Makefile

joyopdf_url = http://kokugo.bunka.go.jp/kokugo_nihongo/joho/kijun/naikaku/pdf/joyokanjihyo_20101130.pdf
pdfbox_url = http://www-us.apache.org/dist/pdfbox/2.0.2/pdfbox-app-2.0.2.jar

cachedir = cache
moduledir = joyodb
joyopdf = $(cachedir)/joyokanjihyo_20101130.pdf
joyotxt = $(cachedir)/joyokanjihyo_20101130.txt
pdfbox = $(cachedir)/pdfbox-app-2.0.2.jar

all: $(cachedir) $(joyotxt) $(moduledir)/__init__.py

$(cachedir):
	mkdir -p $(cachedir)

$(joyotxt): $(joyopdf) $(pdfbox)
	java -jar $(pdfbox) ExtractText $(joyopdf) -console > $(joyotxt)
	@# pdfbox generates ASCII bells (\x07) for some reason.
	sed -e "s/\x07//g" -i $(joyotxt)
$(joyopdf):
	wget $(joyopdf_url) -O $(joyopdf) 
$(pdfbox):
	wget $(pdfbox_url) -O $(pdfbox)

$(moduledir)/__init__.py: $(moduledir)/__init__.py.in
	@echo "Building $@."
	@echo "# Autogenerated from $(moduledir)/__init__.py.in; do not edit." \
	    > $(moduledir)/__init__.py

	sed -e "s,\$$JOYOHYO_TXT,'`readlink -f $(joyotxt)`',"\
	    $(moduledir)/__init__.py.in \
	    >> $(moduledir)/__init__.py

wikipedia_url = 'https://en.wikipedia.org/w/index.php?title=List_of_jōyō_kanji&oldid=727326828'
wikipedia_html = $(cachedir)/List_of_joyo_kanji.html
kanjidic_url = ftp.monash.edu.au::nihongo/kanjidic_comb_utf8
kanjidic = $(cachedir)/kanjidic_comb_utf8
jmdict_url = ftp.monash.edu.au::nihongo/JMdict
jmdict = $(cachedir)/JMdict

test: all $(wikipedia_html) $(kanjidic) $(jmdict)
	python3 test/test.py

$(wikipedia_html):
	wget $(wikipedia_url) -O $(wikipedia_html)

$(kanjidic):
	rsync -z -q $(kanjidic_url) $(kanjidic)
$(jmdict):
	rsync -z -q $(jmdict_url) $(jmdict)

clean:
	rm $(cachedir)/* $(moduledir)/__init__.py

.PHONY: all clean test