如何将TCppWeb中的网页存为文本（用代码）？？

[ramile]

如何将TCppWeb中的网页存为文本（用代码）？？

[freshman2003]

本文来自FAQ。
把Get（）放在 CppWebBrowser1的OnDocumentComplete()中。
bool __fastcall TForm1::Get(void)
{
IHTMLDocument2 *pDoc=NULL;
IHTMLElement *body;
LPDISPATCH pDisp = NULL;
pDisp =CppWebBrowser1->Document;
if(pDisp)
{
if (SUCCEEDED(pDisp->QueryInterface(IID_IHTMLDocument2, (void**)&pDoc)))
{
pDisp->Release();
wchar_t *txt;
pDoc->get_body(&body);
body->get_outerText(&txt);
pDoc->Release();
body->Release();
Memo1->Text=txt;
return true;
}
}
return false;
}

[Friecin]

郁于IE控件的限制，有生成临时文件...
void __fastcall TfrmMain::Button1Click(TObject *Sender)
{
IHTMLDocument2* HTMLDocument = NULL;
IPersistFile* PersistFile = NULL;
if(!CppWebBrowser->Busy && CppWebBrowser->Document &&
AnsiString(CppWebBrowser->LocationURL) != "about:blank" &&
SUCCEEDED(CppWebBrowser->Document->QueryInterface(
IID_IHTMLDocument2, (LPVOID*)&HTMLDocument)))
{
// 保存网页文档
if(SUCCEEDED(HTMLDocument->QueryInterface(IID_IPersistFile,
(LPVOID*)&PersistFile)))
{

AnsiString TempFile = ExtractFilePath(ParamStr(0))+
"0000html_temp";
PersistFile->Save(WideString(TempFile), false);
PersistFile->Release();

TMemoryStream* pStream = new TMemoryStream;
if(pStream)
{
// 将所有的临时文件内容读入TStrings，并显示
if(FileExists(TempFile))
{
pStream->LoadFromFile(TempFile);
// DataModule->WriteBlob(pStream);
DeleteFile(TempFile);
}
}
delete pStream;
}
}
}
//---------------------------------------------------------------------------
void __fastcall DataModule::WriteBlob(const TMemoryStream* pStream)
{
pQuery->Edit();
TBlobField * pField=(TBlobField *)pQuery->FieldByName("HTML_FIELD");
TBlobStream * pmem=new TBlobStream(pField,bmWrite);
pmem->Seek(0,soFromBeginning);
pStream->SaveToStream(pmem);
delete pmem;
pQuery->Post();
}
上面的代码写得真得很好

方法二:
方法2：
不生成临时文件，但由于CppWebBrowser的一个Bug，const int Size = 0x100000;只能预先指定，虽然也可以透过IHTMLDocument来取得大致的大小，呵呵太麻烦了....
10000不够用自己放大呗....
//---------------------------------------------------------------------------
void GetHtmlSourceFromWebBrowser( TCppWebBrowser *WebBrowser,TMemoryStream* pMemoryStream )
{
IStream *pIStream;
HGLOBAL hGlobal;
IPersistStreamInit *psi;

const int Size = 0x100000;

if( WebBrowser->Document == NULL )
return;

hGlobal = GlobalAlloc( GPTR, Size );
if ( 0 == hGlobal )
{
ShowMessage( "GlobalAlloc Error" );
return;
}

OleCheck( CreateStreamOnHGlobal( hGlobal, true, &pIStream ) );
try
{
OleCheck( WebBrowser->Document->QueryInterface( __uuidof(IPersistStreamInit), (void **)&psi ) );
try
{
OleCheck( psi->Save(pIStream,false) );

pMemoryStream->SetSize(Size);
CopyMemory( pMemoryStream->Memory, hGlobal, Size);

}
__finally
{
delete psi;
}
}
__finally
{
delete pIStream;
}
}
//---------------------------------------------------------------------------
void __fastcall TForm1::Button1Click(TObject *Sender)
{

TMemoryStream* pStream = new TMemoryStream();

GetHtmlSourceFromWebBrowser(CppWebBrowser1, pStream);

DataModule->WriteBlob(pStream);

if(pStream) delete pStream;
}
//---------------------------------------------------------------------------
他开辟了很大的内存,用于保存流，下面是msdn上的IPersistStreamInit定义
interface IPersistStream : IPersist {
//IUnknown members and GetClassID from IPersist
HRESULT IsDirty(void);
HRESULT Load(IStream *pStm);
HRESULT Save(IStream *pStm, BOOL fClearDirty);
HRESULT GetSizeMax(ULARGE_INTEGER *pcbSize); };
interface IPersistStreamInit : IPersist { HRESULT InitNew(void); };
这里涌到的save如下
Instructs the object to save its persistent data at the current seek offset of the given stream. The object must save its data and return without holding the IStream pointer. The fClearDirty argument tells the object whether it should consider itself clean after this call-a client may be making a copy of the object as opposed to saving its data permanently, and this flag distinguishes the cases.


方法三：

void __fastcall TForm1::CppWebBrowser1DocumentComplete(TObject *Sender,
LPDISPATCH pDisp, Variant *URL)
{
//为了保证同步，应该把获取Document内容的代码放到OnDocumentComplete事件
IHTMLDocument2 *document;
IHTMLElement *body;
BSTR title,source,content;

(void *)document=CppWebBrowser1->Document;

document->get_title(&title);
document->get_body(&body);
body->get_outerHTML(&source);
body->get_outerText(&content);

Edit1->Text=title;
Edit2->Text=*URL;
Memo1->Text=source;
Memo2->Text=content;

TMemoryStream* pms = new TMemoryStream();
TStringStream* pss = new TStringStream(AnsiString(source));
pms->CopyFrom(pss,pss->Size);

// WriteToDB(pms);

delete pms,pss;

}