# check4sig - Dataset Caption File Watermark Detection Utility
#
# Purpose:
# This function scans .caption files in a dataset directory to identify and edit
# files containing watermark-related text. It's particularly useful for cleaning
# image datasets where caption files might indicate the presence of watermarks.
#
# Usage: check4sig /path/to/dataset
#
# Function workflow:
# 1. Validates the provided directory path
# 2. Searches all .caption files for "signature" or "watermark" keywords
# 3. Opens any matching files in nvim for manual review/editing
#
# Parameters:
# $1 - Target directory containing .caption files
#
# Returns:
# - 1 if directory is invalid or missing
# - Opens nvim if matching files are found
# - Prints message if no matches are found
check4sig() {
    target_dir="$1"
    if [[ -z "$target_dir" ]]; then
        echo "Please provide a target directory."
        return 1
    fi

    if [[ ! -d "$target_dir" ]]; then
        echo "The provided target directory does not exist."
        return 1
    fi

    found_files=()
    for file in "$target_dir"/*.caption; do
        if [[ -f "$file" ]]; then
            if grep -q -e "signature" -e "watermark" "$file"; then
                found_files+=("$file")
            fi
        fi
    done

    if [[ ${#found_files[@]} -eq 0 ]]; then
        echo "No 'signature' or 'watermark' found in any .caption files."
    else
        echo "Opening files in nvim: ${found_files[@]}"
        nvim "${found_files[@]}"
    fi
}