多個辨識器範例

發行項
06/13/2023

此範例示範 MicrosoftTablet PC Automation 應用程式開發介面 (API) 用於手寫辨識的進階功能。

其中包含下列各項：

列舉已安裝的辨識器
使用特定語言辨識器建立辨識器內容
使用筆劃集合序列化辨識結果
將筆劃集合組織成 InkDisp 物件內的自訂集合
將筆跡物件序列化至 ISF) 檔案 (，並從筆跡序列化格式 擷取物件
設定辨識器輸入輔助線
使用同步和非同步辨識

筆跡標頭

首先，包含平板電腦自動化介面的標頭。這些會與 Microsoft Windows XP Tablet PC Edition 軟體發展工具組 (SDK) 一起安裝。

#include <msinkaut.h>
#include <msinkaut_i.c>

EventSinks.h 檔案會定義 IInkEventsImpl 和 IInkRecognitionEventsImpl 介面。

#include "EventSinks.h"

列舉已安裝的辨識器

應用程式的 LoadMenu 方法會將可用的辨識器填入 [建立新筆觸] 功能表。 會建立 InkRecognizers。如果InkRecognizers物件的Languages屬性不是空的，則辨識器是文字辨識器，而且其Name屬性的值會新增至功能表。

// Create the enumerator for the installed recognizers
hr = m_spIInkRecognizers.CoCreateInstance(CLSID_InkRecognizers);
...
    // Filter out non-language recognizers by checking for
    // the languages supported by the recognizer - there is not
    // any if it is a gesture or object recognizer.
    CComVariant vLanguages;
    if (SUCCEEDED(spIInkRecognizer->get_Languages(&vLanguages)))
    {
        if ((VT_ARRAY == (VT_ARRAY & vLanguages.vt))           // it should be an array
            && (NULL != vLanguages.parray)
            && (0 < vLanguages.parray->rgsabound[0].cElements)) // with at least one element
        {
            // This is a language recognizer. Add its name to the menu.
            CComBSTR bstrName;
            if (SUCCEEDED(spIInkRecognizer->get_Name(&bstrName)))
                ...
        }
    }

建立筆跡收集器

應用程式的 OnCreate 方法會建立 InkCollector 物件、將它連接到其事件來源，並啟用筆跡收集。

// Create an ink collector object.
hr = m_spIInkCollector.CoCreateInstance(CLSID_InkCollector);

// Establish a connection to the collector's event source.
hr = IInkCollectorEventsImpl<CMultiRecoApp>::DispEventAdvise(m_spIInkCollector);

// Enable ink input in the m_wndInput window
hr = m_spIInkCollector->put_hWnd((long)m_wndInput.m_hWnd);
hr = m_spIInkCollector->put_Enabled(VARIANT_TRUE);

建立辨識器內容

應用程式的 CreateRecoCoNtext 方法會建立並初始化新的辨識器內容，並設定相關聯語言支援的指南。 IInkRecognizer物件的CreateRecognizerCoNtext方法會建立語言的IInkRecognizerCoNtext2物件。如有必要，會取代舊的辨識器內容。內容會連線到其事件來源。最後，會檢查辨識器內容 的功能屬性， 以參考辨識器內容所支援的內容。

// Create a recognizer context
CComPtr<IInkRecognizerContext2> spNewContext;
if (FAILED(pIInkRecognizer2->CreateRecognizerContext(&spNewContext)))
    return false;

// Replace the current context with the new one
if (m_spIInkRecoContext != NULL)
{
    // Close the connection to the recognition events source
    IInkRecognitionEventsImpl<CMultiRecoApp>::DispEventUnadvise(m_spIInkRecoContext);
}
m_spIInkRecoContext.Attach(spNewContext.Detach());

// Establish a connection with the recognizer context's event source
if (FAILED(IInkRecognitionEventsImpl<CMultiRecoApp>::DispEventAdvise(m_spIInkRecoContext)))
    ...

// Set the guide if it's supported by the recognizer and has been created 
int cRows = 0, cColumns = 0;
InkRecognizerCapabilities dwCapabilities = IRC_DontCare;
if (SUCCEEDED(pIInkRecognizer->get_Capabilities(&dwCapabilities)))
    ...

收集筆劃並顯示辨識結果

應用程式的 OnStroke 方法會更新筆跡收集器的 InkStrokes 、取消現有的非同步辨識要求，以及在辨識器內容上建立辨識要求。

// Add the new stroke to the current collection
hr = m_spIInkStrokes->Add(pIInkStroke);

if (SUCCEEDED(hr))
{
    // Cancel the previous background recognition requests
    // which have not been processed yet
    m_spIInkRecoContext->StopBackgroundRecognition();

    // Ask the context to update the recognition results with newly added strokes
    // When the results are ready, the recognizer context returns them
    // through the corresponding event RecognitionWithAlternates
    CComVariant vCustomData;
    m_spIInkRecoContext->BackgroundRecognize(vCustomData);
}

應用程式的 OnRecognition 方法會將辨識要求的結果傳送至輸出視窗的 UpdateString 方法。

// Update the output window with the new results
m_wndResults.UpdateString(bstrRecognizedString);

刪除筆劃和辨識結果

應用程式的 OnClear 方法會刪除 InkDisp 物件中的所有筆劃和辨識結果，並清除視窗。辨識器內容與其 InkStrokes 集合的關聯已移除。

// Detach the current stroke collection from the recognizer context and release it
if (m_spIInkRecoContext != NULL)
    m_spIInkRecoContext->putref_Strokes(NULL);

m_spIInkStrokes.Release();

// Clear the custom strokes collection
if (m_spIInkCustomStrokes != NULL)
    m_spIInkCustomStrokes->Clear();

// Delete all strokes from the Ink object
// Passing NULL as a stroke collection pointer means asking to delete all strokes
m_spIInkDisp->DeleteStrokes(NULL);

// Get a new stroke collection from the ink object
...
// Ask for an empty collection by passing an empty variant 
if (SUCCEEDED(m_spIInkDisp->CreateStrokes(v, &m_spIInkStrokes)))
{
    // Attach it to the recognizer context
    if (FAILED(m_spIInkRecoContext->putref_Strokes(m_spIInkStrokes)))
        ...
}

變更辨識器內容

當使用者在 [建立新筆觸] 功能表中選取辨識器時，會呼叫應用程式的 OnNewStrokes 方法。儲存目前的 InkStrokes 。如果選取不同的語言辨識器，則會建立新的辨識器內容。然後，新的 InkStrokes 會附加至新的辨識器內容。

// Save the current stroke collection if there is any
if (m_spIInkRecoContext != NULL)
{
    // Cancel the previous background recognition requests
    // which have not been processed yet
    m_spIInkRecoContext->StopBackgroundRecognition();
    
    // Let the context know that there'll be no more input 
    // for the attached stroke collection
    m_spIInkRecoContext->EndInkInput();

    // Add the stroke collection to the Ink object's CustomStrokes collection
    SaveStrokeCollection();
}
...
// If a different recognizer was selected, create a new recognizer context
// Else, reuse the same recognizer context
if (wID != m_nCmdRecognizer)
{
    // Get a pointer to the recognizer object from the recognizer collection  
    CComPtr<IInkRecognizer> spIInkRecognizer;
    if ((m_spIInkRecognizers == NULL)
        || FAILED(m_spIInkRecognizers->Item(wID - ID_RECOGNIZER_FIRST,
                                             &spIInkRecognizer))
        || (false == CreateRecoContext(spIInkRecognizer)))
    {
        // restore the cursor
        ::SetCursor(hCursor);
        return 0;
    }

    // Update the status bar
    m_bstrCurRecoName.Empty();
    spIInkRecognizer->get_Name(&m_bstrCurRecoName);
    UpdateStatusBar();

    // Store the selected recognizer's command id
    m_nCmdRecognizer = wID;
}

然後它會呼叫 StartNewStrokeCollection，這會建立空的 InkStrokes ，並將它附加至辨識器內容。

儲存辨識器內容的筆劃集合

應用程式的 SaveStrokeCollection 方法會檢查現有的辨識器內容，並完成目前筆劃集合的辨識。然後將 InkStrokes 集合新增至筆跡物件的 CustomStrokes 。

if (m_spIInkRecoContext != NULL)
{
    if (SUCCEEDED(m_spIInkStrokes->get_Count(&lCount)) && 0 != lCount)
    {
        CComPtr<IInkRecognitionResult> spIInkRecoResult;
        InkRecognitionStatus RecognitionStatus;
        if (SUCCEEDED(m_spIInkRecoContext->Recognize(&RecognitionStatus, &spIInkRecoResult)))
        {
            if (SUCCEEDED(spIInkRecoResult->SetResultOnStrokes()))
            {
                CComBSTR bstr;
                spIInkRecoResult->get_TopString(&bstr);
                m_wndResults.UpdateString(bstr);
            }
            ...
        }
    }
    // Detach the stroke collection from the old recognizer context
    m_spIInkRecoContext->putref_Strokes(NULL);
}

// Now add it to the ink's custom strokes collection
// Each item (stroke collection) of the custom strokes must be identified
// by a unique string. Here we generate a GUID for this.
if ((0 != lCount) && (m_spIInkCustomStrokes != NULL))
{
    GUID guid;
    WCHAR szGuid[40]; // format: "{xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx}"
    if (SUCCEEDED(::CoCreateGuid(&guid)) 
        && (::StringFromGUID2(guid, szGuid, countof(szGuid)) != 0))
    {
        CComBSTR bstrGuid(szGuid);
        if (FAILED(m_spIInkCustomStrokes->Add(bstrGuid, m_spIInkStrokes)))
            ...

共用方式為