Usfm nonverse multichapter (#224)

* Added test case to fail * Don't preserve _curTextType across verses. * Properly close non-verse segment at the beginning of a verse para - Add extra info when exception occurs while tokenizing in UsfmTextBase --------- Co-authored-by: Damien Daspit <[email protected]>
sillsdev · Jul 12, 2024 · 2659e19 · 2659e19
1 parent 04276d7
commit 2659e19
Show file tree

Hide file tree

Showing 3 changed files with 43 additions and 2 deletions.
diff --git a/src/SIL.Machine/Corpora/ScriptureRefUsfmParserHandlerBase.cs b/src/SIL.Machine/Corpora/ScriptureRefUsfmParserHandlerBase.cs
@@ -68,7 +68,10 @@ string pubNumber
             }
             else
             {
-                EndVerseText(state);
+                if (CurrentTextType == ScriptureTextType.NonVerse)
+                    EndNonVerseText(state);
+                else
+                    EndVerseText(state);
                 UpdateVerseRef(state.VerseRef, marker);
                 StartVerseText(state);
             }

diff --git a/src/SIL.Machine/Corpora/UsfmTextBase.cs b/src/SIL.Machine/Corpora/UsfmTextBase.cs
@@ -36,8 +36,25 @@ protected override IEnumerable<TextRow> GetVersesInDocOrder()
         {
             string usfm = ReadUsfm();
             var rowCollector = new TextRowCollector(this);
+
+            var tokenizer = new UsfmTokenizer(_stylesheet);
+            IReadOnlyList<UsfmToken> tokens;
+            try
+            {
+                tokens = tokenizer.Tokenize(usfm, _includeMarkers);
+            }
+            catch (Exception ex)
+            {
+                var sb = new StringBuilder();
+                sb.Append($"An error occurred while tokenizing the text '{Id}`");
+                if (!string.IsNullOrEmpty(Project))
+                    sb.Append($" in project '{Project}'");
+                sb.Append($". Error: '{ex.Message}'");
+                throw new InvalidOperationException(sb.ToString(), ex);
+            }
+
             var parser = new UsfmParser(
-                usfm,
+                tokens,
                 rowCollector,
                 _stylesheet,
                 Versification,

diff --git a/tests/SIL.Machine.Tests/Corpora/UsfmMemoryTextTests.cs b/tests/SIL.Machine.Tests/Corpora/UsfmMemoryTextTests.cs
@@ -70,6 +70,27 @@ public void GetRows_DuplicateVerseWithTable()
         Assert.That(rows, Has.Length.EqualTo(5));
     }
 
+    [Test]
+    public void GetRows_VersePara_BeginningNonVerseSegment()
+    {
+        // a verse paragraph that begins with a non-verse segment followed by a verse segment
+        TextRow[] rows = GetRows(
+            @"\id MAT - Test
+\c 1
+\q1
+\f \fr 119 \ft World \f*
+\v 1 First verse in line!?!
+\c 2
+\d
+description
+\b
+",
+            includeAllText: true
+        );
+
+        Assert.That(rows, Has.Length.EqualTo(4));
+    }
+
     private static TextRow[] GetRows(string usfm, bool includeMarkers = false, bool includeAllText = false)
     {
         UsfmMemoryText text =