src/unicoderange.cpp

   1 #include<unicoderange.h>
   2
   3 static unsigned int hex2int(char* s){
   4         int res=0;
   5         int i=0, mul=1;
   6         while(s[i+1]!='\0') i++;
   7
   8         while(i>=0){
   9                 if (s[i] > '9') res += mul * (s[i]-'A'+10);
  10                 else res += mul * (s[i]-'0');
  11                 i--;
  12                 mul*=16;
  13         }
  14         return res;
  15 }
  16
  17 UnicodeRange::UnicodeRange(const gchar* value){
  18         gchar* val = (gchar*) value;
  19         while(val[0] != '\0'){
  20                 if (val[0]=='U' && val[1]=='+'){
  21                         val += add_range(val);
  22                 } else {
  23 //                      g_warning("adding unichar. unichar=%c", g_utf8_get_char(&val[0]));
  24                         this->unichars.push_back(g_utf8_get_char(&val[0]));
  25                         val++;
  26                 }
  27                 //skip spaces or commas
  28                 while(val[0]==' ' || val[0]==',') val++;
  29         }
  30 }
  31
  32 int
  33 UnicodeRange::add_range(gchar* val){
  34                 Urange r;
  35                 //U+
  36                 val+=2;
  37                 int i=0, count=2;
  38                 while(val[i]!='\0' && val[i]!='-' && val[i]!=' ' && val[i]!=',') i++;
  39                 r.start = (gchar*) malloc((i+1)*sizeof(gchar*));
  40                 strncpy(r.start, val, i);
  41                 r.start[i] = '\0';
  42                 val+=i;
  43                 count+=i;
  44                 i=0;
  45                 if (val[0]=='-'){
  46                         val++;
  47                         while(val[i]!='\0' && val[i]!='-' && val[i]!=' ' && val[i]!=',') i++;
  48                         r.end = (gchar*) malloc((i+1)*sizeof(gchar*));
  49                         strncpy(r.end, val, i);
  50                         r.end[i] = '\0';
  51                         val+=i;
  52                         count+=i;
  53                 } else {
  54                         r.end=NULL;
  55                 }
  56 //              g_warning("adding range. from %s to %s", r.start, r.end);
  57                 this->range.push_back(r);
  58                 return count+1;
  59 }
  60
  61 bool UnicodeRange::contains(gchar unicode){
  62         for(unsigned int i=0;i<this->unichars.size();i++){
  63                 if ((gunichar) unicode == this->unichars[i]) return true;
  64         }
  65
  66         unsigned int unival;
  67         unival = g_utf8_get_char (&unicode);
  68         g_warning("unival=%d", unival);
  69         char uni[9] = "00000000";
  70         uni[8]= '\0';
  71         unsigned char val;
  72         for (unsigned int i=7; unival>0; i--){
  73                 val = unival & 0xf;
  74                 unival = unival >> 4;
  75                 if (val < 10) uni[i] = '0' + val;
  76                 else uni[i] = 'A'+ val - 10;
  77         }
  78 //      g_warning("uni=%s", uni);
  79
  80         bool found;
  81         for(unsigned int i=0;i<this->range.size();i++){
  82                 Urange r = this->range[i];
  83                 if (r.end){
  84 //                      g_warning("hex2int: start=%d", hex2int(r.start));
  85 //                      g_warning("hex2int: end=%d", hex2int(r.end));
  86                         if (unival >= hex2int(r.start) && unival <= hex2int(r.end)) return true;
  87                 } else {
  88                         found = true;
  89                         for (int pos=0;pos<8;pos++){
  90                                 if (uni[pos]!='?' && uni[pos]!=r.start[pos]) found = false;
  91                         }
  92                         if (found) return true;
  93                 }
  94         }
  95         return false;
  96 }