add iso8859-9 caseMap

jruby · Apr 18, 2018 · e7f99c8 · e7f99c8
1 parent 6f27648
commit e7f99c8
Show file tree

Hide file tree

Showing 2 changed files with 52 additions and 1 deletion.
diff --git a/src/org/jcodings/specific/ISO8859_3Encoding.java b/src/org/jcodings/specific/ISO8859_3Encoding.java
@@ -62,7 +62,7 @@ else if ((ISO8859_3CtypeTable[code] & CharacterType.BIT_UPPER) != 0 && (flags &
             } else if ((ISO8859_3CtypeTable[code] & CharacterType.BIT_LOWER) != 0 && (flags & Config.CASE_UPCASE) != 0) {
                 flags |= Config.CASE_MODIFIED;
                 if (code == 'i') {
-                    code = (flags & Config.CASE_FOLD_TURKISH_AZERI) != 0 ? DOTLESS_i : 'I';
+                    code = (flags & Config.CASE_FOLD_TURKISH_AZERI) != 0 ? I_WITH_DOT_ABOVE : 'I';
                 } else if (code == DOTLESS_i) {
                     code = 'I';
                 } else if (code >= 0xB0 && code <= 0xBF) {

diff --git a/src/org/jcodings/specific/ISO8859_9Encoding.java b/src/org/jcodings/specific/ISO8859_9Encoding.java
@@ -19,14 +19,65 @@
  */
 package org.jcodings.specific;
 
+import org.jcodings.Config;
 import org.jcodings.ISOEncoding;
+import org.jcodings.IntHolder;
+import org.jcodings.constants.CharacterType;
 
 public final class ISO8859_9Encoding extends ISOEncoding {
 
     protected ISO8859_9Encoding() {
         super("ISO-8859-9", ISO8859_9CtypeTable, ISO8859_9ToLowerCaseTable, ISO8859_9CaseFoldMap);
     }
 
+    static final int DOTLESS_i = 0xFD;
+    static final int I_WITH_DOT_ABOVE = 0xDD;
+
+    @Override
+    public int caseMap(IntHolder flagP, byte[] bytes, IntHolder pp, int end, byte[] to, int toP, int toEnd) {
+        int toStart = toP;
+        int flags = flagP.value;
+
+        while (pp.value < end && toP < toEnd) {
+            int code = bytes[pp.value++] & 0xff;
+            if (code == SHARP_s) {
+                if ((flags & Config.CASE_UPCASE) != 0) {
+                    flags |= Config.CASE_MODIFIED;
+                    to[toP++] = 'S';
+                    code = (flags & Config.CASE_TITLECASE) != 0 ? 's' : 'S';
+                } else if ((flags & Config.CASE_FOLD) != 0) {
+                    flags |= Config.CASE_MODIFIED;
+                    to[toP++] = 's';
+                    code = 's';
+                }
+            }
+            else if (code == 0xAA || code == 0xB5 || code == 0xBA || code == 0xFF);
+            else if ((ISO8859_9CtypeTable[code] & CharacterType.BIT_UPPER) != 0 && (flags & (Config.CASE_DOWNCASE | Config.CASE_FOLD)) != 0) {
+                flags |= Config.CASE_MODIFIED;
+                if (code == 'I') {
+                    code = (flags & Config.CASE_FOLD_TURKISH_AZERI) != 0 ? DOTLESS_i : 'i';
+                } else {
+                    code = LowerCaseTable[code];
+                }
+            } else if ((ISO8859_9CtypeTable[code] & CharacterType.BIT_LOWER) != 0 && (flags & Config.CASE_UPCASE) != 0) {
+                flags |= Config.CASE_MODIFIED;
+                if (code == 'i') {
+                    code = (flags & Config.CASE_FOLD_TURKISH_AZERI) != 0 ? I_WITH_DOT_ABOVE : 'I';
+                } else if (code == DOTLESS_i) {
+                    code = 'I';
+                } else {
+                    code -= 0x20;
+                }
+            }
+            to[toP++] = (byte)code;
+            if ((flags & Config.CASE_TITLECASE) != 0) {
+                flags ^= (Config.CASE_UPCASE | Config.CASE_DOWNCASE | Config.CASE_TITLECASE);
+            }
+        }
+        flagP.value = flags;
+        return toP - toStart;
+    }
+
     static final short ISO8859_9CtypeTable[] = {
         0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008,
         0x4008, 0x420c, 0x4209, 0x4208, 0x4208, 0x4208, 0x4008, 0x4008,