|
1
|
|
|
# -*- coding: utf-8 -*- |
|
2
|
|
|
import logging |
|
3
|
|
|
|
|
4
|
|
|
import pyjobsweb.lib.geolocation as geolocation |
|
5
|
|
|
from pyjobsweb import model |
|
6
|
|
|
from pyjobsweb.commands import AppContextCommand |
|
7
|
|
|
from pyjobsweb.lib.lock import acquire_inter_process_lock |
|
8
|
|
|
|
|
9
|
|
|
|
|
10
|
|
|
class GeocodeCommand(AppContextCommand): |
|
11
|
|
|
""" |
|
12
|
|
|
Start geotagging operations for job offers and/or companies |
|
13
|
|
|
""" |
|
14
|
|
|
def __init__(self, *args, **kwargs): |
|
15
|
|
|
super(GeocodeCommand, self).__init__(args, kwargs) |
|
16
|
|
|
self._geolocator = geolocation.Geolocator() |
|
17
|
|
|
self._logger = logging.getLogger(__name__) |
|
18
|
|
|
|
|
19
|
|
|
def get_parser(self, prog_name): |
|
20
|
|
|
parser = super(GeocodeCommand, self).get_parser(prog_name) |
|
21
|
|
|
|
|
22
|
|
|
jobs_help_msg = 'geocode job offers stored in the jobs table' |
|
23
|
|
|
parser.add_argument('-j', '--jobs', |
|
24
|
|
|
help=jobs_help_msg, |
|
25
|
|
|
dest='geocode_jobs', |
|
26
|
|
|
action='store_const', const=True) |
|
27
|
|
|
|
|
28
|
|
|
companies_help_msg = 'geocode the companies stored in the companies ' \ |
|
29
|
|
|
'table' |
|
30
|
|
|
parser.add_argument('-co', '--companies', |
|
31
|
|
|
help=companies_help_msg, |
|
32
|
|
|
dest='geocode_companies', |
|
33
|
|
|
action='store_const', const=True) |
|
34
|
|
|
|
|
35
|
|
|
return parser |
|
36
|
|
|
|
|
37
|
|
|
def _logging(self, logging_level, message): |
|
38
|
|
|
self._logger.log(logging_level, message) |
|
39
|
|
|
|
|
40
|
|
|
def _job_id_logging(self, job_id, logging_level, message): |
|
41
|
|
|
log_msg = u'[Job offer id: %s] %s' % (job_id, message) |
|
42
|
|
|
self._logging(logging_level, log_msg) |
|
43
|
|
|
|
|
44
|
|
|
def _company_id_logging(self, company_id, logging_level, message): |
|
45
|
|
|
log_msg = u'[Company: %s] %s' % (company_id, message) |
|
46
|
|
|
self._logging(logging_level, log_msg) |
|
47
|
|
|
|
|
48
|
|
|
def _geocode(self, cls_to_geocode, id_logger): |
|
49
|
|
|
to_geoloc = cls_to_geocode.get_pending_geolocations() |
|
50
|
|
|
|
|
51
|
|
|
log_msg = 'Computing required geolocations...' |
|
52
|
|
|
self._logging(logging.INFO, log_msg) |
|
53
|
|
|
|
|
54
|
|
|
for obj in to_geoloc: |
|
55
|
|
|
obj_id = obj.id |
|
56
|
|
|
obj_address = obj.address |
|
57
|
|
|
|
|
58
|
|
|
log_msg = 'Resolving address: %s.' % obj_address |
|
59
|
|
|
id_logger(obj_id, logging.INFO, log_msg) |
|
60
|
|
|
|
|
61
|
|
|
try: |
|
62
|
|
|
location = self._geolocator.geocode(obj_address) |
|
63
|
|
|
except geolocation.GeolocationFailure as e: |
|
64
|
|
|
cls_to_geocode.set_address_is_valid(obj_id, False) |
|
65
|
|
|
cls_to_geocode.set_geolocation_is_valid(obj_id, False) |
|
66
|
|
|
id_logger(obj_id, logging.ERROR, e) |
|
67
|
|
|
except geolocation.TemporaryError as e: |
|
68
|
|
|
cls_to_geocode.set_geolocation_is_valid(obj_id, False) |
|
69
|
|
|
id_logger(obj_id, logging.WARNING, e) |
|
70
|
|
|
except geolocation.GeolocationError as e: |
|
71
|
|
|
cls_to_geocode.set_geolocation_is_valid(obj_id, False) |
|
72
|
|
|
id_logger(obj_id, logging.ERROR, e) |
|
73
|
|
|
else: |
|
74
|
|
|
log_msg = 'Successful geocoding: (lat: %s, lon: %s).' % ( |
|
75
|
|
|
location.latitude, location.longitude) |
|
76
|
|
|
id_logger(obj_id, logging.INFO, log_msg) |
|
77
|
|
|
|
|
78
|
|
|
cls_to_geocode.set_geolocation( |
|
79
|
|
|
obj_id, location.latitude, location.longitude) |
|
80
|
|
|
|
|
81
|
|
|
log_msg = 'Geolocations computed.' |
|
82
|
|
|
self._logging(logging.INFO, log_msg) |
|
83
|
|
|
|
|
84
|
|
|
def _geocode_job_offers(self): |
|
85
|
|
|
log_msg = 'Starting job offers geocoding operations...' |
|
86
|
|
|
self._logging(logging.INFO, log_msg) |
|
87
|
|
|
|
|
88
|
|
|
with acquire_inter_process_lock('geocode_job_offers') as acquired: |
|
89
|
|
|
if not acquired: |
|
90
|
|
|
err_msg = 'Another process is already performing geocoding' \ |
|
91
|
|
|
'operations on the job offers, aborting now.' |
|
92
|
|
|
self._logging(logging.WARNING, err_msg) |
|
93
|
|
|
else: |
|
94
|
|
|
self._geocode(model.JobAlchemy, self._job_id_logging) |
|
95
|
|
|
|
|
96
|
|
|
def _geocode_companies(self): |
|
97
|
|
|
log_msg = 'Starting companies geocoding operations...' |
|
98
|
|
|
self._logging(logging.INFO, log_msg) |
|
99
|
|
|
|
|
100
|
|
|
with acquire_inter_process_lock('geocode_companies') as acquired: |
|
101
|
|
|
if not acquired: |
|
102
|
|
|
err_msg = 'Another process is already performing geocoding ' \ |
|
103
|
|
|
'operations on the companies, aborting now.' |
|
104
|
|
|
self._logging(logging.WARNING, err_msg) |
|
105
|
|
|
else: |
|
106
|
|
|
self._geocode(model.CompanyAlchemy, self._company_id_logging) |
|
107
|
|
|
|
|
108
|
|
|
def take_action(self, parsed_args): |
|
109
|
|
|
super(GeocodeCommand, self).take_action(parsed_args) |
|
110
|
|
|
|
|
111
|
|
|
log_msg = 'Starting geocoding operations.' |
|
112
|
|
|
self._logging(logging.INFO, log_msg) |
|
113
|
|
|
|
|
114
|
|
|
# We do not run geocoding operation for the job offers and companies in |
|
115
|
|
|
# different sub-processes to prevent overloading the geocoding API. This |
|
116
|
|
|
# is why geocoding operations are run sequentially. |
|
117
|
|
|
if parsed_args.geocode_jobs: |
|
118
|
|
|
self._geocode_job_offers() |
|
119
|
|
|
|
|
120
|
|
|
if parsed_args.geocode_companies: |
|
121
|
|
|
self._geocode_companies() |
|
122
|
|
|
|
|
123
|
|
|
log_msg = 'Geocoding operations done.' |
|
124
|
|
|
self._logging(logging.INFO, log_msg) |
|
125
|
|
|
|
pyjobs /
web
