use custom string to bigint reader in place of strtol

for better compatibility with default janet number reader

src/core/bigint.c

@@ -83,29 +83,6 @@ static const JanetAbstractType bi_uint64_type = {
     uint64_unmarshal
 };
 
-static int parse_int64(const char *str, bi_int64 *box) {
-    char *endptr;
-    int base = ((str[0] == '0' && (str[1] == 'x' || str[1] == 'X')) ||
-                (str[0] == '-' && str[1] == '0' && (str[2] == 'x' || str[2] == 'X'))) ? 16 : 10;
-    errno = 0;
-    *box = (bi_int64)strtoll(str, &endptr, base);
-    if ((errno == ERANGE && (*box == LLONG_MAX || *box == LLONG_MIN)) ||
-            (errno != 0 && *box == 0) ||
-            (endptr == str)) return 0;
-    return 1;
-}
-
-static int parse_uint64(const char *str, bi_uint64 *box) {
-    char *endptr;
-    int base = (str[0] == '0' && (str[1] == 'x' || str[1] == 'X')) ? 16 : 10;
-    errno = 0;
-    *box = (bi_int64)strtoull(str, &endptr, base);
-    if ((errno == ERANGE && (*box == ULLONG_MAX)) ||
-            (errno != 0 && *box == 0) ||
-            (endptr == str)) return 0;
-    return 1;
-}
-
 
 static bi_int64 check_bi_int64(Janet x) {
     switch (janet_type(x)) {
@@ -117,7 +94,8 @@ static bi_int64 check_bi_int64(Janet x) {
         }
         case JANET_STRING: {
             bi_int64 value;
-            if (parse_int64((const char *)janet_unwrap_string(x), &value))
+            const uint8_t *str = janet_unwrap_string(x);
+            if (janet_scan_int64(str, janet_string_length(str), &value))
                 return value;
             break;
         }
@@ -142,7 +120,8 @@ static bi_uint64 check_bi_uint64(Janet x) {
         }
         case JANET_STRING: {
             bi_uint64 value;
-            if (parse_uint64((const char *)janet_unwrap_string(x), &value))
+            const uint8_t *str = janet_unwrap_string(x);
+            if (janet_scan_uint64(str, janet_string_length(str), &value))
                 return value;
             break;
         }

src/core/strtod.c

@@ -361,3 +361,101 @@ error:
     free(mant.digits);
     return 1;
 }
+
+#ifdef JANET_BIGINT
+
+static int scan_bigint(
+    const uint8_t *str,
+    int32_t len,
+    uint64_t *out,
+    int *neg
+) {
+    const uint8_t *end = str + len;
+    int seenadigit = 0;
+    int base = 10;
+    *neg = 0;
+    *out = 0;
+    uint64_t accum = 0;
+    /* len max is INT64_MAX in base 2 with _ between each bits */
+    /* '2r' + 64 bits + 63 _  + sign = 130 => 150 for some leading  */
+    /* zeros */
+    if (len > 150) return 0;
+    /* Get sign */
+    if (str >= end) return 0;
+    if (*str == '-') {
+        *neg = 1;
+        str++;
+    } else if (*str == '+') {
+        str++;
+    }
+    /* Check for leading 0x or digit digit r */
+    if (str + 1 < end && str[0] == '0' && str[1] == 'x') {
+        base = 16;
+        str += 2;
+    } else if (str + 1 < end  &&
+               str[0] >= '0' && str[0] <= '9' &&
+               str[1] == 'r') {
+        base = str[0] - '0';
+        str += 2;
+    } else if (str + 2 < end  &&
+               str[0] >= '0' && str[0] <= '9' &&
+               str[1] >= '0' && str[1] <= '9' &&
+               str[2] == 'r') {
+        base = 10 * (str[0] - '0') + (str[1] - '0');
+        if (base < 2 || base > 36) return 0;
+        str += 3;
+    }
+
+    /* Skip leading zeros */
+    while (str < end && *str == '0') {
+        seenadigit = 1;
+        str++;
+    }
+    /* Parse significant digits */
+    while (str < end) {
+        if (*str == '_') {
+            if (!seenadigit) return 0;
+        } else {
+            int digit = digit_lookup[*str & 0x7F];
+            if (*str > 127 || digit >= base) return 0;
+            if (accum > (UINT64_MAX - digit) / base) return 0;
+            accum = accum * base + digit;
+            seenadigit = 1;
+        }
+        str++;
+    }
+
+    if (!seenadigit) return 0;
+    *out = accum;
+    return 1;
+}
+
+int janet_scan_int64(const uint8_t *str, int32_t len, int64_t *out) {
+    int neg;
+    uint64_t bi;
+    if (scan_bigint(str, len, &bi, &neg)) {
+        if (neg && bi <= 0x8000000000000000UL) {
+            *out = -bi;
+            return 1;
+        }
+        if (!neg && bi <= 0x7FFFFFFFFFFFFFFFUL) {
+            *out = bi;
+            return 1;
+        }
+    }
+    return 0;
+}
+
+int janet_scan_uint64(const uint8_t *str, int32_t len, uint64_t *out) {
+    int neg;
+    uint64_t bi;
+    if (scan_bigint(str, len, &bi, &neg)) {
+        if (!neg) {
+            *out = bi;
+            return 1;
+        }
+    }
+    return 0;
+}
+
+#endif

src/include/janet.h

@@ -1338,6 +1338,8 @@ JANET_API Janet janet_bigint_int64(int64_t x);
 JANET_API Janet janet_bigint_uint64(uint64_t x);
 JANET_API int64_t janet_checkbigint_int64(Janet x);
 JANET_API uint64_t janet_checkbigint_uint64(Janet x);
+JANET_API int janet_scan_int64(const uint8_t *str, int32_t len, int64_t *out);
+JANET_API int janet_scan_uint64(const uint8_t *str, int32_t len, uint64_t *out);
 #endif
 
 

test/suite6.janet

@@ -35,7 +35,8 @@
    (def b (u64 0x1fffffffffffff))
    (def b (u64 (math/pow 2 53)))
    # from string
-   (def c (u64 "0xffffffffffffffff"))
+   (def c (u64 "0xffff_ffff_ffff_ffff"))
+   (def c (u64 "32rvv_vv_vv_vv"))
    (def d (u64 "123456789"))))
 
 (assert-no-error
@@ -47,7 +48,7 @@
    (def b (i64 0x1fffffffffffff))
    (def b (i64 (math/pow 2 53)))
    # from string 
-   (def c (i64 "0x7fffffffffffffff"))
+   (def c (i64 "0x7fff_ffff_ffff_ffff"))
    (def d (i64 "123456789"))))
 
 (assert-error
@@ -61,12 +62,12 @@
    # just to big
    (def d (u64 "123456789123456789123456789"))))
 
-(assert (:== (:/ (u64 "0xffffffffffffffff") 8 2) "0xfffffffffffffff") "bigint operations")
+(assert (:== (:/ (u64 "0xffff_ffff_ffff_ffff") 8 2) "0xfffffffffffffff") "bigint operations")
 (assert (let [a (u64 0xff)] (:== (:+ a a a a) (:* a 2 2))) "bigint operations")
 
 (assert-error
  "trap INT64_MIN / -1"
- (:/ (bigint/int64 "-0x8000000000000000") -1))
+ (:/ (bigint/int64 "-0x8000_0000_0000_0000") -1))
 
 # in place operators
 (assert (let [a (u64 1e10)] (:+! a 1000000 "1000000" "0xffff") (:== a 10002065535)) "in place operators")