## Include this file in top-level srcdir to compile FSTs
## Copyright: Sámediggi/Divvun/UiT
## Licence: GPL v3+

####### Automake targets: ########

APERTIUM_ANALYSER_HFST=
APERTIUM_ANALYSERS_GT_HFST=
APERTIUM_ANALYSERS_AP_HFST=
APERTIUM_ANALYSERS_ATTGZ=

APERTIUM_GENERATORS_HFST=
APERTIUM_GENERATOR_ATTGZ=
APERTIUM_CG_BIN=

APERTIUM_GENERATED_TAGSET_RELABEL_SRCS=

if WANT_APERTIUM
# The multiple parallel pattern rules further down are fragile (or: make is
# fragile). We need to specify here as the first target to build the dependency
# of the first pattern target (ie the non-last pattern dependency). Without
# this make will give up. DO NOT REMOVE analyser-mt-gt-desc.hfst BELOW!!!
APERTIUM_ANALYSER_HFST+=analyser-mt-gt-desc.hfst analyser-mt-gt-desc.hfstol

# Build list of target language transducers from the list of target languages:
# => analyser-mt-gt-desc.nob.hfst etc.
APERTIUM_ANALYSERS_GT_HFST+=\
	$(addsuffix .hfstol,$(addprefix analyser-mt-gt-desc.,$(APERTIUM_TARGET_LANGS)))

# Build list of apertium tagged files by replacing -gt- with -apertium-:
# analyser-mt-gt-desc.nob.hfst => analyser-mt-apertium-desc.nob.hfst etc.
APERTIUM_ANALYSERS_AP_HFST+=\
	$(subst -gt-,-apertium-,$(APERTIUM_ANALYSERS_GT_HFST))

# Build list of att.gz files by adding the suffix .att.gz to the basenames
# of the filenames:
# analyser-mt-apertium-desc.nob.hfst => analyser-mt-apertium-desc.nob.att.gz
APERTIUM_ANALYSERS_ATTGZ+=\
	$(addsuffix .att.gz,$(basename $(APERTIUM_ANALYSERS_AP_HFST)))

# Build a list of Apertium style target filenames:
#
# Analysers:
# apertium-sme-nob.sme-nob.LR.att.gz
# apertium-smn-sme.sme-smn.LR.att.gz
#
# Generators:
# apertium-sme-nob.sme-nob.RL.att.gz
#
# Apertium filename pattern:
#
# dirname-with-lang-pair.lang-pair.LR
# - dirname-with-lang-pair = fixed string, same as containing AP dir name
# - lang-pair = the pair in the direction being used, as in: from-to
# - LR = analyser
# - RL = generator

# Loop over target languages and pair names and build the final target names
# except the suffix part:
APERTIUM_PAIRS=$(shell for ll in $(APERTIUM_TARGET_LANGS); do\
    for ld in $(APERTIUM_TARGET_LANG_NAME); do\
        if test $$(echo $$ld | grep $$ll); then\
            echo "$$ld.$(GTLANG)-$$ll" ;\
        fi\
    done ;\
done)
# Add the suffixes:
# "apertium-sme-nob.sme-nob" -> "apertium-sme-nob.sme-nob.LR.att.gz":
APERTIUM_MORPH_PAIRS=$(addsuffix .LR.att.gz,$(APERTIUM_PAIRS))

# Apertium style target filenames with stem variable % instead of target
# language, used in a pattern rule further down:
APERTIUM_PAIRS_VAR=$(shell for ll in $(APERTIUM_TARGET_LANGS); do\
    for ld in $(APERTIUM_TARGET_LANG_NAME); do\
        if test $$(echo $$ld | grep $$ll); then\
            echo "$$ld.$(GTLANG)-%" ;\
        fi\
    done ;\
done)
# "apertium-sme-%.sme-nob" -> "apertium-sme-%.sme-nob.LR.att.gz":
APERTIUM_MORPH_PAIRS_V=$(addsuffix .LR.att.gz,$(APERTIUM_PAIRS_VAR))

APERTIUM_GENERATORS_HFST+=generator-mt-gt-norm.hfstol \
						  generator-mt-apertium-norm.hfstol

APERTIUM_GENERATOR_ATTGZ+=apertium-$(GTLANG).$(GTLANG).RL.att.gz

APERTIUM_GENERATED_TAGSET_RELABEL_SRCS+=tagsets/apertium.relabel

if WANT_SYNTAX
APERTIUM_CG_BIN+=$(GTLANG).mor.rlx.bin \
				 $(GTLANG).syn.rlx.bin
#				 $(GTLANG).dep.rlx.bin
endif # WANT_SYNTAX

# Apertium - GTDivvun filename correspondences:
# apertium-sme-nob.sme-nob.LR.att.gz = analyser-mt-apertium-desc.nob.att.gz
# apertium-sme.sme.RL.att.gz = generator-mt-apertium-norm.att.gz
# sme.mor.rlx.bin = disambiguation-mt-apertium.bin
# sme.syn.rlx.bin = functions-mt-apertium.bin
# sme.dep.rlx.bin = dependency-mt-apertium.bin

endif # WANT_APERTIUM

if CAN_HFST
hfstdatadir=$(datadir)/giella/$(GTLANG)
noinst_DATA=$(APERTIUM_ANALYSER_HFST) \
			$(APERTIUM_ANALYSERS_GT_HFST) \
			$(APERTIUM_ANALYSERS_AP_HFST) \
			$(APERTIUM_GENERATORS_HFST) \
			$(APERTIUM_CG_SRC) \
			$(APERTIUM_GENERATED_TAGSET_RELABEL_SRCS)
hfstdata_DATA=$(APERTIUM_MORPH_PAIRS) \
			  $(APERTIUM_GENERATOR_ATTGZ) \
			  $(APERTIUM_CG_BIN)
endif # CAN_HFST

####### Build rules: ########

# Variable to hold filename for file containing all tags used
# (all symbols starting or ending with '+')
TAG_FILE=tagsets/gttags.txt

# Variable to hold filename for fst used as source for extracting tags,
# sans extension
TAG_EXTRACTION_FST=analyser-mt-gt-desc

# Convert GT-tags to Apertium-tags:
tagsets/apertiumtags.txt: tagsets/gttags.txt
	$(AM_V_GEN)tr -d '+' < $< \
		| tr '/[:upper:]' '_[:lower:]' \
		| awk '{print "%<"$$0"%>"}' > $@

# Generate relabel file for replacing Giella tags with Apertium tags:
tagsets/apertium.relabel: tagsets/gttags.txt \
						  tagsets/apertiumtags.txt
	$(AM_V_GEN)paste $^ | tr -d '%' > $@



# Apertium generator - GTD tags. All filtering happens here except dialect tags happens here:
generator-mt-gt-norm.tmp.hfst: $(top_builddir)/src/generator-raw-gt-desc.hfst \
			$(top_builddir)/src/filters/make-optional-transitivity-tags.hfst  \
			$(top_builddir)/src/filters/make-optional-semantic-tags.hfst      \
			$(top_builddir)/src/filters/make-optional-v1-tags.hfst            \
			$(top_builddir)/src/filters/remove-number-string-tags.hfst        \
			$(top_builddir)/src/filters/remove-NG-strings.hfst                \
			$(top_builddir)/src/filters/remove-usage-tags.hfst                \
			$(top_builddir)/src/filters/remove-error-strings.hfst             \
			$(top_builddir)/src/filters/remove-orthography-tags.hfst          \
			$(top_builddir)/src/filters/remove-Orth_IPA-strings.hfst          \
			$(top_builddir)/src/filters/remove-orig_lang-tags.hfst            \
			$(top_builddir)/src/filters/remove-hyphenation-marks.hfst         \
			$(top_builddir)/src/filters/remove-infl_deriv-borders.hfst        \
			$(top_builddir)/src/filters/remove-word-boundary.hfst             \
			$(top_builddir)/src/orthography/downcase-derived_proper-strings.compose.hfst
	$(AM_V_HXFST)$(PRINTF) "set xerox-composition ON \n\
			read regex \
		@\"$(top_builddir)/src/filters/make-optional-semantic-tags.hfst\"     \
	.o. @\"$(top_builddir)/src/filters/make-optional-transitivity-tags.hfst\" \
	.o. @\"$(top_builddir)/src/filters/make-optional-v1-tags.hfst\"           \
	.o. @\"$(top_builddir)/src/filters/remove-number-string-tags.hfst\"       \
	.o. @\"$(top_builddir)/src/filters/remove-usage-tags.hfst\"               \
	.o. @\"$(top_builddir)/src/filters/remove-error-strings.hfst\"            \
	.o. @\"$(top_builddir)/src/filters/remove-NG-strings.hfst\"               \
	.o. @\"$(top_builddir)/src/filters/remove-orig_lang-tags.hfst\"           \
	.o. @\"$(top_builddir)/src/filters/remove-orthography-tags.hfst\"         \
	.o. @\"$(top_builddir)/src/filters/remove-Orth_IPA-strings.hfst\"         \
	.o. @\"$<\" \
	.o. @\"$(top_builddir)/src/orthography/downcase-derived_proper-strings.compose.hfst\" \
	.o. @\"$(top_builddir)/src/filters/remove-word-boundary.hfst\"            \
	.o. @\"$(top_builddir)/src/filters/remove-hyphenation-marks.hfst\"        \
	.o. @\"$(top_builddir)/src/filters/remove-infl_deriv-borders.hfst\"       \
   	;\n\
        save stack $@\n\
        quit\n" | $(HFST_XFST) -p $(MORE_VERBOSITY) $(HFST_FORMAT)

# Apertium analyser - GTD tags. Target language independent filtering
# happens here:
analyser-mt-gt-desc.tmp.hfst: $(top_builddir)/src/generator-raw-gt-desc.hfst \
			$(top_builddir)/src/filters/remove-dialect-tags.hfst             \
			$(top_builddir)/src/filters/remove-variant-tags.hfst             \
			$(top_builddir)/src/filters/remove-number-string-tags.hfst       \
			$(top_builddir)/src/filters/remove-usage-tags.hfst               \
			$(top_builddir)/src/filters/remove-hyphenation-marks.hfst        \
			$(top_builddir)/src/filters/remove-infl_deriv-borders.hfst       \
			$(top_builddir)/src/filters/remove-orthography-tags.hfst         \
			$(top_builddir)/src/filters/remove-orig_lang-tags.hfst           \
			$(top_builddir)/src/filters/remove-Orth_IPA-strings.hfst         \
			$(top_builddir)/src/filters/remove-word-boundary.hfst            \
			$(top_builddir)/src/orthography/downcase-derived_proper-strings.compose.hfst
	$(AM_V_HXFST)$(PRINTF) "set xerox-composition ON \n\
			read regex \
	    @\"$(top_builddir)/src/filters/remove-dialect-tags.hfst\"             \
	.o. @\"$(top_builddir)/src/filters/remove-variant-tags.hfst\"             \
	.o. @\"$(top_builddir)/src/filters/remove-number-string-tags.hfst\"       \
	.o. @\"$(top_builddir)/src/filters/remove-usage-tags.hfst\"               \
	.o. @\"$(top_builddir)/src/filters/remove-orig_lang-tags.hfst\"           \
	.o. @\"$(top_builddir)/src/filters/remove-orthography-tags.hfst\"         \
	.o. @\"$(top_builddir)/src/filters/remove-Orth_IPA-strings.hfst\"         \
	.o. @\"$<\" \
	.o. @\"$(top_builddir)/src/orthography/downcase-derived_proper-strings.compose.hfst\" \
	.o. @\"$(top_builddir)/src/filters/remove-word-boundary.hfst\"            \
	.o. @\"$(top_builddir)/src/filters/remove-hyphenation-marks.hfst\"        \
	.o. @\"$(top_builddir)/src/filters/remove-infl_deriv-borders.hfst\"       \
   	;\n\
        save stack $@\n\
        quit\n" | $(HFST_XFST) -p $(MORE_VERBOSITY) $(HFST_FORMAT)

# Target language specific filtering - build depends on the availability
# of a filter file.
# WITH filter file available (the source file is inverted with the .i modifier):
analyser-mt-gt-desc.%.tmp.hfst: analyser-mt-gt-desc.hfst \
			$(top_builddir)/src/filters/remove-derivation-strings.%.hfst
	$(AM_V_RGX2FST)$(PRINTF) "\
		@\"$(top_builddir)/src/filters/remove-derivation-strings.$*.hfst\" \
   	.o. @\"$<\".i \
    	;" \
		| $(HFST_REGEXP2FST) $(HFST_FLAGS) $(HFST_FORMAT) \
			-S --xerox-composition=ON \
		| $(HFST_FST2FST) $(HFSTFLAGS) -t \
		-o $@

# With NO filter file available, invert the file to allow the general pattern
# *.tmp.hfst -> invert -> *.hfst
# to apply correctly:
analyser-mt-gt-desc.%.tmp.hfst: analyser-mt-gt-desc.hfst
	$(AM_V_INVERT)$(HFST_INVERT) -i $< \
		| $(HFST_FST2FST) $(HFSTFLAGS) -t \
		-o $@

# Relabel GTDivvun tags to Apertium tags - generator, step 1:
generator-mt-apertium-norm.tmp1.hfst: \
			generator-mt-gt-norm.hfst \
			tagsets/apertium.relabel
	$(AM_V_HSUBST)$(HFST_SUBSTITUTE) $(HFSTFLAGS) -i $< \
		-F tagsets/apertium.relabel \
		| $(HFST_FST2FST) $(HFSTFLAGS) -t \
		-o $@

# Relabel GTDivvun tags to Apertium tags - generator, step 2a:
# * relabel file
#   AND
# * general tag change regex file
generator-mt-apertium-norm.%.hfst: \
			generator-mt-apertium-norm.%1.hfst \
			tagsets/modify-tags.hfst \
			$(srcdir)/tagsets/apertium.postproc.relabel
	$(AM_V_RGX2FST)$(PRINTF) "\
		@\"tagsets/modify-tags.hfst\".i \
   	.o. @\"$<\" \
    	;" \
		| $(HFST_REGEXP2FST) $(HFST_FLAGS) -S --xerox-composition=ON \
		| $(HFST_SUBSTITUTE) $(HFSTFLAGS) \
		-F $(srcdir)/tagsets/apertium.postproc.relabel \
		-o $@

# Relabel GTDivvun tags to Apertium tags - generator, step 2b:
# * relabel file ONLY
generator-mt-apertium-norm.%.hfst: \
			generator-mt-apertium-norm.%1.hfst \
			$(srcdir)/tagsets/apertium.postproc.relabel
	$(AM_V_HSUBST)$(HFST_SUBSTITUTE) $(HFSTFLAGS) -i $< \
		-F $(srcdir)/tagsets/apertium.postproc.relabel \
		-o $@

# Relabel GTDivvun tags to Apertium tags - generator, step 2c:
# * general tag change regex file ONLY
generator-mt-apertium-norm.%.hfst: \
			generator-mt-apertium-norm.%1.hfst \
			tagsets/modify-tags.hfst
	$(AM_V_RGX2FST)$(PRINTF) "\
		@\"tagsets/modify-tags.hfst\".i \
   	.o. @\"$<\" \
    	;" \
		| $(HFST_REGEXP2FST) $(HFST_FLAGS) -S --xerox-composition=ON \
		-o $@

# Relabel GTD tags to Apertium tags - analyser:
# % = TARGETLANG
# Step 1a - gt -> apertium.tmp1
analyser-mt-apertium-desc.%.tmp1.hfst: \
			analyser-mt-gt-desc.%.hfst \
			tagsets/apertium.relabel
	$(AM_V_HSUBST)$(HFST_SUBSTITUTE) $(HFSTFLAGS) -i $< \
		-F tagsets/apertium.relabel \
		-o $@

# Relabel GTD tags to Apertium tags - analyser:
# % = TARGETLANG
# Step 1b.a - apertium.tmp1 -> apertium.tmp
# * relabel file
#   AND
# * general tag change regex file
analyser-mt-apertium-desc.%.tmp.hfst: \
			analyser-mt-apertium-desc.%.tmp1.hfst \
			tagsets/modify-tags.hfst \
			$(srcdir)/tagsets/apertium.postproc.relabel
	$(AM_V_RGX2FST)$(PRINTF) "\
	   		@\"$<\" \
		.o. @\"tagsets/modify-tags.hfst\" \
    	;" \
		| $(HFST_REGEXP2FST) $(HFST_FLAGS) -S --xerox-composition=ON \
		| $(HFST_SUBSTITUTE) $(HFSTFLAGS)  \
		-F $(srcdir)/tagsets/apertium.postproc.relabel \
		-o $@

# Step 1b.b - apertium.tmp1 -> apertium.tmp
# * relabel file ONLY
analyser-mt-apertium-desc.%.tmp.hfst: \
			analyser-mt-apertium-desc.%.tmp1.hfst \
			$(srcdir)/tagsets/apertium.postproc.relabel
	$(AM_V_HSUBST)$(HFST_SUBSTITUTE) $(HFSTFLAGS) -i $< \
		-F $(srcdir)/tagsets/apertium.postproc.relabel \
		-o $@

# Step 1b.c - apertium.tmp1 -> apertium.tmp
# * general tag change regex file ONLY
analyser-mt-apertium-desc.%.tmp.hfst: \
			analyser-mt-apertium-desc.%.tmp1.hfst \
			tagsets/modify-tags.hfst
	$(AM_V_RGX2FST)$(PRINTF) "\
   		@\"$<\" \
	.o. @\"tagsets/modify-tags.hfst\" \
    	;" \
		| $(HFST_REGEXP2FST) $(HFST_FLAGS) -S --xerox-composition=ON \
		-o $@

# Relabel GTD tags to Apertium tags - analyser:
# % = TARGETLANG
# Step 2a - apertium.tmp -> apertium
#         - BOTH target language specific modifying AND relabeling
analyser-mt-apertium-desc.%.hfst: analyser-mt-apertium-desc.%.tmp.hfst \
								  tagsets/modify-tags.%.hfst \
								  $(srcdir)/tagsets/apertium.%.relabel
	$(AM_V_RGX2FST)$(PRINTF) "\
   		@\"$<\" \
	.o. @\"tagsets/modify-tags.$*.hfst\" \
    	;" \
		| $(HFST_REGEXP2FST) $(HFST_FLAGS) -S --xerox-composition=ON \
		| $(HFST_SUBSTITUTE) $(HFSTFLAGS) \
		-F $(srcdir)/tagsets/apertium.$*.relabel \
		| $(HFST_FST2FST) $(HFSTFLAGS) -t \
		-o $@

# Relabel GTD tags to Apertium tags - analyser:
# % = TARGETLANG
# Step 2b - apertium.tmp -> apertium
#         - ONLY target language specific relabeling
analyser-mt-apertium-desc.%.hfst: analyser-mt-apertium-desc.%.tmp.hfst \
								  $(srcdir)/tagsets/apertium.%.relabel
	$(AM_V_HSUBST)$(HFST_SUBSTITUTE) $(HFSTFLAGS) -i $< \
		-F $(srcdir)/tagsets/apertium.$*.relabel \
		| $(HFST_FST2FST) $(HFSTFLAGS) -t \
		-o $@

# Relabel GTD tags to Apertium tags - analyser:
# % = TARGETLANG
# Step 2c - apertium.tmp -> apertium
#         - ONLY target language specific modifying
analyser-mt-apertium-desc.%.hfst: analyser-mt-apertium-desc.%.tmp.hfst \
								  tagsets/modify-tags.%.hfst
	$(AM_V_RGX2FST)$(PRINTF) "\
   		@\"$<\" \
	.o. @\"tagsets/modify-tags.$*.hfst\" \
    	;" \
		| $(HFST_REGEXP2FST) $(HFST_FLAGS) -S --xerox-composition=ON \
		| $(HFST_FST2FST) $(HFSTFLAGS) -t \
		-o $@

# Relabel GTD tags to Apertium tags - analyser:
# % = TARGETLANG
# Step 2d - apertium.tmp -> apertium
#         - NO target language specific processing - just copy
analyser-mt-apertium-desc.%.hfst: analyser-mt-apertium-desc.%.tmp.hfst
	$(AM_V_CP)cp -f $< $@

%.att: %.hfst
	$(AM_V_FST2TXT)$(HFST_FST2FST) $(HFSTFLAGS) -w -i $< \
		| $(HFST_FST2TXT) -o $@

%.att.gz: %.att
	$(AM_V_GZIP)$(GZIP) -9 -c $< > $@

##### Disambiguation and syntactic parsing targets
# Copy files from the src dirs:
dependency-mt-gt.cg3: \
				$(GIELLA_SHARED)/smi/src/syntax/dependency.cg3
	$(AM_V_CP)cp -f $< $@

functions-mt-gt.cg3: \
				$(GIELLA_SHARED)/smi/src/syntax/functions.cg3
	$(AM_V_CP)cp -f $< $@

%-mt-gt.cg3: $(top_srcdir)/src/syntax/%.cg3
	$(AM_V_GEN)grep -v '#RemoveFromApertium' < $< > $@

%.cg3: $(top_srcdir)/src/syntax/%.cg3
	$(AM_V_GEN)grep -v '#RemoveFromApertium' < $< > $@

# Make sure included files are copied in before starting to compile:
$(APERTIUM_CG_SRC): $(APERTIUM_CG_INCLUDE_SRC)

# Change the tags from GTDivvun format to Apertium format:
%-apertium.bin: %-gt.bin $(srcdir)/tagsets/gt2apertium.cg3relabel
	$(AM_V_GEN)$(CG_RELABEL) $< $(srcdir)/tagsets/gt2apertium.cg3relabel $@

#### Copy to Apertium filenames
$(APERTIUM_MORPH_PAIRS_V): \
				analyser-mt-apertium-desc.%.att.gz
	$(AM_V_CP)cp -f $< $@

apertium-$(GTLANG).$(GTLANG).RL.att.gz: generator-mt-apertium-norm.att.gz
	$(AM_V_CP)cp -f $< $@

$(GTLANG).mor.rlx.bin: disambiguation-mt-apertium.bin
	$(AM_V_CP)cp -f $< $@

$(GTLANG).syn.rlx.bin: functions-mt-apertium.bin
	$(AM_V_CP)cp -f $< $@

#### HFST transducer fallback target:
#### - inversion is needed FOR THE ANALYSER because the hfst model is upside
#### down compared to Xerox AT APPLICATION TIME. The other command makes the
#### final transducer ready for use.
analyser%.hfst: analyser%.tmp.hfst
	$(AM_V_INVERT)$(HFST_INVERT) $< -o $@

# If you override this goal, you have to remove dialect tags in the overriding goal!
generator-mt-gt%.hfst: generator-mt-gt%.tmp.hfst \
			$(top_builddir)/src/filters/remove-dialect-tags.hfst
	$(AM_V_HXFST)$(PRINTF) "set xerox-composition ON \n\
			read regex \
	    @\"$(top_builddir)/src/filters/remove-dialect-tags.hfst\"             \
	.o. @\"$<\" \
	;\n\
        save stack $@\n\
        quit\n" | $(HFST_XFST) -p $(MORE_VERBOSITY)

# General fallback target for generators:
generator%.hfst: generator%.tmp.hfst
	$(AM_V_CP)cp -f $< $@

#### Include files:
include $(top_srcdir)/am-shared/lookup-include.am
include $(top_srcdir)/am-shared/vislcg3-include.am
include $(top_srcdir)/am-shared/hfst-format-include.am
include $(top_srcdir)/am-shared/tag-extraction-include.am
include $(top_srcdir)/am-shared/silent_build-include.am