This commit is contained in:
Christoph Califice
2025-10-09 20:05:31 -03:00
parent ed22ef22bc
commit 0a5f88d75a
1442 changed files with 101562 additions and 0 deletions

View File

@@ -0,0 +1,158 @@
name: "JulesJordan"
sceneByURL:
- action: scrapeXPath
url:
- girlgirl.com/
- manuelferrara.com/
- spermswallowers.com/
scraper: sceneScraper
- action: scrapeXPath
url:
- julesjordan.com/
- theassfactory.com/
scraper: newStyleScraper
galleryByURL:
- action: scrapeXPath
url:
- girlgirl.com/
- manuelferrara.com/
- spermswallowers.com/
scraper: sceneScraper
- action: scrapeXPath
url:
- julesjordan.com/
- theassfactory.com/
scraper: newStyleScraper
performerByURL:
- action: scrapeXPath
url:
- julesjordan.com/trial/models/
scraper: performerScraper
movieByURL:
- action: scrapeXPath
url:
- julesjordan.com/trial/dvds/
scraper: movieScraper
xPathScrapers:
sceneScraper:
common:
$details: //div[@class="backgroundcolor_info"]
$galinfo: //div[@class="gallery_info"]
scene:
Title: &titleAttr
selector: //div[@class="title_bar"]/span/text()
Date: &dateAttr
selector: $details//div[@class="cell update_date"]/text()
postProcess:
- parseDate: 01/02/2006
Details: &detailsAttr
selector: $details//span[@class="update_description"]/text()
concat: " "
Tags: &tagsAttr
Name: //span[@class="update_tags"]/a/text()
Performers: &performersAttr
Name: $details//span[@class="update_models"]/a/text()|$galinfo//span[@class="update_models"]/div/a/text()
URL: $details//span[@class="update_models"]/a/@href|$galinfo//span[@class="update_models"]/div/a/@href
Studio: &studioAttr
Name:
selector: //base/@href
postProcess:
- replace:
- regex: (.+\/\/|www.|\..+)
with:
- map:
julesjordan: Jules Jordan
girlgirl: Girl Girl
manuelferrara: Manuel Ferrara
theassfactory: The Ass Factory
spermswallowers: Sperm Swallowers
Movies:
Name: $details//span[@class="update_dvds"]/a/text()
URL: $details//span[@class="update_dvds"]/a/@href
Image:
selector: //script[contains(text(),'useimage')]/text()
postProcess:
- replace:
- regex: (?:.+useimage = "([^"]+).+)
with: $1
- regex: (^\/)
with: https://www.julesjordan.com/
gallery:
Title: *titleAttr
Date: *dateAttr
Details: *detailsAttr
Tags: *tagsAttr
Performers: *performersAttr
Studio: *studioAttr
newStyleScraper:
common:
$scene: //div[@class="movie-player"]
scene:
Title: &newTitleAttr $scene//div[@class="movie_title"]
Date: &newDateAttr
selector: $scene//span[contains(., "Date:")]/following-sibling::text()
postProcess:
- parseDate: 01/02/2006
Details: &newDetailsAttr $scene//span[contains(., "Description:")]/following-sibling::text()
Tags: &newTagsAttr
Name: $scene//a[contains(@href, "categories")]/text()
Performers: &newPerformersAttr
Name: $scene//a[contains(@href, "models")]/text()
URL: $scene//a[contains(@href, "models")]/@href
Studio: *studioAttr
Movies:
Name: $scene//span[contains(., "Movie:")]/following-sibling::a/text()
URL: $scene//span[contains(., "Movie:")]/following-sibling::a/@href
Image: $scene//video/@poster
gallery:
Title: *newTitleAttr
Date: *newDateAttr
Details: *newDetailsAttr
Tags: *newTagsAttr
Performers: *newPerformersAttr
Studio: *studioAttr
performerScraper:
performer:
Name: //span[@class='title_bar_hilite']/text()
Birthdate:
selector: //div[@class='model_bio model_bio_info']//text()[contains(.,'Age')]
postProcess:
- replace:
- regex: "Age:"
with:
- parseDate: January 2, 2006
Measurements:
selector: //div[@class='model_bio model_bio_info']//text()[contains(.,'Measurements')]
postProcess:
- replace:
- regex: "Measurements:"
with:
- regex: " "
with:
Height:
selector: //div[@class='model_bio model_bio_info']//text()[contains(.,'Height')]
postProcess:
- replace:
- regex: "Height:"
with:
- feetToCm: true
Image:
selector: //img[contains(@class,'model_bio_thumb')]/@src0_3x
movieScraper:
movie:
Name: //meta[@property="og:title"]/@content
Studio:
Name:
fixed: Jules Jordan Video
FrontImage: //img[contains(@class, "dvd_box")]/@src0_3x
# Last Updated September 19, 2023