Bug fixes

This commit is contained in:
Edward Betts 2021-04-15 12:28:02 +01:00
parent c3094355d7
commit 0df666742e

311
lookup.py
View file

@ -1,11 +1,16 @@
#!/usr/bin/python3 #!/usr/bin/python3
from flask import Flask, render_template, request, jsonify from flask import Flask, render_template, request, jsonify, redirect, url_for
import requests import requests
import os import os
import json import json
import urllib.parse import urllib.parse
import random import random
import simplejson
import psycopg2
from geopy.distance import distance
# select gid, code, name from scotland where st_contains(geom, ST_Transform(ST_SetSRID(ST_MakePoint(-4.177, 55.7644), 4326), 27700));
commons_cat_start = 'https://commons.wikimedia.org/wiki/Category:' commons_cat_start = 'https://commons.wikimedia.org/wiki/Category:'
use_cache = False use_cache = False
@ -23,6 +28,8 @@ headers = {
OVERPASS_URL = 'https://lz4.overpass-api.de' OVERPASS_URL = 'https://lz4.overpass-api.de'
wikidata_query_api_url = 'https://query.wikidata.org/bigdata/namespace/wdq/sparql' wikidata_query_api_url = 'https://query.wikidata.org/bigdata/namespace/wdq/sparql'
wikidata_url = 'https://www.wikidata.org/w/api.php' wikidata_url = 'https://www.wikidata.org/w/api.php'
wd_entity = 'http://www.wikidata.org/entity/Q'
city_of_london_qid = 'Q23311'
samples = [ samples = [
(50.8326, -0.2689, 'Adur'), (50.8326, -0.2689, 'Adur'),
@ -82,24 +89,194 @@ def random_location():
return render_template('random.html', lat=lat, lon=lon, result=result, elements=elements) return render_template('random.html', lat=lat, lon=lon, result=result, elements=elements)
def do_lookup(elements, lat, lon): @app.route("/wikidata_tag")
commons_cat = osm_lookup(elements) def wikidata_tag():
lat = float(request.args.get('lat'))
lon = float(request.args.get('lon'))
scotland_code = get_scotland_code(lat, lon)
if scotland_code:
rows = lookup_scottish_parish_in_wikidata(scotland_code)
hit = commons_from_rows(rows)
elements = []
result = build_dict(hit, lat, lon)
else:
elements = get_osm_elements(lat, lon)
result = do_lookup(elements, lat, lon)
return render_template('wikidata_tag.html', lat=lat, lon=lon, result=result, elements=elements)
@app.route("/detail")
def detail_page():
try:
lat, lon = [float(request.args.get(param)) for param in ('lat', 'lon')]
except TypeError:
return redirect(url_for('index'))
reply = lat_lon_to_wikidata(lat, lon)
return render_template('random.html', lat=lat, lon=lon, **reply)
def bounding_box_area(element):
bbox = element['bounds']
x = distance((bbox['maxlat'], bbox['minlon']), (bbox['maxlat'], bbox['maxlon']))
y = distance((bbox['minlat'], bbox['maxlon']), (bbox['maxlat'], bbox['minlon']))
return x.km * y.km
def wd_to_qid(wd):
# expecting {'type': 'url', 'value': 'https://www.wikidata.org/wiki/Q30'}
if wd['type'] == 'uri':
return wd_uri_to_qid(wd['value'])
def wd_uri_to_qid(value):
assert value.startswith(wd_entity)
return value[len(wd_entity) - 1:]
def build_dict(hit, lat, lon):
coords = {'lat': lat, 'lon': lon} coords = {'lat': lat, 'lon': lon}
if commons_cat is None: if hit is None:
return dict(commons_cat=None, missing=True, coords=coords) return dict(commons_cat=None, missing=True, coords=coords)
commons_cat = hit['commons_cat']
url = commons_cat_start + urllib.parse.quote(commons_cat.replace(' ', '_')) url = commons_cat_start + urllib.parse.quote(commons_cat.replace(' ', '_'))
return dict(commons_cat={'title': commons_cat, 'url': url}, coords=coords) return dict(commons_cat={'title': commons_cat, 'url': url},
coords=coords,
admin_level=hit.get('admin_level'),
wikidata=hit['wikidata'])
def do_lookup(elements, lat, lon):
try:
hit = osm_lookup(elements, lat, lon)
except QueryError as e:
return {
'query': e.query,
'error': e.r.text,
'query_url': 'https://query.wikidata.org/#' + e.query,
}
return build_dict(hit, lat, lon)
def get_scotland_code(lat, lon):
conn = psycopg2.connect(dbname='geocode', user='geocode', password='ooK3ohgh', host='localhost')
cur = conn.cursor()
point = f'ST_Transform(ST_SetSRID(ST_MakePoint({lon}, {lat}), 4326), 27700)'
cur.execute(f'select code, name from scotland where st_contains(geom, {point});')
row = cur.fetchone()
# expand search, disabled for now 2020-04-20
if not row:
cur.execute(f'select code, name from scotland where ST_DWithin(geom, {point}, 100);')
row = cur.fetchone()
conn.close()
if row:
return row[0]
def wdqs_geosearch_query(lat, lon):
if isinstance(lat, float):
lat = f'{lat:f}'
if isinstance(lon, float):
lon = f'{lon:f}'
query_template = '''
SELECT DISTINCT ?item ?distance ?itemLabel ?isa ?isaLabel ?commonsCat ?commonsSiteLink WHERE {
{
SELECT DISTINCT ?item ?location ?distance ?isa WHERE {
?item wdt:P31/wdt:P279* wd:Q486972.
?item wdt:P31 ?isa .
SERVICE wikibase:around {
?item wdt:P625 ?location.
bd:serviceParam wikibase:center "Point(LON LAT)"^^geo:wktLiteral;
wikibase:radius 5;
wikibase:distance ?distance.
}
}
}
MINUS { ?item wdt:P582 ?endTime . }
OPTIONAL { ?item wdt:P373 ?commonsCat. }
OPTIONAL { ?commonsSiteLink schema:about ?item;
schema:isPartOf <https://commons.wikimedia.org/>. }
SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
} ORDER BY (?distance)'''
query = query_template.replace('LAT', lat).replace('LON', lon)
reply = wdqs(query)
return reply['results']['bindings']
def wdqs_geosearch(lat, lon):
default_max_dist = 1
rows = wdqs_geosearch_query(lat, lon)
max_dist = {
'Q188509': 1, # suburb
'Q3957': 2, # town
'Q532': 1, # village
'Q5084': 1, # hamlet
'Q515': 2, # city
'Q1549591': 3, # big city
}
for row in rows:
isa = wd_uri_to_qid(row['isa']['value'])
if ('commonsCat' not in row and 'commonsSiteLink' not in row and isa not in max_dist):
continue
distance = float(row['distance']['value'])
if distance > max_dist.get(isa, default_max_dist):
continue
if 'commonsCat' not in row and 'commonsSiteLink' not in row:
break
return row
def lat_lon_to_wikidata(lat, lon):
scotland_code = get_scotland_code(lat, lon)
if scotland_code:
rows = lookup_scottish_parish_in_wikidata(scotland_code)
hit = commons_from_rows(rows)
elements = []
result = build_dict(hit, lat, lon)
return {'elements': elements, 'result': result}
elements = get_osm_elements(lat, lon)
result = do_lookup(elements, lat, lon)
# special case because the City of London is admin_level=6 in OSM
if result['wikidata'] == city_of_london_qid:
return {'elements': elements, 'result': result}
admin_level = result['admin_level']
if not admin_level or admin_level >= 7:
return {'elements': elements, 'result': result}
row = wdqs_geosearch(lat, lon)
if row:
hit = commons_from_rows([row])
elements = []
result = build_dict(hit, lat, lon)
return {'elements': elements, 'result': result}
@app.route("/") @app.route("/")
def index(): def index():
q = request.args.get('q')
if q and q.strip():
lat, lon = [v.strip() for v in q.split(',', 1)]
return redirect(url_for('detail_page', lat=lat, lon=lon))
lat = request.args.get('lat') lat = request.args.get('lat')
lon = request.args.get('lon') lon = request.args.get('lon')
if lat is None or lon is None: if lat is None or lon is None:
samples.sort(key=lambda row: row[2])
return render_template('index.html', samples=samples) return render_template('index.html', samples=samples)
elements = get_osm_elements(lat, lon) return jsonify(lat_lon_to_wikidata(lat, lon)['result'])
ret = do_lookup(elements, lat, lon)
return jsonify(ret)
def wikidata_api_call(params): def wikidata_api_call(params):
call_params = { call_params = {
@ -160,30 +337,72 @@ out bb tags qt;'''
return run_query(oql) return run_query(oql)
def lookup_gss_in_wikidata(gss): def lookup_scottish_parish_in_wikidata(code):
query = ''' query = '''
SELECT ?item ?itemLabel ?commonsSiteLink ?commonsCat WHERE { SELECT ?item ?itemLabel ?commonsSiteLink ?commonsCat WHERE {
?item wdt:P836 "GSS" . ?item wdt:P528 "CODE" .
?item wdt:P31 wd:Q5124673 .
OPTIONAL { ?commonsSiteLink schema:about ?item ; OPTIONAL { ?commonsSiteLink schema:about ?item ;
schema:isPartOf <https://commons.wikimedia.org/> } schema:isPartOf <https://commons.wikimedia.org/> }
OPTIONAL { ?item wdt:P373 ?commonsCat } OPTIONAL { ?item wdt:P373 ?commonsCat }
SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". } SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
} }
'''.replace('GSS', gss) '''.replace('CODE', code)
reply = wdqs(query)
return reply['results']['bindings']
def lookup_gss_in_wikidata(gss):
query = '''
SELECT ?item ?itemLabel ?commonsSiteLink ?commonsCat WHERE {
?item wdt:P836 GSS .
OPTIONAL { ?commonsSiteLink schema:about ?item ;
schema:isPartOf <https://commons.wikimedia.org/> }
OPTIONAL { ?item wdt:P373 ?commonsCat }
SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
}
'''.replace('GSS', repr(gss))
reply = wdqs(query)
return reply['results']['bindings']
def lookup_wikidata_by_name(name, lat, lon):
query = '''
SELECT DISTINCT ?item ?itemLabel ?commonsSiteLink ?commonsCat WHERE {
?item rdfs:label LABEL@en .
FILTER NOT EXISTS { ?item wdt:P31 wd:Q17362920 } .# ignore Wikimedia duplicated page
OPTIONAL { ?commonsSiteLink schema:about ?item ;
schema:isPartOf <https://commons.wikimedia.org/> }
OPTIONAL { ?item wdt:P373 ?commonsCat }
?item wdt:P625 ?coords .
FILTER(geof:distance(?coords, "Point(LON LAT)"^^geo:wktLiteral) < 10)
FILTER(?commonsCat || ?commonsSiteLink)
SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
}
'''.replace('LABEL', repr(name)).replace('LAT', str(lat)).replace('LON', str(lon))
reply = wdqs(query) reply = wdqs(query)
return reply['results']['bindings'] return reply['results']['bindings']
def unescape_title(t): def unescape_title(t):
return urllib.parse.unquote(t.replace('_', ' ')) return urllib.parse.unquote(t.replace('_', ' '))
def get_commons_cat_from_gss(gss): def commons_from_rows(rows):
rows = lookup_gss_in_wikidata(gss)
for row in rows: for row in rows:
if 'commonsCat' in row: if 'commonsCat' in row:
return row['commonsCat']['value'] qid = wd_to_qid(row['item'])
return {'wikidata': qid,
'commons_cat': row['commonsCat']['value']}
if 'commonsSiteLink' in row: if 'commonsSiteLink' in row:
site_link = row['commonsSiteLink']['value'] site_link = row['commonsSiteLink']['value']
return unescape_title(site_link[len(commons_cat_start):]) qid = wd_to_qid(row['item'])
cat = unescape_title(site_link[len(commons_cat_start):])
return {'wikidata': qid, 'commons_cat': cat}
def get_commons_cat_from_gss(gss):
print('GSS:', gss)
rows = lookup_gss_in_wikidata(gss)
return commons_from_rows(rows)
def get_osm_elements(lat, lon): def get_osm_elements(lat, lon):
filename = f'cache/{lat}_{lon}.json' filename = f'cache/{lat}_{lon}.json'
@ -198,24 +417,74 @@ def get_osm_elements(lat, lon):
return elements return elements
def osm_lookup(elements): def osm_lookup(elements, lat, lon):
is_in = [] is_in = []
for e in elements: elements.sort(key=lambda e: bounding_box_area(e))
if False:
for e in sorted(elements, key=lambda e: e['area']):
try: try:
admin_level = int(e['tags']['admin_level']) admin_level = int(e['tags']['admin_level'])
except (ValueError, KeyError): except (ValueError, KeyError):
admin_level = None
if admin_level is None:
if e['id'] == 6038068: # Great Britain
continue
if 'place' in e['tags'] or e['tags'].get('type') == 'boundary':
is_in.append((99, e['tags']))
continue continue
is_in.append((admin_level, e['tags'])) is_in.append((admin_level, e['tags']))
for _, tags in sorted(is_in, key=lambda i: i[0], reverse=True): # for _, tags in sorted(is_in, key=lambda i: i[0], reverse=True):
for e in elements:
if 'tags' not in e:
continue
tags = e['tags']
admin_level_tag = tags.get('admin_level')
admin_level = int(admin_level_tag) if admin_level_tag and admin_level_tag.isdigit() else None
if not admin_level and tags.get('boundary') != 'political':
continue
if 'wikidata' in tags: if 'wikidata' in tags:
qid = tags['wikidata'] qid = tags['wikidata']
return qid_to_commons_category(qid) commons = qid_to_commons_category(qid)
if commons:
return {
'wikidata': qid,
'commons_cat': commons,
'admin_level': admin_level,
}
gss = tags.get('ref:gss') gss = tags.get('ref:gss')
if not gss: if gss:
ret = get_commons_cat_from_gss(gss)
if ret:
ret['admin_level'] = admin_level
return ret
name = tags.get('name')
if not name:
continue continue
return get_commons_cat_from_gss(gss) if name.endswith(' CP'):
name = name[:-3]
rows = lookup_wikidata_by_name(name, lat, lon)
if len(rows) == 1:
ret = commons_from_rows(rows)
if ret:
ret['admin_level'] = admin_level
return ret
has_wikidata_tag = [e['tags'] for e in elements if 'wikidata' in e['tags']]
if len(has_wikidata_tag) != 1:
return
qid = has_wikidata_tag[0]['wikidata']
return {
'wikidata': qid,
'commons_cat': qid_to_commons_category(qid),
'admin_level': admin_level,
}
if __name__ == '__main__': if __name__ == '__main__':