strconv: fix atoi() and its tests (#23737)

Bruno-Vdr · web-flow · commit 9bed50d76d04 · 2025-02-17T12:07:26.000+02:00
diff --git a/vlib/strconv/atoi.v b/vlib/strconv/atoi.v
@@ -8,7 +8,6 @@ module strconv
 // int_size = 32 << (~u32(0) >> 63)
 // max_u64 = u64(u64(1 << 63) - 1)
 const int_size = 32
-const max_u64 = u64(18446744073709551615)
 
 @[inline]
 pub fn byte_to_lower(c u8) u8 {
@@ -218,34 +217,58 @@ pub fn parse_int(_s string, base int, _bit_size int) !i64 {
 }
 
 // atoi is equivalent to parse_int(s, 10, 0), converted to type int.
+// It follows V scanner as much as observed.
 @[direct_array_access]
 pub fn atoi(s string) !int {
 	if s == '' {
-		return error('strconv.atoi: parsing "": invalid syntax')
-	}
-	if (int_size == 32 && (0 < s.len && s.len < 10))
-		|| (int_size == 64 && (0 < s.len && s.len < 19)) {
-		// Fast path for small integers that fit int type.
-		mut start_idx := 0
-		if s[0] == `-` || s[0] == `+` {
-			start_idx++
-			if s.len - start_idx < 1 {
-				// return 0, &NumError{fnAtoi, s0, ErrSyntax}
-				return error('strconv.atoi: parsing "${s}": invalid syntax')
-			}
+		return error('strconv.atoi: parsing "": empty string')
+	}
+
+	mut start_idx := 0
+	mut sign := i64(1)
+
+	if s[0] == `-` || s[0] == `+` {
+		start_idx++
+		if s[0] == `-` {
+			sign = -1
 		}
-		mut n := 0
-		for i in start_idx .. s.len {
-			ch := s[i] - `0`
-			if ch > 9 {
-				// return 0, &NumError{fnAtoi, s0, ErrSyntax}
-				return error('strconv.atoi: parsing "${s}": invalid syntax')
+	}
+
+	if s.len - start_idx < 1 {
+		return error('strconv.atoi: parsing "${s}": no number after sign')
+	}
+
+	if s[start_idx] == `_` || s[s.len - 1] == `_` {
+		return error('strconv.atoi: parsing "${s}": values cannot start or end with underscores')
+	}
+
+	mut x := i64(0)
+	mut underscored := false
+	for i in start_idx .. s.len {
+		c := s[i] - `0`
+		if c == 47 { // 47 = Ascii(`_`) -  ascii(`0`) = 95 - 48.
+			if underscored == true { // Two consecutives underscore
+				return error('strconv.atoi: parsing "${s}": consecutives underscores are not allowed')
+			}
+			underscored = true
+			continue // Skip underscore
+		} else {
+			if c > 9 {
+				return error('strconv.atoi: parsing "${s}": invalid radix 10 character')
+			}
+			underscored = false
+			x = (x * 10) + (c * sign)
+			if sign == 1 && x > i64_max_int32 {
+				return error('strconv.atoi: parsing "${s}": integer overflow')
+			} else {
+				if x < i64_min_int32 {
+					return error('strconv.atoi: parsing "${s}": integer underflow')
+				}
 			}
-			n = n * 10 + int(ch)
 		}
-		return if s[0] == `-` { -n } else { n }
 	}
-	// Slow path for invalid, big, or underscored integers.
-	int64 := parse_int(s, 10, 0)!
-	return int(int64)
+	return int(x)
 }
+
+const i64_min_int32 = i64(-2147483647) - 1 // msvc has a bug that treats just i64(min_int) as 2147483648 :-(; this is a workaround for it
+const i64_max_int32 = i64(2147483646) + 1
diff --git a/vlib/strconv/atoi_test.v b/vlib/strconv/atoi_test.v
@@ -1,28 +1,66 @@
 import strconv
 
 fn test_atoi() {
-	assert strconv.atoi('16')! == 16
-	assert strconv.atoi('+16')! == 16
-	assert strconv.atoi('-16')! == -16
-
-	// invalid strings
-	if x := strconv.atoi('str') {
-		println(x)
-		assert false
-	} else {
-		assert true
+	struct StrVal { // Inner test struct
+		str_value string
+		int_value int
 	}
-	if x := strconv.atoi('string_longer_than_10_chars') {
-		println(x)
-		assert false
-	} else {
-		assert true
+
+	// Parsing of theses value should succeed.
+	ok := [
+		StrVal{'1', 1},
+		StrVal{'-1', -1},
+		StrVal{'0', 0},
+		StrVal{'+0', 0},
+		StrVal{'-0', 0},
+		StrVal{'-0_00', 0},
+		StrVal{'+0_00', 0},
+		StrVal{'+1', 1},
+		StrVal{'+1024', 1024},
+		StrVal{'+3_14159', 314159},
+		StrVal{'-1_00_1', -1001},
+		StrVal{'-1_024', -1024},
+		StrVal{'123_456_789', 123456789},
+		StrVal{'00000006', 6},
+		StrVal{'0_0_0_0_0_0_0_6', 6},
+		StrVal{'2147483647', 2147483647}, // Signed 32bits max.
+		StrVal{'-2147483648', -2147483648}, // Signed 32bits min.
+	]
+
+	// Check that extracted int value matches its string.
+	for v in ok {
+		// println('Parsing ${v.str_value} should equals ${v.int_value}')
+		assert strconv.atoi(v.str_value)! == v.int_value
 	}
-	if x := strconv.atoi('') {
-		println(x)
-		assert false
-	} else {
-		assert true
+
+	// Parsing of these values should fail !
+	ko := [
+		'', // Empty string
+		'-', // Only sign
+		'+', // Only sign
+		'_', // Only Underscore
+		'_10', // Start with underscore
+		'+_10', // Start with underscore after sign.
+		'-_16', // Start with underscore after sign.
+		'123_', // End with underscore
+		'-3__14159', // Two consecutives underscore.
+		'-3_14159A', // Non radix 10 char.
+		'A42', // Non radix 10 char.
+		'-2147483649', // 32bits underflow by 1.
+		'+2147483648', // 32 bit overflow by 1.
+		'+3147483648', // 32 bit overflow by a lot.
+		'-2147244836470', // Large underflow.
+		'+86842255899621148766244',
+	]
+
+	for v in ko {
+		if r := strconv.atoi(v) {
+			// These conversions should fail so force assertion !
+			assert false, 'The string ${v} int extraction should not succeed or be considered as valid ${r}).'
+		} else {
+			// println('Parsing fails as it should for : "${v}')
+			assert true
+		}
 	}
 }
 
