import holoviews as hv
import dask.dataframe as dd
import datashader as ds, datashader.geo

from holoviews import opts
from holoviews.operation.datashader import datashade, rasterize

hv.extension('bokeh')

df = dd.read_parquet('../data/earthquakes.parq', engine='fastparquet').persist()

print('%s Rows' % len(df))
print('Columns:', list(df.columns))

x, y = ds.geo.lnglat_to_meters(df.longitude, df.latitude)
ddf = df.assign(x=x, y=y).persist()

points = hv.Points(ddf, ['x', 'y'])

datashade(points).opts(width=700, height=500, bgcolor="lightgray")

datashade.streams

# Exercise: Plot the earthquake locations ('longitude' and 'latitude' columns)
# Warning: Don't try to display hv.Points() directly; it's too big! Use datashade() or rasterize() for any display
# Optional: Change the cmap on the datashade operation to inferno

# from datashader.colors import inferno


from holoviews.element.tiles import EsriImagery

tiles = EsriImagery().opts(xaxis=None, yaxis=None, width=700, height=500)
tiles * datashade(points)

# Exercise: Overlay the earthquake data on top of the Wikipedia tile source

selected = points.select(depth=(None, 0))
selected.data = selected.data.persist()
tiles * rasterize(selected, aggregator=ds.mean('mag')).opts(colorbar=True)

# Exercise: Use the ds.min or ds.max aggregator to visualize other fields
# Optional: Eliminate outliers by using select


dset = hv.Dataset(ddf)
grouped = dset.to(hv.Points, ['x', 'y'], groupby=['type'], dynamic=True)
tiles.opts(alpha=0.4, bgcolor="black") * datashade(grouped).opts(
    opts.RGB(width=600, height=500, xaxis=None, yaxis=None, tools=['hover']))

# Exercise: Facet a subset of the types as an NdLayout
# Hint: You can reuse the existing grouped variable or select a subset before using the .to method