From 16120d8b39c3e1a27e5de4fb850a272faf1fe4bf Mon Sep 17 00:00:00 2001
From: Kovid Goyal <kovid@kovidgoyal.net>
Date: Tue, 3 Sep 2024 09:23:52 +0530
Subject: [PATCH] ...

---
 src/calibre/gui2/tts2/develop.py |  2 +-
 src/calibre/gui2/tts2/piper.py   | 22 ++++++++++++----------
 2 files changed, 13 insertions(+), 11 deletions(-)

diff --git a/src/calibre/gui2/tts2/develop.py b/src/calibre/gui2/tts2/develop.py
index dfcd5e2650..1e61168f2b 100644
--- a/src/calibre/gui2/tts2/develop.py
+++ b/src/calibre/gui2/tts2/develop.py
@@ -42,7 +42,7 @@ class MainWindow(MainWindow):
             marked_text.append(c.position())
             if not c.movePosition(QTextCursor.MoveOperation.NextWord, QTextCursor.MoveMode.KeepAnchor):
                 break
-            marked_text.append(c.selectedText())
+            marked_text.append(c.selectedText().replace('\u2029', '\n'))
             c.setPosition(c.position())
         c.setPosition(0)
         self.marked_text = marked_text
diff --git a/src/calibre/gui2/tts2/piper.py b/src/calibre/gui2/tts2/piper.py
index 281125943c..0725a75fd6 100644
--- a/src/calibre/gui2/tts2/piper.py
+++ b/src/calibre/gui2/tts2/piper.py
@@ -167,16 +167,18 @@ def split_long_sentences(sentence: str, offset: int, lang: str = 'en', limit: in
 def split_into_utterances(text: str, counter: count, lang: str = 'en'):
     text = re.sub(r'\n{2,}', PARAGRAPH_SEPARATOR, text.replace('\r', '')).replace('\n', ' ')
     for start, length in sentence_positions(text, lang):
-        sentence = text[start:start+length].rstrip().replace('\n', ' ')
-        for start, sentence in split_long_sentences(sentence, start, lang):
-            payload = json.dumps({'text': sentence}).encode('utf-8')
-            ba = QByteArray()
-            ba.reserve(len(payload) + 1)
-            ba.append(payload)
-            ba.append(UTTERANCE_SEPARATOR)
-            u = Utterance(id=next(counter), payload_size=len(ba), audio_data=QByteArray(), left_to_write=ba, start=start, length=len(sentence))
-            debug(f'Utterance created {u.id}: {sentence}')
-            yield u
+        sentence = text[start:start+length].rstrip().replace('\n', ' ').strip()
+        if sentence:
+            for start, sentence in split_long_sentences(sentence, start, lang):
+                payload = json.dumps({'text': sentence}).encode('utf-8')
+                ba = QByteArray()
+                ba.reserve(len(payload) + 1)
+                ba.append(payload)
+                ba.append(UTTERANCE_SEPARATOR)
+                u = Utterance(id=next(counter), payload_size=len(ba), audio_data=QByteArray(),
+                              left_to_write=ba, start=start, length=len(sentence))
+                debug(f'Utterance created {u.id}: {sentence}')
+                yield u
 
 
 class Piper(TTSBackend):