blob: 2458e404fdd9e5ba5dbeae74b2149ffdcd393d8d [file] [log] [blame]
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001"""
2Python implementation of the io module.
3"""
4
5import os
6import abc
7import codecs
Benjamin Peterson59406a92009-03-26 17:10:29 +00008import warnings
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00009# Import _thread instead of threading to reduce startup cost
10try:
11 from _thread import allocate_lock as Lock
12except ImportError:
13 from _dummy_thread import allocate_lock as Lock
14
15import io
16from io import __all__
Benjamin Peterson8d5fd4e2009-04-02 01:03:26 +000017from io import SEEK_SET, SEEK_CUR, SEEK_END
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000018
19# open() uses st_blksize whenever we can
20DEFAULT_BUFFER_SIZE = 8 * 1024 # bytes
21
22# NOTE: Base classes defined here are registered with the "official" ABCs
23# defined in io.py. We don't use real inheritance though, because we don't
24# want to inherit the C implementations.
25
26
27class BlockingIOError(IOError):
28
29 """Exception raised when I/O would block on a non-blocking I/O stream."""
30
31 def __init__(self, errno, strerror, characters_written=0):
32 super().__init__(errno, strerror)
33 if not isinstance(characters_written, int):
34 raise TypeError("characters_written must be a integer")
35 self.characters_written = characters_written
36
37
Benjamin Peterson9990e8c2009-04-18 14:47:50 +000038def open(file: (str, bytes), mode: str = "r", buffering: int = None,
39 encoding: str = None, errors: str = None,
40 newline: str = None, closefd: bool = True) -> "IOBase":
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000041
42 r"""Open file and return a stream. Raise IOError upon failure.
43
44 file is either a text or byte string giving the name (and the path
45 if the file isn't in the current working directory) of the file to
46 be opened or an integer file descriptor of the file to be
47 wrapped. (If a file descriptor is given, it is closed when the
48 returned I/O object is closed, unless closefd is set to False.)
49
50 mode is an optional string that specifies the mode in which the file
51 is opened. It defaults to 'r' which means open for reading in text
52 mode. Other common values are 'w' for writing (truncating the file if
53 it already exists), and 'a' for appending (which on some Unix systems,
54 means that all writes append to the end of the file regardless of the
55 current seek position). In text mode, if encoding is not specified the
56 encoding used is platform dependent. (For reading and writing raw
57 bytes use binary mode and leave encoding unspecified.) The available
58 modes are:
59
60 ========= ===============================================================
61 Character Meaning
62 --------- ---------------------------------------------------------------
63 'r' open for reading (default)
64 'w' open for writing, truncating the file first
65 'a' open for writing, appending to the end of the file if it exists
66 'b' binary mode
67 't' text mode (default)
68 '+' open a disk file for updating (reading and writing)
69 'U' universal newline mode (for backwards compatibility; unneeded
70 for new code)
71 ========= ===============================================================
72
73 The default mode is 'rt' (open for reading text). For binary random
74 access, the mode 'w+b' opens and truncates the file to 0 bytes, while
75 'r+b' opens the file without truncation.
76
77 Python distinguishes between files opened in binary and text modes,
78 even when the underlying operating system doesn't. Files opened in
79 binary mode (appending 'b' to the mode argument) return contents as
80 bytes objects without any decoding. In text mode (the default, or when
81 't' is appended to the mode argument), the contents of the file are
82 returned as strings, the bytes having been first decoded using a
83 platform-dependent encoding or using the specified encoding if given.
84
Antoine Pitrou45a43722009-12-19 21:09:58 +000085 buffering is an optional integer used to set the buffering policy.
86 Pass 0 to switch buffering off (only allowed in binary mode), 1 to select
87 line buffering (only usable in text mode), and an integer > 1 to indicate
88 the size of a fixed-size chunk buffer. When no buffering argument is
89 given, the default buffering policy works as follows:
90
91 * Binary files are buffered in fixed-size chunks; the size of the buffer
92 is chosen using a heuristic trying to determine the underlying device's
93 "block size" and falling back on `io.DEFAULT_BUFFER_SIZE`.
94 On many systems, the buffer will typically be 4096 or 8192 bytes long.
95
96 * "Interactive" text files (files for which isatty() returns True)
97 use line buffering. Other text files use the policy described above
98 for binary files.
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +000099
100 encoding is the name of the encoding used to decode or encode the
101 file. This should only be used in text mode. The default encoding is
102 platform dependent, but any encoding supported by Python can be
103 passed. See the codecs module for the list of supported encodings.
104
105 errors is an optional string that specifies how encoding errors are to
106 be handled---this argument should not be used in binary mode. Pass
107 'strict' to raise a ValueError exception if there is an encoding error
108 (the default of None has the same effect), or pass 'ignore' to ignore
109 errors. (Note that ignoring encoding errors can lead to data loss.)
110 See the documentation for codecs.register for a list of the permitted
111 encoding error strings.
112
113 newline controls how universal newlines works (it only applies to text
114 mode). It can be None, '', '\n', '\r', and '\r\n'. It works as
115 follows:
116
117 * On input, if newline is None, universal newlines mode is
118 enabled. Lines in the input can end in '\n', '\r', or '\r\n', and
119 these are translated into '\n' before being returned to the
120 caller. If it is '', universal newline mode is enabled, but line
121 endings are returned to the caller untranslated. If it has any of
122 the other legal values, input lines are only terminated by the given
123 string, and the line ending is returned to the caller untranslated.
124
125 * On output, if newline is None, any '\n' characters written are
126 translated to the system default line separator, os.linesep. If
127 newline is '', no translation takes place. If newline is any of the
128 other legal values, any '\n' characters written are translated to
129 the given string.
130
131 If closefd is False, the underlying file descriptor will be kept open
132 when the file is closed. This does not work when a file name is given
133 and must be True in that case.
134
135 open() returns a file object whose type depends on the mode, and
136 through which the standard file operations such as reading and writing
137 are performed. When open() is used to open a file in a text mode ('w',
138 'r', 'wt', 'rt', etc.), it returns a TextIOWrapper. When used to open
139 a file in a binary mode, the returned class varies: in read binary
140 mode, it returns a BufferedReader; in write binary and append binary
141 modes, it returns a BufferedWriter, and in read/write mode, it returns
142 a BufferedRandom.
143
144 It is also possible to use a string or bytearray as a file for both
145 reading and writing. For strings StringIO can be used like a file
146 opened in a text mode, and for bytes a BytesIO can be used like a file
147 opened in a binary mode.
148 """
149 if not isinstance(file, (str, bytes, int)):
150 raise TypeError("invalid file: %r" % file)
151 if not isinstance(mode, str):
152 raise TypeError("invalid mode: %r" % mode)
153 if buffering is not None and not isinstance(buffering, int):
154 raise TypeError("invalid buffering: %r" % buffering)
155 if encoding is not None and not isinstance(encoding, str):
156 raise TypeError("invalid encoding: %r" % encoding)
157 if errors is not None and not isinstance(errors, str):
158 raise TypeError("invalid errors: %r" % errors)
159 modes = set(mode)
160 if modes - set("arwb+tU") or len(mode) > len(modes):
161 raise ValueError("invalid mode: %r" % mode)
162 reading = "r" in modes
163 writing = "w" in modes
164 appending = "a" in modes
165 updating = "+" in modes
166 text = "t" in modes
167 binary = "b" in modes
168 if "U" in modes:
169 if writing or appending:
170 raise ValueError("can't use U and writing mode at once")
171 reading = True
172 if text and binary:
173 raise ValueError("can't have text and binary mode at once")
174 if reading + writing + appending > 1:
175 raise ValueError("can't have read/write/append mode at once")
176 if not (reading or writing or appending):
177 raise ValueError("must have exactly one of read/write/append mode")
178 if binary and encoding is not None:
179 raise ValueError("binary mode doesn't take an encoding argument")
180 if binary and errors is not None:
181 raise ValueError("binary mode doesn't take an errors argument")
182 if binary and newline is not None:
183 raise ValueError("binary mode doesn't take a newline argument")
184 raw = FileIO(file,
185 (reading and "r" or "") +
186 (writing and "w" or "") +
187 (appending and "a" or "") +
188 (updating and "+" or ""),
189 closefd)
190 if buffering is None:
191 buffering = -1
192 line_buffering = False
193 if buffering == 1 or buffering < 0 and raw.isatty():
194 buffering = -1
195 line_buffering = True
196 if buffering < 0:
197 buffering = DEFAULT_BUFFER_SIZE
198 try:
199 bs = os.fstat(raw.fileno()).st_blksize
200 except (os.error, AttributeError):
201 pass
202 else:
203 if bs > 1:
204 buffering = bs
205 if buffering < 0:
206 raise ValueError("invalid buffering size")
207 if buffering == 0:
208 if binary:
209 return raw
210 raise ValueError("can't have unbuffered text I/O")
211 if updating:
212 buffer = BufferedRandom(raw, buffering)
213 elif writing or appending:
214 buffer = BufferedWriter(raw, buffering)
215 elif reading:
216 buffer = BufferedReader(raw, buffering)
217 else:
218 raise ValueError("unknown mode: %r" % mode)
219 if binary:
220 return buffer
221 text = TextIOWrapper(buffer, encoding, errors, newline, line_buffering)
222 text.mode = mode
223 return text
224
225
226class DocDescriptor:
227 """Helper for builtins.open.__doc__
228 """
229 def __get__(self, obj, typ):
230 return (
231 "open(file, mode='r', buffering=None, encoding=None, "
232 "errors=None, newline=None, closefd=True)\n\n" +
233 open.__doc__)
234
235class OpenWrapper:
236 """Wrapper for builtins.open
237
238 Trick so that open won't become a bound method when stored
239 as a class variable (as dbm.dumb does).
240
241 See initstdio() in Python/pythonrun.c.
242 """
243 __doc__ = DocDescriptor()
244
245 def __new__(cls, *args, **kwargs):
246 return open(*args, **kwargs)
247
248
249class UnsupportedOperation(ValueError, IOError):
250 pass
251
252
253class IOBase(metaclass=abc.ABCMeta):
254
255 """The abstract base class for all I/O classes, acting on streams of
256 bytes. There is no public constructor.
257
258 This class provides dummy implementations for many methods that
259 derived classes can override selectively; the default implementations
260 represent a file that cannot be read, written or seeked.
261
262 Even though IOBase does not declare read, readinto, or write because
263 their signatures will vary, implementations and clients should
264 consider those methods part of the interface. Also, implementations
265 may raise a IOError when operations they do not support are called.
266
267 The basic type used for binary data read from or written to a file is
268 bytes. bytearrays are accepted too, and in some cases (such as
269 readinto) needed. Text I/O classes work with str data.
270
271 Note that calling any method (even inquiries) on a closed stream is
272 undefined. Implementations may raise IOError in this case.
273
274 IOBase (and its subclasses) support the iterator protocol, meaning
275 that an IOBase object can be iterated over yielding the lines in a
276 stream.
277
278 IOBase also supports the :keyword:`with` statement. In this example,
279 fp is closed after the suite of the with statement is complete:
280
281 with open('spam.txt', 'r') as fp:
282 fp.write('Spam and eggs!')
283 """
284
285 ### Internal ###
286
287 def _unsupported(self, name: str) -> IOError:
288 """Internal: raise an exception for unsupported operations."""
289 raise UnsupportedOperation("%s.%s() not supported" %
290 (self.__class__.__name__, name))
291
292 ### Positioning ###
293
294 def seek(self, pos: int, whence: int = 0) -> int:
295 """Change stream position.
296
297 Change the stream position to byte offset offset. offset is
298 interpreted relative to the position indicated by whence. Values
299 for whence are:
300
301 * 0 -- start of stream (the default); offset should be zero or positive
302 * 1 -- current stream position; offset may be negative
303 * 2 -- end of stream; offset is usually negative
304
305 Return the new absolute position.
306 """
307 self._unsupported("seek")
308
309 def tell(self) -> int:
310 """Return current stream position."""
311 return self.seek(0, 1)
312
313 def truncate(self, pos: int = None) -> int:
314 """Truncate file to size bytes.
315
316 Size defaults to the current IO position as reported by tell(). Return
317 the new size.
318 """
319 self._unsupported("truncate")
320
321 ### Flush and close ###
322
323 def flush(self) -> None:
324 """Flush write buffers, if applicable.
325
326 This is not implemented for read-only and non-blocking streams.
327 """
328 # XXX Should this return the number of bytes written???
329
330 __closed = False
331
332 def close(self) -> None:
333 """Flush and close the IO object.
334
335 This method has no effect if the file is already closed.
336 """
337 if not self.__closed:
338 try:
339 self.flush()
340 except IOError:
341 pass # If flush() fails, just give up
342 self.__closed = True
343
344 def __del__(self) -> None:
345 """Destructor. Calls close()."""
346 # The try/except block is in case this is called at program
347 # exit time, when it's possible that globals have already been
348 # deleted, and then the close() call might fail. Since
349 # there's nothing we can do about such failures and they annoy
350 # the end users, we suppress the traceback.
351 try:
352 self.close()
353 except:
354 pass
355
356 ### Inquiries ###
357
358 def seekable(self) -> bool:
359 """Return whether object supports random access.
360
361 If False, seek(), tell() and truncate() will raise IOError.
362 This method may need to do a test seek().
363 """
364 return False
365
366 def _checkSeekable(self, msg=None):
367 """Internal: raise an IOError if file is not seekable
368 """
369 if not self.seekable():
370 raise IOError("File or stream is not seekable."
371 if msg is None else msg)
372
373
374 def readable(self) -> bool:
375 """Return whether object was opened for reading.
376
377 If False, read() will raise IOError.
378 """
379 return False
380
381 def _checkReadable(self, msg=None):
382 """Internal: raise an IOError if file is not readable
383 """
384 if not self.readable():
385 raise IOError("File or stream is not readable."
386 if msg is None else msg)
387
388 def writable(self) -> bool:
389 """Return whether object was opened for writing.
390
391 If False, write() and truncate() will raise IOError.
392 """
393 return False
394
395 def _checkWritable(self, msg=None):
396 """Internal: raise an IOError if file is not writable
397 """
398 if not self.writable():
399 raise IOError("File or stream is not writable."
400 if msg is None else msg)
401
402 @property
403 def closed(self):
404 """closed: bool. True iff the file has been closed.
405
406 For backwards compatibility, this is a property, not a predicate.
407 """
408 return self.__closed
409
410 def _checkClosed(self, msg=None):
411 """Internal: raise an ValueError if file is closed
412 """
413 if self.closed:
414 raise ValueError("I/O operation on closed file."
415 if msg is None else msg)
416
417 ### Context manager ###
418
419 def __enter__(self) -> "IOBase": # That's a forward reference
420 """Context management protocol. Returns self."""
421 self._checkClosed()
422 return self
423
424 def __exit__(self, *args) -> None:
425 """Context management protocol. Calls close()"""
426 self.close()
427
428 ### Lower-level APIs ###
429
430 # XXX Should these be present even if unimplemented?
431
432 def fileno(self) -> int:
433 """Returns underlying file descriptor if one exists.
434
435 An IOError is raised if the IO object does not use a file descriptor.
436 """
437 self._unsupported("fileno")
438
439 def isatty(self) -> bool:
440 """Return whether this is an 'interactive' stream.
441
442 Return False if it can't be determined.
443 """
444 self._checkClosed()
445 return False
446
447 ### Readline[s] and writelines ###
448
449 def readline(self, limit: int = -1) -> bytes:
450 r"""Read and return a line from the stream.
451
452 If limit is specified, at most limit bytes will be read.
453
454 The line terminator is always b'\n' for binary files; for text
455 files, the newlines argument to open can be used to select the line
456 terminator(s) recognized.
457 """
458 # For backwards compatibility, a (slowish) readline().
459 if hasattr(self, "peek"):
460 def nreadahead():
461 readahead = self.peek(1)
462 if not readahead:
463 return 1
464 n = (readahead.find(b"\n") + 1) or len(readahead)
465 if limit >= 0:
466 n = min(n, limit)
467 return n
468 else:
469 def nreadahead():
470 return 1
471 if limit is None:
472 limit = -1
Benjamin Petersonb01138a2009-04-24 22:59:52 +0000473 elif not isinstance(limit, int):
474 raise TypeError("limit must be an integer")
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000475 res = bytearray()
476 while limit < 0 or len(res) < limit:
477 b = self.read(nreadahead())
478 if not b:
479 break
480 res += b
481 if res.endswith(b"\n"):
482 break
483 return bytes(res)
484
485 def __iter__(self):
486 self._checkClosed()
487 return self
488
489 def __next__(self):
490 line = self.readline()
491 if not line:
492 raise StopIteration
493 return line
494
495 def readlines(self, hint=None):
496 """Return a list of lines from the stream.
497
498 hint can be specified to control the number of lines read: no more
499 lines will be read if the total size (in bytes/characters) of all
500 lines so far exceeds hint.
501 """
502 if hint is None or hint <= 0:
503 return list(self)
504 n = 0
505 lines = []
506 for line in self:
507 lines.append(line)
508 n += len(line)
509 if n >= hint:
510 break
511 return lines
512
513 def writelines(self, lines):
514 self._checkClosed()
515 for line in lines:
516 self.write(line)
517
518io.IOBase.register(IOBase)
519
520
521class RawIOBase(IOBase):
522
523 """Base class for raw binary I/O."""
524
525 # The read() method is implemented by calling readinto(); derived
526 # classes that want to support read() only need to implement
527 # readinto() as a primitive operation. In general, readinto() can be
528 # more efficient than read().
529
530 # (It would be tempting to also provide an implementation of
531 # readinto() in terms of read(), in case the latter is a more suitable
532 # primitive operation, but that would lead to nasty recursion in case
533 # a subclass doesn't implement either.)
534
535 def read(self, n: int = -1) -> bytes:
536 """Read and return up to n bytes.
537
538 Returns an empty bytes object on EOF, or None if the object is
539 set not to block and has no data to read.
540 """
541 if n is None:
542 n = -1
543 if n < 0:
544 return self.readall()
545 b = bytearray(n.__index__())
546 n = self.readinto(b)
547 del b[n:]
548 return bytes(b)
549
550 def readall(self):
551 """Read until EOF, using multiple read() call."""
552 res = bytearray()
553 while True:
554 data = self.read(DEFAULT_BUFFER_SIZE)
555 if not data:
556 break
557 res += data
558 return bytes(res)
559
560 def readinto(self, b: bytearray) -> int:
561 """Read up to len(b) bytes into b.
562
563 Returns number of bytes read (0 for EOF), or None if the object
564 is set not to block as has no data to read.
565 """
566 self._unsupported("readinto")
567
568 def write(self, b: bytes) -> int:
569 """Write the given buffer to the IO stream.
570
571 Returns the number of bytes written, which may be less than len(b).
572 """
573 self._unsupported("write")
574
575io.RawIOBase.register(RawIOBase)
576from _io import FileIO
577RawIOBase.register(FileIO)
578
579
580class BufferedIOBase(IOBase):
581
582 """Base class for buffered IO objects.
583
584 The main difference with RawIOBase is that the read() method
585 supports omitting the size argument, and does not have a default
586 implementation that defers to readinto().
587
588 In addition, read(), readinto() and write() may raise
589 BlockingIOError if the underlying raw stream is in non-blocking
590 mode and not ready; unlike their raw counterparts, they will never
591 return None.
592
593 A typical implementation should not inherit from a RawIOBase
594 implementation, but wrap one.
595 """
596
597 def read(self, n: int = None) -> bytes:
598 """Read and return up to n bytes.
599
600 If the argument is omitted, None, or negative, reads and
601 returns all data until EOF.
602
603 If the argument is positive, and the underlying raw stream is
604 not 'interactive', multiple raw reads may be issued to satisfy
605 the byte count (unless EOF is reached first). But for
606 interactive raw streams (XXX and for pipes?), at most one raw
607 read will be issued, and a short result does not imply that
608 EOF is imminent.
609
610 Returns an empty bytes array on EOF.
611
612 Raises BlockingIOError if the underlying raw stream has no
613 data at the moment.
614 """
615 self._unsupported("read")
616
617 def read1(self, n: int=None) -> bytes:
618 """Read up to n bytes with at most one read() system call."""
619 self._unsupported("read1")
620
621 def readinto(self, b: bytearray) -> int:
622 """Read up to len(b) bytes into b.
623
624 Like read(), this may issue multiple reads to the underlying raw
625 stream, unless the latter is 'interactive'.
626
627 Returns the number of bytes read (0 for EOF).
628
629 Raises BlockingIOError if the underlying raw stream has no
630 data at the moment.
631 """
632 # XXX This ought to work with anything that supports the buffer API
633 data = self.read(len(b))
634 n = len(data)
635 try:
636 b[:n] = data
637 except TypeError as err:
638 import array
639 if not isinstance(b, array.array):
640 raise err
641 b[:n] = array.array('b', data)
642 return n
643
644 def write(self, b: bytes) -> int:
645 """Write the given buffer to the IO stream.
646
647 Return the number of bytes written, which is never less than
648 len(b).
649
650 Raises BlockingIOError if the buffer is full and the
651 underlying raw stream cannot accept more data at the moment.
652 """
653 self._unsupported("write")
654
Benjamin Petersond2e0c792009-05-01 20:40:59 +0000655 def detach(self) -> None:
656 """
657 Separate the underlying raw stream from the buffer and return it.
658
659 After the raw stream has been detached, the buffer is in an unusable
660 state.
661 """
662 self._unsupported("detach")
663
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000664io.BufferedIOBase.register(BufferedIOBase)
665
666
667class _BufferedIOMixin(BufferedIOBase):
668
669 """A mixin implementation of BufferedIOBase with an underlying raw stream.
670
671 This passes most requests on to the underlying raw stream. It
672 does *not* provide implementations of read(), readinto() or
673 write().
674 """
675
676 def __init__(self, raw):
677 self.raw = raw
678
679 ### Positioning ###
680
681 def seek(self, pos, whence=0):
682 new_position = self.raw.seek(pos, whence)
683 if new_position < 0:
684 raise IOError("seek() returned an invalid position")
685 return new_position
686
687 def tell(self):
688 pos = self.raw.tell()
689 if pos < 0:
690 raise IOError("tell() returned an invalid position")
691 return pos
692
693 def truncate(self, pos=None):
694 # Flush the stream. We're mixing buffered I/O with lower-level I/O,
695 # and a flush may be necessary to synch both views of the current
696 # file state.
697 self.flush()
698
699 if pos is None:
700 pos = self.tell()
701 # XXX: Should seek() be used, instead of passing the position
702 # XXX directly to truncate?
703 return self.raw.truncate(pos)
704
705 ### Flush and close ###
706
707 def flush(self):
708 self.raw.flush()
709
710 def close(self):
Benjamin Petersond2e0c792009-05-01 20:40:59 +0000711 if not self.closed and self.raw is not None:
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000712 try:
713 self.flush()
714 except IOError:
715 pass # If flush() fails, just give up
716 self.raw.close()
717
Benjamin Petersond2e0c792009-05-01 20:40:59 +0000718 def detach(self):
719 if self.raw is None:
720 raise ValueError("raw stream already detached")
721 self.flush()
722 raw = self.raw
723 self.raw = None
724 return raw
725
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000726 ### Inquiries ###
727
728 def seekable(self):
729 return self.raw.seekable()
730
731 def readable(self):
732 return self.raw.readable()
733
734 def writable(self):
735 return self.raw.writable()
736
737 @property
738 def closed(self):
739 return self.raw.closed
740
741 @property
742 def name(self):
743 return self.raw.name
744
745 @property
746 def mode(self):
747 return self.raw.mode
748
Antoine Pitrou716c4442009-05-23 19:04:03 +0000749 def __repr__(self):
750 clsname = self.__class__.__name__
751 try:
752 name = self.name
753 except AttributeError:
754 return "<_pyio.{0}>".format(clsname)
755 else:
756 return "<_pyio.{0} name={1!r}>".format(clsname, name)
757
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000758 ### Lower-level APIs ###
759
760 def fileno(self):
761 return self.raw.fileno()
762
763 def isatty(self):
764 return self.raw.isatty()
765
766
767class BytesIO(BufferedIOBase):
768
769 """Buffered I/O implementation using an in-memory bytes buffer."""
770
771 def __init__(self, initial_bytes=None):
772 buf = bytearray()
773 if initial_bytes is not None:
774 buf += initial_bytes
775 self._buffer = buf
776 self._pos = 0
777
778 def getvalue(self):
779 """Return the bytes value (contents) of the buffer
780 """
781 if self.closed:
782 raise ValueError("getvalue on closed file")
783 return bytes(self._buffer)
784
785 def read(self, n=None):
786 if self.closed:
787 raise ValueError("read from closed file")
788 if n is None:
789 n = -1
790 if n < 0:
791 n = len(self._buffer)
792 if len(self._buffer) <= self._pos:
793 return b""
794 newpos = min(len(self._buffer), self._pos + n)
795 b = self._buffer[self._pos : newpos]
796 self._pos = newpos
797 return bytes(b)
798
799 def read1(self, n):
800 """This is the same as read.
801 """
802 return self.read(n)
803
804 def write(self, b):
805 if self.closed:
806 raise ValueError("write to closed file")
807 if isinstance(b, str):
808 raise TypeError("can't write str to binary stream")
809 n = len(b)
810 if n == 0:
811 return 0
812 pos = self._pos
813 if pos > len(self._buffer):
814 # Inserts null bytes between the current end of the file
815 # and the new write position.
816 padding = b'\x00' * (pos - len(self._buffer))
817 self._buffer += padding
818 self._buffer[pos:pos + n] = b
819 self._pos += n
820 return n
821
822 def seek(self, pos, whence=0):
823 if self.closed:
824 raise ValueError("seek on closed file")
825 try:
826 pos = pos.__index__()
827 except AttributeError as err:
828 raise TypeError("an integer is required") from err
829 if whence == 0:
830 if pos < 0:
831 raise ValueError("negative seek position %r" % (pos,))
832 self._pos = pos
833 elif whence == 1:
834 self._pos = max(0, self._pos + pos)
835 elif whence == 2:
836 self._pos = max(0, len(self._buffer) + pos)
837 else:
838 raise ValueError("invalid whence value")
839 return self._pos
840
841 def tell(self):
842 if self.closed:
843 raise ValueError("tell on closed file")
844 return self._pos
845
846 def truncate(self, pos=None):
847 if self.closed:
848 raise ValueError("truncate on closed file")
849 if pos is None:
850 pos = self._pos
851 elif pos < 0:
852 raise ValueError("negative truncate position %r" % (pos,))
853 del self._buffer[pos:]
854 return self.seek(pos)
855
856 def readable(self):
857 return True
858
859 def writable(self):
860 return True
861
862 def seekable(self):
863 return True
864
865
866class BufferedReader(_BufferedIOMixin):
867
868 """BufferedReader(raw[, buffer_size])
869
870 A buffer for a readable, sequential BaseRawIO object.
871
872 The constructor creates a BufferedReader for the given readable raw
873 stream and buffer_size. If buffer_size is omitted, DEFAULT_BUFFER_SIZE
874 is used.
875 """
876
877 def __init__(self, raw, buffer_size=DEFAULT_BUFFER_SIZE):
878 """Create a new buffered reader using the given readable raw IO object.
879 """
Antoine Pitroucf4c7492009-04-19 00:09:36 +0000880 if not raw.readable():
881 raise IOError('"raw" argument must be readable.')
882
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +0000883 _BufferedIOMixin.__init__(self, raw)
884 if buffer_size <= 0:
885 raise ValueError("invalid buffer size")
886 self.buffer_size = buffer_size
887 self._reset_read_buf()
888 self._read_lock = Lock()
889
890 def _reset_read_buf(self):
891 self._read_buf = b""
892 self._read_pos = 0
893
894 def read(self, n=None):
895 """Read n bytes.
896
897 Returns exactly n bytes of data unless the underlying raw IO
898 stream reaches EOF or if the call would block in non-blocking
899 mode. If n is negative, read until EOF or until read() would
900 block.
901 """
902 if n is not None and n < -1:
903 raise ValueError("invalid number of bytes to read")
904 with self._read_lock:
905 return self._read_unlocked(n)
906
907 def _read_unlocked(self, n=None):
908 nodata_val = b""
909 empty_values = (b"", None)
910 buf = self._read_buf
911 pos = self._read_pos
912
913 # Special case for when the number of bytes to read is unspecified.
914 if n is None or n == -1:
915 self._reset_read_buf()
916 chunks = [buf[pos:]] # Strip the consumed bytes.
917 current_size = 0
918 while True:
919 # Read until EOF or until read() would block.
920 chunk = self.raw.read()
921 if chunk in empty_values:
922 nodata_val = chunk
923 break
924 current_size += len(chunk)
925 chunks.append(chunk)
926 return b"".join(chunks) or nodata_val
927
928 # The number of bytes to read is specified, return at most n bytes.
929 avail = len(buf) - pos # Length of the available buffered data.
930 if n <= avail:
931 # Fast path: the data to read is fully buffered.
932 self._read_pos += n
933 return buf[pos:pos+n]
934 # Slow path: read from the stream until enough bytes are read,
935 # or until an EOF occurs or until read() would block.
936 chunks = [buf[pos:]]
937 wanted = max(self.buffer_size, n)
938 while avail < n:
939 chunk = self.raw.read(wanted)
940 if chunk in empty_values:
941 nodata_val = chunk
942 break
943 avail += len(chunk)
944 chunks.append(chunk)
945 # n is more then avail only when an EOF occurred or when
946 # read() would have blocked.
947 n = min(n, avail)
948 out = b"".join(chunks)
949 self._read_buf = out[n:] # Save the extra data in the buffer.
950 self._read_pos = 0
951 return out[:n] if out else nodata_val
952
953 def peek(self, n=0):
954 """Returns buffered bytes without advancing the position.
955
956 The argument indicates a desired minimal number of bytes; we
957 do at most one raw read to satisfy it. We never return more
958 than self.buffer_size.
959 """
960 with self._read_lock:
961 return self._peek_unlocked(n)
962
963 def _peek_unlocked(self, n=0):
964 want = min(n, self.buffer_size)
965 have = len(self._read_buf) - self._read_pos
966 if have < want or have <= 0:
967 to_read = self.buffer_size - have
968 current = self.raw.read(to_read)
969 if current:
970 self._read_buf = self._read_buf[self._read_pos:] + current
971 self._read_pos = 0
972 return self._read_buf[self._read_pos:]
973
974 def read1(self, n):
975 """Reads up to n bytes, with at most one read() system call."""
976 # Returns up to n bytes. If at least one byte is buffered, we
977 # only return buffered bytes. Otherwise, we do one raw read.
978 if n < 0:
979 raise ValueError("number of bytes to read must be positive")
980 if n == 0:
981 return b""
982 with self._read_lock:
983 self._peek_unlocked(1)
984 return self._read_unlocked(
985 min(n, len(self._read_buf) - self._read_pos))
986
987 def tell(self):
988 return _BufferedIOMixin.tell(self) - len(self._read_buf) + self._read_pos
989
990 def seek(self, pos, whence=0):
991 if not (0 <= whence <= 2):
992 raise ValueError("invalid whence value")
993 with self._read_lock:
994 if whence == 1:
995 pos -= len(self._read_buf) - self._read_pos
996 pos = _BufferedIOMixin.seek(self, pos, whence)
997 self._reset_read_buf()
998 return pos
999
1000class BufferedWriter(_BufferedIOMixin):
1001
1002 """A buffer for a writeable sequential RawIO object.
1003
1004 The constructor creates a BufferedWriter for the given writeable raw
1005 stream. If the buffer_size is not given, it defaults to
Benjamin Peterson59406a92009-03-26 17:10:29 +00001006 DEFAULT_BUFFER_SIZE.
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001007 """
1008
Benjamin Peterson59406a92009-03-26 17:10:29 +00001009 _warning_stack_offset = 2
1010
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001011 def __init__(self, raw,
1012 buffer_size=DEFAULT_BUFFER_SIZE, max_buffer_size=None):
Antoine Pitroucf4c7492009-04-19 00:09:36 +00001013 if not raw.writable():
1014 raise IOError('"raw" argument must be writable.')
1015
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001016 _BufferedIOMixin.__init__(self, raw)
1017 if buffer_size <= 0:
1018 raise ValueError("invalid buffer size")
Benjamin Peterson59406a92009-03-26 17:10:29 +00001019 if max_buffer_size is not None:
1020 warnings.warn("max_buffer_size is deprecated", DeprecationWarning,
1021 self._warning_stack_offset)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001022 self.buffer_size = buffer_size
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001023 self._write_buf = bytearray()
1024 self._write_lock = Lock()
1025
1026 def write(self, b):
1027 if self.closed:
1028 raise ValueError("write to closed file")
1029 if isinstance(b, str):
1030 raise TypeError("can't write str to binary stream")
1031 with self._write_lock:
1032 # XXX we can implement some more tricks to try and avoid
1033 # partial writes
1034 if len(self._write_buf) > self.buffer_size:
1035 # We're full, so let's pre-flush the buffer
1036 try:
1037 self._flush_unlocked()
1038 except BlockingIOError as e:
1039 # We can't accept anything else.
1040 # XXX Why not just let the exception pass through?
1041 raise BlockingIOError(e.errno, e.strerror, 0)
1042 before = len(self._write_buf)
1043 self._write_buf.extend(b)
1044 written = len(self._write_buf) - before
1045 if len(self._write_buf) > self.buffer_size:
1046 try:
1047 self._flush_unlocked()
1048 except BlockingIOError as e:
Benjamin Peterson394ee002009-03-05 22:33:59 +00001049 if len(self._write_buf) > self.buffer_size:
1050 # We've hit the buffer_size. We have to accept a partial
1051 # write and cut back our buffer.
1052 overage = len(self._write_buf) - self.buffer_size
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001053 written -= overage
Benjamin Peterson394ee002009-03-05 22:33:59 +00001054 self._write_buf = self._write_buf[:self.buffer_size]
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001055 raise BlockingIOError(e.errno, e.strerror, written)
1056 return written
1057
1058 def truncate(self, pos=None):
1059 with self._write_lock:
1060 self._flush_unlocked()
1061 if pos is None:
1062 pos = self.raw.tell()
1063 return self.raw.truncate(pos)
1064
1065 def flush(self):
1066 with self._write_lock:
1067 self._flush_unlocked()
1068
1069 def _flush_unlocked(self):
1070 if self.closed:
1071 raise ValueError("flush of closed file")
1072 written = 0
1073 try:
1074 while self._write_buf:
1075 n = self.raw.write(self._write_buf)
1076 if n > len(self._write_buf) or n < 0:
1077 raise IOError("write() returned incorrect number of bytes")
1078 del self._write_buf[:n]
1079 written += n
1080 except BlockingIOError as e:
1081 n = e.characters_written
1082 del self._write_buf[:n]
1083 written += n
1084 raise BlockingIOError(e.errno, e.strerror, written)
1085
1086 def tell(self):
1087 return _BufferedIOMixin.tell(self) + len(self._write_buf)
1088
1089 def seek(self, pos, whence=0):
1090 if not (0 <= whence <= 2):
1091 raise ValueError("invalid whence")
1092 with self._write_lock:
1093 self._flush_unlocked()
1094 return _BufferedIOMixin.seek(self, pos, whence)
1095
1096
1097class BufferedRWPair(BufferedIOBase):
1098
1099 """A buffered reader and writer object together.
1100
1101 A buffered reader object and buffered writer object put together to
1102 form a sequential IO object that can read and write. This is typically
1103 used with a socket or two-way pipe.
1104
1105 reader and writer are RawIOBase objects that are readable and
1106 writeable respectively. If the buffer_size is omitted it defaults to
Benjamin Peterson59406a92009-03-26 17:10:29 +00001107 DEFAULT_BUFFER_SIZE.
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001108 """
1109
1110 # XXX The usefulness of this (compared to having two separate IO
1111 # objects) is questionable.
1112
1113 def __init__(self, reader, writer,
1114 buffer_size=DEFAULT_BUFFER_SIZE, max_buffer_size=None):
1115 """Constructor.
1116
1117 The arguments are two RawIO instances.
1118 """
Benjamin Peterson59406a92009-03-26 17:10:29 +00001119 if max_buffer_size is not None:
1120 warnings.warn("max_buffer_size is deprecated", DeprecationWarning, 2)
Antoine Pitroucf4c7492009-04-19 00:09:36 +00001121
1122 if not reader.readable():
1123 raise IOError('"reader" argument must be readable.')
1124
1125 if not writer.writable():
1126 raise IOError('"writer" argument must be writable.')
1127
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001128 self.reader = BufferedReader(reader, buffer_size)
Benjamin Peterson59406a92009-03-26 17:10:29 +00001129 self.writer = BufferedWriter(writer, buffer_size)
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001130
1131 def read(self, n=None):
1132 if n is None:
1133 n = -1
1134 return self.reader.read(n)
1135
1136 def readinto(self, b):
1137 return self.reader.readinto(b)
1138
1139 def write(self, b):
1140 return self.writer.write(b)
1141
1142 def peek(self, n=0):
1143 return self.reader.peek(n)
1144
1145 def read1(self, n):
1146 return self.reader.read1(n)
1147
1148 def readable(self):
1149 return self.reader.readable()
1150
1151 def writable(self):
1152 return self.writer.writable()
1153
1154 def flush(self):
1155 return self.writer.flush()
1156
1157 def close(self):
1158 self.writer.close()
1159 self.reader.close()
1160
1161 def isatty(self):
1162 return self.reader.isatty() or self.writer.isatty()
1163
1164 @property
1165 def closed(self):
1166 return self.writer.closed
1167
1168
1169class BufferedRandom(BufferedWriter, BufferedReader):
1170
1171 """A buffered interface to random access streams.
1172
1173 The constructor creates a reader and writer for a seekable stream,
1174 raw, given in the first argument. If the buffer_size is omitted it
Benjamin Peterson59406a92009-03-26 17:10:29 +00001175 defaults to DEFAULT_BUFFER_SIZE.
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001176 """
1177
Benjamin Peterson59406a92009-03-26 17:10:29 +00001178 _warning_stack_offset = 3
1179
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001180 def __init__(self, raw,
1181 buffer_size=DEFAULT_BUFFER_SIZE, max_buffer_size=None):
1182 raw._checkSeekable()
1183 BufferedReader.__init__(self, raw, buffer_size)
1184 BufferedWriter.__init__(self, raw, buffer_size, max_buffer_size)
1185
1186 def seek(self, pos, whence=0):
1187 if not (0 <= whence <= 2):
1188 raise ValueError("invalid whence")
1189 self.flush()
1190 if self._read_buf:
1191 # Undo read ahead.
1192 with self._read_lock:
1193 self.raw.seek(self._read_pos - len(self._read_buf), 1)
1194 # First do the raw seek, then empty the read buffer, so that
1195 # if the raw seek fails, we don't lose buffered data forever.
1196 pos = self.raw.seek(pos, whence)
1197 with self._read_lock:
1198 self._reset_read_buf()
1199 if pos < 0:
1200 raise IOError("seek() returned invalid position")
1201 return pos
1202
1203 def tell(self):
1204 if self._write_buf:
1205 return BufferedWriter.tell(self)
1206 else:
1207 return BufferedReader.tell(self)
1208
1209 def truncate(self, pos=None):
1210 if pos is None:
1211 pos = self.tell()
1212 # Use seek to flush the read buffer.
1213 self.seek(pos)
1214 return BufferedWriter.truncate(self)
1215
1216 def read(self, n=None):
1217 if n is None:
1218 n = -1
1219 self.flush()
1220 return BufferedReader.read(self, n)
1221
1222 def readinto(self, b):
1223 self.flush()
1224 return BufferedReader.readinto(self, b)
1225
1226 def peek(self, n=0):
1227 self.flush()
1228 return BufferedReader.peek(self, n)
1229
1230 def read1(self, n):
1231 self.flush()
1232 return BufferedReader.read1(self, n)
1233
1234 def write(self, b):
1235 if self._read_buf:
1236 # Undo readahead
1237 with self._read_lock:
1238 self.raw.seek(self._read_pos - len(self._read_buf), 1)
1239 self._reset_read_buf()
1240 return BufferedWriter.write(self, b)
1241
1242
1243class TextIOBase(IOBase):
1244
1245 """Base class for text I/O.
1246
1247 This class provides a character and line based interface to stream
1248 I/O. There is no readinto method because Python's character strings
1249 are immutable. There is no public constructor.
1250 """
1251
1252 def read(self, n: int = -1) -> str:
1253 """Read at most n characters from stream.
1254
1255 Read from underlying buffer until we have n characters or we hit EOF.
1256 If n is negative or omitted, read until EOF.
1257 """
1258 self._unsupported("read")
1259
1260 def write(self, s: str) -> int:
1261 """Write string s to stream."""
1262 self._unsupported("write")
1263
1264 def truncate(self, pos: int = None) -> int:
1265 """Truncate size to pos."""
1266 self._unsupported("truncate")
1267
1268 def readline(self) -> str:
1269 """Read until newline or EOF.
1270
1271 Returns an empty string if EOF is hit immediately.
1272 """
1273 self._unsupported("readline")
1274
Benjamin Petersond2e0c792009-05-01 20:40:59 +00001275 def detach(self) -> None:
1276 """
1277 Separate the underlying buffer from the TextIOBase and return it.
1278
1279 After the underlying buffer has been detached, the TextIO is in an
1280 unusable state.
1281 """
1282 self._unsupported("detach")
1283
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001284 @property
1285 def encoding(self):
1286 """Subclasses should override."""
1287 return None
1288
1289 @property
1290 def newlines(self):
1291 """Line endings translated so far.
1292
1293 Only line endings translated during reading are considered.
1294
1295 Subclasses should override.
1296 """
1297 return None
1298
Benjamin Peterson0926ad12009-06-06 18:02:12 +00001299 @property
1300 def errors(self):
1301 """Error setting of the decoder or encoder.
1302
1303 Subclasses should override."""
1304 return None
1305
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001306io.TextIOBase.register(TextIOBase)
1307
1308
1309class IncrementalNewlineDecoder(codecs.IncrementalDecoder):
1310 r"""Codec used when reading a file in universal newlines mode. It wraps
1311 another incremental decoder, translating \r\n and \r into \n. It also
1312 records the types of newlines encountered. When used with
1313 translate=False, it ensures that the newline sequence is returned in
1314 one piece.
1315 """
1316 def __init__(self, decoder, translate, errors='strict'):
1317 codecs.IncrementalDecoder.__init__(self, errors=errors)
1318 self.translate = translate
1319 self.decoder = decoder
1320 self.seennl = 0
1321 self.pendingcr = False
1322
1323 def decode(self, input, final=False):
1324 # decode input (with the eventual \r from a previous pass)
1325 if self.decoder is None:
1326 output = input
1327 else:
1328 output = self.decoder.decode(input, final=final)
1329 if self.pendingcr and (output or final):
1330 output = "\r" + output
1331 self.pendingcr = False
1332
1333 # retain last \r even when not translating data:
1334 # then readline() is sure to get \r\n in one pass
1335 if output.endswith("\r") and not final:
1336 output = output[:-1]
1337 self.pendingcr = True
1338
1339 # Record which newlines are read
1340 crlf = output.count('\r\n')
1341 cr = output.count('\r') - crlf
1342 lf = output.count('\n') - crlf
1343 self.seennl |= (lf and self._LF) | (cr and self._CR) \
1344 | (crlf and self._CRLF)
1345
1346 if self.translate:
1347 if crlf:
1348 output = output.replace("\r\n", "\n")
1349 if cr:
1350 output = output.replace("\r", "\n")
1351
1352 return output
1353
1354 def getstate(self):
1355 if self.decoder is None:
1356 buf = b""
1357 flag = 0
1358 else:
1359 buf, flag = self.decoder.getstate()
1360 flag <<= 1
1361 if self.pendingcr:
1362 flag |= 1
1363 return buf, flag
1364
1365 def setstate(self, state):
1366 buf, flag = state
1367 self.pendingcr = bool(flag & 1)
1368 if self.decoder is not None:
1369 self.decoder.setstate((buf, flag >> 1))
1370
1371 def reset(self):
1372 self.seennl = 0
1373 self.pendingcr = False
1374 if self.decoder is not None:
1375 self.decoder.reset()
1376
1377 _LF = 1
1378 _CR = 2
1379 _CRLF = 4
1380
1381 @property
1382 def newlines(self):
1383 return (None,
1384 "\n",
1385 "\r",
1386 ("\r", "\n"),
1387 "\r\n",
1388 ("\n", "\r\n"),
1389 ("\r", "\r\n"),
1390 ("\r", "\n", "\r\n")
1391 )[self.seennl]
1392
1393
1394class TextIOWrapper(TextIOBase):
1395
1396 r"""Character and line based layer over a BufferedIOBase object, buffer.
1397
1398 encoding gives the name of the encoding that the stream will be
1399 decoded or encoded with. It defaults to locale.getpreferredencoding.
1400
1401 errors determines the strictness of encoding and decoding (see the
1402 codecs.register) and defaults to "strict".
1403
1404 newline can be None, '', '\n', '\r', or '\r\n'. It controls the
1405 handling of line endings. If it is None, universal newlines is
1406 enabled. With this enabled, on input, the lines endings '\n', '\r',
1407 or '\r\n' are translated to '\n' before being returned to the
1408 caller. Conversely, on output, '\n' is translated to the system
1409 default line seperator, os.linesep. If newline is any other of its
1410 legal values, that newline becomes the newline when the file is read
1411 and it is returned untranslated. On output, '\n' is converted to the
1412 newline.
1413
1414 If line_buffering is True, a call to flush is implied when a call to
1415 write contains a newline character.
1416 """
1417
1418 _CHUNK_SIZE = 2048
1419
1420 def __init__(self, buffer, encoding=None, errors=None, newline=None,
1421 line_buffering=False):
1422 if newline is not None and not isinstance(newline, str):
1423 raise TypeError("illegal newline type: %r" % (type(newline),))
1424 if newline not in (None, "", "\n", "\r", "\r\n"):
1425 raise ValueError("illegal newline value: %r" % (newline,))
1426 if encoding is None:
1427 try:
1428 encoding = os.device_encoding(buffer.fileno())
1429 except (AttributeError, UnsupportedOperation):
1430 pass
1431 if encoding is None:
1432 try:
1433 import locale
1434 except ImportError:
1435 # Importing locale may fail if Python is being built
1436 encoding = "ascii"
1437 else:
1438 encoding = locale.getpreferredencoding()
1439
1440 if not isinstance(encoding, str):
1441 raise ValueError("invalid encoding: %r" % encoding)
1442
1443 if errors is None:
1444 errors = "strict"
1445 else:
1446 if not isinstance(errors, str):
1447 raise ValueError("invalid errors: %r" % errors)
1448
1449 self.buffer = buffer
1450 self._line_buffering = line_buffering
1451 self._encoding = encoding
1452 self._errors = errors
1453 self._readuniversal = not newline
1454 self._readtranslate = newline is None
1455 self._readnl = newline
1456 self._writetranslate = newline != ''
1457 self._writenl = newline or os.linesep
1458 self._encoder = None
1459 self._decoder = None
1460 self._decoded_chars = '' # buffer for text returned from decoder
1461 self._decoded_chars_used = 0 # offset into _decoded_chars for read()
1462 self._snapshot = None # info for reconstructing decoder state
1463 self._seekable = self._telling = self.buffer.seekable()
1464
Antoine Pitroue4501852009-05-14 18:55:55 +00001465 if self._seekable and self.writable():
1466 position = self.buffer.tell()
1467 if position != 0:
1468 try:
1469 self._get_encoder().setstate(0)
1470 except LookupError:
1471 # Sometimes the encoder doesn't exist
1472 pass
1473
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001474 # self._snapshot is either None, or a tuple (dec_flags, next_input)
1475 # where dec_flags is the second (integer) item of the decoder state
1476 # and next_input is the chunk of input bytes that comes next after the
1477 # snapshot point. We use this to reconstruct decoder states in tell().
1478
1479 # Naming convention:
1480 # - "bytes_..." for integer variables that count input bytes
1481 # - "chars_..." for integer variables that count decoded characters
1482
Benjamin Petersonc4c0eae2009-03-09 00:07:03 +00001483 def __repr__(self):
Antoine Pitrou716c4442009-05-23 19:04:03 +00001484 try:
1485 name = self.name
1486 except AttributeError:
1487 return "<_pyio.TextIOWrapper encoding={0!r}>".format(self.encoding)
1488 else:
1489 return "<_pyio.TextIOWrapper name={0!r} encoding={1!r}>".format(
1490 name, self.encoding)
Benjamin Petersonc4c0eae2009-03-09 00:07:03 +00001491
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001492 @property
1493 def encoding(self):
1494 return self._encoding
1495
1496 @property
1497 def errors(self):
1498 return self._errors
1499
1500 @property
1501 def line_buffering(self):
1502 return self._line_buffering
1503
1504 def seekable(self):
1505 return self._seekable
1506
1507 def readable(self):
1508 return self.buffer.readable()
1509
1510 def writable(self):
1511 return self.buffer.writable()
1512
1513 def flush(self):
1514 self.buffer.flush()
1515 self._telling = self._seekable
1516
1517 def close(self):
Benjamin Petersond2e0c792009-05-01 20:40:59 +00001518 if self.buffer is not None:
1519 try:
1520 self.flush()
1521 except IOError:
1522 pass # If flush() fails, just give up
1523 self.buffer.close()
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001524
1525 @property
1526 def closed(self):
1527 return self.buffer.closed
1528
1529 @property
1530 def name(self):
1531 return self.buffer.name
1532
1533 def fileno(self):
1534 return self.buffer.fileno()
1535
1536 def isatty(self):
1537 return self.buffer.isatty()
1538
1539 def write(self, s: str):
1540 if self.closed:
1541 raise ValueError("write to closed file")
1542 if not isinstance(s, str):
1543 raise TypeError("can't write %s to text stream" %
1544 s.__class__.__name__)
1545 length = len(s)
1546 haslf = (self._writetranslate or self._line_buffering) and "\n" in s
1547 if haslf and self._writetranslate and self._writenl != "\n":
1548 s = s.replace("\n", self._writenl)
1549 encoder = self._encoder or self._get_encoder()
1550 # XXX What if we were just reading?
1551 b = encoder.encode(s)
1552 self.buffer.write(b)
1553 if self._line_buffering and (haslf or "\r" in s):
1554 self.flush()
1555 self._snapshot = None
1556 if self._decoder:
1557 self._decoder.reset()
1558 return length
1559
1560 def _get_encoder(self):
1561 make_encoder = codecs.getincrementalencoder(self._encoding)
1562 self._encoder = make_encoder(self._errors)
1563 return self._encoder
1564
1565 def _get_decoder(self):
1566 make_decoder = codecs.getincrementaldecoder(self._encoding)
1567 decoder = make_decoder(self._errors)
1568 if self._readuniversal:
1569 decoder = IncrementalNewlineDecoder(decoder, self._readtranslate)
1570 self._decoder = decoder
1571 return decoder
1572
1573 # The following three methods implement an ADT for _decoded_chars.
1574 # Text returned from the decoder is buffered here until the client
1575 # requests it by calling our read() or readline() method.
1576 def _set_decoded_chars(self, chars):
1577 """Set the _decoded_chars buffer."""
1578 self._decoded_chars = chars
1579 self._decoded_chars_used = 0
1580
1581 def _get_decoded_chars(self, n=None):
1582 """Advance into the _decoded_chars buffer."""
1583 offset = self._decoded_chars_used
1584 if n is None:
1585 chars = self._decoded_chars[offset:]
1586 else:
1587 chars = self._decoded_chars[offset:offset + n]
1588 self._decoded_chars_used += len(chars)
1589 return chars
1590
1591 def _rewind_decoded_chars(self, n):
1592 """Rewind the _decoded_chars buffer."""
1593 if self._decoded_chars_used < n:
1594 raise AssertionError("rewind decoded_chars out of bounds")
1595 self._decoded_chars_used -= n
1596
1597 def _read_chunk(self):
1598 """
1599 Read and decode the next chunk of data from the BufferedReader.
1600 """
1601
1602 # The return value is True unless EOF was reached. The decoded
1603 # string is placed in self._decoded_chars (replacing its previous
1604 # value). The entire input chunk is sent to the decoder, though
1605 # some of it may remain buffered in the decoder, yet to be
1606 # converted.
1607
1608 if self._decoder is None:
1609 raise ValueError("no decoder")
1610
1611 if self._telling:
1612 # To prepare for tell(), we need to snapshot a point in the
1613 # file where the decoder's input buffer is empty.
1614
1615 dec_buffer, dec_flags = self._decoder.getstate()
1616 # Given this, we know there was a valid snapshot point
1617 # len(dec_buffer) bytes ago with decoder state (b'', dec_flags).
1618
1619 # Read a chunk, decode it, and put the result in self._decoded_chars.
1620 input_chunk = self.buffer.read1(self._CHUNK_SIZE)
1621 eof = not input_chunk
1622 self._set_decoded_chars(self._decoder.decode(input_chunk, eof))
1623
1624 if self._telling:
1625 # At the snapshot point, len(dec_buffer) bytes before the read,
1626 # the next input to be decoded is dec_buffer + input_chunk.
1627 self._snapshot = (dec_flags, dec_buffer + input_chunk)
1628
1629 return not eof
1630
1631 def _pack_cookie(self, position, dec_flags=0,
1632 bytes_to_feed=0, need_eof=0, chars_to_skip=0):
1633 # The meaning of a tell() cookie is: seek to position, set the
1634 # decoder flags to dec_flags, read bytes_to_feed bytes, feed them
1635 # into the decoder with need_eof as the EOF flag, then skip
1636 # chars_to_skip characters of the decoded result. For most simple
1637 # decoders, tell() will often just give a byte offset in the file.
1638 return (position | (dec_flags<<64) | (bytes_to_feed<<128) |
1639 (chars_to_skip<<192) | bool(need_eof)<<256)
1640
1641 def _unpack_cookie(self, bigint):
1642 rest, position = divmod(bigint, 1<<64)
1643 rest, dec_flags = divmod(rest, 1<<64)
1644 rest, bytes_to_feed = divmod(rest, 1<<64)
1645 need_eof, chars_to_skip = divmod(rest, 1<<64)
1646 return position, dec_flags, bytes_to_feed, need_eof, chars_to_skip
1647
1648 def tell(self):
1649 if not self._seekable:
1650 raise IOError("underlying stream is not seekable")
1651 if not self._telling:
1652 raise IOError("telling position disabled by next() call")
1653 self.flush()
1654 position = self.buffer.tell()
1655 decoder = self._decoder
1656 if decoder is None or self._snapshot is None:
1657 if self._decoded_chars:
1658 # This should never happen.
1659 raise AssertionError("pending decoded text")
1660 return position
1661
1662 # Skip backward to the snapshot point (see _read_chunk).
1663 dec_flags, next_input = self._snapshot
1664 position -= len(next_input)
1665
1666 # How many decoded characters have been used up since the snapshot?
1667 chars_to_skip = self._decoded_chars_used
1668 if chars_to_skip == 0:
1669 # We haven't moved from the snapshot point.
1670 return self._pack_cookie(position, dec_flags)
1671
1672 # Starting from the snapshot position, we will walk the decoder
1673 # forward until it gives us enough decoded characters.
1674 saved_state = decoder.getstate()
1675 try:
1676 # Note our initial start point.
1677 decoder.setstate((b'', dec_flags))
1678 start_pos = position
1679 start_flags, bytes_fed, chars_decoded = dec_flags, 0, 0
1680 need_eof = 0
1681
1682 # Feed the decoder one byte at a time. As we go, note the
1683 # nearest "safe start point" before the current location
1684 # (a point where the decoder has nothing buffered, so seek()
1685 # can safely start from there and advance to this location).
1686 next_byte = bytearray(1)
1687 for next_byte[0] in next_input:
1688 bytes_fed += 1
1689 chars_decoded += len(decoder.decode(next_byte))
1690 dec_buffer, dec_flags = decoder.getstate()
1691 if not dec_buffer and chars_decoded <= chars_to_skip:
1692 # Decoder buffer is empty, so this is a safe start point.
1693 start_pos += bytes_fed
1694 chars_to_skip -= chars_decoded
1695 start_flags, bytes_fed, chars_decoded = dec_flags, 0, 0
1696 if chars_decoded >= chars_to_skip:
1697 break
1698 else:
1699 # We didn't get enough decoded data; signal EOF to get more.
1700 chars_decoded += len(decoder.decode(b'', final=True))
1701 need_eof = 1
1702 if chars_decoded < chars_to_skip:
1703 raise IOError("can't reconstruct logical file position")
1704
1705 # The returned cookie corresponds to the last safe start point.
1706 return self._pack_cookie(
1707 start_pos, start_flags, bytes_fed, need_eof, chars_to_skip)
1708 finally:
1709 decoder.setstate(saved_state)
1710
1711 def truncate(self, pos=None):
1712 self.flush()
1713 if pos is None:
1714 pos = self.tell()
1715 self.seek(pos)
1716 return self.buffer.truncate()
1717
Benjamin Petersond2e0c792009-05-01 20:40:59 +00001718 def detach(self):
1719 if self.buffer is None:
1720 raise ValueError("buffer is already detached")
1721 self.flush()
1722 buffer = self.buffer
1723 self.buffer = None
1724 return buffer
1725
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001726 def seek(self, cookie, whence=0):
1727 if self.closed:
1728 raise ValueError("tell on closed file")
1729 if not self._seekable:
1730 raise IOError("underlying stream is not seekable")
1731 if whence == 1: # seek relative to current position
1732 if cookie != 0:
1733 raise IOError("can't do nonzero cur-relative seeks")
1734 # Seeking to the current position should attempt to
1735 # sync the underlying buffer with the current position.
1736 whence = 0
1737 cookie = self.tell()
1738 if whence == 2: # seek relative to end of file
1739 if cookie != 0:
1740 raise IOError("can't do nonzero end-relative seeks")
1741 self.flush()
1742 position = self.buffer.seek(0, 2)
1743 self._set_decoded_chars('')
1744 self._snapshot = None
1745 if self._decoder:
1746 self._decoder.reset()
1747 return position
1748 if whence != 0:
1749 raise ValueError("invalid whence (%r, should be 0, 1 or 2)" %
1750 (whence,))
1751 if cookie < 0:
1752 raise ValueError("negative seek position %r" % (cookie,))
1753 self.flush()
1754
1755 # The strategy of seek() is to go back to the safe start point
1756 # and replay the effect of read(chars_to_skip) from there.
1757 start_pos, dec_flags, bytes_to_feed, need_eof, chars_to_skip = \
1758 self._unpack_cookie(cookie)
1759
1760 # Seek back to the safe start point.
1761 self.buffer.seek(start_pos)
1762 self._set_decoded_chars('')
1763 self._snapshot = None
1764
1765 # Restore the decoder to its state from the safe start point.
Benjamin Peterson9363a652009-03-05 00:42:09 +00001766 if cookie == 0 and self._decoder:
1767 self._decoder.reset()
1768 elif self._decoder or dec_flags or chars_to_skip:
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001769 self._decoder = self._decoder or self._get_decoder()
1770 self._decoder.setstate((b'', dec_flags))
1771 self._snapshot = (dec_flags, b'')
1772
1773 if chars_to_skip:
1774 # Just like _read_chunk, feed the decoder and save a snapshot.
1775 input_chunk = self.buffer.read(bytes_to_feed)
1776 self._set_decoded_chars(
1777 self._decoder.decode(input_chunk, need_eof))
1778 self._snapshot = (dec_flags, input_chunk)
1779
1780 # Skip chars_to_skip of the decoded characters.
1781 if len(self._decoded_chars) < chars_to_skip:
1782 raise IOError("can't restore logical file position")
1783 self._decoded_chars_used = chars_to_skip
1784
Antoine Pitroue4501852009-05-14 18:55:55 +00001785 # Finally, reset the encoder (merely useful for proper BOM handling)
1786 try:
1787 encoder = self._encoder or self._get_encoder()
1788 except LookupError:
1789 # Sometimes the encoder doesn't exist
1790 pass
1791 else:
1792 if cookie != 0:
1793 encoder.setstate(0)
1794 else:
1795 encoder.reset()
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001796 return cookie
1797
1798 def read(self, n=None):
Benjamin Petersona1b49012009-03-31 23:11:32 +00001799 self._checkReadable()
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001800 if n is None:
1801 n = -1
1802 decoder = self._decoder or self._get_decoder()
1803 if n < 0:
1804 # Read everything.
1805 result = (self._get_decoded_chars() +
1806 decoder.decode(self.buffer.read(), final=True))
1807 self._set_decoded_chars('')
1808 self._snapshot = None
1809 return result
1810 else:
1811 # Keep reading chunks until we have n characters to return.
1812 eof = False
1813 result = self._get_decoded_chars(n)
1814 while len(result) < n and not eof:
1815 eof = not self._read_chunk()
1816 result += self._get_decoded_chars(n - len(result))
1817 return result
1818
1819 def __next__(self):
1820 self._telling = False
1821 line = self.readline()
1822 if not line:
1823 self._snapshot = None
1824 self._telling = self._seekable
1825 raise StopIteration
1826 return line
1827
1828 def readline(self, limit=None):
1829 if self.closed:
1830 raise ValueError("read from closed file")
1831 if limit is None:
1832 limit = -1
Benjamin Petersonb01138a2009-04-24 22:59:52 +00001833 elif not isinstance(limit, int):
1834 raise TypeError("limit must be an integer")
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001835
1836 # Grab all the decoded text (we will rewind any extra bits later).
1837 line = self._get_decoded_chars()
1838
1839 start = 0
1840 # Make the decoder if it doesn't already exist.
1841 if not self._decoder:
1842 self._get_decoder()
1843
1844 pos = endpos = None
1845 while True:
1846 if self._readtranslate:
1847 # Newlines are already translated, only search for \n
1848 pos = line.find('\n', start)
1849 if pos >= 0:
1850 endpos = pos + 1
1851 break
1852 else:
1853 start = len(line)
1854
1855 elif self._readuniversal:
1856 # Universal newline search. Find any of \r, \r\n, \n
1857 # The decoder ensures that \r\n are not split in two pieces
1858
1859 # In C we'd look for these in parallel of course.
1860 nlpos = line.find("\n", start)
1861 crpos = line.find("\r", start)
1862 if crpos == -1:
1863 if nlpos == -1:
1864 # Nothing found
1865 start = len(line)
1866 else:
1867 # Found \n
1868 endpos = nlpos + 1
1869 break
1870 elif nlpos == -1:
1871 # Found lone \r
1872 endpos = crpos + 1
1873 break
1874 elif nlpos < crpos:
1875 # Found \n
1876 endpos = nlpos + 1
1877 break
1878 elif nlpos == crpos + 1:
1879 # Found \r\n
1880 endpos = crpos + 2
1881 break
1882 else:
1883 # Found \r
1884 endpos = crpos + 1
1885 break
1886 else:
1887 # non-universal
1888 pos = line.find(self._readnl)
1889 if pos >= 0:
1890 endpos = pos + len(self._readnl)
1891 break
1892
1893 if limit >= 0 and len(line) >= limit:
1894 endpos = limit # reached length limit
1895 break
1896
1897 # No line ending seen yet - get more data'
1898 while self._read_chunk():
1899 if self._decoded_chars:
1900 break
1901 if self._decoded_chars:
1902 line += self._get_decoded_chars()
1903 else:
1904 # end of file
1905 self._set_decoded_chars('')
1906 self._snapshot = None
1907 return line
1908
1909 if limit >= 0 and endpos > limit:
1910 endpos = limit # don't exceed limit
1911
1912 # Rewind _decoded_chars to just after the line ending we found.
1913 self._rewind_decoded_chars(len(line) - endpos)
1914 return line[:endpos]
1915
1916 @property
1917 def newlines(self):
1918 return self._decoder.newlines if self._decoder else None
1919
1920
1921class StringIO(TextIOWrapper):
1922 """Text I/O implementation using an in-memory buffer.
1923
1924 The initial_value argument sets the value of object. The newline
1925 argument is like the one of TextIOWrapper's constructor.
1926 """
1927
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001928 def __init__(self, initial_value="", newline="\n"):
1929 super(StringIO, self).__init__(BytesIO(),
1930 encoding="utf-8",
1931 errors="strict",
1932 newline=newline)
Antoine Pitrou11446482009-04-04 14:09:30 +00001933 # Issue #5645: make universal newlines semantics the same as in the
1934 # C version, even under Windows.
1935 if newline is None:
1936 self._writetranslate = False
Georg Brandl194da4a2009-08-13 09:34:05 +00001937 if initial_value is not None:
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001938 if not isinstance(initial_value, str):
Georg Brandl194da4a2009-08-13 09:34:05 +00001939 raise TypeError("initial_value must be str or None, not {0}"
1940 .format(type(initial_value).__name__))
Benjamin Peterson4fa88fa2009-03-04 00:14:51 +00001941 initial_value = str(initial_value)
1942 self.write(initial_value)
1943 self.seek(0)
1944
1945 def getvalue(self):
1946 self.flush()
1947 return self.buffer.getvalue().decode(self._encoding, self._errors)
Benjamin Peterson9fd459a2009-03-09 00:09:44 +00001948
1949 def __repr__(self):
1950 # TextIOWrapper tells the encoding in its repr. In StringIO,
1951 # that's a implementation detail.
1952 return object.__repr__(self)
Benjamin Petersonb487e632009-03-21 03:08:31 +00001953
1954 @property
Benjamin Peterson0926ad12009-06-06 18:02:12 +00001955 def errors(self):
1956 return None
1957
1958 @property
Benjamin Petersonb487e632009-03-21 03:08:31 +00001959 def encoding(self):
1960 return None
Benjamin Petersond2e0c792009-05-01 20:40:59 +00001961
1962 def detach(self):
1963 # This doesn't make sense on StringIO.
1964 self._unsupported("detach")