blob: e43f4a0360849e34b1154c36db00663b8de63862 [file] [log] [blame]
Marc Kupietz1a422662024-03-16 09:34:10 +01001SRC_DIR ?= test/resources/DNB
2BUILD_DIR = build
3TARGET_DIR ?= target
4
Marc Kupietz3d82f562024-03-16 10:19:03 +01005.PHONY: all clean test krill
Marc Kupietz1a422662024-03-16 09:34:10 +01006
Marc Kupietz3d82f562024-03-16 10:19:03 +01007.PRECIOUS: $(TARGET_DIR)/%.zip $(TARGET_DIR)/%.i5.xml $(TARGET_DIR)/%.tar
Marc Kupietz1a422662024-03-16 09:34:10 +01008
9all: $(TARGET_DIR)/dnb.i5.xml
10
Marc Kupietz3d82f562024-03-16 10:19:03 +010011krill: $(TARGET_DIR)/dnb.krill.tar
12
Marc Kupietz1a422662024-03-16 09:34:10 +010013$(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
18test: $(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 Kupietz3d82f562024-03-16 10:19:03 +010046%.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 Kupietz1a422662024-03-16 09:34:10 +010051
52json: *.krill.tar
53 rm -rf json
54 mkdir -p json
55 for f in $^; do tar -C json -xf $$f; done
56
57clean:
58 rm -rf $(BUILD_DIR) $(TARGET_DIR)
59