PtlParamExtractText.h

// PtlParamExtractText.h
// Copyright (C) 2018-2024 Antenna House,Inc.
// All rights reserved.
#ifndef PDFTK_PTLPARAMEXTRACTTEXT_H_
#define PDFTK_PTLPARAMEXTRACTTEXT_H_
 
#include "PtlParamString.h"
 
namespace PdfTk {
 
class PtlRect;

 
class PDFTK_API PtlParamExtractText
{
public:
    // 抽出するテキスト
    enum TEXT_TYPE {
        TEXT_RAW            = 1,    
        TEXT_SORT          = 2     
    };

    enum UNI_FLAG {
        AHEXTRACTTEXT_UNI_C0_CONTROLS          = 0x00000001,   
        AHEXTRACTTEXT_UNI_SPACE                  = 0x00000002,   
        AHEXTRACTTEXT_UNI_NO_BREAK_SPACE        = 0x00000004,   
        AHEXTRACTTEXT_UNI_SOFT_HYPHEN          = 0x00000008,   
        AHEXTRACTTEXT_UNI_SPACES                = 0x00000010,   
        AHEXTRACTTEXT_UNI_IDEOGRAPHIC_SPACE      = 0x00000020,   
        AHEXTRACTTEXT_UNI_VARIATION_SELECTOR    = 0x00000040,   
        AHEXTRACTTEXT_UNI_REPLACEMENT_CHARACTER  = 0x00000080,   
        AHEXTRACTTEXT_UNI_ALL                  = 0x0FFFFFFF    
    };
 
private:
    int m_dataParamExtractText;
public:
    int getHandle() const;
 
public:
    PtlParamExtractText();

    PtlParamExtractText(const PtlParamExtractText& obj);

    ~PtlParamExtractText();
 
public:
    PtlParamExtractText& operator=(const PtlParamExtractText& obj);
 
public:
    void setTextType(TEXT_TYPE type);

    void appendRect(const PtlRect& rectMM);

    void setIgnoreActualText(bool ignoreActualText);
    
    void setTextOverlapRatio(float overlapRatio);

    void setOverlapAsLine(float overlap);

    void setSameTextOmitRatio(float overlap);

    void setDifferentTextOmitRatio(float overlap, bool applyAtDifferentColor);

    void setUnicodeToSpace(int flag);

    void setUnicodeToSpace(CP_UChar uni);

    void setUnicodeToRemove(int flag);

    void setUnicodeToRemove(CP_UChar uni);

    void setInsertReturn(bool insert);
};
 
} // namespace PdfTk
 
#endif // PDFTK_PTLPARAMEXTRACTTEXT_H_