DataExtractionModule.h

Go to the documentation of this file.

//---------------------------------------------------------------------------------------
// Copyright (c) 2001-2024 by Apryse Software Inc. All Rights Reserved.
// Consult legal.txt regarding legal and license information.
//---------------------------------------------------------------------------------------
// !Warning! This file is autogenerated, modify the .codegen file, not this one
// (any changes here will be wiped out during the autogen process)

#ifndef PDFTRON_H_CPPPDFDataExtractionModule
#define PDFTRON_H_CPPPDFDataExtractionModule
#include <C/PDF/TRN_DataExtractionModule.h>

#include <PDF/DataExtractionOptions.h>
#include <Common/BasicTypes.h>
#include <Common/UString.h>
#include <PDF/PDFDoc.h>

namespace pdftron { namespace PDF { 

class DataExtractionModule
{
public:
    /*
     * Data Extraction Engines
     */
    enum DataExtractionEngine {
        e_Tabular = 0,
        e_Form = 1,
        e_DocStructure = 2,
        e_FormKeyValue = 3
    };

    static bool IsModuleAvailable(DataExtractionEngine engine);
    
    static UString ExtractData(const UString& input_pdf_file, DataExtractionEngine engine, DataExtractionOptions* options = 0);
    
    static void ExtractData(const UString& input_pdf_file, const UString& output_json_file, DataExtractionEngine engine, DataExtractionOptions* options = 0);

    static void DetectAndAddFormFieldsToPDF(PDFDoc& doc, DataExtractionOptions* options = 0);
    
    static void ExtractToXLSX(const UString& input_pdf_file, const UString& output_xlsx_file, DataExtractionOptions* options = 0);
    
    static void ExtractToXLSX(const UString& input_pdf_file, Filters::Filter& output_xlsx_stream, DataExtractionOptions* options = 0);

};

#include <Impl/DataExtractionModule.inl>
} //end PDF
} //end pdftron


#endif //PDFTRON_H_CPPPDFDataExtractionModule