8_ortho2phono_files.sh

#!/usr/bin/env bash
# Wrapper to take a single cleaned up transcript and phonologize it
# 2015-11-26 - adapted to the winnipeglena corpus in its 4 versions

#########VARIABLES
#Variables that have been passed by the user
datafolder=$1
#########
#datafolder=`readlink -f ${1:-./data}`
root=`readlink -f ${2:-../..}` #Hmmm


phonologize=$root/phonologization/scripts/phonologize
clusterize=$root/algoComp

# must exist and contain ortholines files
ortfolder=$datafolder/ortho

# will be created to store phonologized files
resfolder=$datafolder/phono
mkdir -p $resfolder

# to store intermediate files
temp=`mktemp -d $resfolder/eraseme-XXXX`

# create command for each corpus version
list_cmd=$temp/cmd
list_opt=$temp/opt
rm -rf $list_cmd $list_opt
for subdir in `find $ortfolder/WL_* -type d`
do
    version=`basename ${subdir#$resfolder}`
    name=phonol-$version
    mkdir -p $resfolder/$version

    for f in $subdir/*-ortholines.txt
    do
      	g=${f##*/}
        keyf=${g%.*}
        echo "$phonologize $f $resfolder/$version/${keyf}-tags.txt" >> $list_cmd
    	echo "-V -cwd -j y -o $temp/$name.log -N $name" >> $list_opt
    done
done

#echo about to clusterize
# run them in parallel
$clusterize/clusterize_list.sh $list_cmd $list_opt
rm -rf $temp

echo done clusterizing

# TODO remove this loop and generate gold within the phonolgize step
for f in $resfolder/*/*-tags.txt
do
    gold=`echo $f | sed 's/tags/gold/'`
    echo "creating gold version for $f"
    sed 's/;esyll//g' $f |
        sed 's/ //g' |
        sed 's/;eword/ /g' |
        sed 's/ $//g' | tr -s ' ' > $gold
done

exit