Ejemplo n.º 1
0
def by_place_id(list_places, city_id):
    config = Config('./config/config.yml')
    Key = Keygen()
    keyAPI = Key.get_key_geocode()
    if not keyAPI:
        sys.exit()

    cnf = {'googlemaps': {'geocoding': {'key': keyAPI}}}
    config.set(cnf)
    language = 'uk'
    doc_factory = DocFactory(config.get('mongodb'))

    spider = Spider(loader_factory=LoaderFactory,
                    gmap_parser=MapFactory.spain,
                    doc_factory=doc_factory,
                    language=language,
                    config=config,
                    use_cache=True)
    for loc in list_places:
        objects = spider.get_gmap_place_id(loc)
        gmap = {}
        try:
            gmap = objects[0].get_document()
            gmap["city_id"] = city_id
            exixts = db.ukraine_city_sublocal.find({
                "city_id": city_id,
                "code": gmap['code']
            }).count()
            if exixts < 1:
                gmap.pop('_id', None)
                db.ukraine_city_sublocal.save(gmap)
                print(gmap)

        except Exception as e:
            print(str(e))
Ejemplo n.º 2
0
from argparse import ArgumentParser
import sys
import json
import requests
import pymongo
from bson.json_util import dumps
from lib.keygen.gmap_keygen import Keygen

# from lib.parser.wiki.Spain import Spain as ParserSpain
country = 'Spain'
config = Config('./config/config.yml')
mongo_config = config.get('mongodb')
conn = pymongo.MongoClient(mongo_config['host'], mongo_config['port'])
Key = Keygen()
cnf = {'geocoding': {'key': Key.get_key_geocode()}}
config.set(cnf)
db = conn.location
coll = db.sinoplik_romania
# print(config.get('googlemaps').get('geocoding').get('key'))
doc_factory = DocFactory(config.get('mongodb'))
# try:
# 	skiprows = sys.argv[1]
# except Exception as e:
# 	skiprows = 0
# df = pd.read_csv('./data/spain/Spain_notDublicate.csv',  skiprows=int(skiprows), low_memory=False)
loader = Loader.loader_with_mongodb(config.get('mongodb'))
headers = {'User-Agent': 'Mozilla/5.0'}

language = 'ro'
# spider = Spider(
#     loader_factory=LoaderFactory,