app/srt_to_ssml.py

import re
from .translator import translate_text


def sub_to_ssml(input, output, lang, voice, gender, from_lang, to_lang):
  srtFile = input

  # Output file name
  outputFile = output

  #------- SSML Options -------
      # Language
  # language = "en-US"
  language = lang
      # Voice Name - To not specify a voice, put nothing between the quotes or set value to None
  # voiceName = "en-US-DavisNeural"
  voiceName = voice

  durationAttributeName = "duration"
      # Whether to escape special characters in the text. Possible Values: True, False
  enableCharacterEscape = True

  #------- Advanced SSML Options -------
      # SSML Version
  ssmlVersion = "1.0"
      # Whether to include the xmlns:xsi and xsi:schemaLocation attributes in the <speak> tag.
  includeSchemaLocation = True   # Possible Values: True, False

  # Output File Encoding
  chosenFileEncoding = "utf_8_sig" # utf_8_sig for BOM, utf_8 for no BOM

  def escapeChars(enableCharacterEscape, text):
      if enableCharacterEscape:
          text = text.replace("&", "&amp;")
          text = text.replace('"', "&quot;")
          text = text.replace("'", "&apos;")
          text = text.replace("<", "&lt;")
          text = text.replace(">", "&gt;")
      return text


  #======================================== Parse SRT File ================================================
  # Open an srt file and read the lines into a list
  with open(srtFile, 'r', encoding='utf-8-sig') as f:
      lines = f.readlines()

  # Matches the following example with regex:    00:00:20,130 --> 00:00:23,419
  subtitleTimeLineRegex = re.compile(r'\d\d:\d\d:\d\d,\d\d\d --> \d\d:\d\d:\d\d,\d\d\d')

  # Create a dictionary
  subsDict = {}

  # Enumerate lines, and if a line in lines contains only an integer, put that number in the key, and a dictionary in the value
  # The dictionary contains the start, ending, and duration of the subtitles as well as the text
  # The next line uses the syntax HH:MM:SS,MMM --> HH:MM:SS,MMM . Get the difference between the two times and put that in the dictionary
  # For the line after that, put the text in the dictionary
  for lineNum, line in enumerate(lines):
      line = line.strip()
      # If line has no text
      if line.isdigit() and subtitleTimeLineRegex.match(lines[lineNum + 1]):
          lineWithTimestamps = lines[lineNum + 1].strip()
          lineWithSubtitleText = lines[lineNum + 2].strip()

          # If there are more lines after the subtitle text, add them to the text
          count = 3
          while True:
              # Check if the next line is blank or not
              if (lineNum+count) < len(lines) and lines[lineNum + count].strip():
                  lineWithSubtitleText += ' ' + lines[lineNum + count].strip()
                  count += 1
              else:
                  break

          # Create empty dictionary with keys for start and end times and subtitle text
          subsDict[line] = {'start_ms': '', 'end_ms': '', 'duration_ms': '', 'text': '', 'break_until_next': ''}

          time = lineWithTimestamps.split(' --> ')
          time1 = time[0].split(':')
          time2 = time[1].split(':')
          # Converts the time to milliseconds
          processedTime1 = int(time1[0]) * 3600000 + int(time1[1]) * 60000 + int(time1[2].split(',')[0]) * 1000 + int(time1[2].split(',')[1]) #/ 1000 #Uncomment to turn into seconds
          processedTime2 = int(time2[0]) * 3600000 + int(time2[1]) * 60000 + int(time2[2].split(',')[0]) * 1000 + int(time2[2].split(',')[1]) #/ 1000 #Uncomment to turn into seconds
          timeDifferenceMs = str(processedTime2 - processedTime1)
          # Set the keys in the dictionary to the values
          subsDict[line]['start_ms'] = str(processedTime1)
          subsDict[line]['end_ms'] = str(processedTime2)
          subsDict[line]['duration_ms'] = timeDifferenceMs
          subsDict[line]['text'] = lineWithSubtitleText
          if lineNum > 0:
              # Goes back to previous line's dictionary and writes difference in time to current line
              subsDict[str(int(line)-1)]['break_until_next'] = str(processedTime1 - int(subsDict[str(int(line) - 1)]['end_ms']))
          else:
              subsDict[line]['break_until_next'] = '0'

  #=========================================== Create SSML File ============================================
  # Make voice tag if applicable
  if voiceName is None or voiceName == '' or voiceName.lower() == 'none':
      voiceTag = ''
      voiceTagEnd = ''
  else:
    voiceTag = f'<voice xml:lang="{language}" xml:gender="{gender}" name="' + voiceName + '">'
    voiceTagEnd = '</voice>'

  # Encoding with utf-8-sig adds BOM to the beginning of the file, because use with Azure requires it
    output_string = f'<speak version="{ssmlVersion}" xml:lang="{language}">{voiceTag}\n'

    for key, value in subsDict.items():
        if not value['break_until_next'] or value['break_until_next'] == '0':
            breakTimeString = ''
        else:
            breakTime = str(value['break_until_next'])
            breakTimeString = f'<break time="{breakTime}ms"/>'
    
        text = escapeChars(enableCharacterEscape, value['text'])
        translated_text = translate_text(text, from_lang, to_lang)
        texToWrite = (f'\t<prosody {durationAttributeName}="{value["duration_ms"]}ms">{translated_text}</prosody>{breakTimeString}\n')
        output_string += texToWrite
    
    output_string += f'{voiceTagEnd}</speak>'
    with open(outputFile, 'w', encoding=chosenFileEncoding) as f:
        f.write(output_string)
      
    return output_string