name: MeanBitches performerByURL: - action: scrapeXPath url: - meanbitches.com/tour/models/ scraper: performerScraper sceneByURL: - action: scrapeXPath url: - meanbitches.com/tour/updates/ scraper: sceneScraper xPathScrapers: performerScraper: performer: Name: //div[contains(@class,'title_bar') and not(contains(text(),'Updates'))]/text() Image: selector: //tr/td/img[@class="thumbs"]/@src postProcess: - replace: - regex: ^/ with: http://meanbitches.com/ sceneScraper: common: $titlebar: //table[@id="episode_title"]/tbody/tr/ scene: Title: $titlebar/td[contains(comment(),"Title")]/text() Details: //div[@class="episode_description"]/text() Studio: Name: fixed: Mean Bitches URL: fixed: https://www.meanbitches.com/ Performers: Name: selector: $titlebar/td[contains(comment(),"Featured Models")]/text() postProcess: - replace: - regex: 'Featuring:\s*(.*)' with: $1 - regex: '\xa0' # non printable char after the comma with: "" split: "," Image: selector: //td[@class="episode_thumbs"][@align="center"]//img/@src postProcess: - replace: - regex: ^/ with: http://meanbitches.com/ Date: selector: $titlebar/td[contains(text(),"Added")]/text() postProcess: - replace: - regex: "Added(.*)" with: $1 - parseDate: 01/02/2006 # Last Updated December 26, 2022