[de] fix casing rules (#11947)

* [de] fix UPPERCASE_SENTENCE_START and DE_CASE

* [de] fix test fail in casing rule

* [de] remove unnecessary parts in code

* [de] fix bereiten misread as adjective

* [de] other case fixes

* [de] remove invalid parts and add new abbreviation

* [de] fix parsing error

languagetool-core/src/main/resources/org/languagetool/resource/segment.srx

@@ -5164,7 +5164,7 @@
 </rule>
 <!-- Don't split at "bla bla... yada yada" -->
 <rule break="no">
-<beforebreak>[\[\(]?\.\.\.[\]\)]?[\u00A0\s]{1,2}</beforebreak>
+<beforebreak>[\[\(]?(\.\.\.|…)[\]\)]?[\u00A0\s]{1,2}</beforebreak>
 <afterbreak>\p{Ll}</afterbreak>
 </rule>
 <!-- Don't split [.?!] when they're quoted -->
@@ -5245,7 +5245,7 @@
 </rule>
 <!-- German abbreviations -->
 <rule break="no">
-<beforebreak>\b(betr|Geb|Stk|ggü|Mag|mtl|Flgh?|[Pp]arl|Bsp|versch|[Dd]iesbzgl|[Zz]ykl|[Dd]bzgl[Ss]tellv|d|Übers|[Bb]zw|Ab[hkst]|[Ee]ig|[Aa]bzü?gl|\d+-tlg|tlg|[Gg]gfls|[Ff]achspr|[Ll]tda|[Ee]inschl|[Vv]mtl|[Ss]tellv|Ev|[Bb]ezgl|lit|Abzw|[Vv]sl|ahd|Akk|aktual|[Öö]ffentl|prof|allg|alltagsspr|altdt|alttest|amerikan|Anh|Ank|Anm|Art|[Aa]utom|Auftragsnr|Az|Bat|bayr|Bde?|bearb|Bed|Bem|bes|bez|wsl|vsl|Bez|Bhf|Blvd|[Bb]spw|btto|bw|Dtl|[Gg]esetzl|[Ee]lektr|Dez|[Jj]gdfr|[Ee]ff)\.[\u00A0\s]{1,2}</beforebreak>
+<beforebreak>\b(betr|Geb|Stk|ggü|Mag|mtl|Flgh?|[Pp]arl|Bsp|versch|[Dd]iesbzgl|[Zz]ykl|[Dd]bzgl[Ss]tellv|[Ss]tv|d|Übers|[Bb]zw|Ab[hkst]|[Ee]ig|[Aa]bzü?gl|\d+-tlg|tlg|[Gg]gfls|[Ff]achspr|[Ll]tda|[Ee]inschl|[Vv]mtl|[Ss]tellv|Ev|[Bb]ezgl|lit|Abzw|[Vv]sl|ahd|Akk|aktual|[Öö]ffentl|prof|allg|alltagsspr|altdt|alttest|amerikan|Anh|Ank|Anm|Art|[Aa]utom|Auftragsnr|Az|Bat|bayr|Bde?|bearb|Bed|Bem|bes|bez|wsl|vsl|Bez|Bhf|Blvd|[Bb]spw|btto|bw|Dtl|[Gg]esetzl|[Ee]lektr|Dez|[Jj]gdfr|[Ee]ff|M)\.[\u00A0\s]{1,2}</beforebreak>
 <afterbreak></afterbreak>
 </rule>
 <rule break="no">
@@ -5257,6 +5257,11 @@
 <beforebreak>\b([Ee]tc)\.[\u00A0\s]{1,2}</beforebreak>
 <afterbreak>\p{Lu}</afterbreak>
 </rule>
+<!-- Don't split at "o. Ä." -->
+<rule break="no">
+  <beforebreak>\bo\.[\u00A0\s]{1,2}Ä\.</beforebreak>
+  <afterbreak></afterbreak>
+</rule>
 <rule break="no">
 <beforebreak>\b(cts?|[Cc]a|chem|chin|Chr|cresc|[Dd]at|desgl|ders|dgl|Dipl|Dir?|Doz?|durchg|durchges|Dr|[Dd]t|ebd|Ed|[Ee]igt?l|akt|[Ee]ngl|Erg|al|et[cw]|Etw|ev|[Ee]vtl?|[Ee]xkl|Expl|Exz)\.[\u00A0\s]{1,2}</beforebreak>
 <afterbreak></afterbreak>
@@ -5299,15 +5304,15 @@
 </rule>
 <!-- Break rules -->
 <rule break="yes">
-<beforebreak>[\.!?…][\u0002|'|"|“|«|‹|\)|\]|\}¹²³]?[\u00A0\s]+</beforebreak>
+<beforebreak>[\.!?][\u0002|'|"|“|«|‹|\)|\]|\}¹²³]?[\u00A0\s]+</beforebreak>
 <afterbreak></afterbreak>
 </rule>
 <rule break="yes">
-<beforebreak>[\.!?…]['"“\p{Pe}\u00BB\u201D]?</beforebreak>
+<beforebreak>[\.!?]['"“\p{Pe}\u00BB\u201D]?</beforebreak>
 <afterbreak>\p{Lu}[^\p{Lu}]</afterbreak>
 </rule>
 <rule break="yes">
-<beforebreak>[\u00A0\s]\p{L}[\.!?…][\u00A0\s]{1,2}</beforebreak>
+<beforebreak>[\u00A0\s]\p{L}[\.!?][\u00A0\s]{1,2}</beforebreak>
 <afterbreak>\p{Lu}\p{Ll}</afterbreak>
 </rule>
 <!-- z.B. 2 sentences: “Liebst du mich?” “Ja!” -->
@@ -6936,9 +6941,9 @@
 <beforebreak>[\.!?]</beforebreak>
 <afterbreak>\S*@</afterbreak>
 </rule>
-<!-- Some proper nouns: Yahoo!, Mission: Impossible-->
+<!-- Some proper nouns: Yahoo!, NAU!, Mission: Impossible-->
 <rule break="no">
-<beforebreak>Yahoo![\s\u00A0]</beforebreak>
+<beforebreak>(Yahoo|NAU)![\s\u00A0]</beforebreak>
 <afterbreak>\p{Ll}</afterbreak>
 </rule>
 <rule break="no">

languagetool-language-modules/de/src/main/java/org/languagetool/rules/de/CaseRule.java

@@ -83,12 +83,21 @@ public class CaseRule extends Rule {
 
   private static final String[] UNDEFINED_QUANTIFIERS = {"viel", "nichts", "nix", "wenig", "allerlei"};
 
+  // Used for subordinate clause detection after verbs (e.g. "Überlegen, wie man...")
+
   private static final String[] INTERROGATIVE_PARTICLES = {"was", "wodurch", "wofür", "womit", "woran", "worauf", "woraus", "wovon", "wie"};
 
   private static final String[] POSSESSIVE_INDICATORS = {"einer", "eines", "der", "des", "dieser", "dieses"};
 
   private static final String[] DAS_VERB_EXCEPTIONS = {"nur", "sogar", "auch", "die", "alle", "viele", "zu"};
 
+  // Used for short question fragments after a colon (e.g. "Ich frage mich: Warum?")
+  // Intentionally different from INTERROGATIVE_PARTICLES, which targets subordinate clauses.
+
+  private static final String[] COLON_QUESTION_WORDS = {"warum", "wieso", "weshalb", "wer", "was", "wann", "wo", "wie", "wozu"};
+
+  private static final String[] COLON_QUESTION_CONJUNCTIONS = {"und", "oder", "aber", "denn"};
+
   /*
    * These are words that Morphy only knows as non-nouns (or not at all).
    * The proper solution is to add all those to our Morphy data, but as a simple
@@ -1042,10 +1051,15 @@ public class CaseRule extends Rule {
         !isInvisibleSeparator(i-1, tokens) &&
         !language.getDefaultSpellingRule().isMisspelled(lcWord)) {
       if (":".equals(tokens[i - 1].getToken())) {
+        // allow short question sentences like "Warum? Und warum?" after colon
+        if (isQuestionEquivalentAfterColon(i, tokens)) {
+          return;
+        }
+
         AnalyzedTokenReadings[] subarray = new AnalyzedTokenReadings[i];
         System.arraycopy(tokens, 0, subarray, 0, i);
+
         if (isVerbFollowing(i, tokens, lowercaseReadings) || getTokensWithPosTagStartingWithCount(subarray, "VER") == 0) {
-          // no error
         } else {
           addRuleMatch(ruleMatches, sentence, COLON_MESSAGE, tokens[i], lcWord);
         }
@@ -1073,14 +1087,14 @@ public class CaseRule extends Rule {
 
   private boolean isNounWithVerbReading(int i, AnalyzedTokenReadings[] tokens) {
     return tokens[i].hasPosTagStartingWith("SUB") &&
-    		tokens[i].hasPosTagStartingWith("VER:INF");
-	}
+        tokens[i].hasPosTagStartingWith("VER:INF");
+  }
 
   private boolean isInvisibleSeparator(int i, AnalyzedTokenReadings[] tokens) {  // u2063 is used internally by our browser add-on
     return i >= 0 && i < tokens.length && tokens[i].getToken().length() > 0 && tokens[i].getToken().charAt(0) == '\u2063';
   }
 
-	private boolean isVerbFollowing(int i, AnalyzedTokenReadings[] tokens, AnalyzedTokenReadings lowercaseReadings) {
+  private boolean isVerbFollowing(int i, AnalyzedTokenReadings[] tokens, AnalyzedTokenReadings lowercaseReadings) {
     AnalyzedTokenReadings[] subarray = new AnalyzedTokenReadings[ tokens.length - i ];
     System.arraycopy(tokens, i, subarray, 0, subarray.length);
     if (lowercaseReadings != null) {
@@ -1091,6 +1105,10 @@ public class CaseRule extends Rule {
     return getTokensWithPosTagStartingWithCount(subarray, "VER:") != 0;
 }
 
+  private boolean isColonQuestionWord(String word) {
+    return StringUtils.equalsAnyIgnoreCase(word, COLON_QUESTION_WORDS);
+  }
+
   private void addRuleMatch(List<RuleMatch> ruleMatches, AnalyzedSentence sentence, String msg, AnalyzedTokenReadings tokenReadings, String fixedWord) {
     RuleMatch ruleMatch = new RuleMatch(this, sentence, tokenReadings.getStartPos(), tokenReadings.getEndPos(), msg);
     ruleMatch.setSuggestedReplacement(fixedWord);
@@ -1295,6 +1313,33 @@ public class CaseRule extends Rule {
     return true;
   }
 
+  /**
+   * Returns true if the token at position i starts a short question fragment
+   * that forms a valid Satzäquivalent after a colon, per Duden rules.
+   * Handles:
+   *   single word:          "Ich frage mich: Warum?"
+   *   with conjunction:     "Ich frage mich: Und warum?"
+   */
+  @VisibleForTesting
+  boolean isQuestionEquivalentAfterColon(int i, AnalyzedTokenReadings[] tokens) {
+    if (i < tokens.length - 1) {
+      String word = tokens[i].getToken();
+      String next = tokens[i + 1].getToken();
+      // "Warum?"
+      if (isColonQuestionWord(word) && "?".equals(next)) {
+        return true;
+      }
+      // "Und warum?"
+      if (StringUtils.equalsAnyIgnoreCase(word, COLON_QUESTION_CONJUNCTIONS)
+          && i < tokens.length - 2
+          && isColonQuestionWord(tokens[i + 1].getToken())
+          && "?".equals(tokens[i + 2].getToken())) {
+        return true;
+      }
+    }
+    return false;
+  }
+
   private AnalyzedTokenReadings lookup(String word) {
     try {
       return ((GermanTagger) language.getTagger()).lookup(word);
@@ -1303,3 +1348,4 @@ public class CaseRule extends Rule {
     }
   }
 }
+

languagetool-language-modules/de/src/main/resources/org/languagetool/resource/de/case_rule_exceptions.txt

@@ -50,6 +50,7 @@ Alpine[ns] Museums? (der|Münchens?)
 Agile Coach(e?|ing)s?
 Amnesty Internationals?
 Andreas Scheuers?
+Ankleiden?
 Anlage Sonstiges
 Angewandte[mnr]? Kunst
 Angewandte[mnr]? [A-ZÖÄÜ][a-zäöü]+(wissenschaft|forschung)(en)?
@@ -59,6 +60,7 @@ Angewandte[mnr]? (Optik|Informatik|Ethik|Geophysik|Physik|Makroökonomie|Linguis
 Anne Will
 Anarchistischen? Pogo-Partei
 Anthony Blinkens?
+Annies?
 Apokalyptische[nrm]? Reiters?
 Apostolische[nrm]? Palast(s|es)?
 Aqua Vitals?
@@ -195,6 +197,7 @@ Director Legal
 Director of Legal
 Die Toten Hosen
 Die Leiden des jungen Werthers
+Die Mumie
 [Dd]ie Schöne und das Biest
 Digital Assets?|IQ|Garage|Pushs?|Traces?|Pushing|Cockpits?|Ads?|Signals?|Designs?|Designer[ns]?|Engineerings?|Innovation|Media|Content|Community|Detox|Analyst|Leader[ns]?|Consultants?|Equipments?|Onboarding|Transformations?|Learnings?|Events?|Hubs?|Masterings?|Marketers?|Radios?|Experiences?|Kits?|Wallets?|Turbines?|Workspaces?|Leaderships?|Profiles?|News?|Managements?|Economy|Summits?|Fitness
 Digital Film Productions?
@@ -549,11 +552,13 @@ Kaukasische[nsm] Eichhörnchens?
 Kubanischen? Revolution
 kaum Neues
 Kessel Buntes
+KfW(-\d+)?(-[A-ZÄÖÜa-zäöüß]+)?
 Kitchen Impossible
 Klein (Nordendes?|Borstels?)
 Kleine[rnm]? Sauerampfers?
 Kleinen? Weser|Donau
 Knollige[rnm]? Hahnenfu(ß|ss)
+Kommunale[nms]? Integrationszentrum
 Lost (Places?|Islands?)
 Lost and Found
 Landesverband(e?s)? Baye?rischer? Omnibusunternehmen
@@ -772,6 +777,7 @@ Polar 8
 Potemkinschen? Dörfern?
 Politische[rn]? Direktors?
 Polytechnischen? (.*schule|Instituts?)
+Polytechnische[nr]? Universität
 Polnische[rnm]? (Korridors?|Fremdenverkehrsamt(e?s)?)
 Praktische[mns]? Jahr
 Present Progressive
@@ -782,6 +788,7 @@ Prima Terras?
 Product[/-]Market Fit
 Progressive Web Apps?
 Progressive Rocks?|Metals?
+ProMinent Dosiertechnik AG
 Psychologie Heute
 Psychologische[nsm]? Institute?s?
 Psychiatrische[rn]? Klinik ([A-ZÖÄÜ][a-zäöüß\-]+)
@@ -847,6 +854,7 @@ Schiefe[nr]? Turms?
 Schiefen? Ebene
 Schiefen? Schlachtordnung
 Schlesische[rnm]? Kriege?s?
+Schlesische[rn]? Universität
 Schlesische[snm]? Tors?
 Schnellen? Eingreiftruppe
 Schön Klinik
@@ -866,6 +874,7 @@ Schweizerischen? Konferenz
 Schweizerische[nrm]? Verband für
 Schweizerische[nrm]? Gehörlosenbund
 Schweizerische[nrm]? Fu(ss|ß)ballverband
+Schweizerische[rn]? Trassenvergabestelle(-[A-ZÄÖÜ][a-zäöüß]+)?
 Schweizer Illustrierten?
 Schweizerischen? Akademischen? Gesellschaft für Germanistik
 Seiner? Exzellenz
@@ -956,6 +965,7 @@ Theologischen? Fakultät
 Thüringischen? Landeszeitung
 Thüringer Allgemeinen?
 Thorsten Lahm
+THYMUSKIN Sensitive Shampoo
 Tiger Global
 Tote[nsm] Meer
 Total Gross|Values?|Blocking
@@ -1011,6 +1021,7 @@ Wei(ß|ss)e[nrm]? Zwerg(e|s|en|es)?|Sonntags?|Tod|Rings?|Burgunder[ns]?
 (WELT|Welt) Digital
 Wenn und Aber
 Werd Verlag
+Western Union International Bank
 Westliche[nrm]? Gorillas?
 Westfälische[nrm]? Friede(ns?)?|Pforte|Bucht|Nachrichten|Wilhelms-Universität|Hochschule|Anzeigers?
 Wilde[nrm]? Westens?|Kaisers?|Berge?s?

languagetool-language-modules/de/src/main/resources/org/languagetool/resource/de/disambiguation.xml

@@ -1327,6 +1327,11 @@ Copyright © 2013 Markus Brenneis, Daniel Naber, Jan Schreiber
             <token postag="VER:1:PLU.*" postag_regexp="yes" />
             <token postag="SUB:.*" postag_regexp="yes" />
         </antipattern>
+        <antipattern><!-- die Höhe bereitet/bereiten Kopfzerbrechen -->
+            <token postag="SUB:.*" postag_regexp="yes"/>
+            <token regexp="yes">bereitet?|bereiten</token>
+            <token postag="SUB:.*" postag_regexp="yes"/>
+        </antipattern>
         <pattern>
             <unify>
                 <feature id="number"/><feature id="case"/><feature id="gender"/>

languagetool-language-modules/de/src/test/java/org/languagetool/rules/de/CaseRuleTest.java

@@ -383,6 +383,14 @@ public class CaseRuleTest {
     assertBad("Das ist es: Kein Satz.");
     assertBad("Wen magst du lieber: Die Giants oder die Dragons?");
 
+    assertGood("Ich frage mich: Warum?");
+    assertGood("Ich frage mich: Wieso?");
+    assertGood("Ich frage mich: Weshalb?");
+    assertGood("Ich frage mich: Und warum?");
+    assertGood("Ich frage mich: Oder wieso?");
+    assertGood("Ich frage mich: Aber warum?");
+    assertBad("Ich frage mich: Warum Das so ist.");
+
     assertGood("Das wirklich Wichtige ist dies:");
     assertGood("Das wirklich wichtige Verfahren ist dies:");
     //assertBad("Das wirklich wichtige ist dies:");

languagetool-language-modules/de/src/test/java/org/languagetool/rules/de/UppercaseSentenceStartRuleTest.java

@@ -79,6 +79,9 @@ public class UppercaseSentenceStartRuleTest {
     assertEquals(1, matches3.size());
     assertEquals(19, matches3.get(0).getFromPos());
     assertEquals(22, matches3.get(0).getToPos());
+
+    assertEquals(0, lt.check("Willkommen… im Berlin.").size());
+    assertEquals(0, lt.check("Die neue Kollektion von NAU! ist jetzt online.").size());
   }
 
 }