Starlark Geospatial sample

geo.py

# this is a Qri starlark transform script, use it with qri: https://qri.io
load("http.star", "http")
load("geo.star", "geo")

def download(ctx):
  # Download list of 311 complaints, currently capped to 10000 responses for testing purposes
  complaints = http.get("https://data.cityofnewyork.us/resource/fhrw-4uyv.json?$offset=0&$limit=10000")
  # Download the New York Borough Boundaries
  boros = http.get("http://data.beta.nyc//dataset/68c0332f-c3bb-4a78-a0c1-32af515892d6/resource/7c164faa-4458-4ff2-9ef0-09db00b509ef/download/42c737fd496f4d6683bba25fb0e86e1dnycboroughboundaries.geojson")

  return {
    "complaints" : complaints.json(),
    # send boros over as a JSON string, parseGeoJSON will do the decoding
    "boros": boros.text(),
  }


def transform(ds, ctx):
  complaints = ctx.download["complaints"]
  boundaries, properties = geo.parseGeoJSON(ctx.download["boros"])

  # bouroughs data specifies a number of polygons for each borough,
  # combine them all into MultiPolygons, one for each borough
  boro_names = [ boro['borough'] for boro in properties]
  geoms = reduce(append_polygon, zip(boro_names, boundaries), {})
  geoms = [geo.MultiPolygon(geoms[x]) for x in geoms]
  boro_names = list(set(boro_names))

  # dict of complaint-counts, keyed by borough name
  boro_counts = dict(zip(boro_names, [0]*len(boro_names)))
  
  for complaint in complaints:
    # only use complaints that have lat & lng values
    if 'latitude' in complaint and 'longitude' in complaint:
      point = geo.Point(float(complaint['longitude']), float(complaint['latitude']))
      
      for boro_name, geom in zip(boro_names, geoms):
        if geo.within(point, geom):
          boro_counts[boro_name] += 1
  
  ds.set_body(boro_counts)


def append_polygon(acc, prop):
  polys = acc.get(prop[0], [])
  polys.append(prop[1])
  acc[prop[0]] = polys
  return acc

def reduce(fn, l, v):
	for _, el in enumerate(l):
		v = fn(v, el)
	return v