Python: bokeh not generating world's countries map as expected

Question

I am trying to generate world's countries map colored by some values but some countries are missing from the generated map. Here are my codes:

Get the geojson data

import urllib
url = 'https://raw.githubusercontent.com/datasets/geo-boundaries-world-110m/master/countries.geojson'
testfile = urllib.URLopener()
if os.path.exists('countries.json'):
    print "file already exists"
else:
    testfile.retrieve(url, "countries.json")

Use bokeh to generate the map

import json,pandas
from collections import OrderedDict
from bokeh.plotting import figure, show, output_file, ColumnDataSource
from bokeh.models import HoverTool
user_by_country_count = pandas.read_csv('data.csv')

with open('countries.json','r') as f:
    geodata = json.load(f)
f.close()
geodata_features = geodata['features']
country_xs = []
country_ys = []
country_names = []
country_num_users = []
country_colors = []
colors = ['#CCE5FF','#CCCCFF','#9999FF','#6666FF','#3333FF',
         '#0000FF','#0000CC','#000099','#000066','#0000CC']

for aCountry in geodata_features:
    coords = aCountry['geometry']['coordinates'][0]
    country_xs.append(map(lambda x:x[0],coords))
    country_ys.append(map(lambda x:x[1],coords))
    cName = aCountry['properties']['name']
    country_names.append(cName)
    if cName in user_by_country_count['Country'].values:            
        num_users = user_by_country_count['Count'][user_by_country_count[user_by_country_count.Country==cName].index[0]]
        country_num_users.append(num_users)
        country_colors.append(colors[int(np.log(num_users))])
    else:
        country_num_users.append(0)
        country_colors.append("#00FF80")
source = ColumnDataSource(
    data = dict(
        x=country_xs,
        y=country_ys,
        color=country_colors,
        name=country_names,
        users=country_num_users,
    )
)
output_file("global.html", title="global.py example")
TOOLS="pan,wheel_zoom,box_zoom,reset,hover,save"
p = figure(title="Upwork Users Location", tools=TOOLS)
p.patches('x', 'y',
    fill_color='color', fill_alpha=0.7,
    line_color="white", line_width=0.5,
    source=source)
hover = p.select(dict(type=HoverTool))
hover.point_policy = "follow_mouse"
hover.tooltips = OrderedDict([
    ("Name", "@name"),
    ("Number of Users", "@users"),
])
show(p)

I suspect that the problem might be in the geojson data. When I looked carefully, it looks like some coordinates are given as list of numbers while others are given as list of list of numbers. But this geojson was used before by many people so I wonder if anybody else would have encountered a similar issue.

Answer 1

Argentina and some other countries are of the type MultiPolygon, Brazil for example is of the type Polygon. Countries that have islands or separate lands are of the type MultiPolygon. So coordinates for MultiPolygon should contain one more level of arrays, and you should handle that:

#!/usr/bin/python2
import json,pandas
from collections import OrderedDict
from bokeh.plotting import figure, show, output_file, ColumnDataSource
from bokeh.models import HoverTool
import math
user_by_country_count = pandas.read_csv('data.csv')

with open('countries.json','r') as f:
    geodata = json.load(f)

geodata_features = geodata['features']
country_xs = []
country_ys = []
country_names = []
country_num_users = []
country_colors = []
colors = ['#CCE5FF','#CCCCFF','#9999FF','#6666FF','#3333FF',
         '#0000FF','#0000CC','#000099','#000066','#0000CC']

for aCountry in geodata_features:
    cName = aCountry['properties']['name']
    country_names.append(cName)

    geometry_type = aCountry['geometry']['type']
    if geometry_type == "MultiPolygon":
        for poly_coords in aCountry['geometry']['coordinates']:
            coords = poly_coords[0]

            country_xs.append(map(lambda x:x[0],coords))
            country_ys.append(map(lambda x:x[1],coords))
    else:
        coords = aCountry['geometry']['coordinates'][0]
        country_xs.append(map(lambda x:x[0],coords))
        country_ys.append(map(lambda x:x[1],coords))

    if cName in user_by_country_count['Country'].values:
        num_users = user_by_country_count['Count'][user_by_country_count[user_by_country_count.Country==cName].index[0]]
        country_num_users.append(num_users)
        country_colors.append(colors[int(math.log(num_users))])
    else:
        country_num_users.append(0)
        country_colors.append("#00FF80")

source = ColumnDataSource(
    data = dict(
        x=country_xs,
        y=country_ys,
        color=country_colors,
        name=country_names,
        users=country_num_users,
    )
)
output_file("global.html", title="global.py example")
TOOLS="pan,wheel_zoom,box_zoom,reset,hover,save"
p = figure(title="Upwork Users Location", tools=TOOLS)
p.patches('x', 'y',
    fill_color='color', fill_alpha=0.7,
    line_color="white", line_width=0.5,
    source=source)
hover = p.select(dict(type=HoverTool))
hover.point_policy = "follow_mouse"
hover.tooltips = OrderedDict([
    ("Name", "@name"),
    ("Number of Users", "@users"),
])
show(p)