| Marc Kupietz | 84da39d | 2024-07-25 20:23:50 +0200 | [diff] [blame] | 1 | ifneq (,$(filter test,$(MAKECMDGOALS))) |
| 2 | SRC_DIR = test/resources/DNB |
| 3 | YEARS=13 18 |
| 4 | else |
| 5 | SRC_DIR ?= /mnt/data/KorAP@DNB |
| Marc Kupietz | 13e2858 | 2024-04-19 11:26:01 +0200 | [diff] [blame] | 6 | YEARS ?= $(shell seq -w 2012 2024 | sed 's/^.*\([0-9][0-9]\)/\1/') |
| Marc Kupietz | 84da39d | 2024-07-25 20:23:50 +0200 | [diff] [blame] | 7 | endif |
| Marc Kupietz | deb9546 | 2024-04-15 20:26:14 +0200 | [diff] [blame] | 8 | |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 9 | BUILD_DIR = build |
| 10 | TARGET_DIR ?= target |
| Marc Kupietz | b769699 | 2024-03-18 17:41:15 +0100 | [diff] [blame] | 11 | DEPLOY_HOST ?= compute.ids-mannheim.de |
| 12 | DEPLOY_USER ?= korap |
| 13 | DEPLOY_PATH ?= /export/netapp/korap4dnb |
| Marc Kupietz | 48c6a68 | 2024-03-24 15:27:08 +0100 | [diff] [blame] | 14 | MAX_THREADS ?= $(shell nproc) |
| Marc Kupietz | 975996e | 2024-04-13 13:46:37 +0200 | [diff] [blame] | 15 | MAKE ?= make -j $(shell nproc) |
| Marc Kupietz | 3146b63 | 2024-04-23 09:40:25 +0200 | [diff] [blame] | 16 | KORAPXML2CONLLU_HEAP ?= $(shell echo "$$(($(MAX_THREADS) * 2500))") |
| Marc Kupietz | d059d2d | 2024-04-15 18:58:39 +0200 | [diff] [blame] | 17 | KORAPXML2CONLLU ?= java -Xmx$(KORAPXML2CONLLU_HEAP)m -jar lib/korapxml2conllu.jar |
| Marc Kupietz | 68b4f8a | 2024-05-26 18:27:40 +0200 | [diff] [blame] | 18 | SAXON ?= java -Djava.util.logging.config.file=/logging.properties -cp lib/saxon-ee-12.4.jar:lib/xmlresolver-5.2.2.jar:lib/textclassifier.jar:lib/xmlresolver-5.2.2-data.jar net.sf.saxon.Transform -expand:off |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 19 | |
| Marc Kupietz | 84da39d | 2024-07-25 20:23:50 +0200 | [diff] [blame] | 20 | |
| 21 | |
| Marc Kupietz | 5652c81 | 2024-04-19 11:42:00 +0200 | [diff] [blame] | 22 | .DELETE_ON_ERROR: |
| 23 | |
| Marc Kupietz | 84da39d | 2024-07-25 20:23:50 +0200 | [diff] [blame] | 24 | .PHONY: all clean test i5 i5valid krill index deploy show-server-log show-server-status |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 25 | |
| Marc Kupietz | 5652c81 | 2024-04-19 11:42:00 +0200 | [diff] [blame] | 26 | .PRECIOUS: $(TARGET_DIR)/%.i5.xml $(TARGET_DIR)/dnb%.pre.i5.xml %.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] | 27 | |
| Marc Kupietz | feed26a | 2024-03-17 19:16:23 +0100 | [diff] [blame] | 28 | all: index |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 29 | |
| Marc Kupietz | deb9546 | 2024-04-15 20:26:14 +0200 | [diff] [blame] | 30 | krill: $(foreach year,$(YEARS),$(TARGET_DIR)/dnb$(year).krill.tar) |
| Marc Kupietz | 3d82f56 | 2024-03-16 10:19:03 +0100 | [diff] [blame] | 31 | |
| Marc Kupietz | ab1f3ac | 2024-04-16 14:56:43 +0200 | [diff] [blame] | 32 | index: $(TARGET_DIR)/dnb.index |
| Marc Kupietz | deb9546 | 2024-04-15 20:26:14 +0200 | [diff] [blame] | 33 | |
| Marc Kupietz | 84da39d | 2024-07-25 20:23:50 +0200 | [diff] [blame] | 34 | EPUBS := $(wildcard $(SRC_DIR)/**/*.epub) |
| 35 | |
| Marc Kupietz | 8a1e465 | 2024-04-21 15:55:03 +0200 | [diff] [blame] | 36 | $(TARGET_DIR)/dnb%.i5.xml: $(TARGET_DIR)/dnb%.pre.i5.xml xslt/pass2.xsl xslt/pass3.xsl models/dereko_domains_s.classifier |
| Marc Kupietz | 28f48e1 | 2024-04-14 16:17:05 +0200 | [diff] [blame] | 37 | $(SAXON) -xsl:xslt/pass2.xsl $< | $(SAXON) -xsl:xslt/pass3.xsl - > $@ |
| Marc Kupietz | fa445da | 2024-03-16 10:46:22 +0100 | [diff] [blame] | 38 | |
| Marc Kupietz | 84da39d | 2024-07-25 20:23:50 +0200 | [diff] [blame] | 39 | $(TARGET_DIR)/dnb%.pre.i5.xml: $(patsubst %.epub,$(TARGET_DIR)/%.i5.xml,$(notdir $(EPUBS))) |
| 40 | echo $(EPUBS) |
| 41 | @find $(SRC_DIR) -type f -name '*.epub' | while read src; do \ |
| 42 | echo $(TARGET_DIR)/$$(basename $${src%.epub}).i5.xml; \ |
| 43 | done | sort -u > $(TARGET_DIR)/filelist$*.txt |
| Marc Kupietz | deb9546 | 2024-04-15 20:26:14 +0200 | [diff] [blame] | 44 | sed -i -e 's/ /\n/g; /^$$/d' $(TARGET_DIR)/filelist$*.txt |
| 45 | head -n -1 xslt/idsCorpus-template.xml | sed -e 's/{YY}/$*/' > $@ |
| Marc Kupietz | b926f25 | 2024-04-14 08:57:11 +0200 | [diff] [blame] | 46 | @while IFS= read -r f; do \ |
| Marc Kupietz | deb9546 | 2024-04-15 20:26:14 +0200 | [diff] [blame] | 47 | if head -500 "$$f" | grep -Eq '<pubDate type="year">..$*'; then \ |
| Marc Kupietz | b926f25 | 2024-04-14 08:57:11 +0200 | [diff] [blame] | 48 | cat "$$f" >> $@; \ |
| 49 | fi; \ |
| Marc Kupietz | deb9546 | 2024-04-15 20:26:14 +0200 | [diff] [blame] | 50 | done < $(TARGET_DIR)/filelist$*.txt |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 51 | tail -n 1 xslt/idsCorpus-template.xml >> $@ |
| 52 | |
| Marc Kupietz | 161d011 | 2024-07-25 11:09:04 +0200 | [diff] [blame] | 53 | |
| Marc Kupietz | 8467724 | 2024-04-27 11:46:07 +0200 | [diff] [blame] | 54 | test: models/dereko_domains_s.classifier i5valid test/test-xml.sh |
| Marc Kupietz | 73a26bf | 2024-04-26 17:36:44 +0200 | [diff] [blame] | 55 | bash test/test-xml.sh |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 56 | |
| Marc Kupietz | deb9546 | 2024-04-15 20:26:14 +0200 | [diff] [blame] | 57 | i5: $(foreach year,$(YEARS),$(TARGET_DIR)/dnb$(year).i5.xml) |
| Marc Kupietz | 157e079 | 2024-04-13 12:46:42 +0200 | [diff] [blame] | 58 | |
| Marc Kupietz | deb9546 | 2024-04-15 20:26:14 +0200 | [diff] [blame] | 59 | i5valid: i5 |
| 60 | xmllint --noout --valid $(foreach year,$(YEARS),$(TARGET_DIR)/dnb$(year).i5.xml) |
| Marc Kupietz | 157e079 | 2024-04-13 12:46:42 +0200 | [diff] [blame] | 61 | |
| Marc Kupietz | 84da39d | 2024-07-25 20:23:50 +0200 | [diff] [blame] | 62 | |
| 63 | $(BUILD_DIR)/%: $(SRC_DIR)/**/%.epub |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 64 | mkdir -p $@ |
| 65 | echo "Converting $< to $@" |
| 66 | unzip -q -o $< -d $@ |
| Marc Kupietz | 51714ed | 2024-04-13 12:47:39 +0200 | [diff] [blame] | 67 | chmod -R ug+rwX $@ |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 68 | |
| Marc Kupietz | dfbc3f4 | 2024-03-16 11:46:41 +0100 | [diff] [blame] | 69 | $(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] | 70 | mkdir -p $(TARGET_DIR) |
| 71 | echo "Converting $< to $@" |
| Marc Kupietz | 74fb31d | 2024-04-19 11:44:49 +0200 | [diff] [blame] | 72 | $(SAXON) -xsl:xslt/epub2i5.xsl $(shell find $< -name "*.opf") > $@ || (echo "WARN: ignoring invalid $@" && > $@) |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 73 | |
| 74 | %.zip: %.i5.xml |
| 75 | tei2korapxml -l warn -s -tk - < $< > $@ |
| 76 | |
| 77 | %.tree_tagger.zip: %.zip |
| Marc Kupietz | fa445da | 2024-03-16 10:46:22 +0100 | [diff] [blame] | 78 | $(KORAPXML2CONLLU) $< | pv | docker run --rm -i korap/conllu2treetagger -l german | conllu2korapxml > $@ |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 79 | |
| 80 | %.spacy.zip: %.zip |
| Marc Kupietz | fa445da | 2024-03-16 10:46:22 +0100 | [diff] [blame] | 81 | $(KORAPXML2CONLLU) $< | pv | docker run --rm -i korap/conllu2spacy | conllu2korapxml > $@ |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 82 | |
| Marc Kupietz | 8ea9228 | 2024-03-24 14:55:55 +0100 | [diff] [blame] | 83 | models/de.marmot: |
| 84 | mkdir -p models |
| Marc Kupietz | e8c80f3 | 2024-03-24 15:27:28 +0100 | [diff] [blame] | 85 | 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] | 86 | |
| 87 | models/german.mco: |
| 88 | mkdir -p models |
| Marc Kupietz | e8c80f3 | 2024-03-24 15:27:28 +0100 | [diff] [blame] | 89 | curl -sL -o $@ https://corpora.ids-mannheim.de/tools/$@ |
| Marc Kupietz | 8ea9228 | 2024-03-24 14:55:55 +0100 | [diff] [blame] | 90 | |
| Marc Kupietz | a553865 | 2024-04-21 15:49:30 +0200 | [diff] [blame] | 91 | models/dereko_domains_s.classifier: |
| 92 | mkdir -p models |
| 93 | curl -sL -o $@ https://corpora.ids-mannheim.de/tools/$@ |
| 94 | |
| Marc Kupietz | 8ea9228 | 2024-03-24 14:55:55 +0100 | [diff] [blame] | 95 | %.marmot-malt.zip: %.zip models/de.marmot models/german.mco |
| Marc Kupietz | ccf0904 | 2024-04-21 17:19:36 +0200 | [diff] [blame] | 96 | $(KORAPXML2CONLLU) -T $(MAX_THREADS) -t marmot:models/de.marmot -P malt:models/german.mco $< | conllu2korapxml -f "marmot dependency:malt" > $@ |
| Marc Kupietz | 8ea9228 | 2024-03-24 14:55:55 +0100 | [diff] [blame] | 97 | |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 98 | %.ud.zip: %.zip |
| Marc Kupietz | fa445da | 2024-03-16 10:46:22 +0100 | [diff] [blame] | 99 | $(KORAPXML2CONLLU) $< | pv | ./scripts/udpipe2 | conllu2korapxml > $@ |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 100 | |
| Marc Kupietz | d653bb8 | 2024-04-21 12:52:51 +0200 | [diff] [blame] | 101 | %.krill.tar: %.zip %.marmot-malt.zip %.tree_tagger.zip |
| Marc Kupietz | 8759751 | 2024-04-15 18:31:34 +0200 | [diff] [blame] | 102 | mkdir -p ${BUILD_DIR}/krill/$(basename $@) |
| Marc Kupietz | 3d82f56 | 2024-03-16 10:19:03 +0100 | [diff] [blame] | 103 | mkdir -p $(basename $@) |
| Marc Kupietz | d653bb8 | 2024-04-21 12:52:51 +0200 | [diff] [blame] | 104 | korapxml2krill archive --quiet -w -z -cfg krill-korap4dnb.cfg -c ${BUILD_DIR}/krill/$(basename $@)/korapxml2krill.cache -j $(MAX_THREADS) -te ${BUILD_DIR}/krill/$(basename $@) --non-word-tokens --meta I5 -i $< -i $(word 2,$^) -i $(word 3,$^) -o $(basename $@) |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 105 | |
| Marc Kupietz | fc32862 | 2024-03-16 14:31:34 +0100 | [diff] [blame] | 106 | %.json: %.krill.tar |
| 107 | rm -rf $@ |
| 108 | mkdir -p $@ |
| 109 | for f in $<; do tar -C $@ -xf $$f; done |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 110 | |
| Marc Kupietz | ab1f3ac | 2024-04-16 14:56:43 +0200 | [diff] [blame] | 111 | $(TARGET_DIR)/dnb.index: $(foreach year,$(YEARS),$(TARGET_DIR)/dnb$(year).json) |
| Marc Kupietz | 22ad4b7 | 2024-03-17 16:24:21 +0100 | [diff] [blame] | 112 | rm -rf $@ |
| Marc Kupietz | ab1f3ac | 2024-04-16 14:56:43 +0200 | [diff] [blame] | 113 | java -jar lib/Krill-Indexer.jar -c lib/krill.conf -i $(subst " ",;,$^) -o $@ |
| Marc Kupietz | 22ad4b7 | 2024-03-17 16:24:21 +0100 | [diff] [blame] | 114 | |
| Marc Kupietz | ab1f3ac | 2024-04-16 14:56:43 +0200 | [diff] [blame] | 115 | $(TARGET_DIR)/dnb.index.tar.xz: $(TARGET_DIR)/dnb.index |
| Marc Kupietz | 22ad4b7 | 2024-03-17 16:24:21 +0100 | [diff] [blame] | 116 | tar -I 'xz -T0' -C $(dir $<) -cf $@ $(notdir $<) |
| 117 | |
| Marc Kupietz | b769699 | 2024-03-18 17:41:15 +0100 | [diff] [blame] | 118 | deploy: $(TARGET_DIR)/dnb.index.tar.xz korap4dnb-compose.yml |
| 119 | rsync -v $^ $(DEPLOY_USER)@$(DEPLOY_HOST):$(DEPLOY_PATH)/ |
| 120 | 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" |
| 121 | |
| 122 | show-server-log: |
| 123 | ssh $(DEPLOY_USER)@$(DEPLOY_HOST) "cd $(DEPLOY_PATH) && docker compose -p korap4dnb --profile=lite -f korap4dnb-compose.yml logs -f" |
| 124 | |
| 125 | show-server-status: |
| 126 | ssh $(DEPLOY_USER)@$(DEPLOY_HOST) "cd $(DEPLOY_PATH) && docker compose -p korap4dnb --profile=lite -f korap4dnb-compose.yml ps" |
| 127 | |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 128 | clean: |
| 129 | rm -rf $(BUILD_DIR) $(TARGET_DIR) |