From c01f6ce7f6c70b72ceef439996c146f7b4744f98 Mon Sep 17 00:00:00 2001
From: "David A. Madore" <david+git@madore.org>
Date: Thu, 29 Apr 2010 00:20:54 +0200
Subject: More proper handling of CJK and Hangul.

---
 .../madore/android/unicodeMap/UnicodeDatabase.java | 88 +++++++++++++++++-----
 1 file changed, 71 insertions(+), 17 deletions(-)

(limited to 'src/org/madore/android/unicodeMap/UnicodeDatabase.java')

diff --git a/src/org/madore/android/unicodeMap/UnicodeDatabase.java b/src/org/madore/android/unicodeMap/UnicodeDatabase.java
index df91010..efdf5eb 100644
--- a/src/org/madore/android/unicodeMap/UnicodeDatabase.java
+++ b/src/org/madore/android/unicodeMap/UnicodeDatabase.java
@@ -1,6 +1,8 @@
 package org.madore.android.unicodeMap;
 
+import java.util.Collections;
 import java.util.Iterator;
+import java.util.Comparator;
 import java.util.List;
 import java.util.ArrayList;
 import java.util.NoSuchElementException;
@@ -145,27 +147,23 @@ public class UnicodeDatabase {
 
     protected final static String[] queryColumns = { "id", "name", "category" };
 
-    public static UnicodeCharacter getSingleSpecial(int codePoint) {
-	if ( UnicodeCharacter.SpecialRange.isCjkUnifiedIdeograph(codePoint) )
-	    return new UnicodeCharacter(codePoint,
-					UnicodeCharacter.SpecialRange.cjkUnifiedIdeographName(codePoint),
-					UnicodeCharacter.Category.OTHER_LETTER);
-	if ( UnicodeCharacter.SpecialRange.isHangulSyllable(codePoint) )
-	    return new UnicodeCharacter(codePoint,
-					UnicodeCharacter.SpecialRange.hangulSyllableName(codePoint),
-					UnicodeCharacter.Category.OTHER_LETTER);
-	return null;
+    public static UnicodeCharacter getSingleSpecial(UnicodeCharacter.SpecialRange rng,
+						    int codePoint) {
+	return new UnicodeCharacter(codePoint,
+				    rng.getName(codePoint),
+				    rng.getCategory());
     }
 
     public UnicodeCharacter getSingle(int codePoint) {
-	UnicodeCharacter ch = getSingleSpecial(codePoint);
-	if ( ch != null )
-	    return ch;
+	for ( UnicodeCharacter.SpecialRange rng : UnicodeCharacter.SpecialRange.values() )
+	    if ( rng.belongs(codePoint) )
+		return getSingleSpecial(rng, codePoint);
 	final Cursor c
 	    = db.query(UNICODE_TABLE_NAME, queryColumns,
 		       "id=?",
 		       new String[] { Integer.toString(codePoint) },
 		       null, null, null, null);
+	UnicodeCharacter ch;
 	if ( c.getCount() > 0 ) {
 	    c.moveToFirst();
 	    ch = new UnicodeCharacter(c.getInt(0),
@@ -230,13 +228,51 @@ public class UnicodeDatabase {
 	}
     }
 
+    private static class SpecialRangeIterable
+	implements Iterable<UnicodeCharacter> {
+	protected final UnicodeCharacter.SpecialRange rng;
+	protected final int from;  protected final int to;
+	public SpecialRangeIterable(UnicodeCharacter.SpecialRange rng,
+				    int from, int to) {
+	    this.rng = rng;
+	    this.from = from;
+	    this.to = to;
+	}
+	public Iterator<UnicodeCharacter> iterator() {
+	    return (new Iterator<UnicodeCharacter>() {
+		int i = from;
+		public boolean hasNext() { return i < to; }
+		public UnicodeCharacter next() {
+		    if ( i < to )
+			return getSingleSpecial(rng, i++);
+		    else
+			throw new NoSuchElementException();
+		}
+		public void remove() {
+		    throw new UnsupportedOperationException();
+		}
+	    });
+	}
+    }
+
     public Iterable<UnicodeCharacter> getRange(int from, int to) {
+	List<UnicodeCharacter> annoyance = null;
+	boolean muchAnnoyance = false;
 	for ( UnicodeCharacter.SpecialRange rng : UnicodeCharacter.SpecialRange.values() ) {
 	    if ( rng.inside(from, to) ) {
-		List<UnicodeCharacter> list = new ArrayList<UnicodeCharacter>(to-from);
-		for ( int i=from ; i<to ; i++ )
-		    list.add(getSingleSpecial(i));
-		return list;
+		return new SpecialRangeIterable(rng, from, to);
+	    } else if ( rng.interCount(from, to) > 0 ) {
+		if ( annoyance == null )
+		    annoyance
+			= new ArrayList<UnicodeCharacter>(rng.interCount(from, to));
+		else
+		    muchAnnoyance = true;
+		int from0 = (from<rng.getFrom())?rng.getFrom():from;
+		int to0 = (to>rng.getTo())?rng.getTo():to;
+		Iterable<UnicodeCharacter> rangeIt
+		    = new SpecialRangeIterable(rng, from0, to0);
+		for ( UnicodeCharacter ch : rangeIt )
+		    annoyance.add(ch);
 	    }
 	}
 	final Cursor c
@@ -245,6 +281,24 @@ public class UnicodeDatabase {
 		       new String[] { Integer.toString(from),
 				      Integer.toString(to) },
 		       null, null, "id", null);
+	if ( annoyance != null ) {
+	    Iterable<UnicodeCharacter> cursorIt
+		= new CursorIterable(c);
+	    for ( UnicodeCharacter ch : cursorIt ) {
+		annoyance.add(ch);
+		muchAnnoyance = true;
+	    }
+	    if ( muchAnnoyance )
+		Collections.sort(annoyance, new Comparator<UnicodeCharacter>() {
+			public int compare(UnicodeCharacter ch1,
+					   UnicodeCharacter ch2) {
+			    int cp1 = ch1.getCodePoint();
+			    int cp2 = ch2.getCodePoint();
+			    return (cp1<cp2)?-1:(cp1>cp2)?1:0;
+			}
+		    });
+	    return annoyance;
+	}
 	return new CursorIterable(c);
     }
 
-- 
cgit v1.2.3