| .DELETE_ON_ERROR: |
| |
| all: json |
| |
| ./output/corpus.p5.xml: eng2tei.py |
| mkdir -p output |
| python eng2tei.py `find /export/coin/ICC/ICC-EN_written_XML_April2023 -name "*.xml"` |
| |
| icc-eng.p5.xml: ./output/corpus.p5.xml |
| cat $< | xmllint - > $@ |
| |
| icc-eng.zip: icc-eng.p5.xml |
| pv $< | tei2korapxml -s -tk - > $@ |
| |
| icc-eng.ud.zip: icc-eng.zip |
| korapxml2conllu $< | pv | /usr/local/kl/bin/udpipe2 -r -m english-partut-ud-2.10-220711 | conllu2korapxml > $@ |
| |
| |
| icc-eng.krill.tar: icc-eng.zip icc-eng.ud.zip |
| korapxml2krill archive -w -cfg /vol/corpora/ICC/icc-eng.cfg -j 0 --meta ICC -i icc-eng.zip -i icc-eng.ud.zip -o icc-eng.krill |
| |
| json: icc-eng.krill.tar |
| rm -rf json |
| mkdir -p json |
| tar -C json -xf icc-eng.krill.tar |
| rsync -avz --delete json korap@korap-worker-07:/opt/korap/icc/eng/KorAP-Docker/ |
| ssh korap@korap-worker-07 cd /opt/korap/icc/eng/KorAP-Docker && rm -rf index && mkdir -p index && docker run -u root --rm -v /opt/korap/icc/eng/KorAP-Docker:/data:z korap/kustvakt Krill-Indexer.jar -c /kustvakt/kustvakt-lite.conf -i /data/json -o /data/index/ |