From 924a41ddcaa09e88d01caf2b018208bbd4da9745 Mon Sep 17 00:00:00 2001
From: Arya Kiran <arya@projectsegfau.lt>
Date: Mon, 13 Feb 2023 19:29:44 +0530
Subject: [PATCH] add support for organizations

---
 pages/user.go      | 34 +++++++++++++++++++++++++++-------
 utils/ugcpolicy.go |  5 +++--
 views/user.html    |  7 +++++--
 3 files changed, 35 insertions(+), 11 deletions(-)

diff --git a/pages/user.go b/pages/user.go
index a59ef7a..86a6b18 100644
--- a/pages/user.go
+++ b/pages/user.go
@@ -23,12 +23,13 @@ type User struct {
 	StatusEmoji   string
 	AvatarUrl     string
 	Location      string
+	Email         string
 	Timezone      string
 	Following     int64
 	Followers     int64
 	Link          string
 	Social        []string
-	Organization  []string
+	Organizations []string
 	Company       string
 	Type          string
 	Contributions string
@@ -115,7 +116,6 @@ func HandleUser(c *fiber.Ctx) error {
 
 	// scraping
 	Scrape := User{
-		Link:   link,
 		Readme: string(readmeOutput),
 	}
 
@@ -131,7 +131,6 @@ func HandleUser(c *fiber.Ctx) error {
 
 	sc1.Visit("https://github.com/" + c.Params("user") + "/")
 	sc := colly.NewCollector(colly.AllowedDomains("github.com"), colly.UserAgent(UserAgent))
-
 	if Scrape.Type == "http://schema.org/Person" {
 		// Bio
 		sc.OnHTML("div[data-bio-text]", func(e *colly.HTMLElement) {
@@ -149,7 +148,7 @@ func HandleUser(c *fiber.Ctx) error {
 			e.ForEach("li[itemprop*='social']", func(i int, el *colly.HTMLElement) {
 				Scrape.Social = append(Scrape.Social, el.ChildText("a.Link--primary"))
 			})
-
+			Scrape.Link = e.ChildText("a[itemprop*='url']")
 		})
 		// Followers/Following
 		sc.OnHTML("a[href*='https://github.com/"+c.Params("user")+"?tab=followers' i]", func(e *colly.HTMLElement) {
@@ -178,21 +177,42 @@ func HandleUser(c *fiber.Ctx) error {
 		// Organizations
 		sc.OnHTML("div.mt-3", func(e *colly.HTMLElement) {
 			e.ForEach("a[data-hovercard-type*='organization']", func(i int, el *colly.HTMLElement) {
-				Scrape.Organization = append(Scrape.Organization, el.Attr("aria-label"))
+				Scrape.Organizations = append(Scrape.Organizations, el.Attr("aria-label"))
 			})
 		})
 
 	} else {
+		sc.OnHTML("div.container-xl div.flex-md-items-center div.flex-1", func(e *colly.HTMLElement) {
+			Scrape.Bio = e.ChildText("div.color-fg-muted div")
+			Scrape.Followers, err = strconv.ParseInt(e.ChildText("a[href*='/orgs/"+c.Params("user")+"/followers' i] span"), 10, 64)
+			Scrape.Name = e.ChildText("h1.h2")
+			Scrape.Location = e.ChildText("span[itemprop*='location']")
+			Scrape.Link = e.ChildText("a[itemprop*='url']")
+			Scrape.Email = e.ChildText("a[itemprop*='email']")
+			e.ForEach("a.Link--primary", func(i int, el *colly.HTMLElement) {
+				Scrape.Social = append(Scrape.Social, el.Attr("href"))
+			})
+		})
 		sc.OnHTML("img[alt*='@"+c.Params("user")+"' i]", func(e *colly.HTMLElement) {
 			Scrape.AvatarUrl = e.Attr("src")
+			Scrape.Login = e.Attr("alt")
 		})
-		log.Println("Bio and Location routes cannot be scraped for organizations")
 	}
 	sc.Visit("https://github.com/" + c.Params("user") + "/")
 	// Fixing the output a bit
 	Scrape.AvatarUrl = strings.TrimPrefix(Scrape.AvatarUrl, "https://avatars.githubusercontent.com/u/")
 	Scrape.AvatarUrl = "/avatar/" + Scrape.AvatarUrl
-	Scrape.StatusEmoji = emoji.Parse(":"+Scrape.StatusEmoji+":")
+	if Scrape.StatusEmoji != "" {
+		Scrape.StatusEmoji = emoji.Parse(":" + Scrape.StatusEmoji + ":")
+	}
+	Scrape.Login = strings.TrimPrefix(Scrape.Login, "@") // Only for orgs
+	if strings.HasPrefix(Scrape.Link, "https://") {
+		Scrape.Link = strings.TrimPrefix(Scrape.Link, "https://")
+	} else if strings.HasPrefix(Scrape.Link, "http://") {
+		Scrape.Link = strings.TrimPrefix(Scrape.Link, "http://")
+	} else {
+		log.Println("Has no prefix")
+	}
 	// Add scrape-based info to userArray
 	userArray = append(userArray, Scrape)
 
diff --git a/utils/ugcpolicy.go b/utils/ugcpolicy.go
index 665aa41..ffcb4a8 100644
--- a/utils/ugcpolicy.go
+++ b/utils/ugcpolicy.go
@@ -1,8 +1,10 @@
 package utils
+
 import (
-	"regexp"
 	"github.com/microcosm-cc/bluemonday"
+	"regexp"
 )
+
 // copied from bluemonday's GitHub repostiory, with some adaptations
 func UGCPolicy() *bluemonday.Policy {
 
@@ -194,4 +196,3 @@ func UGCPolicy() *bluemonday.Policy {
 
 	return p
 }
-
diff --git a/views/user.html b/views/user.html
index 8a5eabd..6d052f2 100644
--- a/views/user.html
+++ b/views/user.html
@@ -29,14 +29,17 @@
                 {{ if .Link }}
                 <p>🌐 <a href="https://{{.Link}}" target="_blank">{{.Link}}</a></p>
                 {{ end }}
+				{{ if .Email }}
+				<p>✉️  {{.Email}}</p>
+				{{ end }}
 				{{ if .Social }}
 				{{range .Social}}
 				<p>🔗 <a href="{{.}}" target="_blank">{{.}}</a></p>
 				{{ end }}
 				{{ end }}
-				{{ if .Organization }}
+				{{ if .Organizations }}
 				<p>Organizations: 
-				{{range .Organization}}
+				{{range .Organizations}}
 				<a href="/{{.}}" target="_blank">{{.}}</a> 
 				{{ end }}
 				</p>