Guido van Rossum | 4acc25b | 2000-02-02 15:10:15 +0000 | [diff] [blame] | 1 | """File-like objects that read from or write to a string buffer. |
| 2 | |
| 3 | This implements (nearly) all stdio methods. |
| 4 | |
| 5 | f = StringIO() # ready for writing |
| 6 | f = StringIO(buf) # ready for reading |
| 7 | f.close() # explicitly release resources held |
| 8 | flag = f.isatty() # always false |
| 9 | pos = f.tell() # get current position |
| 10 | f.seek(pos) # set current position |
| 11 | f.seek(pos, mode) # mode 0: absolute; 1: relative; 2: relative to EOF |
| 12 | buf = f.read() # read until EOF |
| 13 | buf = f.read(n) # read up to n bytes |
| 14 | buf = f.readline() # read until end of line ('\n') or EOF |
| 15 | list = f.readlines()# list of f.readline() results until EOF |
| 16 | f.write(buf) # write at current position |
| 17 | f.writelines(list) # for line in list: f.write(line) |
| 18 | f.getvalue() # return whole file's contents as a string |
| 19 | |
| 20 | Notes: |
| 21 | - Using a real file is often faster (but less convenient). |
Guido van Rossum | 98d9fd3 | 2000-02-28 15:12:25 +0000 | [diff] [blame] | 22 | - There's also a much faster implementation in C, called cStringIO, but |
| 23 | it's not subclassable. |
Guido van Rossum | 4acc25b | 2000-02-02 15:10:15 +0000 | [diff] [blame] | 24 | - fileno() is left unimplemented so that code which uses it triggers |
| 25 | an exception early. |
| 26 | - Seeking far beyond EOF and then writing will insert real null |
| 27 | bytes that occupy space in the buffer. |
| 28 | - There's a simple test set (see end of this file). |
| 29 | """ |
Guido van Rossum | 85d8945 | 1994-06-23 11:53:27 +0000 | [diff] [blame] | 30 | |
| 31 | import string |
| 32 | |
| 33 | class StringIO: |
| 34 | def __init__(self, buf = ''): |
| 35 | self.buf = buf |
Guido van Rossum | d390102 | 1996-06-19 14:54:19 +0000 | [diff] [blame] | 36 | self.len = len(buf) |
| 37 | self.buflist = [] |
Guido van Rossum | 85d8945 | 1994-06-23 11:53:27 +0000 | [diff] [blame] | 38 | self.pos = 0 |
| 39 | self.closed = 0 |
Guido van Rossum | 0daa13b | 1995-06-22 18:45:04 +0000 | [diff] [blame] | 40 | self.softspace = 0 |
Guido van Rossum | 85d8945 | 1994-06-23 11:53:27 +0000 | [diff] [blame] | 41 | def close(self): |
| 42 | if not self.closed: |
| 43 | self.closed = 1 |
| 44 | del self.buf, self.pos |
| 45 | def isatty(self): |
Fred Drake | ce4ba89 | 1998-08-18 17:43:08 +0000 | [diff] [blame] | 46 | if self.closed: |
| 47 | raise ValueError, "I/O operation on closed file" |
Guido van Rossum | 85d8945 | 1994-06-23 11:53:27 +0000 | [diff] [blame] | 48 | return 0 |
| 49 | def seek(self, pos, mode = 0): |
Fred Drake | ce4ba89 | 1998-08-18 17:43:08 +0000 | [diff] [blame] | 50 | if self.closed: |
| 51 | raise ValueError, "I/O operation on closed file" |
Guido van Rossum | d390102 | 1996-06-19 14:54:19 +0000 | [diff] [blame] | 52 | if self.buflist: |
| 53 | self.buf = self.buf + string.joinfields(self.buflist, '') |
| 54 | self.buflist = [] |
Guido van Rossum | 85d8945 | 1994-06-23 11:53:27 +0000 | [diff] [blame] | 55 | if mode == 1: |
| 56 | pos = pos + self.pos |
| 57 | elif mode == 2: |
Guido van Rossum | d390102 | 1996-06-19 14:54:19 +0000 | [diff] [blame] | 58 | pos = pos + self.len |
Guido van Rossum | 85d8945 | 1994-06-23 11:53:27 +0000 | [diff] [blame] | 59 | self.pos = max(0, pos) |
| 60 | def tell(self): |
Fred Drake | ce4ba89 | 1998-08-18 17:43:08 +0000 | [diff] [blame] | 61 | if self.closed: |
| 62 | raise ValueError, "I/O operation on closed file" |
Guido van Rossum | 85d8945 | 1994-06-23 11:53:27 +0000 | [diff] [blame] | 63 | return self.pos |
Guido van Rossum | f7476c5 | 1996-01-25 17:34:14 +0000 | [diff] [blame] | 64 | def read(self, n = -1): |
Fred Drake | ce4ba89 | 1998-08-18 17:43:08 +0000 | [diff] [blame] | 65 | if self.closed: |
| 66 | raise ValueError, "I/O operation on closed file" |
Guido van Rossum | d390102 | 1996-06-19 14:54:19 +0000 | [diff] [blame] | 67 | if self.buflist: |
| 68 | self.buf = self.buf + string.joinfields(self.buflist, '') |
| 69 | self.buflist = [] |
Guido van Rossum | f7476c5 | 1996-01-25 17:34:14 +0000 | [diff] [blame] | 70 | if n < 0: |
Guido van Rossum | d390102 | 1996-06-19 14:54:19 +0000 | [diff] [blame] | 71 | newpos = self.len |
Guido van Rossum | 85d8945 | 1994-06-23 11:53:27 +0000 | [diff] [blame] | 72 | else: |
Guido van Rossum | d390102 | 1996-06-19 14:54:19 +0000 | [diff] [blame] | 73 | newpos = min(self.pos+n, self.len) |
Guido van Rossum | 85d8945 | 1994-06-23 11:53:27 +0000 | [diff] [blame] | 74 | r = self.buf[self.pos:newpos] |
| 75 | self.pos = newpos |
| 76 | return r |
Guido van Rossum | 2e2525f | 1997-07-25 14:56:01 +0000 | [diff] [blame] | 77 | def readline(self, length=None): |
Fred Drake | ce4ba89 | 1998-08-18 17:43:08 +0000 | [diff] [blame] | 78 | if self.closed: |
| 79 | raise ValueError, "I/O operation on closed file" |
Guido van Rossum | d390102 | 1996-06-19 14:54:19 +0000 | [diff] [blame] | 80 | if self.buflist: |
| 81 | self.buf = self.buf + string.joinfields(self.buflist, '') |
| 82 | self.buflist = [] |
Guido van Rossum | 85d8945 | 1994-06-23 11:53:27 +0000 | [diff] [blame] | 83 | i = string.find(self.buf, '\n', self.pos) |
| 84 | if i < 0: |
Guido van Rossum | d390102 | 1996-06-19 14:54:19 +0000 | [diff] [blame] | 85 | newpos = self.len |
Guido van Rossum | 85d8945 | 1994-06-23 11:53:27 +0000 | [diff] [blame] | 86 | else: |
| 87 | newpos = i+1 |
Guido van Rossum | 2e2525f | 1997-07-25 14:56:01 +0000 | [diff] [blame] | 88 | if length is not None: |
| 89 | if self.pos + length < newpos: |
| 90 | newpos = self.pos + length |
Guido van Rossum | 85d8945 | 1994-06-23 11:53:27 +0000 | [diff] [blame] | 91 | r = self.buf[self.pos:newpos] |
| 92 | self.pos = newpos |
| 93 | return r |
Martin v. Löwis | a05fa1d | 2000-09-19 11:07:44 +0000 | [diff] [blame] | 94 | def readlines(self, sizehint = 0): |
| 95 | total = 0 |
Guido van Rossum | 85d8945 | 1994-06-23 11:53:27 +0000 | [diff] [blame] | 96 | lines = [] |
| 97 | line = self.readline() |
| 98 | while line: |
| 99 | lines.append(line) |
Martin v. Löwis | a05fa1d | 2000-09-19 11:07:44 +0000 | [diff] [blame] | 100 | total += len(line) |
| 101 | if 0 < sizehint <= total: |
| 102 | break |
Guido van Rossum | 85d8945 | 1994-06-23 11:53:27 +0000 | [diff] [blame] | 103 | line = self.readline() |
| 104 | return lines |
| 105 | def write(self, s): |
Fred Drake | ce4ba89 | 1998-08-18 17:43:08 +0000 | [diff] [blame] | 106 | if self.closed: |
| 107 | raise ValueError, "I/O operation on closed file" |
Guido van Rossum | 85d8945 | 1994-06-23 11:53:27 +0000 | [diff] [blame] | 108 | if not s: return |
Guido van Rossum | d390102 | 1996-06-19 14:54:19 +0000 | [diff] [blame] | 109 | if self.pos > self.len: |
| 110 | self.buflist.append('\0'*(self.pos - self.len)) |
| 111 | self.len = self.pos |
Guido van Rossum | 85d8945 | 1994-06-23 11:53:27 +0000 | [diff] [blame] | 112 | newpos = self.pos + len(s) |
Guido van Rossum | d390102 | 1996-06-19 14:54:19 +0000 | [diff] [blame] | 113 | if self.pos < self.len: |
| 114 | if self.buflist: |
| 115 | self.buf = self.buf + string.joinfields(self.buflist, '') |
| 116 | self.buflist = [] |
| 117 | self.buflist = [self.buf[:self.pos], s, self.buf[newpos:]] |
| 118 | self.buf = '' |
| 119 | else: |
| 120 | self.buflist.append(s) |
| 121 | self.len = newpos |
Guido van Rossum | 85d8945 | 1994-06-23 11:53:27 +0000 | [diff] [blame] | 122 | self.pos = newpos |
| 123 | def writelines(self, list): |
| 124 | self.write(string.joinfields(list, '')) |
Sjoerd Mullender | aa3e163 | 1995-07-19 11:21:21 +0000 | [diff] [blame] | 125 | def flush(self): |
Fred Drake | ce4ba89 | 1998-08-18 17:43:08 +0000 | [diff] [blame] | 126 | if self.closed: |
| 127 | raise ValueError, "I/O operation on closed file" |
Guido van Rossum | 85d8945 | 1994-06-23 11:53:27 +0000 | [diff] [blame] | 128 | def getvalue(self): |
Guido van Rossum | d390102 | 1996-06-19 14:54:19 +0000 | [diff] [blame] | 129 | if self.buflist: |
| 130 | self.buf = self.buf + string.joinfields(self.buflist, '') |
| 131 | self.buflist = [] |
Guido van Rossum | 85d8945 | 1994-06-23 11:53:27 +0000 | [diff] [blame] | 132 | return self.buf |
| 133 | |
| 134 | |
| 135 | # A little test suite |
| 136 | |
| 137 | def test(): |
| 138 | import sys |
| 139 | if sys.argv[1:]: |
| 140 | file = sys.argv[1] |
| 141 | else: |
| 142 | file = '/etc/passwd' |
| 143 | lines = open(file, 'r').readlines() |
| 144 | text = open(file, 'r').read() |
| 145 | f = StringIO() |
| 146 | for line in lines[:-2]: |
| 147 | f.write(line) |
| 148 | f.writelines(lines[-2:]) |
| 149 | if f.getvalue() != text: |
| 150 | raise RuntimeError, 'write failed' |
| 151 | length = f.tell() |
| 152 | print 'File length =', length |
| 153 | f.seek(len(lines[0])) |
| 154 | f.write(lines[1]) |
| 155 | f.seek(0) |
| 156 | print 'First line =', `f.readline()` |
| 157 | here = f.tell() |
| 158 | line = f.readline() |
| 159 | print 'Second line =', `line` |
| 160 | f.seek(-len(line), 1) |
| 161 | line2 = f.read(len(line)) |
| 162 | if line != line2: |
| 163 | raise RuntimeError, 'bad result after seek back' |
| 164 | f.seek(len(line2), 1) |
| 165 | list = f.readlines() |
| 166 | line = list[-1] |
| 167 | f.seek(f.tell() - len(line)) |
| 168 | line2 = f.read() |
| 169 | if line != line2: |
| 170 | raise RuntimeError, 'bad result after seek back from EOF' |
| 171 | print 'Read', len(list), 'more lines' |
| 172 | print 'File length =', f.tell() |
| 173 | if f.tell() != length: |
| 174 | raise RuntimeError, 'bad length' |
| 175 | f.close() |
| 176 | |
| 177 | if __name__ == '__main__': |
| 178 | test() |