Refactor validation service

app/controllers/bulkrax/guided_imports_controller.rb

@@ -85,7 +85,7 @@ def render_invalid_uploaded_files_response
     # @param admin_set_id [String, nil] optional admin set ID for validation context
     # @return [Hash] validation result data
     def run_validation(csv_file, zip_file, admin_set_id: nil)
-      CsvValidationService.validate(csv_file: csv_file, zip_file: zip_file, admin_set_id: admin_set_id)
+      CsvParser.validate_csv(csv_file: csv_file, zip_file: zip_file, admin_set_id: admin_set_id)
     end
 
     def importer_params

app/controllers/bulkrax/importers_controller.rb

@@ -70,7 +70,7 @@ def new
     # GET /importers/sample_csv_file
     def sample_csv_file
       admin_set_id = params[:admin_set_id].presence
-      sample = Bulkrax::CsvValidationService.generate_template(models: 'all', output: 'file', admin_set_id: admin_set_id)
+      sample = Bulkrax::CsvParser.generate_template(models: 'all', output: 'file', admin_set_id: admin_set_id)
       send_file sample, filename: File.basename(sample), type: 'text/csv', disposition: 'attachment'
     rescue StandardError => e
       flash[:error] = "Unable to generate sample CSV file: #{e.message}"

app/parsers/bulkrax/csv_parser.rb

@@ -4,7 +4,10 @@ module Bulkrax
   class CsvParser < ApplicationParser # rubocop:disable Metrics/ClassLength
     include ErroredEntries
     include ExportBehavior
+    include CsvParser::CsvTemplateGeneration
+    include CsvParser::CsvValidation
     attr_writer :collections, :file_sets, :works
+    attr_accessor :validation_mode
 
     def self.export_supported?
       true
@@ -15,8 +18,10 @@ def records(_opts = {})
 
       file_for_import = only_updates ? parser_fields['partial_import_file_path'] : import_file_path
       csv_data = entry_class.read_data(file_for_import)
-      importer.parser_fields['total'] = csv_data.count
-      importer.save
+      unless validation_mode
+        importer.parser_fields['total'] = csv_data.count
+        importer.save
+      end
 
       @records = csv_data.map { |record_data| entry_class.data_for_entry(record_data, nil, self) }
       @records

app/parsers/concerns/bulkrax/csv_parser/csv_template_generation.rb

@@ -0,0 +1,73 @@
+# frozen_string_literal: true
+
+module Bulkrax
+  class CsvParser < ApplicationParser
+    module CsvTemplateGeneration
+      extend ActiveSupport::Concern
+
+      class_methods do
+        # Generate a CSV template for the specified models.
+        #
+        # @param models [Array<String>, String] Model names or 'all' for all available models
+        # @param output [String] Output format: 'file' or 'csv_string'
+        # @param admin_set_id [String, nil] Optional admin set ID for context
+        # @param args [Hash] Additional arguments passed to output method (e.g., file_path)
+        # @return [String] File path (for 'file' output) or CSV string (for 'csv_string' output)
+        def generate_template(models: [], output: 'file', admin_set_id: nil, **args)
+          raise NameError, "Hyrax is not defined" unless defined?(::Hyrax)
+          TemplateContext.new(models: models, admin_set_id: admin_set_id).send("to_#{output}", **args)
+        end
+      end
+
+      ##
+      # Holds state for a single template generation run.
+      # Provides the interface expected by CsvTemplate:: components.
+      class TemplateContext
+        attr_reader :mappings, :all_models, :admin_set_id, :field_analyzer, :mapping_manager
+
+        def initialize(models: nil, admin_set_id: nil)
+          @admin_set_id = admin_set_id
+          @mapping_manager = CsvTemplate::MappingManager.new
+          @mappings = @mapping_manager.mappings
+          @field_analyzer = CsvTemplate::FieldAnalyzer.new(@mappings, admin_set_id)
+          @all_models = CsvTemplate::ModelLoader.new(Array.wrap(models)).models
+          @csv_builder = CsvTemplate::CsvBuilder.new(self)
+        end
+
+        def to_file(file_path: nil)
+          file_path ||= CsvTemplate::FilePathGenerator.default_path(@admin_set_id)
+          @csv_builder.write_to_file(file_path)
+          file_path
+        end
+
+        def to_csv_string
+          @csv_builder.generate_string
+        end
+
+        def field_metadata_for_all_models
+          @field_metadata ||= @all_models.each_with_object({}) do |model, hash|
+            field_list = @field_analyzer.find_or_create_field_list_for(model_name: model)
+            hash[model] = {
+              properties: field_list.dig(model, "properties") || [],
+              required_terms: field_list.dig(model, "required_terms") || [],
+              controlled_vocab_terms: field_list.dig(model, "controlled_vocab_terms") || []
+            }
+          end
+        end
+
+        def valid_headers_for_models
+          @valid_headers ||= begin
+            column_builder = CsvTemplate::ColumnBuilder.new(self)
+            all_columns = column_builder.all_columns
+            all_columns - CsvTemplate::CsvBuilder::IGNORED_PROPERTIES
+                             rescue StandardError => e
+                               Rails.logger.error("Error building valid headers: #{e.message}")
+                               standard_fields = %w[model source_identifier parent parents file]
+                               model_fields = field_metadata_for_all_models.values.flat_map { |m| m[:properties] }
+                               (standard_fields + model_fields).uniq
+          end
+        end
+      end
+    end
+  end
+end

app/parsers/concerns/bulkrax/csv_parser/csv_validation.rb

@@ -0,0 +1,257 @@
+# frozen_string_literal: true
+
+module Bulkrax
+  class CsvParser < ApplicationParser
+    module CsvValidation # rubocop:disable Metrics/ModuleLength
+      extend ActiveSupport::Concern
+
+      included do
+        # Lightweight struct used to satisfy the CsvTemplate::ColumnBuilder
+        # interface without constructing a full template context.
+        ValidationContext = Struct.new(:mapping_manager, :field_analyzer, :all_models, :mappings, keyword_init: true)
+      end
+
+      class_methods do
+        # Validate a CSV (and optional zip) without a persisted Importer record.
+        #
+        # @param csv_file [File, ActionDispatch::Http::UploadedFile, String] path or file object
+        # @param zip_file [File, ActionDispatch::Http::UploadedFile, nil]
+        # @param admin_set_id [String, nil]
+        # @return [Hash] validation result compatible with the guided import UI
+        def validate_csv(csv_file:, zip_file: nil, admin_set_id: nil) # rubocop:disable Metrics/MethodLength, Metrics/AbcSize
+          file_path = csv_file.respond_to?(:path) ? csv_file.path : csv_file.to_s
+
+          # 1. Read headers — use CsvEntry.read_data so header normalisation
+          #    (special-char stripping, symbolisation) is identical to a real import.
+          raw_csv = CsvEntry.read_data(file_path)
+          headers = raw_csv.headers.map(&:to_s)
+
+          # 2. Field mappings / column name resolution
+          mapping_manager = CsvTemplate::MappingManager.new
+          mappings        = mapping_manager.mappings
+
+          source_id_key  = resolve_validation_key(mapping_manager, flag: 'source_identifier',                    default: :source_identifier)
+          parent_key     = resolve_validation_key(mapping_manager, flag: 'related_parents_field_mapping',        default: :parents)
+          children_key   = resolve_validation_key(mapping_manager, flag: 'related_children_field_mapping',       default: :children)
+          file_key       = resolve_validation_key(mapping_manager, key: 'file',                                  default: :file)
+
+          # 3. Parse rows — CsvEntry.read_data already filters blank rows and
+          #    returns symbol-keyed rows (same as a real import).
+          csv_data = parse_validation_rows(raw_csv, source_id_key, parent_key, children_key, file_key)
+
+          # 4. Field metadata
+          all_models     = csv_data.map { |r| r[:model] }.compact.uniq
+          field_analyzer = CsvTemplate::FieldAnalyzer.new(mappings, admin_set_id)
+          field_metadata = build_validation_field_metadata(all_models, field_analyzer)
+
+          # 5. Valid-header set (drives unrecognised-header detection)
+          valid_headers = build_valid_validation_headers(mapping_manager, field_analyzer, all_models, mappings, field_metadata)
+
+          # 6. Suffixed variants seen in this specific CSV (e.g. title_1, creator_2)
+          suffixed_headers = headers.select { |h| h.match?(/_\d+\z/) }
+          valid_headers = (valid_headers + suffixed_headers).uniq
+
+          # 7. Header-level checks
+          missing_required = find_missing_required_headers(headers, field_metadata, mapping_manager)
+          unrecognized     = find_unrecognized_validation_headers(headers, valid_headers)
+
+          # 8. Row-level validators
+          parent_split      = resolve_parent_split_pattern(mappings)
+          all_ids           = csv_data.map { |r| r[:source_identifier] }.compact.to_set
+          validator_context = {
+            errors: [],
+            warnings: [],
+            seen_ids: {},
+            all_ids: all_ids,
+            source_identifier: source_id_key.to_s,
+            parent_split_pattern: parent_split,
+            mappings: mappings,
+            field_metadata: field_metadata
+          }
+
+          csv_data.each_with_index do |record, index|
+            row_number = index + 2 # 1-indexed, plus header row
+            Bulkrax.csv_row_validators.each { |v| v.call(record, row_number, validator_context) }
+          end
+
+          # 9. File validation
+          file_validator = CsvTemplate::FileValidator.new(csv_data, zip_file, admin_set_id)
+
+          # 10. Item hierarchy for UI display
+          collections, works, file_sets = extract_validation_items(csv_data)
+
+          # 11. Assemble result
+          row_errors  = validator_context[:errors]
+          has_errors  = missing_required.any? || headers.blank? || csv_data.empty? ||
+                        file_validator.missing_files.any? || row_errors.any?
+          has_warnings = unrecognized.any? || file_validator.possible_missing_files?
+
+          result = {
+            headers: headers,
+            missingRequired: missing_required,
+            unrecognized: unrecognized,
+            rowCount: csv_data.length,
+            isValid: !has_errors,
+            hasWarnings: has_warnings,
+            rowErrors: row_errors,
+            collections: collections,
+            works: works,
+            fileSets: file_sets,
+            totalItems: csv_data.length,
+            fileReferences: file_validator.count_references,
+            missingFiles: file_validator.missing_files,
+            foundFiles: file_validator.found_files_count,
+            zipIncluded: file_validator.zip_included?
+          }
+
+          apply_rights_statement_validation_override!(result, missing_required)
+          result
+        end
+
+        private
+
+        # Resolve a symbol key from mappings for use as a record hash key.
+        # Returns a Symbol matching the parser's symbol-keyed record hash.
+        def resolve_validation_key(mapping_manager, key: nil, flag: nil, default:)
+          options = mapping_manager.resolve_column_name(key: key, flag: flag, default: default.to_s)
+          options.first&.to_sym || default
+        end
+
+        # Parse rows from a CsvEntry.read_data result into the canonical record shape.
+        # CsvEntry.read_data returns CSV::Row objects with symbol headers; blank rows
+        # are already filtered by CsvWrapper.
+        def parse_validation_rows(raw_csv, source_id_key, parent_key, children_key, file_key)
+          raw_csv.map do |row|
+            # CSV::Row#to_h converts symbol headers → string-keyed hash
+            row_hash = row.to_h.transform_keys(&:to_s)
+            {
+              source_identifier: row[source_id_key],
+              model: row[:model],
+              parent: row[parent_key],
+              children: row[children_key],
+              file: row[file_key],
+              raw_row: row_hash
+            }
+          end
+        rescue StandardError => e
+          Rails.logger.error("CsvParser.validate_csv: error parsing rows – #{e.message}")
+          []
+        end
+
+        def build_validation_field_metadata(all_models, field_analyzer)
+          all_models.each_with_object({}) do |model, hash|
+            field_list = field_analyzer.find_or_create_field_list_for(model_name: model)
+            hash[model] = {
+              properties: field_list.dig(model, 'properties') || [],
+              required_terms: field_list.dig(model, 'required_terms') || [],
+              controlled_vocab_terms: field_list.dig(model, 'controlled_vocab_terms') || []
+            }
+          end
+        end
+
+        def build_valid_validation_headers(mapping_manager, field_analyzer, all_models, mappings, field_metadata)
+          svc = ValidationContext.new(
+            mapping_manager: mapping_manager,
+            field_analyzer: field_analyzer,
+            all_models: all_models,
+            mappings: mappings
+          )
+          all_cols = CsvTemplate::ColumnBuilder.new(svc).all_columns
+          all_cols - CsvTemplate::CsvBuilder::IGNORED_PROPERTIES
+        rescue StandardError => e
+          Rails.logger.error("CsvParser.validate_csv: error building valid headers – #{e.message}")
+          standard = %w[model source_identifier parent parents file]
+          model_fields = field_metadata.values.flat_map { |m| m[:properties] }
+          (standard + model_fields).uniq
+        end
+
+        def find_missing_required_headers(headers, field_metadata, mapping_manager)
+          csv_keys = headers.map { |h| mapping_manager.mapped_to_key(h).sub(/_\d+\z/, '') }.uniq
+          missing = []
+          field_metadata.each do |model, meta|
+            (meta[:required_terms] || []).each do |field|
+              missing << { model: model, field: field } unless csv_keys.include?(field)
+            end
+          end
+          missing.uniq
+        end
+
+        def find_unrecognized_validation_headers(headers, valid_headers)
+          checker = DidYouMean::SpellChecker.new(dictionary: valid_headers)
+          headers
+            .reject { |h| valid_headers.include?(h) || valid_headers.include?(h.sub(/_\d+\z/, '')) }
+            .index_with { |h| checker.correct(h).first }
+        end
+
+        def resolve_parent_split_pattern(mappings)
+          split_val = mappings.dig('parents', 'split') || mappings.dig(:parents, :split)
+          return nil if split_val.blank?
+          return Bulkrax::DEFAULT_MULTI_VALUE_ELEMENT_SPLIT_ON if split_val == true
+
+          split_val
+        end
+
+        def extract_validation_items(csv_data) # rubocop:disable Metrics/MethodLength
+          child_to_parents = build_child_to_parents_map(csv_data)
+          collections = []
+          works       = []
+          file_sets   = []
+
+          csv_data.each do |item|
+            categorise_validation_item(item, child_to_parents, collections, works, file_sets)
+          end
+
+          [collections, works, file_sets]
+        end
+
+        def build_child_to_parents_map(csv_data)
+          Hash.new { |h, k| h[k] = [] }.tap do |map|
+            csv_data.each do |item|
+              parse_relationship_field(item[:children]).each do |child_id|
+                map[child_id] << item[:source_identifier]
+              end
+            end
+          end
+        end
+
+        def categorise_validation_item(item, child_to_parents, collections, works, file_sets)
+          item_id   = item[:source_identifier]
+          title     = item[:raw_row]['title'] || item_id
+          model_str = item[:model].to_s
+
+          if model_str.casecmp('collection').zero? || model_str.casecmp('collectionresource').zero?
+            explicit = parse_relationship_field(item[:parent])
+            inferred = child_to_parents[item_id] || []
+            collections << { id: item_id, title: title, type: 'collection',
+                             parentIds: (explicit + inferred).uniq,
+                             childIds: parse_relationship_field(item[:children]) }
+          elsif model_str.casecmp('fileset').zero? || model_str.casecmp('hyrax::fileset').zero?
+            file_sets << { id: item_id, title: title, type: 'file_set' }
+          else
+            explicit = parse_relationship_field(item[:parent])
+            inferred = child_to_parents[item_id] || []
+            works << { id: item_id, title: title, type: 'work',
+                       parentIds: (explicit + inferred).uniq,
+                       childIds: parse_relationship_field(item[:children]) }
+          end
+        end
+
+        def parse_relationship_field(value)
+          return [] if value.blank?
+          value.to_s.split('|').map(&:strip).reject(&:blank?)
+        end
+
+        def apply_rights_statement_validation_override!(result, missing_required)
+          only_rights = missing_required.present? &&
+                        missing_required.all? { |h| h[:field].to_s == 'rights_statement' }
+          return unless only_rights && !result[:isValid]
+          return if result[:headers].blank?
+          return if result[:missingFiles]&.any?
+
+          result[:isValid]     = true
+          result[:hasWarnings] = true
+        end
+      end
+    end
+  end
+end