VTT writer: port SRT writer improvments (#343)

src/main/python/ttconv/vtt/paragraph.py

@@ -25,6 +25,7 @@
 
 """WebVTT paragraph"""
 
+import re
 from fractions import Fraction
 from typing import Optional, Union
 
@@ -34,6 +35,8 @@
 class VttParagraph:
   """VTT paragraph definition class"""
 
+  _EOL_SEQ_RE = re.compile(r"\n{2,}")
+
   def __init__(self, identifier: int):
     self._id: int = identifier
     self._begin: Optional[ClockTime] = None
@@ -61,9 +64,14 @@ def get_end(self) -> Optional[ClockTime]:
     """Returns the paragraph end time code"""
     return self._end
 
-  def is_only_whitespace(self):
-    """Returns whether the paragraph tex contains only whitespace"""
-    return self._text.isspace()
+  def is_only_whitespace_or_empty(self):
+    """Returns whether the paragraph text contains only whitespace or is empty"""
+    return len(self._text) == 0 or self._text.isspace()
+
+  def normalize_eol(self):
+    """Remove line breaks at the beginning and end of the paragraph, and replace
+    line break sequences with a single line break"""
+    self._text = VttParagraph._EOL_SEQ_RE.sub("\n", self._text).strip("\n\r")
 
   def append_text(self, text: str):
     """Appends text to the paragraph"""

src/main/python/ttconv/vtt/writer.py

@@ -102,6 +102,13 @@ def append_element(self, element: model.ContentElement, begin: Fraction, end: Op
 
       for elem in list(element):
         self.append_element(elem, begin, end)
+
+      self._paragraphs[-1].normalize_eol()
+
+      if self._paragraphs[-1].is_only_whitespace_or_empty():
+        LOGGER.debug("Removing empty paragraph.")
+        self._paragraphs.pop()
+        self._captions_counter -= 1
 
     if isinstance(element, model.Span):
       is_bold = style.is_element_bold(element)
@@ -185,7 +192,7 @@ def finish(self):
     LOGGER.debug("Check and process the last VTT paragraph.")
 
     if self._paragraphs and self._paragraphs[-1].get_end() is None:
-      if self._paragraphs[-1].is_only_whitespace():
+      if self._paragraphs[-1].is_only_whitespace_or_empty():
         # if the last paragraph contains only whitespace, remove it
         LOGGER.debug("Removing empty unbounded last paragraph.")
         self._paragraphs.pop()

src/test/python/test_vtt_paragraph.py

@@ -38,19 +38,16 @@ class VttParagraphTest(unittest.TestCase):
   def test_paragraph(self):
     paragraph = VttParagraph(123)
 
-    self.assertEqual("123\nNone --> None\n", str(paragraph))
     self.assertRaisesRegex(ValueError, "VTT paragraph begin time code must be set.", paragraph.to_string)
 
     paragraph.set_begin(Fraction(1234, 1))
     self.assertEqual("00:20:34.000", str(paragraph.get_begin()))
 
-    self.assertEqual("123\n00:20:34.000 --> None\n", str(paragraph))
     self.assertRaisesRegex(ValueError, "VTT paragraph end time code must be set.", paragraph.to_string)
 
     paragraph.set_end(Fraction(1234, 1))
     self.assertEqual("00:20:34.000", str(paragraph.get_end()))
 
-    self.assertEqual("123\n00:20:34.000 --> 00:20:34.000\n", str(paragraph))
     self.assertRaisesRegex(ValueError, "VTT paragraph end time code must be greater than the begin time code.", paragraph.to_string)
 
     paragraph.set_end(Fraction(2345, 1))

src/test/python/test_vtt_writer.py

@@ -222,5 +222,22 @@ def _check_output_vtt(self, model: ContentDocument, vtt: str, path: str):
     else:
       self.assertEqual(8, len(vtt), msg=f"Could not convert {path}")
 
+  def test_empty_isds(self):
+    tree = et.parse('src/test/resources/ttml/imsc-tests/imsc1/ttml/timing/BasicTiming010.ttml')
+    doc = imsc_reader.to_model(tree)
+    srt_from_model = vtt_writer.from_model(doc)
+
+    self.assertEqual(srt_from_model, """WEBVTT
+
+1
+00:00:10.000 --> 00:00:24.400
+This text must appear at 10 seconds and disappear at 24.4 seconds
+
+2
+00:00:25.000 --> 00:00:35.000
+This text must appear at 25 seconds and disappear at 35 seconds
+""")
+
+
 if __name__ == '__main__':
   unittest.main()