Merge branch 'master' into potential-unit-test-for-ids

languagetool-core/src/main/java/org/languagetool/Language.java

@@ -821,6 +821,9 @@ protected int getPriorityForId(String id) {
     if (id.equalsIgnoreCase("TOO_LONG_SENTENCE")) {
       return -101;  // don't hide spelling errors
     }
+    if (id.equals("REPETITIONS_STYLE")) {  // category
+      return -55;  // don't let style issues hide more important errors
+    }
     if (id.contains("STYLE")) {  // category
       return -50;  // don't let style issues hide more important errors
     }

languagetool-core/src/main/java/org/languagetool/rules/Categories.java

@@ -32,7 +32,7 @@ public final class Categories {
 
   public static final Categories REDUNDANCY = make("REDUNDANCY", "category_redundancy");
 
-  public static final Categories REPETITIONS_STYLE = make("STYLE", "cateogry_repetitions_style");
+  public static final Categories REPETITIONS_STYLE = make("REPETITIONS_STYLE", "cateogry_repetitions_style");
 
   /** General style issues not covered by other categories, like overly verbose wording. */
   public static final Categories STYLE = make("STYLE", "category_style");

languagetool-core/src/main/java/org/languagetool/rules/WordRepeatBeginningRule.java

@@ -38,7 +38,7 @@ public class WordRepeatBeginningRule extends TextLevelRule {
 
   public WordRepeatBeginningRule(ResourceBundle messages, Language language) {
     super(messages);
-    super.setCategory(Categories.STYLE.getCategory(messages));
+    super.setCategory(Categories.REPETITIONS_STYLE.getCategory(messages));
     setLocQualityIssueType(ITSIssueType.Style);
   }
 

languagetool-core/src/main/java/org/languagetool/rules/spelling/multitoken/MultitokenSpellerFilter.java

@@ -53,6 +53,14 @@ public RuleMatch acceptRuleMatch(RuleMatch match, Map<String, String> arguments,
     if (replacements.isEmpty()) {
       return null;
     }
+    if (patternTokenPos==1) {
+      List<String> capitalizedReplacements = new ArrayList<>();
+      for (String replacement : replacements) {
+        String capitalized = StringTools.uppercaseFirstChar(replacement);
+        capitalizedReplacements.add(capitalized);
+      }
+      replacements = capitalizedReplacements;
+    }
     match.setSuggestedReplacements(replacements);
     return match;
   }

languagetool-core/src/main/resources/org/languagetool/resource/spelling_global.txt

@@ -528,6 +528,7 @@ Lockheed Martin
 Guinness
 MTV
 Rolling Stones
+Like a Rolling Stone
 Beatles
 Madonna
 Rihanna
@@ -1197,6 +1198,7 @@ Clint Eastwood
 Russell Crowe
 Netscape
 Microsoft Dynamics
+Microsoft Designer
 MS Dynamics
 JDK
 Sotheby's
@@ -3861,6 +3863,7 @@ Sunset Strip
 Sunset Blvd
 Aaron Swartz
 Michael van Gerwen
+van Gerwen
 Simon Wiesenthal
 Adam Driver
 Ashley Graham
@@ -6394,6 +6397,7 @@ Windows Media Player
 Windows Movie Maker
 Wu Tang Clan
 Ho Chi Minh
+Hồ Chí Minh
 Des Moines
 Transylvania County
 MTV Movie Awards
@@ -8537,8 +8541,8 @@ Márta Pardavi
 Mónica González
 Mónica Gómez
 Mônica Benício
-N'golo Kanté
-N’golo Kanté
+N'Golo Kanté
+N’Golo Kanté
 Neven Subotić
 Niko Kovač Wie
 Nora Szász
@@ -8710,6 +8714,7 @@ Eric Cantona
 Éric Toledano
 Étienne Schneider
 Ólafur Elíasson
+Olafur Eliasson
 Óscar Alberto Martínez
 Òscar Grau
 Bartolomé de las Casas
@@ -19169,7 +19174,9 @@ San Giorgio
 Jackson Hinkle
 Anna Linnikova
 San Sebastián
-Office Europeen de Lutte Antifraude
+São Sebastião
+Office européen de lutte antifraude
+Office Européen de Lutte Antifraude
 Nissin Foods
 Colin Burgess
 Angus Young
@@ -19181,11 +19188,62 @@ Rob Bailey
 Peter Clack
 Bon Scott
 Paul Matters
-Mark Evans
 Simon Wright
 Chris Slade
 Malcolm Young
 Larry Van Kriedt
 Dave Evans
 Emil Forsberg
 Van Baerlestraat
+Josep Pla
+Wolfgang Sawallisch
+Eugen Jochum
+Orchestre de la Suisse Romande
+Carmina Burana
+Catulli Carmina
+Emmy Awards
+Emmy Award
+Charles Trépel
+Johan Hultin
+Antonio Dal Masetto
+Juan Benet Goitia
+Juan Benet
+Yefim Bronfman
+Leif Ove Andsnes
+Jelly Roll Morton
+Jelly Roll
+Andrea Petković
+Bonastruc ça Porta
+Hiroshima mon amour
+Emmanuelle Riva
+Jorge Manrique
+Julio Manrique
+Juan Bernat
+Fab Morvan
+Michael Smith
+Luke Humphries
+Gerwyn Price
+Nathan Aspinall
+Danny Noppert
+Rob Cross
+Jonny Clayton
+Damon Heta
+Dave Chisnall
+Dirk van Duijvenbode
+James Wade
+Joe Cullen
+Gabriel Clemens
+Martin Schindler
+Dimitri Van den Bergh
+Van den Bergh
+Ross Smith
+Chris Dobey
+Stephen Bunting
+Ryan Searle
+Andrew Gilding
+Gary Anderson
+Josh Rock
+José de Sousa
+Daryl Gurney
+Brendan Dolan
+Raymond van Barneveld

languagetool-core/src/test/java/org/languagetool/rules/WordListValidatorTest.java

@@ -33,7 +33,7 @@ public class WordListValidatorTest {
   private static final String VALID_CHARS =
           "[ 0-9a-zA-ZöäüÖÄÜßëçèéáàóòŁÈÉÁÀÓÒÍãñí&*_:\\\\" +
           "___INSERT___" +
-          "Œ€ūαΑβΒγɣΓδΔεΕζΖηΗθΘιΙκΚλΛμΜνΝξΞοΟπΠρΡσΣτΤυΥφΦχΧψΨωΩάΆέΈίΊήΉύΎϊϋΰΐœţłń" +
+          "Œ€ūαΑβΒγɣΓδΔεΕζΖηΗθΘιΙκΚλΛμΜνΝξΞοΟπΠρΡσΣτΤυΥφΦχΧψΨωΩάΆέΈίΊήΉύΎϊϋΰΐœţłńÿ" +
           "ČŚśŌōżúïÎôêâû" +
           "ÇÃÕÚÊÂÔ" +
           "ă" +

languagetool-language-modules/be/pom.xml

@@ -43,6 +43,11 @@
             <groupId>org.languagetool</groupId>
             <artifactId>languagetool-core</artifactId>
         </dependency>
+        <dependency>
+            <!-- Hunspell dictionary for Belarusian language -->
+            <groupId>io.github.belarus</groupId>
+            <artifactId>linguistics.grammardb.spell.languagetool</artifactId>
+        </dependency>
         <dependency>
             <groupId>junit</groupId>
             <artifactId>junit</artifactId>

languagetool-language-modules/ca/src/main/resources/org/languagetool/resource/ca/added.txt

@@ -4,15 +4,22 @@
 # Format: fullform baseform postags (tab separated)
 ########## Dictionary fixes ##########
 ########## Additions ##########
+Aixalà	Aixalà	NPCNSP0
 Anschluss	Anschluss	NPMSO00
 Apiretal	Apiretal	NPCSO00
+Bankman-Fried	Bankman-Fried	NPCNSP0
 Berghof	Berghof	NPMSG00
 Burroughs	Burroughs	NPCNSO0
 CFA	CFA	NCMN000
 EOI	EOI	NCFN000
+Emmy	Emmy	NPMNO00
+Emmys	Emmy	NPMPO00
+Giráldez	Giráldez	NPCNSP0
 Göring	Göring	NPCNSP0
 Jacqueline	Jacqueline	NPFSSP0
+Manrique	Manrique	NPCNSP0
 Musikverein	Musikverein	NPFSO00
+Nahmànides	Nahmànides	NPMSSP0
 Nussbaum	Nussbaum	NPCNSP0
 Schuschnigg	Schuschnigg	NPCNSP0
 TS	TS	NCCN000

languagetool-language-modules/ca/src/main/resources/org/languagetool/resource/ca/multiwords.txt

@@ -1,5 +1,12 @@
 #Catalan multiwords file used for chunking
 #separatorRegExp=[\t;]
+Maria Barrientos;NPFSSP0
+Jesús María Barrientos;NPMSSP0
+Tamara Ndong;NPFSSP0
+Sergi Pompermayer;NPMSSP0
+Julio Manrique;NPMSSP0
+Bonastruc ça Porta;NPMSSP0
+Carmen García;NPFSSP0
 Sant Sebastià;NPMSG00
 Rosa Leveroni;NPFSSP0
 Evast e Blaquerna;NPMSO00
@@ -297,7 +304,6 @@ Esteban González Pons;NPMSSP0
 González Pons;NPCNSP0
 Andoni Ortuzar;NPMSSP0
 Iuri Gagarin;NPMSSP0
-Alcalá de Guadaira;NPCSG00
 Pablo Iglesias;NPMSSP0
 res extensa;NCFS000
 res cogitans;NCFS000
@@ -1954,7 +1960,7 @@ Bruce Lee	NPMSSP0
 Vicent Partal	NPMSSP0
 Malcom X	NPMSSP0
 Sun Tzu	NPMSSP0
-Jon Inarritu	NPMSSP0
+Jon Iñarritu;NPMSSP0
 Theresa May	NPFSSP0
 Angela Merkel	NPFSSP0
 stop-motion	NCFS000
@@ -2099,6 +2105,7 @@ Open Arms	NPCNO00
 moto club	NPMS000
 moto clubs	NPMP000
 Carmina Burana	NCCN000
+Carmina burana	NCCN000
 Mato Grosso	NPMSG00
 Luz de Gas	NPMSO00
 ADN polimerasa	NCFS000

languagetool-language-modules/ca/src/main/resources/org/languagetool/resource/ca/removed.txt

@@ -6,6 +6,7 @@ CFA	CFA	NCMP000
 CFA	CFA	NCMS000
 EOI	EOI	NCFP000
 EOI	EOI	NCFS000
+Maimònides	Maimònides	NPMPSP0
 TS	TS	NPMSO00
 XL	XL	AO0CN0
 ZER	ZER	NCFP000

languagetool-language-modules/ca/src/main/resources/org/languagetool/resource/ca/spelling.txt

@@ -1,14 +1,21 @@
 # Words that extend the spell checker. See ignore.txt for words that should be
 # completely ignored (i.e. not used to create suggestions).
+Aixalà
 Anschluss
 Apiretal
+Bankman-Fried
 Berghof
 Burroughs
 CFA
 EOI
+Emmy
+Emmys
+Giráldez
 Göring
 Jacqueline
+Manrique
 Musikverein
+Nahmànides
 Nussbaum
 Schuschnigg
 TS

languagetool-language-modules/ca/src/main/resources/org/languagetool/rules/ca/check_case.txt

@@ -802,4 +802,5 @@ República Txeca
 República de Sud-àfrica
 sistema mètric decimal
 sistema internacional d'unitats
-Ho Chi Minh
+Ho Chi Minh
+Carmina Burana