Downloads Wordpress images and media files from exported posts

wordpress-download-media.rb

# Downloads Wordpress images and media files from exported posts

# usage: 
# ruby ./wordpress-download-media.rb
# ruby ./wordpress-download-media.rb "_posts" "assets/media" "svg,pdf"

require "open-uri"
require "fileutils"

# default config
source = "_posts"
destination = "assets/media"
filetypes = []

if ARGV.length > 0
    source = ARGV[0]
end

if ARGV.length > 1
    destination = ARGV[1]
end

if ARGV.length > 2
    filetypes = ARGV[2].split(",")
end

# define functions
def open_file_content(file)
    #puts "Open '#{file}'"
    fh = open file
    content = fh.read
    fh.close
    return content
end

def file_ext(file)
    return File.extname(file).strip.downcase[1..-1]
end

def process_wp_post(file, types=[])
    lc_types = types.map(&:downcase)
    content = open_file_content(file)
    arr = []
    pattern = /https?:\/\/w{3}?[a-z0-9\-_.+*#?&=%@!',;:~\$\/]+\/wp-content\/uploads\/[0-9]{4}\/[0-9]{2}\/[a-z0-9\-_.+*#?&=%@!',;:~\$\/]+/mi
    content.scan(pattern) do |match|
        src = match.to_s
        if arr.include?(src) == false && (lc_types.length == 0 || lc_types.include?(file_ext(src)))
            arr.push(src)
        end
    end
    #puts "file: #{file}"
    return arr
end

def download_links(links, dest)
    arr = links.uniq
    puts "Start downloading #{arr.length} from #{links.length} links."
    arr.each do |src|
        download(src, dest)
    end
    return arr.length
end

def download(url, dest)
    destFile = File.join(dest, File.basename(url))
    puts "Download file from: #{url} to: #{destFile}"
    Dir.mkdir(dest) unless Dir.exist?(dest)
    case io = open(url)
        when StringIO then File.open(destFile, 'w') { |f| f.write(io) }
        when Tempfile then io.close; FileUtils.mv(io.path, destFile)
    end
end

# script start
puts filetypes.length == 0 ? "Download .* media types" : "Download #{filetypes.length} media types: #{filetypes}" 

# media download list
downloads = []

# process all html files in directory
i = 0
Dir.glob("#{source}**/*") do |post|
    downloads.concat( process_wp_post(post, filetypes) )
    i += 1
end
count = download_links(downloads, destination)
puts "#{i} posts processed and #{count} files downloaded into #{destination} dir"