delphiXE7HttpEncode编码问题

原作者: [db:作者] 来自: [db:来源] 收藏邀请

近期在做网址编码相关的工作，发现在用 XE5 编译的时候，一切正常，拿到 XE7下就结果错误了。百度了下，谷歌了下，有人提出，但是，我没有找到答案，也许都没有碰到这个问题，也许都己经自己默默的解决了，在此小记一下，方便后人，也方便自己查寻。

例子：原字符 "过年"

httpencode('过年') 结果：

XE5为 %B9%FD%C4%EA

XE7 调用相当函数结果 %E8%BF%87%E5%B9%B4

百思不得其解啊，折腾了很长时间，后来终于想到是不是此函数官方更新修改了，（没办法，人比较笨）

于是查看源码：

XE5的 web.httpapp 中（分string 和 ansistring，我用ansistring 版本得到期望的结果）：

  1 function HTTPEncode(const AStr: string): string;
  2 // The NoConversion set contains characters as specificed in RFC 1738 and
  3 // should not be modified unless the standard changes.
  4 const
  5   NoConversion = [Ord('A')..Ord('Z'), Ord('a')..Ord('z'), Ord('*'), Ord('@'),
  6                   Ord('.'), Ord('_'), Ord('-'), Ord('0')..Ord('9'), Ord('$'),
  7                   Ord('!'), Ord(''''), Ord('('), Ord(')')];
  8 var
  9   Sp, Rp: PChar;
 10 begin
 11   SetLength(Result, Length(AStr) * 3);
 12   Sp := PChar(AStr);
 13   Rp := PChar(Result);
 14   while Sp^ <> #0 do
 15   begin
 16     if Ord(Sp^) in NoConversion then
 17       Rp^ := Sp^
 18     else
 19       if Sp^ = ' ' then
 20         Rp^ := '+'
 21       else
 22       begin
 23         FormatBuf(Rp, 3, string('%%%.2x'), 6, [Ord(Sp^)]);
 24         Inc(Rp,2);
 25       end;
 26     Inc(Rp);
 27     Inc(Sp);
 28   end;
 29   SetLength(Result, Rp - PChar(Result));
 30 end;
 31 
 32 function HTTPDecode(const AStr: string): string;
 33 var
 34   Sp, Rp, Cp: PChar;
 35   S: string;
 36 begin
 37   SetLength(Result, Length(AStr));
 38   Sp := PChar(AStr);
 39   Rp := PChar(Result);
 40   Cp := Sp;
 41   try
 42     while Sp^ <> #0 do
 43     begin
 44       case Sp^ of
 45         '+': Rp^ := ' ';
 46         '%': begin
 47                // Look for an escaped % (%%) or %<hex> encoded character
 48                Inc(Sp);
 49                if Sp^ = '%' then
 50                  Rp^ := '%'
 51                else
 52                begin
 53                  Cp := Sp;
 54                  Inc(Sp);
 55                  if (Cp^ <> #0) and (Sp^ <> #0) then
 56                  begin
 57                    S := Char('$') + Cp^ + Sp^;
 58                    Rp^ := Char(StrToInt(string(S)));
 59                  end
 60                  else
 61                    raise EWebBrokerException.CreateFmt(sErrorDecodingURLText, [Cp - PChar(AStr)]);
 62                end;
 63              end;
 64       else
 65         Rp^ := Sp^;
 66       end;
 67       Inc(Rp);
 68       Inc(Sp);
 69     end;
 70   except
 71     on E:EConvertError do
 72       raise EConvertError.CreateFmt(sInvalidURLEncodedChar,
 73         [Char('%') + Cp^ + Sp^, Cp - PChar(AStr)])
 74   end;
 75   SetLength(Result, Rp - PChar(Result));
 76 end;
 77 
 78 function HTTPEncode(const AStr: AnsiString): AnsiString;
 79 // The NoConversion set contains characters as specificed in RFC 1738 and
 80 // should not be modified unless the standard changes.
 81 const
 82   NoConversion = ['A'..'Z','a'..'z','*','@','.','_','-',
 83                   '0'..'9','$','!','''','(',')'];
 84 var
 85   Sp, Rp: PAnsiChar;
 86 begin
 87   SetLength(Result, Length(AStr) * 3);
 88   Sp := PAnsiChar(AStr);
 89   Rp := PAnsiChar(Result);
 90   while Sp^ <> #0 do
 91   begin
 92     if Sp^ in NoConversion then
 93       Rp^ := Sp^
 94     else
 95       if Sp^ = ' ' then
 96         Rp^ := '+'
 97       else
 98       begin
 99         System.AnsiStrings.FormatBuf(Rp^, 3, AnsiString('%%%.2x'), 6, [Ord(Sp^)]);
100         Inc(Rp,2);
101       end;
102     Inc(Rp);
103     Inc(Sp);
104   end;
105   SetLength(Result, Rp - PAnsiChar(Result));
106 end;

View Code

在XE7中

web.httpapp:

function HTTPEncode(const AStr: string): string;
begin
  Result := TNetEncoding.URL.Encode(AStr);
end;

查看 system.netencoding

找到

function TNetEncoding.DoEncode(const Input: array of Byte): TBytes;
begin
  Result := TEncoding.UTF8.GetBytes(DoEncode(TEncoding.UTF8.GetString(@Input[0])));
end;

查看类定义：  
TURLEncoding = class(TNetEncoding)
  protected
    function DoDecode(const Input: string): string; overload; override;
    function DoEncode(const Input: string): string; overload; override;
  end;

查看函数代码：

function TURLEncoding.DoEncode(const Input: string): string;
// The NoConversion set contains characters as specificed in RFC 1738 and
// should not be modified unless the standard changes.
const
  NoConversion = [Ord('A')..Ord('Z'), Ord('a')..Ord('z'), Ord('*'), Ord('@'),
                  Ord('.'), Ord('_'), Ord('-'), Ord('0')..Ord('9'), Ord('$'),
                  Ord('!'), Ord(''''), Ord('('), Ord(')')];

  procedure AppendByte(B: Byte; var Buffer: PChar);
  const
    Hex = '0123456789ABCDEF';
  begin
    Buffer[0] := '%';
    Buffer[1] := Hex[B shr 4 + Low(string)];
    Buffer[2] := Hex[B and $F + Low(string)];
    Inc(Buffer, 3);
  end;

var
  Sp, Rp: PChar;
  MultibyteChar: TBytes;
  I, ByteCount: Integer;
begin
  // Characters that require more than 1 byte are translated as "percent-encoded byte"
  // which will be encoded with 3 chars per byte -> %XX
  // Example: ?character
  //   Multibyte representation: C391 (2 bytes)
  //   URL encode representation: %C3%91
  //
  // So the worst case is 4 bytes(max) per Char, and 3 characters to represent each byte
  SetLength(Result, Length(Input) * 4 * 3);
  Sp := PChar(Input);
  Rp := PChar(Result);
  SetLength(MultibyteChar, 4);
  while Sp^ <> #0 do
  begin
    if Ord(Sp^) in NoConversion then
    begin
      Rp^ := Sp^;
      Inc(Rp)
    end
    else if Sp^ = ' ' then
    begin
      Rp^ := '+';
      Inc(Rp)
    end
    else
    begin
      if (Ord(Sp^) < 128) then
        // Single byte char
        AppendByte(Ord(Sp^), Rp)
      else
      begin
        // Multi byte char
        ByteCount := TEncoding.UTF8.GetBytes([Sp^], 0, 1, MultibyteChar, 0);
        for I := 0 to ByteCount - 1 do
          AppendByte(MultibyteChar[I], Rp);
      end
    end;
    Inc(Sp);
  end;
  SetLength(Result, Rp - PChar(Result));
end;

View Code

似乎有点不同。

目前是我自己建立一个函数，复制XE5版本的代码放在XE7里面调用，得到希望的结果的。

代码：

function MyHTTPEncode(const AStr: AnsiString): AnsiString;
// The NoConversion set contains characters as specificed in RFC 1738 and
// should not be modified unless the standard changes.
const
  NoConversion = ['A'..'Z','a'..'z','*','@','.','_','-',
                  '0'..'9','$','!','''','(',')'];
var
  Sp, Rp: PAnsiChar;
begin
  SetLength(Result, Length(AStr) * 3);
  Sp := PAnsiChar(AStr);
  Rp := PAnsiChar(Result);
  while Sp^ <> #0 do
  begin
    if Sp^ in NoConversion then
      Rp^ := Sp^
    else
      if Sp^ = ' ' then
        Rp^ := '+'
      else
      begin
        System.AnsiStrings.FormatBuf(Rp^, 3, AnsiString('%%%.2x'), 6, [Ord(Sp^)]);
        Inc(Rp,2);
      end;
    Inc(Rp);
    Inc(Sp);
  end;
  SetLength(Result, Rp - PAnsiChar(Result));
end;