.install_tesseract

#! /bin/bash

[ ! -d "$HOME/dependencies" ] && mkdir $HOME/dependencies ; cd $HOME/dependencies

# The $HOME/dependencies directory is configured to be cached and persisted among builds
#  so our logic can determine whether to fully build based on existence.
if [ ! -d "./tesseract" ]
then
  # Presumably the lack of directory means we need to download and build everything
  mkdir ./tesseract ; cd ./tesseract

  echo "Compiling and installing Tesseract and its library dependencies"

  wget "https://github.com/uclouvain/openjpeg/archive/v2.3.1.zip"
  unzip -q v2.3.1.zip
  cd openjpeg-v2.3.1
  mkdir build
  cd build
  which cmake
  cmake ..
  make
  sudo make install
  sudo make clean

  cd $HOME/dependencies/tesseract

  wget "https://github.com/DanBloomberg/leptonica/releases/download/1.79.0/leptonica-1.79.0.tar.gz"
  tar xzf leptonica-1.79.0.tar.gz
  cd leptonica-1.79.0
  sed -i 's/#define  HAVE_LIBJP2K     0/#define  HAVE_LIBJP2K     1/g' ./src/environ.h
  sed -i 's/-ltiff -ljpeg -lpng -lz -lm/-ltiff -ljpeg -lpng -lz -lm -lopenjp2/g' ./prog/makefile.static
  ./configure
  make
  sudo make install

  cd $HOME/dependencies/tesseract

  wget "https://github.com/tesseract-ocr/tesseract/archive/3.05.02.zip"
  unzip -q 3.05.02.zip
  cd tesseract-3.05.02
  ./autogen.sh
  ./configure
  make
  sudo make install
  sudo ldconfig
  cd $HOME/dependencies/tesseract

  echo "Obtaining Tesseract trained language data"
  mkdir ./tessdata
  curl -L --output ./tessdata/eng.traineddata https://github.com/tesseract-ocr/tessdata/raw/master/eng.traineddata
  curl -L --output ./tessdata/ita.traineddata https://github.com/tesseract-ocr/tessdata/raw/master/ita.traineddata
  # git clone https://github.com/tesseract-ocr/tessdata.git
  sudo cp tessdata/eng.* /usr/local/share/tessdata/
  sudo cp tessdata/ita* /usr/local/share/tessdata
else
  # We've already done the build before so just install libraries

  echo "Installing Tesseract library dependencies"
  cd ./tesseract

  cd openjpeg-v2.3.1/build
  sudo make install
  cd $HOME/dependencies/tesseract

  cd leptonica-1.79.0
  sudo make install
  cd $HOME/dependencies/tesseract

  cd tesseract-3.05.02
  sudo make install
  sudo ldconfig
  cd $HOME/dependencies/tesseract

  echo "Obtaining Tesseract trained language data"
  sudo cp tessdata/eng.* /usr/local/share/tessdata/
  sudo cp tessdata/ita* /usr/local/share/tessdata
fi

# With any luck we get Tesseract in our running infrastructure one way or another
tesseract -v