fix: decode UTF-16LE attributed bodies

CHANGELOG.md

@@ -1,6 +1,7 @@
 # Changelog
 
 ## Unreleased
+- fix: decode UTF-16LE BOM attributed message bodies in plain-text history output (#91, thanks @clawbunny)
 - fix: confirm standard tapback reaction selection in Messages automation (#53, thanks @PeterRosdahl)
 - fix: gate RPC watch reaction metadata on `include_reactions`, not `attachments` (#82)
 - fix: dedupe URL balloon preview duplicates in watch stream without cross-chat/schema regressions (#64, thanks @lesaai)

Sources/IMsgCore/TypedStreamParser.swift

@@ -4,6 +4,12 @@ enum TypedStreamParser {
   static func parseAttributedBody(_ data: Data) -> String {
     guard !data.isEmpty else { return "" }
     let bytes = [UInt8](data)
+    if bytes.count >= 2, bytes[0] == 0xff, bytes[1] == 0xfe {
+      let payload = data.dropFirst(2)
+      if let text = String(data: payload, encoding: .utf16LittleEndian) {
+        return text.trimmingLeadingControlCharacters()
+      }
+    }
     let start = [UInt8(0x01), UInt8(0x2b)]
     let end = [UInt8(0x86), UInt8(0x84)]
     var best = ""

Tests/IMsgCoreTests/UtilityTests.swift

@@ -102,6 +102,15 @@ func typedStreamParserTrimsControlCharacters() {
   #expect(TypedStreamParser.parseAttributedBody(data) == "hello")
 }
 
+@Test
+func typedStreamParserDecodesUTF16LittleEndianBOM() throws {
+  var data = Data([0xff, 0xfe])
+  let body = "hello 🌤️"
+  let encoded = try #require(body.data(using: .utf16LittleEndian))
+  data.append(encoded)
+  #expect(TypedStreamParser.parseAttributedBody(data) == body)
+}
+
 @Test
 func phoneNumberNormalizerFormatsValidNumber() {
   let normalizer = PhoneNumberNormalizer()