Fix for : Laboratuvar'ı kelimesini hatalı olarak işaretliyor. #16

COMU · Feb 22, 2019 · 5b83f59 · 5b83f59
1 parent ce875a9
commit 5b83f59
Show file tree

Hide file tree

Showing 3 changed files with 57 additions and 5 deletions.
diff --git a/libreoffice-tr-tools/lib/runtime/zemberek-lo.jar b/libreoffice-tr-tools/lib/runtime/zemberek-lo.jar
diff --git a/libreoffice-tr-tools/src/lo/tr/tools/spellchecker/ZemberekSpellChecker.java b/libreoffice-tr-tools/src/lo/tr/tools/spellchecker/ZemberekSpellChecker.java
@@ -1,5 +1,6 @@
 package lo.tr.tools.spellchecker;
 
+import _zem.org.antlr.v4.runtime.Token;
 import java.io.IOException;
 import java.util.ArrayList;
 import java.util.Collections;
@@ -19,6 +20,7 @@
 import zemberek.morphology.generator.WordGenerator;
 import zemberek.morphology.lexicon.RootLexicon;
 import zemberek.normalization.TurkishSpellChecker;
+import zemberek.tokenization.TurkishTokenizer;
 
 public class ZemberekSpellChecker {
 
@@ -60,7 +62,7 @@ private ZemberekSpellChecker(TurkishMorphology morphology) {
   /**
    * This is used for debugging purposes.
    */
-  static ZemberekSpellChecker getInstance(RootLexicon lexicon) {
+  static synchronized ZemberekSpellChecker getInstance(RootLexicon lexicon) {
     TurkishMorphology morphology = TurkishMorphology.builder()
         .setLexicon(lexicon)
         .useInformalAnalysis().build();
@@ -87,7 +89,20 @@ public boolean isCorrect(String w) {
         return true;
       }
     }
-    return spellChecker.check(input);
+
+    boolean passed = spellChecker.check(input);
+    if (passed) {
+      return true;
+    }
+
+    List<Token> tokens = TurkishTokenizer.DEFAULT.tokenize(w);
+    if (tokens.size() != 1) {
+      return false;
+    }
+    Token t = tokens.get(0);
+    List<SingleAnalysis> analyses = morphology.getUnidentifiedTokenAnalyzer().analyze(t);
+    return analyses.size() > 0;
+
   }
 
   public List<String> getSuggestions(String s) {
@@ -132,10 +147,10 @@ private List<String> informalWordSuggestions(String s) {
   }
 
   private String getApostrophe(String input) {
-    String apostrophe;
+    String apostrophe = null;
     if (input.indexOf('’') > 0) {
       apostrophe = "’";
-    } else {
+    } else if (input.indexOf('\'') > 0) {
       apostrophe = "'";
     }
     return apostrophe;

diff --git a/libreoffice-tr-tools/test/lo/tr/tools/spellchecker/ZemberekSpellCheckTest.java b/libreoffice-tr-tools/test/lo/tr/tools/spellchecker/ZemberekSpellCheckTest.java
@@ -1,5 +1,6 @@
 package lo.tr.tools.spellchecker;
 
+import java.util.List;
 import org.junit.Assert;
 import org.junit.Test;
 import zemberek.morphology.lexicon.RootLexicon;
@@ -24,7 +25,8 @@ public void testSuggestions1() {
     ZemberekSpellChecker instance = ZemberekSpellChecker.getInstance(lexicon);
     String[] inputs = {"ou", "ku", "okuu", "oyu"};
     for (String input : inputs) {
-      Assert.assertTrue(instance.getSuggestions(input).contains("oku"));
+      List<String> suggestions = instance.getSuggestions(input);
+      Assert.assertTrue(suggestions.contains("oku"));
     }
   }
 
@@ -44,5 +46,40 @@ public void testInformalWords1() {
     Assert.assertTrue(instance.getSuggestions("gidicem").contains("gideceğim"));
   }
 
+  @Test
+  public void testRegularWordWithApostrophe() {
+    RootLexicon lexicon = RootLexicon.fromLines("kitap");
+
+    ZemberekSpellChecker instance = ZemberekSpellChecker.getInstance(lexicon);
+    Assert.assertTrue(instance.isCorrect("kitapta"));
+    Assert.assertTrue(instance.isCorrect("Kitapta"));
+    Assert.assertTrue(instance.isCorrect("Kitap'ta"));
+
+    // expect false.
+    Assert.assertFalse(instance.isCorrect("Kitap'taa"));
+  }
+
+  @Test
+  public void testRegularWordWithApostropheSuggestions() {
+    RootLexicon lexicon = RootLexicon.fromLines("kitap");
+
+    ZemberekSpellChecker instance = ZemberekSpellChecker.getInstance(lexicon);
+    Assert.assertTrue(instance.getSuggestions("ktapta").contains("kitapta"));
+    Assert.assertTrue(instance.getSuggestions("Ktapta").contains("Kitapta"));
+    Assert.assertTrue(instance.getSuggestions("Ktap'ta").contains("Kitap'ta"));
+  }
+
+  @Test
+  public void testRegularWordWithApostrophesIssue16() {
+    RootLexicon lexicon = RootLexicon.fromLines("laboratuvar");
+
+    ZemberekSpellChecker instance = ZemberekSpellChecker.getInstance(lexicon);
+    Assert.assertTrue(instance.isCorrect("laboratuvar"));
+    Assert.assertTrue(instance.isCorrect("laboratuvarda"));
+    Assert.assertTrue(instance.isCorrect("Laboratuvarda"));
+    Assert.assertTrue(instance.isCorrect("Laboratuvar'da"));
+  }
+
+
 
 }