Hi All,
The following code fails if I do not create the IPersistStreamInit
object (and call InitNew()). Loosely speaking, these are steps 3 and 4.
The code fails at pDocument->ParseString(...) because pContainer is
NULL.
My question is, why must I create an intialize a stream to parse an
in-memory document?
Eventually, I'm going to use Richard Lopes's code
(http://groups.google.com/group/microsoft.public.inetsdk.programming.webbrowser_ctl/tree/browse_frm/thread/96ea9f563945a5bb/165efdab2bab6538).
But I'm trying to learn my way around the parser now...
Jeff
//
// Code basis was presented by Asher Kobin
// in 'Lightweight HTML Parsing Using MSHTML'
//
// http://www.codeguru.com/Cpp/I-N/ieprogram/article.php/c4385
//
#define _WIN32_IE 0x0500 // IMarkupServices
#define _WIN32_DCOM // CoInitializeEx(...)
#pragma warning( disable: 4192 )
#pragma warning( disable: 4278 )
#import "mshtml.tlb" // MSHTML Namespace and Guids
#pragma warning( default: 4192 )
#pragma warning( default: 4278 )
#include <windows.h>
#include <atlbase.h> // CComPtr<>, CoInitializeEx, etc
#include <mshtml.h> // MSHTML Object and Interfaces
#include <assert.h>
// Standard C++ Library
#include <iostream>
VOID DumpError( UINT nWhere );
HRESULT hr = S_OK;
int main(int argc, char* argv[])
{
MSHTML::IHTMLDocument2* pDocument = NULL;
//IPersistStreamInit* pStream = NULL;
MSHTML::IMarkupServices* pMarkupServices = NULL;
MSHTML::IMarkupContainer* pContainer = NULL;
MSHTML::IMarkupPointer* pBegin = NULL;
MSHTML::IMarkupPointer* pEnd = NULL;
MSHTML::IHTMLDocument2* pNewDocument = NULL;
CComBSTR Page = OLESTR( "<HTML><HEAD><TITLE>Hello World</TITLE>"
);
Page += OLESTR( "</HEAD><BODY><P>Hello
World</P></<BODY></HTML>" );
try
{
///////////////////////////////////////////////////////////////////
hr = CoInitializeEx( NULL, COINIT_APARTMENTTHREADED );
if( FAILED( hr ) ) { throw 1; }
///////////////////////////////////////////////////////////////////
// Create a Document to bootstrap the process
CoCreateInstance(CLSID_HTMLDocument, NULL,
CLSCTX_INPROC_SERVER, IID_IHTMLDocument2,
reinterpret_cast<PVOID*>( &pDocument) );
if( NULL == pDocument ) { throw 2; }
///////////////////////////////////////////////////////////////////
//hr = pDocument->QueryInterface( IID_IPersistStreamInit,
// reinterpret_cast<PVOID*>(
&pStream ) );
//if( FAILED( hr ) ) { throw 3; }
///////////////////////////////////////////////////////////////////
//hr = pStream->InitNew();
//if( FAILED( hr ) ) { throw 4; }
///////////////////////////////////////////////////////////////////
hr = pDocument->QueryInterface( IID_IMarkupServices,
reinterpret_cast<PVOID*>(
&pMarkupServices ) ) ;
if( FAILED( hr ) ) { throw 5; }
///////////////////////////////////////////////////////////////////
hr = pMarkupServices->CreateMarkupPointer( &pBegin );
if( FAILED( hr ) ) { throw 6; }
///////////////////////////////////////////////////////////////////
hr = pMarkupServices->CreateMarkupPointer( &pEnd );
if( FAILED( hr ) ) { throw 7; }
///////////////////////////////////////////////////////////////////
pMarkupServices->ParseString( Page, 0, &pContainer, pBegin,
pEnd );
if( FAILED( hr ) ) { throw 8; }
///////////////////////////////////////////////////////////////////
pContainer->QueryInterface( IID_IHTMLDocument,
(PVOID*)&pNewDocument );
if( FAILED( hr ) ) { throw 9; }
///////////////////////////////////////////////////////////////////
///////////////////////////////////////////////////////////////////
MSHTML::IHTMLElement* pElement = NULL;
pNewDocument->get_body( &pElement );
if( NULL != pElement )
{
CComBSTR Body;
hr = pElement->get_innerText( &Body );
if( S_OK == hr )
{ std::wcout << (wchar_t*) Body << std::endl; }
pElement->Release();
}
///////////////////////////////////////////////////////////////////
///////////////////////////////////////////////////////////////////
}
catch( UINT e )
{
DumpError( e );
}
if( NULL != pDocument ) { pDocument->Release(); }
// if( NULL != pStream ) { pStream->Release(); }
if( NULL != pMarkupServices ) { pMarkupServices->Release(); }
if( NULL != pContainer ) { pContainer->Release(); }
if( NULL != pBegin ) { pBegin->Release(); }
if( NULL != pEnd ) { pEnd->Release(); }
if( NULL != pNewDocument ) { pNewDocument->Release(); }
CoUninitialize( );
return HRESULT_CODE( hr );
}
VOID DumpError( UINT nWhere )
{
CComBSTR error;
switch( nWhere )
{
case 1:
{
error = L"CoInitializeEx";
break;
}
case 2:
{
error = L"pDocument = NULL";
break;
}
case 3:
{
error = L"pDocument->QueryInterface( IPersistStreamInit )";
break;
}
case 4:
{
error = L"pStream->InitNew()";
break;
}
case 5:
{
error = L"pDocument->QueryInterface( IMarkupServices )";
break;
}
case 6:
{
error = L"pMarkupServices->CreateMarkupPointer( pBegin )";
break;
}
case 7:
{
error = L"pMarkupServices->CreateMarkupPointer( pEnd )";
break;
}
case 8:
{
error = L"pMarkupServices->ParseString()";
break;
}
case 9:
{
error = L"pContainer->QueryInterface( NewDocument )";
break;
}
default:
{
error = L"Unknown";
}
}
std::wcout << std::endl;
std::wcout << L"COM Error: " << std::endl << L" ";
std::wcout << (wchar_t*)error << std::endl ;
std::wcout << std::hex << L" 0x" <<hr;
std::wcout << L" (Win32: " << std::dec << HRESULT_CODE( hr );
std::wcout << L")" << std::endl;
}