Masoutis

masoutis.py

# -*- coding: utf-8 -*-
import json

from scrapy import Request

from src.scraper.shops.spiders.base import SupermarketSpider

FORM_URL = 'https://eshop.masoutis.gr/WcfScanNShopForWeb/OrdersService.svc/GetPromoItemWithListCouponsSubCategories/'

HEADERS = {'Content-Type': 'application/json; charset=utf-8', 'Host': 'eshop.masoutis.gr',
           'Origin': 'https://eshop.masoutis.gr/', 'X-Requested-With': 'XMLHttpRequest',
           'Referer': 'https://eshop.masoutis.gr/'}


def format_price(p):
    try:
        return float(p)
    except:
        return None


class MasoutisSpider(SupermarketSpider):
    name = 'masoutis'
    allowed_domains = ['eshop.masoutis.gr']
    app_key = ''
    app_uid = ''
    app_usl = ''
    download_delay = 15

    def authed_request(self, **kwargs):
        return Request(headers={**HEADERS,
                                'Key': self.app_key,
                                'Uid': self.app_uid,
                                'Usl': self.app_usl},
                       **kwargs)

    def make_cat_request(self, cat_id, page):
        REQ_DATA = {
            "PassKey": "Sc@NnSh0p",
            "Itemcode": cat_id,
            "ItemDescr": "0",
            "IfWeight": page
        }
        return self.authed_request(
            url='https://eshop.masoutis.gr/WcfScanNShopForWeb/OrdersService.svc/GetPromoItemWithListCouponsSubCategories/',
            method='POST',
            callback=self.parse_catalog,
            body=json.dumps(REQ_DATA),
            meta={'page': page, 'cat_id': cat_id})

    def start_requests(self):
        yield Request(url='https://eshop.masoutis.gr/Home/GetGred',
                      headers=HEADERS,
                      method='POST',
                      body=json.dumps({'PassTemp': 'ScanNShop$c@nNSh0p'}),
                      callback=self.request_categories)

    def request_categories(self, response):
        data = json.loads(response.body)
        self.app_key = data['Key']
        self.app_uid = data['Uid']
        self.app_usl = data['Usl']

        yield self.authed_request(
            url='https://eshop.masoutis.gr/WcfScanNShopForWeb/OrdersService.svc/GetScanNShopMenuAllLevels/',
            method='POST',
            body=json.dumps({'PassKey': 'Sc@NnSh0p'}),
            callback=self.parse_categories)

    def parse_categories(self, response):
        data = json.loads(response.body)
        cats = set([c['HeaderMenuItem'] for c in data if c['HeaderMenuItemLinkDescr'] != 'oi-sunergates-mas'])

        for cat_id in cats:
            yield self.make_cat_request(cat_id=cat_id,
                                        page=1)

    def parse_catalog(self, response):
        data = json.loads(response.body)
        cat_id = response.meta['cat_id']

        if len(data) <= 0:
            return None

        for item in data:
            img_url = item['PhotoData']
            offer = item['Discount']
            url = item['ItemDescrLink']

            if offer:
                offer = offer.strip()

            image_item, image_hash = self.get_image_item(img_url)

            yield image_item

            self.insert_item(
                item=item.get('ItemDescr', None),
                key=url,
                offer=offer,
                url=url,
                sale_price=format_price(item.get('PosPrice', None)),
                price=format_price(item.get('StartPrice', None)),
                image=image_hash
            )

        page = response.meta['page']

        next_page = page + 1

        yield self.make_cat_request(cat_id=cat_id, page=next_page)