HTML Proofer options

gistfile1.txt

def link_check
            options = {
                  allow_hash_href: true, # default false,
                  allow_missing_href: true,
                  assume_extension: '.html', # default false,
                  checks: ['Links', 'Images', 'Scripts', 'OpenGraph'],
                  check_external_hash: false,
                  check_sri: false,
                  directory_index_file: 'index.html',
                  disable_external: false, # default false,
                  ignore_empty_alt: true, # default false,
                  enforce_https: false, 
                  extensions: ['.html'],
                  external_only: false,
                  ignore_files: [],
                  ignore_status_codes: [999],
                  log_level: :info,
                  ignore_empty_mailto: false,
                  only_4xx: false,
                  ignore_urls: [Regexp.new('^https://github.com/(orgs/)*NetAppDocs/\S*'), Regexp.new('^https://review.docs.netapp.com/\S*'), Regexp.new('^//cdn.jsdelivr.net/\S*'), Regexp.new('^\.\.\/\S*.html')],  #ignore review domain, protocol relative links and relative links, and github links 
                  swap_urls: {},
                  cache: {},
                  parallel: {}
            }
            ## Create baseurl out of lang and repo name
            lang = 'us-en'
            lang = ENV['LANG_REPO_NAME'].to_s unless ENV['LANG_REPO_NAME'].to_s.empty?
            folder = ENV['PREFIX_REPO_NAME'].to_s
            prefix = "^/#{lang}/#{folder}/"
            
            bc_url = "^/#{lang}/#{folder}" #breadcrumb url drops trailing "/" during check so does not find index.html and fails
            options[:ignore_urls] << Regexp.new(bc_url) #ignore breadcrumb url     
            puts "Baseurl: #{prefix}"
            
            ## Swap baseurl
            options[:swap_urls][Regexp.new(prefix)] = '/'
            
            options[:swap_urls][Regexp.new('^https://mysupport.netapp.com/site')] = 'https://mysupport.netapp.com/site/' #add trailing "/" to avoid redirect error

            #setting cache external and internal timeframe
            options[:cache][:timeframe] = {}
            options[:cache][:timeframe][:external] = '1w'
            options[:cache][:timeframe][:internal] =  '1w'
            
           
            pwd = Dir.pwd

            begin
                proofer = HTMLProofer.check_directory(pwd, options)
                proofer.run
            rescue Exception => e
                puts e
            end

            return 0
        end

        def run()
            link_check
        end