change the method of splitting sentences

sentence_api/tests/test_transcription.py

@@ -287,6 +287,38 @@ def test_split_sentences_splits_at_question_marks():
     assert sentences[2].end_seconds == 10.0
 
 
+def test_split_sentences_splits_at_comma_after_fifty_words():
+    before_comma = " ".join(f"word{index}" for index in range(1, 51))
+    after_comma = "This continuation becomes its own sentence."
+    segment = TranscriptionSegment(
+        start_seconds=0.0,
+        end_seconds=55.0,
+        text=f"{before_comma}, {after_comma}",
+    )
+
+    sentences = split_sentences_at_punctuation([segment])
+
+    assert [s.text for s in sentences] == [
+        f"{before_comma},",
+        after_comma,
+    ]
+    assert sentences[0].start_seconds == 0.0
+    assert sentences[0].end_seconds == sentences[1].start_seconds
+    assert sentences[1].end_seconds == 55.0
+
+
+def test_split_sentences_merges_comma_piece_shorter_than_fifty_words():
+    segment = TranscriptionSegment(
+        start_seconds=0.0,
+        end_seconds=5.0,
+        text="One, two, and three.",
+    )
+
+    sentences = split_sentences_at_punctuation([segment])
+
+    assert [s.text for s in sentences] == ["One, two, and three."]
+
+
 def test_split_sentences_falls_back_to_proportional():
     segments = [
         TranscriptionSegment(