From f2dbc2cbfd8fd91061103681c55902fc03e8d65f Mon Sep 17 00:00:00 2001
From: Peter Chapman <peter@conglomo.co.nz>
Date: Thu, 4 Dec 2025 09:48:47 +1300
Subject: [PATCH] Fix crash when surrogate pair handling multibyte Unicode
 characters

---
 .../PunctuationAnalysis/TextSegment.cs        | 12 ++++----
 .../QuotationMarkFinderTests.cs               | 28 +++++++++++++++++++
 2 files changed, 34 insertions(+), 6 deletions(-)

diff --git a/src/SIL.Machine/PunctuationAnalysis/TextSegment.cs b/src/SIL.Machine/PunctuationAnalysis/TextSegment.cs
index e4a6d06f..229aec56 100644
--- a/src/SIL.Machine/PunctuationAnalysis/TextSegment.cs
+++ b/src/SIL.Machine/PunctuationAnalysis/TextSegment.cs
@@ -187,13 +187,13 @@ public class SurrogatePairString
         public SurrogatePairString(string stringValue)
         {
             _stringValue = stringValue;
-            IEnumerable<(int SurrogatePairIndex, int StringIndex)> indexPairs = _stringValue
-                .Select((c, i) => (c, i))
+            IEnumerable<(int StringIndex, int SurrogatePairIndex)> indexPairs = _stringValue
+                .Select((c, stringIndex) => (c, stringIndex))
                 .Where(tup => !char.IsLowSurrogate(tup.c))
-                .Select((tup, i) => (tup.i, i));
+                .Select((tup, surrogatePairIndex) => (tup.stringIndex, surrogatePairIndex));
             _surrogatePairIndexByStringIndex = new Dictionary<int, int>();
             _stringIndexBySurrogatePairIndex = new Dictionary<int, int>();
-            foreach ((int surrogatePairIndex, int stringIndex) in indexPairs)
+            foreach ((int stringIndex, int surrogatePairIndex) in indexPairs)
             {
                 _surrogatePairIndexByStringIndex[stringIndex] = surrogatePairIndex;
                 _stringIndexBySurrogatePairIndex[surrogatePairIndex] = stringIndex;
@@ -251,11 +251,11 @@ public string Substring(int startSurrogatePairIndex, int length)
 
         public int GetStringIndexForSurrogatePairIndex(int surrogatePairIndex)
         {
-            if (surrogatePairIndex == _surrogatePairIndexByStringIndex.Count)
+            if (surrogatePairIndex == _stringIndexBySurrogatePairIndex.Count)
             {
                 return _stringValue.Length;
             }
-            return _surrogatePairIndexByStringIndex[surrogatePairIndex];
+            return _stringIndexBySurrogatePairIndex[surrogatePairIndex];
         }
     }
 }
diff --git a/tests/SIL.Machine.Tests/PunctuationAnalysis/QuotationMarkFinderTests.cs b/tests/SIL.Machine.Tests/PunctuationAnalysis/QuotationMarkFinderTests.cs
index 0af3f39c..056efa23 100644
--- a/tests/SIL.Machine.Tests/PunctuationAnalysis/QuotationMarkFinderTests.cs
+++ b/tests/SIL.Machine.Tests/PunctuationAnalysis/QuotationMarkFinderTests.cs
@@ -441,4 +441,32 @@ public void ThatItUsesTheQuoteConventionSet()
                 )
         );
     }
+
+    [Test]
+    public void SupportsMultibyteUnicodeCharacters()
+    {
+        var quotationMarkFinder = new QuotationMarkFinder(QuoteConventions.Standard);
+
+        // [grinning face], [left double quotation mark][grinning face with big eyes][right double quotation mark]
+        Assert.That(
+            quotationMarkFinder
+                .FindAllPotentialQuotationMarksInTextSegment(
+                    new TextSegment.Builder().SetText("\U0001f600, \u201c\U0001f603\u201d").Build()
+                )
+                .SequenceEqual(
+                    [
+                        new QuotationMarkStringMatch(
+                            new TextSegment.Builder().SetText("\U0001f600, \u201c\U0001f603\u201d").Build(),
+                            3,
+                            4
+                        ),
+                        new QuotationMarkStringMatch(
+                            new TextSegment.Builder().SetText("\U0001f600, \u201c\U0001f603\u201d").Build(),
+                            5,
+                            6
+                        ),
+                    ]
+                )
+        );
+    }
 }