Fix gallery scraper for Femjoy - thank you eloiselle

This commit is contained in:
Maista6969 2023-08-04 02:57:19 +02:00
parent 83e7d04fdb
commit 6064e06b74

View File

@ -1,26 +1,73 @@
name: Femjoy
galleryByURL:
- action: scrapeXPath
url:
- femjoy.com/post/
scraper: galleryScraper
sceneByQueryFragment:
action: scrapeXPath
queryURL: "{url}"
scraper: sceneScraper
sceneByName:
action: scrapeXPath
queryURL: "https://www.femjoy.com/videos?s={}"
scraper: sceneSearch
sceneByURL:
- action: scrapeXPath
url:
- femjoy.com
- femjoy.com/post/
scraper: sceneScraper
xPathScrapers:
sceneScraper:
common:
$performer: //h1[@class='post_title']/a[starts-with(@href,"/models")]
scene:
Title: //div[@class='post_details']/h1[@class='post_title']/span/text()
Studio:
Name:
Title: &titleSel //h1[@class='post_title']/span[last()]/text()
Studio: &studioAttr
Name:
fixed: Femjoy
Date:
selector: //div[@class='post_details']/h2[@class='post_title']
postProcess:
Date: &dateAttr
selector: //h2[@class='post_title']/text()[2]
postProcess: &datePP
- replace:
- regex: .+on\s(.+)
with: $1
- parseDate: Jan 2, 2006
Performers:
Name: //div[@class='post_details']/h1[@class='post_title']/a[contains(@href,"/models/")]
URL: //div[@class='post_details']/h1[@class='post_title']/a[contains(@href,"/models/")]/@href
#Image: Uses placeholder cover now
Details: //div[@class='post_details']/h2[@class='post_description']/p
# Last Updated October 13, 2022
- regex: '.*released on\s+'
with:
- parseDate: Jan 2, 2006
Director: //*[@class='post_title']/a[starts-with(@href,"/director")]
Performers: &performersAttr
Name: $performer
URL:
selector: $performer/@href
postProcess: &prependDomain
- replace:
- regex: ^
with: https://femjoy.com
Details: &details
selector: //*[@class='post_description']//text()
concat: "\n"
Image: //meta[@name='twitter:image']/@content
sceneSearch:
common:
$scene: //div[@class='post_video']
$preview: //div[@class='post_video']//a[@class='preview']
scene:
Title: $preview/@title
Date:
selector: $scene//span[@class='posted_on']/text()
postProcess: *datePP
URL:
selector: $preview/@href
postProcess: *prependDomain
Image:
selector: $preview/@data-media-poster
galleryScraper:
common:
$performer: //h1[@class='post_title']/a[starts-with(@href,"/models")]
gallery:
Title: *titleSel
Studio: *studioAttr
Date: *dateAttr
Performers: *performersAttr
Details: *details
# Last Updated August 03, 2023