From 7f026091c19339559fb58517218ed649b620e4b7 Mon Sep 17 00:00:00 2001
From: Enkidu93 <lowryec17@gcc.edu>
Date: Fri, 23 Aug 2024 12:51:45 -0400
Subject: [PATCH 1/2] Ignore freestanding ellipses

---
 .../ScriptureRefUsfmParserHandlerBase.cs      |  2 +-
 src/SIL.Machine/Corpora/UsfmTextBase.cs       |  5 +++
 .../Corpora/UsfmMemoryTextTests.cs            | 41 +++++++++++++++++++
 3 files changed, 47 insertions(+), 1 deletion(-)

diff --git a/src/SIL.Machine/Corpora/ScriptureRefUsfmParserHandlerBase.cs b/src/SIL.Machine/Corpora/ScriptureRefUsfmParserHandlerBase.cs
index 363209ed..3fd2eb62 100644
--- a/src/SIL.Machine/Corpora/ScriptureRefUsfmParserHandlerBase.cs
+++ b/src/SIL.Machine/Corpora/ScriptureRefUsfmParserHandlerBase.cs
@@ -171,7 +171,7 @@ public override void EndNote(UsfmParserState state, string marker, bool closed)
         public override void Text(UsfmParserState state, string text)
         {
             // if we hit text in a verse paragraph and we aren't in a verse, then start a non-verse segment
-            if (text.Trim().Length > 0)
+            if (text.Trim().Length > 0 && text.Trim() != "...")
                 CheckConvertVerseParaToNonVerse(state);
         }
 
diff --git a/src/SIL.Machine/Corpora/UsfmTextBase.cs b/src/SIL.Machine/Corpora/UsfmTextBase.cs
index ee3d84f8..f530fd9a 100644
--- a/src/SIL.Machine/Corpora/UsfmTextBase.cs
+++ b/src/SIL.Machine/Corpora/UsfmTextBase.cs
@@ -235,6 +235,11 @@ public override void Text(UsfmParserState state, string text)
                 if (_rowTexts.Count == 0)
                     return;
 
+                if (text.Trim() == "...")
+                {
+                    text = "";
+                }
+
                 StringBuilder rowText = _rowTexts.Peek();
                 if (_text._includeMarkers)
                 {
diff --git a/tests/SIL.Machine.Tests/Corpora/UsfmMemoryTextTests.cs b/tests/SIL.Machine.Tests/Corpora/UsfmMemoryTextTests.cs
index e0712e45..71b7f395 100644
--- a/tests/SIL.Machine.Tests/Corpora/UsfmMemoryTextTests.cs
+++ b/tests/SIL.Machine.Tests/Corpora/UsfmMemoryTextTests.cs
@@ -184,6 +184,47 @@ public void GetRows_OptBreak_BeginningIncludeMarkers()
         });
     }
 
+    [Test]
+    public void GetRows_IgnoreLoneEllipsis()
+    {
+        TextRow[] rows = GetRows(
+            @"\id MAT - Test
+\c 1
+\q1
+\f \fr 119 \ft World \f*
+\v 1 ...
+\v 2 Text
+\c 2
+\d
+description
+\b
+",
+            includeAllText: true
+        );
+        Assert.Multiple(() =>
+        {
+            Assert.That(rows, Has.Length.EqualTo(5), string.Join(",", rows.Select(tr => tr.Text)));
+            Assert.That(rows[1].Text, Is.EqualTo(""));
+        });
+    }
+
+    [Test]
+    public void GetRows_Ellipsis()
+    {
+        TextRow[] rows = GetRows(
+            @"\id MAT - Test
+\c 1
+\v 1 Verse text ... More text
+",
+            includeAllText: true
+        );
+        Assert.Multiple(() =>
+        {
+            Assert.That(rows, Has.Length.EqualTo(1), string.Join(",", rows.Select(tr => tr.Text)));
+            Assert.That(rows[0].Text, Is.EqualTo("Verse text ... More text"));
+        });
+    }
+
     private static TextRow[] GetRows(string usfm, bool includeMarkers = false, bool includeAllText = false)
     {
         UsfmMemoryText text =

From 0d743a7c13a9c17bb99bd1626d5c434a70158b57 Mon Sep 17 00:00:00 2001
From: Enkidu93 <lowryec17@gcc.edu>
Date: Mon, 26 Aug 2024 12:42:45 -0400
Subject: [PATCH 2/2] Move check to text generation

---
 .../Corpora/ScriptureRefUsfmParserHandlerBase.cs         | 2 +-
 src/SIL.Machine/Corpora/UsfmTextBase.cs                  | 9 ++++-----
 2 files changed, 5 insertions(+), 6 deletions(-)

diff --git a/src/SIL.Machine/Corpora/ScriptureRefUsfmParserHandlerBase.cs b/src/SIL.Machine/Corpora/ScriptureRefUsfmParserHandlerBase.cs
index 3fd2eb62..363209ed 100644
--- a/src/SIL.Machine/Corpora/ScriptureRefUsfmParserHandlerBase.cs
+++ b/src/SIL.Machine/Corpora/ScriptureRefUsfmParserHandlerBase.cs
@@ -171,7 +171,7 @@ public override void EndNote(UsfmParserState state, string marker, bool closed)
         public override void Text(UsfmParserState state, string text)
         {
             // if we hit text in a verse paragraph and we aren't in a verse, then start a non-verse segment
-            if (text.Trim().Length > 0 && text.Trim() != "...")
+            if (text.Trim().Length > 0)
                 CheckConvertVerseParaToNonVerse(state);
         }
 
diff --git a/src/SIL.Machine/Corpora/UsfmTextBase.cs b/src/SIL.Machine/Corpora/UsfmTextBase.cs
index f530fd9a..e3777f08 100644
--- a/src/SIL.Machine/Corpora/UsfmTextBase.cs
+++ b/src/SIL.Machine/Corpora/UsfmTextBase.cs
@@ -235,11 +235,6 @@ public override void Text(UsfmParserState state, string text)
                 if (_rowTexts.Count == 0)
                     return;
 
-                if (text.Trim() == "...")
-                {
-                    text = "";
-                }
-
                 StringBuilder rowText = _rowTexts.Peek();
                 if (_text._includeMarkers)
                 {
@@ -279,6 +274,10 @@ protected override void StartVerseText(UsfmParserState state, IReadOnlyList<Scri
             protected override void EndVerseText(UsfmParserState state, IReadOnlyList<ScriptureRef> scriptureRefs)
             {
                 string text = _rowTexts.Pop().ToString();
+                if (text.Trim() == "...")
+                {
+                    text = "";
+                }
                 _rows.AddRange(_text.CreateRows(scriptureRefs, text, _sentenceStart));
                 _sentenceStart = state.Token.Marker == "c" || text.HasSentenceEnding();
             }