html-proofer 5.0.5 → 5.0.6

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 5480fbee53b67ca43a6a5896919e8d6ab82bca8e31b20e0734b60c78a71747c6
-  data.tar.gz: 12d2e42cefdab3b45b8048e7293e2bb0ccb2235fb6ac2b1722b662f1129a7202
+  metadata.gz: 96326ab8326657b7f11d5cc1baddfb82e8dcb115e4a2aa576161642ebbe9afe4
+  data.tar.gz: ce761e92fcfe407fa4711558b60a6453545cdba6b7b97f8be8da22109e451256
 SHA512:
-  metadata.gz: 6789de48b6d391cb36fd1c7b05bba18960e1782e4ef20b9cf15c076efdb147231a7b56c7c79eaad0c4c267a58fbc3355cb8a063334a7224a8901ac8a594908d8
-  data.tar.gz: 17fb7a62b96b1eb3b47dd297f8aedab8e02ffb2eacf3880d1b4ea7b0061ab4ba13dcc20993e76ba36d766d29f6ef8726823e5f73524b01eea5de96cba2455e96
+  metadata.gz: 2c66bf1334c00dacb13d2789f119cf5e0f2da723afd59e82948589e26e54e6e12bb5778b23cc758fe81a9d99d112c501e607eacc2247a3258bda60904d7ec65d
+  data.tar.gz: c81c4482233aa8fe86721de10332dc9f42ed03c4103ff0c77144a6fffc0f715e7964be2a0b935a04fbc586b4f78a2b445eebb9df5912076e910ab387add299cd

data/lib/html_proofer/attribute/url.rb

@@ -3,13 +3,16 @@
 module HTMLProofer
   class Attribute
     class Url < HTMLProofer::Attribute
-      attr_reader :url, :size
+      attr_reader :url, :size, :source, :filename
 
       REMOTE_SCHEMES = ["http", "https"].freeze
 
-      def initialize(runner, link_attribute, base_url: nil, extract_size: false)
+      def initialize(runner, link_attribute, base_url: nil, source: nil, filename: nil, extract_size: false)
         super
 
+        @source = source
+        @filename = filename
+
         if @raw_attribute.nil?
           @url = nil
         else
@@ -115,9 +118,29 @@ module HTMLProofer
       def exists?
         return true if base64?
 
-        return @runner.checked_paths[absolute_path] if @runner.checked_paths.key?(absolute_path)
+        !resolved_path.nil?
+      end
+
+      def resolved_path
+        path_to_resolve = absolute_path
+
+        return @runner.resolved_paths[path_to_resolve] if @runner.resolved_paths.key?(path_to_resolve)
+
+        # extensionless URLs
+        path_with_extension = "#{path_to_resolve}#{@runner.options[:assume_extension]}"
+        resolved = if @runner.options[:assume_extension] && File.file?(path_with_extension)
+          path_with_extension # existence checked implicitly by File.file?
+        # implicit index support
+        elsif File.directory?(path_to_resolve) && !unslashed_directory?(path_to_resolve)
+          path_with_index = File.join(path_to_resolve, @runner.options[:directory_index_file])
+          path_with_index if File.file?(path_with_index)
+        # explicit file or directory
+        elsif File.exist?(path_to_resolve)
+          path_to_resolve
+        end
+        @runner.resolved_paths[path_to_resolve] = resolved
 
-        @runner.checked_paths[absolute_path] = File.exist?(absolute_path)
+        resolved
       end
 
       def base64?
@@ -125,47 +148,23 @@ module HTMLProofer
       end
 
       def absolute_path
-        path = file_path || @runner.current_filename
+        path = full_path || @filename
 
         File.expand_path(path, Dir.pwd)
       end
 
-      def file_path
+      def full_path
         return if path.nil? || path.empty?
 
-        path_dot_ext = ""
-
-        path_dot_ext = path + @runner.options[:assume_extension] unless blank?(@runner.options[:assume_extension])
-
         base = if absolute_path?(path) # path relative to root
-          # either overwrite with root_dir; or, if source is directory, use that; or, just get the current file's dirname
-          @runner.options[:root_dir] || (File.directory?(@runner.current_source) ? @runner.current_source : File.dirname(@runner.current_source))
-        # relative links, path is a file
-        elsif File.exist?(File.expand_path(
-          path,
-          @runner.current_source,
-        )) || File.exist?(File.expand_path(path_dot_ext, @runner.current_source))
-          File.dirname(@runner.current_filename)
-        # relative links in nested dir, path is a file
-        elsif File.exist?(File.join(
-          File.dirname(@runner.current_filename),
-          path,
-        )) || File.exist?(File.join(File.dirname(@runner.current_filename), path_dot_ext))
-          File.dirname(@runner.current_filename)
-        # relative link, path is a directory
+          # either overwrite with root_dir; or, if source is directory, use that; or, just get the source file's dirname
+          @runner.options[:root_dir] || (File.directory?(@source) ? @source : File.dirname(@source))
         else
-          @runner.current_filename
-        end
-
-        file = File.join(base, path)
-
-        if @runner.options[:assume_extension] && File.file?("#{file}#{@runner.options[:assume_extension]}")
-          file = "#{file}#{@runner.options[:assume_extension]}"
-        elsif File.directory?(file) && !unslashed_directory?(file) # implicit index support
-          file = File.join(file, @runner.options[:directory_index_file])
+          # path relative to the file where the link is defined
+          File.dirname(@filename)
         end
 
-        file
+        File.join(base, path)
       end
 
       def unslashed_directory?(file)

data/lib/html_proofer/check/favicon.rb

@@ -19,16 +19,14 @@ module HTMLProofer
           if @favicon.url.protocol_relative?
             add_failure(
               "favicon link #{@favicon.url} is a protocol-relative URL, use explicit https:// instead",
-              line: @favicon.line,
-              content: @favicon.content,
+              element: @favicon,
             )
           elsif @favicon.url.remote?
             add_to_external_urls(@favicon.url, @favicon.line)
           elsif !@favicon.url.exists?
             add_failure(
               "internal favicon #{@favicon.url.raw_attribute} does not exist",
-              line: @favicon.line,
-              content: @favicon.content,
+              element: @favicon,
             )
           end
         else

data/lib/html_proofer/check/images.rb

@@ -14,41 +14,37 @@ module HTMLProofer
           # screenshot filenames should return because of terrible names
           add_failure(
             "image has a terrible filename (#{@img.url.raw_attribute})",
-            line: @img.line,
-            content: @img.content,
+            element: @img,
           ) if terrible_filename?
 
           # does the image exist?
           if missing_src?
-            add_failure("image has no src or srcset attribute", line: @img.line, content: @img.content)
+            add_failure("image has no src or srcset attribute", element: @img)
           elsif @img.url.protocol_relative?
             add_failure(
               "image link #{@img.url} is a protocol-relative URL, use explicit https:// instead",
-              line: @img.line,
-              content: @img.content,
+              element: @img,
             )
           elsif @img.url.remote?
             add_to_external_urls(@img.url, @img.line)
           elsif !@img.url.exists? && !@img.multiple_srcsets? && !@img.multiple_sizes?
             add_failure(
               "internal image #{@img.url.raw_attribute} does not exist",
-              line: @img.line,
-              content: @img.content,
+              element: @img,
             )
           elsif @img.multiple_srcsets? || @img.multiple_sizes?
             @img.srcsets_wo_sizes.each do |srcset|
-              srcset_url = HTMLProofer::Attribute::Url.new(@runner, srcset, base_url: @img.base_url, extract_size: true)
+              srcset_url = HTMLProofer::Attribute::Url.new(@runner, srcset, base_url: @img.base_url, source: @img.url.source, filename: @img.url.filename, extract_size: true)
 
               if srcset_url.protocol_relative?
                 add_failure(
                   "image link #{srcset_url.url} is a protocol-relative URL, use explicit https:// instead",
-                  line: @img.line,
-                  content: @img.content,
+                  element: @img,
                 )
               elsif srcset_url.remote?
                 add_to_external_urls(srcset_url.url, @img.line)
               elsif !srcset_url.exists?
-                add_failure("internal image #{srcset} does not exist", line: @img.line, content: @img.content)
+                add_failure("internal image #{srcset} does not exist", element: @img)
               end
             end
           end
@@ -58,22 +54,19 @@ module HTMLProofer
             if missing_alt_tag? && !ignore_missing_alt?
               add_failure(
                 "image #{@img.url.raw_attribute} does not have an alt attribute",
-                line: @img.line,
-                content: @img.content,
+                element: @img,
               )
             elsif (empty_alt_tag? || alt_all_spaces?) && !ignore_empty_alt?
               add_failure(
                 "image #{@img.url.raw_attribute} has an alt attribute, but no content",
-                line: @img.line,
-                content: @img.content,
+                element: @img,
               )
             end
           end
 
           add_failure(
             "image #{@img.url.raw_attribute} uses the http scheme",
-            line: @img.line,
-            content: @img.content,
+            element: @img,
           ) if @runner.enforce_https? && @img.url.http?
         end
 

data/lib/html_proofer/check/links.rb

@@ -10,7 +10,7 @@ module HTMLProofer
           next if @link.ignore?
 
           if !allow_hash_href? && @link.node["href"] == "#"
-            add_failure("linking to internal hash #, which points to nowhere", line: @link.line, content: @link.content)
+            add_failure("linking to internal hash #, which points to nowhere", element: @link)
             next
           end
 
@@ -18,21 +18,20 @@ module HTMLProofer
           if blank?(@link.url.raw_attribute)
             next if allow_missing_href?
 
-            add_failure("'#{@link.node.name}' tag is missing a reference", line: @link.line, content: @link.content)
+            add_failure("'#{@link.node.name}' tag is missing a reference", element: @link)
             next
           end
 
           # is it even a valid URL?
           unless @link.url.valid?
-            add_failure("#{@link.href} is an invalid URL", line: @link.line, content: @link.content)
+            add_failure("#{@link.href} is an invalid URL", element: @link)
             next
           end
 
           if @link.url.protocol_relative?
             add_failure(
               "#{@link.url} is a protocol-relative URL, use explicit https:// instead",
-              line: @link.line,
-              content: @link.content,
+              element: @link,
             )
             next
           end
@@ -50,7 +49,7 @@ module HTMLProofer
             next if @link.node["rel"] == "dns-prefetch"
 
             unless @link.url.path?
-              add_failure("#{@link.url.raw_attribute} is an invalid URL", line: @link.line, content: @link.content)
+              add_failure("#{@link.url.raw_attribute} is an invalid URL", element: @link)
               next
             end
 
@@ -60,8 +59,7 @@ module HTMLProofer
             if @link.url.unslashed_directory?(@link.url.absolute_path)
               add_failure(
                 "internally linking to a directory #{@link.url.raw_attribute} without trailing slash",
-                line: @link.line,
-                content: @link.content,
+                element: @link,
               )
               next
             end
@@ -88,7 +86,7 @@ module HTMLProofer
         when "http"
           return unless @runner.options[:enforce_https]
 
-          add_failure("#{@link.url.raw_attribute} is not an HTTPS link", line: @link.line, content: @link.content)
+          add_failure("#{@link.url.raw_attribute} is not an HTTPS link", element: @link)
         end
       end
 
@@ -96,14 +94,12 @@ module HTMLProofer
         if @link.url.path.empty?
           add_failure(
             "#{@link.url.raw_attribute} contains no email address",
-            line: @link.line,
-            content: @link.content,
+            element: @link,
           ) unless ignore_empty_mailto?
         elsif !/#{URI::MailTo::EMAIL_REGEXP}/o.match?(@link.url.path)
           add_failure(
             "#{@link.url.raw_attribute} contains an invalid email address",
-            line: @link.line,
-            content: @link.content,
+            element: @link,
           )
         end
       end
@@ -111,8 +107,7 @@ module HTMLProofer
       def handle_tel
         add_failure(
           "#{@link.url.raw_attribute} contains no phone number",
-          line: @link.line,
-          content: @link.content,
+          element: @link,
         ) if @link.url.path.empty?
       end
 
@@ -130,16 +125,14 @@ module HTMLProofer
         if blank?(@link.node["integrity"]) && blank?(@link.node["crossorigin"])
           add_failure(
             "SRI and CORS not provided in: #{@link.url.raw_attribute}",
-            line: @link.line,
-            content: @link.content,
+            element: @link,
           )
         elsif blank?(@link.node["integrity"])
-          add_failure("Integrity is missing in: #{@link.url.raw_attribute}", line: @link.line, content: @link.content)
+          add_failure("Integrity is missing in: #{@link.url.raw_attribute}", element: @link)
         elsif blank?(@link.node["crossorigin"])
           add_failure(
             "CORS not provided for external resource in: #{@link.link.url.raw_attribute}",
-            line: @link.line,
-            content: @link.content,
+            element: @link,
           )
         end
       end

data/lib/html_proofer/check/open_graph.rb

@@ -11,24 +11,22 @@ module HTMLProofer
 
           # does the open_graph exist?
           if missing_content?
-            add_failure("open graph has no content attribute", line: @open_graph.line, content: @open_graph.content)
+            add_failure("open graph has no content attribute", element: @open_graph)
           elsif empty_content?
-            add_failure("open graph content attribute is empty", line: @open_graph.line, content: @open_graph.content)
+            add_failure("open graph content attribute is empty", element: @open_graph)
           elsif !@open_graph.url.valid?
-            add_failure("#{@open_graph.src} is an invalid URL", line: @open_graph.line)
+            add_failure("#{@open_graph.src} is an invalid URL", element: @open_graph)
           elsif @open_graph.url.protocol_relative?
             add_failure(
               "open graph link #{@open_graph.url} is a protocol-relative URL, use explicit https:// instead",
-              line: @open_graph.line,
-              content: @open_graph.content,
+              element: @open_graph,
             )
           elsif @open_graph.url.remote?
             add_to_external_urls(@open_graph.url, @open_graph.line)
           else
             add_failure(
               "internal open graph #{@open_graph.url.raw_attribute} does not exist",
-              line: @open_graph.line,
-              content: @open_graph.content,
+              element: @open_graph,
             ) unless @open_graph.url.exists?
           end
         end

data/lib/html_proofer/check/scripts.rb

@@ -12,12 +12,11 @@ module HTMLProofer
 
           # does the script exist?
           if missing_src?
-            add_failure("script is empty and has no src attribute", line: @script.line, content: @script.content)
+            add_failure("script is empty and has no src attribute", element: @script)
           elsif @script.url.protocol_relative?
             add_failure(
               "script link #{@script.url} is a protocol-relative URL, use explicit https:// instead",
-              line: @script.line,
-              content: @script.content,
+              element: @script,
             )
           elsif @script.url.remote?
             add_to_external_urls(@script.url, @script.line)
@@ -25,8 +24,7 @@ module HTMLProofer
           elsif !@script.url.exists?
             add_failure(
               "internal script reference #{@script.src} does not exist",
-              line: @script.line,
-              content: @script.content,
+              element: @script,
             )
           end
         end
@@ -42,20 +40,17 @@ module HTMLProofer
         if blank?(@script.node["integrity"]) && blank?(@script.node["crossorigin"])
           add_failure(
             "SRI and CORS not provided in: #{@script.url.raw_attribute}",
-            line: @script.line,
-            content: @script.content,
+            element: @script,
           )
         elsif blank?(@script.node["integrity"])
           add_failure(
             "Integrity is missing in: #{@script.url.raw_attribute}",
-            line: @script.line,
-            content: @script.content,
+            element: @script,
           )
         elsif blank?(@script.node["crossorigin"])
           add_failure(
             "CORS not provided for external resource in: #{@script.url.raw_attribute}",
-            line: @script.line,
-            content: @script.content,
+            element: @script,
           )
         end
       end

data/lib/html_proofer/check.rb

@@ -24,14 +24,14 @@ module HTMLProofer
       raise NotImplementedError, "HTMLProofer::Check subclasses must implement #run"
     end
 
-    def add_failure(description, line: nil, status: nil, content: nil)
+    def add_failure(description, element: nil, line: nil, status: nil, content: nil)
       @failures << Failure.new(
         @runner.current_filename,
         short_name,
         description,
-        line: line,
+        line: element.nil? ? line : element.line,
         status: status,
-        content: content,
+        content: element.nil? ? content : element.content,
       )
     end
 
@@ -45,8 +45,8 @@ module HTMLProofer
       @internal_urls[url_string] = [] if @internal_urls[url_string].nil?
 
       metadata = {
-        source: @runner.current_source,
-        filename: @runner.current_filename,
+        source: url.source,
+        filename: url.filename,
         line: line,
         base_url: base_url,
         found: false,
@@ -59,7 +59,7 @@ module HTMLProofer
 
       @external_urls[url_string] = [] if @external_urls[url_string].nil?
 
-      @external_urls[url_string] << { filename: @runner.current_filename, line: line }
+      @external_urls[url_string] << { filename: url.filename, line: line }
     end
 
     class << self

data/lib/html_proofer/element.rb

@@ -16,7 +16,7 @@ module HTMLProofer
       swap_attributes!
 
       @base_url = base_url
-      @url = Attribute::Url.new(runner, link_attribute, base_url: base_url)
+      @url = Attribute::Url.new(runner, link_attribute, base_url: base_url, source: @runner.current_source, filename: @runner.current_filename)
 
       @line = node.line
       @content = node.content

data/lib/html_proofer/runner.rb

@@ -6,8 +6,8 @@ module HTMLProofer
   class Runner
     include HTMLProofer::Utils
 
-    attr_reader :options, :cache, :logger, :internal_urls, :external_urls, :checked_paths, :current_check
-    attr_accessor :current_filename, :current_source, :reporter
+    attr_reader :options, :cache, :logger, :internal_urls, :external_urls, :resolved_paths, :current_check, :current_filename, :current_source
+    attr_accessor :reporter
 
     URL_TYPES = [:external, :internal].freeze
 
@@ -26,7 +26,7 @@ module HTMLProofer
 
       @before_request = []
 
-      @checked_paths = {}
+      @resolved_paths = {}
 
       @current_check = nil
       @current_source = nil

data/lib/html_proofer/url_validator/internal.rb

@@ -29,15 +29,11 @@ module HTMLProofer
           matched_count_to_log = pluralize(matched_files.count, "reference", "references")
           @logger.log(:debug, "(#{i + 1} / #{links.count}) Internal link #{link}: Checking #{matched_count_to_log}")
           matched_files.each do |metadata|
-            url = HTMLProofer::Attribute::Url.new(@runner, link, base_url: metadata[:base_url])
+            url = HTMLProofer::Attribute::Url.new(@runner, link, base_url: metadata[:base_url], source: metadata[:source], filename: metadata[:filename])
 
-            @runner.current_source = metadata[:source]
-            @runner.current_filename = metadata[:filename]
-
-            target_file_path = url.absolute_path
-            unless file_exists?(target_file_path)
+            unless url.exists?
               @failed_checks << Failure.new(
-                @runner.current_filename,
+                metadata[:filename],
                 "Links > Internal",
                 "internally linking to #{url}, which does not exist",
                 line: metadata[:line],
@@ -51,6 +47,7 @@ module HTMLProofer
             hash_exists = hash_exists_for_url?(url)
             if hash_exists.nil?
               # the hash needs to be checked in the target file, we collect the url and metadata
+              target_file_path = url.resolved_path
               unless file_paths_hashes_to_check.key?(target_file_path)
                 file_paths_hashes_to_check[target_file_path] = {}
               end
@@ -62,7 +59,7 @@ module HTMLProofer
             end
             unless hash_exists
               @failed_checks << Failure.new(
-                @runner.current_filename,
+                metadata[:filename],
                 "Links > Internal",
                 "internally linking to #{url}; the file exists, but the hash '#{url.hash}' does not",
                 line: metadata[:line],
@@ -109,12 +106,6 @@ module HTMLProofer
         @failed_checks
       end
 
-      private def file_exists?(absolute_path)
-        return @runner.checked_paths[absolute_path] if @runner.checked_paths.key?(absolute_path)
-
-        @runner.checked_paths[absolute_path] = File.exist?(absolute_path)
-      end
-
       # verify the hash w/o just based on the URL, w/o looking at the target file
       # => returns nil if the has could not be verified
       private def hash_exists_for_url?(url)

data/lib/html_proofer/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module HTMLProofer
-  VERSION = "5.0.5"
+  VERSION = "5.0.6"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: html-proofer
 version: !ruby/object:Gem::Version
-  version: 5.0.5
+  version: 5.0.6
 platform: ruby
 authors:
 - Garen Torikian
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2023-03-07 00:00:00.000000000 Z
+date: 2023-03-15 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: addressable