detransliterator

Table of Contents

• Installation
• Usage
• License

Installation

pip install detransliterator

Usage

as a software library

from detransliterator import Detransliterator

detransliterator = Detransliterator('latin2nqo_001.35')
latin = "musa dunbuya"
nqo = detransliterator.detransliterate(latin, beam_size=5)
assert nqo == "ߡߎߛߊ߫ ߘߎ߲ߓߎߦߊ"

as a console tool

# detransliteration tool
python -m detransliterator.tool --help

# csv/tsv extraction tool
python -m detransliterator.csv_tool --help

example: detransliterate a stream

echo "musa dunbuya" | python -m detransliterator.tool --model-name latin2nqo_001.35

example: extract a column from a csv file

cat test_csv_no_header.csv \
| python -m detransliterator.csv_tool \
    extract-column --column-ix 1 \
> tmp_col1.txt

example: extract a column from a tsv file

cat test_tsv_with_header.tsv \
| python -m detransliterator.csv_tool \
    extract-column --column-ix 1 --skip-lines 1 --csv-formatting-params delimiter tab \
> tmp_col2.txt

example: detransliterate a column from a csv file

cat test_tsv_no_header.tsv \
| python -m detransliterator.csv_tool extract-column --column-ix 1 \
    --csv-formatting-params delimiter tab \
| python -m detransliterator.tool --model-name latin2nqo_001.35 \
> tmp_detransliterated_1.nqo

example: detransliterate a column from a tsv file

cat test_csv_with_header.csv \
| python -m detransliterator.csv_tool extract-column --column-ix 1 --skip-lines 1 \
| python -m detransliterator.tool --model-name latin2nqo_001.35 \
> tmp_detransliterated_2.nqo

example: use a particular GPU

CUDA_VISIBLE_DEVICES="1" echo "musa dunbuya" | python -m detransliterator.tool

Detransliteration Models

License

detransliterator is distributed under the terms of the MIT license.