Bazaar

bazaar.py

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

from scrapy import Request

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

BASE_URL = 'https://www.bazaar-online.gr/CachedServices/api/Product/GetScrolledProductList?lang=el&templateCode=productList&configCode=&itemId={item_id}&itemPath={item_path}&pageNumber={page}&pageSize=96&sortKey=BestSelling-Descending'


class BazaarSpider(SupermarketSpider):
    name = 'bazaar'
    allowed_domains = ['bazaar-online.gr']
    start_urls = ['https://www.bazaar-online.gr/']

    def parse(self, response):
        cats = response.css('ul.nav-1 ul.nav-2 li.nav-2-item > a.title').xpath('@href').getall()

        for cat in cats:
            if not cat or not cat.startswith('/') or cat == '/prosfores':
                continue

            yield Request(url=f'https://www.bazaar-online.gr{cat}', callback=self.start_pagination)

    def start_pagination(self, response):
        item_id = re.findall(r"(?<=ItemId:')\d+(?=')", response.body.decode())[0]
        item_path = re.findall(r"(?<=ItemPathChecksum:')-?\d+(?=')", response.body.decode())[0]
        page = 1

        meta = dict(item_id=item_id, item_path=item_path, page=page)

        yield Request(url=BASE_URL.format(item_id=item_id, item_path=item_path, page=page),
                      callback=self.parse_catalog, meta=meta)

    def parse_catalog(self, response):
        data = json.loads(response.body)

        if len(data['data']['Items']) <= 0 or len(data['data']['Items'][0]) <= 0:
            return None

        for item in data['data']['Items'][0]:
            title = item['Title']
            url = item['CanonicalUrl']
            vendor_id = item['Id']
            price = format_price(item['Prices']['Price']['DeletedPrice'])
            sale_price = item['Prices']['Price']['CurrentPriceN']
            image_url = 'https:' + item['Images'][0]['List'].split('?')[0]

            image_item, image_hash = self.get_image_item(image_url)

            yield image_item

            self.insert_item(
                item=title,
                key=str(vendor_id),
                url=url,
                price=price,
                sale_price=sale_price,
                image=image_hash
            )

        meta = response.meta

        meta['page'] += 1

        yield Request(url=BASE_URL.format(item_id=meta['item_id'],
                                          item_path=meta['item_path'],
                                          page=meta['page']),
                      meta=meta,
                      callback=self.parse_catalog)