89 lines
2.8 KiB
YAML
89 lines
2.8 KiB
YAML
name: DorcelVision
|
|
sceneByURL:
|
|
- action: scrapeXPath
|
|
url:
|
|
- https://www.dorcelvision.com/en/movies
|
|
scraper: sceneScraper
|
|
|
|
movieByURL:
|
|
- action: scrapeXPath
|
|
url:
|
|
- https://www.dorcelvision.com/en/movies
|
|
scraper: movieScraper
|
|
sceneByName:
|
|
action: scrapeXPath
|
|
queryURL: https://www.dorcelvision.com/en/search?type=4&keyword={}
|
|
scraper: sceneSearch
|
|
sceneByQueryFragment:
|
|
action: scrapeXPath
|
|
queryURL: "{url}"
|
|
scraper: sceneScraper
|
|
|
|
|
|
xPathScrapers:
|
|
sceneSearch:
|
|
common:
|
|
$card: //div[@id="entriesList"]/div
|
|
scene:
|
|
Title: $card/details/div/h3
|
|
URL:
|
|
selector: $card/a[@class="movies rolloverv2"]/@href
|
|
postProcess:
|
|
- replace:
|
|
- regex: (.*)
|
|
with: https://www.dorcelvision.com$1
|
|
|
|
Image:
|
|
selector: $card/a[@class="movies rolloverv2"]/img[last()]/@data-src
|
|
postProcess:
|
|
- replace:
|
|
- regex: (.*)
|
|
with: https://www.dorcelvision.com$1
|
|
Details: $card/details//p[last()]
|
|
|
|
|
|
sceneScraper:
|
|
scene:
|
|
Title: &titleSel //meta[@property="og:title"]/@content
|
|
Details: &detailsAttr
|
|
selector: //div[@id="biography"]/*/text()
|
|
concat: "\n\n"
|
|
Image: &imageAttr
|
|
selector: //a[contains(@class, "cover")]/@href
|
|
postProcess:
|
|
- replace:
|
|
- regex: (.*)
|
|
with: https://dorcelvision.com$1
|
|
Date: &dateAttr
|
|
selector: //div[@class="row informations"]/div/div/strong[contains(., "Production year")]/parent::div/following-sibling::div
|
|
postProcess:
|
|
- replace:
|
|
- regex: ^(\d+)$
|
|
with: $1-01-01
|
|
- parseDate: 2006-01-02
|
|
Performers:
|
|
Name: //div[contains(@class, "casting")]//div[contains(@class, "slider-xl")]//div[@class="col-xs-2"]//a/span/text()
|
|
Studio: &studioAttr
|
|
Name: //div[@class="row informations"]/div/div/strong[contains(., "Studio")]/parent::div/following-sibling::div
|
|
Movies:
|
|
Name: //meta[@property="og:title"]/@content
|
|
URL: //meta[@property="og:url"]/@content
|
|
URL: //meta[@property="og:url"]/@content
|
|
movieScraper:
|
|
movie:
|
|
Name: *titleSel
|
|
Date: *dateAttr
|
|
Duration:
|
|
selector: //div[@class="informations"]//span[@class="duration"]/text()
|
|
postProcess:
|
|
- replace:
|
|
- regex: \smin
|
|
with: ":00"
|
|
Studio: *studioAttr
|
|
Director:
|
|
selector: //div[@class="informations"]//strong[contains(., "Directed by")]/following-sibling::a/text()
|
|
FrontImage: *imageAttr
|
|
Synopsis: *detailsAttr
|
|
# Reference: https://github.com/PAhelper/PhoenixAdult.bundle/blob/master/Contents/Code/siteDorcelVision.py
|
|
# Last Updated December 13, 2023
|