Ignore freestanding ellipses

sillsdev · Aug 23, 2024 · 7f02609 · 7f02609
1 parent 1220953
commit 7f02609
Show file tree

Hide file tree

Showing 3 changed files with 47 additions and 1 deletion.
diff --git 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
@@ -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
@@ -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 =