Merge pull request #669 from scientist-softserv/i667-install-pdfjs

I667 install pdfjs

Gemfile

@@ -146,3 +146,5 @@ gem 'iiif_print', git: 'https://github.com/scientist-softserv/iiif_print.git', b
 gem 'order_already'
 gem 'redis', '~> 4.0'
 gem 'redlock', '~> 1.2.1'
+# locking to 5.1.0 because it was throwing errors on 5.3.0 in development
+gem 'simple_form', '5.1.0'

Gemfile.lock

@@ -1035,7 +1035,7 @@ GEM
       faraday (>= 0.17.5, < 3.a)
       jwt (>= 1.5, < 3.0)
       multi_json (~> 1.10)
-    simple_form (5.3.0)
+    simple_form (5.1.0)
       actionpack (>= 5.2)
       activemodel (>= 5.2)
     simplecov (0.16.1)
@@ -1225,6 +1225,7 @@ DEPENDENCIES
   secure_headers
   selenium-webdriver
   sentry-raven
+  simple_form (= 5.1.0)
   simplecov
   solr_wrapper (~> 2.0)
   spring (~> 1.7)

app/controllers/flipflop/strategies_controller_decorator.rb

@@ -0,0 +1,19 @@
+# frozen_string_literal: true
+
+# OVERRIDE Flipflop v2.7.1 to allow for custom `Action` labels
+
+module Flipflop
+  module StrategiesControllerDecorator
+    def enable?
+      values = StrategiesController::ENABLE_VALUES | ADDITIONAL_ENABLE_VALUES
+      values.include?(params[:commit])
+    end
+
+    def features_url(**_kargs)
+      hyrax.admin_features_path
+    end
+    ADDITIONAL_ENABLE_VALUES = FeaturesHelper::FEATURE_ACTION_LABELS.map { |_, v| v[:on] }.to_set.freeze
+  end
+end
+
+Flipflop::StrategiesController.prepend(Flipflop::StrategiesControllerDecorator)

app/controllers/hyrax/admin/strategies_controller_decorator.rb

@@ -0,0 +1,15 @@
+# frozen_string_literal: true
+
+# OVERRIDE Hyrax 2.9.6 to support Flipflop 2.7.1 features
+
+module Hyrax
+  module Admin
+    module StrategiesControllerDecorator
+      def features_url(**_kargs)
+        hyrax.admin_features_path
+      end
+    end
+  end
+end
+
+Hyrax::Admin::StrategiesController.prepend(Hyrax::Admin::StrategiesControllerDecorator)

app/helpers/features_helper.rb

@@ -0,0 +1,21 @@
+# frozen_string_literal: true
+
+module FeaturesHelper
+  def status_for(feature)
+    status = @feature_set.status(feature)
+    label = @feature_set.status(feature) == :enabled ? :on : :off
+    FEATURE_ACTION_LABELS.fetch(feature.name.to_sym, label => status)[label]
+  end
+
+  def on(feature)
+    FEATURE_ACTION_LABELS[feature]&.[](:on) || 'on'
+  end
+
+  def off(feature)
+    FEATURE_ACTION_LABELS[feature]&.[](:off) || 'off'
+  end
+
+  FEATURE_ACTION_LABELS = {
+    default_pdf_viewer: { on: 'PDF.js', off: 'UV' }
+  }.freeze
+end

app/helpers/pdf_js_helper.rb

@@ -0,0 +1,48 @@
+# frozen_string_literal: true
+
+module PdfJsHelper
+  def pdf_js_url(file_set_presenter)
+    # assumes that the download path exists if the file set has been characterized
+    url = if file_set_presenter.mime_type
+            "/pdf.js/web/viewer.html?file=#{hyrax.download_path(file_set_presenter.id)}"
+          else
+            file_set_presenter.solr_document["import_url_ssim"].first
+          end
+
+    url + "##{query_param}"
+  end
+
+  def pdf_file_set_presenter(presenter)
+    # currently only supports one pdf per work, falls back to the first pdf file set in ordered members
+
+    # Commenting this line out because even PDFs that were not split will still have a representative media
+    # which will be used first in this logic, consider uncommenting once all imports finish
+    # representative_presenter(presenter) ||
+    external_pdf(presenter)
+  end
+
+  def external_pdf(presenter)
+    reader, archival = pdf_file_set_presenters(presenter.file_set_presenters).partition do |fsp|
+      fsp.solr_document["import_url_ssim"]&.first&.include? "READER"
+    end
+
+    reader.first || archival.first
+  end
+
+  def pdf_file_set_presenters(presenters)
+    presenters.select(&:pdf?).presence || presenters.select do |file_set_presenter|
+      file_set_presenter.solr_document["label_ssi"].downcase.end_with? ".pdf"
+    end
+  end
+
+  def representative_presenter(presenter)
+    presenter.file_set_presenters.find { |file_set_presenter| file_set_presenter.id == presenter.representative_id }
+  end
+
+  def query_param
+    search_params = current_search_session.try(:query_params) || {}
+    q = search_params['q'].presence || ''
+
+    "search=#{q}&phrase=true"
+  end
+end

app/indexers/hyrax/file_set_indexer_decorator.rb

@@ -0,0 +1,34 @@
+# frozen_string_literal: true
+
+# OVERRIDE Hyrax 3.5.0 to add PDF text to solr document when using the default PDF viewer (PDF.js)
+
+module Hyrax
+  module FileSetIndexerDecorator
+    def generate_solr_document
+      return super unless Flipflop.default_pdf_viewer?
+
+      super.tap do |solr_doc|
+        solr_doc['all_text_timv'] = solr_doc['all_text_tsimv'] = pdf_text
+      end
+    end
+
+    private
+
+      def pdf_text
+        return unless object.pdf?
+        return unless object.original_file&.content.is_a? String
+
+        text = IO.popen(['pdftotext', '-', '-'], 'r+b') do |pdftotext|
+          pdftotext.write(object.original_file.content)
+          pdftotext.close_write
+          pdftotext.read
+        end
+
+        text.tr("\n", ' ')
+            .squeeze(' ')
+            .encode('UTF-8', 'binary', invalid: :replace, undef: :replace, replace: '') # remove non-UTF-8 characters
+      end
+  end
+end
+
+Hyrax::FileSetIndexer.prepend(Hyrax::FileSetIndexerDecorator)

app/presenters/hyku/work_show_presenter.rb

@@ -29,6 +29,23 @@ def video_embed_viewer
       :video_embed_viewer
     end
 
+    def pdf_viewer?
+      return unless Flipflop.default_pdf_viewer?
+      return unless file_set_presenters.any?(&:pdf?) || pdf_extension?
+
+      # If all of the member_presenters are file_set presenters, return true
+      # this also means that there are no child works
+      member_presenters.all? { |presenter| presenter.is_a? Hyrax::FileSetPresenter }
+    end
+
+    def pdf_extension?
+      file_set_presenters.any? { |fsp| fsp.label.downcase.end_with?('.pdf') }
+    end
+
+    def viewer?
+      iiif_viewer? || video_embed_viewer? || pdf_viewer?
+    end
+
     private
 
       def extract_from_identifier(rgx)

app/views/hyrax/admin/features/index.html.erb

@@ -0,0 +1,76 @@
+<%
+  # OVERRIDE Hyrax 2.9.6 to make the default PDF viewer switch more intuitive and some styling changes.
+  # Instead of saying on/off it says PDF.js/IIIF Print.
+%>
+
+<% provide :page_header do %>
+  <h1><span class="fa fa-wrench" aria-hidden="true"></span> <%= t('.header') %></h1>
+<% end %>
+<div class="flip row">
+  <div class="col-md-12">
+    <div class="panel">
+      <div class="panel-body">
+        <div class="table-responsive">
+          <table class="table table-striped">
+            <thead>
+              <tr>
+                <th></th>
+                <th class="name"><%= t('.feature') %></th>
+                <th class="description"><%= t('.description') %></th>
+                <th class="action"><%= t('.action') %></th>
+              </tr>
+            </thead>
+            <tbody>
+            <% @feature_set.grouped_features.each do |group, features| -%>
+              <% if @feature_set.grouped? -%>
+                <tr class="group">
+                  <td></td>
+                  <td class="name" colspan="<%= 2 + @feature_set.strategies.size -%>">
+                    <h2>
+                      <%= t(group ? group.name : :default, scope: [:flipflop, :groups], default: group ? group.title : nil) -%>
+                    </h2>
+                  </td>
+                </tr>
+              <% end -%>
+              <% features.each do |feature| %>
+              <tr data-feature="<%= feature.name.dasherize.parameterize %>">
+                <td class="status">
+                  <span class="<%= @feature_set.status(feature) -%>"><%= status_for(feature) -%></span>
+                </td>
+                <td class="name"><%= feature.name.humanize -%></td>
+                <td class="description"><%= feature.description -%></td>
+
+                <% @feature_set.strategies.each do |strategy| -%>
+                  <% next unless strategy.is_a? Flipflop::Strategies::ActiveRecordStrategy %>
+                  <%# OVERRIDE to add min-width so Actions column can display all toggles on the same line %>
+                  <%# adjust min-width as needed for future overrides %>
+                  <td class="toggle" style="min-width: 110px;" data-strategy="<%= strategy.name.dasherize.parameterize %>">
+                    <div class="toolbar">
+                      <%= form_tag(hyrax.admin_feature_strategy_path(feature.key, strategy.key), method: :put) do -%>
+                        <div class="btn-group">
+                          <%# OVERRIDE to use helper, see FeaturesHelper %>
+                          <%= submit_tag on(feature.name.to_sym),
+                            type: "submit",
+                            class: Flipflop.enabled?(feature.name.to_sym) ? 'active' : nil,
+                            disabled: !strategy.switchable? -%>
+
+                          <%# OVERRIDE to use helper, see FeaturesHelper %>
+                          <%= submit_tag off(feature.name.to_sym),
+                            type: "submit",
+                            class: Flipflop.enabled?(feature.name.to_sym) ? nil : 'active',
+                            disabled: !strategy.switchable? -%>
+                        </div>
+                      <% end -%>
+                    </div>
+                  </td>
+                <% end -%>
+              </tr>
+              <% end -%>
+            <% end -%>
+            </tbody>
+          </table>
+        </div>
+      </div>
+    </div>
+  </div>
+</div>

app/views/hyrax/base/_pdf_js.html.erb

@@ -0,0 +1,7 @@
+<div class="viewer-wrapper">
+  <iframe id="pdf-viewer" aria-label="image view"
+    src="<%= pdf_js_url(file_set_presenter) %>"
+    allowfullscreen="true"
+    frameborder="0"
+  ></iframe>
+</div>

app/views/hyrax/base/show.html.erb

@@ -15,16 +15,21 @@
       <div class="panel-body">
         <div class="row">
           <%= render 'workflow_actions_widget', presenter: @presenter %>
-          <% if @presenter.universal_viewer? || @presenter.video_embed_viewer? %>
+          <% if @presenter.pdf_viewer? %>
+            <div class="col-sm-12">
+              <%= render 'pdf_js', file_set_presenter: pdf_file_set_presenter(@presenter) %>
+            </div>
+          <% elsif @presenter.iiif_viewer? || @presenter.video_embed_viewer? %>
             <div class="col-sm-12">
               <%= render 'representative_media', presenter: @presenter, viewer: true %>
             </div>
-          <% end %>
-          <div class="col-sm-5 text-center">
-            <%= render 'representative_media', presenter: @presenter, viewer: false unless (@presenter.universal_viewer? || @presenter.video_embed_viewer?) %>
-            <%= render 'citations', presenter: @presenter %>
-            <%= render 'social_media' %>
-          </div>
+          <% else %>          
+            <div class="col-sm-5 text-center">
+              <%= render 'representative_media', presenter: @presenter, viewer: false unless (@presenter.universal_viewer? || @presenter.video_embed_viewer?) %>
+              <%= render 'citations', presenter: @presenter %>
+              <%= render 'social_media' %>
+            </div>
+            <% end %>
           <div class="col-sm-7">
             <%= render 'metadata', presenter: @presenter %>
           </div>

config/features.rb

@@ -6,4 +6,9 @@
   feature :show_share_button,
           default: true,
           description: "Shows the 'Share Your Work' button on the homepage."
+
+  # Flipflop.default_pdf_viewer? returning `true` means we use PDF.js and `false` means we use IIIF Print.
+  feature :default_pdf_viewer,
+          default: true,
+          description: "Choose PDF.js or Universal Viewer to render PDFs. UV uses IIIF Print and requires PDF spltting with OCR. Switching from PDF.js to the UV may require re-ingesting of the PDF."
  end