Merge pull request #10667 from languagetool-org/cb-changes-20240619

Cb changes 20240619
languagetool-org · Jun 20, 2024 · 4d5b0de · 4d5b0de
2 parents d71107e + b878bc9
commit 4d5b0de
Show file tree

Hide file tree

Showing 14 changed files with 98 additions and 6 deletions.
diff --git a/languagetool-core/src/main/resources/org/languagetool/resource/spelling_global.txt b/languagetool-core/src/main/resources/org/languagetool/resource/spelling_global.txt
@@ -5066,6 +5066,7 @@ Dorothy Dandridge
 Next.js
 Express.js
 Cash App
+Fortescue
 Fortescue Metals
 BHP Billiton
 Sole 24 Ore
@@ -25279,6 +25280,20 @@ María Zamora
 Apple Intelligence
 HiBob
 Sequoia
+Hiob Ludolf
+Adolf Rosenberger
+Éric Ciotti
+Trooping the Colour
+Viola Amherd
+Gareth Southgate
+Priscilla Chan
+Assemblée nationale
+Lloret de Mar
+Ernest Shackleton
+Guide Michelin
+DE-CIX
+Deutsche Commercial Internet Exchange
+Andrew Forrest
 Renault Koleos
 Son Sant Joan
 Son Sant Joan Airport

diff --git a/...getool-language-modules/de/src/main/java/org/languagetool/rules/de/GermanSpellerRule.java b/...getool-language-modules/de/src/main/java/org/languagetool/rules/de/GermanSpellerRule.java
@@ -61,7 +61,7 @@ public class GermanSpellerRule extends CompoundAwareHunspellRule {
 
   private static final String adjSuffix = "(affin|basiert|konform|widrig|fähig|haltig|bedingt|gerecht|würdig|relevant|" +
     "übergreifend|tauglich|untauglich|artig|bezogen|orientiert|fremd|liebend|hassend|bildend|hemmend|abhängig|zentriert|" +
-    "förmig|mäßig|pflichtig|ähnlich|spezifisch|verträglich|technisch|typisch|frei|arm|freundlich|feindlich|gemäß|neutral|seitig|begeistert|geeignet|ungeeignet|berechtigt|sicher|süchtig|resistent)";
+    "förmig|mäßig|pflichtig|ähnlich|spezifisch|verträglich|technisch|typisch|frei|arm|freundlich|feindlich|gemäß|neutral|seitig|begeistert|geeignet|ungeeignet|berechtigt|sicher|süchtig|resistent|verachtend)";
   private static final Pattern missingAdjPattern =
     compile("[a-zöäüß]{3,25}" + adjSuffix + "(er|es|en|em|e)?");
   private static final Pattern compoundPatternWithHeit = compile(".*(heit|keit|ion|ität|schaft|ung|tät)s");

diff --git a/...l-language-modules/de/src/main/resources/org/languagetool/resource/de/hunspell/ignore.txt b/...l-language-modules/de/src/main/resources/org/languagetool/resource/de/hunspell/ignore.txt
@@ -243509,6 +243509,11 @@ VKZ #abk
 Waldo/S
 Wiech/S
 Wieck/S
+Amherd/S #name
+Southgate/S #name
+IXP
+IXPs
+ISPs
 weird #eng
 Literacy #eng
 Backwaters

diff --git a/...language-modules/de/src/main/resources/org/languagetool/resource/de/hunspell/spelling.txt b/...language-modules/de/src/main/resources/org/languagetool/resource/de/hunspell/spelling.txt
@@ -63617,6 +63617,21 @@ DFKI # name
 Deltoid/S
 Deltoide/SN
 unstandardisiert/A
+Pepsinwein/S
+Bitterorange/N
+Bitterorangenschale/N
+Sequoia/S
+weg_besteuern
+demokratieverachtend/A
+menschenrechtsverachtend
+ressentimentgetragen/A
+weg_argumentieren
+oxidbasiert/A
+rauflustig/A
+beflocken
+beflockt/A
+unbeflockt/A
+risikoadjustiert/A
 Thujon
 Thujone/N
 Plastifizierung

diff --git a/...anguage-modules/de/src/main/resources/org/languagetool/resource/de/multitoken-suggest.txt b/...anguage-modules/de/src/main/resources/org/languagetool/resource/de/multitoken-suggest.txt
@@ -2933,6 +2933,8 @@ Commander Spock/S
 Taebaek Sanmaek/S
 Thomas Thabane/S
 Alexander Wiech/S
+Viola Amherd/S
+Embedded System/S
 Frank Yablans
 Julius Prott/S
 Julius Perotti/S

diff --git a/languagetool-language-modules/de/src/main/resources/org/languagetool/rules/de/replace.txt b/languagetool-language-modules/de/src/main/resources/org/languagetool/rules/de/replace.txt
@@ -724,6 +724,9 @@ Augenmakeup=Augen-Make-up
 Augenmakeups=Augen-Make-ups
 Restauranttip=Restauranttipp
 Restauranttips=Restauranttipps
+Eric Ciotti=Éric Ciotti
+Luis Figo=Luís Figo
+Antonio Costa=António Costa
 Kompentz=Kompetenz
 Kukulkan=K'uk'ulkan	Meinten Sie die gefiederte Schlange der Maya?
 Bohéme=Bohème|Boheme
diff --git a/...l-language-modules/en/src/main/resources/org/languagetool/resource/en/hunspell/ignore.txt b/...l-language-modules/en/src/main/resources/org/languagetool/resource/en/hunspell/ignore.txt
@@ -10408,6 +10408,15 @@ BSIP
 ToU
 DSAR
 enum
+dedupe
+dedupes
+deduped
+deduping
+Southgate
+IXP
+IXPs
+ISPs
+shouty
 α
 α-activity
 β

diff --git a/...language-modules/en/src/main/resources/org/languagetool/resource/en/hunspell/spelling.txt b/...language-modules/en/src/main/resources/org/languagetool/resource/en/hunspell/spelling.txt
@@ -340,9 +340,16 @@ underprovisioning
 optoelectronic
 optoelectronics
 derogative
+Kipchak
+Kipchaks
+incrementality
+idempotence
+idempotency
 experientially
 suppressions
 invalidations
+opt-in
+opt-ins
 tapback
 tapbacks
 royalty-free

diff --git a/...ge-modules/en/src/main/resources/org/languagetool/resource/en/hunspell/spelling_en-GB.txt b/...ge-modules/en/src/main/resources/org/languagetool/resource/en/hunspell/spelling_en-GB.txt
@@ -2,6 +2,8 @@
 # To add a word to all variants (en-US, en-GB,...) use spelling.txt
 # For multi-word terms, refer to https://github.com/languagetool-org/languagetool/issues/700
 callisthenics
+megadonor
+megadonors
 remodeller
 remodellers
 nanofibre

diff --git a/...ge-modules/en/src/main/resources/org/languagetool/resource/en/hunspell/spelling_en-US.txt b/...ge-modules/en/src/main/resources/org/languagetool/resource/en/hunspell/spelling_en-US.txt
@@ -2,6 +2,8 @@
 # To add a word to all variants (en-US, en-GB,...) use spelling.txt
 # For multi-word terms, refer to https://github.com/languagetool-org/languagetool/issues/700
 calisthenics
+megadonor
+megadonors
 remodeler
 remodelers
 nanofiber

diff --git a/...getool-language-modules/en/src/main/resources/org/languagetool/resource/en/multiwords.txt b/...getool-language-modules/en/src/main/resources/org/languagetool/resource/en/multiwords.txt
@@ -7960,5 +7960,6 @@ Mohammad Mokhber	NNP
 Ebrahim Raisi	NNP
 Normalització Lingüística	NNP
 Sukhoi Su-57	NNP
+ad infinitum	RB
 caffe latte	NN:UN
-Maillard reaction	NN
+Maillard reaction	NN
diff --git a/...ool-language-modules/en/src/main/resources/org/languagetool/resource/en/specific_case.txt b/...ool-language-modules/en/src/main/resources/org/languagetool/resource/en/specific_case.txt
@@ -5591,6 +5591,7 @@ Spratly Islands
 Lok Sabha
 Grana Padano
 Parmigiano Reggiano
+Golden Horde
 Boba Fett
 Amon-Ra St. Brown
 Ozark Highlands

diff --git a/languagetool-language-modules/en/src/main/resources/org/languagetool/rules/en/diacritics.txt b/languagetool-language-modules/en/src/main/resources/org/languagetool/rules/en/diacritics.txt
@@ -1387,4 +1387,7 @@ Normalitzacio Linguistica=Normalització Lingüística
 Canelo Alvarez=Canelo Álvarez
 Raphael Glucksmann=Raphaël Glucksmann
 Cesky Krumlov=Český Krumlov
+Eric Ciotti=Éric Ciotti
+Luis Figo=Luís Figo
+Antonio Costa=António Costa
 Raul Mulino=Raúl Mulino
diff --git a/languagetool-language-modules/en/src/main/resources/org/languagetool/rules/en/grammar.xml b/languagetool-language-modules/en/src/main/resources/org/languagetool/rules/en/grammar.xml
@@ -102,7 +102,7 @@ USA
     <!ENTITY typically_transitive_verbs "add|allow|bring|get|keep|let|link|lump|make|nominate|piece|put|see|send|tie">
 
     <!-- words that are probably more often used as a noun, preposition or interjection -->
-    <!ENTITY uncommon_verbs "ooh|key|chevy|neck|shark|chicken|suicide|dam|rice|twin|taboo|core|lullaby|ginger|patronage|caravan|buddy|guy|wive|van|green|toenail|pothole|yellow|safari|crap|bulldog|voodoo|minute|radio|fair|squirrel|ape|fox|badger|ferret|beagle|parrot|rat|arbitrage|privilege|deed|wolf|contango|coo|cat|tot|jut|bulletin|purpose|bitter|zinc|yuck|dot|valet|overnight|automobile|toilet|grandfather|mushroom|luck|paragraph|ambition|concept|field|baby|doll|flower|soldier|dude|volume|letter|id|pipeline|blackberry|bin|pretzel|turtle|panel|muscle|silver|warehouse|wood|machine|magic|clown|traffic|page|monkey|version|bundle|fan|till|dog|reason|garage|chapter|company|conference|chance|candy|room|console|margin|rabbit|article|cheese|while|yes|queen|gas|oil|officer|term|people|evidence|max|captain|case|step|minister|fun|water|excel|church|husband|dinner|police|curl|weed|platform|best|better|worst|twitter|prod|summit|calendar|chair|belt|mesh|curry|spam|sauce|lord|section|sheet|weekend|even|breakfast|tent|word|mad|mine|blue|eclipse|diesel|up|down|content|off|pretty|minor|well|physic|knight|dirty|tree|peter|harry|matt|mike|nick|bob|rick|sally|ralph|carol|war|shrimp|table|kit|gang|pen|soup|founder|proverb|jimmy|marcel|roger|wallpaper|primary|coal|tower|noise|cant|card|advantage|background|invalid|wrong|orgasm|babbitt|waitress|ballpark|carp|devil|cop|hot|major|negative|summer|winter|yip|bird|frog|window|web|cringe|beware|splat|bat|skipper|moon|sex|grass|horseplay|crowd|pain|season|wind|champ|friend|duck|institute|bunch|journey|diagram|clerk|photo|cake|wing|breast|course|nose|bread|certificate|cup|interface|cart|lumber|dish|massacre|crater|cement|stake|badge|crew|task|hole|stomach|doodle|weird|roadmap|mother|court|perch|gut|hat|architect|pit|square|bald|headquarter|mar|buffalo|pm|bellow|fret|scheme|cling">
+    <!ENTITY uncommon_verbs "ooh|key|chevy|neck|shark|chicken|suicide|dam|rice|twin|taboo|core|lullaby|ginger|patronage|caravan|buddy|guy|wive|van|green|toenail|pothole|yellow|safari|crap|bulldog|voodoo|minute|radio|fair|squirrel|ape|fox|badger|ferret|beagle|parrot|rat|arbitrage|privilege|deed|wolf|contango|coo|cat|tot|jut|bulletin|purpose|bitter|zinc|yuck|dot|valet|overnight|automobile|toilet|grandfather|mushroom|luck|paragraph|ambition|concept|field|baby|doll|flower|soldier|dude|volume|letter|id|pipeline|blackberry|bin|pretzel|turtle|panel|muscle|silver|warehouse|wood|machine|magic|clown|traffic|page|monkey|version|bundle|fan|till|dog|reason|garage|chapter|company|conference|chance|candy|room|console|margin|rabbit|article|cheese|while|yes|queen|gas|oil|officer|term|people|evidence|max|captain|case|step|minister|fun|water|excel|church|husband|dinner|police|curl|weed|platform|best|better|worst|twitter|prod|summit|calendar|chair|belt|mesh|curry|spam|sauce|lord|section|sheet|weekend|even|breakfast|tent|word|mad|mine|blue|eclipse|diesel|up|down|content|off|pretty|minor|well|physic|knight|dirty|tree|peter|harry|matt|mike|nick|bob|rick|sally|ralph|carol|war|shrimp|table|kit|gang|pen|soup|founder|proverb|jimmy|marcel|roger|wallpaper|primary|coal|tower|noise|cant|card|advantage|background|invalid|wrong|orgasm|babbitt|waitress|ballpark|carp|devil|cop|hot|major|negative|summer|winter|yip|bird|frog|window|web|cringe|beware|splat|bat|skipper|moon|sex|grass|horseplay|crowd|pain|season|wind|champ|friend|duck|institute|bunch|journey|diagram|clerk|photo|cake|wing|breast|course|nose|bread|certificate|cup|interface|cart|lumber|dish|massacre|crater|cement|stake|badge|crew|task|hole|stomach|doodle|weird|roadmap|mother|court|perch|gut|hat|architect|pit|square|bald|headquarter|mar|buffalo|pm|bellow|fret|scheme|cling|spider">
     <!ENTITY common_verbs_that_are_also_adjectives "live|work|record|open|close|strip|reverse|complete|select|switch|return|warm">
     <!ENTITY company_names_that_are_also_verbs "react|google|zoom|curl|facetime|excel|chevy|skype|photoshop|kindle|discover|aspire|velcro|intuit">
 
@@ -73914,6 +73914,33 @@ USA
                 <token regexp="yes">are|is|were|was</token>
                 <example>The only gas CES has are the Exhibit 1 volumes.</example>
             </antipattern>
+            <antipattern>
+                <token regexp="yes">that|which|who</token>
+                <token min="0">all</token>
+                <token chunk_re="E-NP.*" />
+                <token postag="RB" min="0" />
+                <token>have</token>
+                <token regexp="yes">is|was|were|are</token>
+            </antipattern>
+            <antipattern>
+                <token postag="NNS" />
+                <token>all</token>
+                <token regexp="yes">over|across</token>
+            </antipattern>
+            <antipattern>
+                <token inflected="yes">have</token>
+                <token regexp="yes">scars|clocks|cramps</token>
+            </antipattern>
+            <antipattern>
+                <token inflected="yes">have</token>
+                <token postag="NN:UN?" postag_regexp="yes" />
+                <token>any</token>
+                <token>more</token>
+            </antipattern>
+            <antipattern>
+                <token>is</token>
+                <token>that</token>
+            </antipattern>
             <rule>
                 <antipattern>
                     <token inflected="yes">have</token>
@@ -73931,12 +73958,12 @@ USA
                     <token min="0" regexp="yes">&apostrophe;</token>
                     <token postag="RB" min="0">
                         <exception postag="IN|EX" postag_regexp="yes" />
-                        <exception regexp="yes">nothin[g']?|yesterday|today|not|n't|much</exception>
+                        <exception regexp="yes">nothin[g']?|yesterday|today|not|n't|much|please</exception>
                     </token>
                     <marker>
                         <token postag="VB[ZPG]?" postag_regexp="yes">
                             <exception postag="VBN|RB|IN" postag_regexp="yes" regexp="yes">(?!forward$).+</exception><!-- forward #3377 -->
-                            <exception regexp="yes">breakfasts?|people|company|kids?|concepts?|times?</exception>
+                            <exception regexp="yes">breakfasts?|people|company|kids?|concepts?|times?|sex|is</exception>
                         </token>
                     </marker>
                     <token regexp="yes">
@@ -74034,7 +74061,7 @@ USA
                 <example>Who in legal do I have review the contract for acceptance?</example><!-- #3391 -->
                 <example>The copy of the confirm that I have HAS NO COMMENTS.</example><!-- #3391 -->
                 <example>I had cramps, a backache and headache.</example>
-                <example type="triggers_error">I started my period too, so I had cramps a backache and headache and...</example><!-- #3391, probably caused by #3758 -->
+                <example>I started my period too, so I had cramps a backache and headache and...</example><!-- #3391, probably caused by #3758 -->
                 <example type="triggers_error">I will find out from Donna how much of a problem, if any, they will have compiling the information.</example><!-- #3391 -->
                 <example>I knew I'd break his heart.</example>
             </rule>