mk.wordlists

# We make a Latin wordlist by iterating over three corpora: PHI,
# Perseus and DigiLibLT.  We make a Greek wordlist by iterating over
# the Perseus corpus and adding that to the TLG wordlist.

include mk.common
all: $(BUILD)/lat.words $(BUILD)/grc.words

# Download Perseus corpora and make wordlists

GREEKLIT = $(BUILD)/canonical-greekLit
LATINLIT = $(BUILD)/canonical-latinLit

$(GREEKLIT)/.git/HEAD:
	rm -rf $(GREEKLIT)
	cd $(BUILD) && git clone https://github.com/PerseusDL/canonical-greekLit
	cd $(GREEKLIT) && git checkout

$(LATINLIT)/.git/HEAD:
	rm -rf $(LATINLIT)
	cd $(BUILD) && git clone https://github.com/PerseusDL/canonical-latinLit
	cd $(LATINLIT) && git checkout

$(BUILD)/lat.words-perseus: utils/make_latin_wordlist_perseus.sh $(LATINLIT)/.git/HEAD
	mkdir -p $(BUILD)
	./utils/make_latin_wordlist_perseus.sh $(LATINLIT) > $@

$(BUILD)/grc.words-perseus: utils/make_greek_wordlist_perseus.sh $(GREEKLIT)/.git/HEAD
	mkdir -p $(BUILD)
	./utils/make_greek_wordlist_perseus.sh $(GREEKLIT) > $@

# DigilibLT has to be downloaded separately after registering for an
# account.

#DIGILIBDIR = $(HOME)/digiliblt
DIGILIBDIR = ../../diogenes-web/app/static/texts/DigiLibLT/

$(BUILD)/lat.words-digilib: utils/make_latin_wordlist_digilib.sh
	mkdir -p $(BUILD)
	./utils/make_latin_wordlist_digilib.sh $(DIGILIBDIR) > $@

# PHI and TLG wordlists

PHIDIR = $(HOME)/Classics-Data/phi
TLGDIR = $(HOME)/Classics-Data/tlg_e

$(BUILD)/lat.words-phi: utils/make_latin_wordlist.pl
	mkdir -p $(BUILD) $(DATA)
	find $(PHIDIR) -type f -exec sha256sum '{}' ';' | sed 's:  $(PHIDIR):  :g' > $(DATA)/phisums
	./utils/make_latin_wordlist.pl $(PHIDIR) > $@

$(BUILD)/grc.words-tlg: utils/make_greek_wordlist.pl
	mkdir -p $(BUILD) $(DATA)
	find $(TLGDIR) -type f -exec sha256sum '{}' ';' | sed 's:  $(TLGDIR):  :g' > $(DATA)/tlgsums
	./utils/make_greek_wordlist.pl $(TLGDIR) > $@

$(BUILD)/lat.words: $(BUILD)/lat.words-perseus $(BUILD)/lat.words-digilib $(BUILD)/lat.words-phi
	export LC_ALL=C; cat $(BUILD)/lat.words-perseus $(BUILD)/lat.words-digilib $(BUILD)/lat.words-phi | sort | uniq > $@

$(BUILD)/grc.words: $(BUILD)/grc.words-perseus $(BUILD)/grc.words-tlg
	export LC_ALL=C; cat $(BUILD)/grc.words-perseus $(BUILD)/grc.words-tlg | sort | uniq > $@


clean:
	rm -f $(BUILD)/lat.words-perseus
	rm -f $(BUILD)/lat.words-digilib
	rm -f $(BUILD)/lat.words-phi
	rm -f $(BUILD)/lat.words
	rm -f $(BUILD)/grc.words-perseus
	rm -f $(BUILD)/grc.words-tlg
	rm -f $(BUILD)/grc.words