T343 schoolie (#514)

Resolves #343 by adding google scholar compliant meta tags to HTML for ETDs only, and replacing sitemap with new sitemap that only includes ETDs (for now)

Gemfile

@@ -9,9 +9,6 @@ gem 'rails', '~> 5.2.8.1'
 gem "sqlite3", "~> 1.3.0"
 # Use pg as the production database for Active Record
 gem 'pg'
-# Use sitemap
-# See https://github.com/viseztrance/rails-sitemap
-gem 'sitemap'
 # Use Passenger as the app server
 # Update this when we update the Passenger docker container base image version
 gem 'passenger', '6.0.17', require: "phusion_passenger/rack_handler"
@@ -88,6 +85,8 @@ gem "ffi", "~> 1.15"
 
 gem 'json-canonicalization', '0.3.1' # https://github.com/dryruby/json-canonicalization/issues/2
 
+gem 'schoolie', '0.1.3'
+
 gem 'prawn'
 
 group :development, :test do

Gemfile.lock

@@ -876,6 +876,8 @@ GEM
     sassc (2.4.0)
       ffi (~> 1.9)
     scanf (1.0.0)
+    schoolie (0.1.3)
+      actionview (~> 5)
     select2-rails (3.5.11)
     shacl (0.1.1)
       json-ld (~> 3.1, >= 3.1.7)
@@ -1057,6 +1059,7 @@ DEPENDENCIES
   rsolr (>= 1.0, < 3)
   rspec-rails
   sass-rails (~> 5.0)
+  schoolie (= 0.1.3)
   sidekiq (~> 6)
   simplecov
   sitemap

app/jobs/sitemap_regenerate_job.rb

@@ -4,7 +4,6 @@
 
 class SitemapRegenerateJob < ApplicationJob
   def perform
-    Rake::Task['sitemap:generate'].invoke
-    Rake::Task['sitemap:ping'].invoke
+    Rake::Task['schoolie:sitemap'].invoke
   end
 end

app/models/gw_etd.rb

@@ -5,7 +5,7 @@ class GwEtd < ActiveFedora::Base
   # Change this to restrict which works can be added as a child.
   # self.valid_child_concerns = []
   self.indexer = GwEtdIndexer
-
+  
   validates :title, presence: { message: 'Your work must have a title.' }
 
   property :gw_affiliation, predicate: ::RDF::URI.new('http://scholarspace.library.gwu.edu/ns#gwaffiliation') do |index|

app/models/gw_work.rb

@@ -17,4 +17,4 @@ class GwWork < ActiveFedora::Base
   end
 
   include ::Hyrax::BasicMetadata
-end
+end

app/presenters/hyrax/gw_etd_presenter.rb

@@ -6,5 +6,11 @@ class GwEtdPresenter < GwWorkPresenter
     def permanent_url
       Scholarspace::Application.config.permanent_url_base + "etd/#{id}"
     end
+
+    # scholarly? is used to determine whether or not
+    # the Google Scholar meta tags are rendered
+    def scholarly?
+      true
+    end
   end
 end

app/presenters/hyrax/gw_work_presenter.rb

@@ -5,5 +5,9 @@ class GwWorkPresenter < Hyrax::WorkShowPresenter
     def permanent_url
       Scholarspace::Application.config.permanent_url_base + "work/#{id}"
     end
+
+    def scholarly?
+      false
+    end
   end
 end

app/views/layouts/_head_tag_content.html.erb

@@ -12,8 +12,8 @@ signed in %>
 
 <!-- Twitter card metadata -->
 <%= yield :twitter_meta %>
-<!-- Google Scholar metadata -->
-<%= yield :gscholar_meta %>
+<!-- Schoolie for Google Scholar metadata (replaces :gscholar_meta) -->
+<%= yield :schoolie_meta %>
 
 <title><%= content_for?(:page_title) ? yield(:page_title) : default_page_title %></title>
 

app/views/shared/_citations.html.erb

@@ -0,0 +1,2 @@
+<%= render 'shared/twitter_citations' %>
+<%= render 'shared/schoolie_citations' %>

app/views/shared/_schoolie_citations.html.erb

@@ -0,0 +1,5 @@
+<% content_for(:schoolie_meta) do %>
+  <% if @presenter.scholarly? %>
+    <%schoolie_tags(@presenter).html_safe%>
+  <% end %>
+<% end %>

app/views/shared/_twitter_citations.html.erb

@@ -0,0 +1,15 @@
+<% content_for(:twitter_meta) do %>
+  <meta name="twitter:card" content="product" />
+  <meta name="twitter:site" content="<%= t('hyrax.product_twitter_handle') %>" />
+  <meta name="twitter:creator" content="<%= @presenter.tweeter %>" />
+  <meta property="og:site_name" content="<%= application_name %>" />
+  <meta property="og:type" content="object" />
+  <meta property="og:title" content="<%= @presenter.title.first %>" />
+  <meta property="og:description" content="<%= @presenter.description.first.truncate(200) rescue @presenter.title.first %>" />
+  <meta property="og:image" content="<%= @presenter.download_url %>" />
+  <meta property="og:url" content="<%= polymorphic_url([main_app, @presenter]) %>" />
+  <meta name="twitter:data1" content="<%= @presenter.keyword.join(', ') %>" />
+  <meta name="twitter:label1" content="Keywords" />
+  <meta name="twitter:data2" content="<%= @presenter.rights_statement.first %>" />
+  <meta name="twitter:label2" content="Rights Statement" />
+<% end %>

config/authorities/resource_types.yml

@@ -15,12 +15,16 @@ terms:
     term: Conference Proceeding
   - id: Dataset
     term: Dataset
+  - id: Dissertation
+    term: Dissertation
   - id: Image
     term: Image
   - id: Journal
     term: Journal
   - id: Map or Cartographic Material
     term: Map or Cartographic Material
+  - id: Master's Thesis
+    term: Master's Thesis
   - id: Meeting Minutes
     term: Meeting Minutes
   - id: Newsletter
@@ -39,8 +43,6 @@ terms:
     term: Research Paper
   - id: Software or Program Code
     term: Software or Program Code
-  - id: Thesis or Dissertation
-    term: Thesis or Dissertation
   - id: Video
     term: Video
   - id: Working Paper

config/etd_degree_map.yml

@@ -0,0 +1,20 @@
+Master's Thesis:
+  - M.A.
+  - M.S.
+  - M.P.H.
+  - LL.M.
+  - M.F.A.
+  - M.A.T.
+  - M.P.S.
+  - M.B.A.
+  - M.F.S.
+  - M.Int.St.
+  - M.P.H.
+  - M.P.P.
+Dissertation:
+  - Ph.D.
+  - Ed.D.
+  - D.Engr.
+  - Dr.P.H.
+  - D.Sc.
+  - S.J.D.

config/schoolie.yml

@@ -0,0 +1,10 @@
+static:
+  citation_institution: George Washington University
+attributes:
+  citation_title: title
+  citation_author: creator
+  citation_type: resource_type
+  dc.type: resource_type
+  citation_date: publication_date
+  citation_keywords: keyword
+  citation_pdf_url: download_url

lib/tasks/gwss.rake

@@ -20,9 +20,9 @@ namespace :gwss  do
     end
   end
 
-  desc "Queues a job to (re)generate the sitemap.xml"
+  desc "Executes (immediately) a job to (re)generate the sitemap.xml"
   task "sitemap_queue_generate" => :environment do
-    SitemapRegenerateJob.perform_later
+    SitemapRegenerateJob.perform_now
   end
 
   desc "Creates the default Admin Set if it doesn't exist"
@@ -141,6 +141,10 @@ namespace :gwss  do
       # problem that would be caused by referencing GwEtd first
       # See articles such as http://neethack.com/2015/04/rails-circular-dependency/
       GwWork
+
+      degree_hash = YAML.load_file('config/etd_degree_map.yml')
+      degree_categories = degree_hash.keys # Typically ["Master's Thesis", "Dissertation"]
+
       manifest_file = options[:mfpath]
       if File.exist?(manifest_file)
         mf = File.read(manifest_file)
@@ -153,7 +157,10 @@ namespace :gwss  do
           item_attributes['degree'] = manifest_json['degree'][0]
         end
         # resource_type may need more logic around it, TBD
-        item_attributes['resource_type'] = ['Thesis or Dissertation']
+        if manifest_json['etd_type']
+          item_attributes['resource_type'] = manifest_json['etd_type']
+        end
+        # item_attributes['resource_type'] = ['Thesis or Dissertation']
 
         # dc:rights
         # Always set this license for ETDs
@@ -313,4 +320,48 @@ namespace :gwss  do
     ContentBlock.find_or_create_by(name: "help_page").update!(value: help_page_html.read)
     ContentBlock.find_or_create_by(name: "share_page").update!(value: share_page_html.read)
   end
+
+  desc "Reassigns GwEtd resource_type values to Master's Thesis or Dissertation"
+  task "reassign_etd_resource_types" => :environment do
+    etd_degree_map = YAML.load_file('config/etd_degree_map.yml')
+    degree_etd_map = {}
+    degree_categories = etd_degree_map.keys
+    # Flip etd_degree_map to create degree_etd_map
+    # So that for any given degree, we can get back whether it's a masters or a doctorate
+    degree_categories.each do |degree_category|
+      etd_degree_map[degree_category].each do |degree_name|
+        # upcase each degree (just in case) and ignore "."s
+        degree_etd_map[degree_name.upcase.delete('.')] = degree_category
+      end
+    end
+
+    ids = Hyrax::SolrService.new.get("has_model_ssim:GwEtd", fl: [:id], rows: 1_000_000)
+    ids["response"]["docs"].each do |doc|
+      work = GwEtd.find(doc["id"])
+      if work.degree.nil?
+        puts "GwEtd id=#{doc["id"]} degree is empty! Skipping"
+      else
+        degree_name = work.degree.upcase.delete('.')
+        if degree_etd_map.keys.include?(degree_name)
+          work.resource_type = [degree_etd_map[degree_name]]
+          work.save
+          puts "Reassigned #{degree_name} resource type to #{degree_etd_map[degree_name]}"
+        else
+          puts "Degree name #{degree_name} not found! Skipping"
+        end
+      end
+    end
+  end
+
+  desc "Enumerates degree types present among existing GwEtd works"
+  task "enumerate_degree_types" => :environment do
+    ids = Hyrax::SolrService.new.get("has_model_ssim:GwEtd", fl: [:id], rows: 1_000_000)
+    docs = ids["response"]["docs"]
+    # Map a list of ids to a list of degree values
+    degrees = docs.map {|doc| GwEtd.find(doc["id"]).degree}
+    degree_hash = degrees.tally
+    degree_hash.keys.each do |key|
+      puts "#{key}, #{degree_hash[key]}"
+    end
+  end
 end

lib/tasks/schoolie.rake

@@ -0,0 +1,30 @@
+# frozen_string_literal: true
+
+require 'json'
+require 'nokogiri'
+
+namespace :schoolie do
+  desc "Creates Google Scholar compliant sitemap for GwEtd works"
+  task sitemap: :environment do
+    date_field = 'system_modified_dtsi'
+    result = Hyrax::SolrService.new.get("has_model_ssim:GwEtd",
+                                        fl: "id,#{date_field}",
+                                        rows: 1_000_000)
+    ids = result['response']['docs'].map do |x|
+      ["https://scholarspace.library.gwu.edu/etd/#{x['id'].to_s}", x[date_field].to_s]
+    end
+    builder = Nokogiri::XML::Builder.new do |sitemap|
+      sitemap.urlset("xmlns:xsi": "http://www.w3.org/2001/XMLSchema-instance",
+                 xmlns: "http://www.sitemaps.org/schemas/sitemap/0.9",
+                 "xsi:schemaLocation": "http://www.sitemaps.org/schemas/sitemap/0.9 http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd") {
+                   ids.each { |url, date|
+                     sitemap.url {
+                       sitemap.loc url
+                       sitemap.lastmod date
+                     }
+                   }
+                 }
+    end
+    File.open(Rails.root.join("public", "sitemap.xml"), "w") { |f| f.write(builder.to_xml) }
+  end
+end