# Copyright (C) 2003-2007 Robey Pointer # # This file is part of paramiko. # # Paramiko is free software; you can redistribute it and/or modify it under the # terms of the GNU Lesser General Public License as published by the Free # Software Foundation; either version 2.1 of the License, or (at your option) # any later version. # # Paramiko is distrubuted in the hope that it will be useful, but WITHOUT ANY # WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR # A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more # details. # # You should have received a copy of the GNU Lesser General Public License # along with Paramiko; if not, write to the Free Software Foundation, Inc., # 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA. """ BufferedFile. """ from cStringIO import StringIO class BufferedFile (object): """ Reusable base class to implement python-style file buffering around a simpler stream. """ _DEFAULT_BUFSIZE = 8192 SEEK_SET = 0 SEEK_CUR = 1 SEEK_END = 2 FLAG_READ = 0x1 FLAG_WRITE = 0x2 FLAG_APPEND = 0x4 FLAG_BINARY = 0x10 FLAG_BUFFERED = 0x20 FLAG_LINE_BUFFERED = 0x40 FLAG_UNIVERSAL_NEWLINE = 0x80 def __init__(self): self.newlines = None self._flags = 0 self._bufsize = self._DEFAULT_BUFSIZE self._wbuffer = StringIO() self._rbuffer = '' self._at_trailing_cr = False self._closed = False # pos - position within the file, according to the user # realpos - position according the OS # (these may be different because we buffer for line reading) self._pos = self._realpos = 0 # size only matters for seekable files self._size = 0 def __del__(self): self.close() def __iter__(self): """ Returns an iterator that can be used to iterate over the lines in this file. This iterator happens to return the file itself, since a file is its own iterator. @raise ValueError: if the file is closed. @return: an interator. @rtype: iterator """ if self._closed: raise ValueError('I/O operation on closed file') return self def close(self): """ Close the file. Future read and write operations will fail. """ self.flush() self._closed = True def flush(self): """ Write out any data in the write buffer. This may do nothing if write buffering is not turned on. """ self._write_all(self._wbuffer.getvalue()) self._wbuffer = StringIO() return def next(self): """ Returns the next line from the input, or raises L{StopIteration} when EOF is hit. Unlike python file objects, it's okay to mix calls to C{next} and L{readline}. @raise StopIteration: when the end of the file is reached. @return: a line read from the file. @rtype: str """ line = self.readline() if not line: raise StopIteration return line def read(self, size=None): """ Read at most C{size} bytes from the file (less if we hit the end of the file first). If the C{size} argument is negative or omitted, read all the remaining data in the file. @param size: maximum number of bytes to read @type size: int @return: data read from the file, or an empty string if EOF was encountered immediately @rtype: str """ if self._closed: raise IOError('File is closed') if not (self._flags & self.FLAG_READ): raise IOError('File is not open for reading') if (size is None) or (size < 0): # go for broke result = self._rbuffer self._rbuffer = '' self._pos += len(result) while True: try: new_data = self._read(self._DEFAULT_BUFSIZE) except EOFError: new_data = None if (new_data is None) or (len(new_data) == 0): break result += new_data self._realpos += len(new_data) self._pos += len(new_data) return result if size <= len(self._rbuffer): result = self._rbuffer[:size] self._rbuffer = self._rbuffer[size:] self._pos += len(result) return result while len(self._rbuffer) < size: read_size = size - len(self._rbuffer) if self._flags & self.FLAG_BUFFERED: read_size = max(self._bufsize, read_size) try: new_data = self._read(read_size) except EOFError: new_data = None if (new_data is None) or (len(new_data) == 0): break self._rbuffer += new_data self._realpos += len(new_data) result = self._rbuffer[:size] self._rbuffer = self._rbuffer[size:] self._pos += len(result) return result def readline(self, size=None): """ Read one entire line from the file. A trailing newline character is kept in the string (but may be absent when a file ends with an incomplete line). If the size argument is present and non-negative, it is a maximum byte count (including the trailing newline) and an incomplete line may be returned. An empty string is returned only when EOF is encountered immediately. @note: Unlike stdio's C{fgets()}, the returned string contains null characters (C{'\\0'}) if they occurred in the input. @param size: maximum length of returned string. @type size: int @return: next line of the file, or an empty string if the end of the file has been reached. @rtype: str """ # it's almost silly how complex this function is. if self._closed: raise IOError('File is closed') if not (self._flags & self.FLAG_READ): raise IOError('File not open for reading') line = self._rbuffer while True: if self._at_trailing_cr and (self._flags & self.FLAG_UNIVERSAL_NEWLINE) and (len(line) > 0): # edge case: the newline may be '\r\n' and we may have read # only the first '\r' last time. if line[0] == '\n': line = line[1:] self._record_newline('\r\n') else: self._record_newline('\r') self._at_trailing_cr = False # check size before looking for a linefeed, in case we already have # enough. if (size is not None) and (size >= 0): if len(line) >= size: # truncate line and return self._rbuffer = line[size:] line = line[:size] self._pos += len(line) return line n = size - len(line) else: n = self._bufsize if ('\n' in line) or ((self._flags & self.FLAG_UNIVERSAL_NEWLINE) and ('\r' in line)): break try: new_data = self._read(n) except EOFError: new_data = None if (new_data is None) or (len(new_data) == 0): self._rbuffer = '' self._pos += len(line) return line line += new_data self._realpos += len(new_data) # find the newline pos = line.find('\n') if self._flags & self.FLAG_UNIVERSAL_NEWLINE: rpos = line.find('\r') if (rpos >= 0) and ((rpos < pos) or (pos < 0)): pos = rpos xpos = pos + 1 if (line[pos] == '\r') and (xpos < len(line)) and (line[xpos] == '\n'): xpos += 1 self._rbuffer = line[xpos:] lf = line[pos:xpos] line = line[:pos] + '\n' if (len(self._rbuffer) == 0) and (lf == '\r'): # we could read the line up to a '\r' and there could still be a # '\n' following that we read next time. note that and eat it. self._at_trailing_cr = True else: self._record_newline(lf) self._pos += len(line) return line def readlines(self, sizehint=None): """ Read all remaining lines using L{readline} and return them as a list. If the optional C{sizehint} argument is present, instead of reading up to EOF, whole lines totalling approximately sizehint bytes (possibly after rounding up to an internal buffer size) are read. @param sizehint: desired maximum number of bytes to read. @type sizehint: int @return: list of lines read from the file. @rtype: list """ lines = [] bytes = 0 while True: line = self.readline() if len(line) == 0: break lines.append(line) bytes += len(line) if (sizehint is not None) and (bytes >= sizehint): break return lines def seek(self, offset, whence=0): """ Set the file's current position, like stdio's C{fseek}. Not all file objects support seeking. @note: If a file is opened in append mode (C{'a'} or C{'a+'}), any seek operations will be undone at the next write (as the file position will move back to the end of the file). @param offset: position to move to within the file, relative to C{whence}. @type offset: int @param whence: type of movement: 0 = absolute; 1 = relative to the current position; 2 = relative to the end of the file. @type whence: int @raise IOError: if the file doesn't support random access. """ raise IOError('File does not support seeking.') def tell(self): """ Return the file's current position. This may not be accurate or useful if the underlying file doesn't support random access, or was opened in append mode. @return: file position (in bytes). @rtype: int """ return self._pos def write(self, data): """ Write data to the file. If write buffering is on (C{bufsize} was specified and non-zero), some or all of the data may not actually be written yet. (Use L{flush} or L{close} to force buffered data to be written out.) @param data: data to write. @type data: str """ if self._closed: raise IOError('File is closed') if not (self._flags & self.FLAG_WRITE): raise IOError('File not open for writing') if not (self._flags & self.FLAG_BUFFERED): self._write_all(data) return self._wbuffer.write(data) if self._flags & self.FLAG_LINE_BUFFERED: # only scan the new data for linefeed, to avoid wasting time. last_newline_pos = data.rfind('\n') if last_newline_pos >= 0: wbuf = self._wbuffer.getvalue() last_newline_pos += len(wbuf) - len(data) self._write_all(wbuf[:last_newline_pos + 1]) self._wbuffer = StringIO() self._wbuffer.write(wbuf[last_newline_pos + 1:]) return # even if we're line buffering, if the buffer has grown past the # buffer size, force a flush. if self._wbuffer.tell() >= self._bufsize: self.flush() return def writelines(self, sequence): """ Write a sequence of strings to the file. The sequence can be any iterable object producing strings, typically a list of strings. (The name is intended to match L{readlines}; C{writelines} does not add line separators.) @param sequence: an iterable sequence of strings. @type sequence: sequence """ for line in sequence: self.write(line) return def xreadlines(self): """ Identical to C{iter(f)}. This is a deprecated file interface that predates python iterator support. @return: an iterator. @rtype: iterator """ return self ### overrides... def _read(self, size): """ I{(subclass override)} Read data from the stream. Return C{None} or raise C{EOFError} to indicate EOF. """ raise EOFError() def _write(self, data): """ I{(subclass override)} Write data into the stream. """ raise IOError('write not implemented') def _get_size(self): """ I{(subclass override)} Return the size of the file. This is called from within L{_set_mode} if the file is opened in append mode, so the file position can be tracked and L{seek} and L{tell} will work correctly. If the file is a stream that can't be randomly accessed, you don't need to override this method, """ return 0 ### internals... def _set_mode(self, mode='r', bufsize=-1): """ Subclasses call this method to initialize the BufferedFile. """ if bufsize == 1: # apparently, line buffering only affects writes. reads are only # buffered if you call readline (directly or indirectly: iterating # over a file will indirectly call readline). self._flags |= self.FLAG_BUFFERED | self.FLAG_LINE_BUFFERED elif bufsize > 1: self._bufsize = bufsize self._flags |= self.FLAG_BUFFERED if ('r' in mode) or ('+' in mode): self._flags |= self.FLAG_READ if ('w' in mode) or ('+' in mode): self._flags |= self.FLAG_WRITE if ('a' in mode): self._flags |= self.FLAG_WRITE | self.FLAG_APPEND self._size = self._get_size() self._pos = self._realpos = self._size if ('b' in mode): self._flags |= self.FLAG_BINARY if ('U' in mode): self._flags |= self.FLAG_UNIVERSAL_NEWLINE # built-in file objects have this attribute to store which kinds of # line terminations they've seen: # self.newlines = None def _write_all(self, data): # the underlying stream may be something that does partial writes (like # a socket). while len(data) > 0: count = self._write(data) data = data[count:] if self._flags & self.FLAG_APPEND: self._size += count self._pos = self._realpos = self._size else: self._pos += count self._realpos += count return None def _record_newline(self, newline): # silliness about tracking what kinds of newlines we've seen. # i don't understand why it can be None, a string, or a tuple, instead # of just always being a tuple, but we'll emulate that behavior anyway. if not (self._flags & self.FLAG_UNIVERSAL_NEWLINE): return if self.newlines is None: self.newlines = newline elif (type(self.newlines) is str) and (self.newlines != newline): self.newlines = (self.newlines, newline) elif newline not in self.newlines: self.newlines += (newline,)