Big refactoring of the PDF snapshotting - into service file

1 files changed, 132 insertions, 0 deletions

diff --git a/app/services/pdf_processor_service.rb b/app/services/pdf_processor_service.rb
new file mode 100644
index 0000000..3f97611
--- /dev/null
+++ b/app/services/pdf_processor_service.rb
@@ -0,0 +1,132 @@
+class PdfProcessorService
+  DEFAULT_OPTIONS = {
+    quality: 40,
+    resize_percentage: 50,
+    format: 'jpg'
+  }.freeze
+
+  def initialize(resource, options = {})
+    @resource = resource
+    @options = DEFAULT_OPTIONS.merge(options)
+  end
+
+  def process_pdfs
+    @resource.pdfs.each do |pdf|
+      pdf_path = get_blob_path(pdf)
+      process_single_pdf(pdf_path)
+    end
+  end
+
+  private
+
+  def get_blob_path(pdf)
+    ActiveStorage::Blob.service.send(:path_for, pdf.key)
+  end
+
+  def process_single_pdf(pdf_path)
+    page_count = get_page_count(pdf_path)
+    
+    (0...page_count).each do |index|
+      Rails.logger.info "Processing page #{index + 1} of #{page_count} from #{pdf_path}"
+      process_page(pdf_path, index)
+    rescue StandardError => e
+      Rails.logger.error "Failed to process page #{index + 1}: #{e.message}"
+      raise
+    end
+  end
+
+  def get_page_count(pdf_path)
+    Rails.logger.debug "Checking PDF: #{pdf_path}"
+    Rails.logger.debug "File exists: #{File.exist?(pdf_path)}"
+    Rails.logger.debug "File size: #{File.size(pdf_path)}"
+    Rails.logger.debug "File type: #{`file -b #{pdf_path}`}"
+    
+    # Try multiple methods to get page count
+    identify_output = `identify -format %n "#{pdf_path}" 2>&1`
+    gs_output = `gs -q -dNODISPLAY -c "#{pdf_path} (r) file runpdfbegin pdfpagecount = quit" 2>&1`
+    
+    Rails.logger.debug "Identify output: #{identify_output}"
+    Rails.logger.debug "Ghostscript output: #{gs_output}"
+    
+    count = identify_output.to_i
+    if count <= 0
+      count = gs_output.to_i
+    end
+    
+    Rails.logger.debug "Final page count: #{count}"
+    raise "Invalid page count: #{count}" unless count.positive?
+    count
+  rescue StandardError => e
+    Rails.logger.error "Failed to get page count: #{e.message}"
+    raise
+  end
+
+  def process_page(pdf_path, page_index)
+    temp_files = create_page_images(pdf_path, page_index)
+    attach_processed_image(temp_files[:reduced], page_index)
+  ensure
+    cleanup_temp_files(temp_files)
+  end
+
+  def create_page_images(pdf_path, index)
+    original_path = "page-#{index + 1}.#{@options[:format]}"
+    reduced_path = "reduced-page-#{index + 1}.#{@options[:format]}"
+
+    create_original_image(pdf_path, index, original_path)
+    create_reduced_image(original_path, reduced_path)
+
+    { original: original_path, reduced: reduced_path }
+  end
+
+  def create_original_image(pdf_path, index, output_path)
+    MiniMagick::Tool::Convert.new do |convert|
+      convert << "#{pdf_path}[#{index}]"
+      convert << output_path
+    end
+  rescue StandardError => e
+    Rails.logger.error "Failed to create original image for page #{index + 1} from #{pdf_path}: #{e.message}"
+    raise
+  end
+
+  def create_reduced_image(input_path, output_path)
+    MiniMagick::Tool::Convert.new do |convert|
+      convert << input_path
+      convert << "-quality"
+      convert << @options[:quality].to_s
+      convert.strip
+      convert << "-resize"
+      convert << "#{@options[:resize_percentage]}%"
+      convert << output_path
+    end
+  rescue StandardError => e
+    Rails.logger.error "Failed to create reduced image from #{input_path}: #{e.message}"
+    raise
+  end
+
+  def attach_processed_image(image_path, index)
+    return unless File.exist?(image_path)
+
+    filename = generate_filename(index)
+    image_blob = create_blob(image_path, filename)
+    @resource.pdf_snapshots.attach(image_blob)
+  end
+
+  def generate_filename(index)
+    base_name = @resource.name.strip.downcase.gsub(/\s/, "_")
+    "#{base_name}_page-#{index + 1}.#{@options[:format]}"
+  end
+
+  def create_blob(path, filename)
+    ActiveStorage::Blob.create_and_upload!(
+      io: File.open(path),
+      filename: filename,
+      content_type: "image/#{@options[:format]}"
+    )
+  end
+
+  def cleanup_temp_files(temp_files)
+    temp_files.values.each do |path|
+      File.delete(path) if File.exist?(path)
+    end
+  end
+end 
\ No newline at end of file