Merge pull request #48 from BDisp/column-width

Fixes #47. ColumnWidth needs to differentiate between  non-printable and null characters.

NStack/unicode/Rune.ColumnWidth.cs

@@ -79,20 +79,18 @@ static int bisearch (uint rune, uint [,] table, int max)
 			return 0;
 		}
 
-	        /// <summary>
-	        /// Number of column positions of a wide-character code.   This is used to measure runes as displayed by text-based terminals.
-	        /// </summary>
-	        /// <returns>The width in columns, 0 if the argument is the null character, -1 if the value is not printable, otherwise the number of columsn that the rune occupies.</returns>
-	        /// <param name="r">The red component.</param>
-	        public static int ColumnWidth (Rune rune) 
+		/// <summary>
+		/// Number of column positions of a wide-character code.   This is used to measure runes as displayed by text-based terminals.
+		/// </summary>
+		/// <returns>The width in columns, 0 if the argument is the null character, -1 if the value is not printable, otherwise the number of columns that the rune occupies.</returns>
+		/// <param name="rune">The rune.</param>
+		public static int ColumnWidth (Rune rune)
 	        {
 			uint irune = (uint)rune;
-			if (irune < 32)
-				return 0;
+			if (irune < 32 || (irune >= 0x7f && irune <= 0xa0))
+				return -1;
 			if (irune < 127)
 				return 1;
-			if (irune >= 0x7f && irune <= 0xa0)
-				return 0;
 			/* binary search in table of non-spacing characters */
 			if (bisearch (irune, combining, combining.GetLength (0)-1) != 0)
 				return 0;

NStack/unicode/Rune.cs

@@ -58,9 +58,37 @@ public Rune (uint rune)
 		/// <param name="ch">C# characters.</param>
 		public Rune (char ch)
 		{
+			if (ch >= surrogateMin && ch <= surrogateMax)
+			{
+				throw new ArgumentException("Value in the surrogate range and isn't part of a surrogate pair!");
+			}
 			this.value = (uint)ch;
 		}
 
+		/// <summary>
+		/// Initializes a new instance of the <see cref="T:System.Rune"/> from a surrogate pair value.
+		/// </summary>
+		/// <param name="sgateMin">The high surrogate code points minimum value.</param>
+		/// <param name="sgateMax">The low surrogate code points maximum value.</param>
+		public Rune (uint sgateMin, uint sgateMax)
+		{
+			if (sgateMin < surrogateMin || sgateMax > surrogateMax)
+			{
+				throw new ArgumentOutOfRangeException($"Must be between {surrogateMin:x} and {surrogateMax:x} inclusive!");
+			}
+			this.value = DecodeSurrogatePair(sgateMin, sgateMax);
+		}
+
+		/// <summary>
+		/// Gets a value indicating whether this <see cref="T:System.Rune"/> can be encoded as UTF-8 from a surrogate pair.
+		/// </summary>
+		/// <param name="sgateMin">The high surrogate code points minimum value.</param>
+		/// <param name="sgateMax">The low surrogate code points maximum value.</param>
+		public static uint DecodeSurrogatePair(uint sgateMin, uint sgateMax)
+		{
+			return 0x10000 + ((sgateMin - surrogateMin) * 0x0400) + (sgateMax - lowSurrogateMin);
+		}
+
 		/// <summary>
 		/// Gets a value indicating whether this <see cref="T:System.Rune"/> can be encoded as UTF-8
 		/// </summary>
@@ -79,6 +107,9 @@ public bool IsValid {
 		const uint surrogateMin = 0xd800;
 		const uint surrogateMax = 0xdfff;
 
+		const uint highSurrogateMax = 0xdbff;
+		const uint lowSurrogateMin = 0xdc00;
+
 		const byte t1 = 0x00; // 0000 0000
 		const byte tx = 0x80; // 1000 0000
 		const byte t2 = 0xC0; // 1100 0000
@@ -305,7 +336,7 @@ public static (Rune rune, int size) DecodeLastRune (byte [] buffer, int end = -1
 		/// number of bytes required to encode the rune.
 		/// </summary>
 		/// <returns>The length, or -1 if the rune is not a valid value to encode in UTF-8.</returns>
-		/// <param name="rvalue">Rune to probe.</param>
+		/// <param name="rune">Rune to probe.</param>
 		public static int RuneLen (Rune rune)
 		{
 			var rvalue = rune.value;
@@ -771,8 +802,8 @@ public override int GetHashCode ()
 		public override string ToString ()
 		{
 			var buff = new byte [4];
-			EncodeRune (this, buff, 0);
-			return System.Text.Encoding.UTF8.GetString (buff);
+			var size = EncodeRune (this, buff, 0);
+			return System.Text.Encoding.UTF8.GetString(buff, 0, size);
 		}
 
 		/// <summary>

NStackTests/RuneTest.cs

@@ -1,12 +1,48 @@
-using System;
+using NUnit.Framework;
+using System;
 namespace NStackTests {
 	public class RuneTest {
-		public RuneTest ()
+		Rune a = 'a';
+		Rune b = 'b';
+		Rune c = 123;
+		Rune d = '\u1150';  // 0x1150	ᅐ	Unicode Technical Report #11
+		Rune e = '\u1161';  // 0x1161	ᅡ	null character with column equal to 0
+		Rune f = 31;    // non printable character
+		Rune g = 127;   // non printable character
+
+		[Test]
+		public void TestColumnWidth()
 		{
-			Rune a = 'a';
-			Rune b = 'b';
+			var rt = new RuneTest();
+
+			Assert.AreEqual(1, Rune.ColumnWidth(rt.a));
+			Assert.AreEqual(1, Rune.ColumnWidth(rt.b));
 			var l = a < b;
-			Rune c = 123;
+			Assert.IsTrue(l);
+			Assert.AreEqual(1, Rune.ColumnWidth(rt.c));
+			Assert.AreEqual(2, Rune.ColumnWidth(rt.d));
+			Assert.AreEqual(0, Rune.ColumnWidth(rt.e));
+			Assert.AreEqual(-1, Rune.ColumnWidth(rt.f));
+			Assert.AreEqual(-1, Rune.ColumnWidth(rt.g));
+		}
+
+		[Test]
+		public void TestRune()
+		{
+			Rune a = new Rune('a');
+			Assert.AreEqual("a", a.ToString());
+			Rune b = new Rune(0x0061);
+			Assert.AreEqual("a", b.ToString());
+			Rune c = new Rune('\u0061');
+			Assert.AreEqual("a", c.ToString());
+			Rune d = new Rune(0x10421);
+			Assert.AreEqual("𐐡", d.ToString());
+			Assert.Throws<ArgumentOutOfRangeException>(() => new Rune('\ud799', '\udc21'));
+			Rune e = new Rune('\ud801', '\udc21');
+			Assert.AreEqual("𐐡", e.ToString());
+			Assert.Throws<ArgumentException>(() => new Rune('\ud801'));
+			Rune f = new Rune('\ud83c', '\udf39');
+			Assert.AreEqual("🌹", f.ToString());
 		}
 	}
 }