name: ATK Girlfriends # requires: py_common sceneByFragment: action: script script: - python - ATKGirlfriends.py performerByURL: - action: scrapeXPath url: # Trying to scrape without the www. prefix will result in a redirect loop - www.atkgirlfriends.com/tour/model/ scraper: performerScraper sceneByURL: - action: scrapeXPath url: # Trying to scrape without the www. prefix will result in a redirect loop - www.atkgirlfriends.com/tour/movie/ scraper: sceneScraper xPathScrapers: performerScraper: common: $modelWrap: &modelWrap //div[contains(@class, "model-profile-wrap")] performer: Name: //h1[contains(@class, "page-title")] Gender: fixed: female Ethnicity: selector: $modelWrap/b[contains(text(), "Ethnicity")]/following-sibling::text() HairColor: selector: $modelWrap/b[contains(text(), "Hair Color")]/following-sibling::text() postProcess: - map: black: Black blond: Blonde brown: Brown red: Red white: White Height: selector: $modelWrap/b[contains(text(), "Height")]/following-sibling::text() postProcess: - feetToCm: true Weight: selector: $modelWrap/b[contains(text(), "Weight")]/following-sibling::text() postProcess: - replace: - regex: (\d+).* with: $1 - lbToKg: true Measurements: selector: $modelWrap/b[contains(text(), "Bust Size")]/following-sibling::text() Image: selector: $modelWrap/img/@src sceneScraper: common: $movieWrap: //div[contains(@class, "movie-wrap")] $modelWrap: *modelWrap scene: Title: //title Details: $movieWrap/b[contains(text(), "Description")]/following-sibling::text() Image: selector: //video/@poster | //div[@class="flowplayer minimalist is-splash"]/@style postProcess: - replace: - regex: background-image:url\('(.*)'\); with: $1 Studio: Name: fixed: ATK Girlfriends URL: fixed: https://www.atkgirlfriends.com/ Tags: Name: selector: $movieWrap/b[contains(text(), "Tags")]/following-sibling::text() postProcess: - replace: - regex: \s*,\s* with: "," split: "," Performers: Name: $modelWrap/text()[1] driver: cookies: - CookieURL: https://www.atkgirlfriends.com Cookies: - Name: start_session_galleria Domain: www.atkgirlfriends.com Value: stash # Rotate this value if the scraper is blocked. The first request with the new value should fail. Path: / # Last Updated March 14, 2024