[FOray-commit] SF.net SVN: foray:[12061] trunk/foray/foray-orthography/src/main

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

Revision: 12061
          http://sourceforge.net/p/foray/code/12061
Author:   victormote
Date:     2021-11-14 19:46:18 +0000 (Sun, 14 Nov 2021)
Log Message:
-----------
Allow Punctuation4a to silently return a null. Fix some minor side-effects of that.

Modified Paths:
--------------
    trunk/foray/foray-orthography/src/main/data/dictionaries/eng-999-Latn.dict.xml
    trunk/foray/foray-orthography/src/main/java/org/foray/orthography/LexerLatin1.java
    trunk/foray/foray-orthography/src/main/java/org/foray/orthography/Punctuation4a.java

Modified: trunk/foray/foray-orthography/src/main/data/dictionaries/eng-999-Latn.dict.xml
===================================================================
--- trunk/foray/foray-orthography/src/main/data/dictionaries/eng-999-Latn.dict.xml	2021-11-14 16:56:39 UTC (rev 12060)
+++ trunk/foray/foray-orthography/src/main/data/dictionaries/eng-999-Latn.dict.xml	2021-11-14 19:46:18 UTC (rev 12061)
@@ -53,6 +53,7 @@
 # web site, or submit a patch request.
 -->
 
+<w><t>&amp;</t></w>
 <w><t>a</t></w>
 <phrase><t>a cap-pel-la</t></phrase>
 <phrase><t>a for-ti-o-ri</t></phrase>

Modified: trunk/foray/foray-orthography/src/main/java/org/foray/orthography/LexerLatin1.java
===================================================================
--- trunk/foray/foray-orthography/src/main/java/org/foray/orthography/LexerLatin1.java	2021-11-14 16:56:39 UTC (rev 12060)
+++ trunk/foray/foray-orthography/src/main/java/org/foray/orthography/LexerLatin1.java	2021-11-14 19:46:18 UTC (rev 12061)
@@ -39,6 +39,7 @@
     public CharType isWordChar(final char c) {
         switch (c) {
         case '-': return CharType.ALWAYS_WORD_CHAR;
+        case '&': return CharType.ALWAYS_WORD_CHAR;
         default: return super.isWordChar(c);
         }
     }

Modified: trunk/foray/foray-orthography/src/main/java/org/foray/orthography/Punctuation4a.java
===================================================================
--- trunk/foray/foray-orthography/src/main/java/org/foray/orthography/Punctuation4a.java	2021-11-14 16:56:39 UTC (rev 12060)
+++ trunk/foray/foray-orthography/src/main/java/org/foray/orthography/Punctuation4a.java	2021-11-14 19:46:18 UTC (rev 12061)
@@ -39,8 +39,6 @@
 import org.axsl.common.para.ParaPenalty;
 import org.axsl.orthography.Punctuation;
 
-import org.slf4j.LoggerFactory;
-
 import java.util.ArrayList;
 import java.util.List;
 
@@ -108,8 +106,17 @@
     public static final Punctuation4a CLOSE_SQUARE_BRACKET = new Punctuation4a("]");
 
     /** Common punctuation item. */
+    public static final Punctuation4a FORWARD_SLASH = new Punctuation4a("/");
+
+    /** Common punctuation item. */
+    public static final Punctuation4a ELLIPSIS = new Punctuation4a("\u2026");
+
+    /** Common punctuation item. */
     public static final Punctuation4a ZERO_WIDTH_SPACE = new Punctuation4a("\u200B");
 
+    /** Common punctuation item. */
+    public static final Punctuation4a ZERO_WIDTH_NO_BREAK_SPACE = new Punctuation4a("\uFEFF");
+
     /** The list of common interword items. */
     private static final List<Punctuation4a> COMMON_PUNCTUATION = new ArrayList<Punctuation4a>(25);
     static {
@@ -133,7 +140,10 @@
         COMMON_PUNCTUATION.add(CLOSE_PAREN);
         COMMON_PUNCTUATION.add(OPEN_SQUARE_BRACKET);
         COMMON_PUNCTUATION.add(CLOSE_SQUARE_BRACKET);
+        COMMON_PUNCTUATION.add(FORWARD_SLASH);
+        COMMON_PUNCTUATION.add(ELLIPSIS);
         COMMON_PUNCTUATION.add(ZERO_WIDTH_SPACE);
+        COMMON_PUNCTUATION.add(ZERO_WIDTH_NO_BREAK_SPACE);
 
     }
 
@@ -153,7 +163,7 @@
      * Using this method instead of a constructor allows this class to reduce unnecessary object creation be reusing
      * existing instances of this class instead of creating identical ones.
      * @param content The content of the interword item.
-     * @return An instance of this class wrapping the given content.
+     * @return An instance of this class wrapping the given content, or null if no instance is found.
      */
     public static Punctuation4a findInstance(final CharSequence content) {
         for (int index = 0; index < COMMON_PUNCTUATION.size(); index ++) {
@@ -162,8 +172,7 @@
                 return punctuation;
             }
         }
-        LoggerFactory.getLogger(Punctuation4a.class).warn("Standard punctuation not found: [" + content + "]");
-        return new Punctuation4a(content);
+        return null;
     }
 
     @Override

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