src/unicoderange.cpp

   1 #include "unicoderange.h"
   2
   3 #include <malloc.h>
   4 #include <string.h>
   5
   6 static unsigned int hex2int(char* s){
   7         int res=0;
   8         int i=0, mul=1;
   9         while(s[i+1]!='\0') i++;
  10
  11         while(i>=0){
  12                 if (s[i] > '9') res += mul * (s[i]-'A'+10);
  13                 else res += mul * (s[i]-'0');
  14                 i--;
  15                 mul*=16;
  16         }
  17         return res;
  18 }
  19
  20 UnicodeRange::UnicodeRange(const gchar* value){
  21         gchar* val = (gchar*) value;
  22         while(val[0] != '\0'){
  23                 if (val[0]=='U' && val[1]=='+'){
  24                         val += add_range(val);
  25                 } else {
  26 //                      g_warning("adding unichar. unichar=%c", g_utf8_get_char(&val[0]));
  27                         this->unichars.push_back(g_utf8_get_char(&val[0]));
  28                         val++;
  29                 }
  30                 //skip spaces or commas
  31                 while(val[0]==' ' || val[0]==',') val++;
  32         }
  33 }
  34
  35 int
  36 UnicodeRange::add_range(gchar* val){
  37                 Urange r;
  38                 //U+
  39                 val+=2;
  40                 int i=0, count=2;
  41                 while(val[i]!='\0' && val[i]!='-' && val[i]!=' ' && val[i]!=',') i++;
  42                 r.start = (gchar*) malloc((i+1)*sizeof(gchar*));
  43                 strncpy(r.start, val, i);
  44                 r.start[i] = '\0';
  45                 val+=i;
  46                 count+=i;
  47                 i=0;
  48                 if (val[0]=='-'){
  49                         val++;
  50                         while(val[i]!='\0' && val[i]!='-' && val[i]!=' ' && val[i]!=',') i++;
  51                         r.end = (gchar*) malloc((i+1)*sizeof(gchar*));
  52                         strncpy(r.end, val, i);
  53                         r.end[i] = '\0';
  54                         val+=i;
  55                         count+=i;
  56                 } else {
  57                         r.end=NULL;
  58                 }
  59 //              g_warning("adding range. from %s to %s", r.start, r.end);
  60                 this->range.push_back(r);
  61                 return count+1;
  62 }
  63
  64 bool UnicodeRange::contains(gchar unicode){
  65         for(unsigned int i=0;i<this->unichars.size();i++){
  66                 if ((gunichar) unicode == this->unichars[i]) return true;
  67         }
  68
  69         unsigned int unival;
  70         unival = g_utf8_get_char (&unicode);
  71         g_warning("unival=%d", unival);
  72         char uni[9] = "00000000";
  73         uni[8]= '\0';
  74         unsigned char val;
  75         for (unsigned int i=7; unival>0; i--){
  76                 val = unival & 0xf;
  77                 unival = unival >> 4;
  78                 if (val < 10) uni[i] = '0' + val;
  79                 else uni[i] = 'A'+ val - 10;
  80         }
  81 //      g_warning("uni=%s", uni);
  82
  83         bool found;
  84         for(unsigned int i=0;i<this->range.size();i++){
  85                 Urange r = this->range[i];
  86                 if (r.end){
  87 //                      g_warning("hex2int: start=%d", hex2int(r.start));
  88 //                      g_warning("hex2int: end=%d", hex2int(r.end));
  89                         if (unival >= hex2int(r.start) && unival <= hex2int(r.end)) return true;
  90                 } else {
  91                         found = true;
  92                         for (int pos=0;pos<8;pos++){
  93                                 if (uni[pos]!='?' && uni[pos]!=r.start[pos]) found = false;
  94                         }
  95                         if (found) return true;
  96                 }
  97         }
  98         return false;
  99 }