From b1014d5647810894882d588c5a1658980bc181da Mon Sep 17 00:00:00 2001
From: Azareal <azareal@users.noreply.github.com>
Date: Sat, 7 Mar 2020 13:49:44 +1000
Subject: [PATCH] add archive.org user agent

---
 gen_router.go      | 61 ++++++++++++++++++++++++----------------------
 langs/english.json |  1 +
 router_gen/main.go |  3 +++
 3 files changed, 36 insertions(+), 29 deletions(-)

diff --git a/gen_router.go b/gen_router.go
index 0aae6bd4..3bd7f95e 100644
--- a/gen_router.go
+++ b/gen_router.go
@@ -596,19 +596,20 @@ var agentMapEnum = map[string]int{
 	"twitter": 22,
 	"facebook": 23,
 	"cloudflare": 24,
-	"uptimebot": 25,
-	"slackbot": 26,
-	"apple": 27,
-	"discourse": 28,
-	"alexa": 29,
-	"lynx": 30,
-	"blank": 31,
-	"malformed": 32,
-	"suspicious": 33,
-	"semrush": 34,
-	"dotbot": 35,
-	"aspiegel": 36,
-	"zgrab": 37,
+	"archive_org": 25,
+	"uptimebot": 26,
+	"slackbot": 27,
+	"apple": 28,
+	"discourse": 29,
+	"alexa": 30,
+	"lynx": 31,
+	"blank": 32,
+	"malformed": 33,
+	"suspicious": 34,
+	"semrush": 35,
+	"dotbot": 36,
+	"aspiegel": 37,
+	"zgrab": 38,
 }
 var reverseAgentMapEnum = map[int]string{ 
 	0: "unknown",
@@ -636,19 +637,20 @@ var reverseAgentMapEnum = map[int]string{
 	22: "twitter",
 	23: "facebook",
 	24: "cloudflare",
-	25: "uptimebot",
-	26: "slackbot",
-	27: "apple",
-	28: "discourse",
-	29: "alexa",
-	30: "lynx",
-	31: "blank",
-	32: "malformed",
-	33: "suspicious",
-	34: "semrush",
-	35: "dotbot",
-	36: "aspiegel",
-	37: "zgrab",
+	25: "archive_org",
+	26: "uptimebot",
+	27: "slackbot",
+	28: "apple",
+	29: "discourse",
+	30: "alexa",
+	31: "lynx",
+	32: "blank",
+	33: "malformed",
+	34: "suspicious",
+	35: "semrush",
+	36: "dotbot",
+	37: "aspiegel",
+	38: "zgrab",
 }
 var markToAgent = map[string]string{ 
 	"OPR": "opera",
@@ -672,6 +674,7 @@ var markToAgent = map[string]string{
 	"Exabot": "exabot",
 	"SeznamBot": "seznambot",
 	"CloudFlare": "cloudflare",
+	"archive.org_bot": "archive_org",
 	"Uptimebot": "uptimebot",
 	"Slackbot": "slackbot",
 	"Discordbot": "discord",
@@ -809,7 +812,7 @@ func (r *GenRouter) SuspiciousRequest(req *http.Request, prepend string) {
 		prepend += "\n"
 	}
 	r.DumpRequest(req,prepend+"Suspicious Request")
-	co.AgentViewCounter.Bump(33)
+	co.AgentViewCounter.Bump(34)
 }
 
 func isLocalHost(h string) bool {
@@ -824,7 +827,7 @@ func (r *GenRouter) ServeHTTP(w http.ResponseWriter, req *http.Request) {
 		w.WriteHeader(200) // 400
 		w.Write([]byte(""))
 		r.DumpRequest(req,"Malformed Request T"+strconv.Itoa(typ))
-		co.AgentViewCounter.Bump(32)
+		co.AgentViewCounter.Bump(33)
 	}
 	
 	// Split the Host and Port string
@@ -966,7 +969,7 @@ func (r *GenRouter) ServeHTTP(w http.ResponseWriter, req *http.Request) {
 	
 	ua := strings.TrimSpace(strings.Replace(strings.TrimPrefix(req.UserAgent(),"Mozilla/5.0 ")," Safari/537.36","",-1)) // Noise, no one's going to be running this and it would require some sort of agent ranking system to determine which identifier should be prioritised over another
 	if ua == "" {
-		co.AgentViewCounter.Bump(31)
+		co.AgentViewCounter.Bump(32)
 		if c.Dev.DebugMode {
 			var prepend string
 			for _, char := range req.UserAgent() {
diff --git a/langs/english.json b/langs/english.json
index 6abf5d96..03acb5f4 100644
--- a/langs/english.json
+++ b/langs/english.json
@@ -209,6 +209,7 @@
 		"discord":"Discord",
 		"twitter":"Twitterbot",
 		"cloudflare":"Cloudflare Alwayson",
+		"archive_org":"Archive.org",
 		"uptimebot":"Uptimebot",
 		"slackbot":"Slackbot",
 		"facebook":"FacebookBot",
diff --git a/router_gen/main.go b/router_gen/main.go
index 28327e61..7807cf09 100644
--- a/router_gen/main.go
+++ b/router_gen/main.go
@@ -255,6 +255,7 @@ func main() {
 		"twitter",
 		"facebook",
 		"cloudflare",
+		"archive_org",
 		"uptimebot",
 		"slackbot",
 		"apple",
@@ -298,6 +299,7 @@ func main() {
 		"Exabot",
 		"SeznamBot",
 		"CloudFlare",
+		"archive.org_bot",
 		"Uptimebot",
 		"Slackbot",
 		"Discordbot",
@@ -337,6 +339,7 @@ func main() {
 		"Exabot":              "exabot",
 		"SeznamBot":           "seznambot",
 		"CloudFlare":          "cloudflare", // Track alwayson specifically in case there are other bots?
+		"archive.org_bot":     "archive_org",
 		"Uptimebot":           "uptimebot",
 		"Slackbot":            "slackbot",
 		"Discordbot":          "discord",