fix: utf16be encoding

moonbitlang · Dec 10, 2024 · 8802a11 · 8802a11
1 parent b830031
commit 8802a11
Show file tree

Hide file tree

Showing 3 changed files with 11 additions and 11 deletions.
diff --git a/encoding/decoding_test.mbt b/encoding/decoding_test.mbt
@@ -188,14 +188,14 @@ test "lossy decoding UTF16BE encoded data with UTF8" {
   inspect!(
     buf.to_bytes(),
     content=
-      #|b"\x00\xd1\x00\x65\xd8\x3c\xdf\xc3\x00\x38\x00\xf3\xd8\x3c\xdf\xca"
+      #|b"\x8d\xd1\x6b\x65\xd8\x3c\xdf\xc3\x6e\x38\x6c\xf3\xd8\x3c\xdf\xca"
     ,
   )
   let chars = @encoding.decode_lossy(UTF8, buf.to_bytes())
   inspect!(
     chars.iter().collect(),
     content=
-      #|['\x00', '�', 'e', '�', '�', '\x00', '8', '\x00', '�', '�']
+      #|['�', '�', 'e', '�', '�', 'n', '8', 'l', '�', '�']
     ,
   )
 }
@@ -247,13 +247,13 @@ test "lossy decoding UTF16BE encoded data with UTF16LE" {
   inspect!(
     buf.to_bytes(),
     content=
-      #|b"\x00\xd1\x00\x65\xd8\x3c\xdf\xc3\x00\x38\x00\xf3\xd8\x3c\xdf\xca"
+      #|b"\x8d\xd1\x6b\x65\xd8\x3c\xdf\xc3\x6e\x38\x6c\xf3\xd8\x3c\xdf\xca"
     ,
   )
   let chars = @encoding.decode_lossy(UTF16LE, buf.to_bytes())
   inspect!(
     chars.iter().collect(),
-    content="['턀', '攀', '㳘', '쏟', '㠀', '', '㳘', '쫟']",
+    content="['톍', '敫', '㳘', '쏟', '㡮', '', '㳘', '쫟']",
   )
 }
 
@@ -331,14 +331,14 @@ test "strictly decoding UTF16BE encoded data with UTF8" {
   inspect!(
     buf.to_bytes(),
     content=
-      #|b"\x00\xd1\x00\x65\xd8\x3c\xdf\xc3\x00\x38\x00\xf3\xd8\x3c\xdf\xca"
+      #|b"\x8d\xd1\x6b\x65\xd8\x3c\xdf\xc3\x6e\x38\x6c\xf3\xd8\x3c\xdf\xca"
     ,
   )
   let chars = @encoding.decode_strict(UTF8, buf.to_bytes())
   inspect!(
     chars.iter().collect(),
     content=
-      #|[Ok('\x00'), Err(b"\xd1\x00"), Ok('e'), Err(b"\xd8\x3c"), Err(b"\xdf\xc3"), Ok('\x00'), Ok('8'), Ok('\x00'), Err(b"\xf3\xd8\x3c\xdf"), Err(b"\x00")]
+      #|[Err(b"\x8d"), Err(b"\xd1\x6b"), Ok('e'), Err(b"\xd8\x3c"), Err(b"\xdf\xc3"), Ok('n'), Ok('8'), Ok('l'), Err(b"\xf3\xd8\x3c\xdf"), Err(b"\x00")]
     ,
   )
 }
diff --git a/encoding/encoding.mbt b/encoding/encoding.mbt
@@ -168,7 +168,7 @@ pub fn write_utf16le_char(buf : @buffer.T, value : Char) -> Unit {
 pub fn write_utf16be_char(buf : @buffer.T, value : Char) -> Unit {
   let code = value.to_uint()
   if code < 0x10000 {
-    let b0 = (code >> 0xFF).to_byte()
+    let b0 = (code >> 8).to_byte()
     let b1 = (code & 0xFF).to_byte()
     buf.write_byte(b0)
     buf.write_byte(b1)

diff --git a/encoding/encoding_test.mbt b/encoding/encoding_test.mbt
@@ -51,7 +51,7 @@ test "encoding String to UTF16BE" {
   inspect!(
     bytes,
     content=
-      #|b"\x00\xbb\x00\x66\x00\x2e\x00\x28\x00\xbb\x00\x78\x00\x2e\x00\x66\x00\x28\x00\x78\x00\x20\x00\x78\x00\x29\x00\x29\x00\x28\x00\xbb\x00\x78\x00\x2e\x00\x66\x00\x28\x00\x78\x00\x20\x00\x78\x00\x29\x00\x29"
+      #|b"\x03\xbb\x00\x66\x00\x2e\x00\x28\x03\xbb\x00\x78\x00\x2e\x00\x66\x00\x28\x00\x78\x00\x20\x00\x78\x00\x29\x00\x29\x00\x28\x03\xbb\x00\x78\x00\x2e\x00\x66\x00\x28\x00\x78\x00\x20\x00\x78\x00\x29\x00\x29"
     ,
   )
 }
@@ -147,13 +147,13 @@ test "to_utf16be_bytes" {
   inspect!(
     @encoding.to_utf16be_bytes('α'),
     content=
-      #|b"\x00\xb1"
+      #|b"\x03\xb1"
     ,
   )
   inspect!(
     @encoding.to_utf16be_bytes('啊'),
     content=
-      #|b"\x00\x4a"
+      #|b"\x55\x4a"
     ,
   )
   inspect!(
@@ -215,7 +215,7 @@ test "write_utf16be_char" {
   inspect!(
     buf.to_bytes(),
     content=
-      #|b"\x00\x41\x00\xb1\x00\x4a\xd8\x3d\xde\x26"
+      #|b"\x00\x41\x03\xb1\x55\x4a\xd8\x3d\xde\x26"
     ,
   )
 }