Adjust data model.

refs #14

README.md

@@ -36,4 +36,32 @@ docker run -d --rm --init --ulimit core=0 -p 8070:8070 lfoppiano/grobid:latest-f
 ```
 
 ## Helpful task
-Export the metadata for a collection to CSV: `bin/rake "export_csv[druid:jk956kb4381]"`
+Export the metadata for a collection to CSV: `bin/rake "export_csv[druid:jk956kb4381]"`
+
+## Data model for an article
+```
+{
+    "title": STRING,
+    "authors": [
+        {
+            "first_name": STRING (REQUIRED - includes middle name, initials, etc.),
+            "last_name": STRING (REQUIRED),
+            "affiliations": [
+                "department": STRING,
+                "organization": STRING (REQUIRED)
+            ],
+            "orcid": STRING (for example, https://orcid.org/0000-0003-1527-0030)
+        }
+    ],
+    "abstract": STRING,
+    "keywords": [
+        {
+            value: STRING
+        }
+    ],
+    "related_resource_citation": STRING,
+    "related_resource_doi": STRING (for example, 10.5860/lrts.48n4.8259),
+    "published": BOOLEAN,
+    "collection_druid": STRING (for example, druid:jk956kb4381)
+}
+```

app/components/works/fallback_form_component.html.erb

@@ -3,7 +3,7 @@
     <p>Provide a DOI or a citation here or manually complete the form below.</p>
     <%= form_with url: new_works_path, builder: ShroomFormBuilder, data: { controller: 'submits-with', action: 'submits-with#showStatus', turbo: false } do |form| %>
         <%= form.hidden_field :work_file, value: work_file_id %>
-        <%= form.hidden_field :preprint, value: preprint %>
+        <%= form.hidden_field :published, value: published %>
         <%= form.bs_label :doi, 'DOI:', class: 'mt-2' %>
         <%= form.bs_text_field :doi, class: 'form-control', pattern: 'https://doi.org/10.\d+/.+' %>
         <%= form.bs_help_text 'For example: https://doi.org/10.1177/1940161218781254' %>

app/components/works/fallback_form_component.rb

@@ -3,14 +3,14 @@
 module Works
   # Provides a fallback to provide a DOI or citation.
   class FallbackFormComponent < ViewComponent::Base
-    def initialize(work_form:, work_file_id:, preprint:)
+    def initialize(work_form:, work_file_id:, published:)
       @work_form = work_form
       @work_file_id = work_file_id
-      @preprint = preprint
+      @published = published
       super()
     end
 
-    attr_reader :work_form, :work_file_id, :preprint
+    attr_reader :work_form, :work_file_id, :published
 
     def render?
       work_form.title.blank? && work_file_id

app/controllers/files_controller.rb

@@ -4,7 +4,7 @@
 class FilesController < ApplicationController
   def create
     work_file = WorkFile.create!(file_params)
-    redirect_to new_works_path(work_file:, doi: params[:doi], preprint: params[:preprint])
+    redirect_to new_works_path(work_file:, doi: params[:doi], published: params[:published])
   end
 
   private

app/controllers/works_controller.rb

@@ -77,28 +77,27 @@ def grobid_service
   # rubocop:disable Metrics/AbcSize
   def build_new_work_form(work_file:)
     if params[:citation].present?
-      grobid_service.from_citation(citation: params[:citation], preprint: preprint?)
+      grobid_service.from_citation(citation: params[:citation], published: published?)
     elsif params[:doi].present?
-      grobid_service.from_citation(citation: params[:doi], preprint: preprint?)
+      grobid_service.from_citation(citation: params[:doi], published: published?)
     elsif params.key?(:work_file)
-      grobid_service.from_file(path: work_file.path, preprint: preprint?)
+      grobid_service.from_file(path: work_file.path, published: published?)
     else
-      WorkForm.new(preprint: preprint?)
+      WorkForm.new(published: published?)
     end
   end
   # rubocop:enable Metrics/AbcSize
 
-  def preprint?
-    params[:preprint] == 'true'
+  def published?
+    params[:published] == 'true'
   end
 
   def work_params
     # Perhaps these can be introspected from the model?
     params.require(:work).permit(
       :title, :abstract, :publisher,
-      :published_year, :published_month, :published_day,
-      :related_resource_citation, :preprint, :collection_druid,
-      :doi, :related_resource_doi,
+      :related_resource_citation, :published, :collection_druid,
+      :related_resource_doi,
       authors_attributes: [
         :first_name, :last_name, :orcid, { affiliations_attributes: %i[organization department] }
       ],

app/forms/base_form.rb

@@ -5,7 +5,7 @@ class BaseForm
   include ActiveModel::Model
   include ActiveModel::Attributes
   include ActiveModel::Validations::Callbacks
-  include ActiveModel::Serialization
+  include ActiveModel::Serializers::JSON
 
   def self.model_name
     # Remove the "Form" suffix from the class name.

app/forms/work_form.rb

@@ -28,35 +28,6 @@ def authors_are_valid
 
   attribute :abstract, :string
 
-  # For a preprint, the published date is the date the preprint was published not the actual publication date.
-  attribute :published_year, :integer
-  validates :published_year, numericality: { only_integer: true, in: 1900..Date.current.year }, allow_nil: true
-
-  attribute :published_month, :integer
-  validates :published_month, numericality: { only_integer: true, in: 1..12 }, allow_nil: true
-  validate :published_month_is_valid
-
-  def published_month_is_valid
-    errors.add(:published_month, 'requires a year') if published_year.blank? && published_month.present?
-  end
-
-  attribute :published_day, :integer
-  validates :published_day, numericality: { only_integer: true, in: 1..31 }, allow_nil: true
-  validate :published_day_is_valid
-
-  def published_day_is_valid
-    return unless (published_year.blank? || published_month.blank?) && published_day.present?
-
-    errors.add(:published_day,
-               'requires a year and month')
-  end
-
-  # Preprints don't have publishers
-  attribute :publisher, :string
-
-  attribute :doi, :string
-  validates :doi, format: { with: DoiSupport::REGEX }, allow_blank: true, unless: :preprint?
-
   attribute :keywords, array: true, default: -> { [] }
   before_validation do
     keywords.compact_blank!
@@ -66,18 +37,18 @@ def keywords_attributes=(attributes)
     self.keywords = attributes.map { |_, keyword| KeywordForm.new(keyword) }
   end
 
-  # Preprints have a single related resource.
+  # Published articles have a single related resource.
   attribute :related_resource_citation, :string
-  validates :related_resource_citation, presence: true, if: :preprint?
+  validates :related_resource_citation, presence: true, if: :published?
 
   attribute :related_resource_doi, :string
-  validates :related_resource_doi, format: { with: DoiSupport::REGEX }, allow_blank: true, if: :preprint?
+  validates :related_resource_doi, format: { with: DoiSupport::REGEX }, allow_blank: true, if: :published?
 
-  attribute :preprint, :boolean, default: false
+  attribute :published, :boolean, default: false
 
   attribute :collection_druid, :string
 
-  def preprint?
-    preprint || related_resource_citation.present?
+  def published?
+    published || related_resource_citation.present?
   end
 end

app/services/grobid_service.rb

@@ -13,23 +13,23 @@ def self.from_citation(...)
   end
 
   # @param [String] path the path to the PDF file
-  # @param [Boolean] preprint whether the work is a preprint
+  # @param [Boolean] published whether the work is a published article
   # @return [Work] a Work model with metadata extracted from the PDF
   # @raise [Error] if there is an error extracting metadata from the PDF
-  def from_file(path:, preprint: false)
+  def from_file(path:, published: false)
     @tei = fetch_tei_from_file(path:)
-    @bibtex = fetch_tei_from_file(path:, tei: false) if preprint
+    @bibtex = fetch_tei_from_file(path:, tei: false) if published
     tei_to_work(tei:, bibtex:)
   end
 
   # @param [String] citation for the work
-  # @param [Boolean] preprint whether the work is a preprint
+  # @param [Boolean] published whether the work is a published article
   # @return [Work] a Work model with metadata extracted from the PDF
   # @raise [Error] if there is an error extracting metadata from the PDF
-  def from_citation(citation:, preprint: false)
+  def from_citation(citation:, published: false)
     tei_fragment = fetch_tei_from_citation(citation:)
     @tei = "<TEI xmlns=\"http://www.tei-c.org/ns/1.0\">#{tei_fragment}</TEI>"
-    @bibtex = fetch_tei_from_citation(citation:, tei: false) if preprint
+    @bibtex = fetch_tei_from_citation(citation:, tei: false) if published
     tei_to_work(tei:, bibtex:)
   end
 

app/services/tei_cocina_mapper_service.rb

@@ -9,7 +9,7 @@ def self.call(...)
   end
 
   # @param [Nokogiri::XML::Document] tei_ng_xml
-  # @param [String,nil] related_resource_citation citation for the article this is a preprint of
+  # @param [String,nil] related_resource_citation citation for the article this is a published of
   def initialize(tei_ng_xml:, related_resource_citation: nil)
     @tei_doc = TeiDocument.new(ng_xml: tei_ng_xml)
     @related_resource_citation = related_resource_citation
@@ -40,9 +40,7 @@ def description_params
       title: CocinaDescriptionSupport.title(title: tei_doc.title),
       contributor: tei_doc.authors.map { |author_attrs| CocinaDescriptionSupport.person_contributor(**author_attrs) },
       note: note_params,
-      event: event_params,
       subject: subject_params,
-      identifier: identifier_params,
       relatedResource: related_resource_params
     }.compact
   end
@@ -53,34 +51,14 @@ def note_params
     [CocinaDescriptionSupport.note(type: 'abstract', value: tei_doc.abstract)]
   end
 
-  def event_params
-    return [] if preprint? # If a preprint, these are likely to be for the related resource.
-
-    [].tap do |params|
-      if tei_doc.published_date.present?
-        params << CocinaDescriptionSupport.event_date(date_type: 'publication',
-                                                      date_value: tei_doc.published_date)
-      end
-      if tei_doc.publisher.present?
-        params << CocinaDescriptionSupport.event_contributor(contributor_name_value: tei_doc.publisher)
-      end
-    end
-  end
-
   def subject_params
     return if tei_doc.keywords.blank?
 
     CocinaDescriptionSupport.subjects(values: tei_doc.keywords)
   end
 
-  def identifier_params
-    return if preprint? || tei_doc.doi.blank?
-
-    [CocinaDescriptionSupport.doi_identifier(doi: tei_doc.doi)]
-  end
-
   def related_resource_params
-    return unless preprint?
+    return unless published?
 
     [
       {
@@ -91,7 +69,7 @@ def related_resource_params
     ]
   end
 
-  def preprint?
+  def published?
     related_resource_citation.present?
   end
 

app/services/work_cocina_mapper_service/to_cocina/description_mapper.rb

@@ -35,9 +35,7 @@ def params
           title: CocinaDescriptionSupport.title(title: work_form.title),
           contributor: contributors_params.presence,
           note: note_params.presence,
-          event: event_params.presence,
           subject: subject_params.presence,
-          identifier: identifier_params,
           purl: Sdr::Purl.from_druid(druid:),
           relatedResource: related_resource_params
         }.compact
@@ -69,29 +67,10 @@ def note_params
         end
       end
 
-      # rubocop:disable Metrics/AbcSize
-      def event_params
-        [].tap do |params|
-          date_value = EdtfSupport.to_edtf(year: work_form.published_year, month: work_form.published_month,
-                                           day: work_form.published_day)
-          params << CocinaDescriptionSupport.event_date(date_value:, date_type: 'publication') if date_value.present?
-          if work_form.publisher.present?
-            params << CocinaDescriptionSupport.event_contributor(contributor_name_value: work_form.publisher)
-          end
-        end
-      end
-      # rubocop:enable Metrics/AbcSize
-
       def subject_params
         CocinaDescriptionSupport.subjects(values: work_form.keywords.map(&:value))
       end
 
-      def identifier_params
-        return if work_form.doi.blank?
-
-        [CocinaDescriptionSupport.doi_identifier(doi: work_form.doi)]
-      end
-
       def related_resource_params
         resource_params = {}.tap do |params|
           if work_form.related_resource_citation.present?

app/services/work_cocina_mapper_service/to_work_mapper.rb

@@ -19,48 +19,18 @@ def call
 
     attr_reader :cocina_object
 
-    # rubocop:disable Metrics/AbcSize
     def params
       {
         title: CocinaSupport.title_for(cocina_object:),
         authors: WorkCocinaMapperService::ToWork::AuthorsMapper.call(cocina_object:),
         abstract: cocina_object.description.note.find { |note| note.type == 'abstract' }&.value,
-        published_year: published_date&.year,
-        published_month: EdtfSupport.month_for(edtf: published_date),
-        published_day: EdtfSupport.day_for(edtf: published_date),
-        publisher:,
         keywords:,
-        doi:,
         related_resource_citation:,
         related_resource_doi:,
         collection_druid: CocinaSupport.collection_druid_for(cocina_object:)
       }
     end
 
-    # rubocop:disable Metrics/CyclomaticComplexity
-    # rubocop:disable Metrics/PerceivedComplexity
-    def published_date
-      @published_date ||= begin
-        published_event = cocina_object.description.event.find do |event|
-          event.type == 'deposit' \
-          && event.date.first&.encoding&.code == 'edtf' \
-          && event.date.first&.type == 'publication'
-        end
-        EdtfSupport.parse_with_precision(date: published_event&.date&.first&.value)
-      end
-    end
-
-    def publisher
-      publisher_event = cocina_object.description.event.find do |event|
-        event.type == 'publication' \
-        && event.contributor&.first&.role&.first&.value == 'publisher'
-      end
-      publisher_event&.contributor&.first&.name&.first&.value
-    end
-    # rubocop:enable Metrics/AbcSize
-    # rubocop:enable Metrics/CyclomaticComplexity
-    # rubocop:enable Metrics/PerceivedComplexity
-
     def keywords
       cocina_object.description.subject
                    .select { |subject| subject.type == 'topic' }
@@ -74,10 +44,6 @@ def related_resource_citation
       note.value
     end
 
-    def doi
-      doi_for(cocina_object.description)
-    end
-
     def related_resource_doi
       doi_for(related_resource)
     end

app/views/files/_new.html.erb

@@ -1,11 +1,11 @@
 <%= form_with url: files_path, builder: ShroomFormBuilder, data: { controller: 'submits-with', action: 'submits-with#showStatus' } do |form| %>
     <div class="form-check">
-        <%= form.bs_radio_button :preprint, 'false', required: true %>
-        <%= form.bs_radio_label :preprint_false, 'This is an article or manuscript (published or unpublished).' %>
+        <%= form.bs_radio_button :published, 'true', required: true %>
+        <%= form.bs_radio_label :published_true, 'This is a preprint or copy of an article / manuscript that is forthcoming or has been published elsewhere.' %>
     </div>
     <div class="form-check">
-        <%= form.bs_radio_button :preprint, 'true', required: true %>
-        <%= form.bs_radio_label :preprint_true, 'This is a preprint of an article (forthcoming or published).' %>
+        <%= form.bs_radio_button :published, 'false', required: true %>
+        <%= form.bs_radio_label :published_false, 'This is an article / manuscript that will only be published in SDR or has not been submitted and / or accepted for publication.' %>
     </div>
     <%= form.bs_label :file, 'Upload article:', class: 'mt-2' %>
     <%= form.file_field :file, required: true, accept: 'application/pdf', class: 'form-control', direct_upload: true %>