Adjust lat/long for Norway?

[kaimikael]

Geo version

0.2.8

Orange version

3.27.1

Expected behavior

I think it would make more sense to have the default location of Norway to be in mainland Norway, say somewhere about 62° N, 8° E.

Actual behavior

Currently Norway is represented by a point on Svalbard, which, while admittedly Norwegian territory, is not where you’d look first.

Steps to reproduce the behavior

Show any data containing Norway with Geo Map.

Additional info (worksheets, data, screenshots, ...)

[ajdapretnar]

I could be that the point is computed as a central point due to Norway having some overseas territories, but unsure if that's the case. I was not able to find where to adjust this. :(

[kaimikael]

That would be in orange3-geo-master/orangecontrib/geo/geojson/admin1-NOR.json

[ajdapretnar]

I know, but where in that file? I was only able to find polygon descriptions, couldn't figure out where is the coordinate for the "center of Norway". If you can find it, feel free to submit a pull request. :)

[kaimikael]

Yes, I was too hasty. There is a lat/long, but it’s for Bouvet Island. Weird. I’ll dig a bit more.

[ajdapretnar]

Not just Bouvet, also other counties. @kernc Do you perhaps remember if there is a "center" coordinate or it is computed on-the-fly?

[kaimikael]

Yes, if there isn’t an explicit latitude and longitude given, a “representative point” is computed. So, it should be possible to add lat/long data to the file. I’ll fiddle around and see what I can come up with :-)

[kernc]

What I do remember is that JSONs are generated. Manual edits should probably be made to the generating code. 😅

orange3-geo/orangecontrib/geo/geojson/make-geojson.sh

Lines 81 to 141 in c54a6a8

	# Admin1
	# NOTE: Admin0 and Admin1 are matched on adm0_a3 field
	ogr2ogr -f GeoJSON \
	-select name,iso_a2,adm0_a3,adm1_code,fips,admin,code_hasc,type_en,latitude,longitude \
	out.json ne_10m_admin_1_states_provinces.shp
	sed -i -r 's/"type_en"/"type"/g;
	s/"code_hasc"/"hasc"/g;
	s/"adm1_code"/"_id"/g;
	s/"-1"/null/g;
	s/"-99"/null/g;' out.json
	simplify .7 out.json admin1.json
	rm out.json
	python3 -c "
	import json
	from math import isnan
	from itertools import groupby
	from collections import OrderedDict
	CC = {}
	for line in open('iso3166a2a3.txt'):
	cc2, cc3 = line.split()
	CC[cc3] = cc2
	with open('admin1.json') as f:
	geo = json.load(f)
	def isnull(val):
	return (val is None or
	val in ('-99', '-1', 'NUL', 'NULL') or
	isinstance(val, float) and isnan(val))
	out = {'type': 'FeatureCollection',
	'features': []}
	for feature in geo['features']:
	p = feature['properties']
	if isnull(p['iso_a2']):
	p['iso_a2'] = CC.get(p['adm0_a3'])
	out['features'].append(feature)
	# with open('admin1.json', 'w') as f:
	# json.dump(out, f)
	key = lambda feature: feature['properties']['adm0_a3']
	for cc, features in groupby(sorted(out['features'], key=key), key):
	features = [OrderedDict([
	('type', f['type']),
	('properties', f['properties']),
	('geometry', f['geometry']),
	]) for f in features]
	print(cc, len(features))
	if cc == 'SVN':
	from pprint import pprint
	pprint(features[0])
	with open('admin1-{}.json'.format(cc), 'w') as f:
	json.dump(dict(type='FeatureCollection', features=features), f)
	"
	rm admin1.json
	## Alternatively, don't simplify all admin1 at once (above), but simplify
	## each country individually here
	for file in admin1-*.json; do
	simplify .7 "$file" "$file"
	done

[kernc]

Then again, maybe lat-longs for admin0 (countries) are computed on-the-fly, in which case the override might also fit somewhere here:

orange3-geo/orangecontrib/geo/mapper.py

Lines 111 to 126 in 1266026

	try:
	r = shape.representative_point()
	except ValueError:
	# For GBR, representative point above fails with:
	# ValueError: No Shapely geometry can be created from null value
	r = shape.centroid
	if not r.within(shape):
	max_poly = max([shape] if isinstance(shape, Polygon) else list(shape),
	key=lambda polygon: polygon.area)
	# From https://stackoverflow.com/questions/33311616/find-coordinate-of-closest-point-on-polygon-shapely/33324058#33324058
	poly_ext = max_poly.exterior
	dist = poly_ext.project(r)
	pt = poly_ext.interpolate(dist)
	r = Point(pt.coords[0])
	assert r.within(shape)
	p.update(latitude=r.y, longitude=r.x)

[kaimikael]

I've been looking at the data files and I can’t figure out how they are generated. As we noted, there is info on Bouvet island (also Norwegian territory) in the admin1-NOR.json file, but nowhere else in the orange3-geo-master tree, so if that file is generated, what is it based on?

[kernc]

if that file is generated, what is it based on?

https://github.com/biolab/orange3-geo/blob/master/CONTRIBUTING.md#building-geojson-files

Based on your screenshot, however, I think you should to be looking at admin0.json and the code snippet above.