blob: 47e2346978a9ffac2ff0e289f884f67a04bd8a51 [file] [log] [blame]
Guido van Rossum54f22ed2000-02-04 15:10:34 +00001"""A readline()-style interface to the parts of a multipart message.
2
3The MultiFile class makes each part of a multipart message "feel" like
4an ordinary file, as long as you use fp.readline(). Allows recursive
5use, for nested multipart messages. Probably best used together
6with module mimetools.
7
8Suggested use:
9
10real_fp = open(...)
11fp = MultiFile(real_fp)
12
13"read some lines from fp"
14fp.push(separator)
15while 1:
Tim Peters07e99cb2001-01-14 23:47:14 +000016 "read lines from fp until it returns an empty string" (A)
17 if not fp.next(): break
Guido van Rossum54f22ed2000-02-04 15:10:34 +000018fp.pop()
19"read remaining lines from fp until it returns an empty string"
20
21The latter sequence may be used recursively at (A).
22It is also allowed to use multiple push()...pop() sequences.
23
Thomas Wouters7e474022000-07-16 12:04:32 +000024If seekable is given as 0, the class code will not do the bookkeeping
Guido van Rossum54f22ed2000-02-04 15:10:34 +000025it normally attempts in order to make seeks relative to the beginning of the
26current file part. This may be useful when using MultiFile with a non-
27seekable stream object.
28"""
Guido van Rossum741c81a1992-07-13 14:40:45 +000029
Skip Montanaro269b83b2001-02-06 01:07:02 +000030__all__ = ["MultiFile","Error"]
31
Fred Drake227b1202000-08-17 05:06:49 +000032class Error(Exception):
Tim Peters07e99cb2001-01-14 23:47:14 +000033 pass
Guido van Rossum741c81a1992-07-13 14:40:45 +000034
35class MultiFile:
Guido van Rossum54f22ed2000-02-04 15:10:34 +000036
Tim Peters07e99cb2001-01-14 23:47:14 +000037 seekable = 0
Guido van Rossum54f22ed2000-02-04 15:10:34 +000038
Tim Peters07e99cb2001-01-14 23:47:14 +000039 def __init__(self, fp, seekable=1):
40 self.fp = fp
41 self.stack = [] # Grows down
42 self.level = 0
43 self.last = 0
Guido van Rossum912e56c2001-09-18 14:34:06 +000044 self.readahead = ""
Tim Peters07e99cb2001-01-14 23:47:14 +000045 if seekable:
46 self.seekable = 1
47 self.start = self.fp.tell()
48 self.posstack = [] # Grows down
Guido van Rossum54f22ed2000-02-04 15:10:34 +000049
Tim Peters07e99cb2001-01-14 23:47:14 +000050 def tell(self):
51 if self.level > 0:
52 return self.lastpos
Guido van Rossum912e56c2001-09-18 14:34:06 +000053 return self.fp.tell() - len(self.readahead) - self.start
Guido van Rossum54f22ed2000-02-04 15:10:34 +000054
Tim Peters07e99cb2001-01-14 23:47:14 +000055 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 \
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 Rossum912e56c2001-09-18 14:34:06 +000071 self.readahead = ""
Guido van Rossum54f22ed2000-02-04 15:10:34 +000072
Tim Peters07e99cb2001-01-14 23:47:14 +000073 def readline(self):
Guido van Rossum912e56c2001-09-18 14:34:06 +000074 if not self.readahead:
75 self.readahead = self._readline()
76 line = self.readahead
77 if line:
78 self.readahead = self._readline()
79 if not self.readahead and line[-1:] == "\n":
80 line = line[:-1]
81 return line
82
83 def _readline(self):
Tim Peters07e99cb2001-01-14 23:47:14 +000084 if self.level > 0:
85 return ''
86 line = self.fp.readline()
87 # Real EOF?
88 if not line:
89 self.level = len(self.stack)
90 self.last = (self.level > 0)
91 if self.last:
92 raise Error, 'sudden EOF in MultiFile.readline()'
93 return ''
94 assert self.level == 0
95 # Fast check to see if this is just data
96 if self.is_data(line):
97 return line
98 else:
99 # Ignore trailing whitespace on marker lines
Eric S. Raymonde37340e2001-02-09 16:56:44 +0000100 marker = line.rstrip()
Tim Peters07e99cb2001-01-14 23:47:14 +0000101 # No? OK, try to match a boundary.
102 # Return the line (unstripped) if we don't.
103 for i in range(len(self.stack)):
104 sep = self.stack[i]
105 if marker == self.section_divider(sep):
106 self.last = 0
107 break
108 elif marker == self.end_marker(sep):
109 self.last = 1
110 break
111 else:
112 return line
113 # We only get here if we see a section divider or EOM line
114 if self.seekable:
115 self.lastpos = self.tell() - len(line)
116 self.level = i+1
117 if self.level > 1:
118 raise Error,'Missing endmarker in MultiFile.readline()'
119 return ''
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000120
Tim Peters07e99cb2001-01-14 23:47:14 +0000121 def readlines(self):
122 list = []
123 while 1:
124 line = self.readline()
125 if not line: break
126 list.append(line)
127 return list
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000128
Tim Peters07e99cb2001-01-14 23:47:14 +0000129 def read(self): # Note: no size argument -- read until EOF only!
Fred Drake521c83d2001-03-12 02:56:15 +0000130 return ''.join(self.readlines())
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000131
Tim Peters07e99cb2001-01-14 23:47:14 +0000132 def next(self):
133 while self.readline(): pass
134 if self.level > 1 or self.last:
135 return 0
136 self.level = 0
137 self.last = 0
138 if self.seekable:
139 self.start = self.fp.tell()
140 return 1
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000141
Tim Peters07e99cb2001-01-14 23:47:14 +0000142 def push(self, sep):
143 if self.level > 0:
144 raise Error, 'bad MultiFile.push() call'
145 self.stack.insert(0, sep)
146 if self.seekable:
147 self.posstack.insert(0, self.start)
148 self.start = self.fp.tell()
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000149
Tim Peters07e99cb2001-01-14 23:47:14 +0000150 def pop(self):
151 if self.stack == []:
152 raise Error, 'bad MultiFile.pop() call'
153 if self.level <= 1:
154 self.last = 0
155 else:
156 abslastpos = self.lastpos + self.start
157 self.level = max(0, self.level - 1)
158 del self.stack[0]
159 if self.seekable:
160 self.start = self.posstack[0]
161 del self.posstack[0]
162 if self.level > 0:
163 self.lastpos = abslastpos - self.start
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000164
Tim Peters07e99cb2001-01-14 23:47:14 +0000165 def is_data(self, line):
166 return line[:2] != '--'
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000167
Tim Peters07e99cb2001-01-14 23:47:14 +0000168 def section_divider(self, str):
169 return "--" + str
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000170
Tim Peters07e99cb2001-01-14 23:47:14 +0000171 def end_marker(self, str):
172 return "--" + str + "--"