Advertisement
Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- {
- "start_urls":
- [
- "https://www.amazon.ca/s?k=whitewave+silk&i=grocery&dc&ref=a9_asc_1",
- "https://www.amazon.ca/s?k=international+delight&i=grocery&ref=nb_sb_noss",
- "https://www.amazon.ca/s?k=so+delicious&i=grocery&ref=nb_sb_noss_1"
- ],
- "fields": {
- "name": "//h1[@id='title']/span[@id='productTitle']/text()",
- "price": "//span[@id='priceblock_ourprice']/text()",
- "price_old": "//span[@class='priceBlockStrikePriceString a-text-strike']/text()",
- "reference": "",
- "image": "(//script[contains(.,'ImageBlockATF')])[1]/text()",
- "description": "//div[@id='productDescription']//text()",
- "brand": "",
- "brand_text": "//*[@id='bylineInfo_feature_div']/div/a/text()",
- "categories": "",
- "availability": "//div[contains(@id, 'availabilityInsideBuyBox')]//div[@id='availability']/span/text() | //div[@id='availability_feature_div']/div[@id='availability']/span/text()",
- "rating_average": "(//div[@id='averageCustomerReviews_feature_div']/div[@id='averageCustomerReviews']/span[@class='a-declarative']/span[@id='acrPopover']/@title)[1]",
- "rating_count": "(//span[@id='acrCustomerReviewText']/text())[1]",
- "barcode": "",
- "stock": "",
- "seller": "//div[@id='merchant-info']/a[@id='sellerProfileTriggerId']/text()"
- },
- "links": {
- "detail": "//h2/a[contains(@class,'a-text-normal') and not(contains(@href,'pantry'))]|//div[@class='s-item-container' and not(self::*//img[contains(@src,'pantry')])]//h2/parent::a[contains(@class,'a-text-normal')]",
- "next_page": "(//ul[@class='a-pagination']//a[(contains(.,'Next')) and not(contains(@class,'carousel-goto-nextpage'))])[last()]"
- },
- "regex": {
- "rating_average": ["(\\d+)", 1],
- "rating_count": ["(.*?)ratings", 1],
- "price": ["(\\d.*)", 1],
- "price_old": ["(\\d.*)", 1],
- "brand_text": ["Brand:\\s(.*)", 1],
- "image": ["hiRes\\\":\\\"(http.*?.jpg)", 1],
- "availability": ["(Only \\d+ left )?(.*)", 2]
- }
- }
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement