uni/inf2270/hw3/fasit.c

#include <stdio.h>
#include <stdlib.h>

typedef unsigned char byte;
typedef unsigned long unicode;

// The minimum value of the first `char` to indicate n bytes
#define C4bytes 0xF0
#define C3bytes 0xE0
#define C2bytes 0xC0

// The minimum value for the vm_var_char for the utf8 equivalent to be n bytes
#define U4bytes 0x10000
#define U3bytes 0x0800
#define U2bytes 0x0080

void writebyte(FILE *f, byte b)
{
	fwrite(&b, 1, 1, f);
}

void writeutf8char(FILE *f, unicode u)
{
	if (u < U2bytes)
	{
		writebyte(f, (byte)u);
		return;
	}

	int left;
	int a;
	int b;
	if (u >= U4bytes)
	{
		left = 18;
		a = 0b11110000;
		b = 0b00000111;
	}
	else if (u >= U3bytes)
	{
		left = 12;
		a = 0b11100000;
		b = 0b00001111;
	}
	else
	{
		left = 6;
		a = 0b11000000;
		b = 0b00011111;
	}

	while (1)
	{
		writebyte(f, a | ((u >> left) & b));
		if (left == 0)
			return;
		a = 0b10000000;
		b = 0b00111111;
		left -= 6;
	}
}

int readbyte(FILE *f)
{
	int status;
	byte c;
	status = fread(&c, 1, 1, f);
	if (status <= 0) return -1;
	return (int)c;
}

long readutf8char(FILE *f)
{
	unicode u = 0;
	unicode ch = readbyte(f);

	if (ch == -1)
		return -1;

	if (ch < C2bytes)
		return ch;

	int left;
	int mask;
	if (ch >= C4bytes)
	{
		left = 18;
		mask = 0b00000111;
	}
	else if (ch >= C3bytes)
	{
		left = 12;
		mask = 0b00001111;
	}
	else
	{
		left = 6;
		mask = 0b00011111;
	}

	while (1)
	{
		u |= (ch & mask) << left;

		if (left == 0)
			return u;

		left -= 6;
		mask = 0b00111111;
		ch = readbyte(f);
	}
}