hotfix/TTS audio ext (#60) + tts_cache (#61)

hotfix/dynamic TTS audio extension engines (#60)

self.audio_ext code has been cleaned up, it is now detected instead of hard coded

tts engine has a default audio extension, but we can not be sure that the file actually uses the same extension, for instance in chatterbox a mp3 engine might include some cache in wav format, cached utterances might also be saved by the user in the resources directory

hotfix/tts_cache (#61)

fiX - MycroftAI#2934 was not a functional fix, get_tts returns a string not a cache object
companion PR - MycroftAI#2938

cache was deleted on boot, the min_percent param is set to 100%, "Remove cache data if disk space is running low." actually means "always remove cache data". Value was made configurable (per tts engine) and defaults to 75% now

mycroft/tts/cache.py

@@ -161,6 +161,8 @@ def __init__(self, tts_config, tts_name, audio_file_type):
         self.audio_file_type = audio_file_type
         self.resource_dir = Path(__file__).parent.parent.joinpath("res")
         self.cached_sentences = dict()
+        # curate cache if disk usage is above min %
+        self.min_free_percent = self.config.get("min_free_percent", 75)
 
     def __contains__(self, sha):
         """The cache contains a SHA if it knows of it and it exists on disk."""
@@ -312,7 +314,7 @@ def clear(self):
     def curate(self):
         """Remove cache data if disk space is running low."""
         files_removed = curate_cache(self.temporary_cache_dir,
-                                     min_free_percent=100)
+                                     min_free_percent=self.min_free_percent)
 
         hashes = set([hash_from_path(Path(path)) for path in files_removed])
         for sentence_hash in hashes:

mycroft/tts/tts.py

@@ -23,7 +23,7 @@
 
 import os.path
 from os.path import dirname, exists, isdir, join
-
+from pathlib import Path
 import mycroft.util
 from mycroft.enclosure.api import EnclosureAPI
 from mycroft.configuration import Configuration
@@ -376,15 +376,16 @@ def _execute(self, sentence, ident, listen):
                 # TODO 21.08: remove mutation of audio_file.path.
                 returned_file, phonemes = self.get_tts(
                     sentence, str(audio_file.path))
-                if returned_file.path != audio_file.path:
+                returned_file = Path(returned_file)
+                if returned_file != audio_file.path:
                     warn(
                         DeprecationWarning(
                             f"{self.tts_name} is saving files "
                             "to a different path than requested. If you are "
                             "the maintainer of this plugin, please adhere to "
                             "the file path argument provided. Modified paths "
                             "will be ignored in a future release."))
-                    audio_file = returned_file
+                    audio_file.path = returned_file
                 if phonemes:
                     phoneme_file = self.cache.define_phoneme_file(
                         sentence_hash
@@ -396,8 +397,16 @@ def _execute(self, sentence, ident, listen):
                     audio_file, phoneme_file
                 )
             viseme = self.viseme(phonemes) if phonemes else None
+
+            # tts engine has a default audio extension, but we can not be
+            # sure that the file actually uses the same extension,
+            # for instance a mp3 engine might include some cache in wav format
+            # cached utterances might also be saved by the user in the
+            # resources directory
+            _, audio_ext = os.path.splitext(str(audio_file.path))
+            audio_ext = audio_ext[1:]
             self.queue.put(
-                (self.audio_ext, str(audio_file.path), viseme, ident, l)
+                (audio_ext, str(audio_file.path), viseme, ident, l)
             )
 
     def _get_sentence_from_cache(self, sentence_hash):

test/unittests/tts/test_cache.py

@@ -115,7 +115,7 @@ def test_curate_cache(self, curate_mock):
                                     files['gobo'].path]
         tts_cache.curate()
         curate_mock.assert_called_with(tts_cache.temporary_cache_dir,
-                                       min_free_percent=100)
+                                       min_free_percent=75)
 
         # Verify that the "hashes" kermit and gobo was removed from the
         # dict of hashes.