"""LISBET Command Line Interface (CLI)"""
import argparse
import importlib
import inspect
import logging
from importlib.metadata import version as get_version
from typing import Any
from lisbet.cli.commands import (
configure_annotate_behavior_parser,
configure_compute_embeddings_parser,
configure_evaluate_model_parser,
configure_export_embedder_parser,
configure_fetch_dataset_parser,
configure_fetch_model_parser,
configure_model_info_parser,
configure_reduce_dimensions_parser,
configure_segment_motifs_parser,
configure_select_prototypes_parser,
configure_train_model_parser,
)
[docs]
def lazy_load_handler(
module_path: str, function_name: str, args: dict[str, Any]
) -> None:
"""Dynamically import and call a command handler."""
# Always use absolute imports from the lisbet root package
if not module_path.startswith("lisbet"):
module_path = f"lisbet{module_path}"
module = importlib.import_module(module_path)
handler = getattr(module, function_name)
# Filter arguments to match handler"s signature
valid_args = [p.name for p in inspect.signature(handler).parameters.values()]
filtered_args = {k: v for k, v in args.items() if k in valid_args}
logging.debug("CLI arguments for %s: %s", function_name, filtered_args)
# Execute handler
handler(**filtered_args)
[docs]
def app() -> None:
"""Main entry point for betman CLI."""
parser = argparse.ArgumentParser(
description="betman - Command line interface for LISBET",
add_help=False,
)
# Create basic arguments
parser.add_argument(
"--version",
action="version",
version=f"%(prog)s {get_version('lisbet')}",
help="Show LISBET's version number and exit",
)
parser.add_argument(
"-h",
"--help",
action="help",
default=argparse.SUPPRESS,
help="Show this help message and exit",
)
# Create subcommands
subparsers = parser.add_subparsers(title="commands", dest="command", required=True)
# Configure command parsers
commands = {
"train_model": {
"description": "Train a new classification model",
"module": ".cli.commands.train",
"function": "train_model",
"configure": configure_train_model_parser,
},
"annotate_behavior": {
"description": "Annotate behaviors using trained model",
"module": ".inference",
"function": "annotate_behavior",
"configure": configure_annotate_behavior_parser,
},
"compute_embeddings": {
"description": "Compute behavioral embeddings using trained model",
"module": ".inference",
"function": "compute_embeddings",
"configure": configure_compute_embeddings_parser,
},
"export_embedder": {
"description": "Export embedding model",
"module": ".io",
"function": "export_embedder",
"configure": configure_export_embedder_parser,
},
"reduce_dimensions": {
"description": "Perform dimensionality reduction with UMAP",
"module": ".unsupervised",
"function": "reduce_umap",
"configure": configure_reduce_dimensions_parser,
},
"segment_motifs": {
"description": "Segment behavioral motifs using HMM",
"module": ".unsupervised",
"function": "segment_hmm",
"configure": configure_segment_motifs_parser,
},
"select_prototypes": {
"description": "Select prototype behaviors from multiple annotations",
"module": ".postprocessing",
"function": "select_prototypes",
"configure": configure_select_prototypes_parser,
},
"fetch_dataset": {
"description": "Download public dataset from the internet",
"module": ".hub",
"function": "fetch_dataset",
"configure": configure_fetch_dataset_parser,
},
"fetch_model": {
"description": "Download public model from the internet",
"module": ".hub",
"function": "fetch_model",
"configure": configure_fetch_model_parser,
},
"model_info": {
"description": "Show information about a LISBET model config file",
"module": ".modeling",
"function": "model_info",
"configure": configure_model_info_parser,
},
"evaluate_model": {
"description": "Evaluate a classification model on a labeled dataset",
"module": ".evaluation",
"function": "evaluate",
"configure": configure_evaluate_model_parser,
},
}
# Add all subparsers
for cmd_name, cmd_config in commands.items():
cmd_parser = subparsers.add_parser(
cmd_name,
description=cmd_config["description"],
add_help=False,
formatter_class=RawDefaultsHelpFormatter,
)
cmd_parser.add_argument(
"-h",
"--help",
action="help",
default=argparse.SUPPRESS,
help="Show this help message and exit",
)
cmd_config["configure"](cmd_parser)
# Parse arguments
args = parser.parse_args()
# Setup logging
if args.log_level:
level = getattr(logging, args.log_level)
else:
level_map = {
0: logging.WARNING, # default
1: logging.INFO, # -v
2: logging.DEBUG, # -vv
}
level = level_map.get(args.verbose, logging.DEBUG)
logging.basicConfig(level=level)
logging.getLogger("matplotlib.font_manager").setLevel(logging.WARNING)
logging.getLogger("movement.io.load_poses").setLevel(logging.WARNING)
logging.getLogger("numba").setLevel(logging.WARNING)
# Get command configuration
cmd_config = commands[args.command]
# Temporary workaround for new-style commands
if cmd_config["function"] == "train_model":
module_path = f"lisbet{cmd_config['module']}"
function_name = cmd_config["function"]
module = importlib.import_module(module_path)
handler = getattr(module, function_name)
handler(dict(args._get_kwargs()))
else:
# Fallback
lazy_load_handler(
module_path=f"lisbet{cmd_config['module']}",
function_name=cmd_config["function"],
args=vars(args),
)
