blob: 996b37247d8a28850442559060b17e450a754bb8 [file] [log] [blame]
ifneq (,$(filter test,$(MAKECMDGOALS)))
SRC_DIR = test/resources/DNB
YEARS=13 18 23
else
SRC_DIR ?= ./DeLiKo@DNB
YEARS ?= $(shell seq -w 2024 -1 2005 | sed 's/^.*\([0-9][0-9]\)/\1/')
#YEARS ?= $(shell seq -w 2005 2023 | sed 's/^.*\([0-9][0-9]\)/\1/')
endif
BUILD_DIR = build
TARGET_DIR ?= target
DEPLOY_HOST ?= compute.ids-mannheim.de
DEPLOY_USER ?= korap
DEPLOY_PATH ?= /export/netapp/korap4dnb
MAX_THREADS ?= 8 # $(shell nproc)
MAKE ?= make -j $(shell nproc)
SLACK ?= slack
KORAPXMLTOOL_HEAP ?= $(shell echo "$$(($(MAX_THREADS) * 2500))")
KORAPXMLTOOL ?= java -Xmx$(KORAPXMLTOOL_HEAP)m -jar lib/korapxmltool.jar
MARMOTMALTOOL ?= java -Xmx96000m -jar lib/korapxmltool.jar
SPACYXMLTOOL ?= java -Xmx250g -jar lib/korapxmltool.jar
SAXON ?= java -Djava.util.logging.config.file=/logging.properties -cp lib/saxon-ee-12.5.jar:lib/xmlresolver-5.2.2.jar:lib/textclassifier.jar:lib/xmlresolver-5.2.2-data.jar net.sf.saxon.Transform -expand:off -catalog:"lib/dtds/xhtml11/xhtmlcatalog.xml;lib/dtds/xhtml/dtd/xhtmlcatalog.xml"
.DELETE_ON_ERROR:
.PHONY: all clean test i5 i5valid krill malt index deploy show-server-log show-server-status
.PRECIOUS: $(TARGET_DIR)/%.i5.xml $(TARGET_DIR)/dnb%.pre.i5.xml %.zip %.tree_tagger.zip %.ud.zip %.marmot-malt.zip %.spacy.zip %.i5.xml %.tar
all: index
krill: $(foreach year,$(YEARS),$(TARGET_DIR)/dnb$(year).krill.tar)
index: $(TARGET_DIR)/dnb.index
# EPUBS := ($shell cat .epubs_cache)
EPUBS := $(wildcard $(SRC_DIR)/**/*.epub)
$(TARGET_DIR)/dnb%.i5.xml: $(TARGET_DIR)/dnb%.pre.i5.xml xslt/pass2.xsl xslt/pass3.xsl models/dereko_domains_s.classifier
$(SAXON) -xsl:xslt/pass2.xsl $< | $(SAXON) -xsl:xslt/pass3.xsl - > $@
$(TARGET_DIR)/dnb%.pre.i5.xml: $(patsubst %.epub,$(TARGET_DIR)/%.i5.xml,$(notdir $(EPUBS)))
rm -f $(TARGET_DIR)/filelist$*.txt
head -n -1 xslt/idsCorpus-template.xml | sed -e 's/{YY}/$*/' > $@
@find -L $(SRC_DIR) -type f -name '*.epub' | sort -u | while read src; do \
f=$(TARGET_DIR)/$$(basename $${src%.epub}).i5.xml; \
if ! grep -sq "$$f" $(TARGET_DIR)/filelist$*.txt && head -500 "$$f" | grep -Eq '<pubDate type="year">..$*'; then \
echo $$f >> $(TARGET_DIR)/filelist$*.txt; \
cat "$$f" >> $@; \
fi; \
done
tail -n 1 xslt/idsCorpus-template.xml >> $@
test: models/dereko_domains_s.classifier i5valid test/test-xml.sh
bash test/test-xml.sh
i5: $(foreach year,$(YEARS),$(TARGET_DIR)/dnb$(year).i5.xml)
i5valid: i5
xmllint --noout --valid $(foreach year,$(YEARS),$(TARGET_DIR)/dnb$(year).i5.xml)
$(BUILD_DIR)/%: $(SRC_DIR)/**/%.epub
mkdir -p $@.tmp
echo "Converting $< to $@"
echo "<originalPath>$<</originalPath>" > $@.tmp/originalPath.xml
unzip -q -o $< -d $@.tmp
chmod -R ug+rwX $@.tmp
mv $@.tmp $@
$(TARGET_DIR)/%.i5.xml: $(BUILD_DIR)/% xslt/epub2i5.xsl xslt/idsCorpus-template.xml
mkdir -p $(TARGET_DIR)
echo "Converting $< to $@"
$(SAXON) -xsl:xslt/epub2i5.xsl $(shell find $< -name "*.opf") $(shell if cat `find $< -name originalPath.xml` | grep -iq buchpreis; then echo buchpreis=1; fi)> $@ || \
(sleep 5 && $(SAXON) -xsl:xslt/epub2i5.xsl $(shell find $< -name "*.opf") $(shell if cat `find $< -name originalPath.xml` | grep -iq buchpreis; then echo buchpreis=1; fi) debug=1 > $@) || (echo "WARN: ignoring invalid $@" && > $@)
%.zip: %.i5.xml
tei2korapxml -l warn -s -tk - < $< > $@ || ($(SLACK) "$@ failed" && false)
printf "%s\t%s\n" "$(grep -c '<idsText ' $<)" "$(unzip -l $@ | grep data.xml | wc -l)"
$(SLACK) "$@ created"
%.tree_tagger.zip: %.zip
$(KORAPXMLTOOL) -T 1 -A "docker run --rm -i korap/conllu2treetagger -l german" -f zip --overwrite $<
# $(KORAPXMLTOOL) $< | pv | docker run --rm -i korap/conllu2treetagger -l german | conllu2korapxml > $@
%.spacy.zip: %.zip
$(SPACYXMLTOOL) -T 8 -A "docker run -e SPACY_USE_DEPENDENCIES=True --rm -i korap/conllu2spacy:latest" -f zip --overwrite $<
models/de.marmot:
mkdir -p models
curl -sL -o $@ https://cistern.cis.lmu.de/marmot/models/CURRENT/spmrl/de.marmot
models/german.mco:
mkdir -p models
curl -sL -o $@ https://corpora.ids-mannheim.de/tools/$@
models/dereko_domains_s.classifier:
mkdir -p models
curl -sL -o $@ https://corpora.ids-mannheim.de/tools/$@
%.marmot-malt.zip: %.zip models/de.marmot models/german.mco
$(MARMOTMALTOOL) -T $(MAX_THREADS) -t marmot:models/de.marmot -P malt:models/german.mco -f zip --overwrite $<
malt: $(foreach year,$(YEARS),$(TARGET_DIR)/dnb$(year).marmot-malt.zip)
%.ud.zip: %.zip
$(KORAPXMLTOOL) $< | pv | ./scripts/udpipe2 | conllu2korapxml > $@
%.krill.tar: %.zip %.marmot-malt.zip %.tree_tagger.zip
mkdir -p ${BUILD_DIR}/krill/$(basename $@)
mkdir -p $(basename $@)
K2K_PUBLISHER_STRING=1 K2K_TRANSLATOR_TEXT=1 korapxml2krill archive --quiet -w -z -cfg krill-korap4dnb.cfg -c ${BUILD_DIR}/krill/$(basename $@)/korapxml2krill.cache -j 30 -te ${BUILD_DIR}/krill/$(basename $@) --non-word-tokens --meta I5 -i $< -i $(word 2,$^) -i $(word 3,$^) -o $(basename $@)
$(SLACK) "$(basename $@) krill archive created"
%.json: %.krill.tar
rm -rf $@
mkdir -p $@
for f in $<; do tar -C $@ -xf $$f; done
krill: $(foreach year,$(YEARS),$(TARGET_DIR)/dnb$(year).krill.tar)
$(TARGET_DIR)/dnb.index.tar.xz: $(TARGET_DIR)/dnb.index
tar -I 'xz -T0' -C $(dir $<) -cf $@ $(notdir $<)
deploy: $(TARGET_DIR)/dnb.index.tar.xz korap4dnb-compose.yml
rsync -v $^ $(DEPLOY_USER)@$(DEPLOY_HOST):$(DEPLOY_PATH)/
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"
show-server-log:
ssh $(DEPLOY_USER)@$(DEPLOY_HOST) "cd $(DEPLOY_PATH) && docker compose -p korap4dnb --profile=lite -f korap4dnb-compose.yml logs -f"
show-server-status:
ssh $(DEPLOY_USER)@$(DEPLOY_HOST) "cd $(DEPLOY_PATH) && docker compose -p korap4dnb --profile=lite -f korap4dnb-compose.yml ps"
clean:
rm -rf $(BUILD_DIR) $(TARGET_DIR)
$(TARGET_DIR)/dnb.index: $(foreach year,$(YEARS),$(TARGET_DIR)/dnb$(year).krill.tar)
rm -rf $@
java -jar lib/Krill-Indexer.jar --progress -c lib/krill.conf -i $(subst " ",;,$^) -o $@