Added geocodio geocoder to the Python library
This commit is contained in:
@@ -2,10 +2,12 @@ from google import GoogleMapsBulkGeocoder
|
||||
from here import HereMapsBulkGeocoder
|
||||
from tomtom import TomTomBulkGeocoder
|
||||
from mapbox import MapboxBulkGeocoder
|
||||
from geocodio import GeocodioBulkGeocoder
|
||||
|
||||
BATCH_GEOCODER_CLASS_BY_PROVIDER = {
|
||||
'google': GoogleMapsBulkGeocoder,
|
||||
'heremaps': HereMapsBulkGeocoder,
|
||||
'tomtom': TomTomBulkGeocoder,
|
||||
'mapbox': MapboxBulkGeocoder
|
||||
'mapbox': MapboxBulkGeocoder,
|
||||
'geocodio': GeocodioBulkGeocoder,
|
||||
}
|
||||
|
||||
@@ -0,0 +1,2 @@
|
||||
from geocoder import GeocodioGeocoder
|
||||
from bulk_geocoder import GeocodioBulkGeocoder
|
||||
@@ -0,0 +1,76 @@
|
||||
import requests
|
||||
from cartodb_services import StreetPointBulkGeocoder
|
||||
from cartodb_services.geocodio import GeocodioGeocoder
|
||||
from iso3166 import countries
|
||||
from cartodb_services.tools.country import country_to_iso3
|
||||
|
||||
|
||||
class GeocodioBulkGeocoder(GeocodioGeocoder, StreetPointBulkGeocoder):
|
||||
MAX_BATCH_SIZE = 100 # Setting an upper limit (not stated in the documentation)
|
||||
MIN_BATCHED_SEARCH = 0
|
||||
READ_TIMEOUT = 60
|
||||
CONNECT_TIMEOUT = 10
|
||||
MAX_RETRIES = 1
|
||||
|
||||
def __init__(self, token, logger, service_params=None):
|
||||
GeocodioGeocoder.__init__(self, token, logger, service_params)
|
||||
|
||||
self.connect_timeout = self.CONNECT_TIMEOUT
|
||||
self.read_timeout = self.READ_TIMEOUT
|
||||
self.max_retries = self.MAX_RETRIES
|
||||
|
||||
if service_params is not None:
|
||||
self.connect_timeout = service_params.get('connect_timeout', self.CONNECT_TIMEOUT)
|
||||
self.read_timeout = service_params.get('read_timeout', self.READ_TIMEOUT)
|
||||
self.max_retries = service_params.get('max_retries', self.MAX_RETRIES)
|
||||
|
||||
self.session = requests.Session()
|
||||
|
||||
def _should_use_batch(self, searches):
|
||||
return len(searches) >= self.MIN_BATCHED_SEARCH
|
||||
|
||||
def _serial_geocode(self, searches):
|
||||
results = []
|
||||
for search in searches:
|
||||
elements = self._encoded_elements(search)
|
||||
result = self.geocode_meta(*elements)
|
||||
|
||||
if result:
|
||||
results.append((search[0], result[0], result[1]))
|
||||
else:
|
||||
results.append((search[0], None, None))
|
||||
|
||||
return results
|
||||
|
||||
def _encoded_elements(self, search):
|
||||
(search_id, address, city, state, country) = search
|
||||
address = address.encode('utf-8') if address else None
|
||||
city = city.encode('utf-8') if city else None
|
||||
state = state.encode('utf-8') if state else None
|
||||
country = self._country_code(country) if country else None
|
||||
return address, city, state, country
|
||||
|
||||
def _batch_geocode(self, searches):
|
||||
if len(searches) == 1:
|
||||
return self._serial_geocode(searches)
|
||||
else:
|
||||
frees = []
|
||||
for search in searches:
|
||||
elements = self._encoded_elements(search)
|
||||
free = ', '.join([elem for elem in elements if elem])
|
||||
frees.append(free)
|
||||
|
||||
full_results = self.geocode_free_text_meta(frees)
|
||||
results = []
|
||||
for s, r in zip(searches, full_results):
|
||||
results.append((s[0], r[0], r[1]))
|
||||
return results
|
||||
|
||||
def _country_code(self, country):
|
||||
country_iso3166 = None
|
||||
country_iso3 = country_to_iso3(country)
|
||||
if country_iso3:
|
||||
country_iso3166 = countries.get(country_iso3).alpha2.lower()
|
||||
|
||||
return country_iso3166
|
||||
|
||||
@@ -0,0 +1,125 @@
|
||||
from geocodio import GeocodioClient
|
||||
from geocodio.exceptions import GeocodioAuthError, GeocodioServerError, GeocodioDataError, GeocodioError
|
||||
|
||||
from cartodb_services.tools.qps import qps_retry
|
||||
from cartodb_services.metrics import Traceable
|
||||
from cartodb_services.geocoder import EMPTY_RESPONSE, geocoder_metadata
|
||||
from cartodb_services.tools.exceptions import ServiceException
|
||||
|
||||
|
||||
RELEVANCE_BY_LOCATION_TYPE = {
|
||||
'rooftop': 1,
|
||||
'point': 0.9,
|
||||
'range_interpolation': 0.8,
|
||||
'nearest_rooftop_match': 0.7,
|
||||
'intersection': 0.6,
|
||||
'street_center': 0.5,
|
||||
'place': 0.4,
|
||||
'state': 0.1,
|
||||
}
|
||||
|
||||
|
||||
class GeocodioGeocoder(Traceable):
|
||||
'''
|
||||
Python wrapper for the Geocodio Geocoder service.
|
||||
'''
|
||||
|
||||
def __init__(self, token, logger, service_params=None):
|
||||
service_params = service_params or {}
|
||||
self._token = token
|
||||
self._logger = logger
|
||||
|
||||
self._geocoder = GeocodioClient(self._token)
|
||||
|
||||
def _validate_input(self, searchtext, city=None, state_province=None,
|
||||
country=None):
|
||||
if searchtext and searchtext.strip():
|
||||
return True
|
||||
elif city:
|
||||
return True
|
||||
elif state_province:
|
||||
return True
|
||||
|
||||
return False
|
||||
|
||||
@qps_retry(qps=15, provider='geocodio')
|
||||
def geocode(self, searchtext, city=None, state_province=None,
|
||||
country=None):
|
||||
return self._geocode_meta(searchtext, city, state_province, country)[0]
|
||||
|
||||
def geocode_meta(self, searchtext, city=None, state_province=None,
|
||||
country=None):
|
||||
return self._geocode_meta(searchtext, city, state_province, country)[0]
|
||||
|
||||
@qps_retry(qps=15, provider='geocodio')
|
||||
def _geocode_meta(self, searchtext, city=None, state_province=None,
|
||||
country=None):
|
||||
if not self._validate_input(searchtext, city, state_province, country):
|
||||
return EMPTY_RESPONSE
|
||||
|
||||
try:
|
||||
free_text_components = [searchtext, city, state_province, country]
|
||||
response = self._geocoder.geocode(';'.join([c for c in free_text_components if c is not None and c.strip()]))
|
||||
|
||||
return self._parse_geocoder_response(response)
|
||||
except GeocodioDataError as gde:
|
||||
return EMPTY_RESPONSE
|
||||
except GeocodioAuthError as gae:
|
||||
raise ServiceException('Geocodio authorization error: ' + str(gae), None)
|
||||
except GeocodioServerError as gse:
|
||||
raise ServiceException('geocodio server error: ' + str(gse), None)
|
||||
except GeocodioError as ge:
|
||||
raise ServiceException('Unknown Geocodio error: ' + str(ge), None)
|
||||
|
||||
@qps_retry(qps=15)
|
||||
def geocode_free_text_meta(self, free_searches, country=None):
|
||||
"""
|
||||
:param free_searches: Free text searches
|
||||
:return: list of [x, y] on success, [] on error
|
||||
"""
|
||||
output = []
|
||||
|
||||
try:
|
||||
if country:
|
||||
free_searches = ['{s}, {country}'.format(s, country) for s in free_searches]
|
||||
|
||||
responses = self._geocoder.geocode(free_searches)
|
||||
|
||||
for response in responses:
|
||||
output.append(self._parse_geocoder_response(response))
|
||||
except GeocodioDataError as gde:
|
||||
return EMPTY_RESPONSE
|
||||
except GeocodioAuthError as gae:
|
||||
raise ServiceException('Geocodio authorization error: ' + str(gae), None)
|
||||
except GeocodioServerError as gse:
|
||||
raise ServiceException('geocodio server error: ' + str(gse), None)
|
||||
except GeocodioError as ge:
|
||||
raise ServiceException('Unknown Geocodio error: ' + str(ge), None)
|
||||
|
||||
return output
|
||||
|
||||
def _parse_geocoder_response(self, response):
|
||||
if response is None or not response:
|
||||
return EMPTY_RESPONSE
|
||||
|
||||
if response.get('results') is None or not response.get('results'):
|
||||
return EMPTY_RESPONSE
|
||||
|
||||
if response.coords is None or not response.coords:
|
||||
return EMPTY_RESPONSE
|
||||
|
||||
coords = [None, None]
|
||||
accuracy = None
|
||||
accuracy_type = None
|
||||
|
||||
accuracy = response.accuracy
|
||||
|
||||
if response.coords is not None and response.coords:
|
||||
coords = [response.coords[1], response.coords[0]]
|
||||
|
||||
if response.get('results'):
|
||||
accuracy_type = response.get('results')[0].get('accuracy_type')
|
||||
|
||||
metadata = geocoder_metadata(RELEVANCE_BY_LOCATION_TYPE.get(accuracy_type), response.accuracy, accuracy_type)
|
||||
|
||||
return [coords, metadata]
|
||||
@@ -0,0 +1 @@
|
||||
GEOCODIO_GEOCODER_APIKEY_ROUNDROBIN = 'geocodio_geocoder_apikey_roundrobin'
|
||||
@@ -405,6 +405,8 @@ class GeocoderConfig(ServiceConfig):
|
||||
MAPBOX_GEOCODER_API_KEYS = 'mapbox_geocoder_api_keys'
|
||||
TOMTOM_GEOCODER = 'tomtom'
|
||||
TOMTOM_GEOCODER_API_KEYS = 'tomtom_geocoder_api_keys'
|
||||
GEOCODIO_GEOCODER = 'geocodio'
|
||||
GEOCODIO_GEOCODER_API_KEYS = 'geocodio_geocoder_api_keys'
|
||||
QUOTA_KEY = 'geocoding_quota'
|
||||
SOFT_LIMIT_KEY = 'soft_geocoding_limit'
|
||||
USERNAME_KEY = 'username'
|
||||
@@ -437,6 +439,9 @@ class GeocoderConfig(ServiceConfig):
|
||||
elif self._geocoder_provider == self.TOMTOM_GEOCODER:
|
||||
if not self.tomtom_api_keys:
|
||||
raise ConfigException("""TomTom config is not set up""")
|
||||
elif self._geocoder_provider == self.GEOCODIO_GEOCODER:
|
||||
if not self.geocodio_api_keys:
|
||||
raise ConfigException("""Geocodio config is not set up""")
|
||||
|
||||
return True
|
||||
|
||||
@@ -476,6 +481,10 @@ class GeocoderConfig(ServiceConfig):
|
||||
self._tomtom_api_keys = db_config.tomtom_geocoder_api_keys
|
||||
self._cost_per_hit = 0
|
||||
self._tomtom_service_params = db_config.tomtom_geocoder_service_params
|
||||
elif self._geocoder_provider == self.GEOCODIO_GEOCODER:
|
||||
self._geocodio_api_keys = db_config.geocodio_geocoder_api_keys
|
||||
self._cost_per_hit = 0
|
||||
self._geocodio_service_params = db_config.geocodio_geocoder_service_params
|
||||
|
||||
@property
|
||||
def service_type(self):
|
||||
@@ -489,6 +498,8 @@ class GeocoderConfig(ServiceConfig):
|
||||
return 'geocoder_tomtom'
|
||||
elif self._geocoder_provider == self.NOKIA_GEOCODER:
|
||||
return 'geocoder_here'
|
||||
elif self._geocoder_provider == self.GEOCODIO_GEOCODER:
|
||||
return 'geocoder_geocodio'
|
||||
|
||||
@property
|
||||
def heremaps_geocoder(self):
|
||||
@@ -510,6 +521,10 @@ class GeocoderConfig(ServiceConfig):
|
||||
def tomtom_geocoder(self):
|
||||
return self._geocoder_provider == self.TOMTOM_GEOCODER
|
||||
|
||||
@property
|
||||
def geocodio_geocoder(self):
|
||||
return self._geocoder_provider == self.GEOCODIO_GEOCODER
|
||||
|
||||
@property
|
||||
def google_client_id(self):
|
||||
return self._google_maps_client_id
|
||||
@@ -569,6 +584,14 @@ class GeocoderConfig(ServiceConfig):
|
||||
def tomtom_service_params(self):
|
||||
return self._tomtom_service_params
|
||||
|
||||
@property
|
||||
def geocodio_api_keys(self):
|
||||
return self._geocodio_api_keys
|
||||
|
||||
@property
|
||||
def geocodio_service_params(self):
|
||||
return self._geocodio_service_params
|
||||
|
||||
@property
|
||||
def is_high_resolution(self):
|
||||
return True
|
||||
@@ -600,6 +623,7 @@ class ServicesDBConfig:
|
||||
self._get_mapzen_config()
|
||||
self._get_mapbox_config()
|
||||
self._get_tomtom_config()
|
||||
self._get_geocodio_config()
|
||||
self._get_data_observatory_config()
|
||||
|
||||
def _get_server_config(self):
|
||||
@@ -679,6 +703,16 @@ class ServicesDBConfig:
|
||||
self._tomtom_geocoder_quota = tomtom_conf['geocoder']['monthly_quota']
|
||||
self._tomtom_geocoder_service_params = tomtom_conf['geocoder'].get('service', {})
|
||||
|
||||
def _get_geocodio_config(self):
|
||||
geocodio_conf_json = self._get_conf('geocodio_conf')
|
||||
if not geocodio_conf_json:
|
||||
raise ConfigException('Geocodio configuration missing')
|
||||
else:
|
||||
geocodio_conf = json.loads(geocodio_conf_json)
|
||||
self._geocodio_geocoder_api_keys = geocodio_conf['geocoder']['api_keys']
|
||||
self._geocodio_geocoder_quota = geocodio_conf['geocoder']['monthly_quota']
|
||||
self._geocodio_geocoder_service_params = geocodio_conf['geocoder'].get('service', {})
|
||||
|
||||
def _get_data_observatory_config(self):
|
||||
do_conf_json = self._get_conf('data_observatory_conf')
|
||||
if not do_conf_json:
|
||||
@@ -848,6 +882,18 @@ class ServicesDBConfig:
|
||||
def tomtom_geocoder_service_params(self):
|
||||
return self._tomtom_geocoder_service_params
|
||||
|
||||
@property
|
||||
def geocodio_geocoder_api_keys(self):
|
||||
return self._geocodio_geocoder_api_keys
|
||||
|
||||
@property
|
||||
def geocodio_geocoder_monthly_quota(self):
|
||||
return self.geocodio_geocoder_quota
|
||||
|
||||
@property
|
||||
def geocodio_geocoder_service_params(self):
|
||||
return self._geocodio_geocoder_service_params
|
||||
|
||||
@property
|
||||
def data_observatory_connection_str(self):
|
||||
return self._data_observatory_connection_str
|
||||
|
||||
+128
@@ -0,0 +1,128 @@
|
||||
from dateutil.parser import parse as date_parse
|
||||
from cartodb_services.refactor.service.utils import round_robin
|
||||
from cartodb_services.geocodio.types import GEOCODIO_GEOCODER_APIKEY_ROUNDROBIN
|
||||
|
||||
|
||||
class GeocodioGeocoderConfig(object):
|
||||
"""
|
||||
Configuration needed to operate the Geocodio geocoder service.
|
||||
"""
|
||||
|
||||
def __init__(self,
|
||||
geocoding_quota,
|
||||
soft_geocoding_limit,
|
||||
period_end_date,
|
||||
cost_per_hit,
|
||||
log_path,
|
||||
geocodio_api_keys,
|
||||
username,
|
||||
organization,
|
||||
service_params,
|
||||
GD):
|
||||
self._geocoding_quota = geocoding_quota
|
||||
self._soft_geocoding_limit = soft_geocoding_limit
|
||||
self._period_end_date = period_end_date
|
||||
self._cost_per_hit = cost_per_hit
|
||||
self._log_path = log_path
|
||||
self._geocodio_api_keys = geocodio_api_keys
|
||||
self._username = username
|
||||
self._organization = organization
|
||||
self._service_params = service_params
|
||||
self._GD = GD
|
||||
|
||||
@property
|
||||
def service_type(self):
|
||||
return 'geocoder_geocodio'
|
||||
|
||||
@property
|
||||
def provider(self):
|
||||
return 'geocodio'
|
||||
|
||||
@property
|
||||
def is_high_resolution(self):
|
||||
return True
|
||||
|
||||
@property
|
||||
def geocoding_quota(self):
|
||||
return self._geocoding_quota
|
||||
|
||||
@property
|
||||
def soft_geocoding_limit(self):
|
||||
return self._soft_geocoding_limit
|
||||
|
||||
@property
|
||||
def period_end_date(self):
|
||||
return self._period_end_date
|
||||
|
||||
@property
|
||||
def cost_per_hit(self):
|
||||
return self._cost_per_hit
|
||||
|
||||
@property
|
||||
def log_path(self):
|
||||
return self._log_path
|
||||
|
||||
@property
|
||||
def geocodio_api_key(self):
|
||||
return round_robin(self._geocodio_api_keys, self._GD,
|
||||
GEOCODIO_GEOCODER_APIKEY_ROUNDROBIN)
|
||||
|
||||
@property
|
||||
def username(self):
|
||||
return self._username
|
||||
|
||||
@property
|
||||
def organization(self):
|
||||
return self._organization
|
||||
|
||||
@property
|
||||
def service_params(self):
|
||||
return self._service_params
|
||||
|
||||
# TODO: for BW compat, remove
|
||||
@property
|
||||
def google_geocoder(self):
|
||||
return False
|
||||
|
||||
|
||||
class GeocodioGeocoderConfigBuilder(object):
|
||||
|
||||
def __init__(self, server_conf, user_conf, org_conf, username, orgname, GD):
|
||||
self._server_conf = server_conf
|
||||
self._user_conf = user_conf
|
||||
self._org_conf = org_conf
|
||||
self._username = username
|
||||
self._orgname = orgname
|
||||
self._GD = GD
|
||||
|
||||
def get(self):
|
||||
geocodio_server_conf = self._server_conf.get('geocodio_conf')
|
||||
geocodio_api_keys = geocodio_server_conf['geocoder']['api_keys']
|
||||
geocodio_service_params = geocodio_server_conf['geocoder'].get('service', {})
|
||||
|
||||
geocoding_quota = self._get_quota()
|
||||
soft_geocoding_limit = self._user_conf.get('soft_geocoding_limit').lower() == 'true'
|
||||
cost_per_hit = 0
|
||||
period_end_date_str = self._org_conf.get('period_end_date') or self._user_conf.get('period_end_date')
|
||||
period_end_date = date_parse(period_end_date_str)
|
||||
|
||||
logger_conf = self._server_conf.get('logger_conf')
|
||||
log_path = logger_conf.get('geocoder_log_path', None)
|
||||
|
||||
return GeocodioGeocoderConfig(geocoding_quota,
|
||||
soft_geocoding_limit,
|
||||
period_end_date,
|
||||
cost_per_hit,
|
||||
log_path,
|
||||
geocodio_api_keys,
|
||||
self._username,
|
||||
self._orgname,
|
||||
geocodio_service_params,
|
||||
self._GD)
|
||||
|
||||
def _get_quota(self):
|
||||
geocoding_quota = self._org_conf.get('geocoding_quota') or self._user_conf.get('geocoding_quota')
|
||||
if geocoding_quota is '':
|
||||
return 0
|
||||
|
||||
return int(geocoding_quota)
|
||||
Reference in New Issue
Block a user