From 9ca048ef8ded3074bc4004dae28af95971e8a2fd Mon Sep 17 00:00:00 2001
From: mark-baas <mark.baas@languagetool.com>
Date: Fri, 27 Oct 2023 17:00:59 +0700
Subject: [PATCH] [nl] improve CompoundAcceptor

---
 .../languagetool/rules/nl/CompoundAcceptor.java | 15 +++++++--------
 .../rules/nl/CompoundAcceptorTest.java          | 17 ++++++++---------
 2 files changed, 15 insertions(+), 17 deletions(-)

diff --git a/languagetool-language-modules/nl/src/main/java/org/languagetool/rules/nl/CompoundAcceptor.java b/languagetool-language-modules/nl/src/main/java/org/languagetool/rules/nl/CompoundAcceptor.java
index e80162a1bc95..89c689e92f14 100644
--- a/languagetool-language-modules/nl/src/main/java/org/languagetool/rules/nl/CompoundAcceptor.java
+++ b/languagetool-language-modules/nl/src/main/java/org/languagetool/rules/nl/CompoundAcceptor.java
@@ -38,13 +38,12 @@
 public class CompoundAcceptor {
 
   // compound parts that need an 's' appended to be used as first part of the compound:
-  // "teit", "ing", "heid", "schap", "ker"
-  private final Set<String> needsS1 = new HashSet<>(Arrays.asList(
-    "bedrijfs", "passagiers" //, "dorps", "gezichts", "lijdens", "besturings", "verbrandings", "bestemmings", "schoonheids"
+  private final Set<String> needsS = new HashSet<>(Arrays.asList(
+    "bedrijfs", "passagiers", "dorps", "gezichts", "lijdens", "besturings", "verbrandings", "bestemmings", "schoonheids"
   ));
   // compound parts that must not have an 's' appended to be used as first part of the compound:
-  private final Set<String> noS1 = new HashSet<>(Arrays.asList(
-    "sport", "woning" //, "kinder", "fractie", "zout", "schade", "energie", "gemeente", "dienst", "wereld", "telefoon", "winkel", "aandeel", "zwanger", "papier"
+  private final Set<String> noS = new HashSet<>(Arrays.asList(
+    "woning", "kinder", "fractie", "schade", "energie", "gemeente", "dienst", "wereld", "telefoon", "aandeel", "zwanger", "papier"
   ));
 
   private final MorfologikDutchSpellerRule speller;
@@ -79,11 +78,11 @@ boolean acceptCompound(String part1, String part2) throws IOException {
     boolean okWithDash = false;
     boolean okWithoutS = false;
     if (part1.endsWith("s")) {
-      okWithS = spellingOk(part1.replaceFirst("s$", "")) && spellingOk(part2) && needsS1.contains(part1);
+      okWithS = spellingOk(part1.replaceFirst("s$", "")) && spellingOk(part2) && needsS.contains(part1.toLowerCase());
     } else if ( part1.endsWith("-")) {
       okWithDash = abbrevOk(part1) && spellingOk(part2);
     } else {
-      okWithoutS = spellingOk(part1) && spellingOk(part2) && noS1.contains(part1);
+      okWithoutS = spellingOk(part1) && spellingOk(part2) && noS.contains(part1.toLowerCase());
     }
     //System.out.println(" okWithS: " + okWithS + ", okWithoutS " + okWithoutS);
     return okWithS || okWithDash || okWithoutS;
@@ -96,7 +95,7 @@ private boolean abbrevOk(String nonCompound) {
 
   private boolean spellingOk(String nonCompound) throws IOException {
     AnalyzedSentence as = new AnalyzedSentence(new AnalyzedTokenReadings[] {
-      new AnalyzedTokenReadings(new AnalyzedToken(nonCompound, "FAKE_POS", "fakeLemma"))
+      new AnalyzedTokenReadings(new AnalyzedToken(nonCompound.toLowerCase(), "FAKE_POS", "fakeLemma"))
     });
     RuleMatch[] matches = speller.match(as);
     return matches.length == 0;
diff --git a/languagetool-language-modules/nl/src/test/java/org/languagetool/rules/nl/CompoundAcceptorTest.java b/languagetool-language-modules/nl/src/test/java/org/languagetool/rules/nl/CompoundAcceptorTest.java
index 401b2e1aabc6..94ed8f2fe72b 100644
--- a/languagetool-language-modules/nl/src/test/java/org/languagetool/rules/nl/CompoundAcceptorTest.java
+++ b/languagetool-language-modules/nl/src/test/java/org/languagetool/rules/nl/CompoundAcceptorTest.java
@@ -37,25 +37,24 @@ public void testAcceptCompound() throws IOException {
     assertTrue(acceptor.acceptCompound("VRF-regels"));
     assertFalse(acceptor.acceptCompound("VRFregels"));
 
-    //assertTrue(acceptor.acceptCompound("winkeldiefstal"));
-    //assertFalse(acceptor.acceptCompound("winkelsdiefstal"));
-
     assertTrue(acceptor.acceptCompound("bedrijfsregels"));
     assertFalse(acceptor.acceptCompound("bedrijfregels"));
 
-    //assertTrue(acceptor.acceptCompound("zwangerschap"));
-    //assertFalse(acceptor.acceptCompound("zwangersschap"));
+    assertTrue(acceptor.acceptCompound("Bedrijfsbrommer"));
+    assertFalse(acceptor.acceptCompound("Bedrijfbrommer"));
+
+    assertTrue(acceptor.acceptCompound("zwangerschap"));
+    assertFalse(acceptor.acceptCompound("zwangersschap"));
 
-    //assertTrue(acceptor.acceptCompound("papierversnipperaar"));
-    //assertFalse(acceptor.acceptCompound("papiersversnipperaar"));
+    assertTrue(acceptor.acceptCompound("Papierversnipperaar"));
+    assertFalse(acceptor.acceptCompound("Papiersversnipperaar"));
   }
 
   @Test
   public void testAcceptCompoundInternal() throws IOException {
     CompoundAcceptor acceptor = new CompoundAcceptor();
     assertTrue(acceptor.acceptCompound("passagiers", "schip"));
-    //assertTrue(acceptor.acceptCompound("papier", "versnipperaar"));
-    //assertTrue(acceptor.acceptCompound("winkel", "diefstal"));
+    assertTrue(acceptor.acceptCompound("papier", "versnipperaar"));
   }
 
 }
\ No newline at end of file