1
|
|
|
# -*- coding: utf-8 -*- |
2
|
|
|
|
3
|
|
|
try: |
4
|
|
|
import traceback |
5
|
|
|
import logging |
6
|
|
|
import sys |
7
|
|
|
import hashlib |
8
|
|
|
from osm_poi_matchmaker.dao.data_structure import City, POI_common, POI_address, Street_type |
9
|
|
|
from osm_poi_matchmaker.libs import address |
10
|
|
|
from osm_poi_matchmaker.dao import poi_array_structure |
11
|
|
|
except ImportError as err: |
12
|
|
|
logging.error('Error %s import module: %s', __name__, err) |
13
|
|
|
logging.error(traceback.print_exc()) |
14
|
|
|
sys.exit(128) |
15
|
|
|
|
16
|
|
|
POI_COLS = poi_array_structure.POI_COLS |
17
|
|
|
|
18
|
|
|
|
19
|
|
|
def get_or_create(session, model, **kwargs): |
20
|
|
|
instance = session.query(model).filter_by(**kwargs).first() |
21
|
|
|
if instance: |
22
|
|
|
logging.debug('Already added: %s' ,instance) |
23
|
|
|
return instance |
24
|
|
|
else: |
25
|
|
|
try: |
26
|
|
|
instance = model(**kwargs) |
27
|
|
|
session.add(instance) |
28
|
|
|
return instance |
29
|
|
|
except Exception as e: |
30
|
|
|
logging.error('Can't add to database. (%s)', e) |
|
|
|
|
31
|
|
|
logging.error(traceback.print_exc()) |
32
|
|
|
raise (e) |
33
|
|
|
|
34
|
|
|
|
35
|
|
|
def get_or_create_poi(session, model, **kwargs): |
36
|
|
|
if kwargs['poi_common_id'] is not None: |
37
|
|
|
if kwargs['poi_common_id'] is not None and kwargs['poi_addr_city'] is not None and ( |
38
|
|
|
(kwargs['poi_addr_street'] and kwargs['poi_addr_housenumber'] is not None) or ( |
39
|
|
|
kwargs['poi_conscriptionnumber'] is not None)): |
40
|
|
|
logging.debug('Fully filled basic data record') |
41
|
|
|
else: |
42
|
|
|
logging.warning('Missing record data: %s', kwargs) |
43
|
|
|
instance = session.query(model)\ |
44
|
|
|
.filter_by(poi_common_id=kwargs['poi_common_id'])\ |
45
|
|
|
.filter_by(poi_addr_city=kwargs['poi_addr_city'])\ |
46
|
|
|
.filter_by(poi_addr_street=kwargs['poi_addr_street'])\ |
47
|
|
|
.filter_by(poi_addr_housenumber=kwargs['poi_addr_housenumber'])\ |
48
|
|
|
.filter_by(poi_conscriptionnumber=kwargs['poi_conscriptionnumber'])\ |
49
|
|
|
.filter_by(poi_branch=kwargs['poi_branch'])\ |
50
|
|
|
.first() |
51
|
|
|
if instance: |
52
|
|
|
logging.debug('Already added: %s', instance) |
53
|
|
|
return instance |
54
|
|
|
else: |
55
|
|
|
try: |
56
|
|
|
instance = model(**kwargs) |
57
|
|
|
session.add(instance) |
58
|
|
|
return instance |
59
|
|
|
except Exception as e: |
60
|
|
|
logging.error('Can't add to database. (%s)', e) |
61
|
|
|
logging.error(traceback.print_exc()) |
62
|
|
|
raise (e) |
63
|
|
|
|
64
|
|
|
|
65
|
|
|
def insert_city_dataframe(session, city_df): |
66
|
|
|
city_df.columns = ['city_post_code', 'city_name'] |
67
|
|
|
try: |
68
|
|
|
for index, city_data in city_df.iterrows(): |
69
|
|
|
get_or_create(session, City, city_post_code=city_data['city_post_code'], |
70
|
|
|
city_name=address.clean_city(city_data['city_name'])) |
71
|
|
|
except Exception as e: |
72
|
|
|
|
73
|
|
|
logging.error('Rolled back: %s.', e) |
74
|
|
|
logging.error(city_data) |
75
|
|
|
logging.error(traceback.print_exc()) |
76
|
|
|
session.rollback() |
77
|
|
|
else: |
78
|
|
|
logging.info('Successfully added %s city items to the dataset.', len(city_df)) |
79
|
|
|
session.commit() |
80
|
|
|
|
81
|
|
|
|
82
|
|
|
def insert_street_type_dataframe(session, city_df): |
83
|
|
|
city_df.columns = ['street_type'] |
84
|
|
|
try: |
85
|
|
|
for index, city_data in city_df.iterrows(): |
86
|
|
|
get_or_create(session, Street_type, street_type=city_data['street_type']) |
87
|
|
|
except Exception as e: |
88
|
|
|
logging.error('Rolled back: %s.', e) |
89
|
|
|
logging.error(city_data) |
90
|
|
|
logging.error(traceback.print_exc()) |
91
|
|
|
session.rollback() |
92
|
|
|
else: |
93
|
|
|
logging.info('Successfully added %s street type items to the dataset.', len(city_df)) |
94
|
|
|
session.commit() |
95
|
|
|
|
96
|
|
|
|
97
|
|
|
def insert_common_dataframe(session, common_df): |
98
|
|
|
common_df.columns = ['poi_name', 'poi_tags', 'poi_url_base', 'poi_code'] |
99
|
|
|
try: |
100
|
|
|
for index, poi_common_data in common_df.iterrows(): |
101
|
|
|
get_or_create(session, POI_common, **poi_common_data) |
102
|
|
|
except Exception as e: |
103
|
|
|
logging.error('Rolled back: %s.', e) |
104
|
|
|
logging.error(poi_common_data) |
105
|
|
|
logging.error(traceback.print_exc()) |
106
|
|
|
session.rollback() |
107
|
|
|
else: |
108
|
|
|
logging.info('Successfully added %s common items to the dataset.', len(common_df)) |
109
|
|
|
session.commit() |
110
|
|
|
|
111
|
|
|
|
112
|
|
|
def search_for_postcode(session, city_name): |
113
|
|
|
city_col = session.query(City.city_post_code).filter(City.city_name == city_name).all() |
114
|
|
|
if len(city_col) == 1: |
115
|
|
|
return city_col |
116
|
|
|
else: |
117
|
|
|
logging.info('Cannot determine the post code from city name (%s).', city_name) |
118
|
|
|
return None |
119
|
|
|
|
120
|
|
|
|
121
|
|
|
def insert_poi_dataframe(session, poi_df): |
122
|
|
|
poi_df.columns = POI_COLS |
123
|
|
|
poi_df[['poi_postcode']] = poi_df[['poi_postcode']].fillna('0000') |
124
|
|
|
poi_df[['poi_postcode']] = poi_df[['poi_postcode']].astype(int) |
125
|
|
|
poi_dict = poi_df.to_dict('records') |
126
|
|
|
try: |
127
|
|
|
for poi_data in poi_dict: |
128
|
|
|
city_col = session.query(City.city_id).filter(City.city_name == poi_data['poi_city']).filter( |
129
|
|
|
City.city_post_code == poi_data['poi_postcode']).first() |
130
|
|
|
common_col = session.query(POI_common.pc_id).filter(POI_common.poi_code == poi_data['poi_code']).first() |
131
|
|
|
poi_data['poi_addr_city'] = city_col |
132
|
|
|
poi_data['poi_common_id'] = common_col |
133
|
|
|
if 'poi_name' in poi_data: del poi_data['poi_name'] |
134
|
|
|
if 'poi_code' in poi_data: del poi_data['poi_code'] |
135
|
|
|
get_or_create_poi(session, POI_address, **poi_data) |
136
|
|
|
except Exception as e: |
137
|
|
|
logging.error('Rolled back: %s.', e) |
138
|
|
|
logging.error(poi_data) |
139
|
|
|
logging.error(traceback.print_exc()) |
140
|
|
|
session.rollback() |
141
|
|
|
raise (e) |
142
|
|
|
else: |
143
|
|
|
try: |
144
|
|
|
session.commit() |
145
|
|
|
logging.info('Successfully added %s POI items to the dataset.', len(poi_dict)) |
146
|
|
|
except Exception as e: |
147
|
|
|
logging.error('Unsuccessfull commit: %s.', e) |
148
|
|
|
logging.error(traceback.print_exc()) |
149
|
|
|
|
150
|
|
|
|
151
|
|
|
def insert_type(session, type_data): |
152
|
|
|
try: |
153
|
|
|
for i in type_data: |
154
|
|
|
get_or_create(session, POI_common, **i) |
155
|
|
|
except Exception as e: |
156
|
|
|
logging.error('Rolled back: %s.', e) |
157
|
|
|
logging.error(i) |
158
|
|
|
logging.error(traceback.print_exc()) |
159
|
|
|
session.rollback() |
160
|
|
|
else: |
161
|
|
|
logging.info('Successfully added %s type items to the dataset.', len(type_data)) |
162
|
|
|
session.commit() |
163
|
|
|
|
164
|
|
|
|
165
|
|
|
def insert(session, **kwargs): |
166
|
|
|
try: |
167
|
|
|
city_col = session.query(City.city_id).filter(City.city_name == kwargs['poi_city']).filter( |
168
|
|
|
City.city_post_code == kwargs['poi_postcode']).first() |
169
|
|
|
common_col = session.query(POI_common.pc_id).filter(POI_common.poi_code == kwargs['poi_code']).first() |
170
|
|
|
kwargs['poi_addr_city'] = city_col |
171
|
|
|
kwargs['poi_common_id'] = common_col |
172
|
|
|
kwargs['poi_hash'] = hashlib.sha512( |
173
|
|
|
'{}{}{}{}{}{}'.format(kwargs['poi_code'], kwargs['poi_postcode'], kwargs['poi_city'], |
174
|
|
|
kwargs['poi_addr_street'], kwargs['poi_addr_housenumber'], |
175
|
|
|
kwargs['poi_conscriptionnumber']).lower().replace(' ', '').encode( |
176
|
|
|
'utf-8')).hexdigest() |
177
|
|
|
if 'poi_name' in kwargs: del kwargs['poi_name'] |
178
|
|
|
if 'poi_code' in kwargs: del kwargs['poi_code'] |
179
|
|
|
get_or_create(session, POI_address, **kwargs) |
180
|
|
|
except Exception as e: |
181
|
|
|
logging.error('Rolled back: %s.', e) |
182
|
|
|
logging.error(kwargs) |
183
|
|
|
logging.error(traceback.print_exc()) |
184
|
|
|
session.rollback() |
185
|
|
|
else: |
186
|
|
|
logging.debug('Successfully added the item to the dataset.') |
187
|
|
|
session.commit() |
188
|
|
|
finally: |
189
|
|
|
session.close() |
190
|
|
|
|