html-proofer 2.6.4 → 3.0.0

data/lib/html-proofer/version.rb

@@ -0,0 +1,3 @@
+module HTMLProofer
+  VERSION = '3.0.0'
+end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: html-proofer
 version: !ruby/object:Gem::Version
-  version: 2.6.4
+  version: 3.0.0
 platform: ruby
 authors:
 - Garen Torikian
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2016-01-26 00:00:00.000000000 Z
+date: 2016-03-05 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: mercenary
@@ -210,27 +210,27 @@ description: Test your rendered HTML files to make sure they're accurate.
 email:
 - gjtorikian@gmail.com
 executables:
-- htmlproof
+- htmlproofer
 extensions: []
 extra_rdoc_files: []
 files:
-- bin/htmlproof
-- lib/html/proofer.rb
-- lib/html/proofer/cache.rb
-- lib/html/proofer/check_runner.rb
-- lib/html/proofer/check_runner/issue.rb
-- lib/html/proofer/checkable.rb
-- lib/html/proofer/checks/favicon.rb
-- lib/html/proofer/checks/html.rb
-- lib/html/proofer/checks/images.rb
-- lib/html/proofer/checks/links.rb
-- lib/html/proofer/checks/scripts.rb
-- lib/html/proofer/configuration.rb
-- lib/html/proofer/log.rb
-- lib/html/proofer/url_validator.rb
-- lib/html/proofer/utils.rb
-- lib/html/proofer/version.rb
-- lib/html/proofer/xpathfunctions.rb
+- bin/htmlproofer
+- lib/html-proofer.rb
+- lib/html-proofer/cache.rb
+- lib/html-proofer/check.rb
+- lib/html-proofer/check/favicon.rb
+- lib/html-proofer/check/html.rb
+- lib/html-proofer/check/images.rb
+- lib/html-proofer/check/links.rb
+- lib/html-proofer/check/scripts.rb
+- lib/html-proofer/configuration.rb
+- lib/html-proofer/element.rb
+- lib/html-proofer/issue.rb
+- lib/html-proofer/log.rb
+- lib/html-proofer/runner.rb
+- lib/html-proofer/url_validator.rb
+- lib/html-proofer/utils.rb
+- lib/html-proofer/version.rb
 homepage: https://github.com/gjtorikian/html-proofer
 licenses:
 - MIT

data/lib/html/proofer.rb

@@ -1,191 +0,0 @@
-def require_all(path)
-  glob = File.join(File.dirname(__FILE__), path, '*.rb')
-  Dir[glob].each do |f|
-    require f
-  end
-end
-
-require_all 'proofer'
-require_all 'proofer/check_runner'
-require_all 'proofer/checks'
-
-require 'parallel'
-require 'fileutils'
-
-begin
-  require 'awesome_print'
-rescue LoadError; end
-
-module HTML
-  class Proofer
-    include HTML::Proofer::Utils
-
-    attr_reader :options, :typhoeus_opts, :hydra_opts, :parallel_opts, :validation_opts, :external_urls, :iterable_external_urls
-
-    def initialize(src, opts = {})
-      FileUtils.mkdir_p(STORAGE_DIR) unless File.exist?(STORAGE_DIR)
-
-      @src = src
-
-      if opts[:verbose]
-        warn '`@options[:verbose]` will be removed in a future 3.x.x release: http://git.io/vGHHh'
-      end
-      if opts[:href_ignore]
-        warn '`@options[:href_ignore]` will be renamed in a future 3.x.x release: http://git.io/vGHHy'
-      end
-
-      @proofer_opts = HTML::Proofer::Configuration::PROOFER_DEFAULTS
-
-      @typhoeus_opts = HTML::Proofer::Configuration::TYPHOEUS_DEFAULTS.merge(opts[:typhoeus] || {})
-      opts.delete(:typhoeus)
-
-      @hydra_opts = HTML::Proofer::Configuration::HYDRA_DEFAULTS.merge(opts[:hydra] || {})
-      opts.delete(:hydra)
-
-      # fall back to parallel defaults
-      @parallel_opts = opts[:parallel] || {}
-      opts.delete(:parallel)
-
-      @validation_opts = opts[:validation] || {}
-      opts.delete(:validation)
-
-      @options = @proofer_opts.merge(opts)
-
-      @failed_tests = []
-    end
-
-    def logger
-      @logger ||= HTML::Proofer::Log.new(@options[:verbose], @options[:verbosity])
-    end
-
-    def run
-      logger.log :info, :blue, "Running #{checks} on #{@src} on *#{@options[:ext]}... \n\n"
-
-      if @src.is_a?(Array) && !@options[:disable_external]
-        check_list_of_links
-      else
-        check_directory_of_files
-      end
-
-      if @failed_tests.empty?
-        logger.log :info, :green, 'HTML-Proofer finished successfully.'
-      else
-        print_failed_tests
-      end
-    end
-
-    def check_list_of_links
-      if @options[:href_swap]
-        @src = @src.map do |url|
-          swap(url, @options[:href_swap])
-        end
-      end
-      @external_urls = Hash[*@src.map { |s| [s, nil] }.flatten]
-      validate_urls
-    end
-
-    # Collects any external URLs found in a directory of files. Also collectes
-    # every failed test from check_files_for_internal_woes.
-    # Sends the external URLs to Typhoeus for batch processing.
-    def check_directory_of_files
-      @external_urls = {}
-      results = check_files_for_internal_woes
-
-      results.each do |item|
-        @external_urls.merge!(item[:external_urls])
-        @failed_tests.concat(item[:failed_tests])
-      end
-
-      # TODO: lazy. if we're checking only external links,
-      # we'll just trash all the failed tests. really, we should
-      # just not run those other checks at all.
-      if @options[:external_only]
-        @failed_tests = []
-        validate_urls
-      elsif !@options[:disable_external]
-        validate_urls
-      end
-
-      count = files.length
-      file_text = pluralize(count, 'file', 'files')
-      logger.log :info, :blue, "Ran on #{file_text}!\n\n"
-    end
-
-    # Walks over each implemented check and runs them on the files, in parallel.
-    def check_files_for_internal_woes
-      if @parallel_opts.empty?
-        files.map { |path| check_path(path) }
-      else
-        Parallel.map(files, @parallel_opts) { |path| check_path(path) }
-      end
-    end
-
-    def check_path(path)
-      html = create_nokogiri(path)
-      result = { :external_urls => {}, :failed_tests => [] }
-
-      checks.each do |klass|
-        logger.log :debug, :yellow, "Checking #{klass.to_s.downcase} on #{path} ..."
-        check = Object.const_get(klass).new(@src, path, html, @options, @typhoeus_opts, @hydra_opts, @parallel_opts, @validation_opts)
-        check.run
-        result[:external_urls].merge!(check.external_urls)
-        result[:failed_tests].concat(check.issues) if check.issues.length > 0
-      end
-      result
-    end
-
-    def validate_urls
-      url_validator = HTML::Proofer::UrlValidator.new(logger, @external_urls, @options, @typhoeus_opts, @hydra_opts)
-      @failed_tests.concat(url_validator.run)
-      @iterable_external_urls = url_validator.iterable_external_urls
-    end
-
-    def files
-      if File.directory? @src
-        pattern = File.join(@src, '**', "*#{@options[:ext]}")
-        files = Dir.glob(pattern).select { |fn| File.file? fn }
-        files.reject { |f| ignore_file?(f) }
-      elsif File.extname(@src) == @options[:ext]
-        [@src].reject { |f| ignore_file?(f) }
-      else
-        []
-      end
-    end
-
-    def ignore_file?(file)
-      options[:file_ignore].each do |pattern|
-        return true if pattern.is_a?(String) && pattern == file
-        return true if pattern.is_a?(Regexp) && pattern =~ file
-      end
-
-      false
-    end
-
-    def checks
-      return @checks unless @checks.nil?
-      @checks = HTML::Proofer::CheckRunner.checks.map(&:name)
-      @checks.delete('FaviconCheck') unless @options[:check_favicon]
-      @checks.delete('HtmlCheck') unless @options[:check_html]
-      @options[:checks_to_ignore].each do |ignored|
-        @checks.delete(ignored)
-      end
-      @checks
-    end
-
-    def failed_tests
-      return [] if @failed_tests.empty?
-      result = []
-      @failed_tests.each { |f| result << f.to_s }
-      result
-    end
-
-    def print_failed_tests
-      sorted_failures = HTML::Proofer::CheckRunner::SortedIssues.new(@failed_tests, @options[:error_sort], logger)
-
-      sorted_failures.sort_and_report
-      count = @failed_tests.length
-      failure_text = pluralize(count, 'failure', 'failures')
-      fail logger.colorize :red, "HTML-Proofer found #{failure_text}!"
-    end
-  end
-end

data/lib/html/proofer/cache.rb

@@ -1,141 +0,0 @@
-require_relative 'utils'
-
-require 'json'
-require 'active_support/core_ext/string'
-require 'active_support/core_ext/date'
-require 'active_support/core_ext/numeric/time'
-
-module HTML
-  class Proofer
-    class Cache
-      include HTML::Proofer::Utils
-
-      FILENAME = File.join(STORAGE_DIR, 'cache.log')
-
-      attr_accessor :exists, :load, :cache_log, :cache_time
-
-      def initialize(logger, options)
-        @logger = logger
-        @cache_log = {}
-
-        if options.nil? || options.empty?
-          @load = false
-        else
-          @load = true
-          @parsed_timeframe = parsed_timeframe(options[:timeframe] || '30d')
-        end
-        @cache_time = Time.now
-
-        if File.exist?(FILENAME)
-          @exists = true
-          contents = File.read(FILENAME)
-          @cache_log = contents.empty? ? {} : JSON.parse(contents)
-        else
-          @exists = false
-        end
-      end
-
-      def within_timeframe?(time)
-        (@parsed_timeframe..@cache_time).cover?(time)
-      end
-
-      def urls
-        @cache_log['urls'] || []
-      end
-
-      def parsed_timeframe(timeframe)
-        time, date = timeframe.match(/(\d+)(\D)/).captures
-        time = time.to_f
-        case date
-        when 'M'
-          time.months.ago
-        when 'w'
-          time.weeks.ago
-        when 'd'
-          time.days.ago
-        when 'h'
-          time.hours.ago
-        else
-          fail ArgumentError, "#{date} is not a valid timeframe!"
-        end
-      end
-
-      def add(url, filenames, status, msg = '')
-        data = {
-                  :time => @cache_time,
-                  :filenames => filenames,
-                  :status => status,
-                  :message => msg
-               }
-
-        @cache_log[clean_url(url)] = data
-      end
-
-      def detect_url_changes(found)
-        existing_urls = @cache_log.keys.map { |url| clean_url(url) }
-        found_urls = found.keys.map { |url| clean_url(url) }
-
-        # prepare to add new URLs detected
-        additions = found.reject do |url, _|
-          url = clean_url(url)
-          if existing_urls.include?(url)
-            true
-          else
-            @logger.log :debug, :yellow, "Adding #{url} to cache check"
-            false
-          end
-        end
-
-        new_link_count = additions.length
-        new_link_text = pluralize(new_link_count, 'link', 'links')
-        @logger.log :info, :blue, "Adding #{new_link_text} to the cache..."
-
-        # remove from cache URLs that no longer exist
-        del = 0
-        @cache_log.delete_if do |url, _|
-          url = clean_url(url)
-          if !found_urls.include?(url)
-            @logger.log :debug, :yellow, "Removing #{url} from cache check"
-            del += 1
-            true
-          else
-            false
-          end
-        end
-
-        del_link_text = pluralize(del, 'link', 'links')
-        @logger.log :info, :blue, "Removing #{del_link_text} from the cache..."
-
-        additions
-      end
-
-      def write
-        File.write(FILENAME, @cache_log.to_json)
-      end
-
-      def load?
-        @load.nil?
-      end
-
-
-      # FIXME: there seems to be some discrepenacy where Typhoeus occasionally adds
-      # a trailing slash to URL strings, which causes issues with the cache
-      def slashless_url(url)
-        url.chomp('/')
-      end
-
-      # FIXME: it seems that Typhoeus actually acts on escaped URLs,
-      # but there's no way to get at that information, and the cache
-      # stores unescaped URLs. Because of this, some links, such as
-      # github.com/search/issues?q=is:open+is:issue+fig are not matched
-      # as github.com/search/issues?q=is%3Aopen+is%3Aissue+fig
-      def unescape_url(url)
-        Addressable::URI.unescape(url)
-      end
-
-      def clean_url(url)
-        slashless_url(unescape_url(url))
-      end
-    end
-  end
-end

data/lib/html/proofer/check_runner.rb

@@ -1,70 +0,0 @@
-# encoding: utf-8
-
-module HTML
-  class Proofer
-    # Mostly handles issue management and collecting of external URLs.
-    class CheckRunner
-
-      attr_reader :issues, :src, :path, :options, :typhoeus_opts, :hydra_opts, :parallel_opts, \
-                  :validation_opts, :external_urls, :href_ignores, :url_ignores, :alt_ignores, \
-                  :empty_alt_ignore, :allow_hash_href
-
-      def initialize(src, path, html, options, typhoeus_opts, hydra_opts, parallel_opts, validation_opts)
-        @src    = src
-        @path   = path
-        @html   = remove_ignored(html)
-        @options = options
-        @typhoeus_opts = typhoeus_opts
-        @hydra_opts = hydra_opts
-        @parallel_opts = parallel_opts
-        @validation_opts = validation_opts
-        @issues = []
-        @href_ignores = @options[:href_ignore]
-        @url_ignores = @options[:url_ignore]
-        @alt_ignores = @options[:alt_ignore]
-        @empty_alt_ignore = @options[:empty_alt_ignore]
-        @allow_hash_href = @options[:allow_hash_href]
-        @external_urls = {}
-      end
-
-      def run
-        fail NotImplementedError, 'HTML::Proofer::CheckRunner subclasses must implement #run'
-      end
-
-      def add_issue(desc, line_number = nil, status = -1)
-        @issues << Issue.new(@path, desc, line_number, status)
-      end
-
-      def add_to_external_urls(url, line)
-        return if @external_urls[url]
-        add_path_for_url(url)
-      end
-
-      def add_path_for_url(url)
-        if @external_urls[url]
-          @external_urls[url] << @path
-        else
-          @external_urls[url] = [@path]
-        end
-      end
-
-      def self.checks
-        classes = []
-
-        ObjectSpace.each_object(Class) do |c|
-          next unless c.superclass == self
-          classes << c
-        end
-
-        classes
-      end
-
-      private
-
-      def remove_ignored(html)
-        html.css('code, pre, tt').each(&:unlink)
-        html
-      end
-    end
-  end
-end