Make seed optional; remove unneeded code

sillsdev · Nov 18, 2024 · 5376918 · 5376918
1 parent 6ca6027
commit 5376918
Show file tree

Hide file tree

Showing 3 changed files with 50 additions and 10 deletions.
diff --git a/src/SIL.Machine/Corpora/CorporaExtensions.cs b/src/SIL.Machine/Corpora/CorporaExtensions.cs
@@ -538,7 +538,7 @@ public static INParallelTextCorpus AlignMany(
             return nParallelTextCorpus;
         }
 
-        public static ITextCorpus ChooseRandom(this IEnumerable<ITextCorpus> corpora, int seed)
+        public static ITextCorpus ChooseRandom(this IEnumerable<ITextCorpus> corpora, int? seed = null)
         {
             return new MergedTextCorpus(corpora, MergeRule.Random, seed);
         }

diff --git a/src/SIL.Machine/Corpora/NParallelTextCorpus.cs b/src/SIL.Machine/Corpora/NParallelTextCorpus.cs
@@ -155,15 +155,6 @@ private IEnumerable<NParallelTextRow> GetRows(IList<IEnumerator<TextRow>> enumer
                         && minRefIndexes.Any(i => !completed[i] && currentRows[i].IsInRange) //and at least one of the min rows is not completed and in a range
                     )
                     {
-                        if (
-                            rangeInfo.IsInRange
-                            && nonMinRefIndexes.Any(i =>
-                                !completed[i] && currentRows[i].IsInRange && !currentRows[i].IsEmpty
-                            ) //At least one of the non-min rows is not completed, is in a range, and has content
-                        )
-                        {
-                            yield return rangeInfo.CreateRow();
-                        }
                         foreach (int i in minRefIndexes)
                             rangeInfo.AddTextRow(enumerators[i].Current, i);
                         foreach (int i in nonMinRefIndexes)

diff --git a/tests/SIL.Machine.Tests/Corpora/ParallelTextCorpusTests.cs b/tests/SIL.Machine.Tests/Corpora/ParallelTextCorpusTests.cs
@@ -444,6 +444,55 @@ public void GetRows_OverlappingRanges()
         Assert.That(rows[0].IsTargetSentenceStart, Is.True);
     }
 
+    [Test]
+    public void GetRows_OverlappingRangesAndMissingRow()
+    {
+        var sourceCorpus = new DictionaryTextCorpus(
+            new MemoryText(
+                "text1",
+                new[]
+                {
+                    TextRow(
+                        "text1",
+                        1,
+                        "source segment 1 . source segment 2 . source segment 3 .",
+                        TextRowFlags.SentenceStart | TextRowFlags.InRange | TextRowFlags.RangeStart
+                    ),
+                    TextRow("text1", 2, flags: TextRowFlags.InRange),
+                    TextRow("text1", 3, flags: TextRowFlags.InRange)
+                }
+            )
+        );
+        var targetCorpus = new DictionaryTextCorpus(
+            new MemoryText(
+                "text1",
+                new[]
+                {
+                    TextRow(
+                        "text1",
+                        3,
+                        "target segment 3 . target segment 4 .",
+                        TextRowFlags.SentenceStart | TextRowFlags.InRange | TextRowFlags.RangeStart
+                    ),
+                    TextRow("text1", 4, flags: TextRowFlags.InRange)
+                }
+            )
+        );
+
+        var parallelCorpus = new ParallelTextCorpus(sourceCorpus, targetCorpus);
+        ParallelTextRow[] rows = parallelCorpus.ToArray();
+        Assert.That(rows.Length, Is.EqualTo(1));
+        Assert.That(rows[0].SourceRefs, Is.EqualTo(new[] { 1, 2, 3 }));
+        Assert.That(rows[0].TargetRefs, Is.EqualTo(new[] { 3, 4 }));
+        Assert.That(
+            rows[0].SourceSegment,
+            Is.EqualTo("source segment 1 . source segment 2 . source segment 3 .".Split())
+        );
+        Assert.That(rows[0].TargetSegment, Is.EqualTo("target segment 3 . target segment 4 .".Split()));
+        Assert.That(rows[0].IsSourceSentenceStart, Is.True);
+        Assert.That(rows[0].IsTargetSentenceStart, Is.True);
+    }
+
     [Test]
     public void GetRows_AdjacentRangesSameText()
     {