vc 读取wps文件的文本

// ControlWPS.cpp: implementation of the CControlWPS class. // ////////////////////////////////////////////////////////////////////// #include "ControlWPS.h" ////////////////////////////////////////////////////////////////////// // Construction/Destruction ////////////////////////////////////////////////////////////////////// CControlWPS::CControlWPS() : m_pIDisp(NULL) { CoInitialize(NULL); ::VariantInit(&varWorkbook); ::VariantInit(&varSheets); ::VariantInit(&varActiveSheet); } CControlWPS::~CControlWPS() { CloseWPS(); CoUninitialize(); } BOOL CControlWPS::OpenWPSWord( CString File ) { CLSID clsid; // 获取WPS 文档的 CLSID HRESULT hr = CLSIDFromProgID( L"KWPS.Application", &clsid ); if ( FAILED( hr ) ) return FALSE; // 通过CLSID启动WPS hr = CoCreateInstance( clsid, NULL, CLSCTX_LOCAL_SERVER, IID_IDispatch, ( void** )&m_pIDisp ); if ( FAILED( hr ) ) return FALSE; VARIANT varWorkbooks = { 0 }; hr = GetProperty( m_pIDisp, L"Documents", &varWorkbooks ); CComVariant varDocument[ 4 ]; varDocument[ 1 ].vt = VT_BOOL; varDocument[ 1 ].boolVal = VARIANT_FALSE; varDocument[ 3 ].vt = VT_BSTR; varDocument[ 3 ].bstrVal = File.AllocSysString(); hr = InvokeN( varWorkbooks.pdispVal, L"Open", varDocument, 4, &varWorkbook ); ::SysFreeString( varDocument[ 3 ].bstrVal ); if ( FAILED( hr ) ) return FALSE; return TRUE; } BOOL CControlWPS::getWPSWordText( CString& strText ) { HRESULT hr; VARIANT varRange = { 0 }; hr = GetProperty( varWorkbook.pdispVal, L"Range", &varRange ); if ( FAILED( hr ) ) return FALSE; VARIANT var = { 0 }; hr = GetProperty( varRange.pdispVal, L"Text", &var ); if ( FAILED( hr ) ) return FALSE; char* lpszText = _com_util::ConvertBSTRToString( var.bstrVal ); CString cstem( lpszText ); strText = cstem; delete lpszText; return TRUE; } BOOL CControlWPS::OpenWPSPPT( CString File ) { CLSID clsid; // 获取WPS 文档的 CLSID HRESULT hr = CLSIDFromProgID( L"KWPP.Application", &clsid ); if ( FAILED( hr ) ) return FALSE; // 通过CLSID启动WPS hr = CoCreateInstance( clsid, NULL, CLSCTX_LOCAL_SERVER, IID_IDispatch, ( void** )&m_pIDisp ); if ( FAILED( hr ) ) return FALSE; // 使WPS可见 // VARIANT var = {0}; // var.vt = VT_BOOL; // var.boolVal = VARIANT_TRUE; // PutProperty(m_pIDisp, L"Visible", &var); VARIANT varWorkbooks = { 0 }; hr = GetProperty( m_pIDisp, L"Presentations", &varWorkbooks ); CComVariant varDocument[ 4 ]; varDocument[ 1 ].vt = VT_BOOL; varDocument[ 1 ].boolVal = VARIANT_FALSE; varDocument[ 3 ].vt = VT_BSTR; varDocument[ 3 ].bstrVal = File.AllocSysString(); hr = InvokeN( varWorkbooks.pdispVal, L"Open", varDocument, 4, &varWorkbook ); ::SysFreeString( varDocument[ 3 ].bstrVal ); if ( FAILED( hr ) ) return FALSE; } BOOL CControlWPS::getWPSPPTText( CString& strText ) { HRESULT hr; VARIANT varCells = { 0 }; CComVariant varSlides = { 0 }; hr = GetProperty( varWorkbook.pdispVal, L"Slides", &varSlides ); if ( FAILED( hr ) ) return FALSE; VARIANT varCount = { 0 }; hr = GetProperty( varSlides.pdispVal, L"Count", &varCount ); if ( FAILED( hr ) ) return FALSE; long count_slides = varCount.lVal; for ( long i = 1; i <= varCount.lVal; i++ ) { // 获取Worksheet对象 VARIANT varWorksheet = { 0 }; CComVariant varItem[ 1 ]; varItem[ 0 ].vt = VT_I4; varItem[ 0 ].lVal = i; hr = GetPropertyN( varSlides.pdispVal, L"Item", varItem, 1, &varWorksheet ); if ( FAILED( hr ) ) return FALSE; VARIANT varShapes = { 0 }; hr = GetProperty( varWorksheet.pdispVal, L"Shapes", &varShapes ); if ( FAILED( hr ) ) return FALSE; VARIANT varcount_shapes = { 0 }; hr = GetProperty( varShapes.pdispVal, L"Count", &varcount_shapes ); int count_shapes = varcount_shapes.lVal; //提取每张幻灯片中的shapes,并从shape中获得文本放到result.txt中 for ( int j = 1; j < ( count_shapes + 1 ); j++ ) { //VARIANT varshape[ 1 ]; CComVariant varshapItem[ 1 ]; varshapItem[ 0 ].vt = VT_I4; varshapItem[ 0 ].lVal = j; VARIANT varshape = { 0 }; hr = GetPropertyN( varShapes.pdispVal, L"Item", varshapItem, 1, &varshape ); VARIANT varHasTextFrame = { 0 }; hr = GetProperty( varshape.pdispVal, L"HasTextFrame", &varHasTextFrame ); if ( varHasTextFrame.bVal ) { VARIANT varTextFrame = { 0 }; hr = GetProperty( varshape.pdispVal, L"TextFrame", &varTextFrame ); VARIANT varTextRange = { 0 }; hr = GetProperty( varTextFrame.pdispVal, L"TextRange", &varTextRange ); VARIANT varText = { 0 }; hr = GetProperty( varTextRange.pdispVal, L"Text", &varText ); char* lpszText = _com_util::ConvertBSTRToString( varText.bstrVal ); CString cstem( lpszText ); strText += cstem; delete lpszText; if ( FAILED( hr ) ) return FALSE; } } } } // 启动WPS表格 BOOL CControlWPS::OpenWPSExcel(CString File) { CLSID clsid ; // 获取WPS表格的 CLSID HRESULT hr = CLSIDFromProgID( L"KET.Application", &clsid ); if(FAILED(hr)) return FALSE; // 通过CLSID启动WPS hr = CoCreateInstance( clsid, NULL, CLSCTX_LOCAL_SERVER, IID_IDispatch, ( void** )&m_pIDisp ); if(FAILED(hr)) return FALSE; // 使WPS可见 // VARIANT var = {0}; // var.vt = VT_BOOL; // var.boolVal = VARIANT_TRUE; // PutProperty(m_pIDisp, L"Visible", &var); // 获取Workbooks集合 VARIANT varWorkbooks = {0}; hr = GetProperty(m_pIDisp, L"Workbooks", &varWorkbooks); if(FAILED(hr)) return FALSE; // 获取工作表 CComVariant varDocument[4]; varDocument[1].vt = VT_BOOL; varDocument[1].boolVal = VARIANT_FALSE; varDocument[3].vt = VT_BSTR; varDocument[3].bstrVal = File.AllocSysString(); hr = InvokeN(varWorkbooks.pdispVal, L"Open", varDocument, 4, &varWorkbook); ::SysFreeString(varDocument[3].bstrVal); if(FAILED(hr)) return FALSE; // 获取Sheets集合 hr = GetProperty(m_pIDisp, L"Sheets", &varSheets); if(FAILED(hr)) return FALSE; // 获取Count集合 VARIANT varCount = {0}; hr = GetProperty(varSheets.pdispVal, L"Count", &varCount); if(FAILED(hr)) return FALSE; for (long i=1;i<=varCount.lVal;i++) { // 获取Worksheet对象 VARIANT varWorksheet = {0}; CComVariant varItem[1]; varItem[0].vt = VT_I4; varItem[0].lVal = i; hr = GetPropertyN(varSheets.pdispVal, L"Item", varItem,1,&varWorksheet); if(FAILED(hr)) return FALSE; // 获取Name对象 VARIANT varName = {0}; hr = GetProperty(varWorksheet.pdispVal, L"Name", &varName); if(FAILED(hr)) return FALSE; char* lpszText = _com_util::ConvertBSTRToString(varName.bstrVal); CString SheetName( lpszText ); m_SheetNames.Add(SheetName); delete lpszText; } return TRUE; } BOOL CControlWPS::getWPSExcelText( CString& strText ) { CStringArray m_SheetNames; GetSheetNames( m_SheetNames ); for ( int nint = 0; nint < m_SheetNames.GetSize(); nint++ ) { CString cs = m_SheetNames.GetAt( nint ); OpenSheet( cs ); __int64 rows; __int64 column; getRowsAndColumn( rows, column ); int j = 0, i = 0, k = 0; BOOL bRet; CString value; while ( 1 ) { j = 0; while ( 1 ){ value = ""; bRet = ReadCell( value, j, i ); if ( !bRet || j > column ) break; j++; if ( value != "(null)" ) { strText += value; strText += " "; // OutputDebugString( value ); // OutputDebugStringA( " " ); } if ( value == "(null)" || value.IsEmpty() ) { strText += " "; //OutputDebugStringA( " " ); } } strText += "\n"; //OutputDebugStringA( "\n" ); i++; if ( i > rows ) break; } } return TRUE; } BOOL CControlWPS::OpenSheet(CString Sheet) { if(varSheets.vt == 0) return FALSE; for (long i = 0; i < m_SheetNames.GetSize(); i++) { if (m_SheetNames.GetAt(i) == Sheet){ // 获取Worksheet对象 CComVariant varItem[1]; varItem[0].vt = VT_I4; varItem[0].lVal = i+1; HRESULT hr = GetProp