sinclude $(GTLANG)/hyph.mk

hyph: hyph-$(GTLANG).fst
hyph-$(GTLANG).fst: $(GTLANG)/bin/hyph-$(GTLANG).fst
$(GTLANG)/bin/hyph-$(GTLANG).fst: \
		$(GTLANG)/bin/hyph-i$(GTLANG).save \
		$(GTLANG)/bin/$(GTLANG)-norm.fst \
		$(GTLANG)/bin/hyphrules-$(GTLANG).fst
	@echo
	@echo "*** Building hyph-$(GTLANG).fst ***" ;
	@echo
	@printf "read regex @\"$(GTLANG)/bin/hyph-i$(GTLANG).save\" .o. \n\
		@\"$(GTLANG)/bin/$(GTLANG)-norm.fst\" ; \n\
		set flag-is-epsilon ON \n"           > tmp/fst-script
	@printf "read regex @\"$(GTLANG)/bin/hyphrules-$(GTLANG).fst\" ;" >> tmp/fst-script
	@printf "compose net \n\
	save stack $@ \n\
	quit \n"                                     >> tmp/fst-script
	$(CFST) < tmp/fst-script
	@rm -f tmp/fst-script


# This goal is to make a regex for hyphenator rules
hyphrules: hyphrules.fst
hyphrules.fst: $(GTLANG)/bin/hyphrules-$(GTLANG).fst
$(GTLANG)/bin/hyphrules-$(GTLANG).fst: $(GTLANG)/src/hyph-$(GTLANG).txt
	@echo
	@echo "*** Building hyphrules-$(GTLANG).fst ***" ;
	@echo
	printf "source $< \n\
	save stack $@ \n\
	quit \n" > tmp/hyph-script
	$(XFST) < tmp/hyph-script
	rm -f tmp/hyph-script


# This goal is to remove the lexically inserted soft hyphen sign from the
# upper representation, by putting a further upper level without the ^ sign.

# This target should really be renamed to no__soft__hyphen (without underscores)
# To be done.
nohardhyphen: nohardhyphen.fst
nohardhyphen.fst: common/bin/nohardhyphen.fst
common/bin/nohardhyphen.fst: common/src/nohardhyphen.regex
	@echo
	@echo "*** Building nohardhyphen.fst ***" ;
	@echo
	@printf "read regex < $< \n\
	save stack $@ \n\
	quit \n" > tmp/nohardhyphen-script
	$(XFST) < tmp/nohardhyphen-script
	@rm -f tmp/nohardhyphen-script



# This is a normative transducer with lexical hyphenation points, used
# for lower morphosyntactic representation and upper (hyphenated) wordform.

save-ihyph: $(GTLANG)/bin/hyph-i$(GTLANG).save
$(GTLANG)/bin/hyph-i$(GTLANG).save: \
		$(GTLANG)/bin/twol-hyph-$(GTLANG).bin \
		$(NORMFILES)
	@echo
	@echo "*** Building hyph-i$(GTLANG).save ***" ;
	@echo
	printf "compile-source $(NORMFILES) \n\
	read-rules $< \n\
	compose-result \n\
	invert-result \n\
	save-result $@ \n\
	quit \n" > tmp/save-script
	$(LEXC) < tmp/save-script
	rm -f tmp/save-script


# This is a non-inverted version of the previous one. It is used to
# generate wordforms with hyphenation mark for input to speller generation, used
# for upper morphosyntactic representation and lower wordform.

save-hyph: $(GTLANG)/bin/hyph-$(GTLANG).save
$(GTLANG)/bin/hyph-$(GTLANG).save: \
		$(GTLANG)/bin/twol-hyph-$(GTLANG).bin \
		$(NORMFILES)
	@echo
	@echo "*** Building hyph-$(GTLANG).save ***" ;
	@echo
	printf "compile-source $(NORMFILES) \n\
	read-rules $< \n\
	compose-result \n\
	save-result $@ \n\
	quit \n" > tmp/save-script
	$(LEXC) < tmp/save-script
	rm -f tmp/save-script


hyphen-convert.fst: common/bin/hyphen-convert.fst
common/bin/hyphen-convert.fst: common/src/hyphen-convert.regex
	@echo
	@echo "*** Building hyphen-convert.fst ***" ;
	@echo
	@printf "read regex < $< \n\
	save stack $@ \n\
	quit \n" > tmp/hyphen-convert-script
	$(XFST) < tmp/hyphen-convert-script
	@rm -f tmp/hyphen-convert-script


hyphen-remove.fst: common/bin/hyphen-remove.fst
common/bin/hyphen-remove.fst: common/src/hyphen-remove.regex
	@echo
	@echo "*** Building hyphen-remove.fst ***" ;
	@echo
	@printf "read regex < $< \n\
	save stack $@ \n\
	quit \n" > tmp/hyphen-remove-script
	$(XFST) < tmp/hyphen-remove-script
	@rm -f tmp/hyphen-remove-script

#####################################
# Hyphenation testing starts here:
hyphtest: hyphregression hyphwordtypes

hyphregression: \
	$(HYPTESTREPDIR)/regression-$(TESTTOOL)-forrest-$(GTLANG)-$(DATE).xml

hyphwordtypes: \
	$(HYPTESTREPDIR)/wordtypes-$(TESTTOOL)-forrest-$(GTLANG)-$(DATE).xml

#####################################
# The final target is a Forrest doc containing the results and some calculated
# statistics:
$(HYPTESTREPDIR)/%-$(TESTTOOL)-forrest-$(GTLANG)-$(DATE).xml: \
	$(HYPTESTREPDIR)/%-$(TESTTOOL)-$(GTLANG)-$(DATE).xml \
	$(TESTXSLDIR)/hyphtest2document.xsl
	@echo
	@echo "*** Converting to ForrestDoc. Output in:              ***" ;
	@echo
	@echo "$@" ;
	@echo
	@xsltproc \
		--param testlang "'$(GTLANG)'" \
		--param testtype "'$*'" \
		$(TESTXSLDIR)/hyphtest2document.xsl $< | \
		xmllint --encode UTF-8 --output $@ --format -

#####################################
# Collect test results in a simple but sufficient xml format:
.PRECIOUS: $(HYPTESTREPDIR)/%-$(TESTTOOL)-$(GTLANG)-$(DATE).xml
$(HYPTESTREPDIR)/%-$(TESTTOOL)-$(GTLANG)-$(DATE).xml: \
	tmp/hy-%-$(TESTTOOL)-$(GTLANG).txt \
	tmp/hy-%-$(TESTTOOL)-$(GTLANG).txt.out
	@echo
	@echo "*** Collecting $* results, transforming to XML. Output:  ***" ;
	@echo
	@echo "$@" ;
	@echo
	script/hyphen-testres.pl \
		--$(TESTTOOL) \
		--input=$< \
		--output=$<.out \
		--document=$< \
		--date=$(DATE) \
		--version="`cat $<.out.version`" \
		--toolversion="`cat tmp/hyph-$(TESTTOOL)-version.txt`" \
		--xml=$@

#####################################
# Run the actual hyphenation test here:
.PRECIOUS: tmp/hy-%-pl-$(GTLANG).txt.out
tmp/hy-%-pl-$(GTLANG).txt.out: tmp/hy-%-pl-$(GTLANG).txt
	@echo
	@echo "*** Hyphenation testing - $(GTLANG)  ***" ;
	@echo
	cat $<.f1 | $(HYPHEN) -w -m $(GTLANG)/polderland/$(HYPHPATT) > $@
# Add lexicon signature/version here
	@echo "nuvviDspeller	Divvun" > $@.verstmp
	@touch userdict
	$(SPELL) -u8 -0 -m $(GTLANG)/polderland/$(HYPHDIC) < $@.verstmp \
		> $@.verstmp2
	grep 'version' $@.verstmp2 | cut -f2 > $@.version
	@rm -f userdict $@.verstmp $@.verstmp2
# Add hyphenator tool version here
	$(HYPHEN) --version 2> tmp/hyph-$(TESTTOOL)-version.txt

#####################################
# Prepare the hyphenation data -
# Regression data:
tmp/hy-wordtypes-$(TESTTOOL)-$(GTLANG).txt: \
		$(GTLANG)/testing/hyphenation.txt \
		$(GTLANG)/polderland/$(HYPHPATT)
#		prepare-hyphentest
	@echo
	@echo "*** Preparing hyphenation test data - $(GTLANG)  ***" ;
	@echo
	grep -v "^#" $< | grep -v "^\s*$$" | grep -v ' .*	' > $@
	cut -f1 $@ | tr -d '^' > $@.f1

#####################################
# Prepare the hyphenation data -
# Wordtypes data (check hyphenation of a number of different word constructs):
tmp/hy-regression-$(TESTTOOL)-$(GTLANG).txt: \
		$(GTLANG)/polderland/hyph-regressions.txt \
		$(GTLANG)/polderland/$(HYPHPATT) \
		prepare-hyphentest
	@echo
	@echo "*** Preparing hyphenation test data - $(GTLANG)  ***" ;
	@echo
	grep -v "^#" $< | grep -v "^\s*$$" | grep -v ' .*	' > $@
	cut -f1 $@ | tr -d '^' > $@.f1

#####################################
# Make sure the latest lexicon hyphenation file is used:
prepare-hyphentest:
	@echo
	@echo "*** Copying hyphenation lexicon file - $(GTLANG)  ***" ;
	@echo
	cp $(GTLANG)/polderland/$(WINSPELL) $(GTLANG)/polderland/$(HYPHDIC)

#####################################
# Make sure the latest pattern hyphenation file is used:
$(GTLANG)/polderland/$(HYPHPATT): $(MACPROOFDIST)/$(HYPHPATT)
	@echo
	@echo "*** Copying hyphenation pattern lexicon - $(GTLANG)  ***" ;
	@echo
	cp $(MACPROOFDIST)/$(HYPHPATT) $(GTLANG)/polderland/$(HYPHPATT)