Põhja, Loode etc can prefix a geoname

giellalt · Jun 14, 2024 · 21615e3 · 21615e3
1 parent ba18a4d
commit 21615e3
Show file tree

Hide file tree

Showing 3 changed files with 39 additions and 13 deletions.
diff --git a/src/fst/morphology/root.lexc b/src/fst/morphology/root.lexc
@@ -1038,6 +1038,7 @@ LEXICON SingleWords
   Conjunctions ;             !!= * `@CODE@`
   NoninflectingVerbs ;       !!= * `@CODE@`
   NoninflectingAdjectiveVa ; !!= * `@CODE@`
+  ProperNounsGeoPrefixed ;   !!= * `@CODE@`  Kagu-Läti, Kesk-Argentiina etc
   NonCompoundingAdverbs ;    !!= * `@CODE@`
   AdverbsLast ;              !!= * `@CODE@`  an adverb that may be either a simplex word, or the second part of a compound 
 
@@ -1157,10 +1158,6 @@ LEXICON Latter !!= * `@CODE@` the latter part of a compound
   @R.POS.AComp@@R.Case.Gen@ LatterAdjective_ne ;  ! enamaruuduline, enimaruuduline
 
   @R.POS.Adv@ LatterAdjective_v ;        ! harvanähtavaid etc 
-!  @R.POS.Adv@@R.Stem.topelt@ LatterAdjective ;        ! topelthallikas etc 
-!  @R.POS.Adv@@R.Stem.vähe@ LatterAdjective ;        ! vähehallikas etc 
-!  @R.POS.Adv@@R.Stem.üle@ LatterAdjective ;        ! ülehallikas etc 
-!  @R.POS.Adv@@D.Stem.topelt@@D.Stem.vähe@@D.Stem.üle@ LatterAdjective_v ;        ! harvanähtavaid etc 
 
 ! x + noun
   @R.POS.ACRMinus@ LatterNoun ;      
@@ -1185,11 +1182,6 @@ LEXICON Latter !!= * `@CODE@` the latter part of a compound
 
   @R.Der.us@@R.Case.Nom@@D.Stem.Nom@ LatterNoun ;  ! funktsionaalsusriiul
 
-!  @R.POS.Adv@@R.Stem.topelt@ LatterNoun ;        ! topeltriiul etc 
-!  @R.POS.Adv@@R.Stem.vähe@@P.NeedAdj.On@@D.NeedDerNe@ LatterNoun ;  ! vähe + kalor +Der/line; not väheautone
-!  @R.POS.Adv@@R.Stem.üle@ LatterNoun ;        ! ületoodang etc 
-!  @R.POS.Adv@@R.Stem.üle@ LatterProperNounGeo ;          ! ülesiberiline
-
 ! x + deverbal noun (in addition to x + noun)
 
   @R.POS.N@@R.Case.Sem@ LatterDeverbal ;  ! mägedesttulek
@@ -1212,10 +1204,6 @@ LEXICON Latter !!= * `@CODE@` the latter part of a compound
   @R.POS.ACR@@R.Case.Nom@@P.POS.Num@@P.Part.Three@ CardinalNumbersInCompFin ;       ! E10
 
 ! x + adverb
-!  @R.POS.Pref@@P.POS.Adv@@P.NeedPart.Three@ CompoundingAdverbs ;  ! era+ratsa+baas, era+järel+valvama+Der/ja 
-!  @R.POS.N@@R.Case.Nom@@R.Stem.Nom@@P.POS.Adv@@C.Case@@P.NeedPart.Three@ CompoundingAdverbs ;  ! ehitus+järel+valvama+Der/ja 
-!  @R.POS.N@@R.Case.Gen@@P.POS.Adv@@C.Case@@P.NeedPart.Three@ CompoundingAdverbs ;  ! ehitus+järel+valvama+Der/ja 
-
   @R.POS.Pref@@P.POS.Adv@@P.NeedPart.Three@ AdverbsFirst ;  ! era+ratsa+baas, era+järel+valvama+Der/ja 
   @R.POS.Pref@@P.POS.Adv@@P.NeedPart.Three@ AdverbsFirstOrLast ;  ! era+hästi+valvama+Der/ja 
   @R.POS.N@@R.Case.Nom@@R.Stem.Nom@@P.POS.Adv@@C.Case@@P.NeedPart.Three@ AdverbsFirst ;  ! ehitus+järel+valvama+Der/ja 

diff --git a/src/fst/morphology/stems/propernouns.lexc b/src/fst/morphology/stems/propernouns.lexc
@@ -4,6 +4,23 @@ LEXICON ProperNouns
   ProperNounsOther ;
 
 
+LEXICON ProperNounsGeoPrefixed
+  Alam-:Alam-# ProperNounsGeo ; 
+  Edela-:Edela-# ProperNounsGeo ; 
+  Ida-:Ida-# ProperNounsGeo ; 
+  Indo-:Indo-# ProperNounsGeo ; 
+  Kagu-:Kagu-# ProperNounsGeo ; 
+  Kesk-:Kesk-# ProperNounsGeo ; 
+  Kirde-:Kirde-# ProperNounsGeo ; 
+  Ladina-:Ladina-# ProperNounsGeo ; 
+  Loode-:Loode-# ProperNounsGeo ; 
+  Lääne-:Lääne-# ProperNounsGeo ; 
+  Lõuna-:Lõuna-# ProperNounsGeo ; 
+  Põhja-:Põhja-# ProperNounsGeo ; 
+  Suur-:Suur-# ProperNounsGeo ; 
+  Väike-:Väike-# ProperNounsGeo ; 
+  Ülem-:Ülem-# ProperNounsGeo ; 
+
 LEXICON ProperNounsGeo
 
 Aachen+N+Prop:˘Aachen REDEL  "weight: 11 " ; 

diff --git a/src/import/fsgt2final.sh b/src/import/fsgt2final.sh
@@ -684,6 +684,27 @@ LC_COLLATE=C join -a 1 -a 2 -t "_" -e "###" -o 1.2 2.1 2.2 names.tmp1 proper.tmp
 > proper.tmp2 
 
 echo 'LEXICON ProperNouns\n  ProperNounsGeo ;\n  ProperNounsPersons ;\n  ProperNounsOther ;\n' > propernouns.protolexc
+
+echo '\nLEXICON ProperNounsGeoPrefixed' >> propernouns.protolexc
+
+# loend8 originates from Vabamorf; the folowing command creates the following entries
+# cat loend8 | sed "s/^.*$/echo '  &-:&-# ProperNounsGeo ; ' >> propernouns.protolexc/"
+echo '  Alam-:Alam-# ProperNounsGeo ; ' >> propernouns.protolexc
+echo '  Edela-:Edela-# ProperNounsGeo ; ' >> propernouns.protolexc
+echo '  Ida-:Ida-# ProperNounsGeo ; ' >> propernouns.protolexc
+echo '  Indo-:Indo-# ProperNounsGeo ; ' >> propernouns.protolexc
+echo '  Kagu-:Kagu-# ProperNounsGeo ; ' >> propernouns.protolexc
+echo '  Kesk-:Kesk-# ProperNounsGeo ; ' >> propernouns.protolexc
+echo '  Kirde-:Kirde-# ProperNounsGeo ; ' >> propernouns.protolexc
+echo '  Ladina-:Ladina-# ProperNounsGeo ; ' >> propernouns.protolexc
+echo '  Loode-:Loode-# ProperNounsGeo ; ' >> propernouns.protolexc
+echo '  Lääne-:Lääne-# ProperNounsGeo ; ' >> propernouns.protolexc
+echo '  Lõuna-:Lõuna-# ProperNounsGeo ; ' >> propernouns.protolexc
+echo '  Põhja-:Põhja-# ProperNounsGeo ; ' >> propernouns.protolexc
+echo '  Suur-:Suur-# ProperNounsGeo ; ' >> propernouns.protolexc
+echo '  Väike-:Väike-# ProperNounsGeo ; ' >> propernouns.protolexc
+echo '  Ülem-:Ülem-# ProperNounsGeo ; ' >> propernouns.protolexc
+
 echo '\nLEXICON ProperNounsGeo\n' >> propernouns.protolexc
 cat proper.tmp2 | grep '^G_' | sed 's/^G_//' >> propernouns.protolexc