开通博客赚积分
发布资源赚积分
分类

源码开发语言/平台
当前位置：首页 > 源码/资料 > 多国语言处理 > 查看源码
MainFrm.cpp
资源名称：TextClassify.rar　[点击查看]
上传用户：sanxfzhen
上传日期：2014-12-28
资源大小：2324k
文件大小：29k
源码类别：

多国语言处理

开发平台：

Visual C++

MainFrm.cpp：源码内容
  1. // MainFrm.cpp : implementation of the CMainFrame class
  2. //
  4. #include "stdafx.h"
  5. #include "TextClassify.h"
  6. #include "Message.h"
  8. #include "MainFrm.h"
  9. #include "LeftView.h"
  10. #include "TextClassifyView.h"
  13. #include "SegDocDlg.h"
  14. #include "WordSegment.h"
  15. #include "TrainParamsPage.h"
  16. #include "TestParamsDlg.h"
  17. #include "classifier.h"
  18. #include "svmparamspage.h"
  19. #include <direct.h>
  21. #ifdef _DEBUG
  22. #define new DEBUG_NEW
  23. #undef THIS_FILE
  24. static char THIS_FILE[] = __FILE__;
  25. #endif
  27. /////////////////////////////////////////////////////////////////////////////
  28. // CMainFrame
  29. CString CMainFrame::m_strResultDir;
  31. IMPLEMENT_DYNCREATE(CMainFrame, CFrameWnd)
  33. BEGIN_MESSAGE_MAP(CMainFrame, CFrameWnd)
  34. //{{AFX_MSG_MAP(CMainFrame)
  35. ON_WM_CREATE()
  36. ON_MESSAGE(WM_COMPUTATION_FINISH,OnComputationFinish)
  37. ON_COMMAND(ID_TOOLS_WORDSEG, OnToolsWordseg)
  38. ON_WM_CLOSE()
  39. ON_WM_TIMER()
  40. ON_COMMAND(ID_MAIN_TRAINKNN, OnMainTrainknn)
  41. ON_UPDATE_COMMAND_UI(ID_MAIN_TRAINKNN, OnUpdateMainTrainknn)
  42. ON_COMMAND(ID_MAIN_TRAINSVM, OnMainTrainsvm)
  43. ON_UPDATE_COMMAND_UI(ID_MAIN_TRAINSVM, OnUpdateMainTrainsvm)
  44. ON_COMMAND(ID_MAIN_TRAINBAYES,OnMainTrainbayes)
  45. ON_UPDATE_COMMAND_UI(ID_MAIN_TRAINBAYES,OnUpdateMainTrainbayes)
  46. ON_COMMAND(ID_MAIN_OPENMODEL, OnMainOpenmodel)
  47. ON_UPDATE_COMMAND_UI(ID_MAIN_OPENMODEL, OnUpdateMainOpenmodel)
  48. ON_COMMAND(ID_MAIN_RESULTS, OnMainResults)
  49. ON_UPDATE_COMMAND_UI(ID_MAIN_RESULTS, OnUpdateMainResults)
  50. ON_COMMAND(ID_MAIN_TEST, OnMainTest)
  51. ON_UPDATE_COMMAND_UI(ID_MAIN_TEST, OnUpdateMainTest)
  52. //}}AFX_MSG_MAP
  53. END_MESSAGE_MAP()
  55. static UINT indicators[] =
  56. {
  57. ID_SEPARATOR,           // status line indicator
  58. // ID_INDICATOR_CAPS,
  59. // ID_INDICATOR_NUM,
  60. // ID_INDICATOR_SCRL,
  61. // ID_INDICATOR_TIME,
  62. };
  65. UINT CMainFrame::TrainThreadProc(LPVOID pParam)
  66. {
  67. theClassifier.m_theSVM.com_param.running=1;
  68. theClassifier.Train(theClassifier.m_paramClassifier.m_nClassifierType);
  69. theClassifier.m_theSVM.com_param.running=0;
  70. ::PostMessage((HWND)pParam,WM_COMPUTATION_FINISH,0,0);
  71. return 0;
  72. }
  74. UINT CMainFrame::TestThreadProc(LPVOID pParam)
  75. {
  76. if(theClassifier.m_paramClassifier.m_nDocFormat==CClassifierParam::nDF_Directory)
  77. {
  78. if(theClassifier.Classify())
  79. CMainFrame::m_strResultDir = theClassifier.m_paramClassifier.m_strResultDir;
  80. }
  81. else if(theClassifier.m_paramClassifier.m_nDocFormat==CClassifierParam::nDF_Smart)
  82. {
  83. if(theClassifier.ClassifySmart())
  84. CMainFrame::m_strResultDir = theClassifier.m_paramClassifier.m_strResultDir;
  85. }
  86. ::PostMessage((HWND)pParam,WM_COMPUTATION_FINISH,0,0);
  87. return 0;
  88. }
  89. /*
  90. UINT CMainFrame::ConvertThreadProc(LPVOID pParam)
  91. {
  92. CTime startTime;
  93. CTimeSpan totalTime;
  95. CMessage::PrintInfo(_T("开始进行文档格式转换，请稍候..."));
  96. startTime=CTime::GetCurrentTime();
  97. theSaxFileHandler.Convert();
  98. totalTime=CTime::GetCurrentTime()-startTime;
  99. CMessage::PrintInfo(_T("文档格式转换结束，耗时")+totalTime.Format("%H:%M:%S"));
  100. CMessage::PrintStatusInfo("");
  102. ::PostMessage((HWND)pParam,WM_COMPUTATION_FINISH,0,0);
  103. ExitThread(0);
  104. return 0;
  105. }
  107. UINT CMainFrame::SeldocsThreadProc(LPVOID pParam)
  108. {
  109. CTime startTime;
  110. CTimeSpan totalTime;
  111. long lDocNum;
  112. CString str;
  114. CMessage::PrintInfo(_T("开始选择训练文档和测试文档，请稍候..."));
  115. startTime=CTime::GetCurrentTime();
  116. lDocNum=theSelectDocs.SelectDocs();
  117. totalTime=CTime::GetCurrentTime()-startTime;
  118. str.Format("共处理了%d篇文档!",lDocNum);
  119. CMessage::PrintInfo(str);
  120. CMessage::PrintInfo(_T("选择文档结束，耗时")+totalTime.Format("%H:%M:%S"));
  121. CMessage::PrintStatusInfo("");
  123. ::PostMessage((HWND)pParam,WM_COMPUTATION_FINISH,0,0);
  124. ExitThread(0);
  125. return 0;
  126. }
  127. */
  128. UINT CMainFrame::SegDocsThreadProc(LPVOID pParam)
  129. {
  130. CTime startTime;
  131. CTimeSpan totalTime;
  132. startTime=CTime::GetCurrentTime();
  133. CMessage::PrintInfo(_T("分词程序初始化，请稍候..."));
  134. if(!g_wordSeg.InitWorgSegment(theApp.m_strPath.GetBuffer(0)))
  135. {
  136. CMessage::PrintError(_T("分词程序初始化失败!"));
  137. return 0;
  138. }
  139. g_wordSeg.SetSegSetting(CWordSegment::uPlace);
  140. CMessage::PrintInfo(_T("正在进行分词，请稍候..."));
  142. if(theSegDocsParam.m_bMultiDocs)
  143. {
  144. CString strTarget,strSource;
  145. HANDLE hFinder;
  146. LPWIN32_FIND_DATA lpFindFileData;
  147. lpFindFileData  = new WIN32_FIND_DATA;
  148. hFinder = ::FindFirstFile(theSegDocsParam.m_strSource+"\*.*",lpFindFileData );
  149. while(::FindNextFile(hFinder,lpFindFileData))
  150. {
  151. if( !strcmp(lpFindFileData->cFileName,".") || !strcmp(lpFindFileData->cFileName,"..") )
  152. continue;
  154. if(!(lpFindFileData->dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY))
  155. {
  156. strTarget=theSegDocsParam.m_strTarget+"\";
  157. strTarget+=lpFindFileData->cFileName;
  158. strSource=theSegDocsParam.m_strSource+"\";
  159. strSource+=lpFindFileData->cFileName;
  160. g_wordSeg.SegmentFile(strSource.GetBuffer(0),
  161. strTarget.GetBuffer(0),
  162. theSegDocsParam.m_bDelStopwords);
  163. CMessage::PrintStatusInfo(lpFindFileData->cFileName);
  164. }
  165. }
  166. delete lpFindFileData;
  167. }
  168. else
  169. g_wordSeg.SegmentFile(theSegDocsParam.m_strSource.GetBuffer(0),
  170. theSegDocsParam.m_strTarget.GetBuffer(0),
  171. theSegDocsParam.m_bDelStopwords);
  173. g_wordSeg.FreeWordSegment();
  174. totalTime=CTime::GetCurrentTime()-startTime;
  175. CMessage::PrintInfo(_T("分词结束，耗时")+totalTime.Format("%H:%M:%S"));
  176. CMessage::PrintStatusInfo("");
  178. ::PostMessage((HWND)pParam,WM_COMPUTATION_FINISH,0,0);
  179. ExitThread(0);
  180. return 0;
  181. }
  186. /////////////////////////////////////////////////////////////////////////////
  187. // CMainFrame construction/destruction
  189. CMainFrame::CMainFrame()
  190. {
  191. // TODO: add member initialization code here
  192. m_pThread=NULL;
  193. m_strModel.Empty();
  194. m_strResultDir.Empty();
  195. m_nRunning=0;
  196. m_bPaused=false;
  197. m_nClassifierType=0;
  198. }
  200. CMainFrame::~CMainFrame()
  201. {
  202. }
  204. int CMainFrame::OnCreate(LPCREATESTRUCT lpCreateStruct)
  205. {
  206. if (CFrameWnd::OnCreate(lpCreateStruct) == -1)
  207. return -1;
  210. if (!m_wndStatusBar.Create(this) ||
  211. !m_wndStatusBar.SetIndicators(indicators,
  212.   sizeof(indicators)/sizeof(UINT)))
  213. {
  214. TRACE0("Failed to create status barn");
  215. return -1;      // fail to create
  216. }
  218. CMessage::SetStatusWnd(&m_wndStatusBar);
  220. return 0;
  221. }
  223. BOOL CMainFrame::OnCreateClient(LPCREATESTRUCT /*lpcs*/,
  224. CCreateContext* pContext)
  225. {
  226. // create splitter window
  227. if (!m_wndSplitter.CreateStatic(this, 1, 2))
  228. return FALSE;
  230. if (!m_wndSplitter.CreateView(0, 0, RUNTIME_CLASS(CLeftView), CSize(100, 100), pContext) ||
  231. !m_wndSplitter.CreateView(0, 1, RUNTIME_CLASS(CTextClassifyView), CSize(100, 100), pContext))
  232. {
  233. m_wndSplitter.DestroyWindow();
  234. return FALSE;
  235. }
  236. CMessage::SetOutputWnd((CWnd *)m_wndSplitter.GetPane(0, 1));
  237. return TRUE;
  238. }
  240. BOOL CMainFrame::PreCreateWindow(CREATESTRUCT& cs)
  241. {
  242. if( !CFrameWnd::PreCreateWindow(cs) )
  243. return FALSE;
  244. // TODO: Modify the Window class or styles here by modifying
  245. //  the CREATESTRUCT cs
  247. return TRUE;
  248. }
  250. /////////////////////////////////////////////////////////////////////////////
  251. // CMainFrame diagnostics
  253. #ifdef _DEBUG
  254. void CMainFrame::AssertValid() const
  255. {
  256. CFrameWnd::AssertValid();
  257. }
  259. void CMainFrame::Dump(CDumpContext& dc) const
  260. {
  261. CFrameWnd::Dump(dc);
  262. }
  264. #endif //_DEBUG
  266. /////////////////////////////////////////////////////////////////////////////
  267. // CMainFrame message handlers
  269. /*
  270. CSvmclsView* CMainFrame::GetRightPane()
  271. {
  272. CWnd* pWnd = m_wndSplitter.GetPane(0, 1);
  273. CSvmclsView* pView = DYNAMIC_DOWNCAST(CSvmclsView, pWnd);
  274. return pView;
  275. }
  277. CLeftView* CMainFrame::GetLeftView()
  278. {
  279. CWnd* pWnd = m_wndSplitter.GetPane(0, 0);
  280. CLeftView* pView = DYNAMIC_DOWNCAST(CLeftView, pWnd);
  281. return pView;
  282. }
  284. void CMainFrame::OnUpdateViewStyles(CCmdUI* pCmdUI)
  285. {
  286. // TODO: customize or extend this code to handle choices on the
  287. // View menu.
  289. CSvmclsView* pView = GetRightPane(); 
  291. // if the right-hand pane hasn't been created or isn't a view,
  292. // disable commands in our range
  294. if (pView == NULL)
  295. pCmdUI->Enable(FALSE);
  296. else
  297. {
  298. DWORD dwStyle = pView->GetStyle() & LVS_TYPEMASK;
  300. // if the command is ID_VIEW_LINEUP, only enable command
  301. // when we're in LVS_ICON or LVS_SMALLICON mode
  303. if (pCmdUI->m_nID == ID_VIEW_LINEUP)
  304. {
  305. if (dwStyle == LVS_ICON || dwStyle == LVS_SMALLICON)
  306. pCmdUI->Enable();
  307. else
  308. pCmdUI->Enable(FALSE);
  309. }
  310. else
  311. {
  312. // otherwise, use dots to reflect the style of the view
  313. pCmdUI->Enable();
  314. BOOL bChecked = FALSE;
  316. switch (pCmdUI->m_nID)
  317. {
  318. case ID_VIEW_DETAILS:
  319. bChecked = (dwStyle == LVS_REPORT);
  320. break;
  322. case ID_VIEW_SMALLICON:
  323. bChecked = (dwStyle == LVS_SMALLICON);
  324. break;
  326. case ID_VIEW_LARGEICON:
  327. bChecked = (dwStyle == LVS_ICON);
  328. break;
  330. case ID_VIEW_LIST:
  331. bChecked = (dwStyle == LVS_LIST);
  332. break;
  334. default:
  335. bChecked = FALSE;
  336. break;
  337. }
  339. pCmdUI->SetRadio(bChecked ? 1 : 0);
  340. }
  341. }
  342. }
  344. */
  345. void CMainFrame::OnComputationFinish(WPARAM wParam,LPARAM lParam)
  346. {
  347. CeaseComputation();
  348. }
  351. void CMainFrame::CeaseComputation()
  352. {
  353. m_wndStatusBar.SetPaneText(0,"运算全部结束!");
  354. m_nRunning=0;
  355. m_pThread=NULL;
  356. m_bPaused=false;
  357. KillTimer(m_nTimer);
  358. }
  360. void CMainFrame::BeginComputation()
  361. {
  362. m_nRunning=1;
  363. m_nTimer=SetTimer(1001,500,NULL);
  364. m_tmBegin=CTime::GetCurrentTime();
  365. }
  366. /*
  367. void CMainFrame::OnTimer(UINT nIDEvent) 
  368. {
  369. CTimeSpan t=CTime::GetCurrentTime()-m_tmBegin;
  370. m_wndStatusBar.SetPaneText(1,t.Format("耗时%H:%M:%S"),TRUE);
  371. CFrameWnd::OnTimer(nIDEvent);
  372. }
  374. void CMainFrame::OnClose() 
  375. {
  376. GetActiveDocument()->SetModifiedFlag(false);
  377. if(m_nRunning>0)
  378. {
  379. if(AfxMessageBox("正在进行计算，真的要停止吗?",MB_YESNO)==IDYES)
  380. {
  381. StopComputation();
  382. CFrameWnd::OnClose();
  383. }
  384. }
  385. else
  386. CFrameWnd::OnClose();
  387. }
  388. */
  389. void CMainFrame::Train()
  390. {
  391. CMessage::ClearInfo();
  392. BeginComputation();
  393. m_pThread=AfxBeginThread(TrainThreadProc,GetSafeHwnd(),THREAD_PRIORITY_NORMAL);
  394. }
  396. void CMainFrame::Test() 
  397. {
  398. CMessage::ClearInfo();
  399. BeginComputation();
  400. m_pThread=AfxBeginThread(TestThreadProc,GetSafeHwnd(),THREAD_PRIORITY_NORMAL);
  401. }
  403. /*
  404. void CMainFrame::OnMainClassifySinglefile() 
  405. {
  406. static char BASED_CODE szTextFilter[] = "Text Files (*.txt)|*.txt|All Files (*.*)|*.*||";
  407. CFileDialog cfd(TRUE,szTextFilter,NULL,OFN_HIDEREADONLY | OFN_OVERWRITEPROMPT,szTextFilter,NULL);
  408. if (cfd.DoModal()==IDOK) 
  409. {
  410. CString strFile=cfd.GetPathName();
  411. short id;
  412. theClassifier.m_paramClassifier.m_dThreshold     =   60;
  413. theClassifier.m_paramClassifier.m_nKNN           =   35;
  414. theClassifier.m_paramClassifier.m_nClassifyType  =   CClassifierParam::nFT_Single;
  415. theClassifier.m_paramClassifier.m_bEvaluation    =   FALSE;
  416. theClassifier.m_paramClassifier.m_bCopyFiles     =   FALSE;
  417. if(theClassifier.m_paramClassifier.m_nClassifierType==CClassifierParam::nCT_KNN)
  418. id=theClassifier.KNNCategory(strFile.GetBuffer(0));
  419. else if(theClassifier.m_paramClassifier.m_nClassifierType==CClassifierParam::nCT_SVM)
  420. id=theClassifier.SVMCategory(strFile.GetBuffer(0));
  421. else
  422. {
  423. CMessage::PrintError("分类器类型不正确!");
  424. return;
  425. }
  426. CMessage::ClearInfo();
  427. if(id>=0) theClassifier.m_lstTrainCatalogList.GetCataName(id,strFile);
  428. else strFile="无法识别";
  429. CMessage::PrintInfo("文档的类别为: "+strFile);
  430. }
  431. }
  432. */
  434. void CMainFrame::OnMainTest() 
  435. {
  436. CTestParamsDlg testParamsDlg;
  438. if(theClassifier.m_paramClassifier.m_strTestDir.IsEmpty())
  439. theClassifier.m_paramClassifier.m_strTestDir=theApp.m_strPath+"\sample\test";
  440. if(theClassifier.m_paramClassifier.m_strResultDir.IsEmpty())
  441. theClassifier.m_paramClassifier.m_strResultDir=theClassifier.m_paramClassifier.m_txtResultDir;
  442. if((theClassifier.m_paramClassifier.m_nClassifierType==CClassifierParam::nCT_SVM)||
  443. (theClassifier.m_paramClassifier.m_dThreshold>0))
  444. {
  445. theClassifier.m_paramClassifier.m_nKNN=35;
  446. theClassifier.m_paramClassifier.m_dThreshold=0;
  447. }
  449. testParamsDlg.m_strTestDir    = theClassifier.m_paramClassifier.m_strTestDir;
  450. testParamsDlg.m_strResultDir  = theClassifier.m_paramClassifier.m_strResultDir;
  451. testParamsDlg.m_nDocFormat    = theClassifier.m_paramClassifier.m_nDocFormat;
  452. testParamsDlg.m_dThreshold    = theClassifier.m_paramClassifier.m_dThreshold;
  453. testParamsDlg.m_nKNN          = theClassifier.m_paramClassifier.m_nKNN;
  454. testParamsDlg.m_nClassifyType = theClassifier.m_paramClassifier.m_nClassifyType;
  455. testParamsDlg.m_bEvaluation   = theClassifier.m_paramClassifier.m_bEvaluation;
  456. testParamsDlg.m_bCopyFiles    = theClassifier.m_paramClassifier.m_bCopyFiles;
  458. testParamsDlg.SetClassifierType(theClassifier.m_paramClassifier.m_nClassifierType);
  460. if(testParamsDlg.DoModal()==IDOK)
  461. {
  462. if((testParamsDlg.m_nDocFormat==CClassifierParam::nDF_Directory)&&
  463. !CheckPath(testParamsDlg.m_strTestDir)) return;
  464. if(!CheckPath(testParamsDlg.m_strResultDir,true)) return;
  466. theClassifier.m_paramClassifier.m_strTestDir     =   testParamsDlg.m_strTestDir;
  467. theClassifier.m_paramClassifier.m_strResultDir   =   testParamsDlg.m_strResultDir;
  468. theClassifier.m_paramClassifier.m_nDocFormat     =   testParamsDlg.m_nDocFormat; 
  469. theClassifier.m_paramClassifier.m_dThreshold     =   testParamsDlg.m_dThreshold;
  470. theClassifier.m_paramClassifier.m_nKNN           =   testParamsDlg.m_nKNN ;
  471. theClassifier.m_paramClassifier.m_nClassifyType  =   testParamsDlg.m_nClassifyType;
  472. theClassifier.m_paramClassifier.m_bEvaluation    =   testParamsDlg.m_bEvaluation;
  473. theClassifier.m_paramClassifier.m_bCopyFiles     =   testParamsDlg.m_bCopyFiles;
  474. Test();
  475. }
  476. }
  478. /*
  479. void CMainFrame::StopComputation()
  480. {
  481. if((m_nRunning>=1)&&(m_pThread!=NULL))
  482. {
  483. if(TerminateThread(m_pThread->m_hThread,1))
  484. {
  485. CeaseComputation();
  486. }
  487. else
  488. AfxMessageBox("无法停止计算过程!");
  489. }
  490. }
  492. void CMainFrame::OnUpdateMainReadFeatures(CCmdUI* pCmdUI) 
  493. {
  494. pCmdUI->Enable(m_nRunning<=0);
  495. }
  496. */
  498. void CMainFrame::OnUpdateMainTrainknn(CCmdUI* pCmdUI) 
  499. {
  500. pCmdUI->Enable(m_nRunning<=0);
  501. }
  503. void CMainFrame::OnUpdateMainTrainsvm(CCmdUI* pCmdUI) 
  504. {
  505. pCmdUI->Enable(m_nRunning<=0);
  506. }
  508. void CMainFrame::OnUpdateMainTrainbayes(CCmdUI* pCmdUI)
  509. {
  510. pCmdUI->Enable(m_nRunning<=0);
  511. }
  513. void CMainFrame::OnUpdateMainTest(CCmdUI* pCmdUI) 
  514. {
  515. pCmdUI->Enable(m_nRunning<=0&&!m_strModel.IsEmpty());
  516. }
  517. /*
  518. void CMainFrame::OnUpdateMainClassify(CCmdUI* pCmdUI) 
  519. {
  520. pCmdUI->Enable(m_nRunning<=0&&!m_strModel.IsEmpty());
  521. }
  523. void CMainFrame::OnUpdateMainClassifySinglefile(CCmdUI* pCmdUI) 
  524. {
  525. pCmdUI->Enable(m_nRunning<=0&&!m_strModel.IsEmpty());
  526. }
  527. */
  529. void CMainFrame::OnUpdateMainResults(CCmdUI* pCmdUI) 
  530. {
  531. pCmdUI->Enable(m_nRunning<=0&&!m_strResultDir.IsEmpty());
  532. }
  534.   
  535. void CMainFrame::OnUpdateMainOpenmodel(CCmdUI* pCmdUI) 
  536. {
  537. pCmdUI->Enable(m_nRunning<=0);
  538. }
  541. /*
  542. void CMainFrame::OnUpdateMainStoprunning(CCmdUI* pCmdUI) 
  543. {
  544. pCmdUI->Enable(m_nRunning);
  545. }
  547. void CMainFrame::OnUpdateMainPause(CCmdUI* pCmdUI) 
  548. {
  549. pCmdUI->Enable(m_nRunning&&!m_bPaused);
  550. }
  552. void CMainFrame::OnUpdateMainResume(CCmdUI* pCmdUI) 
  553. {
  554. pCmdUI->Enable(m_nRunning&&m_bPaused);
  555. }
  557. void CMainFrame::OnUpdateToolsConvert(CCmdUI* pCmdUI) 
  558. {
  559. pCmdUI->Enable(m_nRunning<=0);
  560. }
  562. void CMainFrame::OnUpdateToolsResults(CCmdUI* pCmdUI) 
  563. {
  564. pCmdUI->Enable(m_nRunning<=0);
  565. }
  567. void CMainFrame::OnMainStoprunning() 
  568. {
  569. if(AfxMessageBox("正在进行计算，真的要停止吗?",MB_YESNO)==IDYES)
  570. StopComputation();
  571. }
  573. */
  574. void CMainFrame::OnMainOpenmodel() 
  575. {
  576. static char BASED_CODE szModelFilter[] = "Model Files (*.prj)|*.prj|All Files (*.*)|*.*||";
  577. CFileDialog cfd(TRUE,szModelFilter,NULL,OFN_HIDEREADONLY | OFN_OVERWRITEPROMPT,szModelFilter,NULL);
  578. if (cfd.DoModal()==IDOK) 
  579. {
  580. m_strModel=cfd.GetPathName();
  581. CMessage::ClearInfo();
  582. if(!theClassifier.OpenModel(m_strModel)) m_strModel.Empty();
  583. }
  584. }
  586. /*
  587. void CMainFrame::OnMainPause() 
  588. {
  589. if(!m_bPaused)
  590. {
  591. m_pThread->SuspendThread();
  592. m_wndStatusBar.SetPaneText(0,"运算被用户暂停了!");
  593. m_bPaused=true;
  594. }
  595. }
  598. void CMainFrame::OnMainResume() 
  599. {
  600. if(m_bPaused)
  601. {
  602. m_pThread->ResumeThread();
  603. m_wndStatusBar.SetPaneText(0,"恢复正常运算了!");
  604. m_bPaused=false;
  605. }
  606. }
  607. */
  609. void CMainFrame::OnMainResults() 
  610. {
  611. if(theClassifier.m_paramClassifier.m_bEvaluation)
  612. theClassifier.Evaluate(theApp.m_strPath);
  613. else
  614. {
  615. CMessage::ClearInfo();
  616. CFile fin;
  617. CString fname=theClassifier.m_paramClassifier.m_strResultDir+"\results.txt";
  618. if(!fin.Open(fname,CFile::modeRead))
  619. {
  620. CMessage::PrintError(_T("分类结果文件")+fname+"不存在!");
  621. return;
  622. }
  623. DWORD flen=fin.GetLength();
  624. char *buffer=new char[flen+1];
  625. flen=fin.ReadHuge(buffer,flen);
  626. buffer[flen]='';
  627. fin.Close();
  628. CMessage::PrintInfo("----------------------分类结果----------------------rn");
  629. CMessage::PrintInfo("文档IDt名称tt所属类别rn");
  630. CMessage::PrintInfo(buffer);
  631. delete[] buffer;
  632. }
  633. }
  635. /*
  636. void CMainFrame::OnToolsResults() 
  637. {
  638. CString strFileName=theApp.m_strPath;
  639. strFileName+="\multieval.exe";
  640. if(WinExec(strFileName,SW_SHOWNORMAL)<32)
  641. AfxMessageBox("分类结果评测程序不存在!");
  642. }
  644. void CMainFrame::OnMainReadFeatures() 
  645. {
  646. static char BASED_CODE szModelFilter[] = "Features Files (*.txt)|*.txt|All Files (*.*)|*.*||";
  647. CFileDialog cfd(TRUE,szModelFilter,NULL,OFN_HIDEREADONLY | OFN_OVERWRITEPROMPT,szModelFilter,NULL);
  648. if (cfd.DoModal()==IDOK) 
  649. {
  650. CString strFileName;
  651. strFileName=cfd.GetPathName();
  652. CMessage::ClearInfo();
  653. theClassifier.m_lstTrainWordList.GetListFromFile(strFileName);
  654. }
  655. }
  656. */
  657. void CMainFrame::OnMainTrainknn() 
  658. {
  659. CPropertySheet paramsSheet;
  660. CTrainParamsPage trainParamsPage;
  662. if(theClassifier.m_paramClassifier.m_txtTrainDir.IsEmpty())
  663. theClassifier.m_paramClassifier.m_txtTrainDir=theApp.m_strPath+"\sample\train";
  664. if(theClassifier.m_paramClassifier.m_txtResultDir.IsEmpty())
  665. theClassifier.m_paramClassifier.m_txtResultDir=theApp.m_strPath+"\sample\knn";
  666. trainParamsPage.m_txtTrainDir  = theClassifier.m_paramClassifier.m_txtTrainDir;
  667. trainParamsPage.m_txtResultDir = theClassifier.m_paramClassifier.m_txtResultDir;
  668. trainParamsPage.m_nFSMode      = theClassifier.m_paramClassifier.m_nFSMode;
  669. trainParamsPage.m_nWordSize    = theClassifier.m_paramClassifier.m_nWordSize;
  670. trainParamsPage.m_nSelMode     = theClassifier.m_paramClassifier.m_nSelMode;
  671. trainParamsPage.m_nOpMode      = theClassifier.m_paramClassifier.m_nOpMode;
  672. trainParamsPage.m_nLanguageType= theClassifier.m_paramClassifier.m_nLanguageType;
  673. trainParamsPage.m_bStem        = theClassifier.m_paramClassifier.m_bStem;
  674. trainParamsPage.m_nWeightMode  = theClassifier.m_paramClassifier.m_nWeightMode;
  676. //paramsSheet.SetWizardMode();
  677. paramsSheet.SetTitle("参数设置");
  678. paramsSheet.AddPage(&trainParamsPage);
  679. if(paramsSheet.DoModal()==IDOK)
  680. {
  681. if(!CheckPath(trainParamsPage.m_txtTrainDir)||
  682. !CheckPath(trainParamsPage.m_txtResultDir,true))
  683. return;
  685. theClassifier.m_paramClassifier.m_txtTrainDir    =   trainParamsPage.m_txtTrainDir;
  686. theClassifier.m_paramClassifier.m_txtResultDir   =   trainParamsPage.m_txtResultDir;
  687. theClassifier.m_paramClassifier.m_nFSMode        =   trainParamsPage.m_nFSMode;
  688. theClassifier.m_paramClassifier.m_nWordSize      =   trainParamsPage.m_nWordSize;
  689. theClassifier.m_paramClassifier.m_nSelMode       =   trainParamsPage.m_nSelMode;
  690. theClassifier.m_paramClassifier.m_nOpMode        =   trainParamsPage.m_nOpMode;
  691. theClassifier.m_paramClassifier.m_nLanguageType  =   trainParamsPage.m_nLanguageType;
  692. theClassifier.m_paramClassifier.m_bStem          =   trainParamsPage.m_bStem;
  693. theClassifier.m_paramClassifier.m_nWeightMode    =   trainParamsPage.m_nWeightMode;
  694. theClassifier.m_paramClassifier.m_nClassifierType=   CClassifierParam::nCT_KNN;
  696. Train();
  697. m_strModel=theClassifier.m_paramClassifier.m_txtResultDir+"\model.prj";
  698. theClassifier.m_paramClassifier.m_strModelFile="model";
  699. }
  700. }
  703. void CMainFrame::OnMainTrainsvm() 
  704. {
  705. // TODO: Add your command handler code here
  706. CPropertySheet paramsSheet;
  707. CSVMParamsPage   svmParamsPage;
  708. CTrainParamsPage trainParamsPage;
  710. if(theClassifier.m_paramClassifier.m_txtTrainDir.IsEmpty())
  711. theClassifier.m_paramClassifier.m_txtTrainDir=theApp.m_strPath+"\sample\train";
  712. if(theClassifier.m_paramClassifier.m_txtResultDir.IsEmpty())
  713. theClassifier.m_paramClassifier.m_txtResultDir=theApp.m_strPath+"\sample\svm";
  714. trainParamsPage.m_txtTrainDir  = theClassifier.m_paramClassifier.m_txtTrainDir;
  715. trainParamsPage.m_txtResultDir = theClassifier.m_paramClassifier.m_txtResultDir;
  716. trainParamsPage.m_nFSMode      = theClassifier.m_paramClassifier.m_nFSMode;
  717. trainParamsPage.m_nWordSize    = theClassifier.m_paramClassifier.m_nWordSize;
  718. trainParamsPage.m_nSelMode     = theClassifier.m_paramClassifier.m_nSelMode;
  719. trainParamsPage.m_nOpMode      = theClassifier.m_paramClassifier.m_nOpMode;
  720. trainParamsPage.m_nLanguageType= theClassifier.m_paramClassifier.m_nLanguageType;
  721. trainParamsPage.m_bStem        = theClassifier.m_paramClassifier.m_bStem;
  722. trainParamsPage.m_nWeightMode  = theClassifier.m_paramClassifier.m_nWeightMode;
  724. svmParamsPage.m_bB=theClassifier.m_theSVM.com_param.biased_Hyperplane;
  725. svmParamsPage.m_bI=theClassifier.m_theSVM.com_param.remove_inconsitant;
  726. svmParamsPage.m_dC=theClassifier.m_theSVM.com_param.C;
  727. svmParamsPage.m_dJ=theClassifier.m_theSVM.com_param.cost_factor;
  728. svmParamsPage.m_bX=theClassifier.m_theSVM.com_param.loo;
  729. svmParamsPage.m_fK=theClassifier.m_theSVM.com_param.search_depth;
  730. svmParamsPage.m_fO=theClassifier.m_theSVM.com_param.rho;
  731. svmParamsPage.m_fP=theClassifier.m_theSVM.com_param.fraction;
  732. svmParamsPage.m_fG=theClassifier.m_theSVM.com_param.rbf_gamma;
  733. svmParamsPage.m_fC=theClassifier.m_theSVM.com_param.poly_c;
  734. svmParamsPage.m_fS=theClassifier.m_theSVM.com_param.poly_s;
  735. svmParamsPage.m_iD=theClassifier.m_theSVM.com_param.poly_degree;
  736. svmParamsPage.m_iKernelType=theClassifier.m_theSVM.com_param.kernel_type;
  737. svmParamsPage.m_fE=theClassifier.m_theSVM.com_param.epsion;
  738. svmParamsPage.m_lgIterationTime=theClassifier.m_theSVM.com_param.iteration_time;
  739. svmParamsPage.m_fM=theClassifier.m_theSVM.com_param.cache_size;
  740. svmParamsPage.m_lgNewVariable=theClassifier.m_theSVM.com_param.new_variable;
  741. svmParamsPage.m_lgMaxQPSize=theClassifier.m_theSVM.com_param.maximum_size;
  742. svmParamsPage.m_bFinal=theClassifier.m_theSVM.com_param.final_test;
  744. //paramsSheet.SetWizardMode();
  745. paramsSheet.SetTitle("参数设置");
  746. paramsSheet.AddPage(&trainParamsPage);
  747. paramsSheet.AddPage(&svmParamsPage);
  748. if(paramsSheet.DoModal()==IDOK)
  749. {
  750. if(!CheckPath(trainParamsPage.m_txtTrainDir)||
  751. !CheckPath(trainParamsPage.m_txtResultDir,true))
  752. return;
  754. theClassifier.m_theSVM.com_param.biased_Hyperplane=svmParamsPage.m_bB ;
  755. theClassifier.m_theSVM.com_param.remove_inconsitant=svmParamsPage.m_bI;
  756. theClassifier.m_theSVM.com_param.C                =  svmParamsPage.m_dC;
  757. theClassifier.m_theSVM.com_param.cost_factor      =  svmParamsPage.m_dJ;
  759. theClassifier.m_theSVM.com_param.loo              =  svmParamsPage.m_bX;
  760. theClassifier.m_theSVM.com_param.search_depth     =  svmParamsPage.m_fK;
  761. theClassifier.m_theSVM.com_param.rho              =  svmParamsPage.m_fO;
  762. theClassifier.m_theSVM.com_param.fraction         =  svmParamsPage.m_fP;
  764. theClassifier.m_theSVM.com_param.rbf_gamma        =  svmParamsPage.m_fG  ;
  765. theClassifier.m_theSVM.com_param.poly_c           =  svmParamsPage.m_fC;
  766. theClassifier.m_theSVM.com_param.poly_s           =  svmParamsPage.m_fS;
  767. theClassifier.m_theSVM.com_param.poly_degree      =  svmParamsPage.m_iD  ;
  768. theClassifier.m_theSVM.com_param.kernel_type      =  svmParamsPage.m_iKernelType ;
  770. theClassifier.m_theSVM.com_param.epsion           =  svmParamsPage.m_fE;
  771. theClassifier.m_theSVM.com_param.iteration_time   =  svmParamsPage.m_lgIterationTime;
  772. theClassifier.m_theSVM.com_param.cache_size       =  svmParamsPage.m_fM  ;
  773. theClassifier.m_theSVM.com_param.new_variable     =  svmParamsPage.m_lgNewVariable;
  774. theClassifier.m_theSVM.com_param.maximum_size     =  svmParamsPage.m_lgMaxQPSize;
  775. theClassifier.m_theSVM.com_param.final_test       =  svmParamsPage.m_bFinal  ;
  777. theClassifier.m_paramClassifier.m_txtTrainDir    =   trainParamsPage.m_txtTrainDir;
  778. theClassifier.m_paramClassifier.m_txtResultDir   =   trainParamsPage.m_txtResultDir;
  779. theClassifier.m_paramClassifier.m_nFSMode        =   trainParamsPage.m_nFSMode;
  780. theClassifier.m_paramClassifier.m_nWordSize      =   trainParamsPage.m_nWordSize;
  781. theClassifier.m_paramClassifier.m_nSelMode       =   trainParamsPage.m_nSelMode;
  782. theClassifier.m_paramClassifier.m_nOpMode        =   trainParamsPage.m_nOpMode;
  783. theClassifier.m_paramClassifier.m_nLanguageType  =   trainParamsPage.m_nLanguageType;
  784. theClassifier.m_paramClassifier.m_bStem          =   trainParamsPage.m_bStem;
  785. theClassifier.m_paramClassifier.m_nWeightMode    =   trainParamsPage.m_nWeightMode;
  786. theClassifier.m_paramClassifier.m_nClassifierType=   CClassifierParam::nCT_SVM;
  788. Train();
  789. m_strModel=theClassifier.m_paramClassifier.m_txtResultDir+"\model.prj";
  790. theClassifier.m_paramClassifier.m_strModelFile="model";
  791. }
  792. }
  794. void CMainFrame::OnMainTrainbayes()
  795. {
  796. CPropertySheet paramsSheet;
  797. CTrainParamsPage trainParamsPage;
  799. if(theClassifier.m_paramClassifier.m_txtTrainDir.IsEmpty())
  800. theClassifier.m_paramClassifier.m_txtTrainDir=theApp.m_strPath+"\sample\train";
  801. if(theClassifier.m_paramClassifier.m_txtResultDir.IsEmpty())
  802. theClassifier.m_paramClassifier.m_txtResultDir=theApp.m_strPath+"\sample\bayes";
  803. trainParamsPage.m_txtTrainDir  = theClassifier.m_paramClassifier.m_txtTrainDir;
  804. trainParamsPage.m_txtResultDir = theClassifier.m_paramClassifier.m_txtResultDir;
  805. trainParamsPage.m_nFSMode      = theClassifier.m_paramClassifier.m_nFSMode;
  806. trainParamsPage.m_nWordSize    = theClassifier.m_paramClassifier.m_nWordSize;
  807. trainParamsPage.m_nSelMode     = theClassifier.m_paramClassifier.m_nSelMode;
  808. trainParamsPage.m_nOpMode      = theClassifier.m_paramClassifier.m_nOpMode;
  809. trainParamsPage.m_nLanguageType= theClassifier.m_paramClassifier.m_nLanguageType;
  810. trainParamsPage.m_bStem        = theClassifier.m_paramClassifier.m_bStem;
  811. trainParamsPage.m_nWeightMode  = theClassifier.m_paramClassifier.m_nWeightMode;
  813. //paramsSheet.SetWizardMode();
  814. paramsSheet.SetTitle("参数设置");
  815. paramsSheet.AddPage(&trainParamsPage);
  816. if(paramsSheet.DoModal()==IDOK)
  817. {
  818. if(!CheckPath(trainParamsPage.m_txtTrainDir)||
  819. !CheckPath(trainParamsPage.m_txtResultDir,true))
  820. return;
  822. theClassifier.m_paramClassifier.m_txtTrainDir    =   trainParamsPage.m_txtTrainDir;
  823. theClassifier.m_paramClassifier.m_txtResultDir   =   trainParamsPage.m_txtResultDir;
  824. theClassifier.m_paramClassifier.m_nFSMode        =   trainParamsPage.m_nFSMode;
  825. theClassifier.m_paramClassifier.m_nWordSize      =   trainParamsPage.m_nWordSize;
  826. theClassifier.m_paramClassifier.m_nSelMode       =   trainParamsPage.m_nSelMode;
  827. theClassifier.m_paramClassifier.m_nOpMode        =   trainParamsPage.m_nOpMode;
  828. theClassifier.m_paramClassifier.m_nLanguageType  =   trainParamsPage.m_nLanguageType;
  829. theClassifier.m_paramClassifier.m_bStem          =   trainParamsPage.m_bStem;
  830. theClassifier.m_paramClassifier.m_nWeightMode    =   trainParamsPage.m_nWeightMode;
  831. theClassifier.m_paramClassifier.m_nClassifierType=   CClassifierParam::nCT_BAYES;
  833. Train();
  834. m_strModel=theClassifier.m_paramClassifier.m_txtResultDir+"\model.prj";
  835. theClassifier.m_paramClassifier.m_strModelFile="model";
  836. }
  837. }
  839. bool CMainFrame::CheckPath(CString &strPath, bool bCreateDir)
  840. {
  841. if(_chdir(strPath)<0)
  842. {
  843. if(bCreateDir)
  844. {
  845. if(_mkdir(strPath)<0)
  846. {
  847. CMessage::PrintError("无法创建目录"+strPath+"!");
  848. return false;
  849. }
  850. }
  851. else
  852. {
  853. CMessage::PrintError("目录"+strPath+"不存在!");
  854. return false;
  855. }
  856. }
  857. if(strPath[strPath.GetLength()-1]=='\')
  858. strPath=strPath.Left(strPath.GetLength()-1);
  859. return true;
  860. }
  862. /*
  863. void CMainFrame::OnToolsConvert() 
  864. {
  865. CConvertDlg dlg;
  866. if(dlg.DoModal()==IDOK)
  867. {
  868. if(dlg.m_nSourceFormat==0&&!CheckPath(dlg.m_strSource)) return;
  869. if(dlg.m_nTargetFormat==0&&!CheckPath(dlg.m_strTarget,true)) return;
  871. strcpy(theSaxFileHandler.m_pTarget,dlg.m_strTarget);
  872. strcpy(theSaxFileHandler.m_pSource,dlg.m_strSource);
  873. theSaxFileHandler.m_nTargetFormat=dlg.m_nTargetFormat;
  874. theSaxFileHandler.m_nSourceFormat=dlg.m_nSourceFormat;
  875. theSaxFileHandler.m_bShowErrors=dlg.m_bShowErrors;
  876. theSaxFileHandler.m_nSplitType=dlg.m_nSplitType;
  877. theSaxFileHandler.m_nDocsSetType=dlg.m_nDocsSetType;
  878. theSaxFileHandler.m_nCategoryNum=dlg.m_nCategoryNum;
  879. CMessage::ClearInfo();
  880. BeginComputation();
  881. m_pThread=AfxBeginThread(ConvertThreadProc,GetSafeHwnd(),THREAD_PRIORITY_NORMAL);
  882. }
  883. }
  885. void CMainFrame::OnToolsSeldoc() 
  886. {
  887. // TODO: Add your command handler code here
  888. CDocSelDlg dlg;
  889. dlg.m_strSource=theSelectDocs.m_strSource;
  890. dlg.m_strTarget=theSelectDocs.m_strTarget;
  891. dlg.m_nPercent=theSelectDocs.m_nPercent;
  893. if(dlg.DoModal()==IDOK)
  894. {
  895. theSelectDocs.m_strSource=dlg.m_strSource;
  896. theSelectDocs.m_strTarget=dlg.m_strTarget;
  897. theSelectDocs.m_nPercent=dlg.m_nPercent;
  898. CMessage::ClearInfo();
  899. BeginComputation();
  900. m_pThread=AfxBeginThread(SeldocsThreadProc,GetSafeHwnd(),THREAD_PRIORITY_NORMAL);
  901. }
  902. }
  903. */
  904. void CMainFrame::OnToolsWordseg() 
  905. {
  906. CSegDocDlg dlg;
  907. if(dlg.DoModal()==IDOK)
  908. {
  909. CMessage::ClearInfo();
  910. BeginComputation();
  911. m_pThread=AfxBeginThread(SegDocsThreadProc,GetSafeHwnd(),THREAD_PRIORITY_NORMAL);
  912. }
  913. }