| Marc Kupietz | 78af9d1 | 2023-07-30 16:49:10 +0200 | [diff] [blame^] | 1 | .DELETE_ON_ERROR: |
| 2 | |
| 3 | all: json |
| 4 | |
| 5 | ./output/corpus.p5.xml: gle2tei.py |
| 6 | mkdir -p output |
| 7 | python irish2tei.py `find ICC-GA-XML -name "*.xml"` |
| 8 | |
| 9 | icc-gle.p5.xml: ./output/GA_corpus.p5.xml |
| 10 | cat $< | xmllint - > $@ |
| 11 | |
| 12 | icc-gle.zip: icc-gle.p5.xml |
| 13 | pv $< | tei2korapxml -s -tk - > $@ |
| 14 | |
| 15 | icc-gle.ud.zip: icc-gle.zip |
| 16 | korapxml2conllu $< | pv | /usr/local/kl/bin/udpipe2 -r -m irish-idt-ud-2.10-220711 | conllu2korapxml > $@ |
| 17 | |
| 18 | |
| 19 | icc-gle.krill.tar: icc-gle.zip icc-gle.ud.zip |
| 20 | /vol/work/kupietz/KorAP/KorAP-XML-Krill/script/korapxml2krill archive -w -cfg /vol/corpora/ICC/icc-gle.cfg -j 0 --meta ICC -i icc-gle.zip -i icc-gle.ud.zip -o icc-gle.krill |
| 21 | |
| 22 | json: icc-gle.krill.tar |
| 23 | rm -rf json |
| 24 | mkdir -p json |
| 25 | tar -C json -xf icc-gle.krill.tar |
| 26 | rsync -avz --delete json korap@korap-worker-07:/opt/korap/icc/gle/ |
| 27 | ssh korap@korap-worker-07 "cd /opt/korap/icc/gle && rm -rf index && mkdir -p index && docker run -u root --rm -v /opt/korap/icc/gle:/data:z korap/kustvakt Krill-Indexer.jar -c /kustvakt/kustvakt-lite.conf -i /data/json -o /data/index/ && INDEX=./index docker-compose --profile=full -p icc-gle restart" |