#!/bin/bash

# Define variables
DUPLICATE_LIST="duplicates.txt"  # Hard-coded duplicate list file
LOG_FILE="deleted_duplicates.log"  # Main log file
TIME_LOG_FILE="deletion_times.log"  # Secondary log file to track deletion time
OUTPUT_LOG_FILE="output.log"  # Third log file for terminal output
SAVED_SPACE_LOG_FILE="saved_space.log"  # Log file for saved space
ERROR_LOG_FILE="error.log"  # Log file for errors
DRY_RUN=false  # Set to true to simulate actions
CONFIRM=false  # Set to true to enable confirmation before deletion

# Function to display usage
usage() {
    echo "Usage: $0 [--dry-run] [--confirm]"
    echo "  --dry-run   Simulate the file removal process."
    echo "  --confirm   Ask for confirmation before each delete."
    exit 1
}

# Parse optional flags
for arg in "$@"; do
    case $arg in
        --dry-run)
            DRY_RUN=true
            shift
            ;;
        --confirm)
            CONFIRM=true
            shift
            ;;
        *)
            ;;
    esac
done

# Create or clear the log files
: > "$LOG_FILE"
: > "$TIME_LOG_FILE"
: > "$OUTPUT_LOG_FILE"
: > "$SAVED_SPACE_LOG_FILE"
: > "$ERROR_LOG_FILE"  # Clear the error log file

# Initialize counters and total saved space
total_files_processed=0
total_files_deleted=0
total_files_skipped=0
total_saved_bytes=0

# Iterate over each duplicate file path in the list
while IFS= read -r file_path; do
    total_files_processed=$((total_files_processed + 1))

    if [ -f "$file_path" ]; then
        # Show a confirmation prompt (if enabled)
        if [ "$CONFIRM" = true ]; then
            read -p "Delete $file_path? (y/n) " response
            if [[ "$response" != "y" ]]; then
                echo "Skipped: $file_path" | tee -a "$LOG_FILE" "$OUTPUT_LOG_FILE"
                total_files_skipped=$((total_files_skipped + 1))
                continue
            fi
        fi

        # Get the size of the file before deletion
        file_size=$(stat -c%s "$file_path")
        total_saved_bytes=$((total_saved_bytes + file_size))

        # Time the deletion or moving process
        start_time=$(date +%s%3N)

        # Check if we're in dry run mode
        if [ "$DRY_RUN" = true ]; then
            echo "[Dry Run] Would delete: $file_path (Size: $file_size bytes)" | tee -a "$LOG_FILE" "$OUTPUT_LOG_FILE"
        else
            # Attempt to move the file
            if mv "$file_path" "$BACKUP_DIR/"; then
                echo "Moved: $file_path" | tee -a "$LOG_FILE" "$OUTPUT_LOG_FILE"
                total_files_deleted=$((total_files_deleted + 1))
            else
                echo "Error moving file: $file_path" | tee -a "$ERROR_LOG_FILE"
            fi
        fi

        end_time=$(date +%s%3N)
        duration=$((end_time - start_time))

        # Log the time taken to the secondary log file
        echo "File: $file_path | Time: ${duration}ms" >> "$TIME_LOG_FILE"
    else
        echo "File not found: $file_path" | tee -a "$LOG_FILE" "$OUTPUT_LOG_FILE"
        total_files_skipped=$((total_files_skipped + 1))
    fi
done < "$DUPLICATE_LIST"

# Convert total saved bytes to human-readable format
human_readable_size=$(numfmt --to=iec-i --suffix=B $total_saved_bytes)

# Log the total saved space in both formats
echo "Total space saved: $total_saved_bytes bytes ($human_readable_size)" | tee -a "$SAVED_SPACE_LOG_FILE" "$OUTPUT_LOG_FILE"

# Summary Report
echo "----- Summary Report -----" | tee -a "$OUTPUT_LOG_FILE"
echo "Total files processed: $total_files_processed" | tee -a "$OUTPUT_LOG_FILE"
echo "Total files deleted: $total_files_deleted" | tee -a "$OUTPUT_LOG_FILE"
echo "Total files skipped: $total_files_skipped" | tee -a "$OUTPUT_LOG_FILE"

echo "Duplicate removal process completed. See $LOG_FILE, $TIME_LOG_FILE, $OUTPUT_LOG_FILE, $SAVED_SPACE_LOG_FILE, and $ERROR_LOG_FILE for details."