inspect_columns.py

# inspect_pierce_columns_v2.py
#
# A focused diagnostic script to inspect the columns of GIS layers.
# VERSION 2: Adds 'ignore_geometry=True' to handle corrupt or non-standard
# geometry without crashing, allowing us to see the attribute columns.

from pathlib import Path
import geopandas as gpd
import fiona

class CONFIG:
    SOURCE_DATA_DIR = Path('pierce_gis_data')
    
    # --- ACTION REQUIRED ---
    # List the layers you want to inspect.
    LAYERS_TO_INSPECT = [
        'Tax_Parcels',
        # Add other layer names here if you have them.
    ]

def find_data_sources(source_dir: Path):
    """Finds all potential data sources (.gdb folders)."""
    return list(source_dir.glob('*.gdb'))

def find_and_load_layer_attributes(gdb_paths: list[Path], layer_name: str):
    """
    Finds a layer and loads ONLY its attribute columns, ignoring geometry.
    """
    for gdb_path in gdb_paths:
        try:
            available_layers = [lyr.lower() for lyr in fiona.listlayers(gdb_path)]
            if layer_name.lower() in available_layers:
                print(f"  -> Found in Geodatabase: {gdb_path.name}")
                # This is the critical change:
                df = gpd.read_file(gdb_path, layer=layer_name, ignore_geometry=True)
                return df
        except Exception as e:
            print(f"  -> ERROR reading attributes from {gdb_path.name}: {e}")
            continue
    return None

def run_inspection():
    """Loads each specified layer's attributes and prints its columns."""
    print(f"--- 🔍 Starting Column Inspection (Ignoring Geometry) 🔍 ---\n")
    gdb_paths = find_data_sources(CONFIG.SOURCE_DATA_DIR)

    if not gdb_paths:
        print(f"FATAL: No '.gdb' directories found in '{CONFIG.SOURCE_DATA_DIR}'.")
        return

    for layer_name in CONFIG.LAYERS_TO_INSPECT:
        print(f"--- Inspecting Attributes For Layer: '{layer_name}' ---")
        df = find_and_load_layer_attributes(gdb_paths, layer_name)
        
        if df is not None:
            print("Columns found:")
            for col in df.columns:
                print(f"  -> {col}")
        else:
            print(f"Layer '{layer_name}' not found or attributes could not be read.")
        print("-" * 40 + "\n")

    print("--- ✅ Inspection Complete ✅ ---")
    print("Use this list of columns to configure the main builder script.")

if __name__ == '__main__':
    run_inspection()