html-proofer 3.18.2 → 3.18.7

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 852efb42767ef8589aa2f469f06c807ab287d78da04043e5a57e4d4a7f8bf6dd
-  data.tar.gz: 065406c8f1a7978d025224dc4ef53137e698eaeec8f097c1375620c8d6e0c5ba
+  metadata.gz: 406cf6b61f0416a7f9f6874cd43e855288bc047ff7e2ec8ba374acdc57c1898b
+  data.tar.gz: fc44f71f75c1b830182d2ef60aed3abcc816ee7671cb7c4d74c174e75edf30be
 SHA512:
-  metadata.gz: c5aabdd3e5087dd1282b4428b7105e62216b22510c1a73a9f80f1e6b2ba18906217c7aaa974b245644de2b325bef8fd3e05c61c528302fb59ce5a353da0c1d04
-  data.tar.gz: 7155e2eed2e31173ad7e8efe4d7ce31b2cf18fa403b560f03428e3ef8ff8f9c60654f9e1ee6a4470ff65f609fdaeeb107e0236c85610212e90b7a9535a650522
+  metadata.gz: 47501eb3aa242423f2e748ed52d3fa5ed8c9ae35c839602d27de0b1712a0c1e22d8cfe20781569d469439ecc75bc3f853100f6aae63ee1c3f6174641cc102f7e
+  data.tar.gz: 30348fc2649b3210038813bf05e24f10d0ca775e1253a2ba04c79c282b1822bc3ed89e45c0652e44889b9bf82a6022fdc25db70407425c144b15875fe0c53ad0

data/lib/html-proofer/cache.rb

@@ -11,6 +11,8 @@ module HTMLProofer
     DEFAULT_STORAGE_DIR = File.join('tmp', '.htmlproofer')
     DEFAULT_CACHE_FILE_NAME = 'cache.log'
 
+    URI_REGEXP = URI::DEFAULT_PARSER.make_regexp
+
     attr_reader :exists, :cache_log, :storage_dir, :cache_file
 
     def initialize(logger, options)
@@ -30,6 +32,8 @@ module HTMLProofer
     end
 
     def within_timeframe?(time)
+      return false if time.nil?
+
       (@parsed_timeframe..@cache_time).cover?(Time.parse(time))
     end
 
@@ -71,10 +75,14 @@ module HTMLProofer
       @cache_log[clean_url(url)] = data
     end
 
-    def detect_url_changes(found)
-      existing_urls = @cache_log.keys.map { |url| clean_url(url) }
+    def detect_url_changes(found, type)
       found_urls = found.keys.map { |url| clean_url(url) }
 
+      # if there were no urls, bail
+      return {} if found_urls.empty?
+
+      existing_urls = @cache_log.keys.map { |url| clean_url(url) }
+
       # prepare to add new URLs detected
       additions = found.reject do |url, _|
         url = clean_url(url)
@@ -91,24 +99,28 @@ module HTMLProofer
       @logger.log :info, "Adding #{new_link_text} to the cache..."
 
       # remove from cache URLs that no longer exist
-      del = 0
+      deletions = 0
       @cache_log.delete_if do |url, _|
         url = clean_url(url)
+
         if found_urls.include?(url)
           false
-        else
+        elsif url_matches_type?(url, type)
           @logger.log :debug, "Removing #{url} from cache check"
-          del += 1
+          deletions += 1
           true
         end
       end
 
-      del_link_text = pluralize(del, 'link', 'links')
+      del_link_text = pluralize(deletions, 'link', 'links')
       @logger.log :info, "Removing #{del_link_text} from the cache..."
 
       additions
     end
 
+    # TODO: Garbage performance--both the external and internal
+    # caches need access to this file. Write a proper versioned
+    # schema in the future
     def write
       File.write(cache_file, @cache_log.to_json)
     end
@@ -117,12 +129,15 @@ module HTMLProofer
       @load.nil?
     end
 
-    def retrieve_urls(urls)
-      urls_to_check = detect_url_changes(urls)
+    def retrieve_urls(urls, type)
+      urls_to_check = detect_url_changes(urls, type)
+
       @cache_log.each_pair do |url, cache|
         next if within_timeframe?(cache['time']) && cache['message'].empty? # these were successes to skip
 
-        urls_to_check[url] = cache['filenames'] # recheck expired links
+        if url_matches_type?(url, type)
+          urls_to_check[url] = cache['filenames'] # recheck expired links
+        end
       end
       urls_to_check
     end
@@ -149,9 +164,9 @@ module HTMLProofer
 
       @cache_file = File.join(storage_dir, cache_file_name)
 
-      return unless File.exist?(cache_file)
+      return unless File.exist?(@cache_file)
 
-      contents = File.read(cache_file)
+      contents = File.read(@cache_file)
       @cache_log = contents.empty? ? {} : JSON.parse(contents)
     end
 
@@ -169,5 +184,10 @@ module HTMLProofer
         @cache_datetime - Rational(measurement / 24.0)
       end.to_time
     end
+
+    def url_matches_type?(url, type)
+      return true if type == :internal && url !~ URI_REGEXP
+      return true if type == :external && url =~ URI_REGEXP
+    end
   end
 end

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

@@ -51,14 +51,16 @@ 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'
 
+        unless @link.path?
+          add_issue("#{@link.href} is an invalid URL", line: line, content: content)
+          next
+        end
+
         add_to_external_urls(@link.href || @link.src)
         next
       elsif @link.internal?
-        if @link.exists? || @link.hash
-          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
+        add_to_internal_urls(@link.href, InternalLink.new(@link, @path, line, content))
+        add_issue("internally linking to #{@link.href}, which does not exist", line: line, content: content) if !@link.exists? && !@link.hash
       end
     end
 

data/lib/html-proofer/element.rb

@@ -24,7 +24,7 @@ module HTMLProofer
         raise e
       end
 
-      @aria_hidden = defined?(@aria_hidden) && @aria_hidden == 'true' ? true : false
+      @aria_hidden = defined?(@aria_hidden) && @aria_hidden == 'true'
 
       @data_proofer_ignore = defined?(@data_proofer_ignore)
 
@@ -74,6 +74,10 @@ module HTMLProofer
       !parts.nil?
     end
 
+    def path?
+      !parts.host.nil? && !parts.path.nil?
+    end
+
     def parts
       @parts ||= Addressable::URI.parse url
     rescue URI::Error, Addressable::URI::InvalidURIError
@@ -216,6 +220,8 @@ module HTMLProofer
     end
 
     def ignores_pattern_check(links)
+      return false unless links.is_a?(Array)
+
       links.each do |ignore|
         case ignore
         when String

data/lib/html-proofer/middleware.rb

@@ -22,7 +22,7 @@ module HTMLProofer
         allow_hash_href: true,
         check_external_hash: true,
         check_html: true,
-        url_ignore: [/.*/], # Don't try to check if local files exist
+        url_ignore: [%r{^/}], # Don't try to check if local files exist
         validation: { report_eof_tags: true }
       }
     end

data/lib/html-proofer/runner.rb

@@ -63,7 +63,9 @@ module HTMLProofer
           swap(url, @options[:url_swap])
         end
       end
-      @external_urls = Hash[*@src.map { |s| [s, nil] }.flatten]
+      @external_urls = @src.each_with_object({}) do |url, hash|
+        hash[url] = nil
+      end
       validate_external_urls
     end
 
@@ -123,7 +125,7 @@ module HTMLProofer
           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]
+          external_urls = check.external_urls.map { |url, file| [swap(url, @options[:url_swap]), file] }.to_h if @options[:url_swap]
           result[:external_urls].merge!(external_urls)
           result[:failures].concat(check.issues)
         end
@@ -236,7 +238,7 @@ module HTMLProofer
     end
 
     def load_internal_cache
-      urls_to_check = @cache.retrieve_urls(@internal_urls)
+      urls_to_check = @cache.retrieve_urls(@internal_urls, :internal)
       cache_text = pluralize(urls_to_check.count, 'internal link', 'internal links')
       @logger.log :info, "Found #{cache_text} in the cache..."
 

data/lib/html-proofer/url_validator.rb

@@ -26,7 +26,7 @@ module HTMLProofer
       @external_urls = remove_query_values
 
       if @cache.use_cache?
-        urls_to_check = @cache.retrieve_urls(@external_urls)
+        urls_to_check = @cache.retrieve_urls(@external_urls, :external)
         external_link_checker(urls_to_check)
         @cache.write
       else
@@ -85,7 +85,7 @@ module HTMLProofer
     # for HEAD. If we've decided to check for hashes, we must do a GET--HEAD is
     # not available as an option.
     def external_link_checker(external_urls)
-      external_urls = Hash[external_urls.sort]
+      external_urls = external_urls.sort.to_h
 
       count = external_urls.length
       check_text = pluralize(count, 'external link', 'external links')

data/lib/html-proofer/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module HTMLProofer
-  VERSION = '3.18.2'
+  VERSION = '3.18.7'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: html-proofer
 version: !ruby/object:Gem::Version
-  version: 3.18.2
+  version: 3.18.7
 platform: ruby
 authors:
 - Garen Torikian
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2020-12-17 00:00:00.000000000 Z
+date: 2021-03-04 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: addressable
@@ -285,9 +285,12 @@ require_paths:
 - lib
 required_ruby_version: !ruby/object:Gem::Requirement
   requirements:
-  - - "~>"
+  - - ">="
+    - !ruby/object:Gem::Version
+      version: 2.4.10
+  - - "<"
     - !ruby/object:Gem::Version
-      version: '2.4'
+      version: '4.0'
 required_rubygems_version: !ruby/object:Gem::Requirement
   requirements:
   - - ">="