| .DELETE_ON_ERROR: |
| |
| all: json |
| |
| icc-ger.disposable.xml: /export/netapp/fi2/luengen/ICC/ICC-German.p5.xml |
| pv $< | ./map-ger-icc-genres.pl > $@ |
| |
| icc-ger.zip: icc-ger.disposable.xml |
| pv $< | tei2korapxml --xmlid-to-textsigle 'ICC.German\.([^.]+\.?[^.]+)\.(.+)@GER/$$1/$$2' -s -tk - > $@ |
| |
| icc-ger.ud.zip: icc-ger.zip |
| korapxml2conllu $< | pv | /usr/local/kl/bin/udpipe2 | conllu2korapxml > $@ |
| |
| icc-ger.tree_tagger.zip: icc-ger.zip |
| /usr/local/kl/korap/Ingestion/treetagger/korap2tt.pl -D . icc-ger.zip |
| |
| icc-ger.krill.tar: icc-ger.zip icc-ger.tree_tagger.zip icc-ger.ud.zip |
| korapxml2krill archive -w -cfg /vol/corpora/ICC/icc-ger.cfg -j 0 --meta ICC -i icc-ger.zip -i icc-ger.tree_tagger.zip -i icc-ger.ud.zip -o icc-ger.krill |
| |
| json: icc-ger.krill.tar |
| rm -rf json |
| mkdir -p json |
| tar -C json -xf icc-ger.krill.tar |
| rsync -avz --delete json korap@korap-worker-07:/opt/korap/icc/ger/ |
| ssh korap@korap-worker-07 "cd /opt/korap/icc/ger && rm -rf index && mkdir -p index && docker run -u root --rm -v /opt/korap/icc/ger:/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-ger restart" |