Merge pull request #2561 from nrg101/add-gallery-by-url-to-arx

add galleryByURL to Arx

Author: feederbox826

scrapers/Arx/Arx.yml

@@ -16,33 +16,37 @@ sceneByURL:
       - transmidnight.com
       - transroommates.com
     scraper: htmlScraper
+galleryByURL:
+  - action: scrapeXPath
+    url: *urls
+    scraper: htmlScraper
 performerByURL:
   - action: scrapeXPath
     url: *urls
     scraper: htmlScraper
 xPathScrapers:
   htmlScraper:
     scene:
-      Title: //meta[@property="og:title"]/@content
-      Date:
+      Title: &titleSel //meta[@property="og:title"]/@content
+      Date: &date
         selector: //h1/following-sibling::div[contains(@class, "items-center")]//span[@class="block"]/text()
         postProcess:
           - parseDate: Jan 2, 2006
-      Details: //span[text()="Description:"]/following-sibling::span
+      Details: &detailsSel //span[text()="Description:"]/following-sibling::span
       Performers:
-        Name: //span[text()="Models:"]/following-sibling::div//a//span/text()
+        Name: &performersSel //span[text()="Models:"]/following-sibling::div//a//span/text()
       Tags:
         Name: &tagsSel //a[contains(@href, "/categories/")]//text()
       URL: &urlSel //meta[@property="og:url"]/@content
       Image: &imageSel //meta[@property="og:image"]/@content
-      Code:
+      Code: &code
         selector: *urlSel
         postProcess:
           - replace:
               - regex: &domainAndCodeRegex 'https?://([^/]+)/scenes/(\d+)/.*'
                 with: $2
       Studio:
-        Name:
+        Name: &studioName
           selector: *imageSel
           postProcess:
             - replace:
@@ -51,7 +55,10 @@ xPathScrapers:
             - map:
                 hhdp: Anal Vault
                 chi: Cuck Hunter
+                hhsf: Honey Trans
                 hhsm: Honey Trans
+                hhsss: Honey Trans
+                jlo: Japan Lust
                 jly: Japan Lust
                 joi: JOI Babes
                 hhles: Les Worship
@@ -66,6 +73,51 @@ xPathScrapers:
             - replace:
                 - regex: *domainAndCodeRegex
                   with: https://www.$1
+    gallery:
+      Title:
+        selector: *titleSel
+        postProcess:
+          - replace:
+              - regex: "Photos for Scene \\| "
+                with: ""
+      Code: *code
+      URL: *urlSel
+      Date:
+        selector: *urlSel
+        postProcess:
+          - replace: &removePhotos
+              - regex: /photos$
+                with: ""
+          - subScraper: *date
+      Studio:
+        Name:
+          selector: *urlSel
+          postProcess:
+            - replace: *removePhotos
+            - subScraper: *studioName
+      Performers:
+        Name:
+          selector: *urlSel
+          postProcess:
+            - replace: *removePhotos
+            - subScraper:
+                selector: *performersSel
+                concat: ","
+          split: ","
+      Tags:
+        Name:
+          selector: *urlSel
+          postProcess:
+            - replace: *removePhotos
+            - subScraper:
+                selector: *tagsSel
+                concat: ","
+          split: ","
+      Details:
+        selector: *urlSel
+        postProcess:
+          - replace: *removePhotos
+          - subScraper: *detailsSel
     performer:
       Name: //meta[@property="og:image:alt"]/@content
       Image:
@@ -76,4 +128,4 @@ xPathScrapers:
                 with: "${1}0x0${2}"
       Tags:
         Name: *tagsSel
-# Last Updated July 28, 2025
+# Last Updated November 7, 2025