replaces kakadu with grok using HT's settings

relevant HT PRs:
* hathitrust/feed#95
* hathitrust/feed#172

grokj2k is available in the mlibrary apt repo

Author: niquerio

Dockerfile

@@ -8,17 +8,16 @@ ARG FEED_VERSION=feed_v1.14.1
 RUN gem install bundler
 
 WORKDIR /tmp
-RUN wget https://kakadusoftware.com/wp-content/uploads/$KAKADU_FILE
-RUN unzip -j -d kakadu $KAKADU_FILE
-RUN mv /tmp/kakadu/*.so /usr/local/lib
-RUN mv /tmp/kakadu/kdu* /usr/local/bin
-RUN echo "/usr/local/lib" > /etc/ld.so.conf.d/kakadu.conf
-RUN ldconfig
+
+RUN curl https://apt.lib.umich.edu/mlibrary-archive-keyring.gpg -o /etc/apt/keyrings/mlibrary-archive-keyring.gpg
+RUN echo "deb [signed-by=/etc/apt/keyrings/mlibrary-archive-keyring.gpg] https://apt.lib.umich.edu bookworm main" > /etc/apt/sources.list.d/mlibrary.list
 
 RUN apt-get update -yqq && apt-get install -yqq --no-install-recommends \
     libtiff-tools\ 
     exiftool \
-    netpbm
+    netpbm\
+    grokj2k
+
 
 ENV APP_PATH=/usr/src/app
 RUN mkdir -p $APP_PATH

lib/compressor.rb

@@ -1,23 +1,15 @@
 require "tiff"
 
-module Kakadu
+module Grok
   JP2_LEVEL_MIN = 5
-  JP2_LAYERS = 8
-  JP2_ORDER = "RLCP"
-  JP2_USE_SOP = "yes"
-  JP2_USE_EPH = "yes"
-  JP2_MODES = '"RESET|RESTART|CAUSAL|ERTERM|SEGMARK"'
-  JP2_SLOPE = 42_988
+
+  # Settings for grk_compress recommended from Roger Espinosa. "-slope"
+  # is a VBR compression mode; the value of 42988 corresponds to pre-6.4
+  # slope of 51180, the current (as of 5/6/2011) recommended setting for
+  # Google digifeeds.
   def self.compress(source, destination, tiffinfo)
     clevels = jp2_clevels(tiffinfo)
-    cmd = "kdu_compress -quiet -i #{source} -o #{destination}" \
-          " 'Clevels=#{clevels}'" \
-          " 'Clayers=#{JP2_LAYERS}'" \
-          " 'Corder=#{JP2_ORDER}'" \
-          " 'Cuse_sop=#{JP2_USE_SOP}'" \
-          " 'Cuse_eph=#{JP2_USE_EPH}'" \
-          " Cmodes=#{JP2_MODES}" \
-          " -no_weights -slope '#{JP2_SLOPE}'"
+    cmd = "grk_compress -i \"#{source}\" -o \"#{destination}\" -p RLCP -n #{clevels} -S -E -M 62 -I -q 32"
     status = Command.new(cmd).run
     LogEntry.info(command: cmd, time: status[:time])
   end
@@ -213,7 +205,7 @@ def compression_type
     "JP2"
   end
 
-  def run(compression_tool = Kakadu)
+  def run(compression_tool = Grok)
     # We don't want any XMP metadata to be copied over on its own. If
     # it's been a while since we last ran exiftool, this might take a sec.
     log_it ExifTool.remove_tiff_metadata(source: image_file.path, destination: sparse_path)

lib/stage/dlxs_compressor.rb

@@ -126,7 +126,7 @@ def pgm_to_bitonal(dir)
 
   # Expand existing jp2 into tif in temp directory
   def expand_jp2(src, dest)
-    cmd = "kdu_expand -quiet -i '#{src}' -o '#{dest}'"
+    cmd = "grk_decompress  -i '#{src}' -o '#{dest}'"
     status = Command.new(cmd).run
     log cmd, status[:time]
   end

spec/compressor_spec.rb

@@ -63,7 +63,7 @@ def tiffinfo(path)
 
       it "runs the compression tool" do
         compressor.run
-        expect(log.entries).to include(match("kdu_compress"))
+        expect(log.entries).to include(match("grk_compress"))
       end
 
       it "copies original metadata to the jpeg2000" do