diff --git a/tools/addSample.sh b/tools/addSample.sh new file mode 100755 index 0000000..a62051e --- /dev/null +++ b/tools/addSample.sh @@ -0,0 +1,9 @@ +#!/bin/sh + +CORPUS_NAME=opensubtitles_sample +CORPUS_PATH=./$CORPUS_NAME +SRC_LANG_ID=1 +TRG_LANG_ID=2 + +./addFastAlignedTM.py $CORPUS_NAME $CORPUS_PATH/src_clean.txt $CORPUS_PATH/src_clean.lem $SRC_LANG_ID $CORPUS_PATH/trg_clean.txt $TRG_LANG_ID $CORPUS_PATH/alignments.txt $CORPUS_PATH/ids_clean.txt +./addSources.py $CORPUS_PATH/ids_sources_censored.txt