[FOray-commit] SF.net SVN: foray:[14944] trunk/foray/foray-common/sr c

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

Revision: 14944
          http://sourceforge.net/p/foray/code/14944
Author:   victormote
Date:     2026-07-29 14:13:05 +0000 (Wed, 29 Jul 2026)
Log Message:
-----------
Allow 4-character language codes.

Modified Paths:
--------------
    trunk/foray/foray-common/src/main/java/org/foray/common/i18n/Language4a.java
    trunk/foray/foray-common/src/test/java/org/foray/common/i18n/Language4aTests.java

Modified: trunk/foray/foray-common/src/main/java/org/foray/common/i18n/Language4a.java
===================================================================
--- trunk/foray/foray-common/src/main/java/org/foray/common/i18n/Language4a.java	2026-07-29 10:22:22 UTC (rev 14943)
+++ trunk/foray/foray-common/src/main/java/org/foray/common/i18n/Language4a.java	2026-07-29 14:13:05 UTC (rev 14944)
@@ -109,13 +109,16 @@
     /** The length, in chars, of the 3-character code. */
     private static final int PATTERN_3CHAR_LENGTH = 3;
 
+    /** The length, in chars, of the 4-character code. */
+    private static final int PATTERN_4CHAR_LENGTH = 4;
+
     /** Regular expression pattern for matching a 2-character code. BCP 47 specifies that the language code is
      * case-insensitive. */
     private static final Pattern PATTERN_2CHAR = Pattern.compile("[A-Za-z]{2}");
 
-    /** Regular expression pattern for matching a 3-character code. BCP 47 specifies that the language code is
-     * case-insensitive. */
-    private static final Pattern PATTERN_3CHAR = Pattern.compile("[A-Za-z]{3}");
+    /** Regular expression pattern for matching a 3-character or 4-character ISO 639 code. BCP 47 specifies that the
+     * language code is case-insensitive. */
+    private static final Pattern PATTERN_3_OR_4CHAR = Pattern.compile("[A-Za-z]{3,4}");
 
     /** The initial capacity of the data structures. */
     private static final int INITIAL_CAPACITY = 600;
@@ -276,6 +279,8 @@
                 Script4a.GREEK));
         Language4a.register(new Language4a("elx", null, "Elamite", "élamite", null));
         Language4a.register(new Language4a("eng", "en", "English", "anglais", Script4a.LATIN));
+        /* The "emen" language code was part of the since-withdrawn ISO 639-6 proposed standard. */
+        Language4a.register(new Language4a("emen", null, "Early Modern English", "[unknown]", Script4a.LATIN));
         Language4a.register(new Language4a("enm", null, "English, Middle(1100-1500)", "anglais moyen(1100-1500)",
                 null));
         Language4a.register(new Language4a("epo", "eo", "Esperanto", "espéranto", null));
@@ -729,8 +734,8 @@
      */
     public Language4a(final String alpha3Code, final String alpha2Code, final String englishName,
             final String frenchName, final Script4a defaultScript) {
-        if (! matches3CharacterPattern(alpha3Code)) {
-            final String template = "Alpha 3 code expected pattern: " + PATTERN_3CHAR.toString() + ", actual: %s";
+        if (! matches3or4CharacterPattern(alpha3Code)) {
+            final String template = "Alpha 3 code expected pattern: " + PATTERN_3_OR_4CHAR.toString() + ", actual: %s";
             throw new IllegalArgumentException(String.format(template, alpha3Code));
         }
         /* Alpha 2 code can be null. */
@@ -889,19 +894,19 @@
     }
 
     /**
-     * Indicates whether a given string fits the pattern of a 3-character language code.
+     * Indicates whether a given string fits the pattern of a 3-character or 4-character language code.
      * This does NOT validate the language code, but only indicates that the pattern fits.
      * This is useful when parsing a string like "eng-USA-Latn" by helping to distinguish which part of the string could
      * be a valid language code.
      * @param testString The string to be tested, e.g. "eng" or "rus".
-     * @return True if and only if {@code testString} fits the pattern of a 3-character language code, i.e. that it
-     * consists of exactly two lowercase characters from the Latin alphabet.
+     * @return True if and only if {@code testString} fits the pattern of a 3-character or 4-character language code,
+     * i.e. that it consists of exactly three or four lowercase characters from the Latin alphabet.
      */
-    public static boolean matches3CharacterPattern(final CharSequence testString) {
+    public static boolean matches3or4CharacterPattern(final CharSequence testString) {
         if (testString == null) {
             return false;
         }
-        final Matcher matcher = PATTERN_3CHAR.matcher(testString);
+        final Matcher matcher = PATTERN_3_OR_4CHAR.matcher(testString);
         return matcher.matches();
     }
 
@@ -920,7 +925,8 @@
         }
         switch (testString.length()) {
         case PATTERN_2CHAR_LENGTH: return matches2CharacterPattern(testString);
-        case PATTERN_3CHAR_LENGTH: return matches3CharacterPattern(testString);
+        case PATTERN_3CHAR_LENGTH: return matches3or4CharacterPattern(testString);
+        case PATTERN_4CHAR_LENGTH: return matches3or4CharacterPattern(testString);
         default: return false;
         }
     }

Modified: trunk/foray/foray-common/src/test/java/org/foray/common/i18n/Language4aTests.java
===================================================================
--- trunk/foray/foray-common/src/test/java/org/foray/common/i18n/Language4aTests.java	2026-07-29 10:22:22 UTC (rev 14943)
+++ trunk/foray/foray-common/src/test/java/org/foray/common/i18n/Language4aTests.java	2026-07-29 14:13:05 UTC (rev 14944)
@@ -86,13 +86,16 @@
     @Test
     public void matches3CharacterPatternTests() {
         /* String should match. */
-        assertThat(Language4a.matches3CharacterPattern("aaa")).isTrue();
+        assertThat(Language4a.matches3or4CharacterPattern("aaa")).isTrue();
 
+        /* String should match. */
+        assertThat(Language4a.matches3or4CharacterPattern("aaaa")).isTrue();
+
         /* String too long. */
-        assertThat(Language4a.matches3CharacterPattern("aaaa")).isFalse();
+        assertThat(Language4a.matches3or4CharacterPattern("aaaaa")).isFalse();
 
         /* String too short. */
-        assertThat(Language4a.matches3CharacterPattern("aa")).isFalse();
+        assertThat(Language4a.matches3or4CharacterPattern("aa")).isFalse();
     }
 
 }

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