libgo/go/mime/multipart/quotedprintable.go

   1 // Copyright 2012 The Go Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style
   3 // license that can be found in the LICENSE file.
   4
   5 // The file define a quoted-printable decoder, as specified in RFC 2045.
   6 // Deviations:
   7 // 1. in addition to "=\r\n", "=\n" is also treated as soft line break.
   8 // 2. it will pass through a '\r' or '\n' not preceded by '=', consistent
   9 //    with other broken QP encoders & decoders.
  10
  11 package multipart
  12
  13 import (
  14         "bufio"
  15         "bytes"
  16         "fmt"
  17         "io"
  18 )
  19
  20 type qpReader struct {
  21         br   *bufio.Reader
  22         rerr error  // last read error
  23         line []byte // to be consumed before more of br
  24 }
  25
  26 func newQuotedPrintableReader(r io.Reader) io.Reader {
  27         return &qpReader{
  28                 br: bufio.NewReader(r),
  29         }
  30 }
  31
  32 func fromHex(b byte) (byte, error) {
  33         switch {
  34         case b >= '0' && b <= '9':
  35                 return b - '0', nil
  36         case b >= 'A' && b <= 'F':
  37                 return b - 'A' + 10, nil
  38         }
  39         return 0, fmt.Errorf("multipart: invalid quoted-printable hex byte 0x%02x", b)
  40 }
  41
  42 func (q *qpReader) readHexByte(v []byte) (b byte, err error) {
  43         if len(v) < 2 {
  44                 return 0, io.ErrUnexpectedEOF
  45         }
  46         var hb, lb byte
  47         if hb, err = fromHex(v[0]); err != nil {
  48                 return 0, err
  49         }
  50         if lb, err = fromHex(v[1]); err != nil {
  51                 return 0, err
  52         }
  53         return hb<<4 | lb, nil
  54 }
  55
  56 func isQPDiscardWhitespace(r rune) bool {
  57         switch r {
  58         case '\n', '\r', ' ', '\t':
  59                 return true
  60         }
  61         return false
  62 }
  63
  64 var (
  65         crlf       = []byte("\r\n")
  66         lf         = []byte("\n")
  67         softSuffix = []byte("=")
  68 )
  69
  70 func (q *qpReader) Read(p []byte) (n int, err error) {
  71         for len(p) > 0 {
  72                 if len(q.line) == 0 {
  73                         if q.rerr != nil {
  74                                 return n, q.rerr
  75                         }
  76                         q.line, q.rerr = q.br.ReadSlice('\n')
  77
  78                         // Does the line end in CRLF instead of just LF?
  79                         hasLF := bytes.HasSuffix(q.line, lf)
  80                         hasCR := bytes.HasSuffix(q.line, crlf)
  81                         wholeLine := q.line
  82                         q.line = bytes.TrimRightFunc(wholeLine, isQPDiscardWhitespace)
  83                         if bytes.HasSuffix(q.line, softSuffix) {
  84                                 rightStripped := wholeLine[len(q.line):]
  85                                 q.line = q.line[:len(q.line)-1]
  86                                 if !bytes.HasPrefix(rightStripped, lf) && !bytes.HasPrefix(rightStripped, crlf) {
  87                                         q.rerr = fmt.Errorf("multipart: invalid bytes after =: %q", rightStripped)
  88                                 }
  89                         } else if hasLF {
  90                                 if hasCR {
  91                                         q.line = append(q.line, '\r', '\n')
  92                                 } else {
  93                                         q.line = append(q.line, '\n')
  94                                 }
  95                         }
  96                         continue
  97                 }
  98                 b := q.line[0]
  99
 100                 switch {
 101                 case b == '=':
 102                         b, err = q.readHexByte(q.line[1:])
 103                         if err != nil {
 104                                 return n, err
 105                         }
 106                         q.line = q.line[2:] // 2 of the 3; other 1 is done below
 107                 case b == '\t' || b == '\r' || b == '\n':
 108                         break
 109                 case b < ' ' || b > '~':
 110                         return n, fmt.Errorf("multipart: invalid unescaped byte 0x%02x in quoted-printable body", b)
 111                 }
 112                 p[0] = b
 113                 p = p[1:]
 114                 q.line = q.line[1:]
 115                 n++
 116         }
 117         return n, nil
 118 }