#include <stdio.h>
#include <ctype.h>
#include "uniformats.h"

/**
 * Prints a single 32-bit Unicode code point as UCS-4
 * (big-endian 32-bit values).
 */
int putUCS4(unsigned long ucp, FILE *outfp)
{
    putc((ucp >> 24) & 0xFF, outfp);
    putc((ucp >> 16) & 0xFF, outfp);
    putc((ucp >> 8) & 0xFF, outfp);
    putc(ucp & 0xFF, outfp);
    return 4;
}

int getUCS4(unsigned long *ucp, FILE *infp)
{
    unsigned char buf[4];
    if (fread(buf, 1, 4, infp) != 4) return EOF;
    *ucp = buf[0]<<24 | buf[1]<<16 | buf[2]<<8 | buf[3];
    return 4;
}

int gobbleUCS4(FILE *infp)
{
    unsigned char signature[] = {0x00, 0x00, 0xFE, 0xFF};
    int k;
    int i;

    for (i=0; i < 4; ++i)
    {
        k = getc(infp);
        if (k == EOF) {
            while (i)
              ungetc(signature[--i], infp);
            return EOF;
        }
        if (k != signature[i]) {
            ungetc(k, infp);
            while (i && ungetc(signature[--i], infp) != EOF)
              continue;
            return i;
        }
    }

    return 4;
}

int spewUCS4(FILE *outfp)
{
    unsigned char signature[] = {0x00, 0x00, 0xFE, 0xFF};
    int i;
    for (i=0; i < 4; ++i)
      putc(signature[i], outfp);
    return 0;
}


/**
 * Prints a single 32-bit Unicode code point as UCS-2
 * (big-endian 16-bit values).
 */
int putUCS2(unsigned long ucp, FILE *outfp)
{
    putc((ucp >> 8) & 0xFF, outfp);
    putc(ucp & 0xFF, outfp);
    return 2;
}

int getUCS2(unsigned long *ucp, FILE *infp)
{
    unsigned char buf[2];
    if (fread(buf, 1, 2, infp) != 2) return EOF;
    *ucp = buf[0]<<8 | buf[1];
    return 2;
}

int gobbleUCS2(FILE *infp)
{
    unsigned char signature[] = {0xFE, 0xFF};
    int k;
    int i;

    for (i=0; i < 2; ++i)
    {
        k = getc(infp);
        if (k == EOF) {
            while (i)
              ungetc(signature[--i], infp);
            return EOF;
        }
        if (k != signature[i]) {
            ungetc(k, infp);
            while (i && ungetc(signature[--i], infp) != EOF)
              continue;
            return i;
        }
    }

    return 2;
}

int spewUCS2(FILE *outfp)
{
    unsigned char signature[] = {0xFE, 0xFF};
    int i;
    for (i=0; i < 2; ++i)
      putc(signature[i], outfp);
    return 0;
}


/**
 * Prints a single 32-bit Unicode code point as UTF-8.
 */
int putUTF8(unsigned long ucp, FILE *outfp)
{
    if (ucp < 0x80) {
        putc(ucp & 0xFF, outfp);
        return 1;
    }
    else if (ucp < 0x800) {
        putc((ucp >> 6) | 0xC0, outfp);
        putc((ucp & 0x3F) | 0x80, outfp);
        return 2;
    }
    else if (ucp < 0x10000) {
        putc((ucp >> 12) | 0xE0, outfp);
        putc(((ucp >> 6) & 0x3F) | 0x80, outfp);
        putc((ucp & 0x3F) | 0x80, outfp);
        return 3;
    }
    else {
        putc((ucp >> 18) | 0xF0, outfp);
        putc(((ucp >> 12) & 0x3F) | 0x80, outfp);
        putc(((ucp >> 6) & 0x3F) | 0x80, outfp);
        putc((ucp & 0x3F) | 0x80, outfp);
        return 4;
    }
}

int getUTF8(unsigned long *ucp, FILE *infp)
{
    int tmp;
    int nbytes;
    int i;

    tmp = getc(infp);
    if (tmp == EOF) return EOF;

    for (nbytes = 0; (tmp << nbytes) & 0x80; ++nbytes)
      continue;

    if (nbytes == 0) {
        *ucp = tmp;
        return 1;
    }
    else if (nbytes == 1 || nbytes > 6) {
        ungetc(tmp, infp);
        return 0;
    }
    else {
        *ucp = (tmp & (0x7F >> nbytes));
        for (i=0; i < nbytes-1; ++i) {
            if ((tmp = getc(infp)) == EOF)
              return EOF;
            *ucp = (*ucp << 6) | (tmp & 0x3F);
        }
        return nbytes;
    }
}

int gobbleUTF8(FILE *infp)
{
    unsigned char signature[] = {0xEF, 0xBB, 0xBF};
    int k;
    int i;

    for (i=0; i < 3; ++i)
    {
        k = getc(infp);
        if (k == EOF) {
            while (i)
              ungetc(signature[--i], infp);
            return EOF;
        }
        if (k != signature[i]) {
            ungetc(k, infp);
            while (i && ungetc(signature[--i], infp) != EOF)
              continue;
            return i;
        }
    }

    return 3;
}

int spewUTF8(FILE *outfp)
{
    unsigned char signature[] = {0xEF, 0xBB, 0xBF};
    int i;
    for (i=0; i < 3; ++i)
      putc(signature[i], outfp);
    return 0;
}


/**
 * Prints a single 32-bit Unicode code point as ASCII
 * (using the question mark '?' if the value is out of
 * the range 0..255).
 */
int putANSI(unsigned long ucp, FILE *outfp)
{
    if (ucp > 0xFF) putc(0x3F, outfp);
    else putc(ucp, outfp);
    return 1;
}

int getANSI(unsigned long *ucp, FILE *infp)
{
    int k = getc(infp);
    if (k == EOF) return EOF;
    if (k & 0x80) {
        ungetc(k, infp);
        return 0;
    }
    *ucp = k & 0x7F;
    return 1;
}


/**
 * Prints a single 32-bit Unicode code point as ASCII
 * (using the LaTeX Omega syntax ^^^^6f7e if the value
 * is out of the range [10, 13, 32..126]).
 */
int putOmega(unsigned long ucp, FILE *outfp)
{
    if (ucp == 10 || ucp == 13 || (ucp >= 32 && ucp <= 126)) {
        putc(ucp, outfp);
        return 1;
    }
    else {
        return fprintf(outfp, "^^^^%04lx", ucp);
    }
}

int getOmega(unsigned long *ucp, FILE *infp)
{
    static int caret_count = 0;
    int k;

    if (caret_count) {
        --caret_count;
        *ucp = '^';
        return 1;
    }

    if ((k=getc(infp)) == EOF)
      return EOF;
    if (k & 0x80) {
        ungetc(k, infp);
        return 0;
    }
    else if (k != '^') {
        *ucp = k & 0x7F;
        return 1;
    }
    else {
        for (caret_count = 1; caret_count < 4; ++caret_count) {
            if ((k=getc(infp)) == EOF)
              return EOF;
            if (k != '^') {
                ungetc(k, infp);
                *ucp = '^';
                return 1;
            }
        }
        if ((k=getc(infp)) == EOF)
          return EOF;
        ungetc(k, infp);
        if (!isxdigit(k)) {
            --caret_count;
            *ucp = '^';
            return 1;
        }
        else {
            if (fscanf(infp, "%4lx", ucp) != 1)
              return 0;
            caret_count = 0;
            return 8;
        }
    }                           
}