Derivative generation on Heroku

Types of original images in the digital collections

TIFF
- Black and white colorspace
- RGB colorspace
PDF

Software needed to generate derivatives:

mediainfo
convert
pdfunite
vips
ffmpeg

Diagnostics for derivative generation software

~~Run the following commands in a heroku dyno.~~

This is now automated in a suite of tests you can run with. ./bin/rspec system_env_spec. See https://github.com/sciencehistory/scihist_digicoll/blob/master/system_env_spec/README.md Below for historical purposes only.

Your results may vary slightly, but anything that is way off should be seen as a red flag.

heroku run bash

mediainfo --version
# Normal output:
# MediaInfo Command line,
# MediaInfoLib - v19.09

convert -version
# Normal output:
# Version: ImageMagick 6.9.10-23 Q16 x86_64 20190101 https://imagemagick.org
# Copyright: © 1999-2019 ImageMagick Studio LLC
# License: https://imagemagick.org/script/license.php
# Features: Cipher DPC Modules OpenMP
# Delegates (built-in): bzlib djvu fftw fontconfig freetype jbig jng jpeg lcms lqr ltdl lzma openexr pangocairo png tiff webp wmf x xml zlib

pdfunite -v
# Normal output:
# pdfunite version 0.86.1

vips --version
# Normal output:
# vips-8.10.6-Tue Mar 23 20:52:58 UTC 2021

ffmpeg -version
# Normal output:
# ffmpeg version 4.2.3 Copyright (c) 2000-2020 the FFmpeg developers
# built with gcc 5.4.0 (Ubuntu 5.4.0-6ubuntu1~16.04.12) 20160609
# configuration: --prefix=/home/work/sffmpeg/build --datadir=/home/work/sffmpeg/build/etc --disable-shared --enable-static --enable-pic --pkg-config-flags=--static --enable-gpl --enable-version3 --disable-doc --disable-debug --disable-ffplay --disable-outdevs --enable-runtime-cpudetect --extra-cflags='-I/home/work/sffmpeg/build/include -static' --extra-ldflags=-L/home/work/sffmpeg/build/lib --extra-ldexeflags=-static --extra-libs='-lstdc++ -lexpat -ldl -lm -lpthread' --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libaom --enable-libmp3lame --enable-libspeex --enable-libtheora --enable-libvorbis --enable-libx264 --enable-libx265 --enable-libvpx --enable-libopus --enable-libfreetype --enable-libass --enable-mbedtls
# libavutil      56. 31.100 / 56. 31.100
# libavcodec     58. 54.100 / 58. 54.100
# libavformat    58. 29.100 / 58. 29.100
# libavdevice    58.  8.100 / 58.  8.100
# libavfilter     7. 57.100 /  7. 57.100
# libswscale      5.  5.100 /  5.  5.100
# libswresample   3.  5.100 /  3.  5.100
# libpostproc    55.  5.100 / 55.  5.100


vips -l | grep -o '[a-z_]*pdf[a-z_]*'
# Normal output:
# pdfload_base
# pdfload
# pdf
# pdfload_buffer
# pdfload_source

cd tmp

PROFILE=`ls ../vendor/bundle/ruby/*/gems/kithe-*/lib/vendor/icc/sRGB2014.icc`
wget https://digital.sciencehistory.org/downloads/m3zcuho -O b_w.tiff
wget https://digital.sciencehistory.org/downloads/1h16b9n -O color.tiff
wget https://digital.sciencehistory.org/downloads/519ucnx -O normal.pdf

vipsthumbnail  color.tiff  --eprofile $PROFILE
vipsthumbnail  b_w.tiff    --eprofile $PROFILE
vipsthumbnail  normal.pdf

identify *.jpg | grep sRGB

# Normal output (ignore warnings):
# tn_b_w.jpg    JPEG 128x108 128x108+0+0 8-bit sRGB 23603B 0.000u 0:00.000
# tn_color.jpg  JPEG 128x96 128x96+0+0   8-bit sRGB 11880B 0.000u 0:00.000
# tn_normal.jpg JPEG 99x128 99x128+0+0   8-bit sRGB  1619B 0.010u 0:00.000

Software setups

We’ve been through several of these since starting to investigate Heroku.

#	Aptfile	Buildpack	Results
1	libvips-tools mediainfo imagemagick poppler-utils	heroku-community/apt heroku/ruby	`vips-8.9.1-Sun Feb 23 08:51:26 UTC 2020` Color TIFFs work B&W TIFFs do not work PDFs work
2	libvips-tools mediainfo imagemagick poppler-utils	heroku-community/apt heroku/ruby	`vips-8.9.1-Sun Feb 23 08:51:26 UTC 2020` PDFs work All TIFFS work Removing `--eprofile srgb_profile_path` from the arguments to `vipsthumbnail` in the code (docs) avoids the error described in issue 942.
3	mediainfo imagemagick poppler-utils	heroku-community/apt https://github.com/machinio/heroku-buildpack-vips heroku/ruby	`vips-8.10.2-Mon Oct 12 16:43:59 UTC 2020` At some point in 2021, `vips -l \| grep -i pdf` started returning blank - no `poppler` support, so PDFs don’t work. All TIFFS work
4	mediainfo imagemagick libglib2.0-0 libglib2.0-dev libpoppler-glib8	heroku-community/apt https://github.com/brandoncc/heroku-buildpack-vips heroku/ruby	`vips-8.10.6-Tue Mar 23 20:52:58 UTC 2021` PDFs work All TIFFS work Combined audio derivatives don’t work
5	mediainfo imagemagick libglib2.0-0 libglib2.0-dev libpoppler-glib8	https://github.com/heroku/heroku-buildpack-activestorage-preview https://buildpack-registry.s3.amazonaws.com/buildpacks/heroku-community/apt.tgz https://github.com/brandoncc/heroku-buildpack-vips heroku/ruby	`vips-8.10.6-Tue Mar 23 20:52:58 UTC 2021` PDFs work All TIFFS work Combined audio derivatives work again (see issue 1448) `libpoppler-glib8` in the aptfile may not be needed (see issue 1455 )
6	`mediainfo` `imagemagick` `libglib2.0-0` `libglib2.0-dev` `libpoppler-glib8` `poppler-utils`	`https://github.com/heroku/heroku-buildpack-activestorage-preview` `heroku-community/apt` `https://github.com/brandoncc/heroku-buildpack-vips` `heroku/ruby`	PDF on-demand stopped working in staging, so we added poppler-utils back into aptfile.
7	`mediainfo` `imagemagick` `libglib2.0-0` `libglib2.0-dev` `libpoppler-glib8` `qpdf` `tesseract-ocr` `tesseract-ocr-eng` `tesseract-ocr-deu` `tesseract-ocr-fra` `tesseract-ocr-spa` `tesseract-error-while-loading-shared-libraries-libarchive-so-13-python` `libarchive13`	`heroku/python` `https://github.com/heroku/heroku-buildpack-activestorage-preview` `https://buildpack-registry.s3.amazonaws.com/buildpacks/heroku-community/apt.tgz` `https://github.com/brandoncc/heroku-buildpack-vips` `https://github.com/fnando/heroku-buildpack-exiftool` `heroku/ruby`	OCR. Note we are removing `poppler-utils`

Note: Row 2 was a band-aid; it violated the rule implicit in the code that all TIFF derivatives should have their derivatives encoded as srgb, including the derivatives of B&W originals. I interpret the documentation as meaning that the icc profile of originals is reused in their derivatives, but further research is needed.
identify -verbose graphics_file.tiff | grep Colorspace can be used to elucidate what happens to various types of original after being processed by vipsthumbnail.