示例#1
0
    def parse(self, response):
        container = response.css('div.rte')
        items = container.css('p')
        lastLoader = None
        for item in items:

            txts = item.css('::text')
            if len(txts) > 2:
                loader = ItemLoader(item=StoreItem())
                for sel in txts:
                    txt = sel.get()
                    if '___' in txt or txt == ' ':
                        continue

                    if 'Tel' in txt:
                        loader.add_value('number', txt[5:])
                    elif 'Address' in txt:
                        loader.add_value('address', txt[9:])
                    else:
                        loader.add_value('brandName', txt)
                        loader.add_value('u_id', txt)
                lastLoader = loader

            iframesrc = item.css('iframe::attr(src)').get()
            if iframesrc != None and lastLoader != None:
                six = iframesrc.index('!2d')
                eix = iframesrc.index('!3d')
                fix = iframesrc.index('!', eix + 3)
                longitude = iframesrc[six + 3:eix]
                latitude = iframesrc[eix + 3:fix]

                lastLoader.add_value('latitude', latitude)
                lastLoader.add_value('longitude', longitude)
                yield loader.load_item()
示例#2
0
 def parse(self, response):
     container = response.css('div.resp-tabs-container div.resp_container')[2]
     items = container.css('div.row')
     for item in items:
         loader = ItemLoader(item=StoreItem(), selector = item)
         loader.add_css('brandName', 'h1::text')
         yield loader.load_item()
示例#3
0
 def parse(self, response):
     for marker in response.css('div.marker'):
         loader = ItemLoader(item=StoreItem(), selector=marker)
         loader.add_css('brandName', 'h4::text')
         loader.add_css('latitude', '::attr(data-lat)')
         loader.add_css('longitude', '::attr(data-lng)')
         loader.add_css('address', 'p.address::text')
         loader.add_value('u_id', marker.css('h4::text').get())
         yield response.follow(marker.css('a')[0],
                               callback=self.parse_store,
                               meta={'store_item': loader.load_item()})
示例#4
0
    def parse_store(self, response):
        brandName = response.css('span.LocationName-geo::text').get()
        number = response.css('div.phone-main::text').get()
        core_address = response.css('div.Core-address')
        address = core_address.css('div.c-AddressRow').css(
            'span::text').getall()
        address = ' '.join(address)
        coords = getcoordinates.get_coordinates(address)

        loader = ItemLoader(item=StoreItem())
        loader.add_value('brandName', brandName)
        loader.add_value('number', number)
        loader.add_value('address', address)
        loader.add_value('latitude', coords['latitude'])
        loader.add_value('longitude', coords['longitude'])
        print(loader.load_item())
示例#5
0
    def parse_stores(self, response):

        for store in response.css('div.locator-result'):
            loader = ItemLoader(item=StoreItem())
            brandName = store.css('h4::text').get().replace(' Spur', '')
            number = store.css('strong[itemprop="telephone"]::text').get()
            address = store.css('p[itemprop="address"]::text').get()
            latitude = store.css(
                'meta[itemprop="latitude"]::attr(content)').get()
            longitude = store.css(
                'meta[itemprop="longitude"]::attr(content)').get()

            loader.add_value('brandName', brandName)
            loader.add_value('number', number)
            loader.add_value('address', address)
            loader.add_value('latitude', latitude)
            loader.add_value('longitude', longitude)
            print(loader.load_item())
示例#6
0
文件: adega.py 项目: kyri33/python
    def parse(self, response):
        items = response.css('div.pos-top')

        for item in items:
            loader = ItemLoader(item=StoreItem())
            brandName = item.css('h3::text').get()
            if brandName == None:
                continue
            loader.add_value('brandName', brandName)
            details = item.css('p::text').getall()
            number = details[0].split('/')[0]
            loader.add_value('number', number)
            address = details[3]
            loader.add_value('address', address)
            coords = get_coordinates(address)
            loader.add_value('latitude', coords['latitude'])
            loader.add_value('longitude', coords['longitude'])
            loader.add_value('u_id', brandName)
            print(loader.load_item())
示例#7
0
    def parse_store(self, response):
        brandName = response.css('h1::text').get().replace('PIATTO',
                                                           '').strip()
        address = ' '.join(
            response.css('div.medium-3')[0].css('p')[0].css('::text').getall())
        number = response.css('div.medium-3')[0].css('p')[1].css(
            '::text').get()
        coords = response.css('iframe::attr(src)').get()
        longitude = coords[coords.index('!2d') + 3:coords.index('!3d')]
        latitude = coords[coords.index('!3d') + 3:coords.index('!2m')]
        u_id = brandName

        loader = ItemLoader(item=StoreItem())
        loader.add_value('brandName', brandName)
        loader.add_value('address', address)
        loader.add_value('number', number)
        loader.add_value('longitude', longitude)
        loader.add_value('latitude', latitude)
        loader.add_value('u_id', u_id)
        print(loader.load_item())
示例#8
0
    def parse_store(self, response):
        brandName = response.css(
            'h2.title-bar-medium-left::text').get().replace('Del Forno',
                                                            '').strip()
        info = response.css('div.contact-us-left').css('li')[0]
        address = info.css('p::text').get()
        coords = info.css('p')[1].css('::text').get()
        latitude = coords[coords.index('-'):coords.index('|')].strip()
        longitude = coords[coords.index('|') + 1:].strip()
        number = response.css('a.my_call::text').get()
        u_id = brandName

        loader = ItemLoader(item=StoreItem())
        loader.add_value('brandName', brandName)
        loader.add_value('address', address)
        loader.add_value('number', number)
        loader.add_value('u_id', u_id)
        loader.add_value('latitude', latitude)
        loader.add_value('longitude', longitude)
        print(loader.load_item())
示例#9
0
    def parse(self, response):

        js = response.css('div.content').css('script::text').get()
        data = js[js.index('[{') :]
        data = data[0: data.index("}]'") + 2]
        
        results = json.loads(data)
        for store in results:
            loader = ItemLoader(item=StoreItem())
            brandName = store['DisplayName'].replace("John Dory's ", '')
            number = store['TelephoneNumber']
            address = store['StreetAddress'] + " " + store['Suburb'] + " " + store['City']
            latitude = store['Latitude']
            longitude = store['Longitude']
            u_id = brandName

            loader.add_value('u_id', u_id)
            loader.add_value('address', address)
            loader.add_value('brandName', brandName)
            loader.add_value('number', number)
            loader.add_value('latitude', latitude)
            loader.add_value('longitude', longitude)
示例#10
0
import sys
import os

sys.path.append("./")

from storesync.items import StoreItem
from scrapy.loader import ItemLoader
from storesync.pipelines import StoresyncPipeline

url = "https://oceanbasket.com/dev/site/ajaxgetbranches?site_id=1&ignore=true"

response = requests.post(url)
if not response:
    print("Error")
    exit(2)

results = json.loads(response.content)

posts = []
items = []

for store in results['branches']:
    loader = ItemLoader(item=StoreItem())
    loader.add_value('u_id', store['id'])
    loader.add_value('address', store['meta_address'])
    loader.add_value('number', store['meta_phone'])
    loader.add_value('latitude', store['meta_latitude'])
    loader.add_value('longitude', store['meta_longitude'])
    loader.add_value('brandName', store['meta_title'])
    # TODO CAN FILTER PROVINCES
    print(loader.load_item())