html-proofer 3.16.0 → 3.17.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 39ffac5e6c23ab7b75d99dd69779be2109ed8b02dc4a5cb0a33740119cb66766
-  data.tar.gz: cdaf142dfde936fa1d88d410499c04fdbbb3b9814bc00ba6800ff3edb8acee50
+  metadata.gz: 9b5a47189e872130e01a2080e6e8ddb1f7f22520098deaf941960194d2338b2b
+  data.tar.gz: 3543c42860956427e8c828861f76e1c5cd984d79054c091c6e5f97e0352b3137
 SHA512:
-  metadata.gz: 34a4253b2eb9674060583c0f9006f9c36204bfd8bbb3e347293874e8db3248153e1efcdb13cbb9a5fc7967aabe478773a6a33c3c2372b4e819ecc0c41af43f36
-  data.tar.gz: 34530aeb5fb2836568aa5f7ebea7f154ef8e58c82ac0803aeba5632f704e7aed6c8111a095546c56b1331fa3dc1b563b08eac7cff0205aee4056815ebd8a6478
+  metadata.gz: 1defd1fb2ee651015231b0ed4d1407ed56e25a907e0602ee9e175b23f3b88938e020f438d2171ceea8d52bc58984901543c97184a3da13c9b08cdaf62200a4aa
+  data.tar.gz: 97eb6120db724822830398a85f14faa4415638e965d21aef0703991e7a14d335f32df65d72bb877be6f0706f24154deea3f08a6bbeb2cdf5d7c392d03ce2af23

data/bin/htmlproofer

@@ -1,7 +1,7 @@
 #!/usr/bin/env ruby
 # frozen_string_literal: true
 
-STDOUT.sync = true
+$stdout.sync = true
 
 $LOAD_PATH.unshift File.join(File.dirname(__FILE__), '..', 'lib')
 

data/lib/html-proofer/cache.rb

@@ -59,6 +59,8 @@ module HTMLProofer
     end
 
     def add(url, filenames, status, msg = '')
+      return unless use_cache?
+
       data = {
         time: @cache_time,
         filenames: filenames,
@@ -92,12 +94,12 @@ module HTMLProofer
       del = 0
       @cache_log.delete_if do |url, _|
         url = clean_url(url)
-        if !found_urls.include?(url)
+        if found_urls.include?(url)
+          false
+        else
           @logger.log :debug, "Removing #{url} from cache check"
           del += 1
           true
-        else
-          false
         end
       end
 
@@ -115,12 +117,11 @@ module HTMLProofer
       @load.nil?
     end
 
-    def retrieve_urls(external_urls)
-      urls_to_check = detect_url_changes(external_urls)
+    def retrieve_urls(urls)
+      urls_to_check = detect_url_changes(urls)
       @cache_log.each_pair do |url, cache|
-        if within_timeframe?(cache['time'])
-          next if cache['message'].empty? # these were successes to skip
-        end
+        next if within_timeframe?(cache['time']) && cache['message'].empty? # these were successes to skip
+
         urls_to_check[url] = cache['filenames'] # recheck expired links
       end
       urls_to_check

data/lib/html-proofer/check.rb

@@ -3,15 +3,17 @@
 module HTMLProofer
   # Mostly handles issue management and collecting of external URLs.
   class Check
-    attr_reader :node, :html, :element, :src, :path, :options, :issues, :external_urls
+    attr_reader :node, :html, :element, :src, :path, :options, :issues, :internal_urls, :external_urls
 
-    def initialize(src, path, html, logger, options)
+    def initialize(src, path, html, logger, cache, options)
       @src    = src
       @path   = path
       @html   = remove_ignored(html)
       @logger = logger
+      @cache = cache
       @options = options
       @issues = []
+      @internal_urls = {}
       @external_urls = {}
     end
 
@@ -26,15 +28,20 @@ module HTMLProofer
 
     def add_issue(desc, line: nil, status: -1, content: nil)
       @issues << Issue.new(@path, desc, line: line, status: status, content: content)
+      false
+    end
+
+    def add_to_internal_urls(url, internal_url)
+      if @internal_urls[url]
+        @internal_urls[url] << internal_url
+      else
+        @internal_urls[url] = [internal_url]
+      end
     end
 
     def add_to_external_urls(url)
       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

data/lib/html-proofer/check/links.rb

@@ -4,6 +4,8 @@ class LinkCheck < ::HTMLProofer::Check
   include HTMLProofer::Utils
 
   def missing_href?
+    return blank?(@link.src) if @node.name == 'source'
+
     blank?(@link.href) && blank?(@link.name) && blank?(@link.id)
   end
 
@@ -12,7 +14,7 @@ class LinkCheck < ::HTMLProofer::Check
   end
 
   def run
-    @html.css('a, link').each do |node|
+    @html.css('a, link, source').each do |node|
       @link = create_element(node)
       line = node.line
       content = node.to_s
@@ -49,23 +51,31 @@ class LinkCheck < ::HTMLProofer::Check
         # curl/Typheous inaccurately return 404s for some links. cc https://git.io/vyCFx
         next if @link.respond_to?(:rel) && @link.rel == 'dns-prefetch'
 
-        add_to_external_urls(@link.href)
+        add_to_external_urls(@link.href || @link.src)
         next
-      elsif @link.internal? && !@link.exists?
-        add_issue("internally linking to #{@link.href}, which does not exist", line: line, content: content)
+      elsif @link.internal?
+        if @link.exists?
+          add_to_internal_urls(@link.href, InternalLink.new(@link, @path, line, content))
+        else
+          add_issue("internally linking to #{@link.href}, which does not exist", line: line, content: content)
+        end
       end
+    end
 
-      # does the local directory have a trailing slash?
-      if @link.unslashed_directory? @link.absolute_path
-        add_issue("internally linking to a directory #{@link.absolute_path} without trailing slash", line: line, content: content)
-        next
-      end
+    external_urls
+  end
 
-      # verify the target hash
-      handle_hash(@link, line, content) if @link.hash
+  def check_internal_link(link, line, content)
+    # does the local directory have a trailing slash?
+    if link.unslashed_directory?(link.absolute_path)
+      add_issue("internally linking to a directory #{link.absolute_path} without trailing slash", line: line, content: content)
+      return false
     end
 
-    external_urls
+    # verify the target hash
+    return handle_hash(link, line, content) if link.hash
+
+    true
   end
 
   def check_schemes(link, line, content)
@@ -94,23 +104,27 @@ class LinkCheck < ::HTMLProofer::Check
   end
 
   def handle_hash(link, line, content)
-    if link.internal? && !hash_check(link.html, link.hash)
-      add_issue("linking to internal hash ##{link.hash} that does not exist", line: line, content: content)
+    if link.internal? && !hash_exists?(link.html, link.hash) # rubocop:disable Style/GuardClause
+      return add_issue("linking to internal hash ##{link.hash} that does not exist", line: line, content: content)
     elsif link.external?
-      external_link_check(link, line, content)
+      return external_link_check(link, line, content)
     end
+
+    true
   end
 
   def external_link_check(link, line, content)
-    if !link.exists?
-      add_issue("trying to find hash of #{link.href}, but #{link.absolute_path} does not exist", line: line, content: content)
+    if link.exists? # rubocop:disable Style/GuardClause
+      target_html = create_nokogiri(link.absolute_path)
+      return add_issue("linking to #{link.href}, but #{link.hash} does not exist", line: line, content: content) unless hash_exists?(target_html, link.hash)
     else
-      target_html = create_nokogiri link.absolute_path
-      add_issue("linking to #{link.href}, but #{link.hash} does not exist", line: line, content: content) unless hash_check target_html, link.hash
+      return add_issue("trying to find hash of #{link.href}, but #{link.absolute_path} does not exist", line: line, content: content)
     end
+
+    true
   end
 
-  def hash_check(html, href_hash)
+  def hash_exists?(html, href_hash)
     decoded_href_hash = Addressable::URI.unescape(href_hash)
     fragment_ids = [href_hash, decoded_href_hash]
     # https://www.w3.org/TR/html5/single-page.html#scroll-to-fragid
@@ -148,7 +162,19 @@ class LinkCheck < ::HTMLProofer::Check
 
   class XpathFunctions
     def case_sensitive_equals(node_set, str_to_match)
-      node_set.find_all { |node| node.to_s. == str_to_match.to_s }
+      node_set.find_all { |node| node.to_s.== str_to_match.to_s }
+    end
+  end
+
+  class InternalLink
+    attr_reader :link, :href, :path, :line, :content
+
+    def initialize(link, path, line, content)
+      @link = link
+      @href = @link.href
+      @path = path
+      @line = line
+      @content = content
     end
   end
 end

data/lib/html-proofer/configuration.rb

@@ -80,7 +80,7 @@ module HTMLProofer
       begin
         JSON.parse(config)
       rescue StandardError
-        raise ArgumentError, "Option '" + option_name + "' did not contain valid JSON."
+        raise ArgumentError, "Option '#{option_name} did not contain valid JSON."
       end
     end
   end

data/lib/html-proofer/element.rb

@@ -108,9 +108,7 @@ module HTMLProofer
       return true if /^javascript:/.match?(url)
 
       # ignore base64 encoded images
-      if %w[ImageCheck FaviconCheck].include? @type
-        return true if /^data:image/.match?(url)
-      end
+      return true if %w[ImageCheck FaviconCheck].include?(@type) && /^data:image/.match?(url)
 
       # ignore user defined URLs
       return true if ignores_pattern_check(@check.options[:url_ignore])
@@ -187,7 +185,7 @@ module HTMLProofer
         end
       elsif File.exist?(File.expand_path(path, @check.src)) || File.exist?(File.expand_path(path_dot_ext, @check.src)) # relative links, path is a file
         base = File.dirname @check.path
-      elsif File.exist?(File.join(File.dirname(@check.path), path)) || File.exist?(File.join(File.dirname(@check.path), path_dot_ext)) # relative links in nested dir, path is a file
+      elsif File.exist?(File.join(File.dirname(@check.path), path)) || File.exist?(File.join(File.dirname(@check.path), path_dot_ext)) # rubocop:disable Lint/DuplicateBranch; relative links in nested dir, path is a file
         base = File.dirname @check.path
       else # relative link, path is a directory
         base = @check.path

data/lib/html-proofer/middleware.rb

@@ -6,6 +6,7 @@ module HTMLProofer
 
     class InvalidHtmlError < StandardError
       def initialize(failures)
+        super
         @failures = failures
       end
 

data/lib/html-proofer/runner.rb

@@ -4,7 +4,7 @@ module HTMLProofer
   class Runner
     include HTMLProofer::Utils
 
-    attr_reader :options, :external_urls, :failures
+    attr_reader :options, :internal_urls, :external_urls, :failures
 
     def initialize(src, opts = {})
       @src = src
@@ -20,6 +20,8 @@ module HTMLProofer
 
       @type = @options.delete(:type)
       @logger = HTMLProofer::Log.new(@options[:log_level])
+      @cache = Cache.new(@logger, @options[:cache])
+      @internal_link_checks = nil
 
       # Add swap patterns for internal domains
       unless @options[:internal_domains].empty?
@@ -30,6 +32,9 @@ module HTMLProofer
         end
       end
 
+      @internal_urls = {}
+      @internal_urls_to_paths = {}
+      @external_urls = {}
       @failures = []
       @before_request = []
     end
@@ -59,15 +64,13 @@ module HTMLProofer
         end
       end
       @external_urls = Hash[*@src.map { |s| [s, nil] }.flatten]
-      validate_urls
+      validate_external_urls
     end
 
     # Collects any external URLs found in a directory of files. Also collectes
     # every failed test from process_files.
     # Sends the external URLs to Typhoeus for batch processing.
     def check_files
-      @external_urls = {}
-
       process_files.each do |item|
         @external_urls.merge!(item[:external_urls])
         @failures.concat(item[:failures])
@@ -78,9 +81,12 @@ module HTMLProofer
       # just not run those other checks at all.
       if @options[:external_only]
         @failures = []
-        validate_urls
+        validate_external_urls
       elsif !@options[:disable_external]
-        validate_urls
+        validate_external_urls
+        validate_internal_urls
+      else
+        validate_internal_urls
       end
     end
 
@@ -101,8 +107,21 @@ module HTMLProofer
       @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, @logger, @options)
+          check = Object.const_get(klass).new(src, path, html, @logger, @cache, @options)
           check.run
+
+          if klass == 'LinkCheck'
+            @internal_link_checks = check
+            check.internal_urls.each_pair do |url, internal_urls|
+              if @internal_urls_to_paths[url]
+                @internal_urls_to_paths[url].concat(internal_urls.map(&:path))
+              else
+                @internal_urls_to_paths[url] = internal_urls.map(&:path)
+              end
+            end
+            @internal_urls.merge!(check.internal_urls)
+          end
+
           external_urls = check.external_urls
           external_urls = Hash[check.external_urls.map { |url, file| [swap(url, @options[:url_swap]), file] }] if @options[:url_swap]
           result[:external_urls].merge!(external_urls)
@@ -113,16 +132,35 @@ module HTMLProofer
     end
 
     def check_path(path)
-      check_parsed create_nokogiri(path), path
+      check_parsed(create_nokogiri(path), path)
     end
 
-    def validate_urls
-      url_validator = HTMLProofer::UrlValidator.new(@logger, @external_urls, @options)
+    def validate_external_urls
+      url_validator = HTMLProofer::UrlValidator.new(@logger, @cache, @external_urls, @options)
       url_validator.before_request = @before_request
       @failures.concat(url_validator.run)
       @external_urls = url_validator.external_urls
     end
 
+    def validate_internal_urls
+      if @cache.use_cache?
+        urls_to_check = load_internal_cache
+
+        urls_to_check.each_pair do |url, internal_urls|
+          result = @internal_link_checks.check_internal_link(internal_urls.first.link, internal_urls.first.line, internal_urls.first.content)
+          code = result ? 200 : 404
+          @cache.add(url, @internal_urls_to_paths[url].sort, code, '') # TODO: blank msg for now
+        end
+        @cache.write
+      else
+        @internal_urls.values.flatten.each do |internal_url|
+          @internal_link_checks.check_internal_link(internal_url.link, internal_url.line, internal_url.content)
+        end
+      end
+
+      @failures.concat(@internal_link_checks.issues) unless @internal_urls.length.zero?
+    end
+
     def files
       @files ||= if @type == :directory
                    @src.map do |src|
@@ -191,5 +229,13 @@ module HTMLProofer
       @before_request << block if block_given?
       @before_request
     end
+
+    def load_internal_cache
+      urls_to_check = @cache.retrieve_urls(@internal_urls)
+      cache_text = pluralize(urls_to_check.count, 'internal link', 'internal links')
+      @logger.log :info, "Found #{cache_text} in the cache..."
+
+      urls_to_check
+    end
   end
 end

data/lib/html-proofer/url_validator.rb

@@ -12,13 +12,13 @@ module HTMLProofer
     attr_reader :external_urls
     attr_writer :before_request
 
-    def initialize(logger, external_urls, options)
+    def initialize(logger, cache, external_urls, options)
       @logger = logger
       @external_urls = external_urls
       @failed_tests = []
       @options = options
       @hydra = Typhoeus::Hydra.new(@options[:hydra])
-      @cache = Cache.new(@logger, @options[:cache])
+      @cache = cache
       @before_request = []
     end
 
@@ -26,7 +26,7 @@ module HTMLProofer
       @external_urls = remove_query_values
 
       if @cache.use_cache?
-        urls_to_check = load_cache
+        urls_to_check = @cache.retrieve_urls(@external_urls)
         external_link_checker(urls_to_check)
         @cache.write
       else
@@ -43,11 +43,11 @@ module HTMLProofer
       iterable_external_urls = @external_urls.dup
       @external_urls.each_key do |url|
         uri = begin
-                Addressable::URI.parse(url)
-              rescue URI::Error, Addressable::URI::InvalidURIError
-                @logger.log :error, "#{url} is an invalid URL"
-                nil
-              end
+          Addressable::URI.parse(url)
+        rescue URI::Error, Addressable::URI::InvalidURIError
+          @logger.log :error, "#{url} is an invalid URL"
+          nil
+        end
         next if uri.nil? || uri.query.nil?
 
         iterable_external_urls.delete(url) unless new_url_query_values?(uri, paths_with_queries)
@@ -74,15 +74,6 @@ module HTMLProofer
       uri.host + uri.path
     end
 
-    def load_cache
-      cache_count = @cache.size
-      cache_text = pluralize(cache_count, 'link', 'links')
-
-      @logger.log :info, "Found #{cache_text} in the cache..."
-
-      @cache.retrieve_urls(@external_urls)
-    end
-
     # Proofer runs faster if we pull out all the external URLs and run the checks
     # at the end. Otherwise, we're halting the consuming process for every file during
     # `process_files`.
@@ -111,11 +102,11 @@ module HTMLProofer
     def establish_queue(external_urls)
       external_urls.each_pair do |url, filenames|
         url = begin
-                 clean_url(url)
-              rescue URI::Error, Addressable::URI::InvalidURIError
-                add_external_issue(filenames, "#{url} is an invalid URL")
-                next
-               end
+          clean_url(url)
+        rescue URI::Error, Addressable::URI::InvalidURIError
+          add_external_issue(filenames, "#{url} is an invalid URL")
+          next
+        end
 
         method = if hash?(url) && @options[:check_external_hash]
                    :get
@@ -129,10 +120,10 @@ module HTMLProofer
     def clean_url(href)
       # catch any obvious issues, like strings in port numbers
       parsed = Addressable::URI.parse(href)
-      if href !~ /^([!#{$&}-;=?-\[\]_a-z~]|%[0-9a-fA-F]{2})+$/
-        parsed.normalize
-      else
+      if href =~ /^([!#{$&}-;=?-\[\]_a-z~]|%[0-9a-fA-F]{2})+$/
         href
+      else
+        parsed.normalize
       end
     end
 

data/lib/html-proofer/utils.rb

@@ -5,7 +5,7 @@ require 'nokogumbo'
 module HTMLProofer
   module Utils
     def pluralize(count, single, plural)
-      "#{count} #{(count == 1 ? single : plural)}"
+      "#{count} #{count == 1 ? single : plural}"
     end
 
     def create_nokogiri(path)

data/lib/html-proofer/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module HTMLProofer
-  VERSION = '3.16.0'
+  VERSION = '3.17.0'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: html-proofer
 version: !ruby/object:Gem::Version
-  version: 3.16.0
+  version: 3.17.0
 platform: ruby
 authors:
 - Garen Torikian
-autorequire: 
+autorequire:
 bindir: bin
 cert_chain: []
-date: 2020-09-11 00:00:00.000000000 Z
+date: 2020-11-14 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: addressable
@@ -293,15 +293,15 @@ homepage: https://github.com/gjtorikian/html-proofer
 licenses:
 - MIT
 metadata: {}
-post_install_message: 
+post_install_message:
 rdoc_options: []
 require_paths:
 - lib
 required_ruby_version: !ruby/object:Gem::Requirement
   requirements:
-  - - ">="
+  - - "~>"
     - !ruby/object:Gem::Version
-      version: '0'
+      version: '2.4'
 required_rubygems_version: !ruby/object:Gem::Requirement
   requirements:
   - - ">="
@@ -309,7 +309,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubygems_version: 3.1.2
-signing_key: 
+signing_key:
 specification_version: 4
 summary: A set of tests to validate your HTML output. These tests check if your image
   references are legitimate, if they have alt tags, if your internal links are working,