|
1
|
|
|
# -*- coding: utf-8 -*- |
|
|
|
|
|
|
2
|
|
|
|
|
3
|
|
|
try: |
|
4
|
|
|
import logging |
|
5
|
|
|
import sys |
|
6
|
|
|
import os |
|
7
|
|
|
import re |
|
|
|
|
|
|
8
|
|
|
import json |
|
9
|
|
|
from osm_poi_matchmaker.libs.soup import save_downloaded_soup |
|
10
|
|
|
from osm_poi_matchmaker.libs.address import extract_street_housenumber_better_2, clean_city, \ |
|
11
|
|
|
extract_javascript_variable, clean_opening_hours |
|
12
|
|
|
from osm_poi_matchmaker.libs.geo import check_hu_boundary |
|
13
|
|
|
from osm_poi_matchmaker.libs.osm_tag_sets import POS_HU_GEN, PAY_CASH |
|
14
|
|
|
from osm_poi_matchmaker.utils.enums import WeekDaysLong |
|
15
|
|
|
from osm_poi_matchmaker.utils.data_provider import DataProvider |
|
16
|
|
|
from osm_poi_matchmaker.utils.enums import FileType |
|
17
|
|
|
except ImportError as err: |
|
18
|
|
|
logging.error('Error %s import module: %s', __name__, err) |
|
19
|
|
|
logging.exception("Exception occurred") |
|
20
|
|
|
sys.exit(128) |
|
21
|
|
|
|
|
22
|
|
|
|
|
23
|
|
|
class hu_rossmann(DataProvider): |
|
|
|
|
|
|
24
|
|
|
|
|
25
|
|
|
def constains(self): |
|
26
|
|
|
self.link = 'https://www.rossmann.hu/uzletkereso' |
|
27
|
|
|
self.POI_COMMON_TAGS = "" |
|
28
|
|
|
self.filetype = FileType.html |
|
29
|
|
|
self.filename = '{}.{}'.format(self.__class__.__name__, self.filetype.name) |
|
30
|
|
|
|
|
31
|
|
View Code Duplication |
def types(self): |
|
|
|
|
|
|
32
|
|
|
self.__types = [{'poi_code': 'hurossmche', 'poi_name': 'Rossmann', 'poi_type': 'chemist', |
|
33
|
|
|
'poi_tags': "{'shop': 'chemist', 'operator': 'Rossmann Magyarország Kft.', \ |
|
|
|
|
|
|
34
|
|
|
'operator:addr': '2225 Üllő, Zsaróka út 8.', 'ref:vatin:hu': '11149769-2-44', \ |
|
35
|
|
|
'ref:vatin': 'HU11149769', 'brand':'Rossmann', 'brand:wikidata': 'Q316004', \ |
|
36
|
|
|
'brand:wikipedia': 'de:Dirk Rossmann GmbH', 'contact:email': '[email protected]', \ |
|
37
|
|
|
'phone': '+36 29 889-800;+36 70 4692 800', 'contact:facebook':'https://www.facebook.com/Rossmann.hu', \ |
|
38
|
|
|
'contact:youtube': 'https://www.youtube.com/channel/UCmUCPmvMLL3IaXRBtx7-J7Q', \ |
|
39
|
|
|
'contact:instagram':'https://www.instagram.com/rossmann_hu', " + POS_HU_GEN + PAY_CASH + "'air_conditioning': 'yes'}", |
|
|
|
|
|
|
40
|
|
|
'poi_url_base': 'https://www.rossmann.hu', 'poi_search_name': 'rossmann', 'osm_search_distance_perfect': 2000, 'osm_search_distance_safe': 200, 'osm_search_distance_unsafe': 3}] |
|
|
|
|
|
|
41
|
|
|
return self.__types |
|
42
|
|
|
|
|
43
|
|
|
def process(self): |
|
44
|
|
|
try: |
|
45
|
|
|
soup = save_downloaded_soup('{}'.format(self.link), os.path.join(self.download_cache, self.filename), |
|
|
|
|
|
|
46
|
|
|
self.filetype, None, self.verify_link) |
|
|
|
|
|
|
47
|
|
|
if soup is not None: |
|
48
|
|
|
# parse the html using beautiful soap and store in variable `soup` |
|
49
|
|
|
text = json.loads(extract_javascript_variable(soup, 'places')) |
|
50
|
|
|
for poi_data in text: |
|
51
|
|
|
poi_data = poi_data['addresses'][0] |
|
52
|
|
|
# Assign: code, postcode, city, name, branch, website, original, street, housenumber, conscriptionnumber, ref, geom |
|
|
|
|
|
|
53
|
|
|
self.data.name = 'Rossmann' |
|
54
|
|
|
self.data.code = 'hurossmche' |
|
55
|
|
|
self.data.city = clean_city(poi_data['city']) |
|
56
|
|
|
self.data.postcode = poi_data.get('zip').strip() |
|
57
|
|
|
for i in range(0, 7): |
|
58
|
|
|
if poi_data['business_hours'][WeekDaysLong(i).name.lower()] is not None: |
|
59
|
|
|
opening, closing = clean_opening_hours(poi_data['business_hours'][WeekDaysLong(i).name.lower()]) |
|
|
|
|
|
|
60
|
|
|
self.data.day_open_close(i, opening, closing) |
|
61
|
|
|
else: |
|
62
|
|
|
self.data.day_open_close(i, None, None) |
|
63
|
|
|
self.data.lat, self.data.lon = check_hu_boundary(poi_data['position'][0], poi_data['position'][1]) |
|
|
|
|
|
|
64
|
|
|
self.data.street, self.data.housenumber, self.data.conscriptionnumber = extract_street_housenumber_better_2( |
|
|
|
|
|
|
65
|
|
|
poi_data['address']) |
|
66
|
|
|
self.data.original = poi_data['address'] |
|
67
|
|
|
self.data.public_holiday_open = False |
|
68
|
|
|
self.data.add() |
|
69
|
|
|
except Exception as e: |
|
|
|
|
|
|
70
|
|
|
logging.exception("Exception occurred") |
|
71
|
|
|
logging.error(e) |
|
72
|
|
|
|
KAMI911 /
osm_poi_matchmaker
Loading history...