Merge branch 'master' into fix-recheck-failure

Author: gjtorikian

.travis.yml

@@ -1,6 +1,5 @@
 language: ruby
 rvm:
-  - 2.3.6
   - 2.4.3
   - 2.5.0
   - 2.6.0

README.md

@@ -321,7 +321,9 @@ You can pass in additional options to configure this validation.
 
 | Option | Description | Default |
 | :----- | :---------- | :------ |
+| `report_eof_tags` | When `check_html` is enabled, HTML markup with mismatched tags are reported as errors | `false`
 | `report_invalid_tags` | When `check_html` is enabled, HTML markup that is unknown to Nokogumbo are reported as errors. | `false`
+| `report_mismatched_tags` | When `check_html` is enabled, HTML markup with tags that are malformed are reported as errors | `false`
 | `report_missing_doctype` | When `check_html` is enabled, HTML markup with missing or out-of-order `DOCTYPE` are reported as errors. | `false`
 | `report_missing_names` | When `check_html` is enabled, HTML markup that are missing entity names are reported as errors. | `false`
 | `report_script_embeds` | When `check_html` is enabled, `script` tags containing markup are reported as errors. | `false`

Rakefile

@@ -13,6 +13,10 @@ RuboCop::RakeTask.new(:rubocop)
 
 task default: %i[spec proof_readme]
 
+task :test do
+  Rake::Task['spec'].invoke
+end
+
 task :proof_readme do
   require 'html-proofer'
   require 'redcarpet'

bin/htmlproofer

@@ -41,6 +41,8 @@ Mercenary.program(:htmlproofer) do |p|
   p.option 'report_missing_names', '--report-missing-names', 'When `check_html` is enabled, HTML markup that are missing entity names are reported as errors (default: `false`)'
   p.option 'report_script_embeds', '--report-script-embeds', 'When `check_html` is enabled, `script` tags containing markup are reported as errors (default: `false`)'
   p.option 'report_missing_doctype', '--report-missing-doctype', 'When `check_html` is enabled, HTML markup with missing or out-of-order `DOCTYPE` are reported as errors (default: `false`)'
+  p.option 'report_eof_tags', '--report-eof-tags', 'When `check_html` is enabled, HTML markup with tags that are malformed are reported as errors (default: `false`)'
+  p.option 'report_mismatched_tags', '--report-mismatched-tags', 'When `check_html` is enabled, HTML markup with mismatched tags are reported as errors (default: `false`)'
   p.option 'log_level', '--log-level <level>', String, 'Sets the logging level, as determined by Yell. One of `:debug`, `:info`, `:warn`, `:error`, or `:fatal`. (default: `:info`)'
   p.option 'only_4xx', '--only-4xx', 'Only reports errors for links that fall within the 4xx status code range'
   p.option 'storage_dir', '--storage-dir PATH', String, 'Directory where to store the cache log (default: "tmp/.htmlproofer")'
@@ -82,6 +84,8 @@ Mercenary.program(:htmlproofer) do |p|
     options[:validation][:report_missing_names] = opts['report_missing_names'] unless opts['report_missing_names'].nil?
     options[:validation][:report_invalid_tags] = opts['report_invalid_tags'] unless opts['report_invalid_tags'].nil?
     options[:validation][:report_missing_doctype] = opts['report_missing_doctype'] unless opts['report_missing_doctype'].nil?
+    options[:validation][:report_eof_tags] = opts['report_eof_tags'] unless opts['report_eof_tags'].nil?
+    options[:validation][:report_mismatched_tags] = opts['report_mismatched_tags'] unless opts['report_mismatched_tags'].nil?
 
     options[:typhoeus] = HTMLProofer::Configuration.parse_json_option('typhoeus_config', opts['typhoeus_config']) unless opts['typhoeus_config'].nil?
 

lib/html-proofer/check.rb

@@ -5,18 +5,19 @@ module HTMLProofer
   class Check
     attr_reader :node, :html, :element, :src, :path, :options, :issues, :external_urls
 
-    def initialize(src, path, html, options)
+    def initialize(src, path, html, logger, options)
       @src    = src
       @path   = path
       @html   = remove_ignored(html)
+      @logger = logger
       @options = options
       @issues = []
       @external_urls = {}
     end
 
     def create_element(node)
       @node = node
-      Element.new(node, self)
+      Element.new(node, self, @logger)
     end
 
     def run

lib/html-proofer/check/html.rb

@@ -6,7 +6,9 @@ class HtmlCheck < ::HTMLProofer::Check
   INVALID_TAG_MSG = /Tag ([\w\-:]+) invalid/.freeze
   INVALID_PREFIX = /Namespace prefix/.freeze
   PARSE_ENTITY_REF = /htmlParseEntityRef: no name/.freeze
-  DOCTYPE_MSG = /The doctype must be the first token in the document/.freeze
+  DOCTYPE_MSG = /Expected a doctype token/.freeze
+  EOF_IN_TAG = /End of input in tag/.freeze
+  MISMATCHED_TAGS = /That tag isn't allowed here/.freeze
 
   def run
     @html.errors.each do |error|
@@ -24,6 +26,10 @@ def report?(message)
       options[:validation][:report_missing_names]
     when DOCTYPE_MSG
       options[:validation][:report_missing_doctype]
+    when EOF_IN_TAG
+      options[:validation][:report_eof_tags]
+    when MISMATCHED_TAGS
+      options[:validation][:report_mismatched_tags]
     else
       true
     end

lib/html-proofer/check/opengraph.rb

@@ -3,8 +3,8 @@
 class OpenGraphElement < ::HTMLProofer::Element
   attr_reader :src
 
-  def initialize(obj, check)
-    super(obj, check)
+  def initialize(obj, check, logger)
+    super(obj, check, logger)
     # Fake up src from the content attribute
     instance_variable_set('@src', @content)
 
@@ -23,7 +23,7 @@ def empty_src?
 
   def run
     @html.css('meta[property="og:url"], meta[property="og:image"]').each do |m|
-      @opengraph = OpenGraphElement.new(m, self)
+      @opengraph = OpenGraphElement.new(m, self, @logger)
 
       next if @opengraph.ignore?
 

lib/html-proofer/configuration.rb

@@ -52,7 +52,9 @@ module Configuration
       report_script_embeds: false,
       report_missing_names: false,
       report_invalid_tags: false,
-      report_missing_doctype: false
+      report_missing_doctype: false,
+      report_eof_tags: false,
+      report_mismatched_tags: false
     }.freeze
 
     CACHE_DEFAULTS = {}.freeze

lib/html-proofer/element.rb

@@ -10,12 +10,18 @@ class Element
 
     attr_reader :id, :name, :alt, :href, :link, :src, :line, :data_proofer_ignore
 
-    def initialize(obj, check)
+    def initialize(obj, check, logger)
+      @logger = logger
       # Construct readable ivars for every element
-      obj.attributes.each_pair do |attribute, value|
-        name = attribute.tr('-:.', '_').to_s.to_sym
-        (class << self; self; end).send(:attr_reader, name)
-        instance_variable_set("@#{name}", value.value)
+      begin
+        obj.attributes.each_pair do |attribute, value|
+          name = attribute.tr('-:.;', '_').to_s.to_sym
+          (class << self; self; end).send(:attr_reader, name)
+          instance_variable_set("@#{name}", value.value)
+        end
+      rescue NameError => e
+        @logger.log :error, "Attribute set `#{obj}` contains an error!"
+        raise e
       end
 
       @aria_hidden = defined?(@aria_hidden) && @aria_hidden == 'true' ? true : false

lib/html-proofer/middleware.rb

@@ -21,7 +21,8 @@ def self.options
         allow_hash_href: true,
         check_external_hash: true,
         check_html: true,
-        url_ignore: [/.*/] # Don't try to check local files exist
+        url_ignore: [/.*/], # Don't try to check if local files exist
+        validation: { report_eof_tags: true }
       }
     end
 

lib/html-proofer/runner.rb

@@ -100,7 +100,7 @@ def check_parsed(html, path)
       @src.each do |src|
         checks.each do |klass|
           @logger.log :debug, "Checking #{klass.to_s.downcase} on #{path} ..."
-          check = Object.const_get(klass).new(src, path, html, @options)
+          check = Object.const_get(klass).new(src, path, html, @logger, @options)
           check.run
           external_urls = check.external_urls
           external_urls = Hash[check.external_urls.map { |url, file| [swap(url, @options[:url_swap]), file] }] if @options[:url_swap]
@@ -147,6 +147,8 @@ def ignore_file?(file)
     def checks
       return @checks if defined?(@checks) && !@checks.nil?
 
+      return (@checks = ['LinkCheck']) if @type == :links
+
       @checks = HTMLProofer::Check.subchecks.map(&:name)
       @checks.delete('FaviconCheck') unless @options[:check_favicon]
       @checks.delete('HtmlCheck') unless @options[:check_html]

lib/html-proofer/utils.rb

@@ -15,7 +15,7 @@ def create_nokogiri(path)
                   path
                 end
 
-      Nokogiri::HTML5(content)
+      Nokogiri::HTML5(content, max_errors: -1)
     end
 
     def swap(href, replacement)

lib/html-proofer/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module HTMLProofer
-  VERSION = '3.15.1'
+  VERSION = '3.15.2'
 end

spec/html-proofer/command_spec.rb

@@ -89,9 +89,9 @@
   end
 
   it 'works with check-html' do
-    broken = "#{FIXTURES_DIR}/html/unmatched_end_tag.html"
-    output = make_bin('--check-html --report-invalid-tags', broken)
-    expect(output).to match('HTML-Proofer finished successfully')
+    broken = "#{FIXTURES_DIR}/html/missing_closing_quotes.html"
+    output = make_bin('--check-html --report-eof-tags', broken)
+    expect(output).to match('1 failure')
   end
 
   it 'works with empty-alt-ignore' do

spec/html-proofer/element_spec.rb

@@ -4,55 +4,55 @@
 
 describe HTMLProofer::Element do
   before(:each) do
-    @check = HTMLProofer::Check.new('', '', Nokogiri::HTML5(''), HTMLProofer::Configuration::PROOFER_DEFAULTS)
+    @check = HTMLProofer::Check.new('', '', Nokogiri::HTML5(''), nil, HTMLProofer::Configuration::PROOFER_DEFAULTS)
   end
 
   describe '#initialize' do
     it 'accepts the xmlns attribute' do
       nokogiri = Nokogiri::HTML5('<a xmlns:cc="http://creativecommons.org/ns#">Creative Commons</a>')
-      checkable = HTMLProofer::Element.new(nokogiri.css('a').first, @check)
+      checkable = HTMLProofer::Element.new(nokogiri.css('a').first, @check, nil)
       expect(checkable.instance_variable_get(:@xmlns_cc)).to eq 'http://creativecommons.org/ns#'
     end
 
     it 'assignes the text node' do
       nokogiri = Nokogiri::HTML5('<p>One')
-      checkable = HTMLProofer::Element.new(nokogiri.css('p').first, @check)
+      checkable = HTMLProofer::Element.new(nokogiri.css('p').first, @check, nil)
       expect(checkable.instance_variable_get(:@text)).to eq 'One'
     end
 
     it 'accepts the content attribute' do
       nokogiri = Nokogiri::HTML5('<meta name="twitter:card" content="summary">')
-      checkable = HTMLProofer::Element.new(nokogiri.css('meta').first, @check)
+      checkable = HTMLProofer::Element.new(nokogiri.css('meta').first, @check, nil)
       expect(checkable.instance_variable_get(:@content)).to eq 'summary'
     end
   end
 
   describe '#ignores_pattern_check' do
     it 'works for regex patterns' do
       nokogiri = Nokogiri::HTML5('<script src=/assets/main.js></script>')
-      checkable = HTMLProofer::Element.new(nokogiri.css('script').first, @check)
+      checkable = HTMLProofer::Element.new(nokogiri.css('script').first, @check, nil)
       expect(checkable.ignores_pattern_check([%r{\/assets\/.*(js|css|png|svg)}])).to eq true
     end
 
     it 'works for string patterns' do
       nokogiri = Nokogiri::HTML5('<script src=/assets/main.js></script>')
-      checkable = HTMLProofer::Element.new(nokogiri.css('script').first, @check)
+      checkable = HTMLProofer::Element.new(nokogiri.css('script').first, @check, nil)
       expect(checkable.ignores_pattern_check(['/assets/main.js'])).to eq true
     end
   end
 
   describe '#url' do
     it 'works for src attributes' do
       nokogiri = Nokogiri::HTML5('<img src=image.png />')
-      checkable = HTMLProofer::Element.new(nokogiri.css('img').first, @check)
+      checkable = HTMLProofer::Element.new(nokogiri.css('img').first, @check, nil)
       expect(checkable.url).to eq 'image.png'
     end
   end
 
   describe '#ignore' do
     it 'works for twitter cards' do
       nokogiri = Nokogiri::HTML5('<meta name="twitter:url" data-proofer-ignore content="http://example.com/soon-to-be-published-url">')
-      checkable = HTMLProofer::Element.new(nokogiri.css('meta').first, @check)
+      checkable = HTMLProofer::Element.new(nokogiri.css('meta').first, @check, nil)
       expect(checkable.ignore?).to eq true
     end
   end

spec/html-proofer/fixtures/html/parse_failure.html

@@ -0,0 +1 @@
+<!DOCTYPE html><html><!-- -- --></a>

spec/html-proofer/fixtures/images/semicolon.html

@@ -0,0 +1,9 @@
+<img
+  class="img-fluid"
+  src="./gpl.png"
+  alt="Frank Blank"
+  ;
+  width="365"
+  ;
+  height="365"
+/><br />

spec/html-proofer/html_spec.rb

@@ -124,4 +124,11 @@
     proofer = run_proofer(file, :file, opts)
     expect(proofer.failed_tests).to eq []
   end
+
+  it 'reports failures' do
+    opts = { check_html: true, validation: { report_mismatched_tags: true } }
+    file = "#{FIXTURES_DIR}/html/parse_failure.html"
+    proofer = run_proofer(file, :file, opts)
+    expect(proofer.failed_tests.first).to match(/ERROR: That tag isn't allowed here/)
+  end
 end

spec/html-proofer/images_spec.rb

@@ -201,4 +201,10 @@
     proofer = run_proofer(relative_images, :file)
     expect(proofer.failed_tests).to eq []
   end
+
+  it 'ignores semicolon outside attribute name' do
+    relative_images = "#{FIXTURES_DIR}/images/semicolon.html"
+    proofer = run_proofer(relative_images, :file)
+    expect(proofer.failed_tests).to eq []
+  end
 end