Merge pull request #219 from denniswebb/publish/fix-kokoro-optional-speed

fix: avoid optional Kokoro speed formatting crash in WebUI
This commit is contained in:
Deniz Şafak authored and GitHub committed 2026-11-01 20:23:27 +03:00
commit 42b430f077
4 files changed
+93 -3

No files matched your search

+2 -2
View File
@@ -222,7 +222,7 @@ def execute_conversion(
voice_resolver, base_voice_spec, request,
log_callback=lambda msg: events.log(msg, level="warning"),
)
logging.info("[executor] Base voice resolved: provider=%s voice=%s speed=%.2f", base_provider, base_voice_choice, base_speed)
logging.info("[executor] Base voice resolved: provider=%s voice=%s speed=%s", base_provider, base_voice_choice, base_speed)
# Use ExitStack for resource management
with ExitStack() as stack:
@@ -309,7 +309,7 @@ def execute_conversion(
voice_resolver, chapter.voice_spec, request,
log_callback=lambda msg: events.log(msg, level="warning"),
)
logging.info("[executor] Chapter %d voice: provider=%s voice=%s speed=%.2f", chapter_idx, chapter_provider, chapter_voice, chapter_speed)
logging.info("[executor] Chapter %d voice: provider=%s voice=%s speed=%s", chapter_idx, chapter_provider, chapter_voice, chapter_speed)
chapter_backend = pipeline_provider.get(chapter_provider, request.language, request.use_gpu)
# Record chapter start for markers
+1 -1
View File
@@ -70,7 +70,7 @@ class AppVoiceResolver:
loaded = resolved
self._cache.set(cache_key, loaded)
logging.info("[resolver] Resolved: spec=%s -> provider=%s resolved=%s speed=%.2f steps=%s",
logging.info("[resolver] Resolved: spec=%s -> provider=%s resolved=%s speed=%s steps=%s",
voice_spec, provider, resolved, speed, steps)
return ResolvedVoice(
provider=provider,
+52
View File
@@ -4,6 +4,7 @@ Uses fake/mock objects for ports (events, pipeline_provider, voice_resolver)
to test the executor without real TTS or audio I/O.
"""
import logging
import tempfile
from pathlib import Path
from typing import Any, List, Optional
@@ -600,6 +601,57 @@ class TestRepeatConversionOutputPath:
assert result.audio_path == Path(tmpdir) / "my_novel.wav"
class TestOptionalVoiceSpeed:
"""Kokoro voices resolve with speed=None; diagnostics must not crash on it."""
class _NoSpeedResolver(FakeVoiceResolver):
def resolve(self, voice_spec: str) -> ResolvedVoice:
resolved = super().resolve(voice_spec)
resolved.speed = None
return resolved
class _SpeedRecordingBackend(FakeBackend):
def __init__(self):
super().__init__()
self.speeds: List[Any] = []
def __call__(self, text: str, *, voice: Any, speed: float = 1.0, **kwargs: Any) -> List:
self.speeds.append(speed)
return super().__call__(text, voice=voice, speed=speed, **kwargs)
def test_none_speed_logs_and_falls_back_to_request_speed(self, caplog):
# caplog's handler re-raises formatting errors (like RichHandler, the
# WebUI console handler), so an unsafe "%.2f" on None fails here.
caplog.set_level(logging.INFO)
with tempfile.TemporaryDirectory() as tmpdir:
req = ConversionRequest(
direct_text="Hello world",
voice="af_heart",
speed=1.3,
save=SaveConfig(mode="custom_folder", output_folder=Path(tmpdir)),
output_format="wav",
)
from abogen.application.conversion_planner import build_conversion_plan
pipeline = FakePipelineProvider()
backend = self._SpeedRecordingBackend()
pipeline.get = lambda *args: backend
events = FakeEvents()
result = execute_conversion(
build_conversion_plan(req),
events,
pipeline,
self._NoSpeedResolver(),
TTSContext(),
)
assert result.audio_path.exists()
assert backend.speeds and set(backend.speeds) == {1.3}
assert not [m for m, level in events.logs if level == "warning"]
assert "Base voice resolved: provider=kokoro voice=af_heart speed=None" in caplog.text
class TestHeadingDedup:
"""Tests for heading dedup in executor."""
+38
View File
@@ -0,0 +1,38 @@
"""Tests for AppVoiceResolver (abogen/application/voice_resolver.py)."""
import logging
from unittest.mock import MagicMock
from abogen.application.conversion_request import ConversionRequest
from abogen.application.voice_resolver import AppVoiceResolver
from abogen.domain.voice_loader import VoiceCache
def _resolver(voice: str, profiles=None, provider: str = "kokoro") -> AppVoiceResolver:
req = ConversionRequest(direct_text="x", voice=voice, tts_provider=provider, speed=1.0)
return AppVoiceResolver(req, profiles or {}, MagicMock(), VoiceCache())
class TestResolveLogging:
"""caplog's handler re-raises formatting errors (like RichHandler, the WebUI
console handler), so unsafe diagnostics on optional values fail here."""
def test_kokoro_voice_keeps_speed_none(self, caplog):
caplog.set_level(logging.INFO)
resolved = _resolver("af_heart").resolve("af_heart")
assert resolved.provider == "kokoro"
assert resolved.resolved_spec == "af_heart"
assert resolved.speed is None
assert "provider=kokoro resolved=af_heart speed=None" in caplog.text
def test_supertonic_profile_speed_is_numeric(self, caplog):
caplog.set_level(logging.INFO)
profiles = {"Narr": {"provider": "supertonic", "voice": "M1", "speed": 1.25, "total_steps": 5}}
resolved = _resolver("profile:Narr", profiles, "supertonic").resolve("profile:Narr")
assert resolved.provider == "supertonic"
assert resolved.speed == 1.25
assert "speed=1.25" in caplog.text