Fix slow Cmd-V paste with mixed clipboard formats

tests/test_plain_text_paste_worker.py

@@ -33,8 +33,8 @@ def tearDown(self):
 
     def board(self, text="hello\n日本語 🦀 e\u0301\r\n", extra=None, behavior=None):
         identity = str(uuid.uuid4())
-        config = dict(name="cmux-paste-test-" + identity, text=text,
-                      representations={"public.utf8-plain-text": text, **(extra or {})},
+        config = dict(name="cmux-paste-test-" + identity, text=text or "",
+                      representations={**({"public.utf8-plain-text": text} if text is not None else {}), **(extra or {})},
                       ready=str(self.root / (identity + "-ready.json")),
                       requested=str(self.root / (identity + "-requested")))
         if behavior:
@@ -92,8 +92,39 @@ def test_unicode_multiline_and_repeat(self):
                 print(json.dumps(dict(helper_startup_ms=ms, directory=directory_index,
                                       repetition=repetition, size=Path(HELPER).stat().st_size)), flush=True)
 
+    def test_plain_text_with_rich_flavors_uses_fast_path(self):
+        for flavor in ["public.html", "public.rtf"]:
+            for text in ["hello\n日本語 🦀 e\u0301\r\n", "Question?", "文本\n" * 100_000]:
+                with self.subTest(flavor=flavor, bytes=len(text.encode())):
+                    board = self.board(text=text, extra={flavor: "unused rich text"})
+                    path = self.directory(board)
+                    result = self.result(path)
+                    self.assertEqual(result["textPayload"]["destination"], {"terminal": {}})
+                    self.assertEqual((path / "text-payload.txt").read_bytes(), text.encode())
+
+    def test_rich_only_and_empty_plain_text_delegate(self):
+        for flavor in ["public.html", "public.rtf", "com.apple.flat-rtfd"]:
+            for text in [None, ""]:
+                with self.subTest(flavor=flavor, text=text):
+                    board = self.board(text=text, extra={flavor: "rich fallback"})
+                    self.result(self.directory(board), expected=73)
+
+    def test_lossy_plain_text_with_rich_flavors_delegates(self):
+        for flavor in ["public.html", "public.rtf"]:
+            for text in ["??", "text\ufffd", "日本語??"]:
+                with self.subTest(flavor=flavor, text=text):
+                    board = self.board(text=text, extra={flavor: "rich fallback"})
+                    self.result(self.directory(board), expected=73)
+
+    def test_loss_markers_without_rich_text_remain_literal(self):
+        text = "??\ufffd"
+        board = self.board(text=text)
+        path = self.directory(board)
+        self.result(path)
+        self.assertEqual((path / "text-payload.txt").read_bytes(), text.encode())
+
     def test_rich_images_and_auxiliary_urls_delegate_without_provider_read(self):
-        for flavor in ["public.html", "public.rtf", "com.apple.flat-rtfd", "public.png",
+        for flavor in ["public.png",
                        "public.tiff", "public.jpeg", "public.file-url", "public.url",
                        "NSFilenamesPboardType", "com.apple.pasteboard.promised-file-url"]:
             with self.subTest(flavor=flavor):

workers/cmux-paste-text/main.m

@@ -121,10 +121,9 @@ static BOOL isPlainTextType(NSString *typeIdentifier) {
 }
 
 static BOOL hasDisallowedType(NSString *typeIdentifier) {
-    if ([typeIdentifier isEqualToString:NSPasteboardTypeHTML] ||
-        [typeIdentifier isEqualToString:NSPasteboardTypeRTF] ||
-        [typeIdentifier isEqualToString:NSPasteboardTypeRTFD] ||
-        [typeIdentifier isEqualToString:NSPasteboardTypeFileURL] ||
+    // RTFD may carry attachments even alongside plain text. Keep its existing
+    // rich-text/image selection policy in the full worker.
+    if ([typeIdentifier isEqualToString:NSPasteboardTypeFileURL] ||
         [typeIdentifier isEqualToString:NSPasteboardTypeURL] ||
         [typeIdentifier isEqualToString:@"NSFilenamesPboardType"] ||
         [typeIdentifier isEqualToString:@"com.apple.pasteboard.promised-file-url"] ||
@@ -190,6 +189,9 @@ static int runWorker(NSArray<NSString *> *arguments) {
 
     NSArray<NSPasteboardType> *types = pasteboard.types ?: @[];
     BOOL hasPlainText = NO;
+    BOOL hasRichText = [types containsObject:NSPasteboardTypeHTML] ||
+        [types containsObject:NSPasteboardTypeRTF] ||
+        [types containsObject:NSPasteboardTypeRTFD];
     for (NSPasteboardType type in types) {
         if (hasDisallowedType(type)) {
             return kIneligibleStatus;
@@ -221,10 +223,25 @@ static int runWorker(NSArray<NSString *> *arguments) {
         return 0;
     }
     if (text == nil) {
+        if (hasRichText) {
+            return kIneligibleStatus;
+        }
         writeResponse(workingDirectory, NO);
         return 0;
     }
 
+    // Match PasteboardTextFidelity.shouldInspectRichTextForPlainTextLoss:
+    // the full worker can recover characters lost by the plain-text exporter.
+    if (hasRichText) {
+        NSUInteger questionMarks = 0;
+        for (NSUInteger index = 0; index < text.length; index++) {
+            unichar character = [text characterAtIndex:index];
+            if (character == 0xFFFD || (character == '?' && ++questionMarks >= 2)) {
+                return kIneligibleStatus;
+            }
+        }
+    }
+
     NSData *payload = [text dataUsingEncoding:NSUTF8StringEncoding allowLossyConversion:NO];
     if (payload == nil || payload.length > kMaximumTextByteCount) {
         return 74;