fix: handle whitespace correctly

json/internal_types.mbt

@@ -33,32 +33,6 @@ fn ParseContext::make(
   }
 }
 
-///|
-priv struct CharClass(Array[(Char, Char)])
-
-///|
-fn CharClass::of(array : Array[(Char, Char)]) -> CharClass {
-  CharClass(array)
-}
-
-///|
-fn CharClass::contains(self : CharClass, c : Char) -> Bool {
-  let CharClass(self) = self
-  for left = 0, right = self.length(); left < right; {
-    let middle = (left + right) / 2
-    let (min, max) = self[middle]
-    if c < min {
-      continue left, middle
-    } else if c > max {
-      continue middle + 1, right
-    } else {
-      break true
-    }
-  } else {
-    false
-  }
-}
-
 ///|
 priv enum Token {
   Null

json/lex_main.mbt

@@ -12,18 +12,6 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-///|
-let non_ascii_whitespace : CharClass = CharClass::of([
-  ('\u00A0', '\u00A0'),
-  ('\u1680', '\u1680'),
-  ('\u2000', '\u200A'),
-  ('\u2028', '\u2029'),
-  ('\u202F', '\u202F'),
-  ('\u205F', '\u205F'),
-  ('\u3000', '\u3000'),
-  ('\uFEFF', '\uFEFF'),
-])
-
 ///|
 fn ParseContext::lex_value(
   ctx : ParseContext,
@@ -87,9 +75,6 @@ fn ParseContext::lex_value(
         return String(s)
       }
       Some(c) => {
-        if c > '\u{7f}' && non_ascii_whitespace.contains(c) {
-          continue
-        }
         let shift = -c.utf16_len()
         ctx.invalid_char(shift~)
       }

json/lex_misc.mbt

@@ -118,10 +118,7 @@ fn ParseContext::lex_skip_whitespace(ctx : ParseContext) -> Unit {
   for {
     match ctx.read_char() {
       Some('\t' | ' ' | '\n' | '\r') => continue
-      Some(c) => {
-        if c > '\u{7f}' && non_ascii_whitespace.contains(c) {
-          continue
-        }
+      Some(_) => {
         ctx.offset -= 1
         break
       }

json/parse_error_test.mbt

@@ -24,7 +24,9 @@ test "parse error branches" {
   guard (try? @json.parse("nullx")) is Err(InvalidChar(_, _)) else {
     fail("expected InvalidChar for trailing data")
   }
-  inspect(@json.parse("\u{00A0}null"), content="Null")
+  guard (try? @json.parse("\u{00A0}null")) is Err(InvalidChar(_, _)) else {
+    fail("expected InvalidChar for leading non-breaking space")
+  }
   guard (try? @json.parse("tx")) is Err(InvalidChar(_, _)) else {
     fail("expected InvalidChar for invalid literal")
   }

json/parse_test.mbt

@@ -222,7 +222,7 @@ test "parses escaped characters" {
 
 ///|
 test "parses whitespace" {
-  let json = test_parse("{\t \u00A0\uFEFF\n\r\u2028\u2029\u2003}")
+  let json = test_parse("{\t \n\r}")
   assert_eq(json, Json::object({}))
 }
 //endregion