Fixes sillsdev/serval#449

sillsdev · Aug 13, 2024 · e49bd29 · e49bd29
1 parent e97df8c
commit e49bd29
Show file tree

Hide file tree

Showing 3 changed files with 68 additions and 4 deletions.
diff --git a/src/SIL.Machine/Corpora/ScriptureRefUsfmParserHandlerBase.cs b/src/SIL.Machine/Corpora/ScriptureRefUsfmParserHandlerBase.cs
@@ -187,6 +187,12 @@ IReadOnlyList<UsfmAttribute> attributes
             CheckConvertVerseParaToNonVerse(state);
         }
 
+        public override void OptBreak(UsfmParserState state)
+        {
+            if (_curTextType.Count == 0)
+                _curTextType.Push(ScriptureTextType.NonVerse);
+        }
+
         protected virtual void StartVerseText(UsfmParserState state, IReadOnlyList<ScriptureRef> scriptureRefs) { }
 
         protected virtual void EndVerseText(UsfmParserState state, IReadOnlyList<ScriptureRef> scriptureRefs) { }

diff --git a/src/SIL.Machine/Corpora/UsfmTextBase.cs b/src/SIL.Machine/Corpora/UsfmTextBase.cs
@@ -216,16 +216,20 @@ public override void EndNote(UsfmParserState state, string marker, bool closed)
 
             public override void OptBreak(UsfmParserState state)
             {
-                base.OptBreak(state);
-
                 if (_text._includeMarkers)
                 {
-                    _rowTexts.Peek().Append("//");
+                    if (_rowTexts.Count > 0)
+                        _rowTexts.Peek().Append("//");
+                    else
+                        _rowTexts.Push(new StringBuilder("//"));
                 }
-                else if (CurrentTextType != ScriptureTextType.Verse || state.IsVerseText)
+                else if (CurrentTextType != ScriptureTextType.Verse || state.IsVerseText) //When is this not true?
                 {
+                    if (_rowTexts.Count == 0)
+                        return;
                     _rowTexts.Peek().TrimEnd();
                 }
+                base.OptBreak(state);
             }
 
             public override void Text(UsfmParserState state, string text)

diff --git a/tests/SIL.Machine.Tests/Corpora/UsfmMemoryTextTests.cs b/tests/SIL.Machine.Tests/Corpora/UsfmMemoryTextTests.cs
@@ -111,6 +111,27 @@ public void GetRows_TriplicateVerse()
         });
     }
 
+    [Test]
+    public void GetRows_OptBreak()
+    {
+        // a verse paragraph that begins with a non-verse segment followed by a verse segment
+        TextRow[] rows = GetRows(
+            @"\id MAT - Test
+\c 1
+\v 1 First verse in line // More text
+\c 2
+\v 1
+",
+            includeAllText: true,
+            includeMarkers: true
+        );
+        Assert.Multiple(() =>
+        {
+            Assert.That(rows, Has.Length.EqualTo(2), string.Join(",", rows.ToList().Select(tr => tr.Text)));
+            Assert.That(rows[0].Text, Is.EqualTo(@"First verse in line // More text"));
+        });
+    }
+
     [Test]
     public void GetRows_VersePara_BeginningNonVerseSegment()
     {
@@ -132,6 +153,39 @@ public void GetRows_VersePara_BeginningNonVerseSegment()
         Assert.That(rows, Has.Length.EqualTo(4), string.Join(",", rows.ToList().Select(tr => tr.Text)));
     }
 
+    [Test]
+    public void GetRows_OptBreakAtBeginning()
+    {
+        TextRow[] rows = GetRows(
+            @"\id MAT - Test
+\li //
+",
+            includeAllText: true
+        );
+        Assert.Multiple(() =>
+        {
+            Assert.That(rows, Has.Length.EqualTo(1), string.Join(",", rows.ToList().Select(tr => tr.Text)));
+            Assert.That(rows[0].Text, Is.EqualTo(""));
+        });
+    }
+
+    [Test]
+    public void GetRows_OptBreakAtBeginningIncludeMarkers()
+    {
+        TextRow[] rows = GetRows(
+            @"\id MAT - Test
+\li //
+",
+            includeAllText: true,
+            includeMarkers: true
+        );
+        Assert.Multiple(() =>
+        {
+            Assert.That(rows, Has.Length.EqualTo(1), string.Join(",", rows.ToList().Select(tr => tr.Text)));
+            Assert.That(rows[0].Text, Is.EqualTo("//"));
+        });
+    }
+
     private static TextRow[] GetRows(string usfm, bool includeMarkers = false, bool includeAllText = false)
     {
         UsfmMemoryText text =