WIP: parakeeet

stt_voice_messages/gtk/sttbox.py

@@ -13,9 +13,12 @@
 # You should have received a copy of the GNU General Public License
 # along with Gajim. If not, see <http://www.gnu.org/licenses/>.
 
-import logging
+from __future__ import annotations
 
-from gi.repository import Gtk
+import logging
+from pathlib import Path
+
+from gi.repository import Gtk, Adw
 
 from gajim.plugins.gajimplugin import GajimPluginConfig
 from gajim.plugins.plugins_i18n import _
@@ -26,49 +29,62 @@ log = logging.getLogger('gajim.p.stt_voice_messages_sttbox')
 
 class STTBox(Gtk.Box):
     def __init__(self,
-                 preview_audio_widget: Gtk.Box,
                  config: GajimPluginConfig,
-                 audio_file: str,
+                 audio_file: Path,
                  ) -> None:
 
-        Gtk.Box.__init__(self, orientation=Gtk.Orientation.VERTICAL, spacing=12)
+        Gtk.Box.__init__(self, orientation=Gtk.Orientation.HORIZONTAL, spacing=6)
 
         self._config = config
-        self._preview_audio = preview_audio_widget
         self._model = None
         self._audio_file = audio_file
         self._text = ''
 
-        self._transcribe_button = Gtk.Button(label=_('Transcribe'))
+        self._transcribe_button = Gtk.Button.new_from_icon_name("lucide-captions-symbolic")
+        self._transcribe_button.set_tooltip_text(_('Transcribe voice message'))
+
+        self._spinner = Adw.Spinner(valign=Gtk.Align.START, visible=False)
 
         self._transcription_label = Gtk.Label(
             label=_('Nothing transcribed yet'))
         self._transcription_label.set_max_width_chars(40)
-        self._transcription_label.set_line_wrap(True)
-
-        self.add(self._transcribe_button)
-        self.add(self._transcription_label)
+        self._transcription_label.set_wrap(True)
 
+        self.append(self._spinner)
+        self.append(self._transcription_label)
+        self._transcribe_button.connect('clicked', self._on_transcribe_clicked)
         self._result = helper.Results('')
 
-        self._transcribe_button.connect('clicked', self._on_transcribe_clicked)
-
-        self.show_all()
+    @property
+    def button(self) -> Gtk.Button:
+        return self._transcribe_button
 
     def _on_transcribe_clicked(self, _button: Gtk.Button) -> None:
-        log.debug('config.data = %s', self._config.data)
-        model_name = self._config.data['model']
-        model = self._config.data[model_name].instance
-        if model is None:
+        log.debug('config._instance = %s', self._config._instance)
+        self._model = self._config._instance
+        if self._model is None:
             return
 
-        self._model = model
+        if self._model.is_loaded:
+            text = _('Transcribing…')
+        elif self._model.will_download:
+            text = _('Downloading ') + self._model.model_id
+        else:
+            text = _('Loading model…')
+        self._transcription_label.set_text(text)
+        self._spinner.set_visible(True)
+        self._task = helper.BackgroundTask(
+            self._model.load, self._on_load_done)
+        self._task.start()
 
-        transcription_task = helper.BackgroundTask(
-            self._model.transcribe(self._result, self._audio_file),
-            self._show_result
+    def _on_load_done(self):
+        self._transcription_label.set_text(_('Transcribing…'))
+        self._task = helper.BackgroundTask(
+            lambda: self._model.recognize(
+                self._result, helper.load_audio(self._audio_file)),
+            self._show_result,
         )
-        transcription_task.start()
+        self._task.start()
 
     def _show_result(self):
         assert self._model is not  None
@@ -77,3 +93,4 @@ class STTBox(Gtk.Box):
             self._transcription_label.set_text(self._text.strip())
         else:
             self._transcription_label.set_text(_('_Have not heard any word!_'))
+        self._spinner.set_visible(False)