From c51b36c125d2f9444a11f30d90763abce1a35bf2 Mon Sep 17 00:00:00 2001
From: donpat1to <patrick@inca-vikingo.de>
Date: Thu, 18 Dec 2025 14:01:51 +0100
Subject: [PATCH] added session detection with requests per task

---
 src/scraper/webdriver.rs | 103 +++++++++++++++++++++++++++++++++++++--
 1 file changed, 100 insertions(+), 3 deletions(-)

diff --git a/src/scraper/webdriver.rs b/src/scraper/webdriver.rs
index 0c1093e..5a744b4 100644
--- a/src/scraper/webdriver.rs
+++ b/src/scraper/webdriver.rs
@@ -276,6 +276,11 @@ pub struct ChromeInstance {
     task_count: usize,
     max_tasks_per_instance: usize,
     proxy_url: Option<String>,
+    
+    // NEU: Session-Management
+    current_session: Arc<Mutex<Option<Client>>>,
+    session_request_count: Arc<Mutex<usize>>,
+    max_requests_per_session: usize,  // z.B. 25
 }
 
 impl ChromeInstance {
@@ -289,17 +294,110 @@ impl ChromeInstance {
             task_count: 0,
             max_tasks_per_instance,
             proxy_url,
+            // NEU
+            current_session: Arc::new(Mutex::new(None)),
+            session_request_count: Arc::new(Mutex::new(0)),
+            max_requests_per_session: 25,  // Konfigurierbar machen!
         })
     }
 
-    pub async fn new_session(&self) -> Result<Client> {
+    pub async fn get_or_renew_session(&self) -> Result<Client> {
+        let mut session_opt = self.current_session.lock().await;
+        let mut request_count = self.session_request_count.lock().await;
+        
+        // Session erneuern wenn:
+        // 1. Keine Session vorhanden
+        // 2. Request-Limit erreicht
+        let needs_renewal = session_opt.is_none() || *request_count >= self.max_requests_per_session;
+        
+        if needs_renewal {
+            // Alte Session schließen
+            if let Some(old_session) = session_opt.take() {
+                crate::util::logger::log_info("Closing old session").await;
+                let _ = old_session.close().await;
+                // Kurze Pause zwischen Sessions
+                sleep(Duration::from_millis(rand::rng().random_range(500..1000))).await;
+            }
+            
+            // Neue Session mit frischem User-Agent erstellen
+            crate::util::logger::log_info(&format!(
+                "Creating new session (requests in last session: {})", 
+                *request_count
+            )).await;
+            
+            let new_session = self.create_fresh_session().await?;
+            *session_opt = Some(new_session.clone());
+            *request_count = 0;
+            
+            Ok(new_session)
+        } else {
+            // Existierende Session verwenden
+            *request_count += 1;
+            Ok(session_opt.as_ref().unwrap().clone())
+        }
+    }
+
+    async fn create_fresh_session(&self) -> Result<Client> {
+        // WICHTIG: User-Agent hier wählen, nicht in chrome_args()!
+        let user_agent = Self::chrome_user_agent();
+        let capabilities = self.chrome_args_with_ua(user_agent);
+        
         ClientBuilder::native()
-            .capabilities(self.chrome_args())
+            .capabilities(capabilities)
             .connect(&self.base_url)
             .await
             .context("Failed to connect to ChromeDriver")
     }
 
+    fn chrome_args_with_ua(&self, user_agent: &str) -> Map<String, Value> {
+        let mut args = vec![
+            "--headless=new".to_string(),
+            "--disable-gpu".to_string(),
+            "--no-sandbox".to_string(),
+            "--disable-dev-shm-usage".to_string(),
+            "--disable-infobars".to_string(),
+            "--disable-extensions".to_string(),
+            "--disable-popup-blocking".to_string(),
+            "--disable-notifications".to_string(),
+            "--disable-autofill".to_string(),
+            "--disable-sync".to_string(),
+            "--disable-default-apps".to_string(),
+            "--disable-translate".to_string(),
+            "--disable-blink-features=AutomationControlled".to_string(),
+            // User-Agent als Parameter!
+            format!("--user-agent={}", user_agent),
+        ];
+        
+        if let Some(ref proxy) = self.proxy_url {
+            args.push(format!("--proxy-server={}", proxy));
+        }
+        
+        let caps = serde_json::json!({
+            "goog:chromeOptions": {
+                "args": args,
+                "excludeSwitches": ["enable-logging", "enable-automation"],
+                "prefs": {
+                    "profile.default_content_setting_values.notifications": 2
+                }
+            }
+        });
+        caps.as_object().cloned().unwrap()
+    }
+
+    pub async fn new_session(&self) -> Result<Client> {
+        // Für Backward-Compatibility, aber sollte get_or_renew_session() nutzen!
+        self.create_fresh_session().await
+    }
+
+    pub fn reset_task_count(&mut self) {
+        self.task_count = 0;
+    }
+
+    pub async fn get_session_stats(&self) -> (usize, usize) {
+        let request_count = *self.session_request_count.lock().await;
+        (self.task_count, request_count)
+    }
+
     pub fn increment_task_count(&mut self) {
         self.task_count +=  1;
     }
@@ -402,7 +500,6 @@ impl ChromeInstance {
         caps.as_object().cloned().unwrap()
     }
 
-
     pub fn chrome_user_agent() -> &'static str {
         static UAS: &[&str] = &[
             "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/124.0.6367.91 Safari/537.36",