[ca] improve rules, more words

languagetool-org · Jan 11, 2025 · 207f988 · 207f988
1 parent cb20c95
commit 207f988
Show file tree

Hide file tree

Showing 6 changed files with 90 additions and 14 deletions.
diff --git a/languagetool-language-modules/ca/src/main/resources/org/languagetool/resource/ca/added.txt b/languagetool-language-modules/ca/src/main/resources/org/languagetool/resource/ca/added.txt
@@ -82,6 +82,7 @@ Mompó	Mompó	NCNSP0
 NESE	NESE	NCFP000
 Nel·lo	Nel·lo	NPCNSP0
 No-Do	No-Do	NPMSO00
+Némirovsky	Némirovsky	NPCNSP0
 OIEC	OIEC	NPFSO00
 Oreo	Oreo	NPCNO00
 PKK	PKK	NPMSO00

diff --git a/...ol-language-modules/ca/src/main/resources/org/languagetool/resource/ca/disambiguation.xml b/...ol-language-modules/ca/src/main/resources/org/languagetool/resource/ca/disambiguation.xml
@@ -10815,7 +10815,7 @@ Copyright (C) 2012 Jaume Ortolà i Font
         </rule>
         <rule>
             <pattern>
-                <token regexp="yes">ha</token>
+                <token>ha</token>
                 <marker>
                     <token postag="N.*" postag_regexp="yes" inflected="yes">tema</token>
                 </marker>

diff --git a/...getool-language-modules/ca/src/main/resources/org/languagetool/resource/ca/multiwords.txt b/...getool-language-modules/ca/src/main/resources/org/languagetool/resource/ca/multiwords.txt
@@ -1,5 +1,6 @@
 #Catalan multiwords file used for chunking
 #separatorRegExp=[\t;]
+Bernat Dedéu;NPMSSP0
 Mohammed bin Salman;NPMSSP0
 coop.;NCFN00
 Costa Blanca;NPFSG00

diff --git a/...uagetool-language-modules/ca/src/main/resources/org/languagetool/resource/ca/spelling.txt b/...uagetool-language-modules/ca/src/main/resources/org/languagetool/resource/ca/spelling.txt
@@ -76,6 +76,7 @@ Mompó
 NESE
 Nel·lo
 No-Do
+Némirovsky
 OIEC
 Oreo
 PKK

diff --git a/languagetool-language-modules/ca/src/main/resources/org/languagetool/rules/ca/grammar.xml b/languagetool-language-modules/ca/src/main/resources/org/languagetool/rules/ca/grammar.xml
@@ -286,6 +286,12 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
                 <example correction="2014">Divendres, 15 d'agost del <marker>2013</marker>.</example>
             </rule>
             <rule id="CA_DATE_WEEKDAY_CURRENTYEAR" name="Data incorrecta (any actual)">
+                <antipattern>
+                    <token regexp="yes">&mesos_any;|&mesos_any_abrev;</token>
+                    <token>d'</token>
+                    <token postag="DD.*" postag_regexp="yes"/>
+                    <token>any</token>
+                </antipattern>
                 <antipattern>
                     <marker>
                         <token regexp="yes">&mesos_any;|&mesos_any_abrev;</token>
@@ -1683,6 +1689,23 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
                 <example correction="calor extrema">Ser pobre és el factor determinant de les morts per <marker>calor extrem</marker>.</example>
             </rule>
         </rulegroup>
+        <rulegroup id="LLUM_MASCULI" name="la llum -> el llum" default="temp_off">
+            <rule>
+                <pattern>
+                    <marker>
+                        <token postag="_GN_F." postag_regexp="yes" regexp="yes">llums?</token>
+                    </marker>
+                    <token regexp="yes" min="0">de|d'</token>
+                    <token inflected="yes" regexp="yes">estacionament|curt|llarg|encreuament|carretera|posició</token>
+                </pattern>
+                <filter class="org.languagetool.rules.ca.ConvertToGenderAndNumberFilter" args="lemmaSelect:[AN].* keepOriginal:true gender:M"/>
+                <message>En aquest context, "llum" és masculí.</message>
+                <example correction="Els llums"><marker>Les llums</marker> de posició.</example>
+                <example correction="Els llums llargs"><marker>Les llums llargues</marker>.</example>
+                <example>Llums curts</example>
+                <example>Llums llargs</example>
+            </rule>
+        </rulegroup>
         <rulegroup id="VESSANT" name="vessant (preferible en masculí)">
             <antipattern>
                 <token>vessant</token>
@@ -9966,19 +9989,28 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
                 <token regexp="yes">santedat|altesa|excel·lència</token>
             </antipattern>
             <antipattern>
-                <marker>
-                    <token>com</token>
-                    <token>a</token>
-                    <token regexp="yes">[mts]e(u|us|ua|ues|va|ves)|[nv]ostr(a|es?)</token>
-                </marker>
+                <token>com</token>
+                <token>a</token>
+                <token regexp="yes">[mts]e(u|us|ua|ues|va|ves)|[nv]ostr(a|es?)</token>
             </antipattern>
             <antipattern>
-                <marker>
-                    <token inflected="yes" regexp="yes">el|un</token>
-                    <token inflected="yes">altre</token>
-                    <token regexp="yes">[mts]e(u|us|ua|ues|va|ves)|[nv]ostr(a|es?)</token>
-                </marker>
+                <token inflected="yes" regexp="yes">el|un</token>
+                <token inflected="yes">altre</token>
+                <token regexp="yes">[mts]e(u|us|ua|ues|va|ves)|[nv]ostr(a|es?)</token>
             </antipattern>
+            <rule default="temp_off">
+                <pattern>
+                    <marker>
+                        <token regexp="yes">a|de|per</token>
+                        <token regexp="yes">[mts]eus?|[nv]ostres?<exception postag="LOC_ADV|_C_LLOC" postag_regexp="yes"/></token>
+                    </marker>
+                    <token postag="N.[MC].*" postag_regexp="yes"/>
+                </pattern>
+                <message>Hi falta un article.</message>
+                <suggestion><match no="1" regexp_match="(?iu)per" regexp_replace="pe"/>l<match no="2" regexp_match="(?iu)^.*[eu](s?)" regexp_replace="$1"/> \2</suggestion>
+                <example correction="dels nostres">Això és <marker>de nostres</marker> germans.</example>
+                <example correction="als nostres">Això va per <marker>a nostres</marker> germans.</example>
+            </rule>
             <rule>
                 <pattern>
                     <token><exception postag="D[DAIN].*|N.*|PP3MSA00|PP3CP000" postag_regexp="yes"/><exception inflected="yes">fer</exception><exception regexp="yes">tan|cal|le</exception></token>
@@ -47297,6 +47329,18 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
                 <example>una mena de presa per subjectar l'oponent</example>
                 <example>Les disposicions preses per apropiar-se'n.</example>
             </rule>
+            <rule>
+                <pattern>
+                    <token>tot</token>
+                    <token>són</token>
+                    <marker>
+                        <token>preses</token>
+                    </marker>
+                </pattern>
+                <message>¿Volíeu dir <suggestion>presses</suggestion> (desig de fer una cosa ràpidament)?</message>
+                <short>Possible confusió</short>
+                <example correction="presses">Ara tot són <marker>preses</marker>!</example>
+            </rule>
             <rule>
                 <pattern>
                     <token regexp="yes" inflected="yes">decisió|mesura</token>
@@ -76209,6 +76253,19 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
         </rulegroup>
     </category>
     <category id="INCORRECT_EXPRESSIONS" name="Expressions incorrectes" type="grammar">
+        <rule id="A_BARCELONA_ESTANT" name="a Barcelona estant -> a Barcelona -> de Barcelona estant" default="temp_off">
+            <pattern>
+                <token>a</token>
+                <token postag="NP..G.*" postag_regexp="yes"/>
+                <token>estant</token>
+            </pattern>
+            <filter class="org.languagetool.rules.AdaptSuggestionsFilter" args="none:none"/>
+            <message>Construcció probablement incorrecta.</message>
+            <suggestion>a \2</suggestion>
+            <suggestion>de \2 \3</suggestion>
+            <example correction="A Barcelona|De Barcelona estant"><marker>A Barcelona estant</marker> es veuen les illes Balears.</example>
+            <example correction="a Eivissa|d'Eivissa estant">Jo recordo un estiu <marker>a Eivissa estant</marker>.</example>
+        </rule>
         <rule id="JA_VEIG" name="ja veig -> ja ho veig">
             <pattern>
                 <token postag="SENT_START"/>
@@ -91532,10 +91589,10 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
             <pattern>
                 <token>(</token>
                 <marker>
-                    <token>veure</token>
+                    <token regexp="yes">veure|veieu</token>
                 </marker>
             </pattern>
-            <message>Com a fórmula de remissió, cal dir <suggestion>vegeu</suggestion>.</message>
+            <message>Com a fórmula de remissió, cal usar la forma en imperatiu <suggestion>vegeu</suggestion>.</message>
             <example correction="vegeu">Maquiavel (<marker>veure</marker> la cita).</example>
             <example>com a persona (veure's a un mateix com a persona)</example>
         </rule>
@@ -103656,6 +103713,21 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
                 <suggestion><match no="3" postag="(V.*)" postag_regexp="yes" postag_replace="$1">sentir</match></suggestion>
                 <example correction="sentir">No es podia <marker>escoltar</marker> res a 70 m.</example>
             </rule>
+            <rule default="temp_off">
+                <pattern>
+                    <token regexp="yes">el|l</token>
+                    <token>que</token>
+                    <token min="0" max="2" regexp="yes">la|gent|les|persones|votants|els|la|persona</token>
+                    <token inflected="yes">voler</token>
+                    <marker>
+                        <token>escoltar</token>
+                    </marker>
+                </pattern>
+                <message>&msg_escoltar_sentir;</message>
+                <suggestion>sentir</suggestion>
+                <example correction="sentir">Només diu als votants el que la gent vol <marker>escoltar</marker>.</example>
+                <example correction="sentir">La gent és molt manipulable, sobretot quant li diuen el que vol <marker>escoltar</marker>.</example>
+            </rule>
         </rulegroup>
         <rule id="TOTA_VEGADA_QUE" name="tota vegada que">
             <pattern>

diff --git a/...l-language-modules/ca/src/main/resources/org/languagetool/rules/ca/replace_multiwords.txt b/...l-language-modules/ca/src/main/resources/org/languagetool/rules/ca/replace_multiwords.txt
@@ -719,4 +719,5 @@ de pura cepa=de soca-rel
 pura cepa=de soca-rel
 caça bombarders|caça-bombarders|caçabombarders=caces bombarders	Forma de plural adequada.
 caçabombarder|caça-bombarder=caça bombarder	S'escriu separat sense guionet.
-costa de marfil|costa de vori=Costa d'Ivori	Nom de país.
+costa de marfil|costa de vori=Costa d'Ivori	Nom de país.
+famílies sòcies=famílies associades	"Associat" és l'adjectiu corresponent al substantiu "soci".
-Original file line number
+Diff line change
@@ Expand Up / @@ -76,6 +76,7 @@ Mompó @@
     NESE
     Nel·lo
     No-Do
+    Némirovsky
     OIEC
     Oreo
     PKK
@@ Expand Down @@