Merge pull request #148 from iautolab/chore/python314-deps

test: migrate test runner from unittest to pytest

Author: MaleicAcid

.github/workflows/ci.yml

@@ -14,7 +14,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        python-version: [ "3.11", "3.12" ]
+        python-version: [ "3.11", "3.12", "3.13", "3.14" ]
         os: [ windows-latest, ubuntu-latest ]
 
     defaults:
@@ -53,7 +53,6 @@ jobs:
         uses: FedericoCarboni/setup-ffmpeg@v2
         id: setup-ffmpeg
 
-      - name: Test with unittest
-        working-directory: ./tests
+      - name: Test with pytest
         run: |
-          uv run --extra full --extra litellm python -m unittest discover -s . -p 'test_*.py'
+          uv run --extra full --extra litellm pytest

README.md

@@ -321,6 +321,9 @@ uv run ruff format --check openlrc/ tests/
 
 # Type check
 uv run pyright openlrc/
+
+# Tests
+uv run --extra full --extra litellm pytest
 ```
 
 For live translation testing as a developer (and for CI usage), set:

pyproject.toml

@@ -7,7 +7,7 @@ name = "openlrc"
 version = "1.7.0a1"
 description = "Transcribe (whisper) and translate (gpt) voice into LRC file."
 authors = [{ name = "Hao Zheng", email = "zhenghaosustc@gmail.com" }]
-requires-python = ">=3.11, <3.13"
+requires-python = ">=3.11, <3.15"
 readme = "README.md"
 license = "MIT"
 keywords = ["openai-gpt3", "whisper", "voice transcribe", "lrc"]
@@ -21,7 +21,7 @@ classifiers = [
 dependencies = [
     "openai>=2.20.0",
     "anthropic>=0.90.0",
-    "tiktoken>=0.8.0,<0.9",
+    "tiktoken>=0.12.0,<0.14",
     "langcodes>=3.3.0,<4",
     "language-data~=1.1",
     "tqdm>=4.65.0,<5",
@@ -32,13 +32,12 @@ dependencies = [
     "lingua-language-detector>=1.3.2,<2",
     "filetype>=1.2.0,<2",
     "jaconvV2>=0.4,<0.5",
-    "spacy>=3.8.5,<4",
+    "spacy>=3.8.13,<3.8.14",  # 3.8.14 has no cp314 wheels/sdist on PyPI (broken release, see explosion/spaCy#13949)
     "pysbd>=0.3.4,<0.4",
     "faster-whisper>=1.1.1,<2",
-    "ffmpeg-normalize>=1.27.5,<2",
+    "ffmpeg-normalize>=1.36.0,<2",
     "google-genai>=1.70.0",
-    "json_repair==0.25.2",
-    "onnxruntime>=1.20.0,<2",
+    "json_repair>=0.50.0,<1",
     "pip>=25.1",
 ]
 
@@ -48,10 +47,10 @@ dependencies = [
 # Only needed when using noise_suppress=True in LRCer.run() or Preprocessor.run().
 full = [
     "dpdfnet>=0.6.0",
-    "numba>=0.58.0",  # first version supporting both Python 3.11 and 3.12
+    "numba>=0.58.0",  # librosa allows >=0.51 which breaks 3.11+ resolution
 ]
 litellm = [
-    "litellm>=1.60,<1.85",
+    "litellm>=1.82,<2",
 ]
 
 [project.urls]
@@ -64,7 +63,15 @@ Homepage = "https://github.com/zh-plus/openlrc"
 openlrc = "openlrc.cli:main"
 
 [dependency-groups]
-dev = ["pyright>=1.1", "ruff>=0.11"]
+dev = [
+    "pyright>=1.1",
+    "ruff>=0.11",
+    "pytest>=8",
+    "pytest-sugar>=1.1.1",
+]
+
+[tool.pytest.ini_options]
+testpaths = ["tests"]
 
 [tool.uv]
 

tests/test_openlrc.py

@@ -12,6 +12,8 @@
 from openlrc.transcribe import TranscriptionInfo
 from openlrc.utils import extend_filename
 
+DATA_DIR = Path(__file__).parent / "data"
+
 # Shared test config — avoids repeating these in every test method.
 _TEST_TRANSCRIPTION = TranscriptionConfig(whisper_model="tiny", compute_type="default", device="cpu")
 
@@ -64,9 +66,9 @@ def _mock_create_chatbot(*args, **kwargs):
 @patch("openlrc.agents.create_chatbot", side_effect=_mock_create_chatbot)
 class TestLRCer(unittest.TestCase):
     def setUp(self) -> None:
-        self.audio_path = Path("data/test_audio.wav")
-        self.video_path = Path("data/test_video.mp4")
-        self.nospeech_video_path = Path("data/test_nospeech_video.mp4")
+        self.audio_path = DATA_DIR / "test_audio.wav"
+        self.video_path = DATA_DIR / "test_video.mp4"
+        self.nospeech_video_path = DATA_DIR / "test_nospeech_video.mp4"
 
     def tearDown(self) -> None:
         def clear_paths(input_path):
@@ -89,7 +91,7 @@ def clear_paths(input_path):
 
         self.video_path.with_suffix(".wav").unlink(missing_ok=True)
 
-        shutil.rmtree("data/preprocessed", ignore_errors=True)
+        shutil.rmtree(DATA_DIR / "preprocessed", ignore_errors=True)
 
     # ------------------------------------------------------------------
     # Pipeline tests (using new config API)
@@ -115,22 +117,22 @@ def test_multiple_audio_transcription_translation(self, _mock_chatbot):
     def test_audio_file_not_found(self, _mock_chatbot):
         lrcer = LRCer(transcription=_TEST_TRANSCRIPTION)
         with self.assertRaises(FileNotFoundError):
-            lrcer.run("data/invalid.mp3")
+            lrcer.run(DATA_DIR / "invalid.mp3")
 
     @patch(
         "openlrc.translate.LLMTranslator.translate", MagicMock(return_value=["test translation1", "test translation2"])
     )
     def test_video_file_transcription_translation(self, _mock_chatbot):
         lrcer = LRCer(transcription=_TEST_TRANSCRIPTION)
-        result = lrcer.run("data/test_video.mp4")
+        result = lrcer.run(DATA_DIR / "test_video.mp4")
         self.assertTrue(result)
 
     @patch(
         "openlrc.translate.LLMTranslator.translate", MagicMock(return_value=["test translation1", "test translation2"])
     )
     def test_nospeech_video_file_transcription_translation(self, _mock_chatbot):
         lrcer = LRCer(transcription=_TEST_TRANSCRIPTION)
-        result = lrcer.run("data/test_nospeech_video.mp4")
+        result = lrcer.run(DATA_DIR / "test_nospeech_video.mp4")
         self.assertTrue(result)
 
     @patch("openlrc.translate.LLMTranslator.translate", MagicMock(side_effect=Exception("test exception")))
@@ -142,7 +144,7 @@ def test_translation_error(self, _mock_chatbot):
     @patch("openlrc.translate.LLMTranslator.translate", MagicMock(side_effect=Exception("test exception")))
     def test_skip_translation(self, _mock_chatbot):
         lrcer = LRCer(transcription=_TEST_TRANSCRIPTION)
-        result = lrcer.run("data/test_video.mp4", skip_trans=True)
+        result = lrcer.run(DATA_DIR / "test_video.mp4", skip_trans=True)
         self.assertTrue(result)
 
     @patch(

tests/test_opt.py

@@ -4,14 +4,17 @@
 import json
 import os
 import unittest
+from pathlib import Path
 
 from openlrc.opt import SubtitleOptimizer
 from openlrc.subtitle import Subtitle
 
+DATA_DIR = Path(__file__).parent / "data"
+
 
 class TestSubtitleOptimizer(unittest.TestCase):
     def setUp(self) -> None:
-        self.subtitle = Subtitle.from_json("data/test_valid_subtitle.json")
+        self.subtitle = Subtitle.from_json(DATA_DIR / "test_valid_subtitle.json")
 
     def test_merge_same(self):
         subtitle = self.subtitle
@@ -75,12 +78,12 @@ def test_save(self):
         subtitle = self.subtitle
         optimizer = SubtitleOptimizer(subtitle)
         optimizer.perform_all()
-        optimizer.save(output_name="data/test_subtitle_optimized.json")
+        optimizer.save(output_name=DATA_DIR / "test_subtitle_optimized.json")
 
-        with open("data/test_subtitle_optimized.json", encoding="utf-8") as f:
+        with open(DATA_DIR / "test_subtitle_optimized.json", encoding="utf-8") as f:
             optimized_subtitle = json.load(f)
 
         self.assertEqual(optimized_subtitle["language"], "zh")
         self.assertEqual(len(optimized_subtitle["segments"]), 8)
 
-        os.remove("data/test_subtitle_optimized.json")
+        os.remove(DATA_DIR / "test_subtitle_optimized.json")

tests/test_preprocess.py

@@ -11,6 +11,8 @@
 
 from openlrc.preprocess import Preprocessor
 
+DATA_DIR = Path(__file__).parent / "data"
+
 # Inject lightweight fakes for the optional noise-suppression stack so these
 # tests run without the openlrc[full] extra installed.
 _dpdfnet = types.ModuleType("dpdfnet")
@@ -43,7 +45,7 @@ def write(self, data):
 
 class TestPreprocessor(unittest.TestCase):
     def tearDown(self) -> None:
-        preprocessed_path = Path("data/preprocessed")
+        preprocessed_path = DATA_DIR / "preprocessed"
         shutil.rmtree(preprocessed_path, ignore_errors=True)
 
     @patch.object(_dpdfnet, "enhance")
@@ -81,7 +83,7 @@ def test_noise_suppression_shape_mismatch_removes_partial_file(self, mock_unlink
     @patch("openlrc.preprocess.FFmpegNormalize")
     def test_loudness_normalization_returns_path_objects(self, mock_norm):
         mock_norm.return_value.run_normalization.return_value = None
-        preprocessor = Preprocessor("data/test_audio.wav")
+        preprocessor = Preprocessor(DATA_DIR / "test_audio.wav")
         ln_paths = preprocessor.loudness_normalization(preprocessor.audio_paths)
         self.assertIsInstance(ln_paths, list)
         self.assertIsInstance(ln_paths[0], Path)

tests/test_subtitle.py

@@ -2,13 +2,16 @@
 #  All rights reserved.
 
 import unittest
+from pathlib import Path
 
 from openlrc.subtitle import Subtitle
 
+DATA_DIR = Path(__file__).parent / "data"
+
 
 class TestSubtitle(unittest.TestCase):
     def setUp(self) -> None:
-        self.subtitle = Subtitle.from_json("data/test_valid_subtitle.json")
+        self.subtitle = Subtitle.from_json(DATA_DIR / "test_valid_subtitle.json")
 
     def check_content(self, subtitle, length=10):
         self.assertEqual(subtitle.lang, "zh")
@@ -24,13 +27,13 @@ def test_load_valid_json(self):
         self.check_content(self.subtitle)
 
     def test_load_lrc(self):
-        subtitle = Subtitle.from_file("data/test_subtitle.lrc")
+        subtitle = Subtitle.from_file(DATA_DIR / "test_subtitle.lrc")
         self.check_content(subtitle, length=7)
 
     def test_save_json(self):
         subtitle = self.subtitle
-        subtitle.save("data/saved.json")
-        loaded_subtitle = Subtitle.from_file("data/saved.json")
+        subtitle.save(DATA_DIR / "saved.json")
+        loaded_subtitle = Subtitle.from_file(DATA_DIR / "saved.json")
         self.check_content(loaded_subtitle)
         loaded_subtitle.filename.unlink()
 

tests/test_transcribe.py

@@ -9,6 +9,8 @@
 
 from openlrc.transcribe import Transcriber, TranscriptionInfo
 
+DATA_DIR = Path(__file__).parent / "data"
+
 return_tuple = (
     [
         Segment(
@@ -44,7 +46,7 @@
 
 class TestTranscriber(unittest.TestCase):
     def setUp(self) -> None:
-        self.audio_path = Path("data/test_audio.wav")
+        self.audio_path = DATA_DIR / "test_audio.wav"
 
     @patch("openlrc.transcribe.BatchedInferencePipeline")
     def test_transcribe_success(self, MockBatchedInferencePipeline):

tests/test_utils.py

@@ -14,12 +14,14 @@
     parse_timestamp,
 )
 
+DATA_DIR = Path(__file__).parent / "data"
+
 
 class TestUtils(unittest.TestCase):
     def setUp(self) -> None:
-        self.audio_file = Path("data/test_audio.wav")
-        self.video_file = Path("data/test_video.mp4")
-        self.unsupported = Path("data/unsupported_file.xyz")
+        self.audio_file = DATA_DIR / "test_audio.wav"
+        self.video_file = DATA_DIR / "test_video.mp4"
+        self.unsupported = DATA_DIR / "unsupported_file.xyz"
 
     def tearDown(self) -> None:
         self.video_file.with_suffix(".wav").unlink(missing_ok=True)