michael@0: /*
michael@0: punycode.c from RFC 3492
michael@0: http://www.nicemice.net/idn/
michael@0: Adam M. Costello
michael@0: http://www.nicemice.net/amc/
michael@0: 
michael@0: This is ANSI C code (C89) implementing Punycode (RFC 3492).
michael@0: 
michael@0: 
michael@0: C. Disclaimer and license
michael@0: 
michael@0:     Regarding this entire document or any portion of it (including
michael@0:     the pseudocode and C code), the author makes no guarantees and
michael@0:     is not responsible for any damage resulting from its use.  The
michael@0:     author grants irrevocable permission to anyone to use, modify,
michael@0:     and distribute it in any way that does not diminish the rights
michael@0:     of anyone else to use, modify, and distribute it, provided that
michael@0:     redistributed derivative works do not contain misleading author or
michael@0:     version information.  Derivative works need not be licensed under
michael@0:     similar terms.
michael@0: */
michael@0: 
michael@0: #include "punycode.h"
michael@0: 
michael@0: /**********************************************************/
michael@0: /* Implementation (would normally go in its own .c file): */
michael@0: 
michael@0: #include <string.h>
michael@0: 
michael@0: /*** Bootstring parameters for Punycode ***/
michael@0: 
michael@0: enum { base = 36, tmin = 1, tmax = 26, skew = 38, damp = 700,
michael@0:        initial_bias = 72, initial_n = 0x80, delimiter = 0x2D };
michael@0: 
michael@0: /* basic(cp) tests whether cp is a basic code point: */
michael@0: #define basic(cp) ((punycode_uint)(cp) < 0x80)
michael@0: 
michael@0: /* delim(cp) tests whether cp is a delimiter: */
michael@0: #define delim(cp) ((cp) == delimiter)
michael@0: 
michael@0: /* decode_digit(cp) returns the numeric value of a basic code */
michael@0: /* point (for use in representing integers) in the range 0 to */
michael@0: /* base-1, or base if cp is does not represent a value.       */
michael@0: 
michael@0: static punycode_uint decode_digit(punycode_uint cp)
michael@0: {
michael@0:   return  cp - 48 < 10 ? cp - 22 :  cp - 65 < 26 ? cp - 65 :
michael@0:           cp - 97 < 26 ? cp - 97 :  base;
michael@0: }
michael@0: 
michael@0: /* encode_digit(d,flag) returns the basic code point whose value      */
michael@0: /* (when used for representing integers) is d, which needs to be in   */
michael@0: /* the range 0 to base-1.  The lowercase form is used unless flag is  */
michael@0: /* nonzero, in which case the uppercase form is used.  The behavior   */
michael@0: /* is undefined if flag is nonzero and digit d has no uppercase form. */
michael@0: 
michael@0: static char encode_digit(punycode_uint d, int flag)
michael@0: {
michael@0:   return d + 22 + 75 * (d < 26) - ((flag != 0) << 5);
michael@0:   /*  0..25 map to ASCII a..z or A..Z */
michael@0:   /* 26..35 map to ASCII 0..9         */
michael@0: }
michael@0: 
michael@0: /* flagged(bcp) tests whether a basic code point is flagged */
michael@0: /* (uppercase).  The behavior is undefined if bcp is not a  */
michael@0: /* basic code point.                                        */
michael@0: 
michael@0: #define flagged(bcp) ((punycode_uint)(bcp) - 65 < 26)
michael@0: 
michael@0: /* encode_basic(bcp,flag) forces a basic code point to lowercase */
michael@0: /* if flag is zero, uppercase if flag is nonzero, and returns    */
michael@0: /* the resulting code point.  The code point is unchanged if it  */
michael@0: /* is caseless.  The behavior is undefined if bcp is not a basic */
michael@0: /* code point.                                                   */
michael@0: 
michael@0: static char encode_basic(punycode_uint bcp, int flag)
michael@0: {
michael@0:   bcp -= (bcp - 97 < 26) << 5;
michael@0:   return bcp + ((!flag && (bcp - 65 < 26)) << 5);
michael@0: }
michael@0: 
michael@0: /*** Platform-specific constants ***/
michael@0: 
michael@0: /* maxint is the maximum value of a punycode_uint variable: */
michael@0: static const punycode_uint maxint = (punycode_uint) -1;
michael@0: /* Because maxint is unsigned, -1 becomes the maximum value. */
michael@0: 
michael@0: /*** Bias adaptation function ***/
michael@0: 
michael@0: static punycode_uint adapt(
michael@0:   punycode_uint delta, punycode_uint numpoints, int firsttime )
michael@0: {
michael@0:   punycode_uint k;
michael@0: 
michael@0:   delta = firsttime ? delta / damp : delta >> 1;
michael@0:   /* delta >> 1 is a faster way of doing delta / 2 */
michael@0:   delta += delta / numpoints;
michael@0: 
michael@0:   for (k = 0;  delta > ((base - tmin) * tmax) / 2;  k += base) {
michael@0:     delta /= base - tmin;
michael@0:   }
michael@0: 
michael@0:   return k + (base - tmin + 1) * delta / (delta + skew);
michael@0: }
michael@0: 
michael@0: /*** Main encode function ***/
michael@0: 
michael@0: enum punycode_status punycode_encode(
michael@0:   punycode_uint input_length,
michael@0:   const punycode_uint input[],
michael@0:   const unsigned char case_flags[],
michael@0:   punycode_uint *output_length,
michael@0:   char output[] )
michael@0: {
michael@0:   punycode_uint n, delta, h, b, out, max_out, bias, j, m, q, k, t;
michael@0: 
michael@0:   /* Initialize the state: */
michael@0: 
michael@0:   n = initial_n;
michael@0:   delta = out = 0;
michael@0:   max_out = *output_length;
michael@0:   bias = initial_bias;
michael@0: 
michael@0:   /* Handle the basic code points: */
michael@0: 
michael@0:   for (j = 0;  j < input_length;  ++j) {
michael@0:     if (basic(input[j])) {
michael@0:       if (max_out - out < 2) return punycode_big_output;
michael@0:       output[out++] =
michael@0:         case_flags ? encode_basic(input[j], case_flags[j]) : (char)input[j];
michael@0:     }
michael@0:     /* else if (input[j] < n) return punycode_bad_input; */
michael@0:     /* (not needed for Punycode with unsigned code points) */
michael@0:   }
michael@0: 
michael@0:   h = b = out;
michael@0: 
michael@0:   /* h is the number of code points that have been handled, b is the  */
michael@0:   /* number of basic code points, and out is the number of characters */
michael@0:   /* that have been output.                                           */
michael@0: 
michael@0:   if (b > 0) output[out++] = delimiter;
michael@0: 
michael@0:   /* Main encoding loop: */
michael@0: 
michael@0:   while (h < input_length) {
michael@0:     /* All non-basic code points < n have been     */
michael@0:     /* handled already.  Find the next larger one: */
michael@0: 
michael@0:     for (m = maxint, j = 0;  j < input_length;  ++j) {
michael@0:       /* if (basic(input[j])) continue; */
michael@0:       /* (not needed for Punycode) */
michael@0:       if (input[j] >= n && input[j] < m) m = input[j];
michael@0:     }
michael@0: 
michael@0:     /* Increase delta enough to advance the decoder's    */
michael@0:     /* <n,i> state to <m,0>, but guard against overflow: */
michael@0: 
michael@0:     if (m - n > (maxint - delta) / (h + 1)) return punycode_overflow;
michael@0:     delta += (m - n) * (h + 1);
michael@0:     n = m;
michael@0: 
michael@0:     for (j = 0;  j < input_length;  ++j) {
michael@0:       /* Punycode does not need to check whether input[j] is basic: */
michael@0:       if (input[j] < n /* || basic(input[j]) */ ) {
michael@0:         if (++delta == 0) return punycode_overflow;
michael@0:       }
michael@0: 
michael@0:       if (input[j] == n) {
michael@0:         /* Represent delta as a generalized variable-length integer: */
michael@0: 
michael@0:         for (q = delta, k = base;  ;  k += base) {
michael@0:           if (out >= max_out) return punycode_big_output;
michael@0:           t = k <= bias /* + tmin */ ? tmin :     /* +tmin not needed */
michael@0:               k >= bias + tmax ? tmax : k - bias;
michael@0:           if (q < t) break;
michael@0:           output[out++] = encode_digit(t + (q - t) % (base - t), 0);
michael@0:           q = (q - t) / (base - t);
michael@0:         }
michael@0: 
michael@0:         output[out++] = encode_digit(q, case_flags && case_flags[j]);
michael@0:         bias = adapt(delta, h + 1, h == b);
michael@0:         delta = 0;
michael@0:         ++h;
michael@0:       }
michael@0:     }
michael@0: 
michael@0:     ++delta, ++n;
michael@0:   }
michael@0: 
michael@0:   *output_length = out;
michael@0:   return punycode_success;
michael@0: }
michael@0: 
michael@0: /*** Main decode function ***/
michael@0: 
michael@0: enum punycode_status punycode_decode(
michael@0:   punycode_uint input_length,
michael@0:   const char input[],
michael@0:   punycode_uint *output_length,
michael@0:   punycode_uint output[],
michael@0:   unsigned char case_flags[] )
michael@0: {
michael@0:   punycode_uint n, out, i, max_out, bias,
michael@0:                  b, j, in, oldi, w, k, digit, t;
michael@0: 
michael@0:   if (!input_length) {
michael@0:     return punycode_bad_input;
michael@0:   }
michael@0: 
michael@0:   /* Initialize the state: */
michael@0: 
michael@0:   n = initial_n;
michael@0:   out = i = 0;
michael@0:   max_out = *output_length;
michael@0:   bias = initial_bias;
michael@0: 
michael@0:   /* Handle the basic code points:  Let b be the number of input code */
michael@0:   /* points before the last delimiter, or 0 if there is none, then    */
michael@0:   /* copy the first b code points to the output.                      */
michael@0: 
michael@0:   for (b = 0, j = input_length - 1 ;  j > 0;  --j) {
michael@0:     if (delim(input[j])) {
michael@0:       b = j;
michael@0:       break;
michael@0:     }
michael@0:   }
michael@0:   if (b > max_out) return punycode_big_output;
michael@0: 
michael@0:   for (j = 0;  j < b;  ++j) {
michael@0:     if (case_flags) case_flags[out] = flagged(input[j]);
michael@0:     if (!basic(input[j])) return punycode_bad_input;
michael@0:     output[out++] = input[j];
michael@0:   }
michael@0: 
michael@0:   /* Main decoding loop:  Start just after the last delimiter if any  */
michael@0:   /* basic code points were copied; start at the beginning otherwise. */
michael@0: 
michael@0:   for (in = b > 0 ? b + 1 : 0;  in < input_length;  ++out) {
michael@0: 
michael@0:     /* in is the index of the next character to be consumed, and */
michael@0:     /* out is the number of code points in the output array.     */
michael@0: 
michael@0:     /* Decode a generalized variable-length integer into delta,  */
michael@0:     /* which gets added to i.  The overflow checking is easier   */
michael@0:     /* if we increase i as we go, then subtract off its starting */
michael@0:     /* value at the end to obtain delta.                         */
michael@0: 
michael@0:     for (oldi = i, w = 1, k = base;  ;  k += base) {
michael@0:       if (in >= input_length) return punycode_bad_input;
michael@0:       digit = decode_digit(input[in++]);
michael@0:       if (digit >= base) return punycode_bad_input;
michael@0:       if (digit > (maxint - i) / w) return punycode_overflow;
michael@0:       i += digit * w;
michael@0:       t = k <= bias /* + tmin */ ? tmin :     /* +tmin not needed */
michael@0:           k >= bias + tmax ? tmax : k - bias;
michael@0:       if (digit < t) break;
michael@0:       if (w > maxint / (base - t)) return punycode_overflow;
michael@0:       w *= (base - t);
michael@0:     }
michael@0: 
michael@0:     bias = adapt(i - oldi, out + 1, oldi == 0);
michael@0: 
michael@0:     /* i was supposed to wrap around from out+1 to 0,   */
michael@0:     /* incrementing n each time, so we'll fix that now: */
michael@0: 
michael@0:     if (i / (out + 1) > maxint - n) return punycode_overflow;
michael@0:     n += i / (out + 1);
michael@0:     i %= (out + 1);
michael@0: 
michael@0:     /* Insert n at position i of the output: */
michael@0: 
michael@0:     /* not needed for Punycode: */
michael@0:     /* if (decode_digit(n) <= base) return punycode_invalid_input; */
michael@0:     if (out >= max_out) return punycode_big_output;
michael@0: 
michael@0:     if (case_flags) {
michael@0:       memmove(case_flags + i + 1, case_flags + i, out - i);
michael@0:       /* Case of last character determines uppercase flag: */
michael@0:       case_flags[i] = flagged(input[in - 1]);
michael@0:     }
michael@0: 
michael@0:     memmove(output + i + 1, output + i, (out - i) * sizeof *output);
michael@0:     output[i++] = n;
michael@0:   }
michael@0: 
michael@0:   *output_length = out;
michael@0:   return punycode_success;
michael@0: }