Merge branch 'master' of https://github.com/apertium/apertium-nob

apertium-nob.nob.acx

@@ -4,6 +4,7 @@
   <char value="’">
     <equiv-char value="&#x2019;"/>
     <equiv-char value="&#x2BC;"/>
+    <equiv-char value="&#8216;"/>
     <equiv-char value="&#8217;"/>
     <equiv-char value="&#39;"/>
   </char>

apertium-nob.nob.dix

@@ -22,7 +22,7 @@
     Generated semi-automatically from the Norsk ordbank.
     http://www.edd.uio.no/prosjekt/ordbanken/
 
-    Grep for "todo" if you're looking for work.
+    Grep for "todo" if you’re looking for work.
 
     Some of the stuff in this file that is still very much work-in-progress
     or barely begun:
@@ -136,7 +136,7 @@
 	<pardefs>
           <!-- Compounding pardefs. -->
 <!-- The cmp tag may be added by transfer to sg.ind forms. Even nouns
-     that can't be analysed in compounds, can be generated in
+     that can’t be analysed in compounds, can be generated in
      compounds, and need to be able to make the sg.ind.cmp form.
      Ø-ep... outputs no epenthetic if used as compound-L -->
 <pardef n="'s_case\Ø_no-cp">
@@ -1545,7 +1545,7 @@
     <!-- Acronyms -->
     <!-- singular/plural todo -->
 <pardef n="km_no-cp__n" c="Note: gives form a dot iff lemma has a dot,
-                     don't include a final dot in the individual entries">
+                     don’t include a final dot in the individual entries">
   <e>       <re>.?</re><p><l></l>          <r><s n="n"/><s n="acr"/></r></p><par n="'s_case\Ø_no-cp"/></e>
 </pardef>
 
@@ -5430,8 +5430,8 @@
 
 <!-- sme-nob agreement-MWE nouns. TODO: use lt-mwpp for these. -->
 <pardef n="bærbar/_datamaskin__n" c="lt-mwpp mwe todo. Also todo: this has an adj already, so should have special transfer rules to insert determiner...
-                                     Definite form has indef noun; better to have no double defnes in 'den bærbare datamaskin' than overdo it in
-                                     'din bærbare datamaskinen'">
+                                     Definite form has indef noun; better to have no double defnes in ’den bærbare datamaskin’ than overdo it in
+                                     ’din bærbare datamaskinen'">
   <e>       <p><l>e<b/>datamaskinene</l>      <r><b/>datamaskin<s n="n"/><s n="f"/><s n="pl"/><s n="def"/></r></p><par n="s_case"/><par n="cp-R"/></e>
   <e>       <p><l>e<b/>datamaskiner</l>       <r><b/>datamaskin<s n="n"/><s n="f"/><s n="pl"/><s n="ind"/></r></p><par n="s_case"/><par n="cp-R"/></e>
   <e>       <p><l>e<b/>datamaskin</l>        <r><b/>datamaskin<s n="n"/><s n="f"/><s n="sg"/><s n="def"/></r></p><par n="s_case"/><par n="cp-R"/></e>
@@ -18361,7 +18361,7 @@
   <e r="RL"><re>([§\+-–] ?)?[0-9]+([.,][0-9]+)*([ \-]+[0-9]+([.,][0-9]+)*)*( *%)?</re><p><l></l><r><s n="det"/><s n="qnt"/><s n="un"/><s n="pl"/></r></p></e>
 </pardef>
 <pardef n="sme-nob-numeros" c="Used in sme-nob, should be RL to avoid messing up other pairs.
-                               Also, the \+ in regexes here can't be trimmed (and would mess up pretransfer in analysis anyway), so keep it RL.
+                               Also, the \+ in regexes here can’t be trimmed (and would mess up pretransfer in analysis anyway), so keep it RL.
                                (Also, this one pardef takes as long as the rest of the dix to compile, so without the RL the analyser also gets slow to compile …)">
   <!-- ordinals: -->
   <e r="RL"><re>[IVXLCDM\-]+</re><i>.</i><par n="første__adj"/></e>
@@ -18395,7 +18395,7 @@
   <e>       <re>[½¹²³¼¾]</re><p><l></l>                  <r><s n="det"/><s n="qnt"/><s n="un"/><s n="pl"/></r></p></e>
 
   <!-- Numbers from 1000 to 2200 (or 20-99) might be years. Other numbers don't get to be years.
-       Noun tagging would be better, but that doesn't work with compounding
+       Noun tagging would be better, but that doesn’t work with compounding
        since inconditional numerals override the unknown-analysis. -->
   <e c="date"><re>1[0-9][0-9][0-9]|2[0-2][0-9][0-9]|[2-9][0-9]</re><p><l></l><r><s n="det"/><s n="qnt"/><s n="un"/><s n="pl"/><s n="date"/></r></p></e>
   <e c="section"> <re>[0-9]+[\-][0-9]+</re><p><l></l><r><s n="det"/><s n="qnt"/><s n="un"/><s n="sg"/></r></p></e>
@@ -18445,13 +18445,10 @@
   <e>       <p><l>/</l><r>/<s n="cm"/></r></p></e>
 </pardef>
 
-<pardef n="aposchar">
-  <e>       <p><l>’</l>     <r>'</r></p></e>
-  <e r="LR"><p><l>'</l>     <r>'</r></p></e>
-</pardef>
-
 <pardef n="cometa">
-  <e>       <par n="aposchar"/><p><l></l><r><s n="apos"/></r></p></e>
+  <e c="left quote">             <p><l>‘</l>     <r>‘<s n="apos"/></r></p></e>
+  <e c="right quote / genitive"> <p><l>’</l>     <r>’<s n="apos"/></r></p></e>
+  <e c="foot">                   <p><l>'</l>     <r>'<s n="n"/><s n="m"/><s n="pl"/><s n="ind"/></r></p></e>
   <e>       <p><l>«</l><r>«<s n="lquot"/></r></p></e>
   <e>       <p><l>»</l><r>»<s n="rquot"/></r></p></e>
 </pardef>
@@ -21090,7 +21087,7 @@
 <e lm="med all tydelighet">        <i>med<b/>all<b/>tydelighet</i><par n="dagevis__adv"/></e>
 <e lm="med ansiktet ned">        <i>med<b/>ansiktet<b/>ned</i><par n="dagevis__adv"/></e>
 <e lm="med ansiktet vendt mot">        <i>med<b/>ansiktet<b/>vendt<b/>mot</i><par n="dagevis__adv"/></e>
-<e r="RL" lm="med dette" c="RL: don't use this entry for [med] [dette laget]"><i>med<b/>dette</i><par n="dagevis__adv"/></e>
+<e r="RL" lm="med dette" c="RL: don’t use this entry for [med] [dette laget]"><i>med<b/>dette</i><par n="dagevis__adv"/></e>
 <e lm="med dobbellås">        <i>med<b/>dobbellås</i><par n="dagevis__adv"/></e>
 <e lm="med en gang">        <i>med<b/>en<b/>gang</i><par n="dagevis__adv"/></e>
 <e lm="med enden mot">        <i>med<b/>enden<b/>mot</i><par n="dagevis__adv"/></e>
@@ -88440,7 +88437,7 @@
 <e lm="cerebral parese-pasient"><i>cerebral<b/>parese-pasient</i><par n="ep_\Ø__n"/></e>
 <e lm="cocker spaniel"><i>cocker<b/>spaniel</i><par n="ep_\Ø__n"/></e>
 <e lm="don juan"><i>don<b/>juan</i><par n="ep_\Ø__n"/></e>
-<e lm="duc d'albe"><i>duc<b/>d'albe</i><par n="re_\Ø__n"/></e>
+<e lm="duc d’albe"><i>duc<b/>d’albe</i><par n="re_\Ø__n"/></e>
 <e lm="eau de cologne"><i>eau<b/>de<b/>cologne</i><par n="re_\Ø__n"/></e>
 <e lm="eau de vie"><i>eau<b/>de<b/>vie</i><par n="re_\Ø__n"/></e>
 <e lm="ex libris"><i>ex<b/>libris</i><par n="ja_\Ø__n"/></e>
@@ -88465,7 +88462,7 @@
 <e lm="peau de pêche"><i>peau<b/>de<b/>pêche</i><par n="re_\Ø__n"/></e>
 <e lm="prae ceteris"><i>prae<b/>ceteris</i><par n="gon__n"/></e>
 <e lm="rock and roll"><i>rock<b/>and<b/>roll</i><par n="ep_\Ø__n"/></e>
-<e lm="rock'n roll"><i>rock'n<b/>roll</i><par n="ep_\Ø__n"/></e>
+<e lm="rock’n roll"><i>rock’n<b/>roll</i><par n="ep_\Ø__n"/></e>
 <e lm="sex appeal"><i>sex<b/>appeal</i><par n="ep_\Ø__n"/></e>
 <e lm="short story"><i>short<b/>story</i><par n="ep_\Ø__n"/></e>
 <e lm="sleeping partner"><i>sleeping<b/>partner</i><par n="uer__n"/></e>
@@ -90682,7 +90679,7 @@
 <e lm="trivia">          <i>trivia</i><par n="MC__n"/></e>
 <e c="tennisvariant" lm="padel"><i>padel</i><par n="MC__n"/></e>
 <e lm="fast track">      <i>fast<b/>track</i><par n="MC__n"/></e>
-<e lm="chargé d'affaires">          <i>chargé<b/>d'affaires</i><par n="MC__n"/></e>
+<e lm="chargé d’affaires">          <i>chargé<b/>d’affaires</i><par n="MC__n"/></e>
 <e lm="subsea">          <i>subsea</i><par n="MC__n"/></e>
 <e lm="fair play">	     <i>fair<b/>play</i><par n="MC__n"/></e>
 <e r="LR" lm="fairplay"> <p><l>fairplay</l><r>fair<b/>play</r></p><par n="MC_no-cp__n"/></e>
@@ -98468,7 +98465,7 @@
 <e lm="brusteinsball"><i>brusteinsball</i><par n="ja__n"/></e>
 <e lm="brutalisering"><i>brutalisering</i><par n="då_\s__n"/></e>
 <e lm="brutilhenger"><i>brutilhenger</i><par n="uer__n"/></e>
-<e lm="brutter'n"><i>brutter'n</i><par n="småen__n"/></e>
+<e lm="brutter’n"><i>brutter’n</i><par n="småen__n"/></e>
 <e lm="bruttoareal"><i>bruttoareal</i><par n="ja__n"/></e>
 <e lm="bruttoavanse"><i>bruttoavanse</i><par n="re_\Ø__n"/></e>
 <e lm="bruttoavkastning"><i>bruttoavkastning</i><par n="då__n"/></e>
@@ -99714,7 +99711,7 @@
 <e lm="colombianer"><i>colombianer</i><par n="uer__n"/></e>
 <e lm="coloradobille"><i>coloradobille</i><par n="re_\Ø__n"/></e>
 <e lm="columbi egg"><i>columbi<b/>egg</i><par n="ul__n"/></e>
-<e lm="commedia dell'arte"><i>commedia<b/>dell'arte</i><par n="MC__n"/></e>
+<e lm="commedia dell’arte"><i>commedia<b/>dell’arte</i><par n="MC__n"/></e>
 <e lm="commedia"><i>commedia</i><par n="MC__n"/></e>
 <e lm="computer"><i>computer</i><par n="uer__n"/></e>
 <e lm="computertomografi"><i>computertomografi</i><par n="ep_\Ø__n"/></e>
@@ -99729,7 +99726,7 @@
 <e lm="copyrightside"><i>copyrightsid</i><par n="enk/e__n"/></e>
 <e lm="cornerflagg"><i>cornerflagg</i><par n="ul__n"/></e>
 <e lm="corps de ballet"><i>corps<b/>de<b/>ballet</i><par n="sør__n"/></e>
-<e lm="ballet d'action"><i>ballet<b/>d'action</i><par n="sør__n"/></e>
+<e lm="ballet d’action"><i>ballet<b/>d’action</i><par n="sør__n"/></e>
 <e lm="corps diplomatique"><i>corps<b/>diplomatique</i><par n="sør__n"/></e>
 <e lm="corsicaner"><i>corsicaner</i><par n="uer__n"/></e>
 <e lm="costaricaner"><i>costaricaner</i><par n="uer__n"/></e>
@@ -104173,7 +104170,7 @@
 <e lm="fatling"><i>fatling</i><par n="då__n"/></e>
 <e lm="fatning"><i>fatning</i><par n="då__n"/></e>
 <e lm="fatteevne"><i>fatteevn</i><par n="enk/e__n"/></e>
-<e lm="fatter'n"><i>fatter'n</i><par n="småen__n"/></e>
+<e lm="fatter’n"><i>fatter’n</i><par n="småen__n"/></e>
 <e lm="fattigdomsbekjempelse"><i>fattigdomsbekjempelse</i><par n="re_\s__n"/></e>
 <e lm="fattigdomsgrense"><i>fattigdomsgrens</i><par n="enk/e__n"/></e>
 <e lm="fattigdomsproblem"><i>fattigdomsproblem</i><par n="ja__n"/></e>
@@ -105060,7 +105057,7 @@
 <e lm="finersag"><i>finersag</i><par n="då__n"/></e>
 <e lm="finfordeling"><i>finfordeling</i><par n="då__n"/></e>
 <e lm="finfølelse"><i>finfølelse</i><par n="re_\s__n"/></e>
-<e lm="finger'n"><i>finger'n</i><par n="småen__n"/></e>
+<e lm="finger’n"><i>finger’n</i><par n="småen__n"/></e>
 <e lm="fingerbredde"><i>fingerbredde</i><par n="re_\Ø__n"/></e>
 <e lm="fingerbreidd"><i>fingerbreidd</i><par n="då__n"/></e>
 <e lm="fingerbøl"><i>fingerbøl</i><par n="ul__n"/></e>
@@ -110320,7 +110317,7 @@
 <e lm="genspleising">    <i>genspleising</i><par n="då__n"/></e>
 <e lm="genteknikk">      <i>genteknikk</i><par n="ep_\Ø__n"/></e>
 <e lm="genteknologi">    <i>genteknologi</i><par n="ep_\Ø__n"/></e>
-<e lm="gentlemen's agreement"><i>gentlemen's<b/>agreement</i><par n="sør__n"/></e>
+<e lm="gentlemen’s agreement"><i>gentlemen’s<b/>agreement</i><par n="sør__n"/></e>
 <e lm="genuafokk">       <i>genuafokk</i><par n="då__n"/></e>
 <e lm="genueser">        <i>genueser</i><par n="uer__n"/></e>
 <e lm="genvitenskap">    <i>genvitenskap</i><par n="ep_\Ø__n"/></e>
@@ -115099,7 +115096,7 @@
 <e lm="hornskei">        <i>hornskei</i><par n="då__n"/></e>
 <e lm="hornskje">        <i>hornskje</i><par n="då__n"/></e>
 <e lm="hornugle">        <i>hornugl</i><par n="enk/e__n"/></e>
-<e lm="hors d'oeuvre">   <i>hors<b/>d'oeuvre</i><par n="mai__n"/></e>
+<e lm="hors d’oeuvre">   <i>hors<b/>d’oeuvre</i><par n="mai__n"/></e>
 <e lm="horse">           <i>hors</i><par n="enk/e__n"/></e>
 <e lm="horv">            <i>horv</i><par n="då__n"/></e>
 <e lm="horving">         <i>horving</i><par n="då__n"/></e>
@@ -127014,7 +127011,7 @@
 <e lm="køye">            <i>køy</i><par n="enk/e__n"/></e>
 <e lm="køyeseng">        <i>køyeseng</i><par n="då__n"/></e>
 <e lm="køying">          <i>køying</i><par n="då__n"/></e>
-<e lm="l'hombre">        <i>l'hombre</i><par n="mai__n"/></e>
+<e lm="l’hombre">        <i>l’hombre</i><par n="mai__n"/></e>
 <e lm="la dolce vita">   <i>la<b/>dolce<b/>vita</i><par n="MC__n"/></e>
 <e lm="la-skure-mentalitet"><i>la-skure-mentalitet</i><par n="ep_\Ø__n"/></e>
 <e lm="labbing">         <i>labbing</i><par n="då__n"/></e>
@@ -129501,7 +129498,7 @@
 <e lm="lottovinner">     <i>lottovinner</i><par n="leser__n"/></e>
 <e lm="lotusblomst">     <i>lotusblomst</i><par n="ep_\Ø__n"/></e>
 <e lm="lotusstilling">   <i>lotusstilling</i><par n="då__n"/></e>
-<e lm="louisd'or">       <i>louisd'or</i><par n="ep_\Ø__n"/></e>
+<e lm="louisd’or">       <i>louisd’or</i><par n="ep_\Ø__n"/></e>
 <e lm="lovarbeid">       <i>lovarbeid</i><par n="marked__n"/></e>
 <e lm="lovart">          <i>lovart</i><par n="MC__n"/></e>
 <e lm="lovavdeling">     <i>lovavdeling</i><par n="då__n"/></e>
@@ -134247,8 +134244,8 @@
 <e lm="mute">            <i>mut</i><par n="enk/e__n"/></e>
 <e lm="muting">          <i>muting</i><par n="då_\s__n"/></e>
 <e lm="mutter">          <i>mut</i><par n="mut/ter__n"/></e>
-<e lm="mutter'n">        <i>mutter'n</i><par n="småen__n"/></e>
-<e lm="pepper'n">        <i>pepper'n</i><par n="småen__n"/></e>
+<e lm="mutter’n">        <i>mutter’n</i><par n="småen__n"/></e>
+<e lm="pepper’n">        <i>pepper’n</i><par n="småen__n"/></e>
 <e lm="mutthet">         <i>mutthet</i><par n="då__n"/></e>
 <e lm="muzak">           <i>muzak</i><par n="ep_\Ø__n"/></e>
 <e lm="myanmarer">       <i>myanmarer</i><par n="uer__n"/></e>
@@ -192683,7 +192680,7 @@
 <e lm="dupuytrens kontraktur">                <i>dupuytrens<b/>kontraktur</i><par n="ep_\Ø__n"/></e>
 <e lm="dårlige boforhold">                    <i>dårlige<b/>boforhold</i><par n="ul__n"/></e>
 <e lm="eczema herpeticum">                    <i>eczema<b/>herpeticum</i><par n="MC__n"/></e>
-<e lm="ehlers-danlos' syndrom">               <i>ehlers-danlos'<b/>syndrom</i><par n="MC__n"/></e>
+<e lm="ehlers-danlos’ syndrom">               <i>ehlers-danlos’<b/>syndrom</i><par n="MC__n"/></e>
 <e lm="ektopisk acth-syndrom">                <i>ektopisk<b/>acth-syndrom</i><par n="ja_\Ø__n"/></e>
 <e lm="erythema infectiosum">                 <i>erythema<b/>infectios</i><par n="vel/um__n"/></e>
 <e lm="erythema nodosum">                     <i>erythema<b/>nodosum</i><par n="bom__n"/></e>
@@ -193717,7 +193714,8 @@
   <!-- not using letter pardefs since that makes the final transducer huge -->
   <e lm="np-guio regex"><re>[A-ZÆØÅ][a-zæøå]+\-</re><p><l/><r><s n="np"/><s n="guess"/></r></p><par n="cp-guio"/></e>
   <e lm="np-guio regex"><re>[A-ZÆØÅ][A-ZÆØÅ]+\-</re><p><l/><r><s n="np"/><s n="guess"/></r></p><par n="cp-guio"/></e>
-  <e lm="np-apos regex"><re>[A-ZÆØÅ][A-ZÆØÅa-zæøå\-]+[SZXszx]'</re><p><l/><r><s n="np"/><s n="guess"/><s n="gen"/></r></p></e>
+  <e lm="np-apos regex"><re>[A-ZÆØÅ][A-ZÆØÅa-zæøå\-]+[SZXszx]</re><p><l>’</l><r>’<s n="np"/><s n="guess"/><s n="gen"/></r></p></e>
+  <!-- Note: ’ has to be outside of regex in order for acx to take effect -->
 
   <!-- Numbers are *not* in inconditional, since that would block compounding with them: -->
   <e>                      <par n="numeros"/></e>

apertium-nob.nob.rlx

@@ -2752,6 +2752,18 @@ SELECT:name-seen-samecase-midsent np + ("(<\\p{Lu}.*>)"r) IF
 # «Sammenlagtpremien er den gjeveste, sa Kilde til NTB. Kilde og Alexis Pinturault kjempet om totalseieren …»
 
 
+
+# Fot-teiknet er berre fot etter tal, elles er det skrivemaskinsapostrof:
+SELECT:fot-fot ("'") IF
+        (-1 siffer)
+        ;
+SELECT:fot-tomme ("'") IF
+        (-2 siffer)
+        (-1 ("'"))              # TODO: kanskje '' som eige oppslag?
+        ;
+REMOVE:fot-apos ("'");
+
+
 ###################################################
 # SECTION Early removal of unlikely / rare forms: #
 ###################################################