34:10 +00001766 endpos = crpos + 1
Guido van Rossum8358db22007-08-18 21:39:55 +00001767 break
Guido van Rossum78892e42007-04-06 17:31:18 +00001768 else:
Guido van Rossum8358db22007-08-18 21:39:55 +00001769 # non-universal
1770 pos = line.find(self._readnl)
1771 if pos >= 0:
Amaury Forgeot d'Arc1ff99102007-11-19 20:34:10 +00001772 endpos = pos + len(self._readnl)
Guido van Rossum8358db22007-08-18 21:39:55 +00001773 break
Guido van Rossum78892e42007-04-06 17:31:18 +00001774
Ka-Ping Yeedbe28e52008-03-20 10:34:07 +00001775 if limit >= 0 and len(line) >= limit:
1776 endpos = limit # reached length limit
1777 break
1778
Guido van Rossum78892e42007-04-06 17:31:18 +00001779 # No line ending seen yet - get more data
Guido van Rossum8358db22007-08-18 21:39:55 +00001780 more_line = ''
Ka-Ping Yeedbe28e52008-03-20 10:34:07 +00001781 while self._read_chunk():
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001782 if self._decoded_chars:
Guido van Rossum78892e42007-04-06 17:31:18 +00001783 break
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001784 if self._decoded_chars:
1785 line += self._get_decoded_chars()
Guido van Rossum8358db22007-08-18 21:39:55 +00001786 else:
1787 # end of file
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001788 self._set_decoded_chars('')
1789 self._snapshot = None
Amaury Forgeot d'Arc1ff99102007-11-19 20:34:10 +00001790 return line
Guido van Rossum78892e42007-04-06 17:31:18 +00001791
Ka-Ping Yeedbe28e52008-03-20 10:34:07 +00001792 if limit >= 0 and endpos > limit:
1793 endpos = limit # don't exceed limit
1794
Ka-Ping Yee593cd6b2008-03-20 10:37:32 +00001795 # Rewind _decoded_chars to just after the line ending we found.
1796 self._rewind_decoded_chars(len(line) - endpos)
Amaury Forgeot d'Arc1ff99102007-11-19 20:34:10 +00001797 return line[:endpos]
Guido van Rossum024da5c2007-05-17 23:59:11 +00001798
Guido van Rossum8358db22007-08-18 21:39:55 +00001799 @property
1800 def newlines(self):
Amaury Forgeot d'Arc1ff99102007-11-19 20:34:10 +00001801 return self._decoder.newlines if self._decoder else None
Guido van Rossum024da5c2007-05-17 23:59:11 +00001802
Alexandre Vassalotti794652d2008-06-11 22:58:36 +00001803class _StringIO(TextIOWrapper):
1804 """Text I/O implementation using an in-memory buffer.
1805
1806 The initial_value argument sets the value of object. The newline
1807 argument is like the one of TextIOWrapper's constructor.
Benjamin Peterson2c5f8282008-04-13 00:27:46 +00001808 """
Guido van Rossum024da5c2007-05-17 23:59:11 +00001809
1810 # XXX This is really slow, but fully functional
1811
Alexandre Vassalotti794652d2008-06-11 22:58:36 +00001812 def __init__(self, initial_value="", newline="\n"):
1813 super(_StringIO, self).__init__(BytesIO(),
1814 encoding="utf-8",
1815 errors="strict",
1816 newline=newline)
Guido van Rossum024da5c2007-05-17 23:59:11 +00001817 if initial_value:
Guido van Rossum3172c5d2007-10-16 18:12:55 +00001818 if not isinstance(initial_value, str):
Guido van Rossum34d19282007-08-09 01:03:29 +00001819 initial_value = str(initial_value)
Guido van Rossum024da5c2007-05-17 23:59:11 +00001820 self.write(initial_value)
1821 self.seek(0)
1822
1823 def getvalue(self):
Guido van Rossum34d19282007-08-09 01:03:29 +00001824 self.flush()
Guido van Rossume7fc50f2007-12-03 22:54:21 +00001825 return self.buffer.getvalue().decode(self._encoding, self._errors)
Alexandre Vassalotti794652d2008-06-11 22:58:36 +00001826
1827try:
1828 import _stringio
1829
1830 # This subclass is a reimplementation of the TextIOWrapper
1831 # interface without any of its text decoding facilities. All the
1832 # stored data is manipulated with the efficient
1833 # _stringio._StringIO extension type. Also, the newline decoding
1834 # mechanism of IncrementalNewlineDecoder is reimplemented here for
1835 # efficiency. Doing otherwise, would require us to implement a
1836 # fake decoder which would add an additional and unnecessary layer
1837 # on top of the _StringIO methods.
1838
1839 class StringIO(_stringio._StringIO, TextIOBase):
1840 """Text I/O implementation using an in-memory buffer.
1841
1842 The initial_value argument sets the value of object. The newline
1843 argument is like the one of TextIOWrapper's constructor.
1844 """
1845
1846 _CHUNK_SIZE = 4096
1847
1848 def __init__(self, initial_value="", newline="\n"):
1849 if newline not in (None, "", "\n", "\r", "\r\n"):
1850 raise ValueError("illegal newline value: %r" % (newline,))
1851
1852 self._readuniversal = not newline
1853 self._readtranslate = newline is None
1854 self._readnl = newline
1855 self._writetranslate = newline != ""
1856 self._writenl = newline or os.linesep
1857 self._pending = ""
1858 self._seennl = 0
1859
1860 # Reset the buffer first, in case __init__ is called
1861 # multiple times.
1862 self.truncate(0)
1863 if initial_value is None:
1864 initial_value = ""
1865 self.write(initial_value)
1866 self.seek(0)
1867
1868 @property
1869 def buffer(self):
1870 raise UnsupportedOperation("%s.buffer attribute is unsupported" %
1871 self.__class__.__name__)
1872
Alexandre Vassalotti3ade6f92008-06-12 01:13:54 +00001873 # XXX Cruft to support the TextIOWrapper API. This would only
1874 # be meaningful if StringIO supported the buffer attribute.
1875 # Hopefully, a better solution, than adding these pseudo-attributes,
1876 # will be found.
1877 @property
1878 def encoding(self):
1879 return "utf-8"
1880
1881 @property
1882 def errors(self):
1883 return "strict"
1884
1885 @property
1886 def line_buffering(self):
1887 return False
1888
Alexandre Vassalotti794652d2008-06-11 22:58:36 +00001889 def _decode_newlines(self, input, final=False):
1890 # decode input (with the eventual \r from a previous pass)
1891 if self._pending:
1892 input = self._pending + input
1893
1894 # retain last \r even when not translating data:
1895 # then readline() is sure to get \r\n in one pass
1896 if input.endswith("\r") and not final:
1897 input = input[:-1]
1898 self._pending = "\r"
1899 else:
1900 self._pending = ""
1901
1902 # Record which newlines are read
1903 crlf = input.count('\r\n')
1904 cr = input.count('\r') - crlf
1905 lf = input.count('\n') - crlf
1906 self._seennl |= (lf and self._LF) | (cr and self._CR) \
1907 | (crlf and self._CRLF)
1908
1909 if self._readtranslate:
1910 if crlf:
1911 output = input.replace("\r\n", "\n")
1912 if cr:
1913 output = input.replace("\r", "\n")
1914 else:
1915 output = input
1916
1917 return output
1918
1919 def writable(self):
1920 return True
1921
1922 def readable(self):
1923 return True
1924
1925 def seekable(self):
1926 return True
1927
1928 _read = _stringio._StringIO.read
1929 _write = _stringio._StringIO.write
1930 _tell = _stringio._StringIO.tell
1931 _seek = _stringio._StringIO.seek
1932 _truncate = _stringio._StringIO.truncate
1933 _getvalue = _stringio._StringIO.getvalue
1934
1935 def getvalue(self) -> str:
1936 """Retrieve the entire contents of the object."""
1937 if self.closed:
1938 raise ValueError("read on closed file")
1939 return self._getvalue()
1940
1941 def write(self, s: str) -> int:
1942 """Write string s to file.
1943
1944 Returns the number of characters written.
1945 """
1946 if self.closed:
1947 raise ValueError("write to closed file")
1948 if not isinstance(s, str):
1949 raise TypeError("can't write %s to text stream" %
1950 s.__class__.__name__)
1951 length = len(s)
1952 if self._writetranslate and self._writenl != "\n":
1953 s = s.replace("\n", self._writenl)
1954 self._pending = ""
1955 self._write(s)
1956 return length
1957
1958 def read(self, n: int = None) -> str:
1959 """Read at most n characters, returned as a string.
1960
1961 If the argument is negative or omitted, read until EOF
1962 is reached. Return an empty string at EOF.
1963 """
1964 if self.closed:
1965 raise ValueError("read to closed file")
1966 if n is None:
1967 n = -1
1968 res = self._pending
1969 if n < 0:
1970 res += self._decode_newlines(self._read(), True)
1971 self._pending = ""
1972 return res
1973 else:
1974 res = self._decode_newlines(self._read(n), True)
1975 self._pending = res[n:]
1976 return res[:n]
1977
1978 def tell(self) -> int:
1979 """Tell the current file position."""
1980 if self.closed:
1981 raise ValueError("tell from closed file")
1982 if self._pending:
1983 return self._tell() - len(self._pending)
1984 else:
1985 return self._tell()
1986
1987 def seek(self, pos: int = None, whence: int = 0) -> int:
1988 """Change stream position.
1989
1990 Seek to character offset pos relative to position indicated by whence:
1991 0 Start of stream (the default). pos should be >= 0;
1992 1 Current position - pos must be 0;
1993 2 End of stream - pos must be 0.
1994 Returns the new absolute position.
1995 """
1996 if self.closed:
1997 raise ValueError("seek from closed file")
1998 self._pending = ""
1999 return self._seek(pos, whence)
2000
2001 def truncate(self, pos: int = None) -> int:
2002 """Truncate size to pos.
2003
2004 The pos argument defaults to the current file position, as
2005 returned by tell(). Imply an absolute seek to pos.
2006 Returns the new absolute position.
2007 """
2008 if self.closed:
2009 raise ValueError("truncate from closed file")
2010 self._pending = ""
2011 return self._truncate(pos)
2012
2013 def readline(self, limit: int = None) -> str:
2014 if self.closed:
2015 raise ValueError("read from closed file")
2016 if limit is None:
2017 limit = -1
2018 if limit >= 0:
2019 # XXX: Hack to support limit argument, for backwards
2020 # XXX compatibility
2021 line = self.readline()
2022 if len(line) <= limit:
2023 return line
2024 line, self._pending = line[:limit], line[limit:] + self._pending
2025 return line
2026
2027 line = self._pending
2028 self._pending = ""
2029
2030 start = 0
2031 pos = endpos = None
2032 while True:
2033 if self._readtranslate:
2034 # Newlines are already translated, only search for \n
2035 pos = line.find('\n', start)
2036 if pos >= 0:
2037 endpos = pos + 1
2038 break
2039 else:
2040 start = len(line)
2041
2042 elif self._readuniversal:
2043 # Universal newline search. Find any of \r, \r\n, \n
2044 # The decoder ensures that \r\n are not split in two pieces
2045
2046 # In C we'd look for these in parallel of course.
2047 nlpos = line.find("\n", start)
2048 crpos = line.find("\r", start)
2049 if crpos == -1:
2050 if nlpos == -1:
2051 # Nothing found
2052 start = len(line)
2053 else:
2054 # Found \n
2055 endpos = nlpos + 1
2056 break
2057 elif nlpos == -1:
2058 # Found lone \r
2059 endpos = crpos + 1
2060 break
2061 elif nlpos < crpos:
2062 # Found \n
2063 endpos = nlpos + 1
2064 break
2065 elif nlpos == crpos + 1:
2066 # Found \r\n
2067 endpos = crpos + 2
2068 break
2069 else:
2070 # Found \r
2071 endpos = crpos + 1
2072 break
2073 else:
2074 # non-universal
2075 pos = line.find(self._readnl)
2076 if pos >= 0:
2077 endpos = pos + len(self._readnl)
2078 break
2079
2080 # No line ending seen yet - get more data
2081 more_line = self.read(self._CHUNK_SIZE)
2082 if more_line:
2083 line += more_line
2084 else:
2085 # end of file
2086 return line
2087
2088 self._pending = line[endpos:]
2089 return line[:endpos]
2090
2091 _LF = 1
2092 _CR = 2
2093 _CRLF = 4
2094
2095 @property
2096 def newlines(self):
2097 return (None,
2098 "\n",
2099 "\r",
2100 ("\r", "\n"),
2101 "\r\n",
2102 ("\n", "\r\n"),
2103 ("\r", "\r\n"),
2104 ("\r", "\n", "\r\n")
2105 )[self._seennl]
2106
2107
2108except ImportError:
2109 StringIO = _StringIO