blob: 38a5a46069f3fdffb979413db4ce88c4d6852417 [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
30import sys
31import string
32
Fred Drake227b1202000-08-17 05:06:49 +000033class Error(Exception):
Tim Peters07e99cb2001-01-14 23:47:14 +000034 pass
Guido van Rossum741c81a1992-07-13 14:40:45 +000035
36class MultiFile:
Guido van Rossum54f22ed2000-02-04 15:10:34 +000037
Tim Peters07e99cb2001-01-14 23:47:14 +000038 seekable = 0
Guido van Rossum54f22ed2000-02-04 15:10:34 +000039
Tim Peters07e99cb2001-01-14 23:47:14 +000040 def __init__(self, fp, seekable=1):
41 self.fp = fp
42 self.stack = [] # Grows down
43 self.level = 0
44 self.last = 0
45 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
53 return self.fp.tell() - 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 Rossum54f22ed2000-02-04 15:10:34 +000071
Tim Peters07e99cb2001-01-14 23:47:14 +000072 def readline(self):
73 if self.level > 0:
74 return ''
75 line = self.fp.readline()
76 # Real EOF?
77 if not line:
78 self.level = len(self.stack)
79 self.last = (self.level > 0)
80 if self.last:
81 raise Error, 'sudden EOF in MultiFile.readline()'
82 return ''
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.
95 for i in range(len(self.stack)):
96 sep = self.stack[i]
97 if marker == self.section_divider(sep):
98 self.last = 0
99 break
100 elif marker == self.end_marker(sep):
101 self.last = 1
102 break
103 else:
104 return line
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)
108 self.level = i+1
109 if self.level > 1:
110 raise Error,'Missing endmarker in MultiFile.readline()'
111 return ''
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000112
Tim Peters07e99cb2001-01-14 23:47:14 +0000113 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 Rossum54f22ed2000-02-04 15:10:34 +0000120
Tim Peters07e99cb2001-01-14 23:47:14 +0000121 def read(self): # Note: no size argument -- read until EOF only!
122 return string.joinfields(self.readlines(), '')
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000123
Tim Peters07e99cb2001-01-14 23:47:14 +0000124 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
130 if self.seekable:
131 self.start = self.fp.tell()
132 return 1
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000133
Tim Peters07e99cb2001-01-14 23:47:14 +0000134 def push(self, sep):
135 if self.level > 0:
136 raise Error, 'bad MultiFile.push() call'
137 self.stack.insert(0, sep)
138 if self.seekable:
139 self.posstack.insert(0, self.start)
140 self.start = self.fp.tell()
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000141
Tim Peters07e99cb2001-01-14 23:47:14 +0000142 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]
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 Rossum54f22ed2000-02-04 15:10:34 +0000156
Tim Peters07e99cb2001-01-14 23:47:14 +0000157 def is_data(self, line):
158 return line[:2] != '--'
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000159
Tim Peters07e99cb2001-01-14 23:47:14 +0000160 def section_divider(self, str):
161 return "--" + str
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000162
Tim Peters07e99cb2001-01-14 23:47:14 +0000163 def end_marker(self, str):
164 return "--" + str + "--"