/**
 * Convert from Windows-1252 to UTF-8
 * Note that Windows-1252 seems to be a superset of Latin-1 (ISO 8859-1).
 *
 * Usage: <code>2utf8 [<var>file</var> ...]</code>
 *
 * BUG: There are a few chars undefined in Windows-1252
 *      which will map to a question mark.
 */

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

static int unicode[32];

/**
 * Get the next byte and convert it to unicode.
 * This is the "win" part of the name.
 * What is returned is actually a unicode code point.
 * BUG? Should this abort when it gets a byte not in Windows-1252?
 * 16 bits is sufficient for the task at hand.
 * All the code points for the characters in Windows-1252
 * are below 0x3000.
 */
int getNext( FILE *infp ) {
    int c = fgetc( infp );
    if ( EOF == c ) {
        return EOF;
    }
    if ( 0x7f < c && c < 0xa0 ) {
        return unicode[c - 128];
    }
    return c;
}

/**
 * Put the unicode character out encoded in UTF-8.
 * This is the "utf" part of the name.
 */
void putBytes( int theChar ) {
    int firstByte, secondByte, thirdByte;
    if ( 0xfff < theChar ) { /* then we need three bytes */
        thirdByte = 0x80 | ( theChar & 0x3f );
        theChar >>= 6;
        secondByte = 0x80 | ( theChar & 0x3f );
        theChar >>= 6;
        firstByte = 0xe0 | theChar;
        putchar(firstByte);
        putchar(secondByte);
        putchar(thirdByte);
    }
    else if ( 0x7f < theChar ) { /* we only need two */
        secondByte = 0x80 | ( theChar & 0x3f );
        theChar >>= 6;
        firstByte = 0xc0 | theChar;
        putchar(firstByte);
        putchar(secondByte);
    }
    else {
        putchar( theChar );
    }
}

void doFiles( int argc, char **argv ) {

    FILE *infp;
    char *infile;
    int c;

    while ( --argc > 0 ) {
        infile = *++argv;

        if ( 0 == strcmp( infile, "-" ) ) {
            infp = stdin;
        }
        /* Must open input in binary mode so we can catch chars with high bit set. */
        else if ( ( infp = fopen( infile, "rb" ) ) == NULL ) {
            fprintf( stderr, "Unable to open input file \"%s\".\n", infile );
            continue;
        }
        while ( ( c = getNext( infp ) ) != EOF ) {
            putBytes( c );
        }
    }

}

int main( int argc, char **argv ) {

    char * oneName[2];

    unicode[0x00] = 0x20ac;
    unicode[0x01] = '?';
    unicode[0x02] = 0x201a;
    unicode[0x03] = 0x0192;
    unicode[0x04] = 0x201e;
    unicode[0x05] = 0x2026;
    unicode[0x06] = 0x2020;
    unicode[0x07] = 0x2021;
    unicode[0x08] = 0x02c6;
    unicode[0x09] = 0x2030;
    unicode[0x0a] = 0x0160;
    unicode[0x0b] = 0x2039;
    unicode[0x0c] = 0x0152;
    unicode[0x0d] = '?';
    unicode[0x0e] = 0x017d;
    unicode[0x0f] = '?';
    unicode[0x10] = '?';
    unicode[0x11] = 0x2018;
    unicode[0x12] = 0x2019;
    unicode[0x13] = 0x201c;
    unicode[0x14] = 0x201d;
    unicode[0x15] = 0x2022;
    unicode[0x16] = 0x2013;
    unicode[0x17] = 0x2014;
    unicode[0x18] = 0x02dc;
    unicode[0x19] = 0x2122;
    unicode[0x1a] = 0x0161;
    unicode[0x1b] = 0x203a;
    unicode[0x1c] = 0x0153;
    unicode[0x1d] = '?';
    unicode[0x1e] = 0x017e;
    unicode[0x1f] = 0x0178;

    if ( argc < 2 ) {
        oneName[0] = argv[0];
        oneName[1] = "-";
        doFiles( 2, oneName );
    }
    else {
        doFiles( argc, argv );
    }

    return 0;
}