train.sh

#! /bin/bash

function has_space {
  [[ "$1" != "${1%[[:space:]]*}" ]] && return 0 || return 1
}

function print_usage_and_exit {
  >&2 echo "USAGE : (tsv | txt) expPath [arguments]"
  exit 1
}

MODE=$1
EXPPATH=$2

if [ -z "$MODE" ];
then
  >&2 echo "ERROR : missing argument 1 (mode)"
  print_usage_and_exit
fi

if [ -z "$EXPPATH" ];
then
  >&2 echo "ERROR : missing argument 2 (expPath)"
  print_usage_and_exit
fi

shift
shift

if [ ! -d "$EXPPATH" ]; then
  >&2 echo "ERROR : directory $EXPPATH doesn't exist"
  print_usage_and_exit
fi

source $EXPPATH"/config"

TRAIN=$(find $CORPUS -type f -name '*train*.conllu')
TRAINRAW=$(find $CORPUS -type f -name '*train*.txt')
DEV=$(find $CORPUS -type f -name '*dev*.conllu')
DEVRAW=$(find $CORPUS -type f -name '*dev*.txt')
TEST=$(find $CORPUS -type f -name '*test*.conllu')
TESTRAW=$(find $CORPUS -type f -name '*test*.txt')

if has_space "$TRAIN" || has_space "$DEV" || has_space "$TEST";
then
  >&2 echo "ERROR : more than 1 match with keyword" $KEYWORD
  >&2 echo "TRAIN : " $TRAIN
  >&2 echo "DEV : " $DEV
  >&2 echo "TEST : " $TEST
  print_usage_and_exit
fi

if test ! -f $TRAIN;
then
  >&2 echo "ERROR : no train file found in" $CORPUS
  >&2 echo "$TRAIN"
  print_usage_and_exit
fi

CURDIR=$(pwd)
cd $EXPPATH"/"data && make -s clean && make -s && cd $CURDIR

MCD=$EXPPATH"/data/*\.mcd"

if [ "$MODE" = "tsv" ]; then
macaon train --model $EXPPATH --mcd $MCD --trainTSV $TRAIN --devTSV $DEV $@ || exit 1
exit 0
fi

if [ "$MODE" = "txt" ]; then
macaon train --model $EXPPATH --mcd $MCD --trainTSV $TRAIN --trainTXT $TRAINRAW --devTSV $DEV --devTXT $DEVRAW $@ || exit 1
exit 0
fi

print_usage_and_exit