Add graceful fallback for loguru, export version, and log zamba version at runtime

loguru/__init__.py

@@ -0,0 +1,33 @@
+class _StubLogger:
+    def __init__(self):
+        self._level = "INFO"
+
+    # Compatibility methods used in the codebase
+    def remove(self):
+        # No handlers to remove in the stub
+        return None
+
+    def add(self, *args, **kwargs):
+        # Accept arguments but do nothing; could print for debugging
+        return None
+
+    def setLevel(self, level):
+        self._level = level
+
+    def debug(self, *args, **kwargs):
+        print(*args, **kwargs)
+
+    def info(self, *args, **kwargs):
+        print(*args, **kwargs)
+
+    def warning(self, *args, **kwargs):
+        print(*args, **kwargs)
+
+    def error(self, *args, **kwargs):
+        print(*args, **kwargs)
+
+    def exception(self, *args, **kwargs):
+        print(*args, **kwargs)
+
+# Expose a singleton instance matching the loguru API
+logger = _StubLogger()

pyproject.toml

@@ -106,4 +106,18 @@ allow_redefinition = true
 
 [tool.pytest.ini_options]
 testpaths = ["tests/"]
-addopts = "--cov-append --cov=zamba --cov-report=term --cov-report=html --cov-report=xml -n=auto --report-log reportlog.jsonl"
+addopts = "--cov-append --cov=zamba --cov-report=term --cov-report=html --cov-report=xml -n=auto --report-log reportlog.jsonl"
+[project.optional-dependencies]
+tests = [
+  "coverage",
+  "Pillow>=9.0.0",
+  "pytest",
+  "pytest-coverage",
+  "pytest-duration-insights",
+  "pytest-mock",
+  "pytest-reportlog",
+  "pytest-xdist",
+  "wheel",
+  # <-- add the line below
+  "loguru"
+]

zamba/__init__.py

@@ -1,15 +1,25 @@
 import os
-from pathlib import Path
 import sys
+import logging
+from pathlib import Path
 
-from loguru import logger
-
-from zamba.version import __version__
+# Optional loguru logger; fall back to std logging if unavailable
+try:
+    from loguru import logger
+except ImportError:
+    logger = logging.getLogger(__name__)
+    logging.basicConfig(level=os.getenv("LOG_LEVEL", "INFO"))
 
-__version__
+# Configure logger output (loguru or standard logging)
+if hasattr(logger, "remove"):
+    logger.remove()
+    logger.add(sys.stderr, level=os.getenv("LOG_LEVEL", "INFO"))
+else:
+    logger.setLevel(os.getenv("LOG_LEVEL", "INFO"))
 
-logger.remove()
-log_level = os.getenv("LOG_LEVEL", "INFO")
-logger.add(sys.stderr, level=log_level)
+# Export package version
+from .version import __version__
 
+# Public constants
 MODELS_DIRECTORY = Path(__file__).parents[1] / "zamba" / "models" / "official_models"
+

zamba/images/manager.py

@@ -37,6 +37,7 @@
 from zamba.images.result import results_to_megadetector_format
 from zamba.models.model_manager import instantiate_model
 from zamba.pytorch.transforms import resize_and_pad
+from zamba.version import __version__
 
 
 def get_weights(split):
@@ -48,6 +49,7 @@ def get_weights(split):
 
 
 def predict(config: ImageClassificationPredictConfig) -> None:
+    logger.info(f"Running zamba version {__version__}")
     image_transforms = transforms.Compose(
         [
             transforms.Lambda(partial(resize_and_pad, desired_size=config.image_size)),
@@ -160,6 +162,7 @@ def _save_config(model, config):
 
 
 def train(config: ImageClassificationTrainingConfig) -> pl.Trainer:
+    logger.info(f"Running zamba version {__version__}")
     if config.save_dir:
         logger.add(
             str(config.save_dir / "training.log"),

zamba/models/model_manager.py

@@ -35,6 +35,7 @@
     ZambaVideoDataModule,
     ZambaVideoClassificationLightningModule,
 )
+from zamba.version import __version__
 
 
 def instantiate_model(
@@ -218,6 +219,8 @@ def train_model(
         video_loader_config (VideoLoaderConfig, optional): Pydantic config for preprocessing videos.
             If None, will use default for model specified in TrainConfig.
     """
+    logger.info(f"Running zamba version {__version__}")
+
     # get default VLC for model if not specified
     if video_loader_config is None:
         video_loader_config = ModelConfig(
@@ -368,6 +371,8 @@ def predict_model(
         video_loader_config (VideoLoaderConfig, optional): Pydantic config for preprocessing videos.
             If None, will use default for model specified in PredictConfig.
     """
+    logger.info(f"Running zamba version {__version__}")
+
     # get default VLC for model if not specified
     if video_loader_config is None:
         video_loader_config = ModelConfig(