arpa2fst

#!/bin/bash
set -e

# constants
model=16000_i
prefix=tam
g_prefix="$prefix.tsaurus"
osyms="$g_prefix.g.osyms"
TRUNK="$(readlink -f "`dirname $0`/../..")"
DECODER="$TRUNK/decoder"

# help message
usage (){
	echo "Usage: `basename $0` [options] <arpa_file> [<fst_file>]"
	echo "Options:"
	echo "   -b <build_directory>"
	echo "       Specify the location of the build directory to use."
	echo "       [Default: $TRUNK/build]"
	echo "   -D <data_directory>"
	echo "       Specify the directory in which to store intermediate files."
	echo "   -d <pronunciation_dictionary>"
	echo "       Specify the pronunciation dictionary to use when compiling the FST."
	echo "   -g"
	echo "       Regenerate the acoustic models before building the FSTs."
	echo "   -z <directory>"
	echo "       Create all intermediate files under this directory."
	echo "   -h , --help"
	echo "       Display this message and exit."
	echo "Arguments:"
	echo "   <arpa_file>"
	echo "       The ARPA file to convert."
	echo "   <fst_file>"
	echo "       Name of output file."
}

# defaults
GEN_AM=0
BUILD_DIR="$TRUNK/build"
DICT="$DECODER/s3data/wsj_all.dic"
DATA_DIR="$DECODER/data"
working_dir="."

[[ $1 == "--help" ]] && { usage; exit 0; }

# parse options
while getopts ":d:D:ghb:Bz:" opt; do
	case $opt in
		D) DATA_DIR="$OPTARG" ;;
		d) DICT="$OPTARG" ;;
		g) GEN_AM=1 ;;
		h) usage; exit 0 ;;
		b) BUILD_DIR="`readlink -f "$OPTARG"`" ;;
		z) working_dir="$OPTARG" ;;
		\?)
			echo "Invalid Option: -$OPTARG" >&2
			usage >&2
			exit 1 ;;
		:)
			echo "Option -$OPTARG requires an additional argument" >&2
			usage >&2
			exit 1 ;;
	esac
done

# make sure that the specified data directory exists
if [[ ! -d "$DATA_DIR" ]]; then
	echo "\"$DATA_DIR\" does not exist or is not a directory." >&2
	exit 1
fi

# shift away the options
shift $(($OPTIND-1))

# make sure the correct number of positional arguments were given
if [[ $# -lt 1 ]]; then
	echo "Error: Missing positional argument" >&2
	usage >&2
	exit 65
fi

arpa_fname="$1"
shift

# make sure the arpa file exists
if [[ ! -e "$arpa_fname" || -d "$arpa_fname" ]]; then
	echo "Error: $arpa_file does not exist or is a directory." >&2
	exit 1
fi

# get fst file name
if [[ $# -gt 0 ]]; then
	fst_fname="$1"
else # default replaces arpa extension with fst extension
	fst_fname="`basename ${arpa_fname%.arpa}.fst`"
fi

# get absolute paths
arpa_fname="`readlink -f "$arpa_fname"`"
fst_fname="`readlink -f "$fst_fname"`"

# make sure the audio model data folder exists
mkdir -p "$DATA_DIR/$model"

# regenerate the AM if requested
if [[ $GEN_AM -eq 1 ]]; then "$DECODER/scripts/gen_am.py" -i "$DECODER/s3data/$model" -o "$DATA_DIR/$model"; fi

# create symlink to dictionary in data directory, which is where ComposeAll expects it
ln -sf "`readlink -f "$DICT"`" "$DATA_DIR/dict"

# create a directory to hold the intermediate files
tmp_data_dir="${working_dir}/intermediate_`basename ${arpa_fname%.arpa}`"
mkdir -p "$tmp_data_dir"
pushd "$tmp_data_dir" >& /dev/null

# compile grammar level FST
echo "compiling $g_prefix.g.fst"
python << EOL
#!/usr/bin/env python
import sys
sys.path.insert(0, '$DECODER/scripts/transducersaurus-0.0.1/python/')
from arpa2fst import ArpaLM
arpa = ArpaLM(
	'$arpa_fname',
	'$g_prefix.g.fst.txt',
	sil='<SIL>',
	prefix='$g_prefix',
	eps='<eps>',
	boff='#b')
arpa.arpa2fst( )
arpa.print_all_syms( )
EOL
# end of python

fstcompile \
	--arc_type=log \
	--acceptor=true \
	"--ssymbols=$g_prefix.g.ssyms" \
	"--isymbols=$g_prefix.g.isyms" \
	"--osymbols=$osyms" \
	--keep_isymbols --keep_osymbols \
	"$g_prefix.g.fst.txt" | fstarcsort --sort_type=ilabel - > "$g_prefix.g.fst"

"${BUILD_DIR:-$TRUNK/build}/decoder/utils/ComposeAll" \
	"$DATA_DIR" \
	"$model" \
	"$g_prefix.g.fst" \
	ptt \
	"$fst_fname" \
	"$prefix"

popd >& /dev/null