PDF page to image conversion on upload

app/controllers/public/resources_controller.rb

@@ -25,30 +25,53 @@ def iiif
     page_number = params[:page] || 1
 
     redirect_resource do |resource|
-      resource.image? ? resource.content_converted_iiif_url(page_number) : resource.content_iiif_url(page_number)
+      if resource.converted_pages?
+        resource.content_converted_pages_iiif_url(page_number.to_i)
+      elsif resource.image?
+        resource.content_converted_iiif_url(page_number)
+      else
+        resource.content_iiif_url(page_number)
+      end
     end
   end
 
   def image_api
     page_number = params[:page] || 1
 
     redirect_resource do |resource|
-      resource.content_image_api_url(
-        page_number,
-        params[:region],
-        params[:size],
-        params[:rotation],
-        params[:quality],
-        params[:format]
-      )
+      if resource.converted_pages?
+        resource.content_converted_pages_image_api_url(
+          page_number,
+          params[:region],
+          params[:size],
+          params[:rotation],
+          params[:quality],
+          params[:format]
+        )
+      else
+        resource.content_image_api_url(
+          page_number,
+          params[:region],
+          params[:size],
+          params[:rotation],
+          params[:quality],
+          params[:format]
+        )
+      end
     end
   end
 
   def info
     page_number = params[:page] || 1
 
     redirect_resource do |resource|
-      resource.image? ? resource.content_converted_info_url(page_number) : resource.content_info_url(page_number)
+      if resource.converted_pages?
+        resource.content_converted_pages_info_url(page_number)
+      elsif resource.image?
+        resource.content_converted_info_url(page_number)
+      else
+        resource.content_info_url(page_number)    
+      end
     end
   end
 
@@ -64,14 +87,30 @@ def manifest
   end
 
   def preview
+    page_number = params[:page] || 1
+
     redirect_resource do |resource|
-      resource.image? ? resource.content_converted_preview_url : resource.content_preview_url
+      if resource.converted_pages?
+        resource.content_converted_pages_preview_url(page_number)
+      elsif resource.image?
+        resource.content_converted_preview_url
+      else
+        resource.content_preview_url
+      end
     end
   end
 
   def thumbnail
+    page_number = params[:page] || 1
+
     redirect_resource do |resource|
-      resource.image? ? resource.content_converted_thumbnail_url : resource.content_thumbnail_url
+      if resource.converted_pages?
+        resource.content_converted_pages_thumbnail_url(page_number)
+      elsif resource.image?
+        resource.content_converted_thumbnail_url
+      else
+        resource.content_thumbnail_url
+      end
     end
   end
 

app/jobs/convert_image_job.rb

@@ -5,35 +5,119 @@ class ConvertImageJob < ApplicationJob
 
   # In the event the job is started before the file is fully uploaded, we'll retry
   retry_on Exceptions::FileNotUploadedError, wait: 10.seconds
+  retry_on Exceptions::PDFExtractionError, wait: 10.seconds
+  retry_on Exceptions::PDFPageConversionError, wait: 10.seconds
 
   def perform(resource_id)
-    # Only convert if the passed resource is an image
     resource = Resource.find(resource_id)
-    return unless resource.image?
-
-    # Only convert if there is content attached
+
+    # Return early if no content is attached
     content = resource.content
     return unless content.attached?
 
     raise Exceptions::FileNotUploadedError unless resource.content_uploaded?
 
+    # Route to appropriate converter based on content type
+    if resource.image?
+      convert_image(resource)
+    elsif resource.pdf?
+      convert_pdf(resource)
+    end
+  end
+
+  private
+
+  # Convert a single image file to TIFF
+  # Maintains backwards compatibility with existing image conversion workflow
+  # @param resource [Resource] The resource containing the image to convert
+  def convert_image(resource)
+    content = resource.content
+
     content.open do |file|
       begin
-        # Convert the image
+        # Convert the image to TIFF using existing service
         filepath = Images::Convert.to_tiff(file)
         filename = Images::Convert.filename content.filename.to_s, FILE_EXTENSION_TIFF
 
-        # Upload the converted content
+        # Upload the converted content to content_converted (single file attachment)
         resource.content_converted.attach(
           io: File.open(filepath),
           content_type: CONTENT_TYPE_TIFF,
           filename:,
         )
       rescue MiniMagick::Error => e
-        # Content cannot be converted
-        Rails.logger.error e.message
+        # Content cannot be converted - log and continue (non-fatal)
+        Rails.logger.error "Error converting image for resource #{resource.id}: #{e.message}"
         Rails.logger.error e.backtrace.join("\n")
+      ensure
+        # Clean up temporary file
+        File.delete(filepath) if filepath && File.exist?(filepath)
+      end
+    end
+  end
+
+  # Convert a multi-page PDF to an ordered set of TIFF files
+  # @param resource [Resource] The resource containing the PDF to convert
+  def convert_pdf(resource)
+    content = resource.content
+    temp_files = []
+
+    begin
+      content.open do |file|
+        # Extract page count from PDF
+        page_count = Images::ConvertPdf.page_count(file)
+
+        # Process each page
+        page_count.times do |page_number|
+          begin
+            # Extract page as intermediate image
+            temp_image_path = Images::ConvertPdf.extract_page(file, page_number)
+            temp_files << temp_image_path
+
+            # Convert extracted page to TIFF
+            tiff_path = Images::ConvertPdf.page_to_tiff(File.new(temp_image_path))
+            temp_files << tiff_path
+
+            # Generate filename for this page (e.g., "document_page_001.tif")
+            base_filename = File.basename(content.filename.to_s, '.*')
+            page_filename = "#{base_filename}_page_#{page_number + 1}.#{FILE_EXTENSION_TIFF}"
+
+            # Attach TIFF to content_converted_pages in order
+            resource.content_converted_pages.attach(
+              io: File.open(tiff_path),
+              content_type: CONTENT_TYPE_TIFF,
+              filename: page_filename,
+            )
+
+            # Clean up intermediate files as we go
+            Images::ConvertPdf.cleanup_temp_files([temp_image_path, tiff_path])
+            temp_files -= [temp_image_path, tiff_path]
+
+          rescue Exceptions::PDFExtractionError => e
+            Rails.logger.error "Failed to extract page #{page_number} from PDF for resource #{resource.id}: #{e.message}"
+            # Continue with next page rather than failing entire job
+          rescue Exceptions::PDFPageConversionError => e
+            Rails.logger.error "Failed to convert page #{page_number} to TIFF for resource #{resource.id}: #{e.message}"
+            # Continue with next page rather than failing entire job
+          end
+        end
+
+        # Store page count on resource for tracking
+        resource.update(pages_count: page_count)
+
+        # Regenerate manifest with converted pages
+        CreateManifestJob.perform_later(resource.id)
+
+        Rails.logger.info "Successfully converted PDF resource #{resource.id} with #{page_count} pages"
+
       end
+    rescue Exceptions::EmptyPDFError => e
+      Rails.logger.error "PDF resource #{resource.id} is empty: #{e.message}"
+    rescue Exceptions::PDFExtractionError => e
+      Rails.logger.error "Failed to extract pages from PDF resource #{resource.id}: #{e.message}"
+    ensure
+      # Ensure all temporary files are cleaned up
+      Images::ConvertPdf.cleanup_temp_files(temp_files)
     end
   end
 end

app/models/concerns/attachable.rb

@@ -18,7 +18,12 @@ def delete_attachments
         attachment = self.send(name)
         return unless attachment.attached?
 
-        attachment.purge
+        # Handle both has_one_attached and has_many_attached
+        if attachment.is_a?(ActiveStorage::Attached::Many)
+          attachment.purge_all
+        else
+          attachment.purge
+        end
       end
     end
 
@@ -31,14 +36,21 @@ def set_attachment_keys
 
       self.class.list_attachments&.each do |name|
         attachment = self.send(name)
-
-        # Only set the storage key for a new record
-        next unless attachment.new_record?
-
-        # If the "storage_key" attribute is set on the metadata, this blob was created as a direct upload
-        next if attachment.metadata && attachment.metadata[:storage_key].present?
-
-        attachment.key = "#{self.storage_key}/#{ActiveStorage::Blob.generate_unique_secure_token}"
+
+        # Handle both has_one_attached and has_many_attached
+        if attachment.is_a?(ActiveStorage::Attached::Many)
+          # For has_many_attached, iterate through each attachment
+          attachment.each do |attach|
+            next unless attach.new_record?
+            next if attach.metadata && attach.metadata[:storage_key].present?
+            attach.key = "#{self.storage_key}/#{ActiveStorage::Blob.generate_unique_secure_token}"
+          end
+        else
+          # For has_one_attached
+          next unless attachment.new_record?
+          next if attachment.metadata && attachment.metadata[:storage_key].present?
+          attachment.key = "#{self.storage_key}/#{ActiveStorage::Blob.generate_unique_secure_token}"
+        end
       end
     end
   end
@@ -47,18 +59,20 @@ def set_attachment_keys
 
     # This method overrides model#has_one_attached in order to facilitate generating a <name>_url method for easily
     # accessing the attachment URL in serializers.
-    def has_one_attached(name, dependent: :purge_later)
-      super
-      generate_url_method name
+    # Pass generate_urls: false when the model defines its own <name>_*_url methods (e.g. page-indexed URLs).
+    def has_one_attached(name, dependent: :purge_later, generate_urls: true)
+      super(name, dependent: dependent)
+      generate_url_method name if generate_urls
       generate_remove_method name
       @attachments << name
     end
 
     # This method overrides model#has_many_attached in order to facilitate generating a <name>_url method for easily
     # accessing the attachment URL in serializers.
-    def has_many_attached(name, dependent: :purge_later)
-      super
-      generate_url_method name
+    # Pass generate_urls: false when the model defines its own <name>_*_url methods (e.g. page-indexed URLs).
+    def has_many_attached(name, dependent: :purge_later, generate_urls: true)
+      super(name, dependent: dependent)
+      generate_url_method name if generate_urls
       generate_remove_method name
       @attachments << name
     end
@@ -163,7 +177,16 @@ def generate_url_method(name)
     end
 
     def attachment_preloads
-      @attachments.map{ |a| { "#{a}_attachment".to_sym => :blob } }
+      @attachments.map do |a|
+        # For has_many_attached, the association is plural: {name}_attachments
+        # For has_one_attached, the association is singular: {name}_attachment
+        # Check if the reflection exists to determine which pattern to use
+        if reflect_on_association("#{a}_attachments")
+          { "#{a}_attachments".to_sym => :blob }
+        else
+          { "#{a}_attachment".to_sym => :blob }
+        end
+      end
     end
 
     def list_attachments