wrw/form.py

   1 import urlparse
   2 import proto
   3
   4 __all__ = ["formdata"]
   5
   6 def formparse(req):
   7     buf = {}
   8     buf.update(urlparse.parse_qsl(req.query))
   9     if req.ihead.get("Content-Type") == "application/x-www-form-urlencoded":
  10         try:
  11             rbody = req.input.read(2 ** 20)
  12         except IOError as exc:
  13             return exc
  14         if len(rbody) >= 2 ** 20:
  15             return ValueError("x-www-form-urlencoded data is absurdly long")
  16         buf.update(urlparse.parse_qsl(rbody))
  17     return buf
  18
  19 class badmultipart(Exception):
  20     pass
  21
  22 class formpart(object):
  23     def __init__(self, form):
  24         self.form = form
  25         self.buf = ""
  26         self.eof = False
  27         self.head = {}
  28
  29     def parsehead(self):
  30         pass
  31
  32     def fillbuf(self, sz):
  33         req = self.form.req
  34         mboundary = "\r\n--" + self.form.boundary + "\r\n"
  35         lboundary = "\r\n--" + self.form.boundary + "--\r\n"
  36         while not self.eof:
  37             p = self.form.buf.find(mboundary)
  38             if p >= 0:
  39                 self.buf += self.form.buf[:p]
  40                 self.form.buf = self.form.buf[p + len(mboundary):]
  41                 self.eof = True
  42                 break
  43             p = self.form.buf.find(lboundary)
  44             if p >= 0:
  45                 self.buf += self.form.buf[:p]
  46                 self.form.buf = self.form.buf[p + len(lboundary):]
  47                 self.eof = True
  48                 self.form.eof = True
  49                 break
  50             self.buf += self.form.buf[:-len(lboundary)]
  51             self.form.buf = self.form.buf[-len(lboundary):]
  52             if sz >= 0 and len(self.buf) >= sz:
  53                 break
  54             while len(self.form.buf) <= len(lboundary):
  55                 ret = req.input.read(8192)
  56                 if ret == "":
  57                     raise badmultipart("Missing last multipart boundary")
  58                 self.form.buf += ret
  59
  60     def read(self, limit=-1):
  61         self.fillbuf(limit)
  62         if limit >= 0:
  63             ret = self.buf[:limit]
  64             self.buf = self.buf[limit:]
  65         else:
  66             ret = self.buf
  67             self.buf = ""
  68         return ret
  69
  70     def readline(self, limit=-1):
  71         last = 0
  72         while True:
  73             p = self.buf.find('\n', last)
  74             if p < 0:
  75                 if self.eof:
  76                     ret = self.buf
  77                     self.buf = ""
  78                     return ret
  79                 last = len(self.buf)
  80                 self.fillbuf(last + 128)
  81             else:
  82                 ret = self.buf[:p + 1]
  83                 self.buf = self.buf[p + 1:]
  84                 return ret
  85
  86     def close(self):
  87         while True:
  88             if self.read(8192) == "":
  89                 break
  90
  91     def __enter__(self):
  92         return self
  93
  94     def __exit__(self, *excinfo):
  95         self.close()
  96         return False
  97
  98     def parsehead(self):
  99         def headline():
 100             ln = self.readline(256)
 101             if ln[-1] != '\n':
 102                 raise badmultipart("Too long header line in part")
 103             return ln.rstrip()
 104
 105         ln = headline()
 106         while True:
 107             if ln == "":
 108                 break
 109             buf = ln
 110             while True:
 111                 ln = headline()
 112                 if not ln[1:].isspace():
 113                     break
 114                 buf += ln.lstrip()
 115             p = buf.find(':')
 116             if p < 0:
 117                 raise badmultipart("Malformed multipart header line")
 118             self.head[buf[:p].strip().lower()] = buf[p + 1:].lstrip()
 119
 120         val, par = proto.pmimehead(self.head.get("content-disposition", ""))
 121         if val != "form-data":
 122             raise badmultipart("Unexpected Content-Disposition in form part: %r" % val)
 123         if not "name" in par:
 124             raise badmultipart("Missing name in form part")
 125         self.name = par["name"]
 126         self.filename = par.get("filename")
 127         val, par = proto.pmimehead(self.head.get("content-type", ""))
 128         self.ctype = val
 129         self.charset = par.get("charset")
 130         encoding = self.head.get("content-transfer-encoding", "binary")
 131         if encoding != "binary":
 132             raise badmultipart("Form part uses unexpected transfer encoding: %r" % encoding)
 133
 134 class multipart(object):
 135     def __init__(self, req):
 136         val, par = proto.pmimehead(req.ihead.get("Content-Type", ""))
 137         if req.method != "POST" or val != "multipart/form-data":
 138             raise badmultipart("Request is not a multipart form")
 139         if "boundary" not in par:
 140             raise badmultipart("Multipart form lacks boundary")
 141         self.boundary = par["boundary"]
 142         self.req = req
 143         self.buf = "\r\n"
 144         self.eof = False
 145         self.lastpart = formpart(self)
 146         self.lastpart.close()
 147
 148     def __iter__(self):
 149         return self
 150
 151     def next(self):
 152         if not self.lastpart.eof:
 153             raise RuntimeError("All form parts must be read entirely")
 154         if self.eof:
 155             raise StopIteration()
 156         self.lastpart = formpart(self)
 157         self.lastpart.parsehead()
 158         return self.lastpart
 159
 160 def formdata(req, onerror=Exception):
 161     data = req.item(formparse)
 162     if isinstance(data, Exception):
 163         if onerror is Exception:
 164             raise data
 165         return onerror
 166     return data