citadel/services/paperless/restore

#!/bin/bash

# Paperless Restore Script
# Location: /home/citadel/services/paperless/restore
# Usage: ./restore <snapshot_id> --test|--production|--extract
#        ./restore --clean
#        ./restore --clean-all

set -euo pipefail

# Configuration
readonly SERVICE_NAME="paperless"
readonly SERVICE_DIR="/home/citadel/services/$SERVICE_NAME"
readonly TEST_DIR="$SERVICE_DIR/test-restore"
readonly CONFIG_FILE="/home/citadel/backup/restic.conf"
readonly SCRIPT_NAME=$(basename "$0")

# Colors
readonly RED='\033[0;31m'
readonly GREEN='\033[0;32m'
readonly YELLOW='\033[1;33m'
readonly BLUE='\033[0;34m'
readonly NC='\033[0m'

# Logging functions
log() { echo -e "${BLUE}[$(date '+%H:%M:%S')]${NC} $1"; }
error() { echo -e "${RED}[ERROR]${NC} $1" >&2; exit 1; }
success() { echo -e "${GREEN}[SUCCESS]${NC} $1"; }
warning() { echo -e "${YELLOW}[WARNING]${NC} $1"; }

# Global variables
SNAPSHOT_ID=""
MODE=""
TEMP_DIR=""

# Show help
show_help() {
    cat << EOF
Usage: $SCRIPT_NAME <snapshot_id> --test|--production|--extract
       $SCRIPT_NAME --clean|--clean-all

Arguments:
    snapshot_id     The restic snapshot ID to restore from

Options:
    --test          Restore to test instance (isolated environment)
    --production    Restore to production instance (NOT YET IMPLEMENTED)
    --extract       Extract snapshot to /tmp directory only
    --clean         Clean test environment (stop containers and remove volumes)
    --clean-all     Clean test environment AND remove all temp directories
    -h, --help      Show this help message

Examples:
    $SCRIPT_NAME abc123 --test        # Restore snapshot abc123 to test instance
    $SCRIPT_NAME abc123 --extract     # Extract snapshot abc123 to /tmp only
    $SCRIPT_NAME latest --production  # Restore latest snapshot to production
    $SCRIPT_NAME --clean              # Clean test environment
    $SCRIPT_NAME --clean-all          # Clean test environment + temp directories

The test instance will be created in: $TEST_DIR
EOF
}

# Cleanup function
cleanup() {
    if [ -n "$TEMP_DIR" ] && [ -d "$TEMP_DIR" ] && [[ "$TEMP_DIR" == /tmp/* ]]; then
        log "Cleaning up temporary directory: $TEMP_DIR"
        rm -rf "$TEMP_DIR"
    fi
}

# Set up cleanup on exit (only for extract mode)
setup_cleanup() {
    if [ "$MODE" = "extract" ]; then
        trap cleanup EXIT
    fi
}

# Parse arguments
parse_arguments() {
    if [ $# -eq 0 ]; then
        show_help
        exit 0
    fi

    while [ $# -gt 0 ]; do
        case "$1" in
            --test)
                MODE="test"
                shift
                ;;
            --production)
                MODE="production"
                shift
                ;;
            --extract)
                MODE="extract"
                shift
                ;;
            --clean)
                MODE="clean"
                shift
                ;;
            --clean-all)
                MODE="clean-all"
                shift
                ;;
            -h|--help)
                show_help
                exit 0
                ;;
            -*)
                error "Unknown option: $1"
                ;;
            *)
                if [ -z "$SNAPSHOT_ID" ]; then
                    SNAPSHOT_ID="$1"
                else
                    error "Too many arguments"
                fi
                shift
                ;;
        esac
    done

    # Validate required arguments
    if [ "$MODE" != "clean" ] && [ "$MODE" != "clean-all" ]; then
        if [ -z "$SNAPSHOT_ID" ]; then
            error "Snapshot ID is required (except for --clean or --clean-all mode)"
        fi
    fi

    if [ -z "$MODE" ]; then
        error "Mode (--test, --production, --extract, --clean, or --clean-all) is required"
    fi

    # Clean modes don't need snapshot ID
    if [ "$MODE" = "clean" ] || [ "$MODE" = "clean-all" ]; then
        if [ -n "$SNAPSHOT_ID" ]; then
            error "Snapshot ID should not be provided with --clean or --clean-all mode"
        fi
    fi
}

# Check prerequisites
check_prerequisites() {
    # Skip restic checks for clean modes
    if [ "$MODE" = "clean" ] || [ "$MODE" = "clean-all" ]; then
        # Only check Docker for clean modes
        if ! command -v docker &>/dev/null; then
            error "Docker not found"
        fi
        
        if ! docker compose version &>/dev/null; then
            error "Docker Compose not found"
        fi
        
        log "Prerequisites OK (clean mode)"
        return
    fi

    # Check configuration file
    if [ ! -f "$CONFIG_FILE" ]; then
        error "Config file not found: $CONFIG_FILE"
    fi

    # Source config
    source "$CONFIG_FILE"

    # Check required variables
    if [ -z "${RESTIC_REPOSITORY:-}" ]; then
        error "RESTIC_REPOSITORY not defined in config"
    fi

    # Test restic access
    if ! restic snapshots &>/dev/null; then
        error "Cannot access Restic repository"
    fi

    # Check Docker (except for extract mode)
    if [ "$MODE" != "extract" ]; then
        if ! command -v docker &>/dev/null; then
            error "Docker not found"
        fi

        # Check docker compose
        if ! docker compose version &>/dev/null; then
            error "Docker Compose not found"
        fi
    fi

    log "Prerequisites OK"
}

# Create secure temporary directory
create_temp_dir() {
    if [ "$MODE" = "extract" ]; then
        # For extract mode, create in /tmp with readable name
        TEMP_DIR="/tmp/paperless-extract-$(date +%Y%m%d-%H%M%S)"
        mkdir -p "$TEMP_DIR"
        chmod 755 "$TEMP_DIR"  # More permissive for extract mode
    else
        # For other modes, use secure temp
        TEMP_DIR=$(mktemp -d "/tmp/paperless-restore-$(date +%Y%m%d-%H%M%S)-XXXXXX")
        chmod 700 "$TEMP_DIR"
    fi

    if [ ! -d "$TEMP_DIR" ]; then
        error "Failed to create temporary directory"
    fi

    log "Created temporary directory: $TEMP_DIR"
}

# Extract snapshot
extract_snapshot() {
    log "Extracting snapshot $SNAPSHOT_ID to temporary directory"

    if ! restic restore "$SNAPSHOT_ID" --target "$TEMP_DIR"; then
        error "Failed to extract snapshot $SNAPSHOT_ID"
    fi

    success "Snapshot extracted successfully"

    # Show what was extracted
    log "Extracted contents:"
    ls -la "$TEMP_DIR" | head -10
    
    # Show paperless data structure if exists
    if [ -d "$TEMP_DIR/tmp/paperless" ]; then
        echo ""
        log "Paperless data structure:"
        ls -la "$TEMP_DIR/tmp/paperless/" | head -10
    fi
}

# Extract only mode
extract_only() {
    log "Starting extract-only mode"
    
    # Create temp directory in /tmp
    create_temp_dir
    
    # Extract snapshot
    extract_snapshot
    
    # Display information about extracted content
    echo ""
    success "Snapshot $SNAPSHOT_ID extracted to: $TEMP_DIR"
    echo ""
    echo "📁 Extracted structure:"
    if [ -d "$TEMP_DIR/tmp/paperless" ]; then
        echo "   Main data location: $TEMP_DIR/tmp/paperless/"
        ls -la "$TEMP_DIR/tmp/paperless/" | head -10
    else
        find "$TEMP_DIR" -maxdepth 2 -type d | head -20
    fi
    echo ""
    echo "📊 Content summary:"
    echo "   Database dumps: $(find "$TEMP_DIR" -name "*.sql" | wc -l) files"
    echo "   Data directories: $(find "$TEMP_DIR/tmp/paperless" -maxdepth 1 -type d 2>/dev/null | grep -v "^$TEMP_DIR/tmp/paperless$" | wc -l) directories"
    echo "   Total size: $(du -sh "$TEMP_DIR" | cut -f1)"
    echo ""
    echo "💡 Manual inspection commands:"
    echo "   cd $TEMP_DIR/tmp/paperless"
    echo "   ls -la"
    echo "   find . -name '*.sql' -exec ls -lh {} +"
    echo ""
    echo "⚠️  The directory will remain until you delete it manually:"
    echo "   rm -rf $TEMP_DIR"
    echo ""
    
    # Don't run cleanup for extract mode - leave directory for user
    trap - EXIT
}

# Clean test environment
clean_test_environment() {
    log "Starting cleanup of test environment"
    
    # Navigate to service directory to ensure we have access to .env
    cd "$SERVICE_DIR"
    
    # Stop test containers
    log "Stopping test containers..."
    if [ -d "$TEST_DIR" ]; then
        cd "$TEST_DIR"
        docker compose --env-file "$SERVICE_DIR/.env" -p paperless-test down --remove-orphans 2>/dev/null || true
        cd "$SERVICE_DIR"
    else
        log "Test directory doesn't exist, checking for running containers anyway..."
    fi
    
    # Try to stop containers by name pattern (in case they're running from different location)
    log "Stopping any remaining paperless test containers..."
    docker stop paperless-webserver-test paperless-db-test paperless-broker-test paperless-gotenberg-test paperless-tika-test 2>/dev/null || true
    docker rm paperless-webserver-test paperless-db-test paperless-broker-test paperless-gotenberg-test paperless-tika-test 2>/dev/null || true
    
    # Remove test volumes
    log "Removing test volumes..."
    local volumes_to_remove=(
        "paperless-ngx_pgdata-test"
        "redisdata-test" 
        "paperless-data-test"
        "paperless-media-test"
        "paperless-export-test"
        "paperless-consume-test"
    )
    
    for volume in "${volumes_to_remove[@]}"; do
        if docker volume ls -q | grep -q "^${volume}$"; then
            log "Removing volume: $volume"
            docker volume rm "$volume" 2>/dev/null || warning "Failed to remove volume: $volume"
        else
            log "Volume not found (already removed): $volume"
        fi
    done
    
    # Clean up any dangling images related to paperless
    log "Cleaning up dangling images..."
    docker image prune -f &>/dev/null || true
    
    success "Test environment cleaned successfully"
    echo ""
    echo "🧹 Cleanup completed!"
    echo "   ✅ Test containers stopped and removed"
    echo "   ✅ Test volumes removed"
    echo "   ✅ Dangling images cleaned"
    echo ""
    echo "💡 The production environment remains untouched."
    echo "💡 The test directory ($TEST_DIR) is preserved."
    echo "💡 You can now run a fresh test restore if needed."
}

# Clean all environment (test + temp directories)
clean_all_environment() {
    log "Starting cleanup of test environment and temp directories"
    
    # First, run the standard test environment cleanup
    clean_test_environment
    
    # Then clean temporary directories
    echo ""
    log "Cleaning temporary directories..."
    
    # Find and remove paperless restore temp directories
    local temp_dirs_restore=()
    local temp_dirs_extract=()
    
    # Use find to safely locate temp directories
    while IFS= read -r -d '' dir; do
        temp_dirs_restore+=("$dir")
    done < <(find /tmp -maxdepth 1 -type d -name "paperless-restore-*" -print0 2>/dev/null)
    
    while IFS= read -r -d '' dir; do
        temp_dirs_extract+=("$dir")
    done < <(find /tmp -maxdepth 1 -type d -name "paperless-extract-*" -print0 2>/dev/null)
    
    # Remove restore temp directories
    if [ ${#temp_dirs_restore[@]} -gt 0 ]; then
        log "Found ${#temp_dirs_restore[@]} restore temp directories to remove"
        for dir in "${temp_dirs_restore[@]}"; do
            if [ -d "$dir" ]; then
                log "Removing: $(basename "$dir")"
                rm -rf "$dir" || warning "Failed to remove: $dir"
            fi
        done
    else
        log "No restore temp directories found"
    fi
    
    # Remove extract temp directories  
    if [ ${#temp_dirs_extract[@]} -gt 0 ]; then
        log "Found ${#temp_dirs_extract[@]} extract temp directories to remove"
        for dir in "${temp_dirs_extract[@]}"; do
            if [ -d "$dir" ]; then
                log "Removing: $(basename "$dir")"
                rm -rf "$dir" || warning "Failed to remove: $dir"
            fi
        done
    else
        log "No extract temp directories found"
    fi
    
    # Show final summary
    echo ""
    success "Complete cleanup finished!"
    echo ""
    echo "🧹 Full cleanup completed!"
    echo "   ✅ Test containers stopped and removed"
    echo "   ✅ Test volumes removed"
    echo "   ✅ Dangling images cleaned"
    echo "   ✅ Restore temp directories removed: ${#temp_dirs_restore[@]}"
    echo "   ✅ Extract temp directories removed: ${#temp_dirs_extract[@]}"
    echo ""
    echo "💡 The production environment remains untouched."
    echo "💡 The test directory ($TEST_DIR) is preserved."
    echo "💡 System is now clean and ready for fresh operations."
}

# Restore to test instance
restore_to_test() {
    log "Starting restore to test instance"

    # Create test directory if it doesn't exist
    mkdir -p "$TEST_DIR"

    # Navigate to test directory
    cd "$TEST_DIR"

    log "Stopping test containers (if running)..."
    docker compose --env-file "$SERVICE_DIR/.env" -p paperless-test down --remove-orphans 2>/dev/null || true

    # Remove existing test volumes to ensure clean restore
    log "Removing existing test volumes..."
    docker volume rm paperless-ngx_pgdata-test redisdata-test \
        paperless-data-test paperless-media-test \
        paperless-export-test paperless-consume-test 2>/dev/null || true

    # Find database dump
    local db_dump
    db_dump=$(find "$TEMP_DIR" -name "*_db_*.sql" | head -1)

    if [ -z "$db_dump" ]; then
        error "No database dump found in snapshot"
    fi

    log "Found database dump: $(basename "$db_dump")"

    # Create PostgreSQL volume before starting database container
    log "Creating PostgreSQL volume..."
    docker volume create paperless-ngx_pgdata-test

    # Start only the database container for restore
    log "Starting test database container..."
    docker compose --env-file "$SERVICE_DIR/.env" -p paperless-test up -d db-test

    # Wait for database to be ready
    log "Waiting for database to be ready..."
    sleep 15

    # Source environment variables
    if [ -f "$SERVICE_DIR/.env" ]; then
        source "$SERVICE_DIR/.env"
    else
        error "Environment file not found: $SERVICE_DIR/.env"
    fi

    # Restore database
    log "Restoring database from dump..."
    if docker compose --env-file "$SERVICE_DIR/.env" -p paperless-test exec -T db-test psql -U "$POSTGRES_USER" -d "$POSTGRES_DB" < "$db_dump"; then
        success "Database restored successfully"
    else
        error "Database restore failed"
    fi

    # Stop database container
    docker compose --env-file "$SERVICE_DIR/.env" -p paperless-test down

    # Restore data volumes using temporary containers
    log "Restoring data volumes..."

    # Create ALL volumes (Docker Compose will not create them since they're external)
    log "Creating all required volumes..."
    docker volume create paperless-ngx_pgdata-test
    docker volume create redisdata-test
    docker volume create paperless-data-test
    docker volume create paperless-media-test
    docker volume create paperless-export-test
    docker volume create paperless-consume-test

    # Function to restore volume data
    restore_volume_data() {
        local source_path="$1"
        local volume_name="$2"
        local container_path="$3"

        if [ -d "$source_path" ]; then
            log "Restoring $volume_name from $source_path"
            docker run --rm \
                -v "$volume_name:$container_path" \
                -v "$source_path:$container_path-source:ro" \
                alpine:latest \
                sh -c "cp -rf $container_path-source/* $container_path/ 2>/dev/null || true"
        else
            warning "Source path not found: $source_path"
        fi
    }

    # Restore each data directory
    restore_volume_data "$TEMP_DIR/tmp/paperless/data" "paperless-data-test" "/data"
    restore_volume_data "$TEMP_DIR/tmp/paperless/media" "paperless-media-test" "/media"
    restore_volume_data "$TEMP_DIR/tmp/paperless/export" "paperless-export-test" "/export"
    restore_volume_data "$TEMP_DIR/tmp/paperless/consume" "paperless-consume-test" "/consume"

    # Start all test containers
    log "Starting test instance..."
    docker compose --env-file "$SERVICE_DIR/.env" -p paperless-test up -d

    # Wait for services to be ready
    log "Waiting for services to start..."
    sleep 30

    # Check if webserver is running
    if docker compose --env-file "$SERVICE_DIR/.env" -p paperless-test ps --filter "status=running" | grep -q "webserver-test"; then
        success "Test instance started successfully"
        echo ""
        echo "🎉 Test instance is ready!"
        echo "📍 Container name: paperless-webserver-test"
        echo "🌐 Access: Configure SWAG to redirect paperless.alouettes.jombi.fr temporarily"
        echo "🔧 Docker compose location: $TEST_DIR"
        echo ""
        echo "💡 To stop the test instance:"
        echo "   cd $TEST_DIR && docker compose --env-file $SERVICE_DIR/.env -p paperless-test down"
        echo ""
        echo "💡 To clean test environment completely:"
        echo "   $SERVICE_DIR/restore --clean"
    else
        error "Test instance failed to start properly"
    fi
}

# Restore to production (not implemented)
restore_to_production() {
    echo ""
    echo "🚧 PRODUCTION RESTORE NOT YET IMPLEMENTED 🚧"
    echo ""
    echo "Production restore functionality is not yet available."
    echo "This feature requires additional safety measures and validation."
    echo ""
    echo "For now, you can:"
    echo "1. Use --test mode to restore to the test instance"
    echo "2. Verify the restored data in the test environment"
    echo "3. Manually copy data from test to production if needed"
    echo ""
    echo "Production restore will be implemented in a future version."
    echo ""
    exit 1
}

# Main function
main() {
    echo "=== Paperless Restore Script ==="

    # Parse arguments
    parse_arguments "$@"

    # Display operation info
    if [ "$MODE" = "clean" ]; then
        log "Mode: $MODE"
    else
        log "Snapshot ID: $SNAPSHOT_ID"
        log "Mode: $MODE"
    fi

    # Check prerequisites
    check_prerequisites

    # Setup cleanup if needed
    setup_cleanup

    # Execute based on mode
    case "$MODE" in
        clean)
            clean_test_environment
            ;;
        clean-all)
            clean_all_environment
            ;;
        extract)
            extract_only
            ;;
        test)
            # Create temporary directory and extract
            create_temp_dir
            extract_snapshot
            restore_to_test
            ;;
        production)
            # Create temporary directory and extract
            create_temp_dir
            extract_snapshot
            restore_to_production
            ;;
        *)
            error "Invalid mode: $MODE"
            ;;
    esac

    success "Operation completed!"
}

# Entry point
main "$@"