Create config boilerplate for apps (#11)

src/spira_training/apps/feature_engineering/configs/audio_config.py

@@ -0,0 +1,13 @@
+from pathlib import Path
+
+from pydantic import BaseModel
+
+class DatasetPaths(BaseModel):
+    patients_csv: Path
+    controls_csv: Path
+    noises_csv: Path
+
+class AudioConfig(BaseModel):
+    dataset_paths: DatasetPaths
+    hop_length: int
+    normalize: bool

src/spira_training/apps/feature_engineering/configs/audio_feature_transformer_config.py

@@ -0,0 +1,31 @@
+from pydantic import BaseModel
+
+
+class AudioFeatureTransformerOptions(BaseModel):
+    use_noise: bool
+    use_overlapping: bool
+    use_padding: bool
+    use_mixture: bool
+
+class NoisyAudioFeatureTransformerConfig(BaseModel):
+    num_noise_control: int
+    num_noise_patient: int
+    noise_max_amp: float
+    noise_min_amp: float
+
+class OverlappedAudioFeatureTransformerConfig(BaseModel):
+    window_length: int
+    step_size: int
+
+class MixedAudioFeatureTransformerConfig(BaseModel):
+    alpha: float
+    beta: float
+
+class AudioFeatureTransformersCollection(BaseModel):
+    noisy_audio: NoisyAudioFeatureTransformerConfig
+    overlapped_audio: OverlappedAudioFeatureTransformerConfig
+    mixed_audio: MixedAudioFeatureTransformerConfig
+
+class AudioFeatureTransformerConfig(BaseModel):
+    options: AudioFeatureTransformerOptions
+    audio_feature_transformers: AudioFeatureTransformersCollection

src/spira_training/apps/feature_engineering/configs/audio_processor_config.py

@@ -0,0 +1,43 @@
+from pydantic import BaseModel
+
+
+class AudioProcessorType(BaseModel):
+    MFCC = "mfcc"
+    SPECTROGRAM = "spectrogram"
+    MELSPECTROGRAM = "melspectrogram"
+
+class MFCCAudioProcessorConfig:
+    sample_rate: int
+    num_mels: int
+    num_mfcc: int
+    log_mels: bool
+    n_fft: int
+    win_length: int
+
+class SpectrogramAudioProcessorConfig:
+    sample_rate: int
+    num_mels: int
+    mel_fmin: float
+    mel_fmax: None
+    num_mfcc: int
+    log_mels: bool
+    n_fft: int
+    num_freq: int
+    win_length: int
+
+class MelspectrogramAudioProcessorConfig:
+    sample_rate: int
+    num_mels: int
+    mel_fmin: float
+    mel_fmax: None
+    num_mfcc: int
+    log_mels: bool
+    n_fft: int
+    num_freq: int
+    win_length: int
+
+class AudioProcessorConfig(BaseModel):
+    feature_type: AudioProcessorType
+    mfcc: MFCCAudioProcessorConfig
+    spectrogram: SpectrogramAudioProcessorConfig
+    melspectrogram: MelspectrogramAudioProcessorConfig

src/spira_training/apps/feature_engineering/configs/feature_engineering_config.py

@@ -0,0 +1,10 @@
+from pydantic import BaseModel
+
+from audio_config import AudioConfig
+from audio_processor_config import AudioProcessorConfig
+from audio_feature_transformer_config import AudioFeatureTransformerConfig
+
+class FeatureEngineeringConfig(BaseModel):
+    audio: AudioConfig
+    audio_processor: AudioProcessorConfig
+    audio_feature_transformer: AudioFeatureTransformerConfig

src/spira_training/apps/feature_engineering/main.py → src/spira_training/apps/feature_engineering/feature_engineering.py

@@ -1,16 +1,14 @@
-from src.spira_training.apps.feature_engineering.app.app import App
+from src.spira_training.shared.core.services.feature_engineering_service import FeatureEngineeringService
 from src.spira_training.shared.core.services.randomizer import Randomizer
 from src.spira_training.shared.ports.config_loader import ConfigLoader
 from src.spira_training.shared.ports.dataset_repository import DatasetRepository
-
-
 def main():
     config = ConfigLoader.load()
     randomizer = Randomizer.initialize_random()
     dataset_repository = DatasetRepository()
+    service = FeatureEngineeringService(config, randomizer, dataset_repository)
 
-    app = App(config, randomizer, dataset_repository)
-    app.execute()
+    service.execute()
 
 if __name__ == "__main__":
     main()

src/spira_training/apps/model_evaluation/model_evaluation_config.py

@@ -0,0 +1,2 @@
+class ModelEvaluationConfig():
+    pass

src/spira_training/apps/model_publish/model_publish_config.py

@@ -0,0 +1,2 @@
+class ModelPublishConfig:
+    pass

src/spira_training/shared/core/models/config.py

@@ -1,2 +1,10 @@
+from src.spira_training.apps.feature_engineering.configs.feature_engineering_config import FeatureEngineeringConfig
+from src.spira_training.apps.model_evaluation.model_evaluation_config import ModelEvaluationConfig
+from src.spira_training.apps.model_publish.model_publish_config import ModelPublishConfig
+
+
 class Config:
-    pass
+    feature_engineering_config: FeatureEngineeringConfig
+    model_trainer_config: ModelTrainerConfig
+    model_evaluation_config: ModelEvaluationConfig
+    model_publish_config: ModelPublishConfig

src/spira_training/shared/core/services/feature_engineering_service.py

@@ -0,0 +1,22 @@
+from src.spira_training.apps.feature_engineering.configs.feature_engineering_config import FeatureEngineeringConfig
+from src.spira_training.shared.core.interfaces.random import Random
+from src.spira_training.shared.ports import dataset_repository
+from src.spira_training.shared.ports.dataset_repository import DatasetRepository
+
+class FeatureEngineeringService:
+    def __init__(self, config: FeatureEngineeringConfig, randomizer: Random, dataset_repository: DatasetRepository):
+        self.config = config
+        self.randomizer = randomizer
+        self.dataset_repository = dataset_repository
+
+    def execute(self) -> None:
+        patients_inputs, controls_inputs, noises = self._load_data(self.config)
+        dataset = self._generate_dataset(self.config, self.randomizer, patients_inputs, controls_inputs, noises)
+        dataset_repository.save_dataset(dataset, self.config.paths.dataset)
+
+    def _load_data(self, config: FeatureEngineeringConfig):
+        pass
+
+    def _generate_dataset(self):
+        pass
+