strconv.atoi: bug fix #18875

vlib/strconv/atoi.v

@@ -89,14 +89,14 @@ pub fn common_parse_uint2(s string, _base int, _bit_size int) (u64, int) {
 	// Use compile-time constants for common cases.
 	cutoff := strconv.max_u64 / u64(base) + u64(1)
 	max_val := if bit_size == 64 { strconv.max_u64 } else { (u64(1) << u64(bit_size)) - u64(1) }
+	basem1 := base - 1
 
 	mut n := u64(0)
 	for i in start_index .. s.len {
 		mut c := s[i]
 
 		// manage underscore inside the number
 		if c == `_` {
-			// println("Here: ${s#[i..]}")
 			if i == start_index || i >= (s.len - 1) {
 				// println("_ limit")
 				return u64(0), 1
@@ -109,21 +109,25 @@ pub fn common_parse_uint2(s string, _base int, _bit_size int) (u64, int) {
 			continue
 		}
 
+		mut sub_count := 0
+
 		// get the 0-9 digit
 		c -= 48 // subtract the rune `0`
 
 		// check if we are in the superior base rune interval [A..Z]
-		if c >= base {
-			c -= 7
-		}
-
-		// check if we are in the superior base rune interval [a..z]
-		if c >= base {
-			c -= 32 // subtract the `A` - `0` rune to obtain the value of the digit
+		if c >= 17 { // (65 - 48)
+			sub_count++
+			c -= 7 // subtract the `A` - `0` rune to obtain the value of the digit
+
+			// check if we are in the superior base rune interval [a..z]
+			if c >= 42 { // (97 - 7 - 48)
+				sub_count++
+				c -= 32 // subtract the `a` - `0` rune to obtain the value of the digit
+			}
 		}
 
 		// check for digit over base
-		if c >= base {
+		if c > basem1 || (sub_count == 0 && c > 9) {
 			return n, i + 1
 		}
 

vlib/strconv/atoi_test.v

@@ -27,6 +27,9 @@ fn test_atoi() {
 }
 
 fn test_parse_int() {
+	// symbols coverage
+	assert strconv.parse_int('1234567890', 10, 32)! == 1234567890
+	assert strconv.parse_int('19aAbBcCdDeEfF', 16, 64)! == 0x19aAbBcCdDeEfF
 	// Different bases
 	assert strconv.parse_int('16', 16, 0)! == 0x16
 	assert strconv.parse_int('16', 8, 0)! == 0o16
@@ -83,6 +86,32 @@ fn test_common_parse_uint2() {
 	assert error == 4
 }
 
+fn test_common_parse_uint2_fail() {
+	mut ascii_characters := [' ', '!', '"', '#', '\$', '%', '&', "'", '(', ')', '*', '+', ',',
+		'-', '.', '/', ':', ';', '<', '=', '>', '?', '@', '[', '\\', ']', '^', '_', '`', '{', '|',
+		'}', '~']
+	mut special_characters := [':', ';', '<', '=', '>', '?', '@', 'X', 'Y', 'Z', '[', '\\', ']',
+		'^', '_', '`']
+
+	num0, err0 := strconv.common_parse_uint2('1Ab', 16, 32)
+	assert num0 == 427
+	assert err0 == 0
+
+	for ch in ascii_characters {
+		// println("ch: [${ch}]")
+		txt_str := '${ch[0]:c}12Ab'
+		num, err := strconv.common_parse_uint2(txt_str, 16, 32)
+		assert err != 0
+	}
+
+	for ch in special_characters {
+		// println("ch: [${ch}]")
+		txt_str := '${ch[0]:c}12Ab'
+		num, err := strconv.common_parse_uint2(txt_str, 16, 32)
+		assert err != 0
+	}
+}
+
 fn test_common_parse_uint2_compatibility() {
 	test_list := [
 		'1234,1234',