#include "StdAfx.h" using _com_util::CheckError; class CHTMLNode : public IHTMLNode { public: typedef CComPtr TComPtr; typedef CComPtr TDocumentPtr; protected: typedef CComPtr TElementCollectionPtr; public: CHTMLNode(const TComPtr& pElement, const TDocumentPtr& pDocument) : m_pElement(pElement), m_pDocument(pDocument) { } virtual THTMLNodePtr GetElementByID(const tstring& rsID)const { if (m_pDocument) { CComPtr pElement; if (SUCCEEDED(m_pDocument->getElementById(bstr_t(rsID.c_str()), &pElement)) && pElement) { TComPtr p(pElement); return THTMLNodePtr(new CHTMLNode(p, m_pDocument)); } } return THTMLNodePtr(); } virtual size_t GetChildCount()const { TElementCollectionPtr pColl = GetElementCollectionPtr(); if (pColl) { LONG celem = 0; HRESULT hr = pColl->get_length(&celem); if (S_OK == hr) return celem; } return 0; } virtual THTMLNodePtr GetChildPtr(size_t nIndex) { TElementCollectionPtr pColl = GetElementCollectionPtr(); if (pColl) { VARIANT varIndex; varIndex.vt = VT_UINT; varIndex.lVal = (LONG)nIndex; VARIANT var2; VariantInit(&var2); TComPtr pDisp; HRESULT hr = pColl->item(varIndex, var2, &pDisp); if (S_OK == hr && pDisp) return THTMLNodePtr(new CHTMLNode(pDisp, m_pDocument)); } return THTMLNodePtr(); } virtual bool Is(EType nType)const { switch (nType) { case Table: { CComPtr pTable; return (SUCCEEDED(m_pElement->QueryInterface(IID_IHTMLTable, reinterpret_cast(&pTable))) && (pTable)); } case TableRow: { CComPtr pRow; return (SUCCEEDED(m_pElement->QueryInterface(IID_IHTMLTableRow, reinterpret_cast(&pRow))) && (pRow)); } case TableColumn: { CComPtr pCol; return (SUCCEEDED(m_pElement->QueryInterface(IID_IHTMLTableCol, reinterpret_cast(&pCol))) && (pCol)); } } return false; } virtual tstring GetAttribute(const tstring& rsAttrName)const { tstring sAttr; CComPtr pElement; if (SUCCEEDED(m_pElement->QueryInterface(IID_IHTMLElement, reinterpret_cast(&pElement))) && pElement) { _variant_t vAttribute; BSTR pbstrAttrName = ::SysAllocString(rsAttrName.c_str()); if (SUCCEEDED(pElement->getAttribute(pbstrAttrName, 1, &vAttribute)) && VT_NULL != vAttribute.vt && VT_EMPTY != vAttribute.vt) { try { _bstr_t b(vAttribute); LPCTSTR psz = b; if (psz) sAttr = psz; } catch (_com_error&) { } } ::SysFreeString(pbstrAttrName); } return sAttr; } virtual tstring GetText()const { tstring sText; CComPtr pElement; if (SUCCEEDED(m_pElement->QueryInterface(IID_IHTMLElement, reinterpret_cast(&pElement))) && pElement) { BSTR bstrText; if (SUCCEEDED(pElement->get_innerText(&bstrText)) && bstrText) { try { sText = _bstr_t(bstrText); } catch (_com_error&) { } ::SysFreeString(bstrText); } } return sText; } protected: virtual TElementCollectionPtr GetElementCollectionPtr()const { TElementCollectionPtr pColl; HRESULT hr = m_pElement->QueryInterface(IID_IHTMLElementCollection, reinterpret_cast(&pColl)); if (FAILED(hr)) { CComPtr pElement; if (SUCCEEDED(m_pElement->QueryInterface(IID_IHTMLElement, reinterpret_cast(&pElement))) && pElement) { CComPtr pDisp; if (SUCCEEDED(pElement->get_children(&pDisp)) && pDisp) pDisp->QueryInterface(IID_IHTMLElementCollection, reinterpret_cast(&pColl)); } } return pColl; } private: TComPtr m_pElement; TDocumentPtr m_pDocument; }; CHTMLParserMS::CHTMLParserMS() : m_bCallUninit(false) { try { CheckError(::CoInitialize(NULL)); m_bCallUninit = true; _com_util::CheckError( ::CoCreateInstance(CLSID_HTMLDocument, NULL, CLSCTX_INPROC_SERVER, IID_IHTMLDocument2, (LPVOID*)&m_pDoc) ); CComPtr pPersist; _com_util::CheckError(m_pDoc->QueryInterface(IID_IPersistStreamInit, (LPVOID*)&pPersist)); _com_util::CheckError(pPersist->InitNew()); _com_util::CheckError(m_pDoc->QueryInterface(IID_IMarkupServices, (LPVOID*)&m_pMS)); if (m_pMS) { _com_util::CheckError(m_pMS->CreateMarkupPointer(&m_pMkStart)); _com_util::CheckError(m_pMS->CreateMarkupPointer(&m_pMkFinish)); } } catch (_com_error&/* e*/) { // show_com_error_msg(e); } } CHTMLParserMS::~CHTMLParserMS() { if (m_bCallUninit) ::CoUninitialize(); } CHTMLParserMS::THTMLNodePtr CHTMLParserMS::ParseString(const tstring& rsHTML) { mir_cslock lck(m_cs); CComPtr pMC; HRESULT hr = m_pMS->ParseString((OLECHAR*)rsHTML.c_str(), 0, &pMC, m_pMkStart, m_pMkFinish); if (SUCCEEDED(hr) && pMC) { CComPtr pNewDoc; hr = pMC->QueryInterface(IID_IHTMLDocument, (LPVOID*)&pNewDoc); if (SUCCEEDED(hr) && pNewDoc) { CComPtr pColl; pNewDoc->get_all(&pColl); CHTMLNode::TDocumentPtr pDoc; pMC->QueryInterface(IID_IHTMLDocument3, (LPVOID*)&pDoc); return THTMLNodePtr(new CHTMLNode(CHTMLNode::TComPtr(pColl), pDoc)); } } return THTMLNodePtr(); } bool CHTMLParserMS::IsInstalled() { bool bResult = true; bool bCallUninit = false; try { CheckError(::CoInitialize(NULL)); bCallUninit = true; CComPtr pDoc; _com_util::CheckError( ::CoCreateInstance(CLSID_HTMLDocument, NULL, CLSCTX_INPROC_SERVER, IID_IHTMLDocument2, reinterpret_cast(&pDoc)) ); } catch (_com_error&/* e*/) { bResult = false; } if (bCallUninit) ::CoUninitialize(); return bResult; } CHTMLEngineMS::CHTMLEngineMS() { } CHTMLEngineMS::~CHTMLEngineMS() { } CHTMLEngineMS::THTMLParserPtr CHTMLEngineMS::GetParserPtr()const { return THTMLParserPtr(new CHTMLParserMS); }