feat: display a spinner while importing expensive things

hfgl/cli.py

@@ -7,6 +7,7 @@
     preprocess_base_command_interface,
     train_base_command_interface,
 )
+from everyvoice.utils import spinner
 from loguru import logger
 from merge_args import merge_args
 
@@ -33,9 +34,10 @@ def preprocess(
     ),
     **kwargs,
 ):
-    from everyvoice.base_cli.helpers import preprocess_base_command
+    with spinner():
+        from everyvoice.base_cli.helpers import preprocess_base_command
 
-    from .config import HiFiGANConfig
+        from .config import HiFiGANConfig
 
     preprocess_base_command(
         model_config=HiFiGANConfig,
@@ -47,11 +49,12 @@ def preprocess(
 @app.command()
 @merge_args(train_base_command_interface)
 def train(**kwargs):
-    from everyvoice.base_cli.helpers import train_base_command
+    with spinner():
+        from everyvoice.base_cli.helpers import train_base_command
 
-    from .config import HiFiGANConfig
-    from .dataset import HiFiGANDataModule
-    from .model import HiFiGAN
+        from .config import HiFiGANConfig
+        from .dataset import HiFiGANDataModule
+        from .model import HiFiGAN
 
     train_base_command(
         model_config=HiFiGANConfig,
@@ -86,9 +89,10 @@ def export(
 ):
     import os
 
-    import torch
+    with spinner():
+        import torch
 
-    from .utils import sizeof_fmt
+        from .utils import sizeof_fmt
 
     orig_size = sizeof_fmt(os.path.getsize(model_path))
     vocoder_ckpt = torch.load(model_path, map_location=torch.device("cpu"))
@@ -132,11 +136,12 @@ def synthesize(
     """Given some Mel spectrograms and a trained model, generate some audio. i.e. perform *copy synthesis*"""
     import sys
 
-    import torch
-    from pydantic import ValidationError
-    from scipy.io.wavfile import write
+    with spinner():
+        import torch
+        from pydantic import ValidationError
+        from scipy.io.wavfile import write
 
-    from .utils import load_hifigan_from_checkpoint, synthesize_data
+        from .utils import load_hifigan_from_checkpoint, synthesize_data
 
     device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
     checkpoint = torch.load(generator_path, map_location=device)