textcodecdetector.h

Go to the documentation of this file.

// edbee - Copyright (c) 2012-2025 by Rick Blommers and contributors
// SPDX-License-Identifier: MIT
 
#pragma once
 
#include "edbee/exports.h"
 
 
class QByteArray;
 
namespace edbee {
 
class TextCodec;

class EDBEE_EXPORT TextCodecDetector {
 
public:
 
    static TextCodec* globalPreferedCodec();
    static void setGlobalPreferedCodec( TextCodec* codec );
 
 
    explicit TextCodecDetector( const QByteArray* buffer=0,  TextCodec* preferedCodec=0 );
    explicit TextCodecDetector( const char* buffer, int length=0, TextCodec* preferedCodec=0 );
    virtual ~TextCodecDetector();
 
 
    virtual TextCodec* detectCodec();

    virtual void setBuffer( const char* buf, int length )
    {
        bufferRef_ = buf;
        bufferLength_ = length;
    }
    virtual void setBuffer( const char* buf, int length ) {…}

    virtual const char*buffer() const { return bufferRef_; }

    virtual int bufferLength() { return bufferLength_; }
 
    virtual void setPreferedCodec( TextCodec* codec=0 );
    virtual TextCodec* preferedCodec() { return preferedCodecRef_; }
 
 
    virtual void setFallbackCodec( TextCodec* codec=0 );
    virtual TextCodec* fallbackCodec() const { return fallbackCodecRef_; }
 
 
 
protected:

    virtual bool isContinuationChar( char b) { return /*-128 <= b && */ b <= -65; }

    virtual bool  isTwoBytesSequence(char b) { return -64 <= b && b <= -33; }

    virtual bool isThreeBytesSequence(char b) { return -32 <= b && b <= -17; }

    virtual bool  isFourBytesSequence(char b) { return -16 <= b && b <= -9; }

    virtual bool  isFiveBytesSequence(char b) { return -8 <= b && b <= -5; }
 
    // If the byte has the form 1110xxx, then it's the first byte of a six-bytes sequence character.
    virtual bool isSixBytesSequence(char b){  return -4 <= b && b <= -3; }
 
public:
    static bool hasUTF8Bom( const char* buffer, int length );
    static bool hasUTF16LEBom( const char* buffer, int length );
    static bool hasUTF16BEBom( const char* buffer, int length );
    static bool hasUTF32LEBom( const char* buffer, int length );
    static bool hasUTF32BEBom( const char* buffer, int length );
 
 
private:
 
    //const QByteArray *bufferRef_;   ///< A reference to the current buffer of data
    const char* bufferRef_;         
    int bufferLength_;              
 
    TextCodec* preferedCodecRef_;  
    TextCodec* fallbackCodecRef_;   
 
 
};
class EDBEE_EXPORT TextCodecDetector {…};
 
} // edbee