hidive_parser.py

import tinycss
import re
import datetime
import math

#Put this script in the same folder as the .css and .vtt
#If you don't know how to get the .css, you'll need to find a different parser

SUBS_FILENAME = 'subs.vtt'
CSS_FILENAME = 'css.css'
#obviously this input method is absurdly clunky. Sorry.

with open(SUBS_FILENAME, 'r', encoding="utf-8") as f:
    subs = f.read()

parser = tinycss.make_parser('page3')
stylesheet = parser.parse_stylesheet_file(CSS_FILENAME)
css_rules = []
#hidive assigns certain text formatting to a bunch of lines at once. our goal is to get those formatting rules (there are usually less than 10 of them per script) and store them in a way that can be matched to individual subtitle lines later
#we make the assumption that any line that is not yellow is a sign. this has generally worked out pretty well in my testing. basically, for yellow (dialogue) lines we ignore the hidive positioning, and for non-yellow lines we try to replicate the positioning as closely as possible. Warning: be wary about rendering songs like dialogue (you should test with something like the araburu ED if you try)
for i in range(0, len(stylesheet.rules)):
    dialogueRule = 0
    signRule = 0
    italicRule = 0
    maybeDialogueRule = 0
    rule = str(stylesheet.rules[i].declarations)
    if (rule.find('font-family: "Swis721 BT"') != -1) and (rule.find('color: yellow') != -1):
        dialogueRule = 1
    elif (rule.find('font-family: "Swis721 BT"') != -1) and (rule.find('color: yellow') == -1):
        maybeDialogueRule = 1
    else: signRule = 1
    if rule.find('font-style: italic') != -1:
        italicRule = 1
    #check for rules that make text bigger or smaller than normal. Example: {color:#FFFFFF; ... font-size:.84em;}
    if re.search('font-size: ([\d\.]+)em', rule) != None:
        font_multiplier = float(re.search('font-size: ([\d\.]+)em', rule).groups()[0])
    else:
        font_multiplier = 1
    newdict = {'selector' : stylesheet.rules[i].selector.as_css(), 'dialogue' : dialogueRule, 'maybedialogue' : maybeDialogueRule, 'sign' : signRule, 'italic' : italicRule, 'font_multiplier' : font_multiplier}
    css_rules.append(newdict)

time_regx = re.compile('\s*(\d+):(\d{2}):(\d{2})\.(\d{3})\s*-->\s*(\d+):(\d{2}):(\d{2})\.(\d{3})\s*(.*)')
dialogue_regx = re.compile('<c\.(.+?_)(\d)>(.+?)</c>')
get_line = 0
lines = []

subs_lines = subs.split('\n')
for i in range(0, len(subs_lines)):
    if get_line == 1:
        line_moniker_pre, line_moniker_post, line_text = dialogue_regx.match(subs_lines[i]).groups()
        for j in css_rules:
            if j['selector'].find(line_moniker_pre + line_moniker_post) != -1:
                line_meta_dict = j
        if (line_moniker_post == '1') or (line_meta_dict['sign'] == 1):
            lines.append(tentative_dict)
        lines[-1].update(line_meta_dict)
        if (line_moniker_post != '1') and (line_meta_dict['sign'] != 1):
            lines[-1].update({'line_text' : lines[-1]['line_text'] + '\\N' + line_text})
        else:
            lines[-1].update({'line_text' : line_text})
        get_line = 0
    if time_regx.match(subs_lines[i]) != None:
        s_h, s_m, s_s, s_ms, e_h, e_m, e_s, e_ms, positioning = time_regx.match(subs_lines[i]).groups()
        start_time = float(int(s_h)*3600 + int(s_m)*60 + int(s_s)) + (float(s_ms)/1000)
        end_time = float(int(e_h)*3600 + int(e_m)*60 + int(e_s)) + (float(e_ms)/1000)
        tentative_dict = {'start_time' : start_time, 'end_time' : end_time, 'positioning' : positioning}
        get_line = 1

#These are meant to deal with aspect ratio. Keep things around 720p. Standard margins for 1280x720 are 150. You do NOT need to change these is your video is 16:9.
width = '1280'
height = '720'
fontsize = '48'
margins = '150'

#HIDIVE's player is 16:9. So if you have 4:3 video, your positioning will be off if you don't have these lines.
if (float(width) / float(height) < 1.7) or (float(width) / float(height) > 1.8):
    playerwidth = float(height) * 1.78
else:
    playerwidth = float(width)


terxt = "[Script Info]\nScriptType: v4.00+\nWrapStyle: 0\nPlayResX: " + width + "\nPlayResY: " + height + "\nScaledBorderAndShadow: yes\nYCbCr Matrix: TV.709\n\n[V4+ Styles]\nFormat: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, OutlineColour, BackColour, Bold, Italic, Underline, StrikeOut, ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline, Shadow, Alignment, MarginL, MarginR, MarginV, Encoding\nStyle: Default,Gandhi Sans," + fontsize + ",&H00FFFFFF,&H000000FF,&H00000000,&HA0000000,-1,0,0,0,100,100,0,0,1,2.4,1,2," + margins + "," + margins + ",40,1\nStyle: Default-alt,Gandhi Sans," + fontsize + ",&H00FFFFFF,&H000000FF,&H00582209,&HA0000000,-1,0,0,0,100,100,0,0,1,2.4,1,2," + margins + "," + margins + ",40,1\nStyle: Songs/Signs,Gandhi Sans," + fontsize + ",&H00FFFFFF,&H000000FF,&H00000000,&HA0000000,-1,0,0,0,100,100,0,0,1,2.4,1,2," + margins + "," + margins + ",40,1\n\n[Events]\nFormat: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text\n"

def get_timestamps_from_seconds(seconds_float):
    int_s = math.floor(seconds_float)
    hms = str(datetime.timedelta(seconds=int_s))
    ms = "%.2f" % round(seconds_float % 1, 2)
    ms = ms[2:]
    return hms, ms

for i in lines:
    if i['dialogue'] == 1 or (i['maybedialogue'] == 1 and re.search('position:([-\d]+)%', i['positioning']) == None):
        starthms, startms = get_timestamps_from_seconds(i['start_time'])
        endhms, endms = get_timestamps_from_seconds(i['end_time'])
        if i['italic'] == 1:
            italic = '\\i1'
        else:
            italic = ''
        if i['maybedialogue'] == 1:
            orientation = '\\an8'
        else:
            orientation = ''
        tags = ''
        if orientation != '' or italic != '':
            tags = '{' + italic + orientation + '}'
        newline = 'Dialogue: 10,' + starthms + '.' + startms + ',' + endhms + '.' + endms + ',Default,,0,0,0,,' + tags + i['line_text'] + '\n'
        terxt = terxt + newline
        

#the reason i do two loops like this is that I want dialogue to be separate from signs/songs in the .ass
for i in lines:
    if i['sign'] == 1 or (i['maybedialogue'] == 1 and re.search('position:([-\d]+)%', i['positioning']) != None):
        starthms, startms = get_timestamps_from_seconds(i['start_time'])
        endhms, endms = get_timestamps_from_seconds(i['end_time'])
        positioning = i['positioning']
        if re.search('position:([-\d]+)%', positioning) != None:
            #if there is horizontal positioning data in the line, we use it...
            x = int(re.search('position:(\d+)%', positioning).groups()[0])
            x = int((playerwidth * x/100) - (0.5 * abs(float(width) - playerwidth)))
            #change the margin so that text can't go past the right edge of the screen
            right_margin = str(max([1, x - int(margins)]))
            an = '7'
        else:
            #...and if there isn't, we center the text.
            x = int(float(width)/2)
            an = '5'
            right_margin = '0'
        y = int(re.search('line:([-\d]+)%', positioning).groups()[0])
        y = int(int(height) * y/100)
        newfontsize = str(int(float(fontsize) * i['font_multiplier']))
        newline = 'Dialogue: 0,' + starthms + '.' + startms + ',' + endhms + '.' + endms + ',Songs/Signs,,0,' + right_margin + ',0,,' + '{\pos(' + str(x) + ',' + str(y) + ')\\an' + an + '\\fs' + newfontsize + '}' + i['line_text'] + '\n'
        terxt = terxt + newline
        
#aegisub freaks out sometimes when you don't set bom
with open("out_araburu.ass", "w", encoding="utf-8-sig") as f:
    f.write(terxt)