fix missing last verse (#212)

* #408 - fix missing last verse * Fix issue without change semantics of EndBook --------- Co-authored-by: Damien Daspit <[email protected]>
sillsdev · Jun 19, 2024 · 4e88953 · 4e88953
1 parent 8c39e3f
commit 4e88953
Show file tree

Hide file tree

Showing 5 changed files with 26 additions and 8 deletions.
diff --git a/src/SIL.Machine/Corpora/UsfmTextUpdater.cs b/src/SIL.Machine/Corpora/UsfmTextUpdater.cs
@@ -38,6 +38,12 @@ public UsfmTextUpdater(
 
         public IReadOnlyList<UsfmToken> Tokens => _tokens;
 
+        public override void EndUsfm(UsfmParserState state)
+        {
+            CollectTokens(state);
+            base.EndUsfm(state);
+        }
+
         public override void StartBook(UsfmParserState state, string marker, string code)
         {
             CollectTokens(state);
@@ -314,7 +320,7 @@ private IReadOnlyList<string> AdvanceRows(IReadOnlyList<ScriptureRef> segScrRefs
                     {
                         compare = rowScrRef.CompareTo(segScrRefs[sourceIndex], compareSegments: false);
                         if (compare > 0)
-                            // source is ahead of row, increment source
+                            // row is ahead of source, increment source
                             sourceIndex++;
                         else
                             break;
@@ -330,7 +336,7 @@ private IReadOnlyList<string> AdvanceRows(IReadOnlyList<ScriptureRef> segScrRefs
                 }
                 if (compare <= 0)
                 {
-                    // row is ahead of source, increment row
+                    // source is ahead row, increment row
                     _rowIndex++;
                 }
             }

diff --git a/tests/SIL.Machine.Tests/Corpora/TestData/usfm/Tes/41MATTes.SFM b/tests/SIL.Machine.Tests/Corpora/TestData/usfm/Tes/41MATTes.SFM
@@ -65,3 +65,5 @@
 \b
 \q1 3.1 part 3
 \q1 3.1 part 4
+\c 4
+\v 1
diff --git a/tests/SIL.Machine.Tests/Corpora/UsfmFileTextTests.cs b/tests/SIL.Machine.Tests/Corpora/UsfmFileTextTests.cs
@@ -13,7 +13,7 @@ public void GetRows_NonEmptyText()
 
         IText text = corpus["MAT"];
         TextRow[] rows = text.GetRows().ToArray();
-        Assert.That(rows, Has.Length.EqualTo(23));
+        Assert.That(rows, Has.Length.EqualTo(24));
 
         Assert.That(rows[0].Ref, Is.EqualTo(ScriptureRef.Parse("MAT 1:1", corpus.Versification)));
         Assert.That(rows[0].Text, Is.EqualTo("Chapter one, verse one."));
@@ -73,7 +73,7 @@ public void GetRows_NonEmptyText_AllText()
 
         IText text = corpus["MAT"];
         TextRow[] rows = text.GetRows().ToArray();
-        Assert.That(rows, Has.Length.EqualTo(49));
+        Assert.That(rows, Has.Length.EqualTo(50));
 
         Assert.That(rows[0].Ref, Is.EqualTo(ScriptureRef.Parse("MAT 1:0/1:h", corpus.Versification)));
         Assert.That(rows[0].Text, Is.EqualTo("Matthew"));
@@ -146,7 +146,7 @@ public void GetRows_SentenceStart()
 
         IText text = corpus["MAT"];
         TextRow[] rows = text.GetRows().ToArray();
-        Assert.That(rows, Has.Length.EqualTo(23));
+        Assert.That(rows, Has.Length.EqualTo(24));
 
         Assert.That(rows[3].Ref, Is.EqualTo(ScriptureRef.Parse("MAT 1:4", corpus.Versification)));
         Assert.That(rows[3].Text, Is.EqualTo("Chapter one, verse four,"));
@@ -179,7 +179,7 @@ public void GetRows_IncludeMarkers()
 
         IText text = corpus["MAT"];
         TextRow[] rows = text.GetRows().ToArray();
-        Assert.That(rows, Has.Length.EqualTo(23));
+        Assert.That(rows, Has.Length.EqualTo(24));
 
         Assert.That(rows[0].Ref, Is.EqualTo(ScriptureRef.Parse("MAT 1:1", corpus.Versification)));
         Assert.That(
@@ -254,7 +254,7 @@ public void GetRows_IncludeMarkers_AllText()
 
         IText text = corpus["MAT"];
         TextRow[] rows = text.GetRows().ToArray();
-        Assert.That(rows, Has.Length.EqualTo(45));
+        Assert.That(rows, Has.Length.EqualTo(46));
 
         Assert.That(rows[2].Ref, Is.EqualTo(ScriptureRef.Parse("MAT 1:0/3:ip", corpus.Versification)));
         Assert.That(rows[2].Text, Is.EqualTo("An introduction to Matthew\\fe + \\ft This is an endnote.\\fe*"));

diff --git a/tests/SIL.Machine.Tests/Corpora/UsfmTextUpdaterTests.cs b/tests/SIL.Machine.Tests/Corpora/UsfmTextUpdaterTests.cs
@@ -375,6 +375,16 @@ public void GetUsfm_Verse_DoubleVaVp()
         );
     }
 
+    [Test]
+    public void GetUsfm_Verse_LastVerse()
+    {
+        var rows = new List<(IReadOnlyList<ScriptureRef>, string)> { (ScrRef("MAT 4:1"), "Updating the last verse.") };
+
+        string target = UpdateUsfm(rows);
+        Assert.That(target, Contains.Substring("\\id MAT - Test\r\n"));
+        Assert.That(target, Contains.Substring("\\v 1 Updating the last verse.\r\n"));
+    }
+
     [Test]
     public void GetUsfm_Verse_PretranslationsBeforeText()
     {

diff --git a/tests/SIL.Machine.Tests/Corpora/UsfmTokenizerTests.cs b/tests/SIL.Machine.Tests/Corpora/UsfmTokenizerTests.cs
@@ -11,7 +11,7 @@ public void Tokenize()
         string usfm = ReadUsfm();
         var tokenizer = new UsfmTokenizer();
         IReadOnlyList<UsfmToken> tokens = tokenizer.Tokenize(usfm);
-        Assert.That(tokens, Has.Count.EqualTo(234));
+        Assert.That(tokens, Has.Count.EqualTo(236));
 
         Assert.That(tokens[0].Type, Is.EqualTo(UsfmTokenType.Book));
         Assert.That(tokens[0].Marker, Is.EqualTo("id"));