Guido van Rossum | aad6761 | 2000-05-08 17:31:04 +0000 | [diff] [blame] | 1 | """A readline()-style interface to the parts of a multipart message. |
| 2 | |
| 3 | The MultiFile class makes each part of a multipart message "feel" like |
| 4 | an ordinary file, as long as you use fp.readline(). Allows recursive |
| 5 | use, for nested multipart messages. Probably best used together |
| 6 | with module mimetools. |
| 7 | |
| 8 | Suggested use: |
| 9 | |
| 10 | real_fp = open(...) |
| 11 | fp = MultiFile(real_fp) |
| 12 | |
| 13 | "read some lines from fp" |
| 14 | fp.push(separator) |
| 15 | while 1: |
| 16 | "read lines from fp until it returns an empty string" (A) |
| 17 | if not fp.next(): break |
| 18 | fp.pop() |
| 19 | "read remaining lines from fp until it returns an empty string" |
| 20 | |
| 21 | The latter sequence may be used recursively at (A). |
| 22 | It is also allowed to use multiple push()...pop() sequences. |
| 23 | |
Thomas Wouters | 7e47402 | 2000-07-16 12:04:32 +0000 | [diff] [blame] | 24 | If seekable is given as 0, the class code will not do the bookkeeping |
Guido van Rossum | aad6761 | 2000-05-08 17:31:04 +0000 | [diff] [blame] | 25 | it normally attempts in order to make seeks relative to the beginning of the |
| 26 | current file part. This may be useful when using MultiFile with a non- |
| 27 | seekable stream object. |
| 28 | """ |
Guido van Rossum | 5c97167 | 1996-07-22 15:23:25 +0000 | [diff] [blame] | 29 | |
| 30 | import sys |
| 31 | import string |
| 32 | |
Guido van Rossum | 8d691c8 | 2000-09-01 19:25:51 +0000 | [diff] [blame] | 33 | class Error(Exception): |
| 34 | pass |
Guido van Rossum | 5c97167 | 1996-07-22 15:23:25 +0000 | [diff] [blame] | 35 | |
| 36 | class MultiFile: |
Guido van Rossum | aad6761 | 2000-05-08 17:31:04 +0000 | [diff] [blame] | 37 | |
Guido van Rossum | e03c050 | 1998-08-12 02:38:11 +0000 | [diff] [blame] | 38 | seekable = 0 |
Guido van Rossum | aad6761 | 2000-05-08 17:31:04 +0000 | [diff] [blame] | 39 | |
Guido van Rossum | e03c050 | 1998-08-12 02:38:11 +0000 | [diff] [blame] | 40 | def __init__(self, fp, seekable=1): |
Guido van Rossum | 5c97167 | 1996-07-22 15:23:25 +0000 | [diff] [blame] | 41 | self.fp = fp |
| 42 | self.stack = [] # Grows down |
| 43 | self.level = 0 |
| 44 | self.last = 0 |
Guido van Rossum | e03c050 | 1998-08-12 02:38:11 +0000 | [diff] [blame] | 45 | if seekable: |
| 46 | self.seekable = 1 |
| 47 | self.start = self.fp.tell() |
| 48 | self.posstack = [] # Grows down |
Guido van Rossum | aad6761 | 2000-05-08 17:31:04 +0000 | [diff] [blame] | 49 | |
Guido van Rossum | 5c97167 | 1996-07-22 15:23:25 +0000 | [diff] [blame] | 50 | def tell(self): |
| 51 | if self.level > 0: |
| 52 | return self.lastpos |
| 53 | return self.fp.tell() - self.start |
Guido van Rossum | aad6761 | 2000-05-08 17:31:04 +0000 | [diff] [blame] | 54 | |
Guido van Rossum | 548703a | 1998-03-26 22:14:20 +0000 | [diff] [blame] | 55 | def seek(self, pos, whence=0): |
| 56 | here = self.tell() |
| 57 | if whence: |
| 58 | if whence == 1: |
| 59 | pos = pos + here |
| 60 | elif whence == 2: |
| 61 | if self.level > 0: |
| 62 | pos = pos + self.lastpos |
| 63 | else: |
| 64 | raise Error, "can't use whence=2 yet" |
| 65 | if not 0 <= pos <= here or \ |
Guido van Rossum | 5c97167 | 1996-07-22 15:23:25 +0000 | [diff] [blame] | 66 | self.level > 0 and pos > self.lastpos: |
| 67 | raise Error, 'bad MultiFile.seek() call' |
| 68 | self.fp.seek(pos + self.start) |
| 69 | self.level = 0 |
| 70 | self.last = 0 |
Guido van Rossum | aad6761 | 2000-05-08 17:31:04 +0000 | [diff] [blame] | 71 | |
Guido van Rossum | 5c97167 | 1996-07-22 15:23:25 +0000 | [diff] [blame] | 72 | def readline(self): |
Guido van Rossum | e03c050 | 1998-08-12 02:38:11 +0000 | [diff] [blame] | 73 | if self.level > 0: |
| 74 | return '' |
Guido van Rossum | 5c97167 | 1996-07-22 15:23:25 +0000 | [diff] [blame] | 75 | line = self.fp.readline() |
Guido van Rossum | e03c050 | 1998-08-12 02:38:11 +0000 | [diff] [blame] | 76 | # Real EOF? |
Guido van Rossum | 5c97167 | 1996-07-22 15:23:25 +0000 | [diff] [blame] | 77 | if not line: |
| 78 | self.level = len(self.stack) |
| 79 | self.last = (self.level > 0) |
| 80 | if self.last: |
Guido van Rossum | e03c050 | 1998-08-12 02:38:11 +0000 | [diff] [blame] | 81 | raise Error, 'sudden EOF in MultiFile.readline()' |
Guido van Rossum | 5c97167 | 1996-07-22 15:23:25 +0000 | [diff] [blame] | 82 | return '' |
Guido van Rossum | e03c050 | 1998-08-12 02:38:11 +0000 | [diff] [blame] | 83 | assert self.level == 0 |
| 84 | # Fast check to see if this is just data |
| 85 | if self.is_data(line): |
| 86 | return line |
| 87 | else: |
| 88 | # Ignore trailing whitespace on marker lines |
| 89 | k = len(line) - 1; |
| 90 | while line[k] in string.whitespace: |
| 91 | k = k - 1 |
| 92 | marker = line[:k+1] |
| 93 | # No? OK, try to match a boundary. |
| 94 | # Return the line (unstripped) if we don't. |
Guido van Rossum | 5c97167 | 1996-07-22 15:23:25 +0000 | [diff] [blame] | 95 | for i in range(len(self.stack)): |
| 96 | sep = self.stack[i] |
Guido van Rossum | e03c050 | 1998-08-12 02:38:11 +0000 | [diff] [blame] | 97 | if marker == self.section_divider(sep): |
Guido van Rossum | 5c97167 | 1996-07-22 15:23:25 +0000 | [diff] [blame] | 98 | self.last = 0 |
| 99 | break |
Guido van Rossum | e03c050 | 1998-08-12 02:38:11 +0000 | [diff] [blame] | 100 | elif marker == self.end_marker(sep): |
Guido van Rossum | 5c97167 | 1996-07-22 15:23:25 +0000 | [diff] [blame] | 101 | self.last = 1 |
| 102 | break |
| 103 | else: |
| 104 | return line |
Guido van Rossum | e03c050 | 1998-08-12 02:38:11 +0000 | [diff] [blame] | 105 | # We only get here if we see a section divider or EOM line |
| 106 | if self.seekable: |
| 107 | self.lastpos = self.tell() - len(line) |
Guido van Rossum | 5c97167 | 1996-07-22 15:23:25 +0000 | [diff] [blame] | 108 | self.level = i+1 |
| 109 | if self.level > 1: |
Guido van Rossum | e03c050 | 1998-08-12 02:38:11 +0000 | [diff] [blame] | 110 | raise Error,'Missing endmarker in MultiFile.readline()' |
Guido van Rossum | 5c97167 | 1996-07-22 15:23:25 +0000 | [diff] [blame] | 111 | return '' |
Guido van Rossum | aad6761 | 2000-05-08 17:31:04 +0000 | [diff] [blame] | 112 | |
Guido van Rossum | 5c97167 | 1996-07-22 15:23:25 +0000 | [diff] [blame] | 113 | def readlines(self): |
| 114 | list = [] |
| 115 | while 1: |
| 116 | line = self.readline() |
| 117 | if not line: break |
| 118 | list.append(line) |
| 119 | return list |
Guido van Rossum | aad6761 | 2000-05-08 17:31:04 +0000 | [diff] [blame] | 120 | |
Guido van Rossum | 5c97167 | 1996-07-22 15:23:25 +0000 | [diff] [blame] | 121 | def read(self): # Note: no size argument -- read until EOF only! |
| 122 | return string.joinfields(self.readlines(), '') |
Guido van Rossum | aad6761 | 2000-05-08 17:31:04 +0000 | [diff] [blame] | 123 | |
Guido van Rossum | 5c97167 | 1996-07-22 15:23:25 +0000 | [diff] [blame] | 124 | def next(self): |
| 125 | while self.readline(): pass |
| 126 | if self.level > 1 or self.last: |
| 127 | return 0 |
| 128 | self.level = 0 |
| 129 | self.last = 0 |
Guido van Rossum | e03c050 | 1998-08-12 02:38:11 +0000 | [diff] [blame] | 130 | if self.seekable: |
| 131 | self.start = self.fp.tell() |
Guido van Rossum | 5c97167 | 1996-07-22 15:23:25 +0000 | [diff] [blame] | 132 | return 1 |
Guido van Rossum | aad6761 | 2000-05-08 17:31:04 +0000 | [diff] [blame] | 133 | |
Guido van Rossum | 5c97167 | 1996-07-22 15:23:25 +0000 | [diff] [blame] | 134 | def push(self, sep): |
| 135 | if self.level > 0: |
| 136 | raise Error, 'bad MultiFile.push() call' |
| 137 | self.stack.insert(0, sep) |
Guido van Rossum | e03c050 | 1998-08-12 02:38:11 +0000 | [diff] [blame] | 138 | if self.seekable: |
| 139 | self.posstack.insert(0, self.start) |
| 140 | self.start = self.fp.tell() |
Guido van Rossum | aad6761 | 2000-05-08 17:31:04 +0000 | [diff] [blame] | 141 | |
Guido van Rossum | 5c97167 | 1996-07-22 15:23:25 +0000 | [diff] [blame] | 142 | def pop(self): |
| 143 | if self.stack == []: |
| 144 | raise Error, 'bad MultiFile.pop() call' |
| 145 | if self.level <= 1: |
| 146 | self.last = 0 |
| 147 | else: |
| 148 | abslastpos = self.lastpos + self.start |
| 149 | self.level = max(0, self.level - 1) |
| 150 | del self.stack[0] |
Guido van Rossum | e03c050 | 1998-08-12 02:38:11 +0000 | [diff] [blame] | 151 | if self.seekable: |
| 152 | self.start = self.posstack[0] |
| 153 | del self.posstack[0] |
| 154 | if self.level > 0: |
| 155 | self.lastpos = abslastpos - self.start |
Guido van Rossum | aad6761 | 2000-05-08 17:31:04 +0000 | [diff] [blame] | 156 | |
Guido van Rossum | e03c050 | 1998-08-12 02:38:11 +0000 | [diff] [blame] | 157 | def is_data(self, line): |
| 158 | return line[:2] <> '--' |
Guido van Rossum | aad6761 | 2000-05-08 17:31:04 +0000 | [diff] [blame] | 159 | |
Guido van Rossum | e03c050 | 1998-08-12 02:38:11 +0000 | [diff] [blame] | 160 | def section_divider(self, str): |
| 161 | return "--" + str |
Guido van Rossum | aad6761 | 2000-05-08 17:31:04 +0000 | [diff] [blame] | 162 | |
Guido van Rossum | e03c050 | 1998-08-12 02:38:11 +0000 | [diff] [blame] | 163 | def end_marker(self, str): |
| 164 | return "--" + str + "--" |