Merge pull request #1040 from indentlabs/2021-10-07

Add word count caches

Author: drusepth

app/jobs/cache_attribute_word_count_job.rb

@@ -0,0 +1,30 @@
+class CacheAttributeWordCountJob < ApplicationJob
+  queue_as :cache
+
+  def perform(*args)
+    attribute_id = args.shift
+    attribute    = Attribute.find_by(id: attribute_id)
+
+    return if attribute.nil?
+    return if attribute.value.nil? || attribute.value.blank?
+
+    word_count = WordCountAnalyzer::Counter.new(
+      ellipsis:          'no_special_treatment',
+      hyperlink:         'count_as_one',
+      contraction:       'count_as_one',
+      hyphenated_word:   'count_as_one',
+      date:              'no_special_treatment',
+      number:            'count',
+      numbered_list:     'ignore',
+      xhtml:             'remove',
+      forward_slash:     'count_as_multiple_except_dates',
+      backslash:         'count_as_one',
+      dotted_line:       'ignore',
+      dashed_line:       'ignore',
+      underscore:        'ignore',
+      stray_punctuation: 'ignore'
+    ).count(attribute.value)
+
+    attribute.update!(word_count_cache: word_count)
+  end
+end

app/jobs/cache_sum_attribute_word_count_job.rb

@@ -0,0 +1,19 @@
+class CacheSumAttributeWordCountJob < ApplicationJob
+  queue_as :cache
+
+  def perform(*args)
+    entity_type = args.shift
+    entity_id   = args.shift
+
+    entity = entity_type.constantize.find_by(id: entity_id)
+    sum_attribute_word_count = Attribute.where(entity_type: entity_type, entity_id: entity_id).sum(:word_count_cache)
+
+    update = entity.word_count_updates.find_or_initialize_by(
+      for_date: DateTime.current,
+    )
+    update.word_count = sum_attribute_word_count
+    update.user_id  ||= entity.user_id
+
+    update.save!
+  end
+end

app/jobs/save_document_revision_job.rb

@@ -4,13 +4,21 @@ class SaveDocumentRevisionJob < ApplicationJob
   def perform(*args)
     document_id = args.shift
 
-    document = Document.find(document_id)
-    return unless document.present?
+    document = Document.find_by(id: document_id)
+    return unless document
 
     # Update cached word count for the document regardless of how often this is called
     new_word_count = document.computed_word_count
     document.update(cached_word_count: new_word_count)
 
+    # Save a WordCountUpdate for this document for today
+    update = document.word_count_updates.find_or_initialize_by(
+      for_date: DateTime.current,
+    )
+    update.word_count = new_word_count
+    update.user_id  ||= document.user_id
+    update.save!
+
     # Make sure we're only storing revisions at least every 5 min
     latest_revision = document.document_revisions.order('created_at DESC').limit(1).first
     if latest_revision.present? && latest_revision.created_at > 5.minutes.ago

app/models/concerns/is_content_page.rb

@@ -18,6 +18,11 @@ module IsContentPage
     has_many :timeline_events,             through: :timeline_event_entities
     has_many :timelines, -> { distinct },  through: :timeline_events
 
+    has_many :word_count_updates, as: :entity, dependent: :destroy
+    def latest_word_count_cache
+      word_count_updates.order('for_date DESC').limit(1).first.try(:word_count) ||  0
+    end
+
     scope :unarchived, -> { where(archived_at: nil) }
     def archive!
       update!(archived_at: DateTime.now)

app/models/documents/document.rb

@@ -26,6 +26,12 @@ class Document < ApplicationRecord
 
   attr_accessor :tagged_text
 
+  # Duplicated from is_content_page since we don't include that here yet
+  has_many :word_count_updates, as: :entity, dependent: :destroy
+  def latest_word_count_cache
+    word_count_updates.order('for_date DESC').limit(1).first.try(:word_count) ||  0
+  end
+
   KEYS_TO_TRIGGER_REVISION_ON_CHANGE = %w(title body synopsis notes_text)
 
   def self.color

app/models/page_data/attribute.rb

@@ -19,6 +19,16 @@ class Attribute < ApplicationRecord
     end
   end
 
+  after_commit do
+    if saved_changes.key?('value')
+      # Cache the updated word count on this attribute
+      CacheAttributeWordCountJob.perform_later(self.id)
+
+      # Cache the updated word count on the page this attribute belongs to
+      CacheSumAttributeWordCountJob.perform_later(self.entity_type, self.entity_id)
+    end
+  end
+
   after_save do
     entity.touch
   end

app/models/word_count_update.rb

@@ -0,0 +1,4 @@
+class WordCountUpdate < ApplicationRecord
+  belongs_to :user
+  belongs_to :entity, polymorphic: true
+end

db/migrate/20211007215520_add_word_count_cache_to_attributes.rb

@@ -0,0 +1,5 @@
+class AddWordCountCacheToAttributes < ActiveRecord::Migration[6.0]
+  def change
+    add_column :attributes, :word_count_cache, :integer
+  end
+end

db/migrate/20211007234707_create_word_count_updates.rb

@@ -0,0 +1,12 @@
+class CreateWordCountUpdates < ActiveRecord::Migration[6.0]
+  def change
+    create_table :word_count_updates do |t|
+      t.references :user, null: false, foreign_key: true
+      t.references :entity, polymorphic: true, null: false
+      t.integer :word_count
+      t.date :for_date
+
+      t.timestamps
+    end
+  end
+end

db/schema.rb

@@ -10,7 +10,7 @@
 #
 # It's strongly recommended that you check this file into your version control system.
 
-ActiveRecord::Schema.define(version: 2021_09_15_030031) do
+ActiveRecord::Schema.define(version: 2021_10_07_234707) do
 
   create_table "active_storage_attachments", force: :cascade do |t|
     t.string "name", null: false
@@ -165,6 +165,7 @@
     t.datetime "created_at"
     t.datetime "updated_at"
     t.datetime "deleted_at"
+    t.integer "word_count_cache"
     t.index ["attribute_field_id", "deleted_at", "entity_id", "entity_type"], name: "attributes_afi_deleted_at_entity_id_entity_type"
     t.index ["attribute_field_id", "deleted_at"], name: "index_attributes_on_attribute_field_id_and_deleted_at"
     t.index ["attribute_field_id", "user_id", "entity_type", "entity_id", "deleted_at"], name: "attributes_afi_ui_et_ei_da"
@@ -3636,6 +3637,18 @@
     t.integer "habitat_id"
   end
 
+  create_table "word_count_updates", force: :cascade do |t|
+    t.integer "user_id", null: false
+    t.string "entity_type", null: false
+    t.integer "entity_id", null: false
+    t.integer "word_count"
+    t.date "for_date"
+    t.datetime "created_at", precision: 6, null: false
+    t.datetime "updated_at", precision: 6, null: false
+    t.index ["entity_type", "entity_id"], name: "index_word_count_updates_on_entity_type_and_entity_id"
+    t.index ["user_id"], name: "index_word_count_updates_on_user_id"
+  end
+
   add_foreign_key "active_storage_attachments", "active_storage_blobs", column: "blob_id"
   add_foreign_key "api_keys", "users"
   add_foreign_key "api_requests", "application_integrations"
@@ -4047,4 +4060,5 @@
   add_foreign_key "vehicles", "users"
   add_foreign_key "votes", "users"
   add_foreign_key "votes", "votables"
+  add_foreign_key "word_count_updates", "users"
 end