updated WebVTT and SRT formatters

youtube_transcript_api/formatters.py

@@ -79,8 +79,19 @@ class TextFormatter(Formatter):
         """
         return '\n\n\n'.join([self.format_transcript(transcript, **kwargs) for transcript in transcripts])
 
+class _TextBasedFormatter(TextFormatter):
+    def _format_timestamp(self, hours, mins, secs, ms):
+        raise NotImplementedError('A subclass of _TextBasedFormatter must implement ' \
+            'their own .format_timestamp() method.')
 
-class WebVTTFormatter(Formatter):
+    def _format_transcript_header(self, lines):
+        raise NotImplementedError('A subclass of _TextBasedFormatter must implement ' \
+            'their own _format_transcript_header method.')
+
+    def _format_transcript_helper(self, i, time_text, line):
+        raise NotImplementedError('A subclass of _TextBasedFormatter must implement ' \
+            'their own _format_transcript_helper method.')    
+            
     def _seconds_to_timestamp(self, time):
         """Helper that converts `time` into a transcript cue timestamp.
 
@@ -95,96 +106,55 @@ class WebVTTFormatter(Formatter):
         '00:00:06.930'
         """
         time = float(time)
-        hours, remainder = divmod(time, 3600)
+        hours_float, remainder = divmod(time, 3600)
-        mins, secs = divmod(remainder, 60)
+        mins_float, secs_float = divmod(remainder, 60)
+        hours, mins, secs = int(hours_float), int(mins_float), int(secs_float)
         ms = int(round((time - int(time))*1000, 2))
-        return "{:02.0f}:{:02.0f}:{:02.0f}.{:03d}".format(hours, mins, secs, ms)
+        return self._format_timestamp(hours, mins, secs, ms)
 
     def format_transcript(self, transcript, **kwargs):
-        """A basic implementation of WEBVTT formatting.
+        """A basic implementation of WEBVTT/SRT formatting.
 
         :param transcript:
-        :reference: https://www.w3.org/TR/webvtt1/#introduction-caption
+        :reference: 
+        https://www.w3.org/TR/webvtt1/#introduction-caption
+        https://www.3playmedia.com/blog/create-srt-file/
         """
         lines = []
         for i, line in enumerate(transcript):
-            if i < len(transcript) - 1:
+            end = line['start'] + line['duration']
-                # Looks ahead, use next start time since duration value
+            time_text = "{} --> {}".format(
-                # would create an overlap between start times.
+                self._seconds_to_timestamp(line['start']),
-                time_text = "{} --> {}".format(
+                self._seconds_to_timestamp(
-                    self._seconds_to_timestamp(line['start']),
+                    transcript[i + 1]['start'] 
-                    self._seconds_to_timestamp(transcript[i + 1]['start'])
+                    if i < len(transcript) - 1 and transcript[i + 1]['start'] < end else end
                 )
-            else:
+            )
-                # Reached the end, cannot look ahead, use duration now.
+            lines.append(self._format_transcript_helper(i, time_text, line))
-                duration = line['start'] + line['duration']
-                time_text = "{} --> {}".format(
-                    self._seconds_to_timestamp(line['start']),
-                    self._seconds_to_timestamp(duration)
-                )
-            lines.append("{}\n{}".format(time_text, line['text']))
 
-        return "WEBVTT\n\n" + "\n\n".join(lines) + "\n"
+        return self._format_transcript_header(lines)
 
-    def format_transcripts(self, transcripts, **kwargs):
-        """A basic implementation of WEBVTT formatting for a list of transcripts.
-
-        :param transcripts:
-        :reference: https://www.w3.org/TR/webvtt1/#introduction-caption
-        """
-        return '\n\n\n'.join([self.format_transcript(transcript, **kwargs) for transcript in transcripts])
-
-class SRTFormatter(Formatter):
-    def _seconds_to_timestamp(self, time):
-        """Helper that converts `time` into a transcript cue timestamp for SRT.
-
-        :param time: a float representing time in seconds.
-        :type time: float
-        :return: a string formatted as a cue timestamp, 'HH:MM:SS,MS'
-        :rtype str
-        :example:
-        >>> self._seconds_to_timestamp(6.93)
-        '00:00:06,930'
-        """
-        time = float(time)
-        hours, remainder = divmod(time, 3600)
-        mins, secs = divmod(remainder, 60)
-        ms = int(round((time - int(time))*1000, 2))
-        return "{:02.0f}:{:02.0f}:{:02.0f},{:03d}".format(hours, mins, secs, ms)
-
-    def format_transcript(self, transcript, **kwargs):
-        """Converts a transcript into SRT formatting.
-
-        :param transcript:
-        :reference: https://www.3playmedia.com/blog/create-srt-file/
-        """
-        lines = []
-        for i, line in enumerate(transcript):
-            if i < len(transcript) - 1:
-                # Looks ahead, use next start time since duration value
-                # would create an overlap between start times.
-                time_text = "{} --> {}".format(
-                    self._seconds_to_timestamp(line['start']),
-                    self._seconds_to_timestamp(transcript[i + 1]['start'])
-                )
-            else:
-                # Reached the end, cannot look ahead, use duration now.
-                duration = line['start'] + line['duration']
-                time_text = "{} --> {}".format(
-                    self._seconds_to_timestamp(line['start']),
-                    self._seconds_to_timestamp(duration)
-                )
-            lines.append("{}\n{}\n{}".format(i + 1, time_text, line['text']))
 
+class SRTFormatter(_TextBasedFormatter):
+    def _format_timestamp(self, hours, mins, secs, ms):
+        return "{:02d}:{:02d}:{:02d},{:03d}".format(hours, mins, secs, ms)
+    
+    def _format_transcript_header(self, lines):
         return "\n\n".join(lines) + "\n"
 
-    def format_transcripts(self, transcripts, **kwargs):
+    def _format_transcript_helper(self, i, time_text, line):
-        """Converts a list of transcripts into SRT formatting.
+        return "{}\n{}\n{}".format(i + 1, time_text, line['text'])
 
-        :param transcript:
+
-        :reference: https://www.3playmedia.com/blog/create-srt-file/
+class WebVTTFormatter(_TextBasedFormatter):
-        """
+    def _format_timestamp(self, hours, mins, secs, ms):
-        return '\n\n\n'.join([self.format_transcript(transcript, **kwargs) for transcript in transcripts])
+        return "{:02d}:{:02d}:{:02d}.{:03d}".format(hours, mins, secs, ms)
+
+    def _format_transcript_header(self, lines):
+        return "WEBVTT\n\n" + "\n\n".join(lines) + "\n"
+
+    def _format_transcript_helper(self, i, time_text, line):
+        return "{}\n{}".format(time_text, line['text'])
 
 
 class FormatterLoader(object):