[FOray-commit] SF.net SVN: foray:[12678] trunk/foray/foray-orthography/src/main/java/org/ foray/orthography/util/SpellChecker.java

[<vic...@us...>]

Revision: 12678
          http://sourceforge.net/p/foray/code/12678
Author:   victormote
Date:     2022-06-21 16:24:03 +0000 (Tue, 21 Jun 2022)
Log Message:
-----------
Process text at the end of terminal elements, to avoid the appearance of continuity of words between them.

Modified Paths:
--------------
    trunk/foray/foray-orthography/src/main/java/org/foray/orthography/util/SpellChecker.java

Modified: trunk/foray/foray-orthography/src/main/java/org/foray/orthography/util/SpellChecker.java
===================================================================
--- trunk/foray/foray-orthography/src/main/java/org/foray/orthography/util/SpellChecker.java	2022-06-21 13:44:29 UTC (rev 12677)
+++ trunk/foray/foray-orthography/src/main/java/org/foray/orthography/util/SpellChecker.java	2022-06-21 16:24:03 UTC (rev 12678)
@@ -149,6 +149,9 @@
     /** The list of elements having no content but that can be placed in the middle of a word. */
     private List<String> elementIgnoreList = Arrays.asList(new String[] {"Page", "MendOut", "Comment", "ToDo"});
 
+    /** The list of elements that should never be straddled by a word. */
+    private List<String> elementTerminalList = Arrays.asList(new String[] {"Para", "Sidenote"});
+
     /** Map whose key is elements that are actually tokens for text, and whose value is that text. */
     private Map<String, String> textElementMap = new HashMap<String, String>();
     {
@@ -248,8 +251,10 @@
         final WritingSystem4a oldWritingSystem = getCurrentWritingSystem();
         final WritingSystem4a newWritingSystem = parseWritingSystem(attributes);
         element.writingSystem = newWritingSystem;
-        checkWords(oldWritingSystem, newWritingSystem);
 
+        if (! ObjectUtils.safeEquals(oldWritingSystem, newWritingSystem)) {
+            checkWords(oldWritingSystem);
+        }
         this.elementStack.push(element);
     }
 
@@ -278,7 +283,15 @@
         }
 
         final WritingSystem4a newWritingSystem = getCurrentWritingSystem();
-        checkWords(oldWritingSystem, newWritingSystem);
+
+        if (! ObjectUtils.safeEquals(oldWritingSystem, newWritingSystem)) {
+            checkWords(oldWritingSystem);
+        }
+
+        /* Whether the writing system has changed or not, if we are at the end of a terminal element, process text. */
+        if (this.elementTerminalList.contains(localName)) {
+            checkWords(oldWritingSystem);
+        }
     }
 
 
@@ -285,20 +298,12 @@
     /**
      * Check for a change in writing system, and, if there is one, flush the text accumulator and spell-check the words
      * in it.
-     * @param oldWritingSystem The old writing system, which will be used to spell-check the accumulated text if the
-     * writing system has changed.
-     * @param newWritingSystem The new writing system.
+     * @param writingSystem The writing system to be used to spell-check the accumulated text.
      */
-    private void checkWords(final WritingSystem4a oldWritingSystem, final WritingSystem4a newWritingSystem) {
-        if (ObjectUtils.safeEquals(oldWritingSystem, newWritingSystem)) {
-            /* Either both are null or they are equal. */
-            return;
-        }
+    private void checkWords(final WritingSystem4a writingSystem) {
+        final Orthography4a orthography = writingSystem == null ? null : this.server.getOrthography(writingSystem);
+        final Lexer lexer = orthography == null ? new LexerJavaBreakIterator(writingSystem) : orthography.getLexer();
 
-        final Orthography4a orthography = oldWritingSystem == null ?
-                null : this.server.getOrthography(oldWritingSystem);
-        final Lexer lexer = orthography == null ? new LexerJavaBreakIterator(oldWritingSystem) : orthography.getLexer();
-
         final List<CharSequence> words = lexer.tokenize(getAndClearText());
 
         if (words == null) {

This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.