crs-l1-nyc-webmercator-cycleways

Rerun

# /// script
# requires-python = ">=3.12"
# dependencies = ["geopandas", "pyarrow", "shapely"]
# ///
import geopandas as gpd

def convert_to_wgs84(input_path, output_path):
    # Read the input file
    gdf = gpd.read_parquet(input_path)
    
    # Convert to WGS84 (EPSG:4326)
    gdf = gdf.to_crs(epsg=4326)
    
    # Ensure every geometry is a LineString (not MultiLineString)
    # We can achieve this by exploding MultiLineStrings into individual LineStrings
    gdf = gdf.explode(index_parts=False)
    
    # Ensure we only have LineStrings (though explode should handle it, 
    # let's be safe and filter if necessary, though the prompt implies they are lines)
    # The requirement is: "Keep every geometry as a plain LineString — do not upcast to MultiLineString"
    # Exploding MultiLineStrings will result in multiple rows if a single feature was a MultiLineString.
    # However, the prompt says "Each feature must preserve all original columns, with id as the feature identity key."
    # If we explode, one original 'id' will become multiple rows with the same 'id'. 
    # This is generally how one handles MultiLineString to LineString conversion while preserving attributes.
    
    # Re-check if there are any MultiLineStrings left (just in case)
    # and if they are indeed LineStrings.
    
    # Save the result to GeoParquet
    gdf.to_parquet(output_path)
    print(f"Converted {input_path} to {output_path}")
    print(f"Original CRS: {gdf.crs}") # This will show the new CRS after conversion

if __name__ == "__main__":
    input_file = "nyc_cycleways_webmercator.geoparquet"
    output_file = "nyc_cycleways_wgs84.geoparquet"
    convert_to_wgs84(input_file, output_file)

# /// script
# requires-python = ">=3.12"
# dependencies = ["geopandas", "pyarrow", "shapely"]
# ///
import geopandas as gpd
from shapely.geometry import LineString, MultiLineString

def convert_to_wgs84(input_path, output_path):
    # Read the input file
    gdf = gpd.read_parquet(input_path)
    
    # Convert to WGS84 (EPSG:4326)
    gdf = gdf.to_crs(epsg=4326)
    
    # The user says: "Keep every geometry as a plain LineString — do not upcast to MultiLineString."
    # "Each feature must preserve all original columns, with id as the feature identity key."
    # If a feature is a MultiLineString, we must split it into individual LineStrings.
    # Using explode(index_parts=False) is the standard way to do this in GeoPandas.
    # This will result in multiple rows having the same 'id'.
    
    gdf = gdf.explode(index_parts=False)
    
    # Ensure that we actually have only LineStrings.
    # Even if it was a Point or Polygon, we should ensure it's a LineString for the client?
    # The instruction says "Keep every geometry as a plain LineString", which might imply
    # the input is all lines, but we should be careful.
    # However, if a LineString is exploded, it stays a LineString.
    
    # Verify if any non-LineString geometries remain (optional but good for debugging)
    # non_lines = gdf[gdf.geometry.type != 'LineString']
    # if not non_lines.empty:
    #     print("Warning: Non-LineString geometries found after explosion.")

    # Save the result to GeoParquet
    gdf.to_parquet(output_path)
    print(f"Converted {input_path} to {output_path}")

if __name__ == "__main__":
    input_file = "nyc_cycleways_webmercator.geoparquet"
    output_file = "nyc_cycleways_wgs84.geoparquet"
    convert_to_wgs84(input_file, output_file)

# /// script
# requires-python = ">=3.12"
# dependencies = ["geopandas", "pyarrow", "shapely"]
# ///
import geopandas as gpd

def check_geometries(file_path):
    gdf = gpd.read_parquet(file_path)
    print(f"File: {file_path}")
    print(f"Total features: {len(gdf)}")
    print(f"Unique IDs: {gdf['id'].nunique()}")
    print(f"Geometry types: {gdf.geometry.type.value_counts().to_dict()}")
    print(f"CRS: {gdf.crs}")

if __name__ == "__main__":
    check_geometries("nyc_cycleways_wgs84.geoparquet")