From d815fa6d0e5316faabeb7c326f23706894e58a0b Mon Sep 17 00:00:00 2001
From: rajchales-monito <rajchales-monito@users.noreply.github.com>
Date: Thu, 30 Jul 2026 13:46:39 +0200
Subject: [PATCH] Improve visitor bot filtering

---
 server.js | 32 ++++++++++++++++++++++++++------
 1 file changed, 26 insertions(+), 6 deletions(-)

diff --git a/server.js b/server.js
index d23eb15..5761863 100644
--- a/server.js
+++ b/server.js
@@ -1835,7 +1835,16 @@ function currentVisitors() {
 function formatVisitorSession(row, thresholds = adminPresenceSettings()) {
   const history = parseJson(row.browsing_history_json, []);
   const device = parseJson(row.device_json, {});
-  const bot = botInfo(device.userAgent);
+  const pageCountValue = new Set(history.map((item) => item.url).filter(Boolean)).size;
+  const sessionSecondsValue = sessionSeconds(row.browsing_history_json, row.first_seen_at);
+  const conversationCountValue = row.visitor_conversation_count || 0;
+  const bot = botInfo(device.userAgent, {
+    pageCount: pageCountValue,
+    visitCount: Math.max(1, history.length),
+    conversationCount: conversationCountValue,
+    sessionSeconds: sessionSecondsValue,
+    referrer: row.referrer
+  });
   const countryCode = row.country_code || countryFromTimezone(row.timezone) || countryFromLanguage(row.language);
   const presence = visitorPresence(row, thresholds);
   return {
@@ -1849,9 +1858,9 @@ function formatVisitorSession(row, thresholds = adminPresenceSettings()) {
     currentUrl: row.current_url,
     referrer: row.referrer,
     browsingHistory: history,
-    pageCount: new Set(history.map((item) => item.url).filter(Boolean)).size,
+    pageCount: pageCountValue,
     visitCount: Math.max(1, history.length),
-    conversationCount: row.visitor_conversation_count || 0,
+    conversationCount: conversationCountValue,
     lastConversationId: row.last_conversation_id,
     isBot: bot.isBot,
     botReason: bot.reason,
@@ -1860,7 +1869,7 @@ function formatVisitorSession(row, thresholds = adminPresenceSettings()) {
     timezone: row.timezone,
     countryCode,
     countryFlag: countryCode ? flagEmoji(countryCode) : "",
-    sessionSeconds: sessionSeconds(row.browsing_history_json, row.first_seen_at),
+    sessionSeconds: sessionSecondsValue,
     firstSeenAt: row.first_seen_at,
     lastActivityAt: row.last_activity_at,
     lastSeenAt: row.last_seen_at
@@ -2245,19 +2254,30 @@ function countryFromLanguage(language) {
   return match ? match[1].toUpperCase() : null;
 }
 
-function botInfo(userAgent) {
+function botInfo(userAgent, activity = {}) {
   const ua = String(userAgent || "");
   const patterns = [
     "googlebot", "bingbot", "slurp", "duckduckbot", "baiduspider", "yandexbot",
     "ahrefsbot", "semrushbot", "mj12bot", "dotbot", "petalbot", "bytespider",
     "facebookexternalhit", "twitterbot", "linkedinbot", "slackbot", "discordbot",
     "whatsapp", "telegrambot", "preview", "crawler", "spider", "headlesschrome",
-    "pingdom", "uptimerobot", "curl", "wget", "python-requests", "axios", "httpclient"
+    "phantomjs", "playwright", "puppeteer", "selenium", "cypress", "chrome-lighthouse",
+    "pagespeed", "lighthouse", "pingdom", "uptimerobot", "curl", "wget",
+    "python-requests", "aiohttp", "go-http-client", "axios", "httpclient", "java/",
+    "okhttp", "libwww-perl", "scrapy", "httpx", "node-fetch", "undici"
   ];
   const lower = ua.toLowerCase();
   const match = patterns.find((pattern) => lower.includes(pattern));
   if (match) return { isBot: true, reason: match };
   if (/\bbot\b/i.test(ua)) return { isBot: true, reason: "bot" };
+  if (!ua.trim()) return { isBot: true, reason: "empty ua" };
+  if (activity.conversationCount > 0) return { isBot: false, reason: "" };
+  const pageCount = Number(activity.pageCount || 0);
+  const visitCount = Number(activity.visitCount || 0);
+  const sessionSecondsValue = Number(activity.sessionSeconds || 0);
+  if (pageCount >= 25 && sessionSecondsValue <= 180) return { isBot: true, reason: "fast crawler" };
+  if (pageCount >= 40) return { isBot: true, reason: "many pages" };
+  if (visitCount >= 60 && !activity.referrer) return { isBot: true, reason: "no-ref crawler" };
   return { isBot: false, reason: "" };
 }