2 punycode.c from RFC 3492
\r
3 http://www.nicemice.net/idn/
\r
5 http://www.nicemice.net/amc/
\r
7 This is ANSI C code (C89) implementing Punycode (RFC 3492).
\r
12 /************************************************************/
\r
13 /* Public interface (would normally go in its own .h file): */
\r
17 enum punycode_status {
\r
19 punycode_bad_input, /* Input is invalid. */
\r
20 punycode_big_output, /* Output would exceed the space provided. */
\r
21 punycode_overflow /* Input needs wider integers to process. */
\r
24 #if UINT_MAX >= (1 << 26) - 1
\r
25 typedef unsigned int punycode_uint;
\r
27 typedef unsigned long punycode_uint;
\r
30 enum punycode_status punycode_encode(
\r
31 punycode_uint input_length,
\r
32 const punycode_uint input[],
\r
33 const unsigned char case_flags[],
\r
34 punycode_uint *output_length,
\r
37 /* punycode_encode() converts Unicode to Punycode. The input */
\r
38 /* is represented as an array of Unicode code points (not code */
\r
39 /* units; surrogate pairs are not allowed), and the output */
\r
40 /* will be represented as an array of ASCII code points. The */
\r
41 /* output string is *not* null-terminated; it will contain */
\r
42 /* zeros if and only if the input contains zeros. (Of course */
\r
43 /* the caller can leave room for a terminator and add one if */
\r
44 /* needed.) The input_length is the number of code points in */
\r
45 /* the input. The output_length is an in/out argument: the */
\r
46 /* caller passes in the maximum number of code points that it */
\r
47 /* can receive, and on successful return it will contain the */
\r
48 /* number of code points actually output. The case_flags array */
\r
49 /* holds input_length boolean values, where nonzero suggests that */
\r
50 /* the corresponding Unicode character be forced to uppercase */
\r
51 /* after being decoded (if possible), and zero suggests that */
\r
52 /* it be forced to lowercase (if possible). ASCII code points */
\r
53 /* are encoded literally, except that ASCII letters are forced */
\r
54 /* to uppercase or lowercase according to the corresponding */
\r
55 /* uppercase flags. If case_flags is a null pointer then ASCII */
\r
56 /* letters are left as they are, and other code points are */
\r
57 /* treated as if their uppercase flags were zero. The return */
\r
58 /* value can be any of the punycode_status values defined above */
\r
59 /* except punycode_bad_input; if not punycode_success, then */
\r
60 /* output_size and output might contain garbage. */
\r
62 enum punycode_status punycode_decode(
\r
63 punycode_uint input_length,
\r
65 punycode_uint *output_length,
\r
66 punycode_uint output[],
\r
67 unsigned char case_flags[] );
\r
69 /* punycode_decode() converts Punycode to Unicode. The input is */
\r
70 /* represented as an array of ASCII code points, and the output */
\r
71 /* will be represented as an array of Unicode code points. The */
\r
72 /* input_length is the number of code points in the input. The */
\r
73 /* output_length is an in/out argument: the caller passes in */
\r
74 /* the maximum number of code points that it can receive, and */
\r
75 /* on successful return it will contain the actual number of */
\r
76 /* code points output. The case_flags array needs room for at */
\r
77 /* least output_length values, or it can be a null pointer if the */
\r
78 /* case information is not needed. A nonzero flag suggests that */
\r
79 /* the corresponding Unicode character be forced to uppercase */
\r
80 /* by the caller (if possible), while zero suggests that it be */
\r
81 /* forced to lowercase (if possible). ASCII code points are */
\r
82 /* output already in the proper case, but their flags will be set */
\r
83 /* appropriately so that applying the flags would be harmless. */
\r
84 /* The return value can be any of the punycode_status values */
\r
85 /* defined above; if not punycode_success, then output_length, */
\r
86 /* output, and case_flags might contain garbage. On success, the */
\r
87 /* decoder will never need to write an output_length greater than */
\r
88 /* input_length, because of how the encoding is defined. */
\r
90 /**********************************************************/
\r
91 /* Implementation (would normally go in its own .c file): */
\r
95 /*** Bootstring parameters for Punycode ***/
\r
97 enum { base = 36, tmin = 1, tmax = 26, skew = 38, damp = 700,
\r
98 initial_bias = 72, initial_n = 0x80, delimiter = 0x2D };
\r
100 /* basic(cp) tests whether cp is a basic code point: */
\r
101 #define basic(cp) ((punycode_uint)(cp) < 0x80)
\r
103 /* delim(cp) tests whether cp is a delimiter: */
\r
104 #define delim(cp) ((cp) == delimiter)
\r
106 /* decode_digit(cp) returns the numeric value of a basic code */
\r
107 /* point (for use in representing integers) in the range 0 to */
\r
108 /* base-1, or base if cp is does not represent a value. */
\r
110 static punycode_uint decode_digit(punycode_uint cp)
\r
112 return cp - 48 < 10 ? cp - 22 : cp - 65 < 26 ? cp - 65 :
\r
113 cp - 97 < 26 ? cp - 97 : base;
\r
116 /* encode_digit(d,flag) returns the basic code point whose value */
\r
117 /* (when used for representing integers) is d, which needs to be in */
\r
118 /* the range 0 to base-1. The lowercase form is used unless flag is */
\r
119 /* nonzero, in which case the uppercase form is used. The behavior */
\r
120 /* is undefined if flag is nonzero and digit d has no uppercase form. */
\r
122 static char encode_digit(punycode_uint d, int flag)
\r
124 return d + 22 + 75 * (d < 26) - ((flag != 0) << 5);
\r
125 /* 0..25 map to ASCII a..z or A..Z */
\r
126 /* 26..35 map to ASCII 0..9 */
\r
129 /* flagged(bcp) tests whether a basic code point is flagged */
\r
130 /* (uppercase). The behavior is undefined if bcp is not a */
\r
131 /* basic code point. */
\r
133 #define flagged(bcp) ((punycode_uint)(bcp) - 65 < 26)
\r
135 /* encode_basic(bcp,flag) forces a basic code point to lowercase */
\r
136 /* if flag is zero, uppercase if flag is nonzero, and returns */
\r
137 /* the resulting code point. The code point is unchanged if it */
\r
138 /* is caseless. The behavior is undefined if bcp is not a basic */
\r
141 static char encode_basic(punycode_uint bcp, int flag)
\r
143 bcp -= (bcp - 97 < 26) << 5;
\r
144 return bcp + ((!flag && (bcp - 65 < 26)) << 5);
\r
147 /*** Platform-specific constants ***/
\r
149 /* maxint is the maximum value of a punycode_uint variable: */
\r
150 static const punycode_uint maxint = -1;
\r
151 /* Because maxint is unsigned, -1 becomes the maximum value. */
\r
153 /*** Bias adaptation function ***/
\r
155 static punycode_uint adapt(
\r
156 punycode_uint delta, punycode_uint numpoints, int firsttime )
\r
160 delta = firsttime ? delta / damp : delta >> 1;
\r
161 /* delta >> 1 is a faster way of doing delta / 2 */
\r
162 delta += delta / numpoints;
\r
164 for (k = 0; delta > ((base - tmin) * tmax) / 2; k += base) {
\r
165 delta /= base - tmin;
\r
168 return k + (base - tmin + 1) * delta / (delta + skew);
\r
171 /*** Main encode function ***/
\r
173 enum punycode_status punycode_encode(
\r
174 punycode_uint input_length,
\r
175 const punycode_uint input[],
\r
176 const unsigned char case_flags[],
\r
177 punycode_uint *output_length,
\r
180 punycode_uint n, delta, h, b, out, max_out, bias, j, m, q, k, t;
\r
182 /* Initialize the state: */
\r
186 max_out = *output_length;
\r
187 bias = initial_bias;
\r
189 /* Handle the basic code points: */
\r
191 for (j = 0; j < input_length; ++j) {
\r
192 if (basic(input[j])) {
\r
193 if (max_out - out < 2) return punycode_big_output;
\r
195 case_flags ? encode_basic(input[j], case_flags[j]) : input[j];
\r
197 /* else if (input[j] < n) return punycode_bad_input; */
\r
198 /* (not needed for Punycode with unsigned code points) */
\r
203 /* h is the number of code points that have been handled, b is the */
\r
204 /* number of basic code points, and out is the number of characters */
\r
205 /* that have been output. */
\r
207 if (b > 0) output[out++] = delimiter;
\r
209 /* Main encoding loop: */
\r
211 while (h < input_length) {
\r
212 /* All non-basic code points < n have been */
\r
213 /* handled already. Find the next larger one: */
\r
215 for (m = maxint, j = 0; j < input_length; ++j) {
\r
216 /* if (basic(input[j])) continue; */
\r
217 /* (not needed for Punycode) */
\r
218 if (input[j] >= n && input[j] < m) m = input[j];
\r
221 /* Increase delta enough to advance the decoder's */
\r
222 /* <n,i> state to <m,0>, but guard against overflow: */
\r
224 if (m - n > (maxint - delta) / (h + 1)) return punycode_overflow;
\r
225 delta += (m - n) * (h + 1);
\r
228 for (j = 0; j < input_length; ++j) {
\r
229 /* Punycode does not need to check whether input[j] is basic: */
\r
230 if (input[j] < n /* || basic(input[j]) */ ) {
\r
231 if (++delta == 0) return punycode_overflow;
\r
234 if (input[j] == n) {
\r
235 /* Represent delta as a generalized variable-length integer: */
\r
237 for (q = delta, k = base; ; k += base) {
\r
238 if (out >= max_out) return punycode_big_output;
\r
239 t = k <= bias /* + tmin */ ? tmin : /* +tmin not needed */
\r
240 k >= bias + tmax ? tmax : k - bias;
\r
242 output[out++] = encode_digit(t + (q - t) % (base - t), 0);
\r
243 q = (q - t) / (base - t);
\r
246 output[out++] = encode_digit(q, case_flags && case_flags[j]);
\r
247 bias = adapt(delta, h + 1, h == b);
\r
256 *output_length = out;
\r
257 return punycode_success;
\r
260 /*** Main decode function ***/
\r
262 enum punycode_status punycode_decode(
\r
263 punycode_uint input_length,
\r
264 const char input[],
\r
265 punycode_uint *output_length,
\r
266 punycode_uint output[],
\r
267 unsigned char case_flags[] )
\r
269 punycode_uint n, out, i, max_out, bias,
\r
270 b, j, in, oldi, w, k, digit, t;
\r
272 /* Initialize the state: */
\r
276 max_out = *output_length;
\r
277 bias = initial_bias;
\r
279 /* Handle the basic code points: Let b be the number of input code */
\r
280 /* points before the last delimiter, or 0 if there is none, then */
\r
281 /* copy the first b code points to the output. */
\r
283 for (b = j = 0; j < input_length; ++j) if (delim(input[j])) b = j;
\r
284 if (b > max_out) return punycode_big_output;
\r
286 for (j = 0; j < b; ++j) {
\r
287 if (case_flags) case_flags[out] = flagged(input[j]);
\r
288 if (!basic(input[j])) return punycode_bad_input;
\r
289 output[out++] = input[j];
\r
292 /* Main decoding loop: Start just after the last delimiter if any */
\r
293 /* basic code points were copied; start at the beginning otherwise. */
\r
295 for (in = b > 0 ? b + 1 : 0; in < input_length; ++out) {
\r
297 /* in is the index of the next character to be consumed, and */
\r
298 /* out is the number of code points in the output array. */
\r
300 /* Decode a generalized variable-length integer into delta, */
\r
301 /* which gets added to i. The overflow checking is easier */
\r
302 /* if we increase i as we go, then subtract off its starting */
\r
303 /* value at the end to obtain delta. */
\r
305 for (oldi = i, w = 1, k = base; ; k += base) {
\r
306 if (in >= input_length) return punycode_bad_input;
\r
307 digit = decode_digit(input[in++]);
\r
308 if (digit >= base) return punycode_bad_input;
\r
309 if (digit > (maxint - i) / w) return punycode_overflow;
\r
311 t = k <= bias /* + tmin */ ? tmin : /* +tmin not needed */
\r
312 k >= bias + tmax ? tmax : k - bias;
\r
313 if (digit < t) break;
\r
314 if (w > maxint / (base - t)) return punycode_overflow;
\r
318 bias = adapt(i - oldi, out + 1, oldi == 0);
\r
320 /* i was supposed to wrap around from out+1 to 0, */
\r
321 /* incrementing n each time, so we'll fix that now: */
\r
323 if (i / (out + 1) > maxint - n) return punycode_overflow;
\r
324 n += i / (out + 1);
\r
327 /* Insert n at position i of the output: */
\r
329 /* not needed for Punycode: */
\r
330 /* if (decode_digit(n) <= base) return punycode_invalid_input; */
\r
331 if (out >= max_out) return punycode_big_output;
\r
334 memmove(case_flags + i + 1, case_flags + i, out - i);
\r
335 /* Case of last character determines uppercase flag: */
\r
336 case_flags[i] = flagged(input[in - 1]);
\r
339 memmove(output + i + 1, output + i, (out - i) * sizeof *output);
\r
343 *output_length = out;
\r
344 return punycode_success;
\r