#!/bin/bash

set -x;

if [ -d /projects/nlpl ]; then
  host="abel";
  root="/projects/nlpl";
  suffix="tcl";
elif [ -d /proj/nlpl ]; then
  host="taito";
  root="/proj/nlpl";
  suffix="lua";
elif [ -d /cluster/shared/nlpl ]; then
  host="saga";
  root="/cluster/shared/nlpl";
  suffix="lua";
elif [ -d /projappl/nlpl ]; then
  host="puhti";
  root="/projappl/nlpl";
  suffix="lua";
fi
module use -a ${root}/software/modules/etc;

if [ -z "${host}" ]; then
  echo "unable to locate NLPL project directory; exit."
  exit 1;
fi

python="python3";
version=$(command date "+%Y%m");
tmp="/tmp/.initialize.${USER}.$$";

while [ $# -gt 0 -a "${1#-}" != "$1" ]; do
  case ${1} in
    --p*)
      python=${2};
      shift 2;
    ;;
    --v*)
      version=${2};
      shift 2;
    ;;
    --nvidia)
      nvidia=yes;
      shift 1;
    ;;
    *)
      echo "initialize: invalid option \`${1}'; exit.";
      exit 1;
    ;;
  esac
done

name=${1}
if [ -z "${name}" ]; then
  echo "initialize: missing obligatory ‘name’ argument; exit.";
  exit 1;
fi

log="${root}/operation/log/${host}/python.${name}.$(date "+%y-%m-%d-%H:%M")";
{
  if [ ${version} = "pypi" ]; then
    case ${name} in
      pytorch)
        key="torch";
      ;;
      *)
        key=${name};
      ;;
    esac;
    version=$(command python3 -m pip install ${key}== 2>&1 \
              | grep "Could not find" \
              | awk '{print $NF}' | sed 's/)$//');
  fi

  if [ -z "$(type -t ${python})" ]; then
    echo "initialize: invalid interpreter ‘${python}’; exit.";
    exit 1;
  fi

  implementation=$(${python} --version 2>&1 | sed 's/^Python //');
  dialect=$(${python} --version 2>&1 | sed 's/^Python \([0-9]\.[0-9]\).*$/\1/');
  family=$(${python} --version 2>&1 | sed 's/^Python \([0-9]\)\..*$/\1/');
  interpreter="python${dialect}";
  requirements="${root}/operation/python/${name}.txt";
  pip="./${interpreter} -m pip";
  if [ ${host} = "abel" -o ${host} = "taito" ]; then
    target="${root}/software/${name}/${version}";
    definition="${root}/software/modulefiles/nlpl-${name}/${version}/${dialect}";
  else
    target="${root}/software/modules/${name}/${version}";
    definition="${root}/software/modules/etc/nlpl-${name}/${version}/${dialect}.lua";
  fi

  if [ -f ${definition} ]; then
    echo "initialize: existing ‘nlpl-${name}/${version}/${dialect}’; exit.";
    exit 1;
  fi
  
  command mkdir -pv ${target}/bin;
  command mkdir -pv $(dirname ${definition});

  cd ${target}/bin;

  command ls > ${tmp};
  
  ${python} -m virtualenv --always-copy ${target};

  #
  # if need be, enable our own more modern GLIBC environment
  #
  if [ ${host} = "abel" -o ${host} = "taito" ]; then
    command mv ${interpreter} .${interpreter};
    command cp -va ${root}/software/glibc/2.23/wrapper ${interpreter};
    command rm python${family};
    command ln -s ${interpreter} python${family};
  fi

  #
  # the 3.7.3 (anaconda-based) Python on Puhti does not seem to provide
  # the most specific reference to the interpreter
  #
  [ ! -f ./${interpreter} ] && command cp python${family} ${interpreter};

  if false; then
  source activate
  command ${pip} install --upgrade pip
  command ${pip} install -U $(command ${pip} list | command tail -n +3 | command awk '{print $1}');
  if [ "$(type -t deactivate)" = "function" ]; then
    deactivate;
  else
    source deactivate;
  fi
  fi

  #
  # our in-house definition files can contain shell directives
  #
  command egrep '^#\$' ${requirements} > ${tmp};
  while read line; do
    eval ${line##\#\$ };
  done < ${tmp};
  
  if [ -n "${nvidia}" -a -f /usr/lib64/libcuda.so ]; then
    command cp -av /usr/lib64/libcuda.so* /usr/lib64/libnvidia* ../lib;
  fi

  command sed \
    -e "s@__root__@${root}@g" \
    -e "s/__name__/${name}/g" \
    -e "s/__version__/${version}/g" \
    -e "s/__dialect__/${dialect}/g" \
    -e "s/__family__/${family}/g" \
    ${root}/operation/python/prologue.${suffix} \
    | grep -v "VIRTUAL_ENV" > ${definition};

  {
    echo;
    for i in $(set +x; module --terse list 2>&1 | grep -v "Currently Loaded"); do
      if [ ${suffix} = "tcl" ]; then
        echo "module load ${i}";
      else
        echo "if not (isloaded(\"${i}\")) then";
	echo "  load(\"${i}\")";
	echo "end";
      fi
    done
    echo;
    if [ ${suffix} = "tcl" ]; then
      echo "prepend-path PYTHONPATH \$root/\$name/\$version/lib/python${dialect}/site-packages";
      echo "prepend-path LD_LIBRARY_PATH \$root/\$name/\$version/lib";
      echo "prepend-path PATH \$root/\$name/\$version/bin/${dialect}";
    else
      echo "prepend_path(\"PYTHONPATH\", pathJoin(root, name, version, \"lib/python${dialect}/site-packages\"))";
      echo "prepend_path(\"LD_LIBRARY_PATH\", pathJoin(root, name, version, \"lib\"))";
      echo "prepend_path(\"PATH\", pathJoin(root, name, version, \"bin/${dialect}\"))";
    fi
      echo;
    case ${name} in
      nltk)
	if [ ${suffix} = "tcl" ]; then
          echo "setenv NLTK_DATA \$root/nltk/\$version/data";
	else
          echo "setenv(\"NLTK_DATA\", pathJoin(root, \"nltk\", version, \"data\"))";
	fi
        echo;
      ;;
    esac
  } >> ${definition};

  #
  # our in-house definition files can contain configuration snippets
  #
  command egrep '^#>' ${requirements} \
  | command sed 's/^#> *//g' >> ${definition};

  command cat ${root}/operation/python/epilogue.${suffix} >> ${definition};

  module --ignore-cache load nlpl-${name}/${version}/${dialect};
  ${pip} install --upgrade pip;

  if [ -f ${requirements} ]; then
    ${pip} install --upgrade -r ${requirements};
    case ${name} in
      nltk)
	if [ ${host} = "abel" -o ${host} = "taito" ]; then
	  data="${root}/software/nltk/${version}/data";
	else
	  data="${root}/software/modules/nltk/${version}/data";
	fi
        if [ ! -d ${data} ]; then
          ${target}/bin/${interpreter} -m nltk.downloader -d ${data} all;
        fi
      ;;
      spacy)
        for i in en de es pt fr it nl xx; do
          ${target}/bin/${interpreter} -m spacy download ${i};
        done
      ;;
    esac
  fi

  cd ${target}/bin;
  command mkdir ./${dialect};
  [ -f .${interpreter} ] && command mv .${interpreter} ./${dialect};

  for i in $(command ls); do
    if [ -f ${i} -a -z "$(command egrep "^${i}$" ${tmp})" ]; then
      if [ -h ${i} ]; then
        command mv ${i} ./${dialect};
      else
	if [ ${host} = "abel" ]; then
	  sed="s@${name}/${version}/bin/.python${dialect}\$";
	  sed="${sed}@${name}/${version}/bin/${dialect}/python${dialect}@";
	else
	  sed="s@/${name}/${version}/bin/python";
	  sed="${sed}@/${name}/${version}/bin/${dialect}/python@";
	fi
        command sed -e ${sed} ${i} > ./${dialect}/${i};
        command chmod +x ./${dialect}/${i};
        command rm -f ${i};
      fi
    fi
  done
  cd ${dialect};
  if [ -f python -a -f .python${dialect} ]; then
    if command cmp python .python${dialect}; then
      command rm python;
      command ln -s python${dialect} python;
    fi
  fi
  #command rm -f ${tmp};
} 2>&1 | command tee ${log}