Guido van Rossum | 54f22ed | 2000-02-04 15:10:34 +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: |
Tim Peters | 07e99cb | 2001-01-14 23:47:14 +0000 | [diff] [blame] | 16 | "read lines from fp until it returns an empty string" (A) |
| 17 | if not fp.next(): break |
Guido van Rossum | 54f22ed | 2000-02-04 15:10:34 +0000 | [diff] [blame] | 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 | 54f22ed | 2000-02-04 15:10:34 +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 | 741c81a | 1992-07-13 14:40:45 +0000 | [diff] [blame] | 29 | |
| 30 | import sys |
| 31 | import string |
| 32 | |
Skip Montanaro | 269b83b | 2001-02-06 01:07:02 +0000 | [diff] [blame^] | 33 | __all__ = ["MultiFile","Error"] |
| 34 | |
Fred Drake | 227b120 | 2000-08-17 05:06:49 +0000 | [diff] [blame] | 35 | class Error(Exception): |
Tim Peters | 07e99cb | 2001-01-14 23:47:14 +0000 | [diff] [blame] | 36 | pass |
Guido van Rossum | 741c81a | 1992-07-13 14:40:45 +0000 | [diff] [blame] | 37 | |
| 38 | class MultiFile: |
Guido van Rossum | 54f22ed | 2000-02-04 15:10:34 +0000 | [diff] [blame] | 39 | |
Tim Peters | 07e99cb | 2001-01-14 23:47:14 +0000 | [diff] [blame] | 40 | seekable = 0 |
Guido van Rossum | 54f22ed | 2000-02-04 15:10:34 +0000 | [diff] [blame] | 41 | |
Tim Peters | 07e99cb | 2001-01-14 23:47:14 +0000 | [diff] [blame] | 42 | def __init__(self, fp, seekable=1): |
| 43 | self.fp = fp |
| 44 | self.stack = [] # Grows down |
| 45 | self.level = 0 |
| 46 | self.last = 0 |
| 47 | if seekable: |
| 48 | self.seekable = 1 |
| 49 | self.start = self.fp.tell() |
| 50 | self.posstack = [] # Grows down |
Guido van Rossum | 54f22ed | 2000-02-04 15:10:34 +0000 | [diff] [blame] | 51 | |
Tim Peters | 07e99cb | 2001-01-14 23:47:14 +0000 | [diff] [blame] | 52 | def tell(self): |
| 53 | if self.level > 0: |
| 54 | return self.lastpos |
| 55 | return self.fp.tell() - self.start |
Guido van Rossum | 54f22ed | 2000-02-04 15:10:34 +0000 | [diff] [blame] | 56 | |
Tim Peters | 07e99cb | 2001-01-14 23:47:14 +0000 | [diff] [blame] | 57 | def seek(self, pos, whence=0): |
| 58 | here = self.tell() |
| 59 | if whence: |
| 60 | if whence == 1: |
| 61 | pos = pos + here |
| 62 | elif whence == 2: |
| 63 | if self.level > 0: |
| 64 | pos = pos + self.lastpos |
| 65 | else: |
| 66 | raise Error, "can't use whence=2 yet" |
| 67 | if not 0 <= pos <= here or \ |
| 68 | self.level > 0 and pos > self.lastpos: |
| 69 | raise Error, 'bad MultiFile.seek() call' |
| 70 | self.fp.seek(pos + self.start) |
| 71 | self.level = 0 |
| 72 | self.last = 0 |
Guido van Rossum | 54f22ed | 2000-02-04 15:10:34 +0000 | [diff] [blame] | 73 | |
Tim Peters | 07e99cb | 2001-01-14 23:47:14 +0000 | [diff] [blame] | 74 | def readline(self): |
| 75 | if self.level > 0: |
| 76 | return '' |
| 77 | line = self.fp.readline() |
| 78 | # Real EOF? |
| 79 | if not line: |
| 80 | self.level = len(self.stack) |
| 81 | self.last = (self.level > 0) |
| 82 | if self.last: |
| 83 | raise Error, 'sudden EOF in MultiFile.readline()' |
| 84 | return '' |
| 85 | assert self.level == 0 |
| 86 | # Fast check to see if this is just data |
| 87 | if self.is_data(line): |
| 88 | return line |
| 89 | else: |
| 90 | # Ignore trailing whitespace on marker lines |
| 91 | k = len(line) - 1 |
| 92 | while line[k] in string.whitespace: |
| 93 | k = k - 1 |
| 94 | marker = line[:k+1] |
| 95 | # No? OK, try to match a boundary. |
| 96 | # Return the line (unstripped) if we don't. |
| 97 | for i in range(len(self.stack)): |
| 98 | sep = self.stack[i] |
| 99 | if marker == self.section_divider(sep): |
| 100 | self.last = 0 |
| 101 | break |
| 102 | elif marker == self.end_marker(sep): |
| 103 | self.last = 1 |
| 104 | break |
| 105 | else: |
| 106 | return line |
| 107 | # We only get here if we see a section divider or EOM line |
| 108 | if self.seekable: |
| 109 | self.lastpos = self.tell() - len(line) |
| 110 | self.level = i+1 |
| 111 | if self.level > 1: |
| 112 | raise Error,'Missing endmarker in MultiFile.readline()' |
| 113 | return '' |
Guido van Rossum | 54f22ed | 2000-02-04 15:10:34 +0000 | [diff] [blame] | 114 | |
Tim Peters | 07e99cb | 2001-01-14 23:47:14 +0000 | [diff] [blame] | 115 | def readlines(self): |
| 116 | list = [] |
| 117 | while 1: |
| 118 | line = self.readline() |
| 119 | if not line: break |
| 120 | list.append(line) |
| 121 | return list |
Guido van Rossum | 54f22ed | 2000-02-04 15:10:34 +0000 | [diff] [blame] | 122 | |
Tim Peters | 07e99cb | 2001-01-14 23:47:14 +0000 | [diff] [blame] | 123 | def read(self): # Note: no size argument -- read until EOF only! |
| 124 | return string.joinfields(self.readlines(), '') |
Guido van Rossum | 54f22ed | 2000-02-04 15:10:34 +0000 | [diff] [blame] | 125 | |
Tim Peters | 07e99cb | 2001-01-14 23:47:14 +0000 | [diff] [blame] | 126 | def next(self): |
| 127 | while self.readline(): pass |
| 128 | if self.level > 1 or self.last: |
| 129 | return 0 |
| 130 | self.level = 0 |
| 131 | self.last = 0 |
| 132 | if self.seekable: |
| 133 | self.start = self.fp.tell() |
| 134 | return 1 |
Guido van Rossum | 54f22ed | 2000-02-04 15:10:34 +0000 | [diff] [blame] | 135 | |
Tim Peters | 07e99cb | 2001-01-14 23:47:14 +0000 | [diff] [blame] | 136 | def push(self, sep): |
| 137 | if self.level > 0: |
| 138 | raise Error, 'bad MultiFile.push() call' |
| 139 | self.stack.insert(0, sep) |
| 140 | if self.seekable: |
| 141 | self.posstack.insert(0, self.start) |
| 142 | self.start = self.fp.tell() |
Guido van Rossum | 54f22ed | 2000-02-04 15:10:34 +0000 | [diff] [blame] | 143 | |
Tim Peters | 07e99cb | 2001-01-14 23:47:14 +0000 | [diff] [blame] | 144 | def pop(self): |
| 145 | if self.stack == []: |
| 146 | raise Error, 'bad MultiFile.pop() call' |
| 147 | if self.level <= 1: |
| 148 | self.last = 0 |
| 149 | else: |
| 150 | abslastpos = self.lastpos + self.start |
| 151 | self.level = max(0, self.level - 1) |
| 152 | del self.stack[0] |
| 153 | if self.seekable: |
| 154 | self.start = self.posstack[0] |
| 155 | del self.posstack[0] |
| 156 | if self.level > 0: |
| 157 | self.lastpos = abslastpos - self.start |
Guido van Rossum | 54f22ed | 2000-02-04 15:10:34 +0000 | [diff] [blame] | 158 | |
Tim Peters | 07e99cb | 2001-01-14 23:47:14 +0000 | [diff] [blame] | 159 | def is_data(self, line): |
| 160 | return line[:2] != '--' |
Guido van Rossum | 54f22ed | 2000-02-04 15:10:34 +0000 | [diff] [blame] | 161 | |
Tim Peters | 07e99cb | 2001-01-14 23:47:14 +0000 | [diff] [blame] | 162 | def section_divider(self, str): |
| 163 | return "--" + str |
Guido van Rossum | 54f22ed | 2000-02-04 15:10:34 +0000 | [diff] [blame] | 164 | |
Tim Peters | 07e99cb | 2001-01-14 23:47:14 +0000 | [diff] [blame] | 165 | def end_marker(self, str): |
| 166 | return "--" + str + "--" |