fixed class.c

class.c

@@ -12,186 +12,106 @@ int testBit(char *bits, int n)
     return bits[n/8] & (1 << n%8);
 }
 
-int isEscapedChar(char char1, char char2){
-    if( char1=='\n'){
-        return 10;
-    }
-    if( char1=='\a'){
-        return 7;
-    }
-    if( char1=='\b'){
-        return 8;
-    }
-    if( char1=='\e'){
-        return 27;
-    }
-    if( char1=='\f'){
-        return 12;
-    }
-    if( char1=='\r'){
-        return 13;
-    }
-    else if( char1=='\t'){
-        return 9;
-    }
-    if( char1=='\v'){
-        return 1;
-    }
-    else if( char1=='\\' && char2=='-'){
-        return 45;
-    }
-    else if ( char1=='\\'){
-        return 92;
-    }
-    else return ; //returning nothing for the moment, returning 0 printed weird character, no idea what to do
-}
-
-int hexaToDecimal(char hex1,char hex2){
-    char hex[2];
-    hex[0]=hex1;hex[1]=hex2;
-    long long decimal = 0, base = 1;
-    int i = 0, value, length;
-
-    length = 2;
-    for(i = length--; i >= 0; i--)
-    {
-        if(hex[i] >= '0' && hex[i] <= '9')
-        {
-            decimal += (hex[i] - 48) * base;
-            base *= 16;
-        }
-        else if(hex[i] >= 'A' && hex[i] <= 'F')
-        {
-            decimal += (hex[i] - 55) * base;
-            base *= 16;
-        }
-        else if(hex[i] >= 'a' && hex[i] <= 'f')
-        {
-            decimal += (hex[i] - 87) * base;
-            base *= 16;
-        }
+int digitValue(int c)
+{
+    switch (c) {
+      case '0'...'9':	return c - '0';
+      case 'A'...'Z':	return c - 'A' + 10;
+      case 'a'...'z':	return c - 'a' + 10;
     }
-    return decimal;
+    return 666;
 }
 
-int octalToDecimal(int octalNumber)
+int isRadix(int r, int c)
 {
-    int decimalNumber = 0, i = 0;
-
-    while(octalNumber != 0)
-    {
-        decimalNumber += (octalNumber%10) * pow(8,i);
-        ++i;
-        octalNumber/=10;
-    }
-
-    i = 1;
-
-    return decimalNumber;
+    return digitValue(c) < r;
 }
 
-int isByte(char char1, char char2, char char3, char char4){
-    if( char1=='\\' && char2=='x' && (char3<=55 && char3>=48) && ((char4<=57 && char4>=48) || (char4<=70 && char4>=65))){
-        printf("ouaiiiiis");
-        return hexaToDecimal(char3,char4);
-    } 
-    else if(char1=='\\' && (char2<=55 && char2>=48) && (char3<=55 && char3>=48) && (char4<=55 && char4>=48)){
-        int octal=((int)char2-48)*100+((int)char3-48)*10+(int)char4-48;
-        return octalToDecimal(octal);
-    }
-    else return 0 ;
-}
 
+int unescapedCharacter(char *cp, char **ep)
+{
+    int c= 0;
+    c= *cp++;
+    if ('\\' == c) {
+	c= *cp++;
+	switch (c) {
+	  case 'n':	c= 10;    break;
+	  case 'a':	c=  7;    break;
+	  case 'b':	c=  8;    break;
+	  case 'e':	c= 27;    break;
+	  case 'f':	c= 12;    break;
+	  case 'r':	c= 13;    break;
+	  case 't':	c=  9;    break;
+	  case 'v':	c=  1;    break;
+	  case '\\':    c= '\\';  break;
+	  case ']':     c= ']';   break;
+	  case '-':     c= '-';   break;
+	  case '0'...'7': {
+	      c -= '0';		// c is now the value of the digit that represented it
+	      if (isRadix(8, *cp)) c= c * 8 + *cp++ - '0';
+	      if (isRadix(8, *cp)) c= c * 8 + *cp++ - '0';
+	      break;
+	  }
+	  case 'x': {
+	      char *op= cp;
+	      c= 0;
+	      if (isRadix(16, *cp)) {
+		  c= c * 16 + digitValue(*cp++);
+		  if (isRadix(16, *cp)) {
+		      c= c * 16 + digitValue(*cp++);
+		      break;
+		  }
+	      }
+	      fprintf(stderr, "illegal hexadecimal escape: \\x%.2s\n", op);
+	      break;
+	  }
+	  default:
+	      fprintf(stderr, "unusual escape: \\%c\n", c);
+	      break;
+	}
+    }
+    if (ep) *ep= cp;
+    return c;
+}
 
 
 
 char *classify(char *spec)
 {
-    char *class= malloc(32);	// 256 bits indicating if character N (bit position N) is in the class or not
-    int index=0;
-    while (spec[index] != '\0' ) {  //*spec
-	// go through spec converting each entry in the spec into a set of bits in the class
-	// test if the next thing is \ .
-	// or test if you have X-Y (where X or Y might be a character or escaped character)
-	//    \n-\r
-       if (spec[index]=='-' && index!=0 && !(index==1 && spec[0]=='^') && spec[index+1]!='\0'){
-
-            if(spec[index-1]<65 || spec[index+1]>122 || (spec[index-1]>90 && spec[index-1]<97) || (spec[index+1]>90 && spec[index+1]<97)){
-                //= if(spec[index-1] or spec[index+1] are not letters)
-
-                if((spec[index-1]>47 && spec[index-1]<57) || (spec[index+1]>47 && spec[index+1]<57) ){ //if it's a digit range
-                    if(spec[index-1]>spec[index+1]){
-                        printf("Error, first digit greater than the second one\n");
-                    }
-                    else{
-                        for(int j=spec[index-1]; j<=spec[index+1] ; j++){
-                            setBit(class,j);
-                            printf("added digit : %c\n",j);
-                            
-                        }
-                        index++;
-                    }   
-                }
-                else{
-                    printf("Error, bad use of range");
-                    exit(1);
-                }
-            }
-            else{
-                if(spec[index-1]>spec[index+1]){
-                printf("Error, first char greater than the second one\n");
-                exit(1);
-                }
-                else if(spec[index-1]>=65 && spec[index-1]<=90 && spec[index+1]>=65 && spec[index+1]<=90 ){
-                    for(int j=spec[index-1]; j<=spec[index+1] ; j++){
-                            setBit(class,j);
-                            printf("added char : %c\n",j);
-                    }
-                    index++;
-                }
-                else if( spec[index-1]>=97 && spec[index-1]<=122 && spec[index+1]>=97 && spec[index+1]<=122 ){
-                    for(int j=spec[index-1]; j<=spec[index+1] ; j++){
-                            setBit(class,j);
-                            printf("added char : %c\n",j);
-                            
-                    }
-                    index++;
-                }
-                else{
-                    printf("Bad use of char range");
-                    exit(1);
-                }
-            }
-
-       }
-       else if((spec[index]>=65 && spec[index]<=90) || (spec[index]>=97 && spec[index]<=122) || (spec[index]<=57 && spec[index]>=48) ){
-           setBit(class,spec[index]);
-           printf("added char : %c\n",spec[index]);
-       }
-
-       else if(isByte(spec[index],spec[index+1],spec[index+2],spec[index+3])){
-           setBit(class,isByte(spec[index],spec[index+1],spec[index+2],spec[index+3]));
-           index=index+4;
-       }
-       else if(isEscapedChar(spec[index],spec[index+1])!=0){
-           if(isEscapedChar(spec[index],spec[index+1])==45){
-               index++;
-           }
-           setBit(class,isEscapedChar(spec[index],spec[index+1]));
-       }
-       
-        printf("%c\n",spec[index]);
-        printf("%i\n",index);
-        index++;
-    }
-
-    if (spec[0]==94) {
-	for (int i= 0;  i < 8;  ++i) class[i] ^= 255;   // invert all bits in the class
-    }
+    char *class= calloc(1, 32);
+
+    int negated= 0;
+    if ('^' == *spec) ++negated, ++spec;
+
+    int prevChar= 0;
+    while (*spec) {
+	if ('-' == *spec && prevChar && spec[1]) {
+	    ++spec;
+	    int thisChar= unescapedCharacter(spec, &spec);
+	    if (thisChar < prevChar) {
+		fprintf(stderr, "range is backwards: \\x%x-\\x%x\n", prevChar, thisChar);
+		int tmp= prevChar;
+		prevChar= thisChar;
+		thisChar= tmp;
+	    }
+	    for (int i= prevChar;  i <= thisChar;  ++i)
+		setBit(class, i);
+	    prevChar= 0;
+	    continue;
+	}
+	prevChar= *spec;
+	setBit(class, unescapedCharacter(spec, &spec));
+    }
+
+    if (negated)
+	for (int i= 0;  i < 32;  ++i)
+	    class[i] ^= 255;   // invert all bits in the class
 
     return class;
 }
 
+
+#ifdef TESTING_CLASS_C
 int main()
 {
     char   *line=0;
@@ -212,4 +132,5 @@ int main()
 
     return 0;
 }
+#endif