Guido van Rossum | 1b5a1d0 | 1996-05-28 23:17:13 +0000 | [diff] [blame^] | 1 | """File-like objects that read/write an array buffer. |
| 2 | |
| 3 | This implements (nearly) all stdio methods. |
| 4 | |
| 5 | f = ArrayIO() # ready for writing |
| 6 | f = ArrayIO(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 | - This is very similar to StringIO. StringIO is faster for reading, |
| 22 | but ArrayIO is faster for writing. |
| 23 | - ArrayIO uses an array object internally, but all its interfaces |
| 24 | accept and return strings. |
| 25 | - Using a real file is often faster (but less convenient). |
| 26 | - fileno() is left unimplemented so that code which uses it triggers |
| 27 | an exception early. |
| 28 | - Seeking far beyond EOF and then writing will insert real null |
| 29 | bytes that occupy space in the buffer. |
| 30 | - There's a simple test set (see end of this file). |
| 31 | """ |
Guido van Rossum | 405ab23 | 1996-02-28 23:58:09 +0000 | [diff] [blame] | 32 | |
| 33 | import string |
Guido van Rossum | 1b5a1d0 | 1996-05-28 23:17:13 +0000 | [diff] [blame^] | 34 | from array import array |
Guido van Rossum | 405ab23 | 1996-02-28 23:58:09 +0000 | [diff] [blame] | 35 | |
Guido van Rossum | 1b5a1d0 | 1996-05-28 23:17:13 +0000 | [diff] [blame^] | 36 | class ArrayIO: |
Guido van Rossum | 405ab23 | 1996-02-28 23:58:09 +0000 | [diff] [blame] | 37 | def __init__(self, buf = ''): |
Guido van Rossum | 1b5a1d0 | 1996-05-28 23:17:13 +0000 | [diff] [blame^] | 38 | self.buf = array('c', buf) |
Guido van Rossum | 405ab23 | 1996-02-28 23:58:09 +0000 | [diff] [blame] | 39 | self.pos = 0 |
| 40 | self.closed = 0 |
Guido van Rossum | 1b5a1d0 | 1996-05-28 23:17:13 +0000 | [diff] [blame^] | 41 | self.softspace = 0 |
| 42 | def close(self): |
| 43 | if not self.closed: |
| 44 | self.closed = 1 |
| 45 | del self.buf, self.pos |
| 46 | def isatty(self): |
| 47 | return 0 |
| 48 | def seek(self, pos, mode = 0): |
| 49 | if mode == 1: |
| 50 | pos = pos + self.pos |
| 51 | elif mode == 2: |
| 52 | pos = pos + len(self.buf) |
| 53 | self.pos = max(0, pos) |
| 54 | def tell(self): |
| 55 | return self.pos |
| 56 | def read(self, n = -1): |
| 57 | if n < 0: |
Guido van Rossum | 405ab23 | 1996-02-28 23:58:09 +0000 | [diff] [blame] | 58 | newpos = len(self.buf) |
| 59 | else: |
Guido van Rossum | 1b5a1d0 | 1996-05-28 23:17:13 +0000 | [diff] [blame^] | 60 | newpos = min(self.pos+n, len(self.buf)) |
Guido van Rossum | 405ab23 | 1996-02-28 23:58:09 +0000 | [diff] [blame] | 61 | r = self.buf[self.pos:newpos].tostring() |
| 62 | self.pos = newpos |
| 63 | return r |
Guido van Rossum | 1b5a1d0 | 1996-05-28 23:17:13 +0000 | [diff] [blame^] | 64 | def readline(self): |
| 65 | i = string.find(self.buf[self.pos:].tostring(), '\n') |
| 66 | if i < 0: |
| 67 | newpos = len(self.buf) |
| 68 | else: |
| 69 | newpos = self.pos+i+1 |
| 70 | r = self.buf[self.pos:newpos].tostring() |
Guido van Rossum | 405ab23 | 1996-02-28 23:58:09 +0000 | [diff] [blame] | 71 | self.pos = newpos |
Guido van Rossum | 1b5a1d0 | 1996-05-28 23:17:13 +0000 | [diff] [blame^] | 72 | return r |
| 73 | def readlines(self): |
| 74 | lines = string.splitfields(self.read(), '\n') |
| 75 | if not lines: |
| 76 | return lines |
| 77 | for i in range(len(lines)-1): |
| 78 | lines[i] = lines[i] + '\n' |
| 79 | if not lines[-1]: |
| 80 | del lines[-1] |
| 81 | return lines |
| 82 | def write(self, s): |
| 83 | if not s: return |
| 84 | a = array('c', s) |
| 85 | n = self.pos - len(self.buf) |
| 86 | if n > 0: |
| 87 | self.buf[len(self.buf):] = array('c', '\0')*n |
| 88 | newpos = self.pos + len(a) |
| 89 | self.buf[self.pos:newpos] = a |
| 90 | self.pos = newpos |
| 91 | def writelines(self, list): |
| 92 | self.write(string.joinfields(list, '')) |
| 93 | def flush(self): |
| 94 | pass |
Guido van Rossum | 405ab23 | 1996-02-28 23:58:09 +0000 | [diff] [blame] | 95 | def getvalue(self): |
Guido van Rossum | 405ab23 | 1996-02-28 23:58:09 +0000 | [diff] [blame] | 96 | return self.buf.tostring() |
| 97 | |
| 98 | |
| 99 | # A little test suite |
Guido van Rossum | 405ab23 | 1996-02-28 23:58:09 +0000 | [diff] [blame] | 100 | |
| 101 | def test(): |
| 102 | import sys |
| 103 | if sys.argv[1:]: |
| 104 | file = sys.argv[1] |
| 105 | else: |
| 106 | file = '/etc/passwd' |
| 107 | lines = open(file, 'r').readlines() |
| 108 | text = open(file, 'r').read() |
| 109 | f = ArrayIO() |
| 110 | for line in lines[:-2]: |
| 111 | f.write(line) |
| 112 | f.writelines(lines[-2:]) |
| 113 | if f.getvalue() != text: |
| 114 | raise RuntimeError, 'write failed' |
| 115 | length = f.tell() |
| 116 | print 'File length =', length |
| 117 | f.seek(len(lines[0])) |
| 118 | f.write(lines[1]) |
| 119 | f.seek(0) |
| 120 | print 'First line =', `f.readline()` |
| 121 | here = f.tell() |
| 122 | line = f.readline() |
| 123 | print 'Second line =', `line` |
| 124 | f.seek(-len(line), 1) |
| 125 | line2 = f.read(len(line)) |
| 126 | if line != line2: |
| 127 | raise RuntimeError, 'bad result after seek back' |
| 128 | f.seek(len(line2), 1) |
| 129 | list = f.readlines() |
| 130 | line = list[-1] |
| 131 | f.seek(f.tell() - len(line)) |
| 132 | line2 = f.read() |
| 133 | if line != line2: |
| 134 | raise RuntimeError, 'bad result after seek back from EOF' |
| 135 | print 'Read', len(list), 'more lines' |
| 136 | print 'File length =', f.tell() |
| 137 | if f.tell() != length: |
| 138 | raise RuntimeError, 'bad length' |
| 139 | f.close() |
| 140 | |
| 141 | if __name__ == '__main__': |
| 142 | test() |