blob: c143581b3db883326b7439dc17a174864ad85c2e [file] [log] [blame]
Marc Kupietz4292bc02026-03-07 16:12:52 +01001SHELL := /bin/bash
Marc Kupietz89a44072025-12-11 07:14:00 +01002SRC_DIR ?= I5
3
4# Discover all *.i5.xml files in SRC_DIR
5I5_FILES := $(wildcard $(SRC_DIR)/*.i5.xml)
6BASENAMES := $(patsubst %.i5.xml,%,$(notdir $(I5_FILES)))
7
Marc Kupietz99ff2b02026-06-05 08:54:37 +02008# Standard TEI P5 support
9TEI_DIR ?= TEI
10TEI_FILES := $(wildcard $(TEI_DIR)/*.xml)
11TEI_ZIP_NAME ?= tei
12TEI_FLAGS ?= --auto-textsigle 'TEI/XYZ.00001' # --xmlid-to-textsigle '([A-Z]+)\.(.*)\.([0-9]+)\.*([0-9])@$$1/$$2/$$3$$4'
13export TEI_FLAGS
14
15ifneq ($(TEI_FILES),)
16BASENAMES += $(TEI_ZIP_NAME)
17endif
18
Marc Kupietz89a44072025-12-11 07:14:00 +010019BUILD_DIR = build
Marc Kupietz15b313d2025-12-11 12:20:56 +010020TARGET_DIR ?= ./target
Marc Kupietz89a44072025-12-11 07:14:00 +010021MAX_THREADS ?= 8 # $(shell nproc)
22MAKE ?= make -j $(shell nproc)
Marc Kupietz15b313d2025-12-11 12:20:56 +010023# KORAPXMLTOOL_HEAP ?= $(shell echo "$$(($(MAX_THREADS) * 2500))")
Marc Kupietz89a44072025-12-11 07:14:00 +010024KORAPXMLTOOL ?= ./bin/korapxmltool
Marc Kupietz15b313d2025-12-11 12:20:56 +010025KORAPXMLTOOL_MODELS_PATH ?= models
Marc Kupietz4292bc02026-03-07 16:12:52 +010026DOCKER_CPU_SHARES ?= # e.g. 512 for lower priority (default Docker value is 1024)
Marc Kupietz89a44072025-12-11 07:14:00 +010027
28.DELETE_ON_ERROR:
29
Marc Kupietz463da332026-03-08 13:19:15 +010030.PHONY: all clean test index korap check-src pre-krill krill
Marc Kupietz89a44072025-12-11 07:14:00 +010031
Marc Kupietz5b3ecc02025-12-15 14:43:09 +010032.PRECIOUS: $(BUILD_DIR)/%.zip $(BUILD_DIR)/%.tree_tagger.zip $(BUILD_DIR)/%.marmot-malt.zip $(BUILD_DIR)/%.spacy.zip $(BUILD_DIR)/%.corenlp.zip $(BUILD_DIR)/%.cmc.zip $(BUILD_DIR)/%.opennlp.zip $(BUILD_DIR)/%.krill.tar %.i5.xml
Marc Kupietz89a44072025-12-11 07:14:00 +010033
Marc Kupietz15b313d2025-12-11 12:20:56 +010034all: check-src korap
Marc Kupietz89a44072025-12-11 07:14:00 +010035
Marc Kupietz15b313d2025-12-11 12:20:56 +010036index: check-src $(TARGET_DIR)/index
37
38check-src:
Marc Kupietz99ff2b02026-06-05 08:54:37 +020039 @if [ ! -d "$(SRC_DIR)" ] && [ ! -d "$(TEI_DIR)" ]; then \
40 echo "Error: Neither SRC_DIR '$(SRC_DIR)' nor TEI_DIR '$(TEI_DIR)' exists."; \
41 echo "Please place your .i5.xml files in '$(SRC_DIR)' or your TEI .xml files in '$(TEI_DIR)'."; \
Marc Kupietz15b313d2025-12-11 12:20:56 +010042 exit 1; \
43 fi
Marc Kupietz99ff2b02026-06-05 08:54:37 +020044 @if [ -z "$$(find "$(SRC_DIR)" -maxdepth 1 -name '*.i5.xml' -print -quit 2>/dev/null)" ] && \
45 [ -z "$$(find "$(TEI_DIR)" -maxdepth 1 -name '*.xml' -print -quit 2>/dev/null)" ]; then \
46 echo "Error: No .i5.xml files found in '$(SRC_DIR)' and no .xml files found in '$(TEI_DIR)'."; \
Marc Kupietz15b313d2025-12-11 12:20:56 +010047 exit 1; \
48 fi
Marc Kupietz89a44072025-12-11 07:14:00 +010049
50$(BUILD_DIR)/%.zip: $(SRC_DIR)/%.i5.xml
51 mkdir -p $(BUILD_DIR)
Marc Kupietz4292bc02026-03-07 16:12:52 +010052 docker run --rm -i $(if $(DOCKER_CPU_SHARES),--cpu-shares $(DOCKER_CPU_SHARES)) korap/tei2korapxml:latest -l warn -s -tk - < $< > $@ 2> >(tee $(@:.zip=.log) >&2)
53# docker run --rm $(if $(DOCKER_CPU_SHARES),--cpu-shares $(DOCKER_CPU_SHARES)) -v $(abspath $<):/input.i5.xml:ro korap/tei2korapxml:latest --progress -l warn -s -tk /input.i5.xml > $@ 2> >(tee $(@:.zip=.log) >&2)
54 printf "%s\t%s\n" "$$(grep -c '<idsText ' $<)" "$$(unzip -l $@ | grep data.xml | wc -l)"
Marc Kupietz89a44072025-12-11 07:14:00 +010055
Marc Kupietz99ff2b02026-06-05 08:54:37 +020056$(BUILD_DIR)/$(TEI_ZIP_NAME).zip: $(TEI_FILES)
57 mkdir -p $(BUILD_DIR)
58 docker run --rm --entrypoint /bin/sh $(if $(DOCKER_CPU_SHARES),--cpu-shares $(DOCKER_CPU_SHARES)) -v "$(abspath $(TEI_DIR)):/input:ro" korap/tei2korapxml:latest -c 'tei2korapxml -l warn -s -tk '"$$TEI_FLAGS"' /input/*.xml' > $@ 2> >(tee $(@:.zip=.log) >&2)
59 printf "%s\t%s\n" "$$(cat $^ | grep -c '<teiHeader')" "$$(unzip -l $@ | grep data.xml | wc -l)"
60
Marc Kupietz89a44072025-12-11 07:14:00 +010061
62$(BUILD_DIR)/%.tree_tagger.zip: $(BUILD_DIR)/%.zip bin/korapxmltool
Marc Kupietz5b3ecc02025-12-15 14:43:09 +010063 $(KORAPXMLTOOL) -j 1 -T treetagger -t zip --force -D $(BUILD_DIR) $<
Marc Kupietz89a44072025-12-11 07:14:00 +010064# $(KORAPXMLTOOL) $< | pv | docker run --rm -i korap/conllu2treetagger -l german | conllu2korapxml > $@
65
66$(BUILD_DIR)/%.spacy.zip: $(BUILD_DIR)/%.zip bin/korapxmltool
Marc Kupietz15b313d2025-12-11 12:20:56 +010067 $(KORAPXMLTOOL) -P spacy -t zip --force -D $(BUILD_DIR) $<
Marc Kupietz89a44072025-12-11 07:14:00 +010068
Marc Kupietz15b313d2025-12-11 12:20:56 +010069lib/Krill-Indexer.jar:
70 mkdir -p lib
71 curl -sL -o $@ https://github.com/korap/Krill/releases/latest/download/Krill-Indexer.jar
Marc Kupietz5b3ecc02025-12-15 14:43:09 +010072
Marc Kupietz89a44072025-12-11 07:14:00 +010073bin/korapxmltool:
74 mkdir -p bin
Marc Kupietz15b313d2025-12-11 12:20:56 +010075 curl -sL -o $@ https://github.com/korap/korapxmltool/releases/latest/download/korapxmltool
Marc Kupietz89a44072025-12-11 07:14:00 +010076 chmod +x $@
77
Marc Kupietz15b313d2025-12-11 12:20:56 +010078$(KORAPXMLTOOL_MODELS_PATH)/de.marmot:
79 mkdir -p $(KORAPXMLTOOL_MODELS_PATH)
Marc Kupietz89a44072025-12-11 07:14:00 +010080 curl -sL -o $@ https://cistern.cis.lmu.de/marmot/models/CURRENT/spmrl/de.marmot
81
Marc Kupietz15b313d2025-12-11 12:20:56 +010082$(KORAPXMLTOOL_MODELS_PATH)/german.mco:
83 mkdir -p $(KORAPXMLTOOL_MODELS_PATH)
Marc Kupietz89a44072025-12-11 07:14:00 +010084 curl -sL -o $@ https://corpora.ids-mannheim.de/tools/$@
85
Marc Kupietz15b313d2025-12-11 12:20:56 +010086$(KORAPXMLTOOL_MODELS_PATH)/dereko_domains_s.classifier:
87 mkdir -p $(KORAPXMLTOOL_MODELS_PATH)
88 curl -sL -o $@ https://corpora.ids-mannheim.de/tools/models/$@
89
90$(KORAPXMLTOOL_MODELS_PATH)/german-fast.tagger:
91 mkdir -p $(KORAPXMLTOOL_MODELS_PATH)
Marc Kupietz89a44072025-12-11 07:14:00 +010092 curl -sL -o $@ https://corpora.ids-mannheim.de/tools/$@
93
Marc Kupietz15b313d2025-12-11 12:20:56 +010094$(KORAPXMLTOOL_MODELS_PATH)/germanSR.ser.gz:
95 mkdir -p $(KORAPXMLTOOL_MODELS_PATH)
Marc Kupietz89a44072025-12-11 07:14:00 +010096 curl -sL -o $@ https://corpora.ids-mannheim.de/tools/$@
97
Marc Kupietz15b313d2025-12-11 12:20:56 +010098$(KORAPXMLTOOL_MODELS_PATH)/de-pos-maxent.bin:
99 mkdir -p $(KORAPXMLTOOL_MODELS_PATH)
Marc Kupietz89a44072025-12-11 07:14:00 +0100100 curl -sL -o $@ https://corpora.ids-mannheim.de/tools/$@
101
Marc Kupietz15b313d2025-12-11 12:20:56 +0100102$(BUILD_DIR)/%.marmot-malt.zip: $(BUILD_DIR)/%.zip $(KORAPXMLTOOL_MODELS_PATH)/de.marmot $(KORAPXMLTOOL_MODELS_PATH)/german.mco bin/korapxmltool
Marc Kupietz89a44072025-12-11 07:14:00 +0100103 $(KORAPXMLTOOL) -T marmot:models/de.marmot -P malt:models/german.mco -t zip --force -D $(BUILD_DIR) $<
Marc Kupietz5b3ecc02025-12-15 14:43:09 +0100104
Marc Kupietz15b313d2025-12-11 12:20:56 +0100105$(BUILD_DIR)/%.corenlp.zip: $(BUILD_DIR)/%.zip $(KORAPXMLTOOL_MODELS_PATH)/german-fast.tagger $(KORAPXMLTOOL_MODELS_PATH)/germanSR.ser.gz bin/korapxmltool
Marc Kupietz89a44072025-12-11 07:14:00 +0100106 $(KORAPXMLTOOL) -T corenlp -P corenlp -t zip --force -D $(BUILD_DIR) $<
107
Marc Kupietz15b313d2025-12-11 12:20:56 +0100108$(BUILD_DIR)/%.opennlp.zip: $(BUILD_DIR)/%.zip $(KORAPXMLTOOL_MODELS_PATH)/de-pos-maxent.bin bin/korapxmltool
Marc Kupietz89a44072025-12-11 07:14:00 +0100109 $(KORAPXMLTOOL) -T opennlp -t zip --force -D $(BUILD_DIR) $<
110
Marc Kupietz800f9bb2025-12-11 17:30:24 +0100111$(BUILD_DIR)/%.cmc.zip: $(BUILD_DIR)/%.zip bin/korapxmltool
Marc Kupietz5b3ecc02025-12-15 14:43:09 +0100112 $(KORAPXMLTOOL) -j 1 -A "docker run --rm -i korap/conllu-cmc -s" -l error -F cmc -t zip --force -D $(BUILD_DIR) $<
Marc Kupietz800f9bb2025-12-11 17:30:24 +0100113
Marc Kupietza2876eb2026-03-07 22:03:17 +0100114$(BUILD_DIR)/%.gender.zip: $(BUILD_DIR)/%.zip bin/conllu-gender
Marc Kupietz463da332026-03-08 13:19:15 +0100115 $(KORAPXMLTOOL) -j 1 -A "bin/conllu-gender -s" -l WARNING -F gender -t zip --force -D $(BUILD_DIR) $<
Marc Kupietza2876eb2026-03-07 22:03:17 +0100116
Marc Kupietz89a44072025-12-11 07:14:00 +0100117# udpipe target removed as requested
118# %.ud.zip: %.zip
119# $(KORAPXMLTOOL) $< | pv | ./scripts/udpipe2 | conllu2korapxml > $@
120
Marc Kupietza2876eb2026-03-07 22:03:17 +0100121KRILL_PREREQS := $(foreach base,$(BASENAMES),$(BUILD_DIR)/$(base).zip $(BUILD_DIR)/$(base).marmot-malt.zip $(BUILD_DIR)/$(base).tree_tagger.zip $(BUILD_DIR)/$(base).spacy.zip $(BUILD_DIR)/$(base).corenlp.zip $(BUILD_DIR)/$(base).opennlp.zip $(BUILD_DIR)/$(base).gender.zip)
Marc Kupietz4292bc02026-03-07 16:12:52 +0100122
123pre-krill: check-src $(KRILL_PREREQS)
124
Marc Kupietza2876eb2026-03-07 22:03:17 +0100125$(BUILD_DIR)/%.krill.tar: $(BUILD_DIR)/%.zip $(BUILD_DIR)/%.marmot-malt.zip $(BUILD_DIR)/%.tree_tagger.zip $(BUILD_DIR)/%.spacy.zip $(BUILD_DIR)/%.corenlp.zip $(BUILD_DIR)/%.opennlp.zip $(BUILD_DIR)/%.gender.zip
Marc Kupietz4292bc02026-03-07 16:12:52 +0100126 $(KORAPXMLTOOL) --non-word-tokens -f -t krill -D $(BUILD_DIR) $(basename $<)*.zip
Marc Kupietz89a44072025-12-11 07:14:00 +0100127
Marc Kupietz463da332026-03-08 13:19:15 +0100128krill: $(foreach base,$(BASENAMES),$(BUILD_DIR)/$(base).krill.tar)
129
Marc Kupietz15b313d2025-12-11 12:20:56 +0100130$(TARGET_DIR)/index: $(foreach base,$(BASENAMES),$(BUILD_DIR)/$(base).krill.tar)
131 make lib/Krill-Indexer.jar
132 touch lib/krill.cfg
133 mkdir -p $(TARGET_DIR)
134 java -jar lib/Krill-Indexer.jar -c lib/krill.cfg --progress -i $(subst " ",;,$^) -o $@
135
136korap: check-src $(TARGET_DIR)/index
137 curl https://raw.githubusercontent.com/KorAP/KorAP-Docker/master/compose.yaml | INDEX='$(TARGET_DIR)/index' docker compose -p korap -f - --profile=lite --profile=example up
Marc Kupietz89a44072025-12-11 07:14:00 +0100138
139$(TARGET_DIR)/index.tar.xz: $(TARGET_DIR)/index
140 tar -I 'xz -T0' -C $(dir $<) -cf $@ $(notdir $<)
141
142clean:
143 rm -rf $(BUILD_DIR) $(TARGET_DIR)