Remove references to keywords

app/models/episode.rb

@@ -21,8 +21,6 @@ class Episode < ApplicationRecord
 
   acts_as_paranoid
 
-  serialize :categories, JSON
-  serialize :keywords, JSON
   serialize :overrides, HashSerializer
 
   belongs_to :podcast, -> { with_deleted }, touch: true
@@ -224,15 +222,11 @@ def overrides
   end
 
   def categories
-    self[:categories] ||= []
+    self[:categories] || []
   end
 
   def categories=(cats)
-    self[:categories] = Array(cats).reject(&:blank?)
-  end
-
-  def keywords
-    self[:keywords] ||= []
+    self[:categories] = sanitize_keywords(cats, false).presence
   end
 
   def copy_media(force = false)
@@ -295,26 +289,18 @@ def set_external_keyword
 
     identifiers = []
     %i[published_at guid].each do |attr|
-      identifiers << sanitize_keyword(send(attr), 10)
+      identifiers << sanitize_keyword(send(attr), 10, true)
     end
-    identifiers << sanitize_keyword(title || "undefined", 20)
+    identifiers << sanitize_keyword(title || "undefined", 20, true)
     self.keyword_xid = identifiers.join("_")
   end
 
-  def sanitize_keyword(kw, length)
-    kw.to_s.downcase.gsub(/[^ a-z0-9_-]/, "").gsub(/\s+/, " ").strip.slice(
-      0,
-      length
-    )
-  end
-
   def sanitize_text
     self.description = sanitize_white_list(description) if description_changed?
     self.content = sanitize_white_list(content) if content_changed?
     self.subtitle = sanitize_text_only(subtitle) if subtitle_changed?
     self.summary = sanitize_links_only(summary) if summary_changed?
     self.title = sanitize_text_only(title) if title_changed?
-    self.keywords = keywords.map { |kw| sanitize_keyword(kw, kw.length) }
   end
 
   def description_with_default

app/models/episode_entry_handler.rb

@@ -3,7 +3,7 @@
 class EpisodeEntryHandler
   ENTRY_ATTRIBUTES = %w[author block categories content description explicit
     feedburner_orig_enclosure_link feedburner_orig_link is_closed_captioned
-    is_perma_link keywords position subtitle summary title url
+    is_perma_link position subtitle summary title url
     season_number episode_number clean_title].freeze
 
   attr_accessor :episode

app/models/feed.rb

@@ -156,16 +156,6 @@ def feed_episodes
     include_in_feed
   end
 
-  def episode_categories_include?(ep, match_tags)
-    tags = match_tags.map { |cat| normalize_category(cat) }
-    cats = (ep || []).categories.map { |cat| normalize_category(cat) }
-    (tags & cats).length > 0
-  end
-
-  def normalize_category(cat)
-    cat.to_s.downcase.gsub(/[^ a-z0-9_-]/, "").gsub(/\s+/, " ").strip
-  end
-
   def publish_to_apple?
     !!apple_config&.publish_to_apple?
   end
@@ -180,29 +170,20 @@ def exclude_tags=(tags)
     self[:exclude_tags] = tags.blank? ? nil : tags
   end
 
-  def use_include_tags?
-    !include_tags.blank?
-  end
-
-  def use_exclude_tags?
-    !exclude_tags.blank?
-  end
-
   def filtered_episodes
     eps = podcast.episodes.published_by(episode_offset_seconds.to_i)
 
-    eps =
-      if use_include_tags?
-        eps.select { |ep| episode_categories_include?(ep, include_tags) }
-      else
-        eps
-      end
+    if include_tags.present?
+      kws = sanitize_keywords(include_tags, true)
+      eps = eps.select { |e| (sanitize_keywords(e.categories, true) & kws).any? }
+    end
 
-    if use_exclude_tags?
-      eps.reject { |ep| episode_categories_include?(ep, exclude_tags) }
-    else
-      eps
+    if exclude_tags.present?
+      kws = sanitize_keywords(exclude_tags, true)
+      eps = eps.reject { |e| (sanitize_keywords(e.categories, true) & kws).any? }
     end
+
+    eps
   end
 
   def enclosure_template

app/models/podcast.rb

@@ -15,8 +15,6 @@ class Podcast < ApplicationRecord
 
   acts_as_paranoid
 
-  serialize :categories, JSON
-  serialize :keywords, JSON
   serialize :restrictions, JSON
 
   has_one :default_feed, -> { default }, class_name: "Feed", validate: true, autosave: true, inverse_of: :podcast
@@ -155,6 +153,14 @@ def managing_editor
     "#{managing_editor_email} (#{managing_editor_name})"
   end
 
+  def categories
+    self[:categories] || []
+  end
+
+  def categories=(cats)
+    self[:categories] = sanitize_keywords(cats, false).presence
+  end
+
   def feed_episodes
     default_feed.feed_episodes
   end

app/models/podcast_feed_handler.rb

@@ -1,5 +1,5 @@
 class PodcastFeedHandler
-  FEED_ATTRS = %w[ complete copyright description explicit keywords language
+  FEED_ATTRS = %w[ complete copyright description explicit language
     subtitle summary title update_frequency update_period
     author managing_editor new_feed_url owners ].freeze
 

app/representers/api/episode_representer.rb

@@ -37,7 +37,6 @@ class Api::EpisodeRepresenter < Api::BaseRepresenter
   property :is_closed_captioned
   property :is_perma_link
   property :include_in_feed?, as: :is_feed_ready
-  property :keywords
   property :categories
   property :position
 

app/representers/api/podcast_representer.rb

@@ -46,7 +46,6 @@ class Api::PodcastRepresenter < Api::BaseRepresenter
     property :managing_editor_email, as: :email
   end
 
-  collection :keywords
   collection :categories
   collection :restrictions
 

app/views/podcasts/show.rss.builder

@@ -82,11 +82,11 @@ xml.rss "xmlns:atom" => "http://www.w3.org/2005/Atom",
       xml.itunes(:summary) { xml.cdata!(itunes_summary(@podcast)) }
     end
 
-    xml.itunes :keywords, @podcast.keywords.join(",") unless @podcast.keywords.blank?
+    # xml.itunes :keywords, @podcast.keywords.join(",") unless @podcast.keywords.blank?
 
     xml.media :copyright, @podcast.copyright unless @podcast.copyright.blank?
     xml.media :thumbnail, url: @feed_image.url if @feed_image
-    xml.media :keywords, @podcast.keywords.join(",") unless @podcast.keywords.blank?
+    # xml.media :keywords, @podcast.keywords.join(",") unless @podcast.keywords.blank?
 
     cat = @itunes_categories.first.try(:name)
     xml.media :category, cat, scheme: "http://www.itunes.com/dtds/podcast-1.0.dtd" unless cat.blank?
@@ -152,7 +152,7 @@ xml.rss "xmlns:atom" => "http://www.w3.org/2005/Atom",
             xml.itunes :image, href: @itunes_image.url
           end
 
-          xml.itunes :keywords, ep.keywords.join(",") unless ep.keywords.blank?
+          # xml.itunes :keywords, ep.keywords.join(",") unless ep.keywords.blank?
           xml.itunes(:isClosedCaptioned, "Yes") if ep.is_closed_captioned
 
           if ep.media?

test/factories/podcast_factory.rb

@@ -16,7 +16,6 @@
     owner_email { "jesse@maximumfun.org" }
     categories { ["Humor", "Entertainment"] }
     explicit { "true" }
-    keywords { ["laffs", "comedy", "good-times"] }
     update_period { "weekly" }
     update_frequency { 1 }
     update_base { 1.year.ago }

test/models/episode_entry_handler_test.rb

@@ -58,7 +58,6 @@
     assert_equal episode.summary,
       "Few forms of contemporary architecture draw as much criticism as the McMansion, a particular type of oversized house that people love to hate. McMansions usually feature 3,000 or more square feet of space and fail to embody a cohesive style or interact with their environment. Kate Wagner, architecture critic and creator of McMansion Hell, is on a mission to illustrate just why these buildings are so terrible. McMansion Hell: The Devil is in the Details Support 99pi and Radiotopia today! Be part of the 5000 backer FreshBooks challenge: FreshBooks will donate $40,000 to Radiotopia if we get 5000 total new donations during this drive. FreshBooks makes intuitive and beautiful cloud accounting software for small businesses."
     assert_equal episode.explicit, "false"
-    assert_equal episode.keywords, ["roman mars", "kate wagner"]
     assert_equal episode.description,
       "Few forms of contemporary architecture draw as much criticism as the McMansion, a particular type of oversized house that people love to hate. McMansions usually feature 3,000 or more square feet of space and fail to embody a cohesive style or interact with their environment. Kate Wagner, architecture critic and creator of McMansion Hell, is on a mission to illustrate just why these buildings are so terrible. McMansion Hell: The Devil is in the Details Support 99pi and Radiotopia today! Be part of the 5000 backer FreshBooks challenge: FreshBooks will donate $40,000 to Radiotopia if we get 5000 total new donations during this drive. FreshBooks makes intuitive and beautiful cloud accounting software for small businesses."
     assert_equal episode.categories,

test/models/episode_test.rb

@@ -166,13 +166,15 @@
     assert_includes episode.keyword_xid, "241 its a title yay"
   end
 
-  it "strips non-alphanumeric characters from all keywords" do
-    episode.update(keywords: ["Here's John,ny!?"])
-    episode.run_callbacks :save do
-      episode.save
-    end
-    refute_nil episode.reload.keywords
-    episode.keywords.each { |k| refute_match(/['?!,]/, k) }
+  it "sanitizes categories" do
+    e = build_stubbed(:episode)
+
+    e.categories = []
+    assert_equal [], e.categories
+    assert_nil e[:categories]
+
+    e.categories = ["foo", " Foo ", "BAR  ", "  foo", "!@  $?"]
+    assert_equal ["foo", "Foo", "BAR", "!@  $?"], e.categories
   end
 
   it "has a valid itunes episode type" do

test/models/podcast_test.rb

@@ -66,6 +66,17 @@
     assert_nil p[:link]
   end
 
+  it "sanitizes categories" do
+    p = build_stubbed(:podcast)
+
+    p.categories = []
+    assert_equal [], p.categories
+    assert_nil p[:categories]
+
+    p.categories = ["foo", " Foo ", "BAR  ", "  foo", "!@  $?"]
+    assert_equal ["foo", "Foo", "BAR", "!@  $?"], p.categories
+  end
+
   describe "publishing" do
     it "creates a publish job on publish" do
       PublishingPipelineState.stub(:start_pipeline!, "published!") do

test/representers/api/podcast_representer_test.rb

@@ -20,10 +20,6 @@
     assert_equal json["owner"]["email"], "jesse@maximumfun.org"
   end
 
-  it "includes keywords" do
-    assert_includes json["keywords"], "laffs"
-  end
-
   it "includes categories" do
     assert_includes json["categories"], "Humor"
   end