app.py

import gradio as gr
import geopandas as gpd
import pandas as pd
import os
import matplotlib.pyplot as plt
from shapely.geometry import shape
from shapely.ops import unary_union
#from datasets import load_dataset

#ds = load_dataset('psalama/NYC_sensitive_sites', data_files=data_files)

def process_buildings(input_gdf, sensitive_sites_gdf, default_building_height_m, multiplier_factor):
    # List to store all intersected sensitive sites
    intersected_sites = []
    
    # List to store all buffers
    buffers = []

    intersection_desc = ""
    
    # Iterate over each building in the input file
    for idx, building in input_gdf.iterrows():
        building_name = building.get('building_name', 'Unnamed building')
        
        # If the 'building_height' field exists and its value is not null or zero for this building,
        # use it as the building height. Otherwise, use the default building height provided by the user.
        if 'building_height' in building and pd.notnull(building['building_height']) and building['building_height'] != 0:
            building_height_m = building['building_height'] * 0.3048
        else:
            building_height_m = default_building_height_m

        buffer_distance_m = building_height_m * multiplier_factor

        # Convert building's geometry to EPSG:3857 for accurate meter-based distance measurement
        building_geometry = gpd.GeoSeries([building['geometry']], crs="EPSG:4326")
        building_geometry_m = building_geometry.to_crs("EPSG:3857")

        # Create a buffer around the building and convert it to a GeoDataFrame
        building_buffer = building_geometry_m.buffer(buffer_distance_m)
        building_buffer_gdf = gpd.GeoDataFrame(geometry=building_buffer, crs="EPSG:3857")
        building_buffer_gdf = building_buffer_gdf.to_crs("EPSG:4326")

        # Convert back to feet for storing and printing, rounding to the nearest foot
        building_height_ft = round(building_height_m / 0.3048)
        buffer_distance_ft = round(buffer_distance_m / 0.3048)

        # Assign additional attributes
        building_buffer_gdf['building_name'] = building_name
        building_buffer_gdf['building_height'] = building_height_ft
        building_buffer_gdf['buffer_distance'] = buffer_distance_ft

        buffers.append(building_buffer_gdf)

        # Check if the buffer intersects with any sensitive sites
        intersects = gpd.overlay(building_buffer_gdf, sensitive_sites_gdf, how='intersection')
        
        if not intersects.empty:
            building_intersect_desc = f"Building {idx} ({building_name}), height: {building_height_ft}, buffer distance: {buffer_distance_ft} is in the vicinity of a sensitive site."
            intersected_sites.append(intersects)
        else:
            building_intersect_desc = f"Building {idx} ({building_name}), height: {building_height_ft}, buffer distance: {buffer_distance_ft} is not in the vicinity of any sensitive sites."

        if intersection_desc == "":
            intersection_desc = building_intersect_desc
        else:
            intersection_desc += "\n" + building_intersect_desc
    
    return buffers, intersected_sites, intersection_desc

def get_max_extent(*gdfs):
    minx = min(gdf.total_bounds[0] for gdf in gdfs)
    miny = min(gdf.total_bounds[1] for gdf in gdfs)
    maxx = max(gdf.total_bounds[2] for gdf in gdfs)
    maxy = max(gdf.total_bounds[3] for gdf in gdfs)
    
    return minx, miny, maxx, maxy

def create_plot(filename, extent, *gdfs):
    fig, ax = plt.subplots(1, 1)
    
    colors = ['blue', 'red', 'green', 'purple', 'orange', 'yellow'] # Extend this list as needed

    for idx, gdf in enumerate(gdfs):
        gdf.plot(ax=ax, color=colors[idx % len(colors)]) # Cycle through colors

    ax.set_xlim(extent[0], extent[2])
    ax.set_ylim(extent[1], extent[3])

    plt.savefig(filename)

def ss_intersect(geojson1, ss_geoselect, multiplier_factor, default_building_height): 
    # Read the GeoJSON files
    input_gdf = gpd.read_file(geojson1.name)

    # Check that CRS is EPSG:4326
    if input_gdf.crs.to_epsg() != 4326:
        raise ValueError("Input GeoJSON files must be in CRS EPSG:4326")
    
    if ss_geoselect==0:
        sensitive_sites_gdf = gpd.read_file("sensitive_sites/NYC_Parks_Properties.geojson")
    else:
        sensitive_sites_gdf = gpd.read_file("sensitive_sites/NYC_Parks_Zones.geojson")

    default_building_height_m = default_building_height * 0.3048

    buffers, intersected_sites, intersection_desc = process_buildings(input_gdf, sensitive_sites_gdf, default_building_height_m, multiplier_factor)

    # Concatenate all buffer GeoDataFrames and save as a GeoJSON file
    buffers_gdf = pd.concat(buffers, ignore_index=True)
    buffers_gdf = buffers_gdf.to_crs("EPSG:4326")
    buffers_gdf.to_file("building_buffers.geojson", driver='GeoJSON')

    # Concatenate all intersected sensitive sites and save as a GeoJSON file
    if intersected_sites:
        intersected_sites_gdf = pd.concat(intersected_sites, ignore_index=True)
        intersected_sites_gdf = intersected_sites_gdf.to_crs("EPSG:4326")
        intersected_sites_gdf.to_file("intersected_sensitive_sites.geojson", driver='GeoJSON')
    else:
        print("No buildings are in the vicinity of any sensitive sites.")

    # Perform the union operation if there is more than one buffer
    if len(buffers) > 1:
        # Perform a unary union on the geometry column of the GeoDataFrame
        buffer_union = unary_union(buffers_gdf['geometry'])

        # Create a new GeoDataFrame from the union result
        buffer_union_gdf = gpd.GeoDataFrame(geometry=[buffer_union], crs="EPSG:4326")

        # Save the union GeoDataFrame as a GeoJSON file
        buffer_union_gdf.to_file("buffer_union.geojson", driver='GeoJSON')


    # Calculate the maximum extent
    extent = get_max_extent(input_gdf, buffers_gdf)

    # Create and save the plot
    create_plot('output_image.png', extent, buffer_union_gdf, intersected_sites_gdf, input_gdf)
            
    # Return the image
    return 'output_image.png', "building_buffers.geojson", "buffer_union.geojson", intersection_desc


iface = gr.Interface(
    fn=ss_intersect, 
    inputs=[
        gr.inputs.File(label="Building Footprints GeoJSON"),
        gr.Radio(["Parks Properties", "Park Zones"], label="Which Sensitive Sites?", info="From NYC DPR", type="index"),
        gr.inputs.Slider(minimum=0.0, maximum=10.0, default=4.3, label="Building Height Multiplier"),
        gr.inputs.Number(default=200, label="Default Building Height"), #Can I make this optional?
    ], 
    outputs=[
        gr.outputs.Image(type="pil", label="Result Image"),
        gr.outputs.File(label="Building Buffers"),
        gr.outputs.File(label="Union of Building Buffers"),
        gr.outputs.Textbox(label="Building and Sensitive Site Vicinities"),
    ],
    examples=[
        ["files/building4test.geojson", "Parks Properties", 4.3, 200],
    ],
    title="Shadow Proximity",
    description="Upload proposed building footprints in a GeoJSON file and select a numeric value to get the building proximity prediction.",
)

iface.launch()