html-proofer 3.18.6 → 3.18.7

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: e2d6deb85601ff3f5e5df02999f1b55f29231e5658f1d4fa9986c002b3cc0263
-  data.tar.gz: e20a007144a866b41b97eb36a964ca1372d693ade77acc69a9c9e706b353c8ef
+  metadata.gz: 406cf6b61f0416a7f9f6874cd43e855288bc047ff7e2ec8ba374acdc57c1898b
+  data.tar.gz: fc44f71f75c1b830182d2ef60aed3abcc816ee7671cb7c4d74c174e75edf30be
 SHA512:
-  metadata.gz: c51154450eb20c04c0b7038effb12ab8d258457ea5560421b5a2a0739342504afebb8fd092fe4378838a076636ff2665eb691f513b8b34b49af2e89ffba988c2
-  data.tar.gz: fcb7bf492588fabc991caf1f779fca14a7b553d3ce00dbf55ba24aebf3c39f7613af6a408d94c3a3faaeb21dceb8265ece91ca9bf4320aee7d4f3dca133ba4de
+  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)
@@ -91,19 +99,20 @@ 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
@@ -113,21 +122,22 @@ module HTMLProofer
     # caches need access to this file. Write a proper versioned
     # schema in the future
     def write
-      file = {}
-      file = JSON.parse(File.read(cache_file)) if File.exist?(cache_file)
-      File.write(cache_file, file.merge(@cache_log).to_json)
+      File.write(cache_file, @cache_log.to_json)
     end
 
     def load?
       @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
@@ -154,9 +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
 
@@ -174,5 +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/runner.rb

@@ -238,7 +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

data/lib/html-proofer/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module HTMLProofer
-  VERSION = '3.18.6'
+  VERSION = '3.18.7'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: html-proofer
 version: !ruby/object:Gem::Version
-  version: 3.18.6
+  version: 3.18.7
 platform: ruby
 authors:
 - Garen Torikian
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2021-02-21 00:00:00.000000000 Z
+date: 2021-03-04 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: addressable