|
| 1 | +package scrape |
| 2 | + |
| 3 | +import ( |
| 4 | +"encoding/json" |
| 5 | +"regexp" |
| 6 | +"strconv" |
| 7 | +"strings" |
| 8 | +"sync" |
| 9 | + |
| 10 | +"github.com/gocolly/colly/v2" |
| 11 | +"github.com/mozillazg/go-slugify" |
| 12 | +"github.com/nleeper/goment" |
| 13 | +"github.com/thoas/go-funk" |
| 14 | +"github.com/xbapps/xbvr/pkg/models" |
| 15 | +) |
| 16 | + |
| 17 | +funcTransVR(wg*sync.WaitGroup,updateSitebool,knownScenes []string,outchan<- models.ScrapedScene,singleSceneURLstring,singeScrapeAdditionalInfostring,limitScrapingbool)error { |
| 18 | +deferwg.Done() |
| 19 | +scraperID:="transvr" |
| 20 | +siteID:="TransVR" |
| 21 | +allowedDomains:= []string{"transvr.com","www.transvr.com"} |
| 22 | +logScrapeStart(scraperID,siteID) |
| 23 | + |
| 24 | +sceneCollector:=createCollector(allowedDomains...) |
| 25 | +siteCollector:=createCollector(allowedDomains...) |
| 26 | +vodCollector:=createCollector(allowedDomains...) |
| 27 | + |
| 28 | +sceneCollector.OnHTML(`html`,func(e*colly.HTMLElement) { |
| 29 | +sc:= models.ScrapedScene{} |
| 30 | +sc.ScraperID=scraperID |
| 31 | +sc.SceneType="VR" |
| 32 | +sc.Studio="TransVR" |
| 33 | +sc.Site=siteID |
| 34 | +sc.HomepageURL=strings.Split(e.Request.URL.String(),"?")[0] |
| 35 | + |
| 36 | +// Title |
| 37 | +sc.Title=strings.Replace(e.ChildText(`title`),"Trans VR: ","",-1) |
| 38 | + |
| 39 | +// Cast |
| 40 | +sc.ActorDetails=make(map[string]models.ActorDetails) |
| 41 | +e.ForEach(`div.trailer_toptitle_left a`,func(idint,e*colly.HTMLElement) { |
| 42 | +sc.Cast=append(sc.Cast,strings.TrimSpace(e.Text)) |
| 43 | +sc.ActorDetails[strings.TrimSpace(e.Text)]= models.ActorDetails{Source:scraperID+" scrape",ProfileUrl:e.Request.AbsoluteURL(e.Attr("href"))} |
| 44 | +}) |
| 45 | + |
| 46 | +// Cover URL |
| 47 | +coverURL:=e.Request.AbsoluteURL(e.ChildAttr("div.player-thumb img","src")) |
| 48 | +sc.Covers=append(sc.Covers,coverURL) |
| 49 | + |
| 50 | +// Scene ID - get from URL |
| 51 | +tmps:=strings.Split(coverURL,"/") |
| 52 | +tmp:=strings.Replace(tmps[len(tmps)-1],".jpg","",-1) |
| 53 | +sc.SiteID=tmp |
| 54 | +sc.SceneID=slugify.Slugify(sc.Site)+"-"+sc.SiteID |
| 55 | + |
| 56 | +// Synopsis |
| 57 | +sc.Synopsis=strings.TrimSpace(e.ChildText(`div.trailerblock p`)) |
| 58 | + |
| 59 | +// Date |
| 60 | +dateString:=strings.Replace(e.ChildText(`div.set_meta`),"Added ","",-1) |
| 61 | +tmpDate,_:=goment.New(dateString,"MMMM D, YYYY") |
| 62 | +sc.Released=tmpDate.Format("YYYY-MM-DD") |
| 63 | + |
| 64 | +// Duration |
| 65 | +r:=regexp.MustCompile(`(?:(\d{2}):)?(\d{2}):(\d{2})`) |
| 66 | +m:=r.FindStringSubmatch(e.ChildText(`div.set_meta`)) |
| 67 | +duration:=0 |
| 68 | +iflen(m)==4 { |
| 69 | +hours,_:=strconv.Atoi("0"+m[1]) |
| 70 | +minutes,_:=strconv.Atoi(m[2]) |
| 71 | +duration=hours*60+minutes |
| 72 | +} |
| 73 | +sc.Duration=duration |
| 74 | + |
| 75 | +sc.TrailerType="scrape_html" |
| 76 | +params:= models.TrailerScrape{SceneUrl:sc.HomepageURL,HtmlElement:"dl8-video source",ContentPath:"src",QualityPath:"quality",ContentBaseUrl:`https://www.transvr.com`} |
| 77 | +strParams,_:=json.Marshal(params) |
| 78 | +sc.TrailerSrc=string(strParams) |
| 79 | + |
| 80 | +// Pull data from vod page - not every scene has a vod link |
| 81 | +ctx:=colly.NewContext() |
| 82 | +ctx.Put("scene",&sc) |
| 83 | +vodURL:=e.ChildAttr("a.downBtnbuy","href") |
| 84 | +if!strings.Contains(vodURL,"/tour") { |
| 85 | +vodCollector.Request("GET",vodURL,nil,ctx,nil) |
| 86 | +} |
| 87 | + |
| 88 | +out<-sc |
| 89 | +}) |
| 90 | + |
| 91 | +vodCollector.OnHTML(`html`,func(e*colly.HTMLElement) { |
| 92 | +sc:=e.Request.Ctx.GetAny("scene").(*models.ScrapedScene) |
| 93 | + |
| 94 | +// Gallery |
| 95 | +sc.Gallery=e.ChildAttrs("div.gallery-group img","data-orig-file") |
| 96 | + |
| 97 | +// Tags |
| 98 | +e.ForEach(`span.meta-tag a`,func(idint,e*colly.HTMLElement) { |
| 99 | +sc.Tags=append(sc.Tags,strings.TrimSpace(e.Text)) |
| 100 | +}) |
| 101 | + |
| 102 | +// Not every page has tags so use categories as well |
| 103 | +e.ForEach(`span.meta-cat a`,func(idint,e*colly.HTMLElement) { |
| 104 | +sc.Tags=append(sc.Tags,strings.TrimSpace(e.Text)) |
| 105 | +}) |
| 106 | + |
| 107 | +}) |
| 108 | + |
| 109 | +siteCollector.OnHTML(`div.videohere a`,func(e*colly.HTMLElement) { |
| 110 | +sceneURL:=e.Request.AbsoluteURL(e.Attr("href")) |
| 111 | + |
| 112 | +if!funk.ContainsString(knownScenes,sceneURL) { |
| 113 | +sceneCollector.Visit(sceneURL) |
| 114 | +} |
| 115 | +}) |
| 116 | + |
| 117 | +siteCollector.OnHTML(`div.pagination li a:not(.active)`,func(e*colly.HTMLElement) { |
| 118 | +if!limitScraping { |
| 119 | +pageURL:=e.Request.AbsoluteURL(e.Attr("href")) |
| 120 | +siteCollector.Visit(pageURL) |
| 121 | +} |
| 122 | +}) |
| 123 | + |
| 124 | +ifsingleSceneURL!="" { |
| 125 | +sceneCollector.Visit(singleSceneURL) |
| 126 | +}else { |
| 127 | +siteCollector.Visit("https://www.transvr.com/tour/categories/movies/1/latest/") |
| 128 | +} |
| 129 | + |
| 130 | +ifupdateSite { |
| 131 | +updateSiteLastUpdate(scraperID) |
| 132 | +} |
| 133 | +logScrapeFinished(scraperID,siteID) |
| 134 | +returnnil |
| 135 | +} |
| 136 | + |
| 137 | +funcinit() { |
| 138 | +registerScraper("transvr","TransVR","https://www.transvr.com/tour/custom_assets/favicon/apple-touch-icon.png","transvr.com",TransVR) |
| 139 | +} |