stimmenfryslan/stimmen/geojson.py

from pygeoif.geometry import mapping
from shapely.geometry import shape
from shapely.geometry.point import Point


def merge_features(geojson, condition, aggregate={}):
    """Merge the geometries using shapely's union for all the geojson's features that
    meet the condition, condition get's passed an item of feature. Then aggregate the properties
    in the aggregate dict using a function that get as input the list of property values off alle
    matched features. Operates inplace."""
    indices = [index for index, feature in enumerate(geojson['features']) if condition(feature)]
    if len(indices) == 0: # also if there is one index, we
        return geojson
    properties = {
        prop: agg([
            properties_[prop]
            for index in indices
            for properties_ in [geojson['features'][index]['properties']]
            if prop in properties_
        ])
        for prop, agg in aggregate.items()
    }
    properties.update({
        key: value
        for index in indices
        for key, value in geojson['features'][index]['properties'].items()
        if key not in aggregate
    })
    if len(indices) == 1:
        geojson['features'][indices[0]]['properties'] = properties
        return geojson

    union = shape(geojson['features'][indices[0]]['geometry'])
    for index in indices[1:]:
        union = union.union(shape(geojson['features'][index]['geometry']))

    for index in indices[::-1]: # reverse, such that the 'todo' indices willnot change.
        del geojson['features'][index]

    geojson['features'].append({
        'geometry': mapping(union),
        'properties': properties
    })
    return geojson


def inject_geojson_regions_into_dataframe(
        geojson, dataframe,
        latitude_column='latitude', longitude_column='longitude',
        region_name_property='name',
        region_name_column='region'
):
    """adds a region_name_column column to the dataframe with the region name as specified
    in the region_name_property of the geojson, by checking which geojson feature geometrically
    contains the longitude and latitude of the dataframe's row. This allows for faster cross
    reference between the geojson and the dataframe compared to always checking shape-point
    containment when cross referencing. Operates in place."""
    shapes = {
        feature['properties'][region_name_property]: shape(feature['geometry'])
        for feature in geojson['features']
    }

    def get_region_name(point):
        nonlocal shapes
        for region_name, region_shape in shapes.items():
            if region_shape.contains(point):
                return region_name

    point_to_region_name = {
        (latitude, longitude): get_region_name(point)
        for latitude, longitude in set(zip(dataframe[latitude_column], dataframe[longitude_column]))
        for point in [Point(longitude, latitude)] # alias
    }

    dataframe[region_name_column] = [
        point_to_region_name[(latitude, longitude)]
        for latitude, longitude in zip(dataframe[latitude_column], dataframe[longitude_column])
    ]
    return dataframe