Move jmdict stuff to its own makefile.

Julien LepillerMon Jun 22 22:45:19+0200 2020

d9acf92

Move jmdict stuff to its own makefile.

Makefile

22
33
all: site
44
5-
JMDICT_LANGS=dut fre ger hun rus slv spa swe
6-
7-
DICOS=dicos/JMdict_e.nani $(addprefix dicos/JMdict_, $(addsuffix .nani, $(JMDICT_LANGS)))
8-
9-
DOWNLOADS=dictionaries/JMdict_e.xml dictionaries/JMdict.xml dictionaries/frequency.tsv
5+
DICOS=
6+
DOWNLOADS=
107
118
include radicals.mk
129
include wadoku.mk
10+
include jmdict.mk
1311
1412
PAGES=blog.scm data.scm documentation.scm e404.scm feeds.scm index.scm mentions.scm
1513

3937
	@rm -f dictionaries/*
4038
	@$(MAKE) $(DOWNLOADS)
4139
42-
# Download JMdict dictionaries from ERDRG
43-
dictionaries/%.xml:
44-
	dl_filename="$(shell basename "$@" | rev | cut -c5- | rev)"; \
45-
	wget ftp://ftp.edrdg.org/pub/Nihongo/"$$dl_filename".gz -O "$$dl_filename.gz"; \
46-
	gunzip "$$dl_filename.gz"; \
47-
	sed -i -e 's|<|\&\<;|g' -e 's|>|\&\>;|g' "$$dl_filename"; \
48-
	sed -i -e 's|&\([^;]\+\);|\1|g' "$$dl_filename"; \
49-
	mv "$$dl_filename" "$@"
50-
51-
# Download frequency analysis run on Wikipedia in 2015
52-
# https://en.wiktionary.org/wiki/Wiktionary:Frequency_lists/Japanese2015_10000
53-
dictionaries/frequency.tsv:
54-
	wget --no-check-certificate \
55-
        https://namakajiri.net/data/wikipedia-20150422-lemmas.tsv -O $@.tmp
56-
	sed -i 's| ||g' $@.tmp
57-
	head -n20000 $@.tmp > $@
58-
	rm $@.tmp
59-
60-
dicos/JMdict_%.nani: dictionaries/JMdict.xml tools/jmdict.scm dictionaries/frequency.tsv $(DICO_MODULES)
61-
	guile -L modules tools/jmdict.scm build \
62-
        $< $(shell echo $@ | sed 's|^.*_\([^.]*\)\..*$$|\1|g') $@
63-
64-
dicos/JMdict_e.nani: dictionaries/JMdict_e.xml tools/jmdict.scm dictionaries/frequency.tsv $(DICO_MODULES)
65-
	guile -L modules tools/jmdict.scm build $< e $@
66-
6740
po/%/LC_MESSAGES/nani.mo: po/%.po
6841
	@mkdir -p $$(dirname $@)
6942
	msgfmt --output-file=$@ $<

jmdict.mk unknown status 1

1+
JMDICT_LANGS=dut fre ger hun rus slv spa swe
2+
DICOS+=dicos/JMdict_e.nani $(addprefix dicos/JMdict_, $(addsuffix .nani, $(JMDICT_LANGS)))
3+
DOWNLOADS+=dictionaries/JMdict_e.xml dictionaries/JMdict.xml dictionaries/frequency.tsv
4+
5+
# Download JMdict dictionaries from ERDRG
6+
dictionaries/%.xml:
7+
	dl_filename="$(shell basename "$@" | rev | cut -c5- | rev)"; \
8+
	wget ftp://ftp.edrdg.org/pub/Nihongo/"$$dl_filename".gz -O "$$dl_filename.gz"; \
9+
	gunzip "$$dl_filename.gz"; \
10+
	sed -i -e 's|&lt;|\&\&lt;;|g' -e 's|&gt;|\&\&gt;;|g' "$$dl_filename"; \
11+
	sed -i -e 's|&\([^;]\+\);|\1|g' "$$dl_filename"; \
12+
	mv "$$dl_filename" "$@"
13+
14+
# Download frequency analysis run on Wikipedia in 2015
15+
# https://en.wiktionary.org/wiki/Wiktionary:Frequency_lists/Japanese2015_10000
16+
dictionaries/frequency.tsv:
17+
	wget --no-check-certificate \
18+
        https://namakajiri.net/data/wikipedia-20150422-lemmas.tsv -O $@.tmp
19+
	sed -i 's| ||g' $@.tmp
20+
	head -n20000 $@.tmp > $@
21+
	rm $@.tmp
22+
23+
dicos/JMdict_%.nani: dictionaries/JMdict.xml tools/jmdict.scm dictionaries/frequency.tsv $(DICO_MODULES)
24+
	guile -L modules tools/jmdict.scm build \
25+
        $< $(shell echo $@ | sed 's|^.*_\([^.]*\)\..*$$|\1|g') $@
26+
27+
dicos/JMdict_e.nani: dictionaries/JMdict_e.xml tools/jmdict.scm dictionaries/frequency.tsv $(DICO_MODULES)
28+
	guile -L modules tools/jmdict.scm build $< e $@

manifest.scm unknown status 1

1+
(specifications->manifest
2+
  '("make" "guile" "haunt"
3+
4+
    ; for download
5+
    "unzip" "libiconv"
6+
7+
    ; for all
8+
    "gettext"))

radicals.mk

66
	wget ftp://ftp.monash.edu/pub/nihongo/kradzip.zip -O dictionaries/kradzip.zip
77
	unzip dictionaries/kradzip.zip radkfilex -d dictionaries
88
	iconv -f euc-jp -t utf-8 dictionaries/radkfilex > $@
9-
	rm radkfilex
9+
	rm dictionaries/radkfilex
1010
1111
dictionaries/kanjidic2.xml:
1212
	wget http://www.edrdg.org/kanjidic/kanjidic2.xml.gz -O $@.gz