Lib/quopri.py

   1 #! /usr/bin/env python
   2
   3 """Conversions to/from quoted-printable transport encoding as per RFC 1521."""
   4
   5 # (Dec 1991 version).
   6
   7 __all__ = ["encode", "decode", "encodestring", "decodestring"]
   8
   9 ESCAPE = '='
  10 MAXLINESIZE = 76
  11 HEX = '0123456789ABCDEF'
  12 EMPTYSTRING = ''
  13
  14 try:
  15     from binascii import a2b_qp, b2a_qp
  16 except ImportError:
  17     a2b_qp = None
  18     b2a_qp = None
  19
  20
  21 def needsquoting(c, quotetabs, header):
  22     """Decide whether a particular character needs to be quoted.
  23
  24     The 'quotetabs' flag indicates whether embedded tabs and spaces should be
  25     quoted.  Note that line-ending tabs and spaces are always encoded, as per
  26     RFC 1521.
  27     """
  28     if c in ' \t':
  29         return quotetabs
  30     # if header, we have to escape _ because _ is used to escape space
  31     if c == '_':
  32         return header
  33     return c == ESCAPE or not (' ' <= c <= '~')
  34
  35 def quote(c):
  36     """Quote a single character."""
  37     i = ord(c)
  38     return ESCAPE + HEX[i//16] + HEX[i%16]
  39
  40
  41
  42 def encode(input, output, quotetabs, header = 0):
  43     """Read 'input', apply quoted-printable encoding, and write to 'output'.
  44
  45     'input' and 'output' are files with readline() and write() methods.
  46     The 'quotetabs' flag indicates whether embedded tabs and spaces should be
  47     quoted.  Note that line-ending tabs and spaces are always encoded, as per
  48     RFC 1521.
  49     The 'header' flag indicates whether we are encoding spaces as _ as per
  50     RFC 1522.
  51     """
  52
  53     if b2a_qp is not None:
  54         data = input.read()
  55         odata = b2a_qp(data, quotetabs = quotetabs, header = header)
  56         output.write(odata)
  57         return
  58
  59     def write(s, output=output, lineEnd='\n'):
  60         # RFC 1521 requires that the line ending in a space or tab must have
  61         # that trailing character encoded.
  62         if s and s[-1:] in ' \t':
  63             output.write(s[:-1] + quote(s[-1]) + lineEnd)
  64         elif s == '.':
  65             output.write(quote(s) + lineEnd)
  66         else:
  67             output.write(s + lineEnd)
  68
  69     prevline = None
  70     while 1:
  71         line = input.readline()
  72         if not line:
  73             break
  74         outline = []
  75         # Strip off any readline induced trailing newline
  76         stripped = ''
  77         if line[-1:] == '\n':
  78             line = line[:-1]
  79             stripped = '\n'
  80         # Calculate the un-length-limited encoded line
  81         for c in line:
  82             if needsquoting(c, quotetabs, header):
  83                 c = quote(c)
  84             if header and c == ' ':
  85                 outline.append('_')
  86             else:
  87                 outline.append(c)
  88         # First, write out the previous line
  89         if prevline is not None:
  90             write(prevline)
  91         # Now see if we need any soft line breaks because of RFC-imposed
  92         # length limitations.  Then do the thisline->prevline dance.
  93         thisline = EMPTYSTRING.join(outline)
  94         while len(thisline) > MAXLINESIZE:
  95             # Don't forget to include the soft line break `=' sign in the
  96             # length calculation!
  97             write(thisline[:MAXLINESIZE-1], lineEnd='=\n')
  98             thisline = thisline[MAXLINESIZE-1:]
  99         # Write out the current line
 100         prevline = thisline
 101     # Write out the last line, without a trailing newline
 102     if prevline is not None:
 103         write(prevline, lineEnd=stripped)
 104
 105 def encodestring(s, quotetabs = 0, header = 0):
 106     if b2a_qp is not None:
 107         return b2a_qp(s, quotetabs = quotetabs, header = header)
 108     from cStringIO import StringIO
 109     infp = StringIO(s)
 110     outfp = StringIO()
 111     encode(infp, outfp, quotetabs, header)
 112     return outfp.getvalue()
 113
 114
 115
 116 def decode(input, output, header = 0):
 117     """Read 'input', apply quoted-printable decoding, and write to 'output'.
 118     'input' and 'output' are files with readline() and write() methods.
 119     If 'header' is true, decode underscore as space (per RFC 1522)."""
 120
 121     if a2b_qp is not None:
 122         data = input.read()
 123         odata = a2b_qp(data, header = header)
 124         output.write(odata)
 125         return
 126
 127     new = ''
 128     while 1:
 129         line = input.readline()
 130         if not line: break
 131         i, n = 0, len(line)
 132         if n > 0 and line[n-1] == '\n':
 133             partial = 0; n = n-1
 134             # Strip trailing whitespace
 135             while n > 0 and line[n-1] in " \t\r":
 136                 n = n-1
 137         else:
 138             partial = 1
 139         while i < n:
 140             c = line[i]
 141             if c == '_' and header:
 142                 new = new + ' '; i = i+1
 143             elif c != ESCAPE:
 144                 new = new + c; i = i+1
 145             elif i+1 == n and not partial:
 146                 partial = 1; break
 147             elif i+1 < n and line[i+1] == ESCAPE:
 148                 new = new + ESCAPE; i = i+2
 149             elif i+2 < n and ishex(line[i+1]) and ishex(line[i+2]):
 150                 new = new + chr(unhex(line[i+1:i+3])); i = i+3
 151             else: # Bad escape sequence -- leave it in
 152                 new = new + c; i = i+1
 153         if not partial:
 154             output.write(new + '\n')
 155             new = ''
 156     if new:
 157         output.write(new)
 158
 159 def decodestring(s, header = 0):
 160     if a2b_qp is not None:
 161         return a2b_qp(s, header = header)
 162     from cStringIO import StringIO
 163     infp = StringIO(s)
 164     outfp = StringIO()
 165     decode(infp, outfp, header = header)
 166     return outfp.getvalue()
 167
 168
 169
 170 # Other helper functions
 171 def ishex(c):
 172     """Return true if the character 'c' is a hexadecimal digit."""
 173     return '0' <= c <= '9' or 'a' <= c <= 'f' or 'A' <= c <= 'F'
 174
 175 def unhex(s):
 176     """Get the integer value of a hexadecimal number."""
 177     bits = 0
 178     for c in s:
 179         if '0' <= c <= '9':
 180             i = ord('0')
 181         elif 'a' <= c <= 'f':
 182             i = ord('a')-10
 183         elif 'A' <= c <= 'F':
 184             i = ord('A')-10
 185         else:
 186             break
 187         bits = bits*16 + (ord(c) - i)
 188     return bits
 189
 190
 191
 192 def main():
 193     import sys
 194     import getopt
 195     try:
 196         opts, args = getopt.getopt(sys.argv[1:], 'td')
 197     except getopt.error, msg:
 198         sys.stdout = sys.stderr
 199         print msg
 200         print "usage: quopri [-t | -d] [file] ..."
 201         print "-t: quote tabs"
 202         print "-d: decode; default encode"
 203         sys.exit(2)
 204     deco = 0
 205     tabs = 0
 206     for o, a in opts:
 207         if o == '-t': tabs = 1
 208         if o == '-d': deco = 1
 209     if tabs and deco:
 210         sys.stdout = sys.stderr
 211         print "-t and -d are mutually exclusive"
 212         sys.exit(2)
 213     if not args: args = ['-']
 214     sts = 0
 215     for file in args:
 216         if file == '-':
 217             fp = sys.stdin
 218         else:
 219             try:
 220                 fp = open(file)
 221             except IOError, msg:
 222                 sys.stderr.write("%s: can't open (%s)\n" % (file, msg))
 223                 sts = 1
 224                 continue
 225         if deco:
 226             decode(fp, sys.stdout)
 227         else:
 228             encode(fp, sys.stdout, tabs)
 229         if fp is not sys.stdin:
 230             fp.close()
 231     if sts:
 232         sys.exit(sts)
 233
 234
 235
 236 if __name__ == '__main__':
 237     main()