ParagraphAbsorber.h

#pragma once
// Copyright (c) 2001-2024 Aspose Pty Ltd. All Rights Reserved.

#include <system/details/array_view.h>
#include <system/array.h>
#include <cstdint>

#include "Aspose.PDF.Cpp/aspose_pdf_api_defs.h"

namespace Aspose
{
namespace Pdf
{
class Document;
class Page;
class Rectangle;
namespace Text
{
class MarkupSection;
class PageMarkup;
class ParagraphAbsorberOptions;
class TextReplaceOptions;
} // namespace Text
} // namespace Pdf
} // namespace Aspose
namespace System
{
namespace Collections
{
namespace Generic
{
template <typename> class List;
} // namespace Generic
} // namespace Collections
} // namespace System

namespace Aspose {

namespace Pdf {

namespace Text {

class ASPOSE_PDF_SHARED_CLASS ParagraphAbsorber : public System::Object
{
    typedef ParagraphAbsorber ThisType;
    typedef System::Object BaseType;
    
    typedef ::System::BaseTypesInfo<BaseType> ThisTypeBaseTypesInfo;
    ASPOSE_PDF_SHARED_RTTI_INFO_DECL();
    
    friend class Aspose::Pdf::Text::MarkupSection;
    friend class Aspose::Pdf::Text::PageMarkup;
    
public:

    ASPOSE_PDF_SHARED_API System::SharedPtr<System::Collections::Generic::List<System::SharedPtr<PageMarkup>>> get_PageMarkups() const;
    ASPOSE_PDF_SHARED_API int32_t get_SectionsSearchDepth() const;
    ASPOSE_PDF_SHARED_API void set_SectionsSearchDepth(int32_t value);
    ASPOSE_PDF_SHARED_API bool get_IsMulticolumnParagraphsAllowed() const;
    ASPOSE_PDF_SHARED_API void set_IsMulticolumnParagraphsAllowed(bool value);
    ASPOSE_PDF_SHARED_API System::SharedPtr<Aspose::Pdf::Text::ParagraphAbsorberOptions> get_ParagraphAbsorberOptions() const;
    ASPOSE_PDF_SHARED_API void set_ParagraphAbsorberOptions(System::SharedPtr<Aspose::Pdf::Text::ParagraphAbsorberOptions> value);
    ASPOSE_PDF_SHARED_API System::SharedPtr<Aspose::Pdf::Text::TextReplaceOptions> get_TextReplaceOptions() const;
    ASPOSE_PDF_SHARED_API void set_TextReplaceOptions(System::SharedPtr<Aspose::Pdf::Text::TextReplaceOptions> value);
    
    ASPOSE_PDF_SHARED_API ParagraphAbsorber();
    ASPOSE_PDF_SHARED_API ParagraphAbsorber(int32_t sectionsSearchDepth);
    ASPOSE_PDF_SHARED_API ParagraphAbsorber(System::SharedPtr<Aspose::Pdf::Text::ParagraphAbsorberOptions> paragraphAbsorberOptions);
    ASPOSE_PDF_SHARED_API ParagraphAbsorber(int32_t sectionsSearchDepth, System::SharedPtr<Aspose::Pdf::Text::ParagraphAbsorberOptions> paragraphAbsorberOptions);
    
    ASPOSE_PDF_SHARED_API void Visit(System::SharedPtr<Document> doc);
    ASPOSE_PDF_SHARED_API void Visit(System::SharedPtr<Page> page);
    
protected:

    static System::ArrayPtr<int32_t> GetGridIndicesForRect(System::SharedPtr<Rectangle> rect, double scaleFactor);
    static System::SharedPtr<Rectangle> GetRectFromGridByIndices(System::Details::ArrayView<int32_t> indices, double scaleFactor);
    static int32_t GetGridIndexForX(double x, double scaleFactor);
    static bool IsCapitalLetter(char16_t c);
    
private:

    int32_t _sectionsSearchDepth;
    double _minBreakingHorizontalFillingLevel;
    double _minBreakingVerticalFillingLevel;
    int32_t _minBreakingHorizontalFillingOnGrid;
    int32_t _minBreakingVerticalFillingOnGrid;
    double _maxUnbreakingVLength;
    double _maxUnbreakingHLength;
    int32_t _maxUnbreakingVLengthOnGrid;
    int32_t _maxUnbreakingHLengthOnGrid;
    System::SharedPtr<System::Collections::Generic::List<System::SharedPtr<PageMarkup>>> _markups;
    System::String _errors;
    bool _isMulticolumnParagraphs;
    System::SharedPtr<Aspose::Pdf::Text::ParagraphAbsorberOptions> _paragraphAbsorberOptions;
    System::SharedPtr<Aspose::Pdf::Text::TextReplaceOptions> _textReplaceOptions;
    
    void ProcessMarkupSections(System::SharedPtr<PageMarkup> markup);
    void Split4Horizontal(System::SharedPtr<MarkupSection> section, System::ArrayPtr<std::vector<uint8_t>> grid, System::SharedPtr<System::Collections::Generic::List<System::SharedPtr<MarkupSection>>>& outSections);
    void Split4Vertical(System::SharedPtr<MarkupSection> section, System::ArrayPtr<std::vector<uint8_t>> grid, System::SharedPtr<System::Collections::Generic::List<System::SharedPtr<MarkupSection>>>& outSections);
    System::SharedPtr<System::Collections::Generic::List<System::SharedPtr<MarkupSection>>> SortSections(System::SharedPtr<System::Collections::Generic::List<System::SharedPtr<MarkupSection>>> unsortedSections);
    void ToCrossSectionParagraphs();
    void ToIsolatedSectionParagraphs();
    
};

} // namespace Text
} // namespace Pdf
} // namespace Aspose