EncodingGuesser.h

Go to the documentation of this file.

/*****************************************************************/
#pragma once
#include <vector>
#include <string>
#include "OTCore/CoreAPIExport.h"
#include "OTCore/TextEncoding.h"
 
namespace ot
{
    class OT_CORE_API_EXPORT EncodingGuesser
    {
    public:
        TextEncoding::EncodingStandard operator()(const std::vector<char>& fileContent);
        TextEncoding::EncodingStandard operator()(const char* fileContent, int64_t size);
    private:
        using byte = unsigned char;
        const byte utf8Bom[3] = { 0xEF, 0xBB, 0xBF }; // This is the UTF-16 BOM after being converted to UTF-8
        const byte bigEndianBom[2] = { 0xFE, 0xFF };
        const byte littleEndianBom[2] = { 0xFF, 0xFE };
 
        bool CheckIfISO(byte* fileContent, int64_t numberOfBytes);
    };
}