Merge commit 'ae88b743f2ddff082db675c2eb8740f55284db34' as 'src/fst/m…

…orphology/ext-Apertium-nno'

src/fst/morphology/ext-Apertium-nno/.gitattributes

@@ -0,0 +1,7 @@
+* text eol=lf
+*.prob binary
+*.dix linguist-language=XML linguist-detectable=true
+*.lrx linguist-language=XML linguist-detectable=true
+*.lsx linguist-language=XML linguist-detectable=true
+*.tsx linguist-language=XML linguist-detectable=true
+*.t[[:digit:]]x linguist-language=XML linguist-detectable=true

src/fst/morphology/ext-Apertium-nno/.gitignore

@@ -0,0 +1,38 @@
+/INSTALL
+/Makefile
+/Makefile.in
+/aclocal.m4
+/ap_include.am
+/apertium-nno.pc
+/autom4te.cache
+/config.log
+/config.status
+/configure
+/configure~
+/install-sh
+/missing
+/modes
+/nno.autogen.att.gz
+/nno.autogen.bin
+/nno.autopgen.bin
+/nno.automorf-no-cp.bin
+/nno.automorf.att.gz
+/nno.automorf.bin
+/nno.rlx.bin
+/nno.syn.rlx.bin
+/nno_e.autogen.bin
+/nno_a_vi.autogen.bin
+/nno_e_me.autogen.bin
+/nno_e_vi.autogen.bin
+/nno.prefs.rlx.bin
+/nno.e_vi_prefs.rlx.bin
+/*.hfst
+/*.zhfst
+/.DS_Store
+/*.mode
+/*-*.prob
+/.deps
+/config.in
+/config
+/test/*-output.txt
+/nno.seg.rlx.bin

src/fst/morphology/ext-Apertium-nno/AUTHORS

@@ -0,0 +1,10 @@
+2009-2021, Kevin Brubeck Unhammer <[email protected]>
+2008-2021, Trond Trosterud <[email protected]>
+2019-2021, Anja Zawadzka Persvold <[email protected]>
+2020-2021, Hallvard Østrem <[email protected]>
+2021, Flammie Pirinen <[email protected]>
+2021, Hallvard Østrem <[email protected]>
+2021, Victoria Troland <[email protected]>
+2019, Jo Christian Oterhals <[email protected]>
+2008-2017, Francis M. Tyers <[email protected]>
+2016, Kartik Mistry <[email protected]>

src/fst/morphology/ext-Apertium-nno/ChangeLog

@@ -0,0 +1,181 @@
+2014-12-28  unhammer
+
+	* [r58177] apertium-nno.nno.dix, apertium-nno.nno.rlx: «over 3. Men
+	  blah» correctly splits sentence now
+
+2014-12-10  unhammer
+
+	* [r57970] apertium-nno.nno.dix: omardit→sørge_for + andre ord
+	  funne i magiske gtweb.uit.no/korp
+
+2014-10-25  unhammer
+
+	* [r57679] apertium-nno.nno.dix: testvoc
+
+	  (I think? just getting this out of the way before the major
+	  slv-ita commit)
+
+2014-10-06  unhammer
+
+	* [r57472] apertium-nno.nno.dix, configure.ac: testvoc
+
+2014-07-24  unhammer
+
+	* [r55966] AUTHORS, apertium-nno.nno.dix: clear up some of the
+	  license mess; nno/nob/sme stuff is GPL2-or-later (as are any of
+	  my contributions to Apertium, if that's ever a question)
+
+2014-07-23  unhammer
+
+	* [r55909] apertium-nno.nno.dix: meir unio-feilretting
+
+2014-07-15  unhammer
+
+	* [r55575] Makefile.am, configure.ac: use $(CGCOMP) as dep for rlx
+	  goals so theyre recompiled on new versions
+
+2014-06-28  unhammer
+
+	* [r55070] apertium-nno.nno.dix: doh
+
+2014-06-24  unhammer
+
+	* [r54910] apertium-nno.nno.dix: doh
+	* [r54909] apertium-nno.nno.dix: tatarisk osb.
+
+2014-06-13  unhammer
+
+	* [r54196] modes.xml: s/dis/disam;s/syn/syntax;s/mor/morph
+
+2014-06-09  unhammer
+
+	* [r54005] Makefile.am, autogen.sh: Put .pc and compiled data in
+	  /share/, not /lib/ (following Debian/autotools standards)
+
+	  pkgconfigdir=$(libdir)/pkgconfig →
+	  pkgconfigdir=$(prefix)/share/pkgconfig AP_LIB → AP_SRC
+	  apertium_XXXdir=$(prefix)/lib/apertium/$(BASENAME)/ →
+	  apertium_XXXdir=$(prefix)/share/apertium/$(BASENAME)/
+
+	  put share/pkgconfig in PKG_CONFIG_PATH in autogen.sh
+
+2014-05-20  unhammer
+
+	* [r53217] Makefile.am, apertium-nno.nno.syn.rlx, modes.xml:
+	  syn.rlx'es (unused) moved to monolingual dirs
+
+2014-05-12  unhammer
+
+	* [r52850] modes.xml: -e
+
+2014-04-16  unhammer
+
+	* [r51984] apertium-nno.nno.dix: testvoc
+
+2014-04-14  unhammer
+
+	* [r51954] apertium-nno.nno.dix: moar testvoc
+
+2014-04-13  unhammer
+
+	* [r51936] Makefile.am, apertium-nno.nno.dix, apertium-nno.nno.rlx:
+	  nno->nob testvoc (and turn on compounding)
+
+2014-04-12  unhammer
+
+	* [r51922] apertium-nno.nno.dix: some nno-nob testvoc and cleanup
+	  of dev dir
+
+2014-04-05  unhammer
+
+	* [r51693] apertium-nno.nno.dix: more #-testvoc, cmp-split RL's
+	  added to no-cmp pardefs
+	* [r51691] apertium-nno.nno.dix: some #-testvoc
+
+2014-04-04  unhammer
+
+	* [r51671] configure.ac: require 3.3.0
+
+2014-04-02  unhammer
+
+	* [r51635] apertium-nno.nno.dix: dan-nor/{nob,nno}.dix merged in;
+	  adjectives pardefs now have un and sp tags; adj testvoc in
+	  sme-nob and nno-nob TODO
+
+2014-03-26  unhammer
+
+	* [r51260] autogen.sh: fix autogen.sh so users don't have to set
+	  PKG_CONFIG_PATH and similar *PATH's
+
+2014-03-24  unhammer
+
+	* [r51164] Makefile.am: install .prob
+
+2014-03-22  unhammer
+
+	* [r51072] Makefile.am: -alt.xsl
+
+2014-03-20  unhammer
+
+	* [r50961] Makefile.am, apertium-nno.nno.dix: adj sint
+
+2014-03-16  unhammer
+
+	* [r50852] apertium-nno.nno.dix: wip: merging in sme nob.dix
+	* [r50843] apertium-nno.nno.dix: uniq
+	* [r50840] apertium-nno.nno.dix: re-add stuff that was added to
+	  nn-nb after this file was split out
+
+2014-03-15  unhammer
+
+	* [r50776] Makefile.am, rem-compounds.xsl: nno-nob depends on nno
+	  and nob, using lt-trim
+
+2014-03-14  unhammer
+
+	* [r50748] Makefile.am, alt.xsl: rm alt.xsl, use lt-comp --alt
+
+2014-02-14  unhammer
+
+	* [r50245] Makefile.am, README, apertium-nno.nno.dix,
+	  rem-compounds.xsl: lt-trim + modes makefile fixes to nor-eng, use
+	  alt.xsl in nno, no rem-compounds
+
+	  also, use compiled analysers/generators from AP_LIBfoo
+
+2013-11-06  unhammer
+
+	* [r48395] Makefile.am: use apertium.m4, supporting
+	  --with-lang1n=/path/to/nno --with-lang1b=/path/to/nob
+
+2013-10-31  unhammer
+
+	* [r48232] configure.ac: configure.ac's updated to use apertium.pc
+	  (or apertium-3.2.pc if that doesn't exist)
+	* [r48230] apertium-nno.nno.dix: mv infreq to languages (didn't I
+	  do this already)
+
+2013-10-15  unhammer
+
+	* [r48117] Makefile.am, alt.xsl, apertium-nno.nno.tsx, nno.prob,
+	  rem-compounds.xsl: nor-eng depends on nno/nob for dix, prob, xsl
+	* [r48112] Makefile.am: use $(TARGETS_COMMON) in _DATA to make it
+	  even more less redundanter
+	* [r48110] Makefile.am: s/LANG2/LANG1/g # there is no LANG2
+	* [r48108] Makefile.am: fix mode building, add EXTRA_DIST goals for
+	  tarballing
+
+	  also, modes.xml for tuk and fix dix-build for hbs
+
+2013-10-14  unhammer
+
+	* [r48087] Makefile.am, apertium-nno.pc.in: as -r48046; put dir and
+	  srcdir in pc.in, minor mode build fixes
+
+2013-10-12  spectre360
+
+	* [r48039] ., AUTHORS, COPYING, ChangeLog, Makefile.am, NEWS,
+	  README, apertium-nno.nno.dix, apertium-nno.nno.rlx,
+	  apertium-nno.pc.in, autogen.sh, configure.ac, modes.xml: add nno
+	  dir
+

src/fst/morphology/ext-Apertium-nno/Makefile.am

@@ -0,0 +1,113 @@
+###############################################################################
+## Makefile for apertium-nno
+###############################################################################
+
+LANG1=nno
+BASENAME=apertium-$(LANG1)
+
+TARGETS_COMMON = $(LANG1).automorf.bin \
+	$(LANG1).autogen.bin \
+	$(LANG1).autogen.att.gz \
+	$(LANG1).autopgen.bin \
+	$(LANG1).automorf.att.gz \
+	$(LANG1).rlx.bin \
+        $(LANG1).seg.rlx.bin                                \
+	$(LANG1).e_vi_prefs.rlx.bin \
+	$(LANG1).prefs.rlx.bin \
+	$(LANG1).syn.rlx.bin
+
+# This include defines goals for install-modes, .deps/.d and .mode files:
+@ap_include@
+
+###############################################################################
+## Norwegian Nynorsk transducer
+###############################################################################
+
+.deps/%.autogen.bin: $(BASENAME).%.dix .deps/.d
+	apertium-validate-dictionary $<
+	lt-comp rl $< $@
+
+$(LANG1).autogen.bin: .deps/$(LANG1).autogen.bin .deps/proper-$(LANG1).autogen.bin
+	lt-append $^ $@
+
+# .deps/improper-$(LANG1).automorf.bin  and  .deps/$(LANG1).automorf.bin:
+.deps/%.automorf.bin: $(BASENAME).%.dix $(BASENAME).$(LANG1).acx .deps/.d
+	apertium-validate-dictionary $<
+	apertium-validate-acx $(BASENAME).$(LANG1).acx
+	lt-comp lr $< $@ $(BASENAME).$(LANG1).acx
+
+$(LANG1).automorf.bin: .deps/$(LANG1).automorf.bin .deps/proper-$(LANG1).automorf.bin
+	lt-append $^ $@
+
+
+$(LANG1).autogen.att.gz: $(LANG1).autogen.bin
+	lt-print $< | gzip -9 -c -n > $@
+
+$(LANG1).automorf.att.gz: $(LANG1).automorf.bin
+	lt-print $< | gzip -9 -c -n > $@
+
+$(LANG1).autopgen.bin: $(BASENAME).post-$(LANG1).dix
+	lt-comp lr $< $@
+
+###############################################################################
+## Distribution
+###############################################################################
+EXTRA_DIST=$(BASENAME).$(LANG1).dix \
+           $(BASENAME).post-$(LANG1).dix \
+           $(BASENAME).proper-$(LANG1).dix \
+	   $(BASENAME).$(LANG1).rlx \
+	   $(BASENAME).$(LANG1).seg.rlx \
+	   $(BASENAME).$(LANG1).syn.rlx \
+           $(LANG1).prob            \
+	   $(BASENAME).$(LANG1).tsx \
+	   $(LANG1).preferences.xml \
+	   modes.xml
+
+###############################################################################
+## Installation stuff
+###############################################################################
+#
+#   apertium_nno_dir: This is where the compiled binaries go
+#   apertium_nno_srcdir: This is where the source files go
+
+apertium_nnodir=$(prefix)/share/apertium/$(BASENAME)/
+apertium_nno_srcdir=$(prefix)/share/apertium/$(BASENAME)/
+
+apertium_nno_DATA=$(TARGETS_COMMON) $(LANG1).prob \
+				  $(LANG1).preferences.xml \
+				  $(BASENAME).$(LANG1).dix \
+				  $(BASENAME).post-$(LANG1).dix \
+				  $(BASENAME).proper-$(LANG1).dix \
+				  $(BASENAME).$(LANG1).rlx \
+				  $(BASENAME).$(LANG1).seg.rlx \
+				  $(BASENAME).$(LANG1).syn.rlx \
+				  $(BASENAME).$(LANG1).tsx
+
+pkgconfigdir = $(prefix)/share/pkgconfig
+pkgconfig_DATA = $(BASENAME).pc
+
+noinst_DATA=modes/$(LANG1)-morph.mode
+
+install-data-local: install-modes
+uninstall-local: uninstall-modes
+
+###############################################################################
+## Cleanup
+###############################################################################
+
+CLEANFILES = $(TARGETS_COMMON)
+clean-local:
+	-rm -rf .deps modes
+
+###############################################################################
+## Test
+###############################################################################
+
+.PHONY: test
+test: all
+	@grep -nE 'lm=".*/([a-zæøå]+)_(\\[^_]+|)_' $(BASENAME).*$(LANG1).dix \
+	    | grep -vE '<i?g/?>'                                             \
+	    | grep -vE  'lm="(.*)(.*)".*/\2_(\\[^_]+|)_'                     \
+	    | sed 's%^%WARNING: lemma vs par/tail inconsistency in %'
+	apertium-regtest test
+	test -e apertium-lint && apertium-lint $(BASENAME).$(LANG1).dix

src/fst/morphology/ext-Apertium-nno/README

@@ -0,0 +1,15 @@
+Some example pipelines:
+
+$ echo Har du lyst til å kjøra eller kjøre no? | lt-proc -we nno.automorf.bin | cg-proc -w nno.rlx.bin | apertium-tagger -g nno.prob | lt-proc -g nno.autogen.bin
+Har du lyst til å køyra eller køyra no?
+
+$ echo Har du lyst til å kjøra eller kjøre no? | lt-proc -we nno.automorf.bin | cg-proc -w nno.rlx.bin | apertium-tagger -g nno.prob | lt-proc -g nno_e.autogen.bin
+Har du lyst til å køyre eller køyre no?
+
+The two generators differ only in whether they give -a or -e as infinitive endings.
+
+
+For more information, see
+- https://wiki.apertium.org/wiki/Apertium-nno
+- https://wiki.apertium.org/wiki/Installation
+

src/fst/morphology/ext-Apertium-nno/apertium-nno.nno.acx

@@ -0,0 +1,10 @@
+<?xml version="1.0"?>
+<analysis-chars>
+<!-- Make apostrophe variants equal ' -->
+  <char value="'">
+    <equiv-char value="&#x2019;"/>
+    <equiv-char value="&#x2BC;"/>
+    <equiv-char value="&#8217;"/>
+    <equiv-char value="&#39;"/>
+  </char>
+</analysis-chars>