61 lines
1.7 KiB
YAML
61 lines
1.7 KiB
YAML
# yaml-language-server: $schema=../validator/scraper.schema.json
|
|
|
|
name: "AnalVids"
|
|
sceneByURL:
|
|
- action: scrapeXPath
|
|
url:
|
|
- analvids.com/watch/
|
|
- pissvids.com/watch/
|
|
scraper: sceneScraper
|
|
performerByURL:
|
|
- action: scrapeXPath
|
|
url:
|
|
- analvids.com/model/
|
|
- pissvids.com/model/
|
|
scraper: performerScraper
|
|
sceneByFragment:
|
|
action: script
|
|
script:
|
|
- python
|
|
# use python3 instead if needed
|
|
- AnalVids.py
|
|
- query
|
|
|
|
xPathScrapers:
|
|
sceneScraper:
|
|
common:
|
|
$title: //h1[contains(@class, "watch__title")]//text()[not(ancestor::span)]
|
|
scene:
|
|
Title:
|
|
selector: $title
|
|
concat: " "
|
|
Date:
|
|
selector: //i[contains(@class, "bi-calendar3")]/text()
|
|
postProcess:
|
|
- parseDate: 2006-01-02
|
|
Details:
|
|
selector: //div[contains(@class, "text-mob-more")]//text()[not(parent::span[contains(@class, "dots")])]
|
|
concat: " "
|
|
Code:
|
|
selector: $title
|
|
postProcess:
|
|
- replace:
|
|
- regex: .+?([A-Z]{2,3}\d+)$|(.+)
|
|
with: $1
|
|
Performers:
|
|
Name: //h1[contains(@class, "watch__title")]//a/text()
|
|
URL: //h1[contains(@class, "watch__title")]//a/@href
|
|
Studio:
|
|
Name: //div[contains(@class, "genres-list")]//a[contains(@href, "/studios/")]/text()
|
|
URL: //div[contains(@class, "genres-list")]//a[contains(@href, "/studios/")]/@href
|
|
Tags:
|
|
Name: //div[contains(@class, "genres-list")]//a[contains(@href, "/genre/")]/text()
|
|
Image: //video/@data-poster
|
|
|
|
performerScraper:
|
|
performer:
|
|
Name: //h1
|
|
Country: //a[contains(@href, "nationality")]
|
|
Image: //div[contains(@class, 'model__left')]//img/@src
|
|
# Last Updated August 16, 2023
|