| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 1 | SRC_DIR ?= test/resources/DNB |
| 2 | BUILD_DIR = build |
| 3 | TARGET_DIR ?= target |
| Marc Kupietz | b769699 | 2024-03-18 17:41:15 +0100 | [diff] [blame] | 4 | DEPLOY_HOST ?= compute.ids-mannheim.de |
| 5 | DEPLOY_USER ?= korap |
| 6 | DEPLOY_PATH ?= /export/netapp/korap4dnb |
| Marc Kupietz | 48c6a68 | 2024-03-24 15:27:08 +0100 | [diff] [blame] | 7 | MAX_THREADS ?= $(shell nproc) |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 8 | |
| Marc Kupietz | b769699 | 2024-03-18 17:41:15 +0100 | [diff] [blame] | 9 | .PHONY: all clean test krill index deploy server-log server-status |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 10 | |
| Marc Kupietz | 8ea9228 | 2024-03-24 14:55:55 +0100 | [diff] [blame] | 11 | .PRECIOUS: %.zip %.tree_tagger.zip %.ud.zip %.marmot-malt.zip %.spacy.zip %.i5.xml %.tar |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 12 | |
| Marc Kupietz | b769699 | 2024-03-18 17:41:15 +0100 | [diff] [blame] | 13 | .DELETE_ON_ERROR: |
| 14 | |
| Marc Kupietz | feed26a | 2024-03-17 19:16:23 +0100 | [diff] [blame] | 15 | all: index |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 16 | |
| Marc Kupietz | 3d82f56 | 2024-03-16 10:19:03 +0100 | [diff] [blame] | 17 | krill: $(TARGET_DIR)/dnb.krill.tar |
| Marc Kupietz | 012c708 | 2024-03-17 16:40:47 +0100 | [diff] [blame] | 18 | index: $(TARGET_DIR)/dnb.index.tar.xz |
| Marc Kupietz | 3d82f56 | 2024-03-16 10:19:03 +0100 | [diff] [blame] | 19 | |
| Marc Kupietz | fa445da | 2024-03-16 10:46:22 +0100 | [diff] [blame] | 20 | KORAPXML2CONLLU ?= java -jar lib/korapxml2conllu.jar |
| 21 | |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 22 | $(TARGET_DIR)/dnb.i5.xml: $(patsubst $(SRC_DIR)/%.epub,$(TARGET_DIR)/%.i5.xml,$(wildcard $(SRC_DIR)/*.epub)) |
| 23 | head -n -1 xslt/idsCorpus-template.xml > $@ |
| 24 | cat $^ >> $@ |
| 25 | tail -n 1 xslt/idsCorpus-template.xml >> $@ |
| 26 | |
| Marc Kupietz | dfbc3f4 | 2024-03-16 11:46:41 +0100 | [diff] [blame] | 27 | test: $(TARGET_DIR)/dnb.i5.xml |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 28 | xmllint --noout --valid $< |
| 29 | |
| 30 | $(BUILD_DIR)/%: $(SRC_DIR)/%.epub |
| 31 | mkdir -p $@ |
| 32 | echo "Converting $< to $@" |
| 33 | unzip -q -o $< -d $@ |
| 34 | |
| Marc Kupietz | dfbc3f4 | 2024-03-16 11:46:41 +0100 | [diff] [blame] | 35 | $(TARGET_DIR)/%.i5.xml: $(BUILD_DIR)/% xslt/epub2i5.xsl xslt/idsCorpus-template.xml |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 36 | mkdir -p $(TARGET_DIR) |
| 37 | echo "Converting $< to $@" |
| 38 | java -jar lib/saxon9ee.jar -xsl:xslt/epub2i5.xsl $</*/content.opf > $@ |
| 39 | |
| 40 | %.zip: %.i5.xml |
| 41 | tei2korapxml -l warn -s -tk - < $< > $@ |
| 42 | |
| 43 | %.tree_tagger.zip: %.zip |
| Marc Kupietz | fa445da | 2024-03-16 10:46:22 +0100 | [diff] [blame] | 44 | $(KORAPXML2CONLLU) $< | pv | docker run --rm -i korap/conllu2treetagger -l german | conllu2korapxml > $@ |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 45 | |
| 46 | %.spacy.zip: %.zip |
| Marc Kupietz | fa445da | 2024-03-16 10:46:22 +0100 | [diff] [blame] | 47 | $(KORAPXML2CONLLU) $< | pv | docker run --rm -i korap/conllu2spacy | conllu2korapxml > $@ |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 48 | |
| Marc Kupietz | 8ea9228 | 2024-03-24 14:55:55 +0100 | [diff] [blame] | 49 | models/de.marmot: |
| 50 | mkdir -p models |
| Marc Kupietz | e8c80f3 | 2024-03-24 15:27:28 +0100 | [diff] [blame] | 51 | curl -sL -o $@ https://cistern.cis.lmu.de/marmot/models/CURRENT/spmrl/de.marmot |
| Marc Kupietz | 8ea9228 | 2024-03-24 14:55:55 +0100 | [diff] [blame] | 52 | |
| 53 | models/german.mco: |
| 54 | mkdir -p models |
| Marc Kupietz | e8c80f3 | 2024-03-24 15:27:28 +0100 | [diff] [blame] | 55 | curl -sL -o $@ https://corpora.ids-mannheim.de/tools/$@ |
| Marc Kupietz | 8ea9228 | 2024-03-24 14:55:55 +0100 | [diff] [blame] | 56 | |
| 57 | %.marmot-malt.zip: %.zip models/de.marmot models/german.mco |
| Marc Kupietz | 48c6a68 | 2024-03-24 15:27:08 +0100 | [diff] [blame] | 58 | $(KORAPXML2CONLLU) -T $(MAX_THREADS) -t marmot:models/de.marmot -P malt:models/german.mco $< | tee $(TARGET_DIR)/dnb.marmot-malt.conllu | conllu2korapxml > $@ |
| Marc Kupietz | 8ea9228 | 2024-03-24 14:55:55 +0100 | [diff] [blame] | 59 | |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 60 | %.ud.zip: %.zip |
| Marc Kupietz | fa445da | 2024-03-16 10:46:22 +0100 | [diff] [blame] | 61 | $(KORAPXML2CONLLU) $< | pv | ./scripts/udpipe2 | conllu2korapxml > $@ |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 62 | |
| Marc Kupietz | 8ea9228 | 2024-03-24 14:55:55 +0100 | [diff] [blame] | 63 | %.krill.tar: %.zip %.marmot-malt.zip %.tree_tagger.zip %.spacy.zip |
| Marc Kupietz | 3d82f56 | 2024-03-16 10:19:03 +0100 | [diff] [blame] | 64 | mkdir -p $(basename $@) |
| Marc Kupietz | 8ea9228 | 2024-03-24 14:55:55 +0100 | [diff] [blame] | 65 | korapxml2krill archive --quiet -w -z -cfg krill-korap4dnb.cfg --non-word-tokens --meta I5 -i $< -i $(word 2,$^) -i $(word 3,$^) -i $(word 4,$^) -o $(basename $@) |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 66 | |
| Marc Kupietz | fc32862 | 2024-03-16 14:31:34 +0100 | [diff] [blame] | 67 | %.json: %.krill.tar |
| 68 | rm -rf $@ |
| 69 | mkdir -p $@ |
| 70 | for f in $<; do tar -C $@ -xf $$f; done |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 71 | |
| Marc Kupietz | 22ad4b7 | 2024-03-17 16:24:21 +0100 | [diff] [blame] | 72 | %.index: %.json |
| 73 | rm -rf $@ |
| Marc Kupietz | 012c708 | 2024-03-17 16:40:47 +0100 | [diff] [blame] | 74 | java -jar lib/Krill-Indexer.jar -c lib/krill.conf -i $< -o $@ |
| Marc Kupietz | 22ad4b7 | 2024-03-17 16:24:21 +0100 | [diff] [blame] | 75 | |
| 76 | %.index.tar.xz: %.index |
| 77 | tar -I 'xz -T0' -C $(dir $<) -cf $@ $(notdir $<) |
| 78 | |
| Marc Kupietz | b769699 | 2024-03-18 17:41:15 +0100 | [diff] [blame] | 79 | deploy: $(TARGET_DIR)/dnb.index.tar.xz korap4dnb-compose.yml |
| 80 | rsync -v $^ $(DEPLOY_USER)@$(DEPLOY_HOST):$(DEPLOY_PATH)/ |
| 81 | ssh $(DEPLOY_USER)@$(DEPLOY_HOST) "mkdir -p $(DEPLOY_PATH) && cd $(DEPLOY_PATH) && docker compose -p korap4dnb --profile=lite -f $(notdir $(word 2,$^)) up -d --dry-run && docker compose -p korap4dnb stop && (mv -f dnb.index dnb.index.bak || true) && tar Jxvf $(notdir $<) && docker compose -p korap4dnb --profile=lite -f $(notdir $(word 2,$^)) up -d" |
| 82 | |
| 83 | show-server-log: |
| 84 | ssh $(DEPLOY_USER)@$(DEPLOY_HOST) "cd $(DEPLOY_PATH) && docker compose -p korap4dnb --profile=lite -f korap4dnb-compose.yml logs -f" |
| 85 | |
| 86 | show-server-status: |
| 87 | ssh $(DEPLOY_USER)@$(DEPLOY_HOST) "cd $(DEPLOY_PATH) && docker compose -p korap4dnb --profile=lite -f korap4dnb-compose.yml ps" |
| 88 | |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 89 | clean: |
| 90 | rm -rf $(BUILD_DIR) $(TARGET_DIR) |
| 91 | |