refactor(python): tidy integer prefix matching and add parse regression tests

src/fable-library-py/src/ints.rs

@@ -1067,9 +1067,9 @@ fn determine_radix(string: &str, style: i32, default_radix: i32) -> i32 {
     // `get` returns None when the string is shorter than two bytes or the
     // boundary falls inside a multi-byte character, so this never panics
     match string.get(..2) {
-        Some("0x") | Some("0X") => 16,
-        Some("0b") | Some("0B") => 2,
-        Some("0o") | Some("0O") => 8,
+        Some("0x" | "0X") => 16,
+        Some("0b" | "0B") => 2,
+        Some("0o" | "0O") => 8,
         _ => default_radix,
     }
 }
@@ -1091,15 +1091,15 @@ fn trim_whitespace(string: &str, style: i32) -> &str {
 /// Removes numeric prefixes (0x, 0b, 0o) when radix is not decimal
 #[inline]
 fn remove_prefix(string: &str, radix: i32) -> &str {
-    if radix != 10 {
-        match string.get(..2) {
-            Some("0x") | Some("0X") | Some("0b") | Some("0B") | Some("0o") | Some("0O") => {
-                &string[2..]
-            }
-            _ => string,
-        }
-    } else {
-        string
+    if radix == 10 {
+        return string;
+    }
+
+    // `get` never panics on a multi-byte boundary; `&string[2..]` is only
+    // reached after matching a two-byte ASCII prefix, so byte 2 is a boundary
+    match string.get(..2) {
+        Some("0x" | "0X" | "0b" | "0B" | "0o" | "0O") => &string[2..],
+        _ => string,
     }
 }
 

src/fable-library-py/tests/test_ints.py

@@ -56,6 +56,10 @@ def test_uint64_create() -> None:
     assert uint64(uint32(-1)) == 4294967295
 
 
+# NumberStyles.Any (511) has no AllowHexSpecifier, so it reaches the prefix
+# check in `determine_radix`. NumberStyles.HexNumber (515) short-circuits to
+# radix 16 and reaches the prefix check in `remove_prefix` instead. Both used
+# to panic on a multi-byte first character.
 @pytest.mark.parametrize(
     "style",
     [

tests/Js/Main/ConvertTests.fs

@@ -160,6 +160,8 @@ let tests =
         (fun () -> Int32.Parse("1FFFFFFFF", System.Globalization.NumberStyles.HexNumber)) |> throwsError ""
         (fun () -> Int32.Parse("5foo", System.Globalization.NumberStyles.HexNumber)) |> throwsError ""
         (fun () -> Int32.Parse("foo5", System.Globalization.NumberStyles.HexNumber)) |> throwsError ""
+        // Multi-byte UTF-8 input must fail to parse, not crash (see #4823)
+        (fun () -> Int32.Parse("–", System.Globalization.NumberStyles.HexNumber)) |> throwsError ""
 
     testCase "System.Int64.Parse works" <| fun () ->
         Int64.Parse("5") |> equal 5L
@@ -207,6 +209,10 @@ let tests =
         tryParse Int32.TryParse 0 "X9TRE34" |> equal (false, 0)
         tryParse Int32.TryParse 0 "9SayWhat12Huh" |> equal (false, 0)
         tryParse Int32.TryParse 0 "-1" |> equal (true, -1)
+        // Multi-byte UTF-8 input must fail to parse, not crash (see #4823)
+        tryParse Int32.TryParse 0 "–" |> equal (false, 0)
+        tryParse Int32.TryParse 0 "é" |> equal (false, 0)
+        tryParse Int32.TryParse 0 "𝟙" |> equal (false, 0)
 
     testCase "BigInt.TryParse works" <| fun () ->
         tryParse bigint.TryParse 0I "4234523548923954" |> equal (true, 4234523548923954I)

tests/Python/TestConvert.fs

@@ -160,6 +160,8 @@ let ``test System.Int32.Parse with hex works`` () =
     (fun () -> Int32.Parse("1FFFFFFFF", System.Globalization.NumberStyles.HexNumber)) |> throwsError ""
     (fun () -> Int32.Parse("5foo", System.Globalization.NumberStyles.HexNumber)) |> throwsError ""
     (fun () -> Int32.Parse("foo5", System.Globalization.NumberStyles.HexNumber)) |> throwsError ""
+    // Multi-byte UTF-8 input must fail to parse, not crash (see #4823)
+    (fun () -> Int32.Parse("–", System.Globalization.NumberStyles.HexNumber)) |> throwsError ""
 
 [<Fact>]
 let ``test System.Int64.Parse works`` () =
@@ -214,6 +216,10 @@ let ``test System.Int32.TryParse works`` () =
     tryParse Int32.TryParse 0 "X9TRE34" |> equal (false, 0)
     tryParse Int32.TryParse 0 "9SayWhat12Huh" |> equal (false, 0)
     tryParse Int32.TryParse 0 "-1" |> equal (true, -1)
+    // Multi-byte UTF-8 input must fail to parse, not crash (see #4823)
+    tryParse Int32.TryParse 0 "–" |> equal (false, 0)
+    tryParse Int32.TryParse 0 "é" |> equal (false, 0)
+    tryParse Int32.TryParse 0 "𝟙" |> equal (false, 0)
 
 [<Fact>]
 let ``test BigInt.TryParse works`` () =