Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- {
- "start_urls": [
- "https://www.elcorteingles.es/altus/deportes/",
- "https://www.elcorteingles.es/trangoworld/deportes/",
- "https://www.elcorteingles.es/izas/deportes/"
- ],
- "fields": {
- "name": "//li[@class='products_list-item']/span/@data-json",
- "price": "//li[@class='products_list-item']/span/@data-json",
- "image": "//img[contains(@class,'js_preview_image')]/@src",
- "brand_text": "//li[@class='products_list-item']/span/@data-json",
- "categories": "//li[@class='products_list-item']/span/@data-json",
- "url": "//h2[contains(@class,'plp_title')]/a/@href"
- },
- "links": {
- "item_detail": {
- "detail": "",
- "next_page": {
- "restrict_xpaths": "//li[@id='pagination-next']/a",
- "callback": "parse_item"
- },
- "parse_start_url": true
- },
- "positioning_spider": {
- "detail": "//h2/a[contains(@class,'product_preview-title')]",
- "next_page": "//li[@id='pagination-next']/a"
- }
- },
- "regex": {
- "price": ["f_price\\\":(.*?),", 1],
- "name": ["name\\\":\\\"(.*?)\\\"", 1],
- "brand_text": ["brand\\\":\\\"(.*?)\\\"", 1],
- "categories": ["hierarchy\\\":.*\\\"(.*?)\\\"\\]", 1]
- },
- "list": {
- "product": "//li[@class='products_list-item']",
- "page": "substring-before(substring-after(//li[@id='pagination-current']/text(), 'Página '), ' de')",
- "default_page": 1
- },
- "parser_modules": [
- "multiple_response_xpath",
- "multiple_item_xpath"
- ]
- }
Add Comment
Please, Sign In to add comment