peterjazenga / simdesign Goto Github PK

Failed to open an xml file, if it contains such a node `<language  
ui="english"/>`

At NativeXml.pas line 4601

          WhiteSpaceNode := TsdWhiteSpace.Create(TNativeXml(FOwner));
          NodeAdd(WhiteSpaceNode);
          inc(FDirectNodeCount);
          WhiteSpaceNode.SetCoreValue(Blanks);

I changed it to:

          WhiteSpaceNode := TsdWhiteSpace.Create(TNativeXml(FOwner));
          FDirectNodeCount := NodeAdd(WhiteSpaceNode);
          //inc(FDirectNodeCount);
          WhiteSpaceNode.SetCoreValue(Blanks);

Please confirm.

See test in rael_bauer

What steps will reproduce the problem?
1.
2.
3.

What is the expected output? What do you see instead?


Please use labels and text to provide additional information.

What steps will reproduce the problem?
1. When create an XML with a large node value, 32KB or greater, function 
TsdBufferWriter.Write(const Buffer; Count: Integer): Longint;
 is failing as TByteArray has a 32KB limit.

What is the expected output? What do you see instead?


What version of the product are you using? On what operating system?
Trunk r71, Windows 7 Professional, Delphi XE2

Please provide any additional information below.

As a quick work around I have added the following to my local NativeXML.pas
type
  TBigByteArray = array[0..2147483646] of byte; // 2GB - 2, This should never be instantiated, it's used as TByteArray indexing is limited to <32KB

and renamed all instances of TByteArray to TBigByteArray

function TsdXmlParser.ReadStringUntilChar(AChar: AnsiChar): Utf8String;
var
  Count: integer;
  StartIdx: integer;
  CurrentChar: Char;//<<<<--This is problem. I am change to AnsiChar
begin
 Count := MakeDataAvailable;

  StartIdx := FUtf8CurrentIdx;
  while not FEndOfStream do
  begin

    CurrentChar := FUtf8Buffer[FUtf8CurrentIdx];//<<<<-- no [DCC Error] NativeXml.pas(6998): E2010 Incompatible types: 'Char' and 'AnsiChar'

Hello,

Bit of background. If I have some misbehaving system that do not like some 
characters (in other words badly made). Like Nordic ones äöå.

Now I can't (Or I could not find) write string that I have already handled with 
some escaping routine. because it becomes something like this : ä -> ä -> 
ä

Or other way would be (But would slow things down) to have check at the 
sdEscapeString part which handles the  '&': would check is that part of escaped 
string. I think this would be so much slower to do.

But having Setter in XML node that would bypass the sdEscapeString and will not 
do any Base64 encoding and so on. then user has to make sure that string is 
correctly escaped. Also it could speed thigs little if you are inserting purely 
numeric values and so, when you can be sure that there is no need for escaping.

-Tee-

What steps will reproduce the problem?
1. If I compile any of the projects (SvgTest or SvgViewer)
2. I try to open any svg file (I downloaded also some that are in the svn as 
"Anarchist_logo.svg")
3. When I try to load the svg file I get "List index out of bound" error.

I'm using Delphi XE and I debugged the source and the error occures in the 
TsdContainerNode.NodeInsert procedure. The index variable passed is 2 while the 
"FNodes" ha no element inside it.


I hope this is enough. I will be glad to make more debug if this can help.

Thanks.
Ibrahim

it is no way to directly assign escaped text to XML attribute, so there is no 
way to control escape of special characters like $0D,$0A etc. for compatibility 
with MSXML parser-based applications

Please add define for XE6

I'd suggest use CompilerVersion to simplify simdesign.inc

{$if CompilerVersion >= 15}
  {$define D7UP}
{$endif}

{$if CompilerVersion >= 18}
  {$define D10UP} // Delphi 2006
{$endif}

{$if CompilerVersion >= 23}
  {$define D15UP} // Delphi XE2
{$endif}

New version NativeXML use very many memory. 

File for test - 

https://drive.google.com/file/d/0B7XT46w-6F3RUkNSQ0R4enBLTzg/view?usp=sharing

EOL normalization in sdWideToUTF8Buffer does not work as expected

if, for example, attribute value contains escaped CRLF ($0D$0A) after call to 
this function it results in $0A$0A, so escaped multiline text does not appear 
as expected

It is no way to skip or select desired EOL normalization style here

      AddChar := True;
      ByteCh := byte(W);

      if ByteCh = $0D then
      begin
// I have to comment this line to skip EOL normalization
//!!!////        ByteCh := $0A;
        LastChar0D := True;
      end else
      begin
        if (ByteCh = $0A) and LastChar0D then
          AddChar := False;
        LastChar0D := False;
      end;

If I check existing "ABC" attribute by "abc" string then method HasAttribute() 
return False.
But method GetAttributeByName() return proper value. 

Problem is with compare way, in HasAttribute is:

  if AttributeName[i] = AName then

but in GetAttributeByName() is:

  if Utf8CompareText(A.Name, AName) = 0 then

Please update HasAttribute to use Utf8CompareText function()

It should be removed.

simdesign.inc modified

Load attached sample.xml and try to read value of Lorem node.

LDoc.LoadFromFile('sample.xml');
LValue:= LDoc.Root.ReadUnicodeString('Lorem'); 

The LValue should contain lorem text but instead it contains only 3 characters 
(#$D#$A#9).