smart unzip of canvas quiz submissions

smart_unzip.sh

#!/bin/bash

# Function to check if required tools are available
check_dependencies() {
    local missing_tools=()
    
    if ! command -v unzip &> /dev/null; then
        missing_tools+=("unzip")
    fi
    
    if ! command -v unrar &> /dev/null && ! command -v rar &> /dev/null && ! command -v unar &> /dev/null; then
        missing_tools+=("unrar, rar, or unar")
    fi
    
    if ! command -v 7z &> /dev/null && ! command -v 7za &> /dev/null; then
        missing_tools+=("7z or 7za")
    fi
    
    if [[ ${#missing_tools[@]} -gt 0 ]]; then
        echo "Warning: Some extraction tools are missing:"
        printf '  - %s\n' "${missing_tools[@]}"
        echo ""
        echo "Install missing tools:"
        echo "  Ubuntu/Debian: sudo apt install unzip unrar-free p7zip-full"
        echo "  CentOS/RHEL:   sudo yum install unzip unrar p7zip"
        echo "  macOS:         brew install unzip unar p7zip"
        echo ""
    fi
}

# Function to get archive type from extension
get_archive_type() {
    local file="$1"
    local extension="${file##*.}"
    echo "$extension" | tr '[:upper:]' '[:lower:]' # Convert to lowercase
}

# Function to get base name without archive extension
get_base_name() {
    local file="$1"
    local base_name=$(basename "$file")
    
    # Remove various archive extensions
    if [[ "$base_name" =~ \.(zip|rar|7z)$ ]]; then
        echo "${base_name%.*}"
    else
        echo "$base_name"
    fi
}

# Function to extract archive based on type
extract_archive() {
    local archive_file="$1"
    local temp_dir="$2"
    local archive_type=$(get_archive_type "$archive_file")
    
    case "$archive_type" in
        "zip")
            if command -v unzip &> /dev/null; then
                unzip -q "$archive_file" -d "$temp_dir" 2>/dev/null
            else
                echo "Error: unzip not found"
                return 1
            fi
            ;;
        "rar")
            if command -v unar &> /dev/null; then
                # macOS preferred tool - unar extracts to current directory by default
                (cd "$temp_dir" && unar -q "$archive_file" 2>/dev/null)
            elif command -v unrar &> /dev/null; then
                unrar x -inul "$archive_file" "$temp_dir/" 2>/dev/null
            elif command -v rar &> /dev/null; then
                rar x -inul "$archive_file" "$temp_dir/" 2>/dev/null
            else
                echo "Error: unar/unrar/rar not found"
                return 1
            fi
            ;;
        "7z")
            if command -v 7z &> /dev/null; then
                7z x "$archive_file" -o"$temp_dir" -y >/dev/null 2>&1
            elif command -v 7za &> /dev/null; then
                7za x "$archive_file" -o"$temp_dir" -y >/dev/null 2>&1
            else
                echo "Error: 7z/7za not found"
                return 1
            fi
            ;;
        *)
            echo "Error: Unsupported archive type: $archive_type"
            return 1
            ;;
    esac
}

# Function to extract archive and rename folder
smart_extract() {
    local archive_file="$1"
    
    # Check if archive file exists
    if [[ ! -f "$archive_file" ]]; then
        echo "Error: Archive file '$archive_file' not found."
        return 1
    fi
    
    # Check if it's a supported archive type
    local archive_type=$(get_archive_type "$archive_file")
    if [[ ! "$archive_type" =~ ^(zip|rar|7z)$ ]]; then
        echo "Error: '$archive_file' is not a supported archive type (zip, rar, 7z)."
        return 1
    fi
    
    # Get the base name without archive extension
    local base_name=$(get_base_name "$archive_file")
    
    # Extract the prefix before the first sequence of numbers
    # This regex matches everything up to (but not including) the first digit
    local new_name=$(echo "$base_name" | sed 's/[0-9].*//')
    
    # Remove trailing underscore if present
    new_name=$(echo "$new_name" | sed 's/_$//')
    
    # Skip if target directory already exists
    if [[ -d "$new_name" ]]; then
        echo "Skipping '$archive_file' - directory '$new_name' already exists"
        return 0
    fi
    
    echo "Extracting '$archive_file' ($(echo "$archive_type" | tr '[:lower:]' '[:upper:]'))..."
    
    # Create a temporary directory for extraction
    local temp_dir=$(mktemp -d)
    
    # Extract to temporary directory
    if ! extract_archive "$archive_file" "$temp_dir"; then
        echo "Error: Failed to extract '$archive_file'"
        rm -rf "$temp_dir"
        return 1
    fi
    
    # Count items in temp directory (exclude hidden files starting with .)
    local item_count=$(ls -1 "$temp_dir" 2>/dev/null | grep -v '^\.' | wc -l)
    
    if [[ $item_count -eq 0 ]]; then
        echo "Warning: No files extracted from '$archive_file'"
        rm -rf "$temp_dir"
        return 1
    elif [[ $item_count -eq 1 ]]; then
        # Single item - move and rename it
        local extracted_item=$(ls -1 "$temp_dir" | grep -v '^\.' | head -1)
        mv "$temp_dir/$extracted_item" "./$new_name"
        echo "✓ Extracted and renamed to: $new_name"
    else
        # Multiple items - move temp dir contents to new folder
        mkdir -p "./$new_name"
        # Move all items except hidden files
        find "$temp_dir" -mindepth 1 -maxdepth 1 ! -name '.*' -exec mv {} "./$new_name/" \;
        echo "✓ Extracted multiple items to folder: $new_name"
    fi
    
    # Clean up temp directory
    rm -rf "$temp_dir"
}

# Function to extract all archives in directory
extract_all() {
    local archive_files=(*.zip *.rar *.7z *.ZIP *.RAR *.7Z)
    local valid_archives=()
    
    # Filter out non-existent files (when no files match the pattern)
    for file in "${archive_files[@]}"; do
        if [[ -f "$file" ]]; then
            valid_archives+=("$file")
        fi
    done
    
    # Check if any archive files exist
    if [[ ${#valid_archives[@]} -eq 0 ]]; then
        echo "No supported archive files (zip, rar, 7z) found in current directory."
        exit 1
    fi
    
    echo "Found ${#valid_archives[@]} archive file(s) in current directory:"
    for file in "${valid_archives[@]}"; do
        local archive_type=$(get_archive_type "$file")
        echo "  $file ($(echo "$archive_type" | tr '[:lower:]' '[:upper:]'))"
    done
    echo ""
    
    local success_count=0
    local skip_count=0
    local error_count=0
    
    # Process each archive file
    for archive_file in "${valid_archives[@]}"; do
        local before_dir_count=$(ls -ld */ 2>/dev/null | wc -l)
        
        if smart_extract "$archive_file"; then
            local after_dir_count=$(ls -ld */ 2>/dev/null | wc -l)
            if [[ $after_dir_count -gt $before_dir_count ]]; then
                ((success_count++))
            else
                ((skip_count++))
            fi
        else
            ((error_count++))
        fi
        echo ""
    done
    
    echo "=== Summary ==="
    echo "Successfully extracted: $success_count"
    echo "Skipped (already exists): $skip_count"
    echo "Errors: $error_count"
    echo "Total processed: ${#valid_archives[@]}"
}

# Main execution
echo "Smart Archive Extractor (ZIP, RAR, 7Z)"
echo "======================================"
echo ""

# Check dependencies
check_dependencies

if [[ $# -eq 0 ]]; then
    echo "Extracting all archive files in current directory..."
    echo ""
    extract_all
else
    # Single file mode
    archive_type=$(get_archive_type "$1")
    if [[ ! "$archive_type" =~ ^(zip|rar|7z)$ ]]; then
        echo "Error: '$1' is not a supported archive type (zip, rar, 7z)."
        exit 1
    fi
    smart_extract "$1"
fi