PdfExtractor.h

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

// C# preprocessor directive: #if ASPOSE_PDF_DRAWING

// C# INACTIVE CODE:
// using ImageFormat = Aspose.Pdf.Drawing.ImageFormat;

// C# preprocessor directive: #else


// C# preprocessor directive: #endif


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

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

namespace Aspose
{
namespace Pdf
{
class Document;
enum class ExtractImageMode;
namespace Facades
{
class PdfContentEditor;
} // namespace Facades
class FileSpecification;
namespace LicenseManagement
{
class VentureLicense;
class VentureLicenseDecoder;
} // namespace LicenseManagement
namespace Text
{
class TextSearchOptions;
} // namespace Text
class XForm;
class XImage;
} // namespace Pdf
} // namespace Aspose
namespace System
{
namespace Collections
{
namespace Generic
{
template <typename, typename> class Dictionary;
template <typename> class IList;
template <typename> class List;
} // namespace Generic
} // namespace Collections
namespace Drawing
{
namespace Imaging
{
class ImageFormat;
} // namespace Imaging
} // namespace Drawing
namespace IO
{
class MemoryStream;
class Stream;
} // namespace IO
namespace Text
{
class Encoding;
} // namespace Text
} // namespace System

namespace Aspose {

namespace Pdf {

namespace Facades {

class ASPOSE_PDF_SHARED_CLASS PdfExtractor final : public Aspose::Pdf::Facades::Facade
{
    typedef PdfExtractor ThisType;
    typedef Aspose::Pdf::Facades::Facade BaseType;
    
    typedef ::System::BaseTypesInfo<BaseType> ThisTypeBaseTypesInfo;
    ASPOSE_PDF_SHARED_RTTI_INFO_DECL();
    
    friend class Aspose::Pdf::LicenseManagement::VentureLicenseDecoder;
    friend class Aspose::Pdf::Facades::PdfContentEditor;
    
public:

    ASPOSE_PDF_SHARED_API int32_t get_StartPage() const;
    ASPOSE_PDF_SHARED_API void set_StartPage(int32_t value);
    ASPOSE_PDF_SHARED_API int32_t get_EndPage() const;
    ASPOSE_PDF_SHARED_API void set_EndPage(int32_t value);
    ASPOSE_PDF_SHARED_API int32_t get_ExtractTextMode() const;
    ASPOSE_PDF_SHARED_API void set_ExtractTextMode(int32_t value);
    ASPOSE_PDF_SHARED_API System::SharedPtr<Aspose::Pdf::Text::TextSearchOptions> get_TextSearchOptions() const;
    ASPOSE_PDF_SHARED_API void set_TextSearchOptions(System::SharedPtr<Aspose::Pdf::Text::TextSearchOptions> value);
    ASPOSE_PDF_SHARED_API Aspose::Pdf::ExtractImageMode get_ExtractImageMode() const;
    ASPOSE_PDF_SHARED_API void set_ExtractImageMode(Aspose::Pdf::ExtractImageMode value);
    ASPOSE_PDF_SHARED_API bool get_IsBidi();
    ASPOSE_PDF_SHARED_API int32_t get_Resolution() const;
    ASPOSE_PDF_SHARED_API void set_Resolution(int32_t value);
    ASPOSE_PDF_SHARED_API System::String get_Password() const;
    ASPOSE_PDF_SHARED_API void set_Password(System::String value);
    
    ASPOSE_PDF_SHARED_API void ExtractText();
    //[Obfuscation(Feature = "virtualization", Exclude = false)]
    ASPOSE_PDF_SHARED_API void ExtractText(System::SharedPtr<System::Text::Encoding> encoding);
    ASPOSE_PDF_SHARED_API void GetText(System::String outputFile);
    ASPOSE_PDF_SHARED_API void GetText(System::SharedPtr<System::IO::Stream> outputStream);
    ASPOSE_PDF_SHARED_API void BindPdf(System::String inputFile) override;
    ASPOSE_PDF_SHARED_API void BindPdf(System::SharedPtr<System::IO::Stream> inputStream) override;
    ASPOSE_PDF_SHARED_API void ExtractImage();
    ASPOSE_PDF_SHARED_API bool HasNextImage();
    ASPOSE_PDF_SHARED_API bool GetNextImage(System::String outputFile);
    ASPOSE_PDF_SHARED_API bool GetNextImage(System::String outputFile, System::SharedPtr<System::Drawing::Imaging::ImageFormat> format);
    ASPOSE_PDF_SHARED_API bool GetNextImage(System::SharedPtr<System::IO::Stream> outputStream, System::SharedPtr<System::Drawing::Imaging::ImageFormat> format);
    ASPOSE_PDF_SHARED_API bool GetNextImage(System::SharedPtr<System::IO::Stream> outputStream);
    ASPOSE_PDF_SHARED_API System::SharedPtr<System::Collections::Generic::IList<System::String>> GetAttachNames();
    ASPOSE_PDF_SHARED_API void ExtractAttachment();
    ASPOSE_PDF_SHARED_API void ExtractAttachment(System::String attachmentFileName);
    ASPOSE_PDF_SHARED_API void GetAttachment(System::String outputPath);
    ASPOSE_PDF_SHARED_API bool HasNextPageText();
    ASPOSE_PDF_SHARED_API void GetNextPageText(System::String outputFile);
    ASPOSE_PDF_SHARED_API void GetNextPageText(System::SharedPtr<System::IO::Stream> outputStream);
    
    ASPOSE_PDF_SHARED_API PdfExtractor();
    ASPOSE_PDF_SHARED_API PdfExtractor(System::SharedPtr<Aspose::Pdf::Document> document);
    
    ASPOSE_PDF_SHARED_API void GetText(System::SharedPtr<System::IO::Stream> outputStream, bool filterNotAscii);
    ASPOSE_PDF_SHARED_API System::ArrayPtr<System::SharedPtr<System::IO::MemoryStream>> GetAttachment();
    ASPOSE_PDF_SHARED_API System::SharedPtr<System::Collections::Generic::List<System::SharedPtr<FileSpecification>>> GetAttachmentInfo();
    
protected:

    bool get__IsObjectLicensed();
    
    void SetVentureLicense(System::SharedPtr<Aspose::Pdf::LicenseManagement::VentureLicense> license) override;
    System::SharedPtr<Aspose::Pdf::LicenseManagement::VentureLicense> GetVentureLicense() override;
    static void InitXFormImages(System::SharedPtr<System::Collections::Generic::List<System::SharedPtr<XImage>>> images, System::SharedPtr<System::Collections::Generic::List<int32_t>> processedObjectIds, System::SharedPtr<XForm> xForm);
    void InitPageImages(System::SharedPtr<System::Collections::Generic::List<System::SharedPtr<XImage>>> images, System::SharedPtr<Aspose::Pdf::Document> document, int32_t page, int32_t endPage);
    static void InitPageXFormImages_ActuallyUsed(System::SharedPtr<System::Collections::Generic::List<System::SharedPtr<XImage>>> images, System::SharedPtr<Aspose::Pdf::Document> document, int32_t page, int32_t endPage);
    void InitPageXFormImages_DefinedInResources(System::SharedPtr<System::Collections::Generic::List<System::SharedPtr<XImage>>> images, System::SharedPtr<Aspose::Pdf::Document> document, int32_t page, int32_t endPage);
    
private:

    int32_t _startPage;
    int32_t _endPage;
    int32_t _currentPage;
    System::SharedPtr<System::Collections::Generic::List<System::SharedPtr<XImage>>> _images;
    int32_t _currentImage;
    int32_t _currentNoNameAttachment;
    System::String _attachmentFileName;
    System::SharedPtr<System::Collections::Generic::List<System::String>> _pageTexts;
    int32_t _extractTextMode;
    Aspose::Pdf::ExtractImageMode _extractImageMode;
    System::SharedPtr<Aspose::Pdf::Text::TextSearchOptions> _textSearchOptions;
    int32_t _resolution;
    System::String _password;
    System::SharedPtr<System::Text::Encoding> _encoding;
    System::SharedPtr<System::Collections::Generic::Dictionary<int32_t, int32_t>> _uniqueObjectIDs;
    System::SharedPtr<System::Collections::Generic::List<int32_t>> _processedObjectIds;
    System::SharedPtr<Aspose::Pdf::LicenseManagement::VentureLicense> _ventureLicense;
    
    void WriteAttachment(System::SharedPtr<FileSpecification> file, System::String outputPath);
    System::ArrayPtr<System::SharedPtr<System::IO::MemoryStream>> GetAttachment(System::String outputPath, bool fileOut);
    System::SharedPtr<System::Collections::Generic::List<System::SharedPtr<FileSpecification>>> GetAllAttachments();
    void InitPageBounds();
    bool IsArabic(char16_t chr);
    bool IsHebriew(char16_t chr);
    void Reset();
    
};

} // namespace Facades
} // namespace Pdf
} // namespace Aspose