Handle internal & external cache merging

lib/html-proofer/cache.rb

@@ -109,8 +109,13 @@ def detect_url_changes(found)
       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)
+      file = {}
+      file = JSON.parse(File.read(cache_file)) if File.exist?(cache_file)
+      File.write(cache_file, file.merge(@cache_log).to_json)
     end
 
     def load?

lib/html-proofer/check/links.rb

@@ -59,11 +59,8 @@ def run
         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
 

lib/html-proofer/element.rb

@@ -24,7 +24,7 @@ def initialize(obj, check, logger)
         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)
 

lib/html-proofer/runner.rb

@@ -63,7 +63,9 @@ def check_list_of_links
           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 @@ def check_parsed(html, path)
           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

lib/html-proofer/url_validator.rb

@@ -85,7 +85,7 @@ def extract_domain_path(uri)
     # 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')

spec/html-proofer/cache_spec.rb

@@ -247,4 +247,58 @@ def read_cache(cache_file)
       Timecop.return
     end
   end
+
+  it 'does recheck failures, regardless of external-only cache' do
+    cache_file_name = '.external.log'
+    cache_location = File.join(storage_dir, cache_file_name)
+
+    file = "#{FIXTURES_DIR}/cache/external_example.html"
+
+    new_time = Time.local(2021, 0o1, 27, 12, 0, 0)
+    Timecop.freeze(new_time)
+
+    File.delete(cache_location) if File.exist?(cache_location)
+
+    run_proofer(file, :file, external_only: true, links_only: true, cache: { timeframe: '1d', cache_file: cache_file_name }.merge(default_cache_options))
+
+    cache = JSON.parse(File.read(cache_location))
+
+    expect(cache.keys.count).to eq(1)
+    expect(cache.keys[0]).to eq('https://github.com/gjtorikian/html-proofer')
+
+    run_proofer(file, :file, links_only: true, cache: { timeframe: '1d', cache_file: cache_file_name }.merge(default_cache_options))
+
+    cache = JSON.parse(File.read(cache_location))
+    expect(cache.keys.count).to eq(1)
+    expect(cache.keys[0]).to eq('https://github.com/gjtorikian/html-proofer')
+
+    Timecop.return
+  end
+
+  it 'does recheck failures, regardless of external and internal cache' do
+    cache_file_name = '.internal_and_external.log'
+    cache_location = File.join(storage_dir, cache_file_name)
+
+    file = "#{FIXTURES_DIR}/cache/internal_and_external_example.html"
+
+    new_time = Time.local(2021, 0o1, 27, 12, 0, 0)
+    Timecop.freeze(new_time)
+
+    File.delete(cache_location) if File.exist?(cache_location)
+
+    run_proofer(file, :file, external_only: true, links_only: true, cache: { timeframe: '1d', cache_file: cache_file_name }.merge(default_cache_options))
+
+    cache = JSON.parse(File.read(cache_location))
+
+    expect(cache.keys.count).to eq(1)
+    expect(cache.keys[0]).to eq('https://github.com/gjtorikian/html-proofer')
+
+    run_proofer(file, :file, links_only: true, cache: { timeframe: '1d', cache_file: cache_file_name }.merge(default_cache_options))
+
+    cache = JSON.parse(File.read(cache_location))
+    expect(cache.keys.count).to eq(2)
+    expect(cache.keys[0]).to eq('https://github.com/gjtorikian/html-proofer')
+
+    Timecop.return
+  end
 end

spec/html-proofer/fixtures/cache/.external.log

@@ -0,0 +1 @@
+{"https://github.com/gjtorikian/html-proofer":{"time":"2021-01-27 12:00:00 +0000","filenames":["spec/html-proofer/fixtures/cache/external_example.html"],"status":200,"message":""}}

spec/html-proofer/fixtures/cache/.internal_and_external.log

@@ -0,0 +1 @@
+{"https://github.com/gjtorikian/html-proofer":{"time":"2021-01-27 12:00:00 +0000","filenames":["spec/html-proofer/fixtures/cache/internal_and_external_example.html"],"status":200,"message":""},"/somewhere.html":{"time":"2021-01-27 12:00:00 +0000","filenames":["spec/html-proofer/fixtures/cache/internal_and_external_example.html"],"status":200,"message":""}}

spec/html-proofer/fixtures/cache/external_example.html

@@ -0,0 +1,11 @@
+<!DOCTYPE html>
+<html>
+  <head>
+    <title>Example</title>
+  </head>
+  <body>
+    <a href="https://github.com/gjtorikian/html-proofer"
+      >The amazing HTMLProofer</a
+    >
+  </body>
+</html>

spec/html-proofer/fixtures/cache/internal_and_external_example.html

@@ -0,0 +1,13 @@
+<!DOCTYPE html>
+<html>
+  <head>
+    <title>Example</title>
+  </head>
+  <body>
+    <a href="https://github.com/gjtorikian/html-proofer"
+      >The amazing HTMLProofer</a
+    >
+
+    <a href="/somewhere.html">Somewhere</a>
+  </body>
+</html>