Utf8TextProcessor.h

Go to the documentation of this file.

/*
 * Copyright (c) 2018 https://www.thecoderscorner.com (Dave Cherry).
 * This product is licensed under an Apache license, see the LICENSE file in the top-level directory.
 */
 
#ifndef TCMENU_UTF8TEXTPROCESSOR_H
#define TCMENU_UTF8TEXTPROCESSOR_H
 
#include <string.h>
#include <inttypes.h>
 
/* the value that defines an error in character conversion, usual procedure is to call reset() on the text processor */
#define TC_UNICODE_CHAR_ERROR 0xffffffff
 
namespace tccore {
 
    enum UnicodeEncodingMode { ENCMODE_UTF8, ENCMODE_EXT_ASCII };
 
    typedef void (*UnicodeCharacterHandler)(void* callbackData, uint32_t convertedChar);
 
    class Utf8TextProcessor {
    public:
        enum DecoderState {
            WAITING_BYTE_0, WAITING_BYTE_1, WAITING_BYTE_2, WAITING_BYTE_3, UTF_CHAR_FOUND
        };
    private:
        DecoderState decoderState = WAITING_BYTE_0;
        uint32_t currentUtfChar = 0U;
        int extraCharsNeeded = 0;
        UnicodeCharacterHandler handler;
        void* userData;
        const UnicodeEncodingMode encodingMode;
    public:
        explicit Utf8TextProcessor(UnicodeCharacterHandler handler, void* userData, UnicodeEncodingMode mode)
                : handler(handler), userData(userData), encodingMode(mode) {}
 
        void reset();
 
        void pushChar(char ch);
 
        void pushChars(const char *str);
 
    private:
        void error(char lastCode);
        bool couldSequenceBeSmaller() const;
 
        void processChar0(char data);
    };
 
}
 
 
 
#endif //TCMENU_UTF8TEXTPROCESSOR_H