"""Model browser - scan models directory for available models.""" from __future__ import annotations import jsonfrom pathlib import Pathfrom typing import Optional from pydantic import BaseModel  class ModelInfo(BaseModel):    """Information about a model in the models directory."""     name: str    path: str    size_gb: Optional[float] = None    architecture: Optional[str] = None    quantization: Optional[str] = None    context_length: Optional[int] = None    has_recipe: bool = False  def scan_models(models_dir: Path, recipe_paths: set[str]) -> list[ModelInfo]:    """    Scan models directory and return list of discovered models.     Args:        models_dir: Path to models directory        recipe_paths: Set of model paths that have recipes     Returns:        List of ModelInfo objects    """    if not models_dir.exists():        return []     models = []     # Look for subdirectories that appear to be models    # A model directory typically has config.json or .safetensors files    for item in models_dir.iterdir():        if not item.is_dir():            continue         # Check if this looks like a model directory        has_config = (item / "config.json").exists()        has_safetensors = any(item.glob("*.safetensors"))         if not (has_config or has_safetensors):            continue         # Extract metadata        model_path = str(item)        name = item.name        architecture = None        quantization = None        context_length = None        size_gb = None         # Try to read config.json for metadata        config_path = item / "config.json"        if config_path.exists():            try:                with open(config_path) as f:                    config = json.load(f)                    architecture = config.get("architectures", [None])[0]                    context_length = config.get("max_position_embeddings")            except Exception:                pass         # Infer quantization from name        name_lower = name.lower()        for quant in ["awq", "gptq", "gguf", "fp16", "bf16", "int8", "int4"]:            if quant in name_lower:                quantization = quant.upper()                break         # Calculate size        try:            total_bytes = sum(f.stat().st_size for f in item.rglob("*") if f.is_file())            size_gb = round(total_bytes / (1024**3), 1)        except Exception:            pass         # Check if has recipe        has_recipe = model_path in recipe_paths         models.append(            ModelInfo(                name=name,                path=model_path,                size_gb=size_gb,                architecture=architecture,                quantization=quantization,                context_length=context_length,                has_recipe=has_recipe,            )        )     # Sort by name    models.sort(key=lambda m: m.name.lower())    return models