regions.py

#!/usr/bin/env python3
from xml.etree import ElementTree
from geometry import *

REGION_TYPES = ['major', 'minor']
REGION_KEY_VALUES = [x + '_region' for x in REGION_TYPES]
SHAPE_KEYS = [x + '_shapes' for x in REGION_TYPES]

def parse_nodes(root):
	nodes = {}
	for child in root:
		if child.tag == 'node':
			lat, lon = float(child.attrib['lat']), float(child.attrib['lon'])
			nodes[child.attrib['id']] = Location(lat, lon)
	return nodes

def parse_way(way, nodes):
	def choose_shapes(way, boundary):
		return (way['major_shapes']
			if boundary == 'major_region'
			else way['minor_shapes'])
	result = {'minor_shapes': [], 'major_shapes': []}
	shape = []
	for child in way:
		if child.tag == 'nd':
			shape.append(child.attrib['ref'])
		elif child.tag == 'tag':
			key = child.attrib['k']
			if key in SHAPE_KEYS:
				raise ValueError(str.format('tag "{}" is not allowed', key))
			result[key] = child.attrib['v']
			if key == 'boundary' and result['boundary'] not in REGION_KEY_VALUES:
				return None # we're not interested in it!
	if shape[-1] != shape[0]:
		raise ValueError('polygon is not closed: %r' % result)
	if 'boundary' in result:
		shape = [nodes[ref] for ref in shape[:-1]]
		choose_shapes(result, result['boundary']).append(Polygon(*shape))
		return result

def parse_boundaries(root, *, nodes):
	for child in root:
		if child.tag == 'way':
			way = parse_way(child, nodes = nodes)
			if way:
				yield way

def parse_regions(filename):
	from katakana import transliterate as transliterate_katakana
	tree = ElementTree.parse(filename)
	root = tree.getroot()
	nodes = parse_nodes(root)
	regions = dict()
	extra_shapes = list()
	for way in parse_boundaries(root, nodes = nodes):
		if 'boundary' in way and 'municipality' in way and 'name:fi' in way:
			def clean(x):
				return x.replace('/', '').replace(';', '')
			way['ref'] = str.format(
				'{municipality}/{name}',
				municipality = str.upper(clean(way['municipality'])),
				name = clean(way['name:fi']),
			)
			# defines a region
			way['via_factor'] = int(way.get('via_factor', 1))
			if way['ref'] in regions:
				raise ValueError(str.format(
					'Region {ref} defined twice',
					ref = repr(way['ref']),
				))
			regions[way['ref']] = way
			del way['boundary']
			if 'external' in way:
				way['boundary'] = 'minor_region'
			for prefix in ['', 'short_', 'internal_']:
				name_key = prefix + 'name:fi'
				ja_name_key = prefix + 'name:ja'
				if name_key in way and way[name_key] and ja_name_key not in way:
					way[ja_name_key] = transliterate_katakana(way[name_key])
		elif 'boundary' in way and 'is_in' in way:
			# adds an extra shape to an existing region
			extra_shapes.append(way)
	for extra_shape in extra_shapes:
		name = extra_shape['is_in']
		try:
			region = regions[name]
		except KeyError:
			raise ValueError(str.format(
				'Extra shape refers to {name} which was not found: {extra_shape}',
				name = repr(name),
				extra_shape = repr(extra_shape),
			))
		for key in SHAPE_KEYS:
			region[key].extend(extra_shape[key])
	return regions