init: scripts diversos (crawlers, conversores, scrapers)

scraper/check_sites_availability.sh

@@ -0,0 +1,65 @@
+#!/bin/bash
+# Verificação disponibilidade sites - Batch 4
+# Author: Descomplicar® Crescimento Digital
+# Copyright: 2025 Descomplicar®
+
+echo "🔍 VERIFICAÇÃO DISPONIBILIDADE - BATCH 4 SITES"
+echo "=============================================="
+echo ""
+
+declare -a SITES=(
+    "https://portalclassicos.com/foruns/index.php"
+    "https://forums.pelicanparts.com/porsche-forums/"
+    "https://forums.pelicanparts.com/bmw-forums/"
+    "https://www.peachparts.com/shopforum/index.php"
+    "https://forums.pelicanparts.com/vw-audi-technical-forum/"
+    "https://forums.pelicanparts.com/saab-technical-forum/"
+    "https://forums.pelicanparts.com/mini-discussion-forum/"
+    "https://www.pelicanparts.com/techarticles/tech_center_main.htm"
+    "https://www.pelicanparts.com/techarticles/Mercedes-Benz/MBZ_Tech_Index.htm"
+    "https://www.pelicanparts.com/BMW/techarticles/tech_main.htm"
+    "https://www.pelicanparts.com/MINI/index-SC.htm"
+    "https://www.pelicanparts.com/techarticles/Audi_tech/Audi_Tech_Index.htm"
+    "https://www.pelicanparts.com/techarticles/Volkswagen_Tech_Index.htm"
+    "https://www.pelicanparts.com/techarticles/Volvo_Tech.htm"
+    "https://www.pelicanparts.com/techarticles/Saab_Tech.htm"
+    "https://www.verdeck.de/blog/"
+    "https://www.verdeck.de/unser-material/"
+    "https://www.lederzentrum.de/wiki/index.php/Das_Lederzentrum_Lederlexikon"
+    "https://pieldetoro.net/web/default.php"
+    "https://www.aircraftinteriorsinternational.com/"
+    "https://www.ainonline.com/"
+    "https://www.railwayinteriorsinternational.com/"
+    "https://www.globalrailwayreview.com/"
+    "https://www.upholsteryresource.com/"
+)
+
+AVAILABLE=0
+FAILED=0
+
+for site in "${SITES[@]}"; do
+    echo -n "Testing: $site ... "
+    
+    # Timeout 10s, seguir redirects, user-agent
+    HTTP_CODE=$(curl -L -s -o /dev/null -w "%{http_code}" \
+        --max-time 10 \
+        -A "Mozilla/5.0 (compatible; CTF-Bot/1.0)" \
+        "$site" 2>/dev/null)
+    
+    if [ "$HTTP_CODE" -ge 200 ] && [ "$HTTP_CODE" -lt 400 ]; then
+        echo "✅ OK (HTTP $HTTP_CODE)"
+        ((AVAILABLE++))
+    else
+        echo "❌ FAILED (HTTP $HTTP_CODE)"
+        ((FAILED++))
+    fi
+    
+    sleep 1
+done
+
+echo ""
+echo "=============================================="
+echo "✅ Disponíveis: $AVAILABLE"
+echo "❌ Falhas: $FAILED"
+echo "📊 Total testado: ${#SITES[@]}"
+echo "=============================================="