#!/bin/bash

unset DISPLAY;
unset LUI;

if [ -z "${LOGONROOT}" ]; then
  echo "parse: unable to determine \$LOGONROOT directory; exit.";
  exit 1;
fi

#
# include a shared set of shell functions and global parameters, including the
# architecture identifier .LOGONOS.
#
. ${LOGONROOT}/etc/library.bash;

tsdb="$LOGONROOT/bin/tsdb"
source="--source";

while [ $# -gt 0 -a "${1#-}" != "$1" ]; do
  case ${1} in
    --binary)
      source="";
      shift 1;
    ;;
    --source)
      source="--source";
      shift 1;
    ;;
    -c|--compress)
      action="compress";
      shift 1;
    ;;
    -t|--thin)
      action="thin";
      shift 1;
    ;;
    --drop)
      action="drop";
      gold=${2};
      shift 2;
    ;;
    --cat)
      cat="--cat";
      shift 1;
    ;;
    *)
      echo "usage: itsdb { -c | -t } directory; exit.";
      exit 1;
  esac
done

if [ -z "${action}" ]; then
  echo "usage: itsdb { -c | -t } directory; exit.";
  exit 1;
fi


if [ ! -d ${1} ]; then
  echo "itsdb: invalid or missing directory \`${1}'; exit.";
  exit 1;
fi

directory="${1}";

if [ -z "${LOGONLOG}" ]; then
  LOGONLOG=${LOGONROOT}/log;
fi
LOG=${LOGONLOG}/itsdb.${date}.log;

case "${action}" in
  compress)
    tmp="/tmp/.itsdb.tmp.${USER}.$$"
    find ${directory} -name relations | sed 's@/relations@@g' > ${tmp}.dbs
    while read i; do
      ${tsdb} -home "${i}" -query "info relations" \
      | egrep '^[a-zA-Z0-9_-]+:$' | sed 's@:$@@g' > ${tmp}.files
      for j in $(cat ${tmp}.files); do
        file="${i}/${j}";
        if [ -s "${file}" ]; then
          echo -n "compressing \`${file}' ... ";
          gzip -f -9 "${file}";
          echo "done";
        fi
      done
    done < ${tmp}.dbs
    /bin/rm -f ${tmp}.dbs ${tmp}.files
  ;;
  thin)
    #
    # _fix_me_
    # the code below hard-wires the grammar to be used, viz. the ERG.  use the
    # `--erg' et al. options much like the `parse' script.     (30-jul-09; oe)
    #
    unset DISPLAY;
    unset LUI;
    #
    # erase the target directory, if need be, and invoke the normalizer
    #
    target="${directory}.1";
    if [ -a ${target} ]; then /bin/rm -rf ${target}; fi
    {
      loadup;

      echo "(lkb:read-script-file-aux \"${LOGONROOT}/lingo/erg/lkb/script\")";
      echo "(setf tsdb::*redwoods-trees-hook* nil)";
      echo "(setf tsdb::*redwoods-semantix-hook* \"mrs::get-mrs-string\")";
      echo "(setf tsdb::*redwoods-thinning-normalize-p* t)";
      echo "(tsdb::browse-trees \"${directory}\" :strip \"${target}\")";
      echo "(excl:exit)";
    } | ${LOGONROOT}/bin/logon ${source} ${cat} \
          -I base -locale no_NO.UTF-8 -qq 2>&1 | tee ${LOG}
  ;;
  drop)
set -x;
    tmp="/tmp/.itsdb.tmp.${USER}.$$"
    ${tsdb} -home "${directory}" -query "info relations" \
    | egrep '^[a-zA-Z0-9_-]+:$' | sed 's@:$@@g' > ${tmp}
    for j in $(cat ${tmp}); do
      file="${directory}/${j}";
      if [ -f ${file} -a -z "$(cat ${file})" ]; then
        echo "cat ${file} > ${gold}/${j}";
      elif [ -s ${file}.gz ]; then
        echo "cat ${file}.gz > ${gold}/${j}.gz";
      else
        echo "itsdb: missing or invalid relation file for \`${j}'; exit."
        exit 1;
      fi
    done
    /bin/rm -f ${tmp}
  ;;
esac