dll/nls/normaliz/IdnToUnicode.c

   1
   2 #define WIN32_NO_STATUS
   3 #include <wine/unicode.h>
   4
   5 #define NDEBUG
   6 #include <debug.h>
   7
   8 /* Taken from Wine kernel32/locale.c */
   9
  10 enum {
  11     BASE = 36,
  12     TMIN = 1,
  13     TMAX = 26,
  14     SKEW = 38,
  15     DAMP = 700,
  16     INIT_BIAS = 72,
  17     INIT_N = 128
  18 };
  19
  20 static inline INT adapt(INT delta, INT numpoints, BOOL firsttime)
  21 {
  22     INT k;
  23
  24     delta /= (firsttime ? DAMP : 2);
  25     delta += delta/numpoints;
  26
  27     for(k=0; delta>((BASE-TMIN)*TMAX)/2; k+=BASE)
  28         delta /= BASE-TMIN;
  29     return k+((BASE-TMIN+1)*delta)/(delta+SKEW);
  30 }
  31
  32 static inline unsigned short get_table_entry( const unsigned short *table, WCHAR ch )
  33 {
  34     return table[table[table[ch >> 8] + ((ch >> 4) & 0x0f)] + (ch & 0xf)];
  35 }
  36
  37 INT WINAPI IdnToUnicode(DWORD dwFlags, LPCWSTR lpASCIICharStr, INT cchASCIIChar,
  38                         LPWSTR lpUnicodeCharStr, INT cchUnicodeChar)
  39 {
  40     extern const unsigned short nameprep_char_type[];
  41
  42     INT i, label_start, label_end, out_label, out = 0;
  43     WCHAR ch;
  44
  45     DPRINT("%x %p %d %p %d\n", dwFlags, lpASCIICharStr, cchASCIIChar,
  46            lpUnicodeCharStr, cchUnicodeChar);
  47
  48     for(label_start=0; label_start<cchASCIIChar;) {
  49         INT n = INIT_N, pos = 0, old_pos, w, k, bias = INIT_BIAS, delim=0, digit, t;
  50
  51         out_label = out;
  52         for(i=label_start; i<cchASCIIChar; i++) {
  53             ch = lpASCIICharStr[i];
  54
  55             if(ch>0x7f || (i!=cchASCIIChar-1 && !ch)) {
  56                 SetLastError(ERROR_INVALID_NAME);
  57                 return 0;
  58             }
  59
  60             if(!ch || ch=='.')
  61                 break;
  62             if(ch == '-')
  63                 delim = i;
  64
  65             if((dwFlags&IDN_USE_STD3_ASCII_RULES) == 0)
  66                 continue;
  67             if((ch>='a' && ch<='z') || (ch>='A' && ch<='Z')
  68                     || (ch>='0' && ch<='9') || ch=='-')
  69                 continue;
  70
  71             SetLastError(ERROR_INVALID_NAME);
  72             return 0;
  73         }
  74         label_end = i;
  75         /* last label may be empty */
  76         if(label_start==label_end && ch) {
  77             SetLastError(ERROR_INVALID_NAME);
  78             return 0;
  79         }
  80
  81         if((dwFlags&IDN_USE_STD3_ASCII_RULES) && (lpASCIICharStr[label_start]=='-' ||
  82                     lpASCIICharStr[label_end-1]=='-')) {
  83             SetLastError(ERROR_INVALID_NAME);
  84             return 0;
  85         }
  86         if(label_end-label_start > 63) {
  87             SetLastError(ERROR_INVALID_NAME);
  88             return 0;
  89         }
  90
  91         if(label_end-label_start<4 ||
  92                 tolowerW(lpASCIICharStr[label_start])!='x' ||
  93                 tolowerW(lpASCIICharStr[label_start+1])!='n' ||
  94                 lpASCIICharStr[label_start+2]!='-' || lpASCIICharStr[label_start+3]!='-') {
  95             if(label_end < cchASCIIChar)
  96                 label_end++;
  97
  98             if(!lpUnicodeCharStr) {
  99                 out += label_end-label_start;
 100             }else if(out+label_end-label_start <= cchUnicodeChar) {
 101                 memcpy(lpUnicodeCharStr+out, lpASCIICharStr+label_start,
 102                         (label_end-label_start)*sizeof(WCHAR));
 103                 out += label_end-label_start;
 104             }else {
 105                 SetLastError(ERROR_INSUFFICIENT_BUFFER);
 106                 return 0;
 107             }
 108
 109             label_start = label_end;
 110             continue;
 111         }
 112
 113         if(delim == label_start+3)
 114             delim++;
 115         if(!lpUnicodeCharStr) {
 116             out += delim-label_start-4;
 117         }else if(out+delim-label_start-4 <= cchUnicodeChar) {
 118             memcpy(lpUnicodeCharStr+out, lpASCIICharStr+label_start+4,
 119                     (delim-label_start-4)*sizeof(WCHAR));
 120             out += delim-label_start-4;
 121         }else {
 122             SetLastError(ERROR_INSUFFICIENT_BUFFER);
 123             return 0;
 124         }
 125         if(out != out_label)
 126             delim++;
 127
 128         for(i=delim; i<label_end;) {
 129             old_pos = pos;
 130             w = 1;
 131             for(k=BASE; ; k+=BASE) {
 132                 ch = i<label_end ? tolowerW(lpASCIICharStr[i++]) : 0;
 133                 if((ch<'a' || ch>'z') && (ch<'0' || ch>'9')) {
 134                     SetLastError(ERROR_INVALID_NAME);
 135                     return 0;
 136                 }
 137                 digit = ch<='9' ? ch-'0'+'z'-'a'+1 : ch-'a';
 138                 pos += digit*w;
 139                 t = k<=bias ? TMIN : k>=bias+TMAX ? TMAX : k-bias;
 140                 if(digit < t)
 141                     break;
 142                 w *= BASE-t;
 143             }
 144             bias = adapt(pos-old_pos, out-out_label+1, old_pos==0);
 145             n += pos/(out-out_label+1);
 146             pos %= out-out_label+1;
 147
 148             if((dwFlags&IDN_ALLOW_UNASSIGNED)==0 &&
 149                     get_table_entry(nameprep_char_type, n)==1/*UNASSIGNED*/) {
 150                 SetLastError(ERROR_INVALID_NAME);
 151                 return 0;
 152             }
 153             if(!lpUnicodeCharStr) {
 154                 out++;
 155             }else if(out+1 <= cchASCIIChar) {
 156                 memmove(lpUnicodeCharStr+out_label+pos+1,
 157                         lpUnicodeCharStr+out_label+pos,
 158                         (out-out_label-pos)*sizeof(WCHAR));
 159                 lpUnicodeCharStr[out_label+pos] = n;
 160                 out++;
 161             }else {
 162                 SetLastError(ERROR_INSUFFICIENT_BUFFER);
 163                 return 0;
 164             }
 165             pos++;
 166         }
 167
 168         if(out-out_label > 63) {
 169             SetLastError(ERROR_INVALID_NAME);
 170             return 0;
 171         }
 172
 173         if(label_end < cchASCIIChar) {
 174             if(!lpUnicodeCharStr) {
 175                 out++;
 176             }else if(out+1 <= cchUnicodeChar) {
 177                 lpUnicodeCharStr[out++] = lpASCIICharStr[label_end];
 178             }else {
 179                 SetLastError(ERROR_INSUFFICIENT_BUFFER);
 180                 return 0;
 181             }
 182         }
 183         label_start = label_end+1;
 184     }
 185
 186     return out;
 187 }