Merge pull request #225 from ThakeeNathees/number-from-string

Number constructor from string

src/pk_compiler.c

@@ -814,7 +814,7 @@ static void eatNumber(Compiler* compiler) {
   char c = *parser->token_start;
 
   // Binary literal.
-  if (c == '0' && peekChar(parser) == 'b') {
+  if (c == '0' && ((peekChar(parser) == 'b') || (peekChar(parser) == 'B'))) {
     eatChar(parser); // Consume '0b'
 
     uint64_t bin = 0;
@@ -845,7 +845,8 @@ static void eatNumber(Compiler* compiler) {
     }
     value = VAR_NUM((double)bin);
 
-  } else if (c == '0' && peekChar(parser) == 'x') {
+  } else if (c == '0' &&
+             ((peekChar(parser) == 'x') || (peekChar(parser) == 'X'))) {
     eatChar(parser); // Consume '0x'
 
     uint64_t hex = 0;
@@ -883,16 +884,19 @@ static void eatNumber(Compiler* compiler) {
     }
 
   } else { // Regular number literal.
+
     while (utilIsDigit(peekChar(parser))) {
       eatChar(parser);
     }
 
+    if (c != '.') { // Number starts with a decimal point.
       if (peekChar(parser) == '.' && utilIsDigit(peekNextChar(parser))) {
         matchChar(parser, '.');
         while (utilIsDigit(peekChar(parser))) {
           eatChar(parser);
         }
       }
+    }
 
     // Parse if in scientific notation format (MeN == M * 10 ** N).
     if (matchChar(parser, 'e') || matchChar(parser, 'E')) {

src/pk_core.c

@@ -769,10 +769,22 @@ static void _ctorBool(PKVM* vm) {
 
 static void _ctorNumber(PKVM* vm) {
   double value;
-  if (!validateNumeric(vm, ARG(1), &value, "Argument 1")) return;
+  if (isNumeric(ARG(1), &value)) {
     RET(VAR_NUM(value));
   }
 
+  if (IS_OBJ_TYPE(ARG(1), OBJ_STRING)) {
+    String* str = (String*) AS_OBJ(ARG(1));
+    double value;
+    const char* err = utilToNumber(str->data, &value);
+    if (err == NULL) RET(VAR_NUM(value));
+    VM_SET_ERROR(vm, newString(vm, err));
+    RET(VAR_NULL);
+  }
+
+  VM_SET_ERROR(vm, newString(vm, "Argument must be numeric or string."));
+}
+
 static void _ctorString(PKVM* vm) {
   if (!pkCheckArgcRange(vm, ARGC, 0, 1)) return;
   if (ARGC == 0) {

src/pk_utils.c

@@ -6,6 +6,11 @@
 
 #include "pk_utils.h"
 
+#include <assert.h>
+#include <errno.h>
+#include <stdlib.h>
+#include <string.h>
+
 // Function implementation, see utils.h for description.
 int utilPowerOf2Ceil(int n) {
   n--;
@@ -94,6 +99,122 @@ uint32_t utilHashString(const char* string) {
 #undef FNV_offset_basis_32_bit
 }
 
+const char* utilToNumber(const char* str, double* num) {
+#define IS_HEX_CHAR(c)            \
+  (('0' <= (c) && (c) <= '9')  || \
+   ('a' <= (c) && (c) <= 'f'))
+
+#define IS_BIN_CHAR(c) (((c) == '0') || ((c) == '1'))
+#define IS_DIGIT(c) (('0' <= (c)) && ((c) <= '9'))
+#define INVALID_NUMERIC_STRING "Invalid numeric string."
+
+  assert((str != NULL) && (num != NULL));
+  uint32_t length = (uint32_t) strlen(str);
+
+  // Consume the sign.
+  double sign = +1;
+  if (*str == '-') {
+    sign = -1;
+    str++;
+  } else if (*str == '+') {
+    str++;
+  }
+
+  // Binary String.
+  if (length >= 3 &&
+      ((strncmp(str, "0b", 2) == 0) || (strncmp(str, "0B", 2) == 0))) {
+
+    uint64_t bin = 0;
+    const char* c = str + 2;
+
+    if (*c == '\0') return INVALID_NUMERIC_STRING;
+
+    do {
+      if (*c == '\0') {
+        *num = sign * bin;
+        return NULL;
+      };
+
+      if (!IS_BIN_CHAR(*c)) return INVALID_NUMERIC_STRING;
+      if ((c - str) > (68 /*STR_BIN_BUFF_SIZE*/ - 2)) { // -2 for '-, \0'.
+        return "Binary literal is too long.";
+      }
+
+      bin = (bin << 1) | (*c - '0');
+      c++;
+
+    } while (true);
+  }
+
+  // Hex String.
+  if (length >= 3 &&
+      ((strncmp(str, "0x", 2) == 0) || (strncmp(str, "0X", 2) == 0))) {
+
+    uint64_t hex = 0;
+    const char* c = str + 2;
+
+    if (*c == '\0') return INVALID_NUMERIC_STRING;
+
+    do {
+      if (*c == '\0') {
+        *num = sign * hex;
+        return NULL;
+      };
+
+      if (!IS_HEX_CHAR(*c)) return INVALID_NUMERIC_STRING;
+      if ((c - str) > (20 /*STR_HEX_BUFF_SIZE*/ - 2)) { // -2 for '-, \0'.
+        return "Hex literal is too long.";
+      }
+
+      uint8_t digit = ('0' <= *c && *c <= '9')
+        ? (uint8_t)(*c - '0')
+        : (uint8_t)((*c - 'a') + 10);
+
+      hex = (hex << 4) | digit;
+      c++;
+
+    } while (true);
+  }
+
+  // Regular number.
+  if (*str == '\0') return INVALID_NUMERIC_STRING;
+
+  const char* c = str;
+  do {
+
+    while (IS_DIGIT(*c)) c++;
+    if (*c == '.') { // TODO: allowing "1." as a valid float.?
+      c++;
+      while (IS_DIGIT(*c)) c++;
+    }
+
+    if (*c == '\0') break; // Done.
+
+    if (*c == 'e' || *c == 'E') {
+      c++;
+      if (*c == '+' || *c == '-') c++;
+
+      if (!IS_DIGIT(*c)) return INVALID_NUMERIC_STRING;
+      while (IS_DIGIT(*c)) c++;
+      if (*c != '\0') return INVALID_NUMERIC_STRING;
+    }
+
+    if (*c != '\0') return INVALID_NUMERIC_STRING;
+
+  } while (false);
+
+  errno = 0;
+  *num = atof(str) * sign;
+  if (errno == ERANGE) return "Numeric string is too long.";
+
+  return NULL;
+
+#undef INVALID_NUMERIC_STRING
+#undef IS_DIGIT
+#undef IS_HEX_CHAR
+#undef IS_BIN_CHAR
+}
+
 /****************************************************************************
  * UTF8                                                                     *
  ****************************************************************************/

src/pk_utils.h

@@ -38,6 +38,11 @@ uint32_t utilHashNumber(double num);
 // Generate a has code for [string].
 uint32_t utilHashString(const char* string);
 
+// Convert the string to number. On success it'll return NULL and set the
+// [num] value. Otherwise it'll return a C literal string containing the error
+// message.
+const char* utilToNumber(const char* str, double* num);
+
 /****************************************************************************
  * UTF8                                                                     *
  ****************************************************************************/

tests/lang/builtin_ty.pk

@@ -2,7 +2,7 @@
 ## Builtin types attributes and methods tests.
 
 ###############################################################################
-## INTEGER
+## Number
 ###############################################################################
 
 sum = 0
@@ -11,6 +11,10 @@ sum = 0
 end
 assert(sum == 0 + 1 + 2 + 3 + 4)
 
+assert(Number("-.23e+6") == -.23e6)
+assert(Number("0b10100101") == 0b10100101)
+assert(Number("-0Xabcdef123") == -0xabcdef123)
+
 ###############################################################################
 ## STRING
 ###############################################################################