TextAbsorber.h

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

#include <system/object.h>
#include <cstdint>

#include "Aspose.PDF.Cpp/Text/TextOptions/TextExtractionOptions.h"
#include "Aspose.PDF.Cpp/aspose_pdf_api_defs.h"

namespace Aspose
{
namespace Pdf
{
class Document;
namespace Engine
{
namespace CommonData
{
namespace Text
{
namespace Segmenting
{
class TextSegmenter;
} // namespace Segmenting
} // namespace Text
} // namespace CommonData
} // namespace Engine
namespace GroupProcessor
{
class TextExtractor;
} // namespace GroupProcessor
class Page;
namespace Text
{
class TextExtractionError;
class TextSearchOptions;
} // namespace Text
class XForm;
} // namespace Pdf
} // namespace Aspose
namespace System
{
namespace Collections
{
namespace Generic
{
template <typename> class List;
} // namespace Generic
} // namespace Collections
class String;
namespace Text
{
class StringBuilder;
} // namespace Text
} // namespace System

namespace Aspose {

namespace Pdf {

namespace Text {

class ASPOSE_PDF_SHARED_CLASS TextAbsorber : public System::Object
{
    typedef TextAbsorber ThisType;
    typedef System::Object BaseType;
    
    typedef ::System::BaseTypesInfo<BaseType> ThisTypeBaseTypesInfo;
    ASPOSE_PDF_SHARED_RTTI_INFO_DECL();
    
    friend class Aspose::Pdf::GroupProcessor::TextExtractor;
    
public:

    virtual ASPOSE_PDF_SHARED_API System::String get_Text();
    ASPOSE_PDF_SHARED_API bool get_HasErrors() const;
    ASPOSE_PDF_SHARED_API System::SharedPtr<System::Collections::Generic::List<System::SharedPtr<TextExtractionError>>> get_Errors() const;
    virtual ASPOSE_PDF_SHARED_API System::SharedPtr<TextExtractionOptions> get_ExtractionOptions();
    virtual ASPOSE_PDF_SHARED_API void set_ExtractionOptions(System::SharedPtr<TextExtractionOptions> value);
    virtual ASPOSE_PDF_SHARED_API System::SharedPtr<Aspose::Pdf::Text::TextSearchOptions> get_TextSearchOptions();
    virtual ASPOSE_PDF_SHARED_API void set_TextSearchOptions(System::SharedPtr<Aspose::Pdf::Text::TextSearchOptions> value);
    
    virtual ASPOSE_PDF_SHARED_API void Visit(System::SharedPtr<Page> page);
    virtual ASPOSE_PDF_SHARED_API void Visit(System::SharedPtr<XForm> form);
    virtual ASPOSE_PDF_SHARED_API void Visit(System::SharedPtr<Document> pdf);
    
    ASPOSE_PDF_SHARED_API TextAbsorber();
    ASPOSE_PDF_SHARED_API TextAbsorber(System::SharedPtr<TextExtractionOptions> extractionOptions);
    ASPOSE_PDF_SHARED_API TextAbsorber(System::SharedPtr<TextExtractionOptions> extractionOptions, System::SharedPtr<Aspose::Pdf::Text::TextSearchOptions> textSearchOptions);
    ASPOSE_PDF_SHARED_API TextAbsorber(System::SharedPtr<Aspose::Pdf::Text::TextSearchOptions> textSearchOptions);
    
protected:

    System::SharedPtr<System::Text::StringBuilder> extractedText;
    
    System::SharedPtr<System::Collections::Generic::List<int32_t>> get_PageTextLengthes() const;
    
    System::String GetTotalText(System::SharedPtr<Aspose::Pdf::Engine::CommonData::Text::Segmenting::TextSegmenter> segmenter, TextExtractionOptions::TextFormattingMode mode);
    
private:

    static const int32_t EvaluationCharsCount;
    System::SharedPtr<TextExtractionOptions> extractionOptions;
    System::SharedPtr<Aspose::Pdf::Text::TextSearchOptions> textSearchOptions;
    bool _hasErrors;
    System::SharedPtr<System::Collections::Generic::List<System::SharedPtr<TextExtractionError>>> _errors;
    System::SharedPtr<System::Collections::Generic::List<int32_t>> pageTextLengthes;
    
    void Init();
    System::String GetExtractedText(System::SharedPtr<Aspose::Pdf::Engine::CommonData::Text::Segmenting::TextSegmenter> segmenter, TextExtractionOptions::TextFormattingMode mode);
    System::String Deligaturize(System::String phrase);
    
};

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