fixing tests and HF_login

.github/workflows/test.yaml

@@ -45,6 +45,8 @@ jobs:
       shell: bash
     - name: Run unit tests
       id: run-tests
+      env:
+          HF_TOKEN: ${{ secrets.HF_TOKEN }}
       run: >
         poetry run pytest \
           --junitxml=pytest.xml \

src/pipepal/audio/tasks/input_output/interface.py

@@ -34,7 +34,7 @@ def get_service(cls, service_data: Dict[str, Any]) -> Any:  # noqa: ANN401
             ValueError: If the service name is unsupported.
         """
         # Use a composite key to uniquely identify instances
-        key: str = cls.get_data_uuid(service_data)
+        key: str = str(cls.get_data_uuid(service_data))
 
         if key not in cls._instances:
             if service_data["service_name"] == DatasetsService.NAME:

src/pipepal/video/tasks/input_output/interface.py

@@ -34,7 +34,7 @@ def get_service(cls, service_data: Dict[str, Any]) -> Any:  # noqa: ANN401
             ValueError: If the service name is unsupported.
         """
         # Use a composite key to uniquely identify instances
-        key: str = cls.get_data_uuid(service_data)
+        key: str = str(cls.get_data_uuid(service_data))
 
         if key not in cls._instances:
             if service_data["service_name"] == FfmpegService.NAME:

src/tests/test_audio_iotask_datasets.py

@@ -1,6 +1,8 @@
 """This module tests the audio's IOTask class."""
 
 
+from typing import Any, Dict, List
+
 import pytest
 from datasets import Dataset
 
@@ -46,7 +48,7 @@ def test_read_audios_from_disk_output_type() -> None:
     """Test the read_audios_from_disk method to check if the output is of type HF datasets."""
     test_input = {
         "data": {
-            "files": ["../../data_for_testing/audio_48khz_mono_16bits.wav", "../../data_for_testing/audio_48khz_stereo_16bits.wav"]
+            "files": ["./data_for_testing/audio_48khz_mono_16bits.wav", "./data_for_testing/audio_48khz_stereo_16bits.wav"]
         },
         "service": {
             "service_name": "Datasets"
@@ -61,15 +63,15 @@ def test_read_audios_from_disk_output_dimensions() -> None:
     This test checks if the dimensions of the output HF datasets object match the input list of audio files.
     Uses mocker to patch the DatasetsService to avoid actual file I/O and simulate reading files.
     """
-    test_input = {
+    test_input: Dict[str, Any] = {
         "data": {
-            "files": ["../../data_for_testing/audio_48khz_mono_16bits.wav", "../../data_for_testing/audio_48khz_stereo_16bits.wav"]
+            "files": ["./data_for_testing/audio_48khz_mono_16bits.wav", "./data_for_testing/audio_48khz_stereo_16bits.wav"]
         },
         "service": {
             "service_name": "Datasets"
         }
     }
-    response = AudioIOTask().read_audios_from_disk(test_input)
+    response: Dict[str, List[str]]  = AudioIOTask().read_audios_from_disk(test_input)
     assert len(response["output"]) == len(test_input["data"]["files"]), "The number of items in the output should match the number of input files."
 
 
@@ -97,7 +99,7 @@ def test_save_HF_dataset_to_disk() -> None:
         },
         "data": {
             "dataset": Dataset.from_dict({"pokemon": ["bulbasaur", "squirtle"], "type": ["grass", "water"]}),
-            "output_path": "../../data_for_testing/output_dataset"
+            "output_path": "./data_for_testing/output_dataset"
         }
     }
 
@@ -107,7 +109,7 @@ def test_save_HF_dataset_to_disk() -> None:
     except Exception as e:
         pytest.fail(f"Function raised an exception with valid input: {e}")
 
-    # shutil.rmtree("../../data_for_testing/output_dataset")
+    # shutil.rmtree("./data_for_testing/output_dataset")
 
 
 def test_upload_HF_dataset_to_HF_hub() -> None:
@@ -163,7 +165,7 @@ def test_read_local_HF_dataset() -> None:
             "service_name": "Datasets"
         },
         "data": {
-            "path": '../../data_for_testing/output_dataset'
+            "path": './data_for_testing/output_dataset'
         }
     }
 

src/tests/test_video_iotask_datasets.py

@@ -1,6 +1,7 @@
 """This module tests the video's IOTask class."""
 
 import os
+from typing import Any, Dict, List
 
 import pytest
 
@@ -39,7 +40,7 @@ def test_extract_audios_from_videos_input_errors() -> None:
         VideoIOTask().extract_audios_from_videos({
             "data": {
                 "files": ["/non/existent/path/file1.mp4"],
-                "output_folder": "../../data_for_testing",
+                "output_folder": "./data_for_testing",
                 "audio_format": "wav",
                 "audio_codec": "pcm_s16le"
             },
@@ -52,8 +53,8 @@ def test_extract_audios_from_videos_output_type() -> None:
     """Test the extract_audios_from_videos method to check if the output is of type list of strings."""
     test_input = {
         "data": {
-                "files": ["../../data_for_testing/video_48khz_stereo_16bits.mp4"],
-                "output_folder": "../../data_for_testing",
+                "files": ["./data_for_testing/video_48khz_stereo_16bits.mp4"],
+                "output_folder": "./data_for_testing",
                 "audio_format": "wav",
                 "audio_codec": "pcm_s16le"
         },
@@ -74,18 +75,18 @@ def test_read_audios_from_disk_output_dimensions() -> None:
      
     This test checks if the dimensions of the output list of audio files match the input list of video files.
     """
-    test_input = {
+    test_input: Dict[str, Any] = {
         "data": {
-            "files": ["../../data_for_testing/video_48khz_stereo_16bits.mp4"],
-            "output_folder": "../../data_for_testing",
+            "files": ["./data_for_testing/video_48khz_stereo_16bits.mp4"],
+            "output_folder": "./data_for_testing",
             "audio_format": "wav",
             "audio_codec": "pcm_s16le"
         },
         "service": {
             "service_name": "ffmpeg"
         }
     }
-    response = VideoIOTask().extract_audios_from_videos(test_input)
+    response: Dict[str, List[str]] = VideoIOTask().extract_audios_from_videos(test_input)
     assert len(response["output"]) == len(test_input["data"]["files"]), "The number of items in the output should match the number of input files."
 
     # Clean up