osm parsed
This commit is contained in:
SDE
2023-07-20 14:04:32 +03:00
parent 113e576d4c
commit b768988a5d
19 changed files with 855 additions and 20 deletions

View File

@@ -0,0 +1,288 @@
import time
import overpass
import copy
from geopy.geocoders import Nominatim
Nominatim_last_request = None
def osm_api_request(data, res_type):
time.sleep(1)
api = overpass.API(timeout=100)
res = api.get(data, responseformat=res_type)
pres_names = []
res_list = []
if res_type[:3].lower() == 'csv':
data_type_list = copy.deepcopy(res[0])
for line in res[1:]:
if line[0] in pres_names:
continue
i = 0
line_Dict = {}
while i < len(data_type_list):
line_Dict.update({data_type_list[i]: line[i]})
i += 1
if 'name:en' in line_Dict and 'name' in line_Dict:
if not line_Dict["name:en"]:
if line_Dict["name"]:
line_Dict["name:en"] = line_Dict["name"]
if not line_Dict["name:en"]:
continue
res_list.append(line_Dict)
pres_names.append(line[0])
else:
res_list = copy.deepcopy(res)
return res_list
def osm_get_area_id_by_params_dict(paramsDict):
area_id = None
try:
# Geocoding request via Nominatim
time.sleep(1)
geolocator = Nominatim(user_agent='TWB')
geo_results = geolocator.geocode(paramsDict, exactly_one=False, limit=3)
if not geo_results:
return area_id
# Searching for relation in result set
country_obj = None
for r in geo_results:
print(r.address, r.raw.get("osm_type"))
if r.raw.get("osm_type") == "relation":
country_obj = r
break
if not country_obj:
return area_id
# Calculating area id
area_id = int(country_obj.raw.get("osm_id")) + 3600000000
except Exception as e:
print(f'osm_get_area_id_by_params_dict Error = {e}')
return area_id
def osm_get_countries():
res_type = 'csv("name", "name:en", "name:ru", "ISO3166-1", "flag", "int_name", "official_name", "ISO3166-1:alpha3", ' \
'"ISO3166-1:numeric", ::lon, ::lat)'
data = 'relation["admin_level"="2"]["ISO3166-1:alpha3"~"^...$"]; ' \
'out center;'
res = osm_api_request(data, res_type)
from operator import itemgetter
res = sorted(res, key=itemgetter('name'))
return res
def osm_get_cities_by_country(country_Dict):
res = []
try:
if 'area_id' in country_Dict and country_Dict['area_id']:
area_id = country_Dict['area_id']
else:
return []
res_type = 'csv("name", "name:ru", "name:en", ::lon, ::lat)'
data = f'area({area_id})->.searchArea;' \
'(node[place~"city|town"](area.searchArea););' \
'out center;'
res = osm_api_request(data, res_type)
except Exception as e:
print(f'osm_get_cities_by_country Error = {e}')
return res
def osm_get_airports(area_id, city_find_str=None):
airports_cities_Dict = {}
try:
if not area_id:
return airports_cities_Dict
res_type = 'csv("name", "name:ru", "name:en", "iata", "icao", "place", "int_name", "addr:country", "city_served", ::lon, ::lat)'
data = f'area({area_id})->.searchArea;' \
f'nwr["aeroway"="aerodrome"]["iata"~"^...$"]["icao"~"^....$"]["abandoned"!~".*"]["landuse"!="military"]["was:landuse"!="military"]["aerodrome:type"!~"airfield|military|private"]["amenity"!="flight_school"]["closed"!="yes"](area.searchArea)->.airports;' \
if city_find_str:
data = f'{data}' \
f'foreach.airports->.elem(nwr(around.elem:20000)[{city_find_str}]->.city; .elem out center; .city out center;);'
else:
data = f'{data} .airports out center;'
res = osm_api_request(data, res_type)
if not city_find_str:
return res
present_IATA_list = []
i = 0
while i < len(res):
if not res[i]['iata'] or not res[i]['icao']:
del res[i]
continue
if res[i]['iata'] and res[i]['iata'] in present_IATA_list:
del res[i]
continue
present_IATA_list.append(res[i]['iata'])
# если аэропорт
if res[i]['iata']:
if i + 1 < len(res) and res[i+1]['place']:
# ищем город на следующих строках
i2 = i + 1
linked = None
while i2 < len(res) and res[i2]['place']:
if not linked:
if (res[i2]['name'] and res[i]['name'] and res[i2]['name'] in res[i]['name']) or \
(res[i2]['name:en'] and res[i]['name:en'] and res[i2]['name:en'] in res[i]['name:en']) or \
(res[i2]['name:ru'] and res[i]['name:ru'] and res[i2]['name:ru'] in res[i]['name:ru']):
linked = i2
i2 += 1
if not linked:
linked = i + 1
res[i]['city'] = copy.deepcopy(res[linked])
if not res[i]['city']['name'] in airports_cities_Dict:
airports_cities_Dict.update({res[i]['city']['name:en']: [res[i]]})
else:
airports_cities_Dict[res[i]['city']['name:en']].append(res[i])
while i < i2:
del res[i]
i2 -= 1
continue
# если не найдена связка с городом
elif res[i]['city_served']:
res[i]['city'] = {'name:en': res[i]['city_served']}
if not res[i]['city']['name:en'] in airports_cities_Dict:
airports_cities_Dict.update({res[i]['city']['name:en']: [res[i]]})
else:
airports_cities_Dict[res[i]['city']['name:en']].append(res[i])
del res[i]
continue
# текущий элемент - не аэропорт - удаляем
else:
del res[i]
continue
i += 1
if res:
airports_cities_Dict[None] = res
except Exception as e:
print(f'osm_get_airports Error = {e}')
if e.args[0] == 25:
return {'error': 'timeout'}
return airports_cities_Dict
def osm_get_country_w_cities_n_airports(country_Dict, area_id):
print(f'{country_Dict["name:en"]}')
if area_id:
country_Dict['area_id'] = area_id
else:
country_Dict['area_id'] = osm_get_area_id_by_params_dict({'country': country_Dict['name']})
airports_Dict = osm_get_airports(country_Dict['area_id'], 'place~"city|town"')
if airports_Dict and 'error' in airports_Dict and airports_Dict['error'] == 'timeout':
airports_Dict = osm_get_airports(country_Dict['area_id'], 'place="city"')
from ReferenceDataApp.funcs import get_countries_key_data, get_cities_by_country_name_en
db_cities = get_cities_by_country_name_en(country_Dict["name:en"])
cities = osm_get_cities_by_country(country_Dict)
for city in cities:
print(f' > {city["name:en"]}')
if airports_Dict and 'error' in airports_Dict and airports_Dict['error'] == 'timeout':
if city['name:en'] in db_cities.keys():
city['area_id'] = db_cities[city['name:en']]
else:
city['area_id'] = osm_get_area_id_by_params_dict(
{'country': country_Dict['name:en'], 'city': city['name:en']})
airports_list = osm_get_airports(city['area_id'])
if airports_list:
city['airports'] = copy.deepcopy(airports_list)
else:
if city['name:en'] in airports_Dict.keys():
city['airports'] = copy.deepcopy(airports_Dict[city['name:en']])
del airports_Dict[city['name:en']]
if not 'airports' in city:
city['airports'] = []
print(f' > > airports count={str(len(city["airports"]))}')
city['parsing_status'] = 'finished'
country_Dict['cities'] = cities
country_Dict['parsing_status'] = 'finished'
airports_wo_city = []
if airports_Dict and None in airports_Dict:
airports_wo_city = airports_Dict[None]
return country_Dict, airports_wo_city
def osm_get_countries_n_cities_n_airports():
airports_wo_city = []
from ReferenceDataApp.funcs import get_countries_key_data, get_cities_by_country_name_en
db_countries = get_countries_key_data()
countries = osm_get_countries()
i = 0
while i < len(countries):
area_id = None
if countries[i]['name:en'] in db_countries.keys():
area_id = db_countries[countries[i]['name:en']]
countries[i], airports_wo_city_for_country = osm_get_country_w_cities_n_airports(countries[i], area_id)
if airports_wo_city_for_country:
airports_wo_city.extend(airports_wo_city_for_country)
return countries