html-proofer 3.18.3 → 3.18.8

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: a7d5f07fea9b72a1731abbdd7066c4d7e5807daf4d622f45b051a240216dac41
-  data.tar.gz: 6cbe2fa0408ba39f759e6fb3a93902d73009ccdab1b083a41a29660855539564
+  metadata.gz: d705eda2b6ad497d076f1dfa565e5c98e0769b9cd83ddee56ab04b8d4ad6722f
+  data.tar.gz: 74d2ca4e6b65e8f1d1467aef9cae90a4804cb285eecf8a5014e8b77c579ef25b
 SHA512:
-  metadata.gz: de2d7e2dff071f8b36e92e7bbfd913308aa8c7b1e4f2d58ee798d5598b64c9167817d55e3091091752858976d1fad5f0df9d62858596add03ff4233da041f476
-  data.tar.gz: 6da898dd1b1ac623df5ef637049bea96872860a51fb486e5637cc59db19e9c609826d0846b89af67eb447fa20d9ad2584e4a5d8671cc0e3c4ce9ce327b55a785
+  metadata.gz: b51003f4125230780bc0335f72148e7b4c964aba772b16cdcfb5bcad941eea7eb0ce428cd8b0d78ebcf08fe2e049b4850fbd4e85f90e57b56b1b854bfb516408
+  data.tar.gz: 2bc6deafa255f91f210bdfc0df14c8aac36f16bff90e2402b4a0657a1ed68ad7c60b4717bdceaff4a819764104a05da6c95e049c83cb83d3f518e83a6638d14b

data/lib/html-proofer/cache.rb

@@ -3,6 +3,7 @@
 require_relative 'utils'
 require 'date'
 require 'json'
+require 'uri'
 
 module HTMLProofer
   class Cache
@@ -11,6 +12,8 @@ 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 +33,8 @@ module HTMLProofer
     end
 
     def within_timeframe?(time)
+      return false if time.nil?
+
       (@parsed_timeframe..@cache_time).cover?(Time.parse(time))
     end
 
@@ -71,10 +76,14 @@ 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 +100,28 @@ 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 +130,15 @@ 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 +165,9 @@ 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 +185,10 @@ 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 = '3.18.3'
+  VERSION = '3.18.8'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: html-proofer
 version: !ruby/object:Gem::Version
-  version: 3.18.3
+  version: 3.18.8
 platform: ruby
 authors:
 - Garen Torikian
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2020-12-29 00:00:00.000000000 Z
+date: 2021-03-04 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: addressable