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 +1,202 @@
name: "GammaEntertainment"
sceneByURL:
- action: scrapeXPath
url:
- alettaoceanempire.com/en/video/
- analacrobats.com/en/video/
- ashleyfires.com/
- bigfatcreampie.com/
- bskow.com/en/video/
- bushybushy.com/
- buttman.com/
- cumshotoasis.com/
- currycreampie.com/
- devilsfilmparodies.com/
- devilsgangbangs.com/
- devonlee.com/
- dylanryder.com/
- familycreep.com/
- fistingcentral.com/
- gapingangels.com/
- girlsandstuds.com/
- grannyghetto.com/
- jaysinxxx.com/
- jonnidarkkoxxx.com/
- lanesisters.com/
- lesbianfactor.com/
- lexingtonsteele.com/
- maledigital.com/en/video/
- maskurbate.com/en/video/
- motherfuckerxxx.com/
- myteenoasis.com/
- openlife.com/
- pantypops.com/
- povblowjobs.com/
- povthis.com/
- silverstonedvd.com/
- silviasaint.com/
- squirtalicious.com/
- squirtinglesbian.com/en/video/
- squirtingorgies.com/en/scene/
- strapattackers.com/
- sunnyleone.com/
- tittycreampies.com/
- transsexualangel.com/en/video/
- transsexualroadtrip.com/
- whiteghetto.com/
scraper: sceneScraper
performerByURL:
- action: scrapeXPath
url:
- activeduty.com/en/pornstar/
- assholefever.com/en/pornstar/
- evilangel.com/en/pornstar/
scraper: performerScraper
movieByURL:
- action: scrapeXPath
url:
- darkx.com/en/movie/
- hardx.com/en/movie/
- xempire.com/en/movie/
scraper: movieScraper
xPathScrapers:
sceneScraper:
common:
$info: //div[@id="sceneInfoPlayer"]/div
$actors: //div[@class="sceneCol sceneColActors"]
$tags: //div[@class="sceneCol sceneColCategories"]
$detailscript: //script[contains(text(), 'sceneDetails')]/text()
scene:
Title: //div[@id="breadcrumb"]/span[3]/text()|//div[@id="breadCrumb"]/span[3]/text()|//div[@id="BreadcrumbMenu"]/span[3]/text()|//h1[@class="sceneTitle"]/text()|//div[@id="playerWrapper"]//h1/text()|//div[@id="scenePlayer"]//h1/text()|//div[@id="scenePlayerScene"]//h1/text()|//h3[@class="sceneTitle"]/text()
Date:
selector: //li[@class="updatedDate"]/text()[2]|//div[@class="updatedDate"]/text()[2]|//p[@class="updatedDate"]/text()[2]|//div[@class="sceneCol updatedDate"]/span/text()|//p[@class="p-updated"]/text()
postProcess:
- replace:
- regex: (\d{4})-(\d{2})-(\d{2})
with: $2-$3-$1
- regex: (\d{2})\/(\d{2})\/(\d{4})
with: $1-$2-$3
- regex: ":"
with:
- parseDate: 01-02-2006
Details:
selector: $detailscript
postProcess:
- replace:
- regex: .+(?:sceneDescription":")(.+)(?:","sceneActors).+
with: $1
# Second regex in case no description is provided.
- regex: .+(?:"sceneDescription":"").+
with:
- regex: <\\\/br>|<br\s\\\/>|<br>|<br\\/>
with: "\n"
- regex: <strong>|<\\\/strong>
with:
Tags:
Name: $tags/a/text()|//div[@class="sceneCol categories"]/a/text()|//div[@class="sceneCol sceneColCategories"]//span[not(@class="categorySeparator")]/text()|//div[@class="sceneCategories"]/a/text()
Performers:
Name: $actors/a/text()|$actors//span[not(@class="actorSeparator")]/text()|//div[@class="sceneCol actors"]/a/text()|//span[@class="slide-title"]/text()|//a[@class="pornstarName"]/strong/text()|//p[@class="starringLinks"]/a/text()|//p[@class="p-actors"]/a/text()
Movies:
Name:
selector: $info/a/img/@title|//a[@class="sceneLink "]/@title|//a[@class="dvdLink "]/@title
URL:
selector: //link[@rel='canonical']/@href|$info/a/@href
concat: "|"
postProcess:
- replace:
- regex: \.com.*[|]
with: ".com"
Image:
selector: //script[contains(text(), 'picPreview')]/text()
postProcess:
- replace:
- regex: .+(?:picPreview":")([\w:]+)(?:[\\\/]+)([\w-\.]+)(?:[\\\/]+)(\w+)(?:[\\\/]+)(\d+)(?:[\\\/]+)([\d_]+)(?:[\\\/]+)(\w+)(?:[\\\/]+)(\d+)(?:[\\\/]+)(\d+)(?:[\\\/]+)([\w]+)(?:[\\\/]+)([\w.]+).+
with: $1//$2/$3/$4/$5/$6/$7/$8/$9/$10
# if using the transport subdomain, parameters need to be passed
# otherwise a cropped square image is returned by default
- regex: (https:\/\/transform.+)
with: $1?width=1920&height=1080&enlarge=true
Studio:
Name:
selector: //*[local-name()="div" or local-name()=p][contains(text(),"is a site owned and operated by Gamma")]|//meta[@name="twitter:domain"]/@content
postProcess:
- replace:
- regex: ([^\.]+)\.com(\s+)is(.+)
with: $1
- regex: \.com$
with: ""
- map:
activeduty: Active Duty
alettaoceanempire: Aletta Ocean Empire
allblackx: All Black X
analacrobats: Anal Acrobats
ashleyfires: Ashley Fires
bigfatcreampie: Big Fat Cream Pie
bskow: BsKow
bushybushy: Bushy Bushy
buttman: Buttman
cumshotoasis: Cumshot Oasis
currycreampie: Curry Creampie
darkx: DarkX
devilsfilmparodies: Devil's Film Parodies
devilsgangbangs: Devil's Gangbangs
devonlee: Devon Lee
dylanryder: Dylan Ryder
eroticax: EroticaX
evilangel: Evil Angel
familycreep: Family Creep
fistingcentral: Fisting Central
gapingangels: Gaping Angels
girlsandstuds: Girls And Studs
grannyghetto: Granny Ghetto
jaysinxxx: Jay Sin XXX
jonnidarkkoxxx: Jonni Darkko XXX
lanesisters: Lane Sisters
lesbianfactor: Lesbian Factor
lesbianx: Lesbian X
lexingtonsteele: Lexington Steele
maledigital: Male Digital
maskurbate: Maskurbate
motherfuckerxxx: Mother Fucker XXX
myteenoasis: My Teen Oasis
openlife: Open Life
outofthefamily: Out Of The Family
pantypops: Panty Pops
povblowjobs: POV Blowjobs
povthis: POV This
silverstonedvd: Silverstone DVD
silviasaint: Silvia Saint
squirtalicious: Squirtalicious
squirtinglesbian: Squirting Lesbian
squirtingorgies: Squirting Orgies
strapattackers: Strap Attackers
sunnyleone: Sunny Leone
tittycreampies: Titty Creampies
transsexualangel: Transsexual Angel
transsexualroadtrip: Transsexual Roadtrip
whiteghetto: White Ghetto
xempire: XEmpire
performerScraper:
performer:
Name: //h1[@class='actorName']
Image: //img[@class='actorPicture']/@src
movieScraper:
movie:
Name: //h3[@class='dvdTitle']//text()
Director: //ul[@class='directedBy']/li/a/text()
Duration: //li[@class='length']/text()
Date:
selector: //li[@class='updatedOn']/text()
postProcess:
- parseDate: 2006-01-02
Synopsis: //p[@class='descriptionText']/text()
Studio:
Name: //a[contains(@class, 'GA_Id_headerLogo')]/span[@class='linkMainCaption']/text()
FrontImage: //a[@class='frontCoverImg']/@href
BackImage: //a[@class='backCoverImg']/@href
# Last Updated February 24, 2024