blob: 7ec798d08a41060761b9f501ff8411146add6c63 [file] [log] [blame]
Marc Kupietz1a422662024-03-16 09:34:10 +01001SRC_DIR ?= test/resources/DNB
2BUILD_DIR = build
3TARGET_DIR ?= target
4
5
6
7.PHONY: all clean test
8
9
10all: $(TARGET_DIR)/dnb.i5.xml
11
12$(TARGET_DIR)/dnb.i5.xml: $(patsubst $(SRC_DIR)/%.epub,$(TARGET_DIR)/%.i5.xml,$(wildcard $(SRC_DIR)/*.epub))
13 head -n -1 xslt/idsCorpus-template.xml > $@
14 cat $^ >> $@
15 tail -n 1 xslt/idsCorpus-template.xml >> $@
16
17test: $(TARGET_DIR)/dnb.i5.xml xslt/epub2i5.xsl
18 xmllint --noout --valid $<
19
20$(BUILD_DIR)/%: $(SRC_DIR)/%.epub
21 mkdir -p $@
22 echo "Converting $< to $@"
23 unzip -q -o $< -d $@
24
25$(TARGET_DIR)/%.i5.xml: $(BUILD_DIR)/% xslt/epub2i5.xsl
26 mkdir -p $(TARGET_DIR)
27 echo "Converting $< to $@"
28 java -jar lib/saxon9ee.jar -xsl:xslt/epub2i5.xsl $</*/content.opf > $@
29
30%.zip: %.i5.xml
31 tei2korapxml -l warn -s -tk - < $< > $@
32
33%.tree_tagger.zip: %.zip
34 korapxml2conllu $< | pv | docker run --rm -i korap/conllu2treetagger -l german | conllu2korapxml > $@
35
36%.spacy.zip: %.zip
37 korapxml2conllu $< | pv | docker run --rm -i korap/conllu2spacy | conllu2korapxml > $@
38
39%.ud.zip: %.zip
40 korapxml2conllu $< | pv | ./scripts/udpipe2 | conllu2korapxml > $@
41
42%.cmc.zip: %.zip
43 korapxml2conllu $< | pv | conllu2cmc -s | conllu2korapxml > $@
44
45%.krill.tar: %.zip %.ud.zip %.cmc.zip
46 korapxml2krill archive --quiet -w -z -cfg krill-kokokom.cfg --non-word-tokens --meta I5 -i $< -i $(word 2,$^) -i $(word 3,$^) -o $(basename $@)
47
48json: *.krill.tar
49 rm -rf json
50 mkdir -p json
51 for f in $^; do tar -C json -xf $$f; done
52
53clean:
54 rm -rf $(BUILD_DIR) $(TARGET_DIR)
55