Extracting TIFF splitting logic from IIIF Print

As I'm thinking through the logic, I'm realizing that we have a splitter
and a strategy;  the splitter leverages the strategy.  As implemented
the splitter is the strategy is the splitter.

Related to:

- scientist-softserv/iiif_print#194
- samvera/bulkrax#760
- scientist-softserv/utk-hyku#343
- https://github.com/scientist-softserv/adventist-dl/issues/330

lib/space_stone/pdf_splitter/pdf_pages_summary.rb

@@ -8,20 +8,21 @@ module PdfSplitter
     PdfPagesSummary = Struct.new(
       :path, :page_count, :width,
       :height, :pixels_per_inch, :color_description,
-      :channels, :bits, keyword_init: true
+      :channels, :bits_per_channel, keyword_init: true
     ) do
       # @return [Array<String, Integer, Integer>]
       def color
-        [color_description, channels, bits]
+        [color_description, channels, bits_per_channel]
       end
       alias_method :ppi, :pixels_per_inch
+      alias_method :bits, :bits_per_channel
 
       # If the underlying extraction couldn't set the various properties, we likely have an
       # invalid_pdf.
       def valid?
         return false if pdf_pages_summary.color_description.nil?
         return false if pdf_pages_summary.channels.nil?
-        return false if pdf_pages_summary.bits.nil?
+        return false if pdf_pages_summary.bits_per_channel.nil?
         return false if pdf_pages_summary.height.nil?
         return false if pdf_pages_summary.page_count.to_i.zero?
 

lib/space_stone/pdf_splitter/pdf_pages_summary/extractor.rb

@@ -56,7 +56,7 @@ def call
           width = 0
           height = 0
           channels = 0
-          bits = 0
+          bits_per_channel = 0
           pixels_per_inch = 0
           Open3.popen3(command) do |_stdin, stdout, _stderr, _wait_thr|
             stdout.read.split("\n").each_with_index do |line, index|
@@ -70,7 +70,7 @@ def call
               width = cells[COL_WIDTH].to_i if cells[COL_WIDTH].to_i > width
               height = cells[COL_HEIGHT].to_i if cells[COL_HEIGHT].to_i > height
               channels = cells[COL_CHANNELS].to_i if cells[COL_CHANNELS].to_i > channels
-              bits = cells[COL_BITS].to_i if cells[COL_BITS].to_i > bits
+              bits_per_channel = cells[COL_BITS].to_i if cells[COL_BITS].to_i > bits_per_channel
 
               # In the case of poppler version < 0.25, we will have no more than 12 columns.  As such,
               # we need to do some alternative magic to calculate this.
@@ -95,7 +95,7 @@ def call
             height: height,
             color_description: color_description,
             channels: channels,
-            bits: bits
+            bits_per_channel: bits_per_channel
           )
         end
         # rubocop:enable Metrics/AbcSize

lib/space_stone/pdf_splitter/strategies.rb

@@ -10,3 +10,4 @@ module Strategies
 
 require 'space_stone/pdf_splitter/strategies/base'
 require 'space_stone/pdf_splitter/strategies/jpg'
+require 'space_stone/pdf_splitter/strategies/tiff'

lib/space_stone/pdf_splitter/strategies/base.rb

@@ -60,6 +60,13 @@ def invalid_pdf?
         attr_reader :pdf_pages_summary, :tmpdir, :baseid, :pdfpath
         private :pdf_pages_summary, :tmpdir, :baseid, :pdfpath
 
+        # @api private
+        def gsdevice
+          return self.class.gsdevice if self.class.gsdevice
+
+          raise NotImplementedError
+        end
+
         private
 
         # entries for each page
@@ -94,12 +101,6 @@ def gsconvert
         end
         # rubocop:enable Metrics/MethodLength
 
-        def gsdevice
-          return self.class.gsdevice if self.class.gsdevice
-
-          raise NotImplementedError
-        end
-
         PAGE_COUNT_REGEXP = %r{^Pages: +(\d+)$}.freeze
 
         def pagecount

lib/space_stone/pdf_splitter/strategies/tiff.rb

@@ -0,0 +1,47 @@
+# frozen_string_literal: true
+
+module SpaceStone
+  module PdfSplitter
+    module Strategies
+      # The purpose of this class is to split the PDF into constituent tiff files.
+      class Tiff < Strategies::Base
+        self.image_extension = 'tiff'
+        self.compression = 'lzw'
+
+        ##
+        # @api private
+        #
+        # @param gsdevice [Object]
+        def gsdevice
+          return @gsdevice if defined?(@gsdevice)
+
+          color = pdf_pages_summary.color_description
+          channels = pdf_pages_summary.channels
+          bpc = pdf_pages_summary.bits_per_channel
+
+          if color == 'gray'
+            # CCITT Group 4 Black and White, if applicable:
+            if bpc == 1
+              self.compression = 'g4'
+              return @gsdevice = 'tiffg4'
+            elsif bpc > 1
+              # 8 Bit Grayscale, if applicable:
+              return @gsdevice = 'tiffgray'
+            end
+          end
+
+          # otherwise color:
+          @gsdevice = colordevice(channels, bpc)
+        end
+
+        def colordevice(channels, bpc)
+          bits = bpc * channels
+          # will be either 8bpc/16bpd color TIFF,
+          #   with any CMYK source transformed to 8bpc RBG
+          bits = 24 unless [24, 48].include? bits
+          "tiff#{bits}nc"
+        end
+      end
+    end
+  end
+end

spec/space_stone/pdf_splitter/pdf_pages_summary/extractor_spec.rb

@@ -26,6 +26,18 @@
         expect(subject.width).to eq(3)
       end
 
+      it 'determines the bits_per_channel of the given PDF' do
+        expect(subject.bits_per_channel).to eq(8)
+      end
+
+      it 'determines the color_description of the given PDF' do
+        expect(subject.color_description).to eq('rgb')
+      end
+
+      it 'determines the channels of the given PDF' do
+        expect(subject.channels).to eq(3)
+      end
+
       it 'determines the color of the given PDF' do
         expect(subject.color).to eq(['rgb', 3, 8])
       end

spec/space_stone/pdf_splitter/pdf_pages_summary_spec.rb

@@ -17,7 +17,8 @@
 
   subject do
     described_class.new(path: __FILE__, page_count: 1, width: 2, height: 3,
-                        pixels_per_inch: 4, color_description: 'rgb', channels: 5, bits: 6)
+                        pixels_per_inch: 4, color_description: 'rgb', channels: 5,
+                        bits_per_channel: 6)
   end
 
   it { is_expected.to respond_to(:valid?) }
@@ -31,4 +32,5 @@
   it { is_expected.to respond_to(:color_description) }
   it { is_expected.to respond_to(:channels) }
   it { is_expected.to respond_to(:bits) }
+  it { is_expected.to respond_to(:bits_per_channel) }
 end

spec/space_stone/pdf_splitter/strategies/base_spec.rb

@@ -14,7 +14,7 @@
 
   describe '#gsdevice' do
     it "expects that you will have set .gsdevice in the subclass" do
-      expect { subject.send(:gsdevice) }.to raise_error(NotImplementedError)
+      expect { subject.gsdevice }.to raise_error(NotImplementedError)
     end
   end
 end

spec/space_stone/pdf_splitter/strategies/jpg_spec.rb

@@ -13,7 +13,7 @@
   end
 
   describe '#gsdevice' do
-    subject { splitter.send(:gsdevice) }
+    subject { splitter.gsdevice }
     it { is_expected.to eq('jpeg') }
   end
 

spec/space_stone/pdf_splitter/strategies/tiff_spec.rb

@@ -0,0 +1,49 @@
+# frozen_string_literal: true
+require 'spec_helper'
+
+RSpec.describe SpaceStone::PdfSplitter::Strategies::Tiff do
+  let(:path) { __FILE__ }
+  let(:pdf_pages_summary) { double(SpaceStone::PdfSplitter::PdfPagesSummary) }
+
+  let(:splitter) { described_class.new(path, pdf_pages_summary: pdf_pages_summary) }
+
+  describe '.compression' do
+    subject { described_class.compression }
+    it { is_expected.to eq('lzw') }
+  end
+
+  describe '.compression?' do
+    subject { described_class.compression? }
+    it { is_expected.to be_truthy }
+  end
+
+  describe '.image_extension' do
+    subject { described_class.image_extension }
+    it { is_expected.to eq('tiff') }
+  end
+
+  describe '#gsdevice' do
+    DEFAULT_SUMMARY_ATTRIBUTES = {
+      page_count: 10,
+      color_description: 'rgb',
+      bits_per_channel: 0,
+      channels: 0
+    }.freeze
+
+    [
+      [{ color_description: 'gray', bits_per_channel: 2 }, 'tiffgray'],
+      [{ color_description: 'gray', bits_per_channel: 1 }, 'tiffg4'],
+      [{ color_description: 'rgb', bits_per_channel: 1 }, 'tiff24nc'],
+      [{ color_description: 'rgb', channels: 8, bits_per_channel: 6 }, 'tiff48nc'],
+      [{ color_description: 'rgb', channels: 8, bits_per_channel: 5 }, 'tiff24nc'],
+      [{ color_description: 'rgb', channels: 8, bits_per_channel: 3 }, 'tiff24nc']
+    ].each do |attributes, expected_value|
+      context 'with #{attributes.inspect}' do
+        it "is expected to be #{expected_value.inspect}" do
+          summary = SpaceStone::PdfSplitter::PdfPagesSummary.new(**DEFAULT_SUMMARY_ATTRIBUTES.merge(attributes))
+          expect(described_class.new(__FILE__, pdf_pages_summary: summary).gsdevice).to eq(expected_value)
+        end
+      end
+    end
+  end
+end