// utf8purify.c
//
// Copyright (C) 2007, 2011 by Clifford Wolf <www.clifford.at>
//
// This program is free software; you can redistribute it and/or modify
// it under the terms of the GNU General Public License as published by
// the Free Software Foundation; either version 2 of the License, or
// (at your option) any later version.
//
// Read a wild mix of utf8 and latin1 from input
// Write pure utf8 to output
//
// A byte walks into a bar and orders a pint. Bartender asks him "What's
// wrong?" Byte says "Parity error." Bartender nods and says "Yeah, I
// thought you looked a bit off."

#include <unistd.h>
#include <stdlib.h>
#include <string.h>
#include <errno.h>
#include <stdio.h>

// lookahead buffer
unsigned char *inputstr = NULL;
unsigned char *outputstr = NULL;
int labuffer[3];

// read a byte from input and shift thru the lookahead buffer
int readbyte()
{
	int retval = labuffer[0];
	labuffer[0] = labuffer[1];
	labuffer[1] = labuffer[2];
	labuffer[2] = -1;

	unsigned char buffer;
	if (inputstr != NULL) {
		if (*inputstr)
			labuffer[2] = *(inputstr++);
	} else {
		if (read(0, &buffer, 1) == 1)
			labuffer[2] = buffer;
	}

	return retval;
}

// write a byte to the output
void writebyte(int ch)
{
	unsigned char buffer = (unsigned)ch;
	if (outputstr) {
		*(outputstr++) = buffer;
		*outputstr = 0;
	} else {
		if (write(1, &buffer, 1) != 1)
			perror("Single byte write failed");
	}
}

void worker()
{
	// fill lookahead buffer
	readbyte();
	readbyte();
	readbyte();

	// main loop
	while (1)
	{
		// read the first byte (rest is in the lookahead buffer)
		int ch = readbyte();

		// we have reached the end of the input
		if (ch < 0)
			break;

		// just normal 7-bit ascii
		if ((ch & 0x80) == 0) {
			writebyte(ch);
			continue;
		}

		// 2-byte utf-8 sequence (starting with '110')
		if ((ch & 0xE0) == 0xC0 && (labuffer[0] & 0xC0) == 0x80) {
			writebyte(ch);
			writebyte(readbyte());
			continue;
		}

		// 3-byte utf-8 sequence (starting with '1110')
		if ((ch & 0xF0) == 0xE0 && (labuffer[0] & 0xC0) == 0x80 &&
				(labuffer[1] & 0xC0) == 0x80) {
			writebyte(ch);
			writebyte(readbyte());
			writebyte(readbyte());
			continue;
		}

		// 4-byte utf-8 sequence (starting with '11110')
		if ((ch & 0xF8) == 0xF0 && (labuffer[0] & 0xC0) == 0x80 &&
				(labuffer[1] & 0xC0) == 0x80 &&
				(labuffer[2] & 0xC0) == 0x80) {
			writebyte(ch);
			writebyte(readbyte());
			writebyte(readbyte());
			writebyte(readbyte());
			continue;
		}

		// so it seams to be a latin1 character
		writebyte(0xC0 | (ch >> 6));
		writebyte(0x80 | (ch & 0x3f));
	}
}

int main(int argc, char **argv)
{
	int i;
	char *buf;
	if (argc <= 1) {
		worker();
	} else {
		for (i = 1; i < argc; i++)
		{
			inputstr = (unsigned char*)argv[i];
			buf = malloc(strlen(argv[i])*2 + 2);
			outputstr = (unsigned char*)buf;
			*outputstr = 0;
			worker();
			if (strcmp(argv[i], buf)) {
				printf("%s: `%s' -> `%s'\n", argv[0], argv[i], buf);
				if (rename(argv[i], buf) < 0)
					fprintf(stderr, "%s: Can't rename `%s' to `%s': %s\n", 
							argv[0], argv[i], buf, strerror(errno));
			}
			free(buf);
		}
	}
	return 0;
}