Add extension to files where extension is missing

fix_extensions.sh

#!/bin/bash

# Function to display help
display_help() {
    echo "Usage: $0 <source_directory> [destination_directory] [-overwrite] [-start <start_index>] [-limit <max_files>]"
    echo
    echo "Parameters:"
    echo "  source_directory        Directory containing files without extensions (use '.' for current directory)"
    echo "  destination_directory   Optional: Directory to save files with added extensions (defaults to source_directory)"
    echo "  -overwrite              Optional flag to overwrite original files"
    echo "  -start <index>          Optional starting file index (default is 0)"
    echo "  -limit <max_files>      Optional maximum number of files to process (default is all)"
    echo
    echo "Example:"
    echo "  $0 ./source_dir ./destination_dir -overwrite -start 5 -limit 10"
    exit 1
}

# Default values
overwrite="no"
start=0
limit=999999  # Process all files by default

# Check if at least the source directory is provided
if [[ $# -lt 1 ]]; then
    display_help
fi

# Parse command-line arguments
while [[ $# -gt 0 ]]; do
    case "$1" in
        -overwrite)
            overwrite="overwrite"
            shift
            ;;
        -start)
            start="$2"
            shift 2
            ;;
        -limit)
            limit="$2"
            shift 2
            ;;
        *)
            if [[ -z "$source_dir" ]]; then
                source_dir="$1"
            elif [[ -z "$destination_dir" ]]; then
                destination_dir="$1"
            else
                echo "Unexpected argument: $1"
                display_help
            fi
            shift
            ;;
    esac
done

# Get absolute source directory if it's relative
if [[ "$source_dir" != /* ]]; then
    source_dir="$(pwd)/$source_dir"
fi

# Set destination directory to the source directory if not provided
if [[ -z "$destination_dir" ]]; then
    destination_dir="$source_dir"
fi

# Create the destination directory if it doesn't exist
mkdir -p "$destination_dir"

# Function to map file descriptions to appropriate extensions
get_extension_by_description() {
    local file_type="$1"

    case "$file_type" in
        *PDF*) echo "pdf" ;;
        *EPUB*) echo "epub" ;;
        *HTML*) echo "html" ;;
        *ASCII*) echo "txt" ;;
        *Zip*) echo "zip" ;;
        *RAR*) echo "rar" ;;
        *JPEG*) echo "jpg" ;;
        *PNG*) echo "png" ;;
        *DjVu*) echo "djvu" ;;
        *Mobipocket\ E-book*) echo "mobi" ;;
        *Microsoft\ Word*) echo "docx" ;;
        *OpenDocument\ Text*) echo "odt" ;;
        *Microsoft\ Excel*) echo "xlsx" ;;
        *OpenDocument\ Spreadsheet*) echo "ods" ;;
        *MP3*) echo "mp3" ;;
        *MPEG*) echo "mpg" ;;
        *Matroska*) echo "mkv" ;;
        *Video*) echo "mp4" ;;
        *Audio*) echo "wav" ;;
        *) echo "unknown" ;;
    esac
}

# Start timer
start_time=$(date +%s)

# Counter for processed files
count=0
errors=()

# Determine action to perform
action="Copying"
if [ "$overwrite" == "overwrite" ]; then
    action="Overwriting"
fi

# Inform the user of the action being taken
echo "$action files from $source_dir to $destination_dir"

# Iterate over each file in the source directory
for file in "$source_dir"/*; do
    # Skip directories and hidden files
    if [ -d "$file" ] || [[ "$file" == .* ]]; then
        continue
    fi

    # Skip files that already have an extension
    if [[ "$file" == *.* ]]; then
        continue
    fi

    # Check if we've reached the start limit
    if [ "$count" -lt "$start" ]; then
        count=$((count + 1))
        continue
    fi

    # Break if we've reached the limit
    if [ "$count" -ge $((start + limit)) ]; then
        break
    fi

    # First try to guess the extension using `file --extension`
    extension=$(file -b --extension "$file" | awk -F'/' '{print $1}')

    # If the extension is "???" or similar (invalid), try the second method
    if [[ "$extension" == "???" || -z "$extension" ]]; then
        # Guess the file type description using `file`
        file_type=$(file -b "$file")
        # Use the function to get a better extension based on the description
        extension=$(get_extension_by_description "$file_type")
    fi

    # If we still can't determine the extension, record the error
    if [ "$extension" == "unknown" ]; then
        errors+=("$file (Type: $file_type)")
        continue
    fi

    # Create a new file path with the detected extension
    filename=$(basename "$file")
    new_file="$destination_dir/$filename.$extension"

    # Perform the file operation based on the overwrite flag
    if [ "$overwrite" == "overwrite" ]; then
        mv "$file" "$new_file"
    else
        cp "$file" "$new_file"
    fi

    count=$((count + 1))
done

# End timer
end_time=$(date +%s)
duration=$((end_time - start_time))

# Summary of processed files
echo "Processed $count files in $duration seconds."

# If there were any errors, report them at the end
if [ ${#errors[@]} -gt 0 ]; then
    echo "The following files could not be processed:"
    for error in "${errors[@]}"; do
        echo "$error"
    done
fi