From 3f28aa2707382250d21df51b175da5476220fd34 Mon Sep 17 00:00:00 2001
From: Flammie A Pirinen <flammie@iki.fi>
Date: Thu, 5 Oct 2023 12:14:29 +0200
Subject: [PATCH] cleanups

---
 src/filters/Makefile.am                       |  6 +++-
 src/filters/remove-DNorm-tags.regex           | 18 +++++++++++
 .../remove-derivation-position-tags.regex     | 10 ++++++
 src/filters/remove-norm-comp-tags.regex       | 23 +++++++++++++
 src/filters/rename-POS_before_Der-tags.regex  | 32 +++++++++++++++++++
 5 files changed, 88 insertions(+), 1 deletion(-)
 create mode 100644 src/filters/remove-DNorm-tags.regex
 create mode 100644 src/filters/remove-derivation-position-tags.regex
 create mode 100644 src/filters/remove-norm-comp-tags.regex
 create mode 100644 src/filters/rename-POS_before_Der-tags.regex

diff --git a/src/filters/Makefile.am b/src/filters/Makefile.am
index 4100d44a..0189bba2 100644
--- a/src/filters/Makefile.am
+++ b/src/filters/Makefile.am
@@ -22,7 +22,11 @@
 GIELLA_FILTER_LOCAL_REGEX_SRCS=\
 					downcase_UCletters.regex\
 					allow_uppercase.regex\
-					remove-bracketed-english-gloss.regex
+					remove-bracketed-english-gloss.regex\
+					remove-DNorm-tags.regex \
+					remove-derivation-position-tags.regex \
+					remove-norm-comp-tags.regex \
+					rename-POS_before_Der-tags.regex
 
 # List any local filter xfscript files here:
 GIELLA_FILTER_LOCAL_XFSCRIPT_SRCS=
diff --git a/src/filters/remove-DNorm-tags.regex b/src/filters/remove-DNorm-tags.regex
new file mode 100644
index 00000000..456f3ed6
--- /dev/null
+++ b/src/filters/remove-DNorm-tags.regex
@@ -0,0 +1,18 @@
+! Divvun & Giellatekno - open source grammars for Sámi and other languages
+! Copyright © 2000-2010 The University of Tromsø & the Norwegian Sámi Parliament
+! http://giellatekno.uit.no & http://divvun.no
+!
+! This program is free software; you can redistribute and/or modify
+! this file under the terms of the GNU General Public License as published by
+! the Free Software Foundation, either version 3 of the License, or
+! (at your option) any later version. The GNU General Public License
+! is found at http://www.gnu.org/licenses/gpl.html. It is
+! also available in the file $GTHOME/LICENSE.txt.
+!
+! Other licensing options are available upon request, please contact
+! giellatekno@uit.no or feedback@divvun.no
+
+# This filter removes Rel-Foc combinations.
+
+0 <- %+Use%/NotDNorm,
+0 <- %+Use%/DNorm;
diff --git a/src/filters/remove-derivation-position-tags.regex b/src/filters/remove-derivation-position-tags.regex
new file mode 100644
index 00000000..7f86484e
--- /dev/null
+++ b/src/filters/remove-derivation-position-tags.regex
@@ -0,0 +1,10 @@
+# This regex removes tags we do not want in the ordinary $(GTLANG).fst
+# That is, we do not give them as output from analysis
+
+# delete the tags governing normative/speller derivation restrictions
+0 <- %+Der,
+0 <- %+Der1,
+0 <- %+Der2,
+0 <- %+Der3,
+0 <- %+Der4,
+0 <- %+Der5;
diff --git a/src/filters/remove-norm-comp-tags.regex b/src/filters/remove-norm-comp-tags.regex
new file mode 100644
index 00000000..8f83ca91
--- /dev/null
+++ b/src/filters/remove-norm-comp-tags.regex
@@ -0,0 +1,23 @@
+# This regex removes tags we do not want in the ordinary $(GTLANG).fst
+# That is, we do not give them as output from analysis
+
+# delete normative tags governing compound behaviour
+# -N- in the tag name stands for Normative
+0 <- %+CmpNP%/All,
+0 <- %+CmpNP%/First,
+0 <- %+CmpNP%/Pref,
+0 <- %+CmpNP%/Last,
+0 <- %+CmpNP%/Suff,
+0 <- %+CmpNP%/None,
+0 <- %+CmpNP%/Only,
+0 <- %+CmpN%/SgN,
+0 <- %+CmpN%/SgG,
+0 <- %+CmpN%/PlG,
+0 <- %+CmpN%/SgNomLeft,
+0 <- %+CmpN%/SgGenLeft,
+0 <- %+CmpN%/PlGenLeft,
+0 <- %+CmpN%/Def,
+0 <- %+CmpN%/DefSgNom,
+0 <- %+CmpN%/DefSgGen,
+0 <- %+CmpN%/DefPlGen
+;
diff --git a/src/filters/rename-POS_before_Der-tags.regex b/src/filters/rename-POS_before_Der-tags.regex
new file mode 100644
index 00000000..9395a0c5
--- /dev/null
+++ b/src/filters/rename-POS_before_Der-tags.regex
@@ -0,0 +1,32 @@
+# This script renames POS tags when in front of derivations, such that
+# only the final POS tag remains. This allows disambiguation to work
+# properly also for derived words, without having to resort to post-processing
+# hacks. The script is language specific.
+
+# POS tags changes:
+[ "+Ex/Adv"  <- "+Adv" ,
+  "+Ex/N"    <- "+N"   ,
+  "+Ex/A"    <- "+A"   ,
+  "+Ex/Attr" <- "+Attr",
+  "+Ex/V"    <- "+V"   ||
+# change always and only when followed by a derivation:
+  _ \[ "+Cmp" ]*
+       [ "+Der"
+       | "+Der1"
+       | "+Der2"
+       | "+Der3"
+       | "+Der4"
+       | "+Der5"
+       ]
+] .o.
+
+# Transitivity tag changes:
+[ "+Ex/TV" <- "+TV" ,
+  "+Ex/IV" <- "+IV" ||
+# change only within the same stem, and only when followed by
+# a verbal derivation:
+  _ \[ "+Cmp" ]*
+       [ "+TV"
+       | "+IV"
+       ]
+] ;