Keep hyphen runs and word-initial hyphens together in NominalWordWrap [patch]

Extensions.Test/StringExtensionsTests.cs

@@ -590,4 +590,47 @@
 		List<string> result = [.. input.NominalWordWrap(1000f, 10f)];
 		CollectionAssert.AreEqual(new List<string> { "well-known" }, result);
 	}
+
+	[TestMethod]
+	public void NominalWordWrapKeepsADoubleHyphenDashTogether()
+	{
+		// 12 chars per line; "It was late-" would fit, but the break must not fall between the two hyphens.
+		string input = "It was late--too late to call";
+		List<string> result = [.. input.NominalWordWrap(12f, 1f)];
+		CollectionAssert.AreEqual(new List<string> { "It was", "late--too", "late to call" }, result);
+	}
+
+	[TestMethod]
+	public void NominalWordWrapBreaksAfterTheLastHyphenOfARun()
+	{
+		string input = "It was late--too";
+		List<string> result = [.. input.NominalWordWrap(13f, 1f)];
+		CollectionAssert.AreEqual(new List<string> { "It was late--", "too" }, result);
+	}
+
+	[TestMethod]
+	public void NominalWordWrapDoesNotBreakAfterAWordInitialHyphen()
+	{
+		List<string> number = [.. "ab -5".NominalWordWrap(4f, 1f)];
+		CollectionAssert.AreEqual(new List<string> { "ab", "-5" }, number);
+
+		List<string> flag = [.. "Use the --force flag".NominalWordWrap(10f, 1f)];
+		CollectionAssert.AreEqual(new List<string> { "Use the", "--force", "flag" }, flag);
+	}
+
+	[TestMethod]
+	public void NominalWordWrapStillBreaksAfterAHyphenThatFollowsALeadingOne()
+	{
+		// "-5-6": the second hyphen has a digit before it, so it is still a break opportunity.
+		List<string> result = [.. "-5-6".NominalWordWrap(3f, 1f)];
+		CollectionAssert.AreEqual(new List<string> { "-5-", "6" }, result);
+	}
+
+	[TestMethod]
+	public void NominalWordWrapDoesNotAddAHyphenAtASoftHyphenAfterALeadingDash()
+	{
+		// The soft hyphen after "--" is a break opportunity, but the visible dash already ends the upper line.
+		List<string> result = [.. "ab --\u00ADforce".NominalWordWrap(5f, 1f)];
+		CollectionAssert.AreEqual(new List<string> { "ab --", "force" }, result);
+	}
 }

Extensions/StringExtensions.cs

@@ -122,9 +122,9 @@
 #endif
 	}
 
 	private static Regex LineEndingRegexUnix { get; } = new(@"(?<!\r)\n", RegexOptions.Compiled);
 	private static Regex LineEndingRegexWindows { get; } = new(@"\r\n", RegexOptions.Compiled);
 	private static Regex LineEndingRegexMac { get; } = new(@"\r(?!\n)", RegexOptions.Compiled);
 
 	// One pass over every kind of line ending. Normalizing in several passes lets an earlier pass
 	// build a sequence the next one reads differently: under Mac, "\r\n\n" became "\r" + "\n", a new CRLF.
@@ -264,7 +264,7 @@
 	/// </summary>
 	private const char SoftHyphen = '­';
 
 	private static IEnumerable<string> NominalWordWrapIterator(string text, int maxCharsPerLine)
 	{
 		string normalized = text.NormalizeLineEndings(LineEndingStyle.Unix);
 		string[] segments = normalized.Split('\n');
@@ -297,7 +297,7 @@
 
 					yield return line.ToString();
 					line.Clear();
 					atLineStart = true;
 				}
 				else if (!atLineStart)
 				{
@@ -364,19 +364,25 @@
 	{
 		// Split a whitespace-delimited word into break-units at hyphenation opportunities:
 		// after a visible hyphen (kept on the left chunk) or at a soft hyphen (removed unless a break lands there).
+		// A visible hyphen is a break opportunity only when it ends its run of hyphens and some other character
+		// comes before it in the word, so "--" stays together and "-5" or "--force" is never split after the dash.
 		List<(string Text, bool HyphenBefore)> chunks = [];
 		StringBuilder current = new();
 		bool hyphenBefore = false;
+		bool seenNonHyphen = false;
 
-		foreach (char ch in word)
+		for (int i = 0; i < word.Length; i++)
 		{
+			char ch = word[i];
 			if (ch == SoftHyphen)
 			{
 				if (current.Length > 0)
 				{
+					// A visible hyphen that did not end the chunk (a leading "--" before a soft hyphen) already shows at a break here.
+					bool endsInHyphen = current[^1] == '-';
 					chunks.Add((current.ToString(), hyphenBefore));
 					current.Clear();
-					hyphenBefore = true;
+					hyphenBefore = !endsInHyphen;
 				}
 				else if (chunks.Count > 0 && chunks[^1].Text[^1] != '-')
 				{
@@ -390,7 +396,11 @@
 			}
 
 			current.Append(ch);
-			if (ch == '-')
+			if (ch != '-')
+			{
+				seenNonHyphen = true;
+			}
+			else if (seenNonHyphen && (i + 1 >= word.Length || word[i + 1] != '-'))
 			{
 				chunks.Add((current.ToString(), hyphenBefore));
 				current.Clear();