Read xml in OpenSubtitles format, output text intended for training machine learning models

subxml2txt.py

#Read xml in OpenSubtitles format, output text intended for training machine learning models

import sys
from os import path
import time
import xml.etree.ElementTree as ET

if len(sys.argv) != 2:
    print("Usage: "+ str(sys.argv[0]) + " input.xml")
    exit(-1)

input_xml_file = sys.argv[1]

if path.exists(input_xml_file) == False:
    print("Error: "+ str(sys.argv[1]) + " not found")
    exit(-2)

root = ET.parse(input_xml_file).getroot()
all_s = root.findall('s')

print("\n<|endoftext|>\n")
previous_timestamp = 0
for type_tag in all_s:
    all_time = type_tag.findall('time')
    if len(all_time) > 0:
        time_tag = all_time[-1]
        value_time = time_tag.get('value')
        ms = int(value_time.split(",")[1])
        time_struct = time.strptime(value_time, "%H:%M:%S,%f")
        timestamp = ((time_struct.tm_hour * 36000) + (time_struct.tm_min * 60) + (time_struct.tm_sec)) * 1000 + ms
        #print('timestamp = ' + str(timestamp))
        if previous_timestamp == 0:
            previous_timestamp = timestamp
        delta_timestamp = timestamp - previous_timestamp
        previous_timestamp = timestamp
        if delta_timestamp > 10000:
            print("<|endoftext|>")

    value_texts = type_tag.itertext()
    print(''.join(value_texts).replace("\n", ""))
print("<|endoftext|>")