| 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 |
| 4 | |
| Marc Kupietz | 3d82f56 | 2024-03-16 10:19:03 +0100 | [diff] [blame] | 5 | .PHONY: all clean test krill |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 6 | |
| Marc Kupietz | 95bce0d | 2024-03-16 10:37:45 +0100 | [diff] [blame^] | 7 | .PRECIOUS: %.zip $(TARGET_DIR)/%.zip $(TARGET_DIR)/%.i5.xml $(TARGET_DIR)/%.tar |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 8 | |
| 9 | all: $(TARGET_DIR)/dnb.i5.xml |
| 10 | |
| Marc Kupietz | 3d82f56 | 2024-03-16 10:19:03 +0100 | [diff] [blame] | 11 | krill: $(TARGET_DIR)/dnb.krill.tar |
| 12 | |
| Marc Kupietz | 1a42266 | 2024-03-16 09:34:10 +0100 | [diff] [blame] | 13 | $(TARGET_DIR)/dnb.i5.xml: $(patsubst $(SRC_DIR)/%.epub,$(TARGET_DIR)/%.i5.xml,$(wildcard $(SRC_DIR)/*.epub)) |
| 14 | head -n -1 xslt/idsCorpus-template.xml > $@ |
| 15 | cat $^ >> $@ |
| 16 | tail -n 1 xslt/idsCorpus-template.xml >> $@ |
| 17 | |
| 18 | test: $(TARGET_DIR)/dnb.i5.xml xslt/epub2i5.xsl |
| 19 | xmllint --noout --valid $< |
| 20 | |
| 21 | $(BUILD_DIR)/%: $(SRC_DIR)/%.epub |
| 22 | mkdir -p $@ |
| 23 | echo "Converting $< to $@" |
| 24 | unzip -q -o $< -d $@ |
| 25 | |
| 26 | $(TARGET_DIR)/%.i5.xml: $(BUILD_DIR)/% xslt/epub2i5.xsl |
| 27 | mkdir -p $(TARGET_DIR) |
| 28 | echo "Converting $< to $@" |
| 29 | java -jar lib/saxon9ee.jar -xsl:xslt/epub2i5.xsl $</*/content.opf > $@ |
| 30 | |
| 31 | %.zip: %.i5.xml |
| 32 | tei2korapxml -l warn -s -tk - < $< > $@ |
| 33 | |
| 34 | %.tree_tagger.zip: %.zip |
| 35 | korapxml2conllu $< | pv | docker run --rm -i korap/conllu2treetagger -l german | conllu2korapxml > $@ |
| 36 | |
| 37 | %.spacy.zip: %.zip |
| 38 | korapxml2conllu $< | pv | docker run --rm -i korap/conllu2spacy | conllu2korapxml > $@ |
| 39 | |
| 40 | %.ud.zip: %.zip |
| 41 | korapxml2conllu $< | pv | ./scripts/udpipe2 | conllu2korapxml > $@ |
| 42 | |
| 43 | %.cmc.zip: %.zip |
| 44 | korapxml2conllu $< | pv | conllu2cmc -s | conllu2korapxml > $@ |
| 45 | |
| Marc Kupietz | 3d82f56 | 2024-03-16 10:19:03 +0100 | [diff] [blame] | 46 | %.krill.tar: %.zip |
| 47 | mkdir -p $(basename $@) |
| 48 | korapxml2krill archive --quiet -w -z -cfg krill-korap4dnb.cfg --non-word-tokens --meta I5 -i $< -o $(basename $@) |
| 49 | # add annotations later |
| 50 | # korapxml2krill archive --quiet -w -z -cfg krill-korap4dnb.cfg --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] | 51 | |
| 52 | json: *.krill.tar |
| 53 | rm -rf json |
| 54 | mkdir -p json |
| 55 | for f in $^; do tar -C json -xf $$f; done |
| 56 | |
| 57 | clean: |
| 58 | rm -rf $(BUILD_DIR) $(TARGET_DIR) |
| 59 | |