Update docker/clam/docker-entrypoint.sh

docker/clam/docker-entrypoint.sh

@@ -4,28 +4,44 @@ MODE=${1:-"scan"}
 
 temp_log_file="/tmp/clamav_scan_$(date +%Y%m%d_%H%M%S).log"
 
+total_files_estimated=0
+total_files_scanned=0
+
 create_temp_log() {
     local log_file="$1"
     cat "$log_file" > "$temp_log_file"
 }
 
-scan() {
-    echo "Running ClamAV scan..."
-    local retry_count=0
-    local max_retries=5
+prepare_estimation() {
+    echo "Preparing file count estimation... (Note: This is an estimate and may include files across filesystems or symlinks if not all conditions are respected)"
+    local scan_dirs=$(find /scan -type d -print) # Lists all directories to scan
 
-    echo "Running scan in optimized mode to reduce resource usage."
-    while ! clamscan -r --max-scansize=100M --max-filesize=50M --max-recursion=5 --log=/var/log/clamav/clamav.log /scan; do
-        retry_count=$((retry_count + 1))
-        if [ "$retry_count" -ge "$max_retries" ]; then
-            echo "Max retries reached. Sending failure report to GlitchTip..."
-            create_temp_log "/var/log/clamav/clamav.log"
-            go-glitch report --dsn "$GLITCHTIP_DSN" "$temp_log_file" || echo "Failed to report scan failure to GlitchTip"
-            return
-        fi
-        echo "Scan failed. Retrying... ($retry_count/$max_retries)"
-        sleep 5
+    # Use parallel to count files across directories, matching clamscan criteria and ignoring symlinks
+    total_files_estimated=$(echo "$scan_dirs" | parallel -j $(nproc) \
+        "find {} -type f -not -type l -size -100M | wc -l" | \
+        awk '{s+=$1} END {print s}')
+
+    echo "Estimated total files to scan: $total_files_estimated"
+}
+
+scan() {
+    prepare_estimation
+
+    echo "Running ClamAV scan on each subdirectory sequentially..."
+    local scan_dirs=$(find /scan -type d -print) # Lists all directories to scan
+
+    # Run clamscan on each subdirectory in sequence, ignoring symlinks
+    for dir in $scan_dirs; do
+        clamscan -r --max-scansize=100M --max-filesize=50M --max-recursion=0 --follow-dir-symlinks=no --cross-filesystems=no --log=/var/log/clamav/clamav_${dir##*/}.log "$dir"
+        scanned_files=$(grep -E "^/.*: (OK|FOUND)" /var/log/clamav/clamav_${dir##*/}.log | wc -l)
+        total_files_scanned=$((total_files_scanned + scanned_files))
+        remaining_files=$((total_files_estimated - total_files_scanned))
+        echo "Progress: $total_files_scanned files scanned, $remaining_files files remaining"
     done
+
+    # Combine logs
+    cat /var/log/clamav/clamav_*.log > /var/log/clamav/clamav.log
+    rm /var/log/clamav/clamav_*.log
 }
 
 report() {