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 +1,77 @@
name: TitanMen
sceneByURL:
- action: scrapeXPath
url:
- titanmen.com
scraper: sceneScraper
movieByURL:
- action: scrapeXPath
url:
- titanmen.com
scraper: movieScraper
xPathScrapers:
sceneScraper:
common:
$details: //div[contains(@class, "scene-page-detail")]
scene:
Title: //h1[@class="scene-header-title"]
Details:
selector: //div[@class="col span_6_of_12"][1]/p/text()
concat: "\n\n"
Director: $details//strong[starts-with(text(), "Director")]/following-sibling::a
Code: //div[@class="rating_box"]/@data-id
Date:
selector: $details//strong[starts-with(text(), "Released")]/following-sibling::text()
postProcess:
- parseDate: Jan 2, 2006
Studio:
Name:
selector: $details//strong[starts-with(text(), "Movie Title")]/following-sibling::a/@href
postProcess:
- replace:
- regex: ^
with: https://www.titanmen.com/
- subScraper: //div[contains(@class, "movie-page-detail")]//strong[starts-with(text(), "Studio")]/following-sibling::a
- map:
Rough: TitanMen Rough
Performers:
Name: $details//strong[starts-with(text(), "Starring")]/following-sibling::a/text()
URL:
selector: $details//strong[starts-with(text(), "Starring")]/following-sibling::a/@href
postProcess:
- replace:
- regex: ^
with: https://www.titanmen.com/
Movies:
Name: $details//strong[starts-with(text(), "Movie")]/following-sibling::a
URL:
selector: $details//strong[starts-with(text(), "Movie")]/following-sibling::a/@href
postProcess:
- replace:
- regex: ^
with: https://www.titanmen.com/
Tags:
Name: $details//strong[starts-with(text(), "Categories")]/following-sibling::a
Image:
selector: //script[contains(text(), concat("scene-image-", //div[@class="rating_box"]/@data-id))]/text()
postProcess:
- replace:
- regex: .*image:\s*"([^"]+).*
with: $1
movieScraper:
movie:
Name:
selector: //div[@class="col span_9_of_12"]/h1/text()
Synopsis:
selector: //div[@class="col span_9_of_12"]/p[1]/text()
concat: "\n\n"
Date:
selector: //div[@class="col span_9_of_12"]/p[3]/text()
postProcess:
- parseDate: Jan 2, 2006
Studio:
Name: //div[contains(@class, "movie-page-detail")]//strong[starts-with(text(), "Studio")]/following-sibling::a
Director: //div[contains(@class, "movie-page-detail")]//strong[starts-with(text(), "Director")]/following-sibling::a
FrontImage: //div[contains(@class, "box-cover")]//img/@src
# Last Updated January 07, 2024