Start work on configuring TTS backends

src/calibre/gui2/tts/linux.py

@@ -13,13 +13,13 @@ from .errors import TTSSystemUnavailable
 class Client:
 
     mark_template = '<mark name="{}"/>'
+    name = 'speechd'
 
     @classmethod
     def escape_marked_text(cls, text):
         return prepare_string_for_xml(text)
 
     def __init__(self, dispatch_on_main_thread):
-        self.create_ssip_client()
         self.status = {'synthesizing': False, 'paused': False}
         self.dispatch_on_main_thread = dispatch_on_main_thread
         self.current_marked_text = None
@@ -27,6 +27,9 @@ class Client:
         self.next_cancel_is_for_pause = False
         self.next_begin_is_for_resume = False
         self.current_callback = None
+        self.settings_applied = False
+        self.ssip_client = None
+        self.system_default_output_module = None
 
     def create_ssip_client(self):
         from speechd.client import Priority, SpawnError, SSIPClient
@@ -37,12 +40,29 @@ class Client:
         self.ssip_client.set_priority(Priority.TEXT)
 
     def __del__(self):
-        if hasattr(self, 'ssip_client'):
+        if self.ssip_client is not None:
             self.ssip_client.cancel()
             self.ssip_client.close()
-            del self.ssip_client
+            self.ssip_client = None
     shutdown = __del__
 
+    def ensure_state(self, use_ssml=False):
+        if self.ssip_client is None:
+            self.create_ssip_client()
+        if self.system_default_output_module is None:
+            self.system_default_output_module = self.ssip_client.get_output_module()
+        if not self.settings_applied:
+            self.apply_settings()
+        self.set_use_ssml(use_ssml)
+
+    def apply_settings(self, new_settings=None):
+        if self.settings_applied:
+            self.shutdown()
+            self.settings_applied = False
+            self.ensure_state()
+        self.settings_applied = True
+        # TODO: Implement this
+
     def set_use_ssml(self, on):
         from speechd.client import DataMode, SSIPCommunicationError
         mode = DataMode.SSML if on else DataMode.TEXT
@@ -50,12 +70,12 @@ class Client:
             self.ssip_client.set_data_mode(mode)
         except SSIPCommunicationError:
             self.ssip_client.close()
-            self.create_ssip_client()
-            self.ssip_client.set_data_mode(mode)
+            self.ssip_client = None
+            self.ensure_state(on)
 
     def speak_simple_text(self, text):
         self.stop()
-        self.set_use_ssml(False)
+        self.ensure_state(use_ssml=False)
         self.current_marked_text = self.last_mark = None
 
         def callback(callback_type, index_mark=None):
@@ -104,7 +124,7 @@ class Client:
             self.dispatch_on_main_thread(partial(callback_wrapper, callback_type, index_mark))
         self.current_callback = cw
 
-        self.set_use_ssml(True)
+        self.ensure_state(use_ssml=True)
         self.ssip_client.speak(text, callback=self.current_callback)
 
     def pause(self):
@@ -131,4 +151,21 @@ class Client:
         self.current_callback = self.current_marked_text = self.last_mark = None
         self.next_cancel_is_for_pause = False
         self.next_begin_is_for_resume = False
-        self.ssip_client.stop()
+        if self.ssip_client is not None:
+            self.ssip_client.stop()
+
+    def config_widget(self, backend_settings, parent):
+        from calibre.gui2.tts.linux_config import Widget
+        return Widget(self, backend_settings, parent)
+
+    def get_voice_data(self):
+        ans = getattr(self, 'voice_data', None)
+        if ans is None:
+            self.ensure_state()
+            ans = self.voice_data = {}
+            output_module = self.ssip_client.get_output_module()
+            for om in self.ssip_client.list_output_modules():
+                self.ssip_client.set_output_module(om)
+                ans[om] = tuple(self.ssip_client.list_synthesis_voices())
+            self.ssip_client.set_output_module(output_module)
+        return ans

src/calibre/gui2/tts/linux_config.py

@@ -0,0 +1,72 @@
+#!/usr/bin/env python
+# vim:fileencoding=utf-8
+# License: GPL v3 Copyright: 2020, Kovid Goyal <kovid at kovidgoyal.net>
+
+from PyQt5.Qt import (
+    QAbstractTableModel, QComboBox, QFontMetrics, QFormLayout, Qt, QTableView,
+    QWidget
+)
+
+from calibre.gui2.preferences.look_feel import BusyCursor
+
+
+class VoicesModel(QAbstractTableModel):
+
+    def __init__(self, voice_data, default_output_module, parent=None):
+        super().__init__(parent)
+        self.voice_data = voice_data
+        self.current_voices = voice_data[default_output_module]
+        self.column_headers = (_('Name'), _('Language'), _('Variant'))
+
+    def rowCount(self, parent=None):
+        return len(self.current_voices)
+
+    def columnCount(self, parent=None):
+        return 3
+
+    def headerData(self, section, orientation, role=Qt.ItemDataRole.DisplayRole):
+        if role == Qt.ItemDataRole.DisplayRole and orientation == Qt.Orientation.Horizontal:
+            return self.column_headers[section]
+        return super().headerData(section, orientation, role)
+
+    def data(self, index, role=Qt.ItemDataRole.DisplayRole):
+        if role == Qt.ItemDataRole.DisplayRole:
+            row = index.row()
+            try:
+                data = self.current_voices[row]
+                return data[index.column()]
+            except IndexError:
+                return
+
+    def change_output_module(self, om):
+        self.beginResetModel()
+        try:
+            self.current_voices = self.voice_data[om]
+        finally:
+            self.endResetModel()
+
+
+class Widget(QWidget):
+
+    def __init__(self, tts_client, initial_backend_settings, parent=None):
+        QWidget.__init__(self, parent)
+        self.l = l = QFormLayout(self)
+        self.tts_client = tts_client
+
+        self.output_modules = om = QComboBox(self)
+        with BusyCursor():
+            self.voice_data = self.tts_client.get_voice_data()
+            self.system_default_output_module = self.tts_client.system_default_output_module
+        om.addItem(_('System default'), self.system_default_output_module)
+        l.addRow(_('Speech synthesizer:'), om)
+
+        self.voices = v = QTableView(self)
+        self.voices_model = VoicesModel(self.voice_data, self.system_default_output_module, parent=v)
+        v.setModel(self.voices_model)
+        v.horizontalHeader().resizeSection(0, QFontMetrics(self.font()).averageCharWidth() * 30)
+        l.addRow(v)
+
+    def sizeHint(self):
+        ans = super().sizeHint()
+        ans.setHeight(max(ans.height(), 600))
+        return ans

src/calibre/gui2/tts/macos.py

@@ -8,6 +8,7 @@ from .common import Event, EventType
 class Client:
 
     mark_template = '[[sync 0x{:x}]]'
+    name = 'nsss'
 
     @classmethod
     def escape_marked_text(cls, text):

src/calibre/gui2/tts/windows.py

@@ -13,6 +13,7 @@ from .common import Event, EventType
 class Client:
 
     mark_template = '<bookmark mark="{}"/>'
+    name = 'sapi'
 
     @classmethod
     def escape_marked_text(cls, text):

src/calibre/gui2/viewer/config.py

@@ -22,3 +22,8 @@ def get_session_pref(name, default=None, group='standalone_misc_settings'):
     sd = vprefs['session_data']
     g = sd.get(group, {}) if group else sd
     return g.get(name, default)
+
+
+def get_pref_group(name):
+    sd = vprefs['session_data']
+    return sd.get(name) or {}

src/calibre/gui2/viewer/tts.py

@@ -2,9 +2,30 @@
 # vim:fileencoding=utf-8
 # License: GPL v3 Copyright: 2020, Kovid Goyal <kovid at kovidgoyal.net>
 
-from PyQt5.Qt import QObject, pyqtSignal
+from PyQt5.Qt import QObject, QVBoxLayout, pyqtSignal
 
 from calibre.gui2 import error_dialog
+from calibre.gui2.viewer.config import get_pref_group, vprefs
+from calibre.gui2.widgets2 import Dialog
+
+
+class Config(Dialog):
+
+    def __init__(self, tts_client, ui_settings, backend_settings, parent):
+        self.tts_client = tts_client
+        self.ui_settings = ui_settings
+        self.backend_settings = backend_settings
+        Dialog.__init__(self, _('Configure Read aloud'), 'read-aloud-config', parent, prefs=vprefs)
+
+    def setup_ui(self):
+        self.l = l = QVBoxLayout(self)
+        self.config_widget = self.tts_client.config_widget(self.backend_settings, self)
+        l.addWidget(self.config_widget)
+        l.addWidget(self.bb)
+
+    def accept(self):
+        self.backend_settings = self.config_widget.backend_settings
+        return super().accept()
 
 
 def add_markup(text_parts):
@@ -23,6 +44,7 @@ class TTS(QObject):
 
     dispatch_on_main_thread_signal = pyqtSignal(object)
     event_received = pyqtSignal(object, object)
+    settings_changed = pyqtSignal(object)
 
     def __init__(self, parent=None):
         QObject.__init__(self, parent)
@@ -80,3 +102,11 @@ class TTS(QObject):
 
     def stop(self, data):
         self.tts_client.stop()
+
+    def configure(self, data):
+        ui_settings = get_pref_group('tts').copy()
+        key = 'tts_' + self.tts_client.name
+        d = Config(self.tts_client, ui_settings, vprefs.get(key) or {}, parent=self.parent())
+        if d.exec_() == d.DialogCode.Accepted:
+            vprefs[key] = d.backend_settings
+            self.settings_changed.emit(d.ui_settings)

src/calibre/gui2/viewer/web_view.py

@@ -351,8 +351,10 @@ class WebPage(QWebEnginePage):
             QApplication.instance().clipboard().setMimeData(md)
 
     def javaScriptConsoleMessage(self, level, msg, linenumber, source_id):
-        prefix = {QWebEnginePage.JavaScriptConsoleMessageLevel.InfoMessageLevel: 'INFO', QWebEnginePage.JavaScriptConsoleMessageLevel.WarningMessageLevel: 'WARNING'}.get(
-                level, 'ERROR')
+        prefix = {
+            QWebEnginePage.JavaScriptConsoleMessageLevel.InfoMessageLevel: 'INFO',
+            QWebEnginePage.JavaScriptConsoleMessageLevel.WarningMessageLevel: 'WARNING'
+        }.get(level, 'ERROR')
         prints('%s: %s:%s: %s' % (prefix, source_id, linenumber, msg), file=sys.stderr)
         try:
             sys.stderr.flush()

src/pyj/read_book/read_aloud.pyj

@@ -117,6 +117,7 @@ class ReadAloud:
             bar.appendChild(cb(None, 'hourglass', _('Pause reading')))
         else:
             bar.appendChild(cb('play', 'play', _('Start reading') if self.state is STOPPED else _('Resume reading')))
+        bar.appendChild(cb('configure', 'cogs', _('Configure Read aloud')))
         bar.appendChild(cb('hide', 'close', _('Close Read aloud')))
         if self.state is not WAITING_FOR_PLAY_TO_START:
             notes_container = bar_container.lastChild
@@ -129,6 +130,10 @@ class ReadAloud:
             else:
                 notes_container.textContent = _('Tap/click on a word to continue from there')
 
+    def configure(self):
+        self.pause()
+        ui_operations.tts('configure')
+
     def play(self):
         if self.state is PAUSED:
             ui_operations.tts('resume')
@@ -194,8 +199,7 @@ class ReadAloud:
             self.state = PLAYING
         elif which is 'end':
             self.state = STOPPED
-            if not self.view.show_next_spine_item():
-                self.hide()
+            self.view.show_next_spine_item()
 
     def send_message(self, type, **kw):
         self.view.iframe_wrapper.send_message('tts', type=type, **kw)

src/pyj/session.pyj

@@ -69,6 +69,7 @@ defaults = {
     'selection_bar_actions': v"['copy', 'lookup', 'highlight', 'remove_highlight', 'search_net', 'clear']",
     'selection_bar_quick_highlights': v"[]",
     'skipped_dialogs': v'{}',
+    'tts': v'{}',
 }
 
 is_local_setting = {
@@ -98,6 +99,7 @@ is_local_setting = {
     'standalone_recently_opened': True,
     'user_stylesheet': True,
     'highlight_style': True,
+    'tts': True,
 }