scraper: SinsVR cast fix (#1823)

* SinsVR Fix for model not present on scrape

Occasionally sinsVR will leave out the model on the scene page resulting in no model being listed for the scene. This falls back to using the trailer URL which contains the models in the scenes

* Debug prompt removal

* Some more tweaks. 

Check to ensure the profile url is valid. Notice occasionally there is a misspelling of the actor name in the url

* Format

pkg/scrape/sinsvr.go

@@ -2,6 +2,7 @@ package scrape
 
 import (
 	"encoding/json"
+	"net/http"
 	"regexp"
 	"strconv"
 	"strings"
@@ -66,15 +67,18 @@ func SinsVR(wg *sync.WaitGroup, updateSite bool, knownScenes []string, out chan<
 		sc.ActorDetails = make(map[string]models.ActorDetails)
 		e.ForEach(`.video-detail__specs div.cell`, func(id int, e *colly.HTMLElement) {
 			c := strings.TrimSpace(e.Text)
+
 			// Cast
 			if strings.Contains(c, "Starring") {
 				e.ForEach(`.cell a`, func(id int, e *colly.HTMLElement) {
+
 					cast := strings.Split(e.Text, ",")
 					sc.Cast = append(sc.Cast, cast...)
 					if len(cast) > 1 {
 						sc.ActorDetails[strings.TrimSpace(e.Text)] = models.ActorDetails{Source: sc.ScraperID + " scrape", ProfileUrl: e.Request.AbsoluteURL(e.Attr("href"))}
 					}
 					sc.ActorDetails[strings.TrimSpace(e.Text)] = models.ActorDetails{Source: sc.ScraperID + " scrape", ProfileUrl: e.Request.AbsoluteURL(e.Attr("href"))}
+
 				})
 			} else {
 				// Released - Date Oct 19, 2019
@@ -85,6 +89,30 @@ func SinsVR(wg *sync.WaitGroup, updateSite bool, knownScenes []string, out chan<
 			}
 		})
 
+		// Fallback incase SinsVR forgets to add the actor to the scene. This uses the trailer URL which contains the models in the scene.
+		if len(sc.Cast) == 0 {
+			trailerURL := e.ChildAttr("div.video-player-container__player source", "src")
+			//The cast is the first part of the trailer file name in the URL
+			re := regexp.MustCompile(`https:\/\/public\.xsinsvr\.com\/video\/.+\/(?P<cast>[A-Za-z_-]+)_trailer`)
+			r := re.FindStringSubmatch(trailerURL)
+			castIndex := re.SubexpIndex("cast")
+			//sinsVR uses _ for whitespace and - to separate cast members
+			cast := strings.Split(strings.ReplaceAll(r[castIndex], "_", " "), "-")
+			sc.Cast = append(sc.Cast, cast...)
+
+			// This should result in the correct model url for sinsVR but occasionally sins has yet to create the model url and will result in a 404
+			for _, name := range cast {
+				profileUrl := `https://xsinsvr.com/model/` + strings.ToLower(strings.ReplaceAll(name, " ", "-"))
+				profileUrlResp, err := http.Head(profileUrl)
+				if err != nil {
+					log.Errorf("Method Head Failed on profileUrlResp %s with error %s", profileUrlResp, err)
+				} else if profileUrlResp.StatusCode == 200 { //The url is not valid don't bother adding it to the ActorDetails
+					sc.ActorDetails[name] = models.ActorDetails{Source: sc.ScraperID + " scrape", ProfileUrl: profileUrl}
+				}
+				defer profileUrlResp.Body.Close()
+			}
+		}
+
 		// Duration
 		durationText := e.ChildText(`div.video-player-container__info div.tn-video-props span`)
 		for _, regex := range durationRegexes {