Fixing SpellExclusion

luwrain · Jun 23, 2022 · 635f30e · 635f30e
1 parent dc5c527
commit 635f30e
Show file tree

Hide file tree

Showing 3 changed files with 22 additions and 4 deletions.
diff --git a/src/main/java/org/luwrain/nlp/SpellChecker.java b/src/main/java/org/luwrain/nlp/SpellChecker.java
@@ -18,9 +18,9 @@
 
 import java.util.*;
 
-
 public interface SpellChecker
 {
  List<SpellProblem> check(String text);
  List<String> suggestCorrections(String word);
+ SpellExclusion getExclusion();
 }
diff --git a/src/main/java/org/luwrain/nlp/SpellText.java b/src/main/java/org/luwrain/nlp/SpellText.java
@@ -28,12 +28,14 @@ public class SpellText
 static public final String
  LOG_COMPONENT = "spelling";
 
+ final SpellExclusion exclusion;
  final String text;
  final List<Fragment> fragments = new ArrayList<>();
  final List<SpellProblem> problems;
 
  public SpellText(String[] text, SpellChecker checker)
  {
+ this.exclusion = checker.getExclusion();
  if (text.length == 0)
  {
  this.text = "";
@@ -93,12 +95,21 @@ public List<SpellProblem> getProblems()
 
  private List<SpellProblem> filterExclusions(String text, List<SpellProblem> source)
  {
+ if (this.exclusion == null)
+ return source;
+ //FIXME: Better to do this on class creation
+ final Set<String> wordsToExclude = new HashSet<>();
+ for(SpellExclusion.Exclusion e: exclusion.getExclusions())
+ {
+ final String word = e.getText().trim();
+ if (!word.isEmpty())
+ wordsToExclude.add(word.toUpperCase());
+ }
  final List<SpellProblem> res = new ArrayList<>();
  for(SpellProblem p: source)
  {
  final String f = text.substring(p.getStart(), p.getEnd()).toUpperCase();
- Log.debug("proba", "fragment " + f);
- if (f.equals(""))
+ if (wordsToExclude.contains(f))
  continue;
  res.add(p);
  }

diff --git a/src/main/java/org/luwrain/nlp/ru/spell/RuSpellChecker.java b/src/main/java/org/luwrain/nlp/ru/spell/RuSpellChecker.java
@@ -31,6 +31,7 @@ public final class RuSpellChecker implements SpellChecker
 {
  final JLanguageTool langTool;
  final Hunspell hunspell;
+ final SpellExclusion exclusion;
 
  public RuSpellChecker(Luwrain luwrain)
  {
@@ -40,13 +41,16 @@ public RuSpellChecker(Luwrain luwrain)
  dictFile = new File(hunspellDataDir, "ru_RU.dic"),
  affFile = new File(hunspellDataDir, "ru_RU.aff");
  this.hunspell = new Hunspell(dictFile.getAbsolutePath(), affFile.getAbsolutePath());
+ this.exclusion = new SpellExclusion(luwrain);
+ this.exclusion.load();
  }
 
  //Just for unit tests
 RuSpellChecker()
  {
  this.langTool = new JLanguageTool(new Russian());
  this.hunspell = null;
+ this.exclusion = null;
  }
 
  @Override public List<SpellProblem> check(String text)
@@ -71,5 +75,8 @@ public RuSpellChecker(Luwrain luwrain)
 return hunspell.suggest(word);
  }
 
-
+ @Override public SpellExclusion getExclusion()
+ {
+ return exclusion;
+ }
 }