run-20260619-015555Z / dc-l1-capetown-waterway-nulls

0:00

Hi, first time asking for help with this. The contractor delivered capetown_waterways.geojson but some features are unusable for our mapping pipeline. Drop any features that cannot represent a valid, usable waterway (a feature needs both a drawable geometry and a proper type to be useful) and keep the rest. I'd also like a top-level dropped_count so I can show my supervisor what we're flagging back at the contractor. Filename waterways_clean.geojson, please. The output should contain only LineString features, each preserving all original columns, with feature_id as the key field. The FeatureCollection must include a top-level dropped_count foreign member (an integer) reporting how many features were removed.

0:00

System info

{
  "adapter": "openrouter",
  "model": "google/gemma-4-26b-a4b-it",
  "base_url": "https://openrouter.ai/api/v1"
}

0:01

Bash ls -F

ls -F

capetown_waterways.geojson

0:26

Write /work/solve.py

content · 73 lines · py

import geopandas as gpd
import json

def clean_waterways(input_file, output_file):
    # Read the input file
    gdf = gpd.read_file(input_file)
    
    initial_count = len(gdf)
    
    # 1. Keep only LineString features
    # Note: MultiLineString might be considered "waterways" in a real-world scenario, 
    # but the instruction specifically says "The output should contain only LineString features".
    # I will stick strictly to the requirement.
    
    # Filtering for LineString
    # In GeoPandas/Shapely, geometry.geom_type will be 'LineString' or 'MultiLineString'
    mask_linestring = gdf.geometry.geom_type == 'LineString'
    
    # 2. Keep only valid geometries
    mask_valid = gdf.geometry.is_valid
    
    # 3. Ensure it has a proper 'type' column. 
    # The prompt says "a feature needs both a drawable geometry and a proper type to be useful"
    # I need to check if a 'type' column exists and if it's not null/empty.
    # If 'type' doesn't exist, it might be a problem. Let's check the columns first.
    
    # Let's refine the mask.
    # Based on the prompt: "Drop any features that cannot represent a valid, usable waterway 
    # (a feature needs both a drawable geometry and a proper type to be useful)"
    
    # First, check if 'type' column exists. If not, all features are technically "invalid" 
    # by the prompt's definition of needing a "proper type". 
    # However, usually, "type" is a column.
    
    if 'type' in gdf.columns:
        mask_type = gdf['type'].notna() & (gdf['type'].astype(str).str.strip() != "")
    else:
        # If 'type' is missing, we can't satisfy the "needs a proper type" requirement.
        # But let's see what's in the file first.
        mask_type = False 

    final_mask = mask_linestring & mask_valid & mask_type
    
    gdf_clean = gdf[final_mask].copy()
    dropped_count = initial_count - len(gdf_clean)
    
    # The prompt says "feature_id as the key field". 
    # GeoJSON doesn't have a "key field" concept in the same way a DB does, 
    # but it might mean the index or a specific column. 
    # However, standard GeoJSON preserves all columns.
    
    # To add a top-level 'dropped_count' to the FeatureCollection, 
    # we can't just use gdf.to_file() because that produces a standard FeatureCollection.
    # We need to manually construct the JSON.
    
    # Convert cleaned GDF to GeoJSON dict
    geojson_dict = gdf_clean.to_json()
    data = json.loads(geojson_dict)
    
    # Add the top-level member
    data['dropped_count'] = dropped_count
    
    # Write to file
    with open(output_file, 'w') as f:
        json.dump(data, f)
        
    print(f"Initial: {initial_count}")
    print(f"Dropped: {dropped_count}")
    print(f"Kept: {len(gdf_clean)}")

if __name__ == "__main__":
    clean_waterways('capetown_waterways.geojson', 'waterways_clean.geojson')

Wrote /work/solve.py (2870 chars).