AI_Spokesgirl/OpenshotService/test.py

from os import listdir
from os.path import isfile, isdir, join
import openshot
import threading
import zhtts
import os 
import urllib
from typing import List
import requests
from pydantic import BaseModel
from bs4 import BeautifulSoup
from PIL import Image,ImageDraw,ImageFont
import pyttsx3
import rpyc
import random
import re
import time
import math
import dataset
from datetime import datetime
from gtts import gTTS
import ffmpy
from difflib import SequenceMatcher
import difflib
from autosub import DEFAULT_CONCURRENCY
from autosub import DEFAULT_SUBTITLE_FORMAT
from pytranscriber.control.ctr_main import Ctr_Main
from pytranscriber.control.ctr_autosub import Ctr_Autosub
import multiprocessing
from itertools import groupby
from operator import itemgetter
from util.parser import parser

dir_sound = 'mp3_track/'
dir_photo = 'photo/'
dir_text = 'text_file/'
dir_video = 'video_material/'
dir_title = 'title/'
dir_subtitle = 'subtitle/'
dir_anchor = 'anchor_raw/'
tmp_video_dir = 'tmp_video/'
video_sub_folder = 'ai_anchor_video/'

dir_list = [dir_sound,dir_photo,dir_text,dir_video,dir_title,dir_subtitle,dir_anchor,tmp_video_dir]


def cKey(r,g,b,fuzz):
    col=openshot.Color()
    col.red=openshot.Keyframe(r)
    col.green=openshot.Keyframe(g)
    col.blue=openshot.Keyframe(b)
    return openshot.ChromaKey(col, openshot.Keyframe(fuzz))

def video_photo_clip(vid=None,layer=None, position=None, end=None
    ,scale_x=1,scale_y=1,location_x=0,location_y=0,ck=None,audio=True):
    clip = openshot.Clip(vid)
    clip.Layer(layer)
    clip.Position(position)
    clip.End(end)
    clip.scale_x=openshot.Keyframe(scale_x)
    clip.scale_y=openshot.Keyframe(scale_y)
    clip.location_x=openshot.Keyframe(location_x)
    clip.location_y=openshot.Keyframe(location_y)
    
    if ck!=None:
        clip.AddEffect(ck)
    if audio==True:
        clip.has_audio=openshot.Keyframe(1)
    else:
        clip.has_audio=openshot.Keyframe(0)
    return clip

def listener_progress(string, percent):
    True

def myunichchar(unicode_char):
        mb_string = unicode_char.encode('big5')
        try:
            unicode_char = unichr(ord(mb_string[0]) << 8 | ord(mb_string[1]))
        except NameError:
            unicode_char = chr(mb_string[0] << 8 | mb_string[1])
        return unicode_char

def get_url_type(url):
    print('---------------------------------------------')
    print(url)
    req = urllib.request.Request(url, method='HEAD', headers={'User-Agent': 'Mozilla/5.0'})
    r = urllib.request.urlopen(req)
    contentType = r.getheader('Content-Type')
    print(contentType)
    print('-------------------------------------------------')
    return contentType
    
def make_dir(name_hash):
    for direct in dir_list:
        if not os.path.isdir(direct):
            os.mkdir(direct)
    try:
        os.mkdir(dir_photo+name_hash)
    except FileExistsError:
        print("~~~~~~Warning~~~~~~~~~Directory " , dir_photo+name_hash ,  " already exists")
    try:
        os.mkdir(dir_text+name_hash)
    except FileExistsError:
        print("~~~~~~Warning~~~~~~~~~Directory " , dir_text+name_hash ,  " already exists")
    try:
        os.mkdir(dir_sound+name_hash)
    except FileExistsError:
        print("~~~~~~Warning~~~~~~~~~Directory " , dir_sound+name_hash ,  " already exists")
    try:
        os.mkdir(dir_anchor+name_hash)
    except FileExistsError:
        print("~~~~~~Warning~~~~~~~~~Directory " , dir_anchor+name_hash ,  " already exists")
    try:
        os.mkdir(dir_subtitle+name_hash)
    except FileExistsError:
        print("~~~~~~Warning~~~~~~~~~Directory " , dir_subtitle+name_hash ,  " already exists")

def file_prepare(name, name_hash,text_content,image_urls,multiLang,lang='zh'):
    make_dir(name_hash)
    img_num = 1
    for imgu in image_urls:
        if get_url_type(imgu) =='video/mp4':
            r=requests.get(imgu)
            f=open(dir_photo+name_hash+"/"+str(img_num)+".mp4",'wb')
            for chunk in r.iter_content(chunk_size=255): 
                if chunk:
                    f.write(chunk)
            f.close()
        else:
            im = Image.open(requests.get(imgu, stream=True).raw)
            im= im.convert("RGB")
            im.save(dir_photo+name_hash+"/"+str(img_num)+".jpg")
        img_num+=1
    #save text
    txt_idx=0
    for txt in text_content:
        text_file = open(dir_text+name_hash+"/"+str(txt_idx)+".txt", "w")
        text_file.write(txt)
        text_file.close()
        txt_idx+=1
    print("text file made")
    #make mp3
    txt_idx = 0
    for txt in text_content:
        if lang!='zh' or multiLang==1:
            if lang!='zh':
                tts = gTTS(txt)
                tts.save(dir_sound+name_hash+"/"+str(txt_idx)+"raw.mp3")
            else:
                tts = gTTS(txt,lang='zh-tw')
                tts.save(dir_sound+name_hash+"/"+str(txt_idx)+"raw.mp3")
            #speed up 
            ff = ffmpy.FFmpeg(inputs={dir_sound+name_hash+"/"+str(txt_idx)+"raw.mp3": None}
                            , outputs={dir_sound+name_hash+"/"+str(txt_idx)+".mp3": ["-filter:a", "atempo=1.2"]})
            ff.run()
            os.remove(dir_sound+name_hash+"/"+str(txt_idx)+"raw.mp3")
        else:
            print('use zhtts')
            tts = zhtts.TTS() 
            tts.text2wav(txt,dir_sound+name_hash+"/"+str(txt_idx)+".mp3")
        txt_idx+=1
    print("mp3 file made")
    #make title as image
    txt2image_title(name, dir_title+name_hash+".png",lang)


def file_prepare_long(name, name_hash,text_content,image_urls,multiLang,lang='zh'):
    make_dir(name_hash)
    img_num = 1
    for imgu in image_urls:
        if get_url_type(imgu) =='video/mp4':
            r=requests.get(imgu)
            f=open(dir_photo+name_hash+"/"+str(img_num)+".mp4",'wb')
            for chunk in r.iter_content(chunk_size=255): 
                if chunk:
                    f.write(chunk)
            f.close()
        else:
            im = Image.open(requests.get(imgu, stream=True).raw)
            im= im.convert("RGB")
            im.save(dir_photo+name_hash+"/"+str(img_num)+".jpg")
        img_num+=1

    #make mp3
    text_parser = parser()
    txt_idx = 0
    for txt in text_content:
        rep_list = text_parser.replace_list(txt)
        for reptxt in rep_list:
            txt = txt.replace(reptxt,'')
        if lang!='zh' or multiLang==1:
            if lang!='zh':
                tts = gTTS(txt)
                tts.save(dir_sound+name_hash+"/"+str(txt_idx)+"raw.mp3")
            else:
                tts = gTTS(txt,lang='zh-tw')
                tts.save(dir_sound+name_hash+"/"+str(txt_idx)+"raw.mp3")
            #speed up 
            ff = ffmpy.FFmpeg(inputs={dir_sound+name_hash+"/"+str(txt_idx)+"raw.mp3": None}
                            , outputs={dir_sound+name_hash+"/"+str(txt_idx)+".mp3": ["-filter:a", "atempo=1.2"]})
            ff.run()
            os.remove(dir_sound+name_hash+"/"+str(txt_idx)+"raw.mp3")
        else:
            print('use zhtts')
            tts = zhtts.TTS() 
            tts.text2wav(txt,dir_sound+name_hash+"/"+str(txt_idx)+".mp3")
        txt_idx+=1
    print("mp3 file made")
    #make title as image
    txt2image_title(name, dir_title+name_hash+".png",lang)

def txt2image(content, save_target,lang='zh'):
    unicode_text = trim_punctuation(content)
    font = ''
    if lang=='zh':
        font = ImageFont.truetype(font="font/DFT_B7.ttc", size=38)
    else :
        font = ImageFont.truetype(font="font/arial.ttf", size=38)
    text_width, text_height = font.getsize(unicode_text)
    canvas = Image.new('RGBA', (700, 500), (255, 0, 0, 0) )
    draw = ImageDraw.Draw(canvas)
    text= unicode_text
    draw.text((5,5), text, (255, 255, 0), font)
    canvas.save(save_target, "PNG")

def txt2image_title(content, save_target, lang='zh'):
    unicode_text = trim_punctuation(content)
    font = ''
    if lang=='zh':
        font = ImageFont.truetype(font="font/DFT_B7.ttc", size=22)
    else :
        font = ImageFont.truetype(font="font/arial.ttf", size=22)
    text_width, text_height = font.getsize(unicode_text)
    canvas = Image.new('RGBA', (510, 500), (255, 0, 0, 0) )
    draw = ImageDraw.Draw(canvas)
    text= unicode_text
    draw.text((5,5), text, (17, 41, 167), font)
    canvas.save(save_target, "PNG")

def call_anchor(fileName,avatar):
    conn = rpyc.classic.connect("192.168.1.105",18812)
    ros = conn.modules.os 
    rsys = conn.modules.sys 
    fr=open(dir_sound+fileName+".mp3",'rb')# voice
    #warning!!!    file my be replaced by other process
    fw=conn.builtins.open('/tmp/output.mp3','wb')

    while True:
        b=fr.read(1024)
        if b:
            fw.write(b)
        else:
            break

    fr.close()
    fw.close()

    val=random.randint(1000000,9999999)
    ros.chdir('/home/jared/to_video')
    ros.system('./p'+str(avatar)+'.sh '+str(val)+' &')

    while True:
        print('waiting...')
        if ros.path.exists('/tmp/results/'+str(val)):
            break
        time.sleep(5)
        print('waiting...')

    fr=conn.builtins.open('/tmp/results/'+str(val)+'.mp4','rb')
    fw=open(dir_anchor+fileName+".mp4",'wb')
    while True:
        b=fr.read(1024)
        if b:
            fw.write(b)
        else:
            break

    fr.close()
    fw.close()
    
def syllable_count(word):
    word = word.lower()
    count = 0
    vowels = "aeiouy"
    if word[0] in vowels:
        count += 1
    for index in range(1, len(word)):
        if word[index] in vowels and word[index - 1] not in vowels:
            count += 1

    if word.endswith("e"):
        count -= 1
    if count == 0:
        count += 1
    return count

def split_sentence(in_str, maxLen):
    re.findall(r'[\u4e00-\u9fff]+', in_str)

    zh_idx = []
    eng_idx= []
    for i in range(len(in_str)):
        if in_str[i] > u'\u4e00' and in_str[i] < u'\u9fff':
            zh_idx.append(i)
        else:
            eng_idx.append(i)

    space_index = [m.start() for m in re.finditer(' ', in_str)]
    for idx in space_index:
        eng_idx.remove(idx)
    
    eng_range_list = []
    for k, g in groupby(enumerate(eng_idx), lambda ix : ix[0] - ix[1]):
        eng_range = list(map(itemgetter(1), g))
        eng_range_list.append(eng_range)

    total_syllable = 0
    for i in range(len(eng_range_list)):
        total_syllable += (syllable_count(in_str[eng_range_list[i][0]:eng_range_list[i][-1]+1])+0.5)
    for i in range(len(zh_idx)):
        total_syllable+=1
    
    #final chchchchchc[en][en][en]
    #[en] is a vocabulary dict with  occurence of image
    zh_eng_idx_list = []
    i = 0
    while i < len(in_str):
        if in_str[i]==' ':
            i+=1
        if i in zh_idx:
            zh_eng_idx_list.append(i)
            i+=1
        if i in eng_idx:
            for ls in eng_range_list:
                if i in ls:
                    zh_eng_idx_list.append(ls)
                    i = ls[-1]+1
                    break
            
    zh_eng_dict_list = [{'content':'','time_ratio':0}]
    idx = 0 
    current_len = 0
    sen_idx = 0
    while idx < len(zh_eng_idx_list):
        str_from_idx = ''
        sylla_cnt = 1
        if type(zh_eng_idx_list[idx])==type([]):
            str_from_idx = in_str[zh_eng_idx_list[idx][0]:zh_eng_idx_list[idx][-1]+1]+' '
            sylla_cnt = syllable_count(str_from_idx)
        else:
            str_from_idx = in_str[zh_eng_idx_list[idx]]
    
      
        if len(zh_eng_dict_list[sen_idx]['content'])+sylla_cnt>=maxLen:
            zh_eng_dict_list[sen_idx]['time_ratio'] = current_len/total_syllable
           
            zh_eng_dict_list.append({'content':'','time_ratio':0})
            sen_idx+=1
            current_len = 0
        else:
            current_len += sylla_cnt
            zh_eng_dict_list[sen_idx]['content'] += str_from_idx
            idx+=1
        
    total_ratio = 0
    for obj in zh_eng_dict_list:
        total_ratio+=obj['time_ratio']
    zh_eng_dict_list[-1]['time_ratio'] = 1-total_ratio
    return zh_eng_dict_list
   
def parse_script(file_path,gt_list):
    with open(file_path, 'r',encoding="utf-8") as f:
        raw_lines = [line.strip() for line in f]
    lines = adjustSub_by_text_similarity(gt_list,raw_lines)
    text_parser = parser()
    #make dict
    dict_list = []
    for idx in range(len(lines)):
        script={}
        rep_ls = text_parser.replace_list(lines[idx])
        line_content = lines[idx]
        for reptxt in rep_ls:
            line_content = line_content.replace(reptxt,'')
        if len(rep_ls)!=0:
            script['image_idx'] = int(rep_ls[0].replace('{','').replace('}',''))
        script['content'] = line_content
        time_raw = raw_lines[idx * 4 +1 ].split(' --> ')
        start = time_raw[0].split(':')
        stop = time_raw[1].split(':')
        script['start'] = float(start[0])*3600 + float(start[1])*60 + float(start[2].replace(',','.'))
        script['stop'] = float(stop[0])*3600 + float(stop[1])*60 + float(stop[2].replace(',','.'))
        dict_list.append(script)
 
    #merge duplicated sentences
    skip_list = []
    script_not_dup_list = []
    for idx in range(len(dict_list)):
        if idx not in skip_list:
            dup_list = []
            found = 0
            for idx_inner in range(len(dict_list)):
                if dict_list[idx_inner]['content'] == dict_list[idx]['content'] and idx <= idx_inner:
                    dup_list.append(idx_inner)
                    skip_list.append(idx_inner)
                    found += 1
                if found != 0 and dict_list[idx_inner]['content']!=dict_list[idx]['content'] and idx <= idx_inner:
                    found = 0
                    break
        
            for dup_idx in dup_list:
                if dup_idx == min(dup_list):
                    dict_list[dup_idx]['type'] = 'lead_sentence'
                else:
                    dict_list[dup_idx]['type'] = 'duplicated'
            dict_list[dup_list[0]]['stop'] = dict_list[dup_list[-1]]['stop']
          
            if dict_list[idx]['type'] == 'lead_sentence':
                script_not_dup_list.append(dict_list[idx])
            
    
    new_idx = 0
    splitted_dict = []
    for dic in script_not_dup_list:
        dic_idx = 0
        accumulated_duration = 0
        duration = dic['stop']-dic['start']

        for sub_dic in split_sentence(dic['content'],13):
            new_dic = {}
            new_dic['index'] = new_idx
            if 'image_idx' in dic:
                new_dic['image_obj'] = {'start':dic['start'],'idx':dic['image_idx']}
            new_idx+=1
            ind_duration = duration * sub_dic['time_ratio']
            new_dic['start'] = dic['start'] + accumulated_duration
            accumulated_duration += ind_duration
            new_dic['content'] = sub_dic['content']
            new_dic['duration'] = ind_duration*0.7
            splitted_dict.append(new_dic)
    return splitted_dict



def adjustSub_by_text_similarity(gts_in,gens_raw):
    #call by value only
    gts = gts_in[:]
    text_parser = parser()
    for i in range(len(gts)):
        rep_ls = text_parser.replace_list(gts[i])
        for reptxt in rep_ls:
            gts[i] = gts[i].replace(reptxt,'')

    gens = []
    for idx in range(int((len(gens_raw)+1)/4)):
        gens.append(gens_raw[idx*4+2])

    combine2 = [''.join([i,j]) for i,j in zip(gts, gts[1:])]
    combine3 = [''.join([i,j,k]) for i,j,k in zip(gts, gts[1:], gts[2:])]
    alls = gts #+ combine2 + combine3
    adjusted = [None]*len(gens)
    duplicated_list = []
    for idx in range(len(gens)):
        match_text = difflib.get_close_matches(gens[idx], alls, cutoff=0.1)
        if len(match_text) != 0:
            if match_text[0] not in duplicated_list:
                adjusted[idx] = match_text[0]
                duplicated_list.append(match_text[0])
            else:
                if match_text[0] == adjusted[idx-1]:
                    adjusted[idx] = match_text[0]
                else:
                    found = 0
                    for mt in match_text:
                        if mt not in duplicated_list:
                            adjusted[idx] = mt
                            found += 1
                            break
                    if found ==0:
                        adjusted[idx] = ' '
        else :
            adjusted[idx] = ' '

    combine2_tag = [''.join([i,j]) for i,j in zip(gts_in, gts_in[1:])]
    combine3_tag = [''.join([i,j,k]) for i,j,k in zip(gts_in, gts_in[1:], gts_in[2:])]
    alls_tag = gts_in #+ combine2_tag + combine3_tag

    for idx in range(len(adjusted)):
        match_text = difflib.get_close_matches(adjusted[idx], alls_tag, cutoff=0.1)
        adjusted[idx] = match_text[0]
    return adjusted

def trim_punctuation(s):
    pat_block = u'[^\u4e00-\u9fff0-9a-zA-Z]+';
    pattern = u'([0-9]+{0}[0-9]+)|{0}'.format(pat_block)
    res = re.sub(pattern, lambda x: x.group(1) if x.group(1) else u" " ,s)
    return res

def splitter(s):
    for sent in re.findall(u'[^!?，。\!\?]+[!? 。\!\?]?', s, flags=re.U):
        yield sent

def split_by_pun(s):
    res = list(splitter(s))
    return res

def generate_subtitle_image_from_dict(name_hash, sub_dict):
    for script in sub_dict:
        sv_path = dir_subtitle + name_hash + '/' + str(script['index'])+'.png'
        sub = script['content']
        txt2image(sub,sv_path)

def generate_subtitle_image(name_hash,text_content):
    img_list = [None]*len(text_content)
    for idx in range(len(text_content)):
        img_list[idx]=[]
        senList = split_by_pun(text_content[idx])
        for inner_idx in range(len(senList)):
            sv_path = dir_subtitle + name_hash +'/'+str(idx)+ str(inner_idx) +'.png'
            sub = senList[inner_idx]
            txt2image(sub,sv_path)
            img_list[idx]+=[{"count":len(sub),"path":sv_path}]
    return img_list

def generate_subtitle_image_ENG(name_hash,text_content):
    img_list = [None]*len(text_content)
    for idx in range(len(text_content)):
        sv_path = dir_subtitle + name_hash +'/'+str(idx)+'.png'
        sub = text_content[idx]
        txt2image(sub, sv_path,lang='eng')
        img_list[idx] = sv_path
    return img_list

def video_writer_init(path):
    w = openshot.FFmpegWriter(path)
    w.SetAudioOptions(True, "aac", 44100, 2, openshot.LAYOUT_STEREO, 3000000)
    w.SetVideoOptions(True, "libx264", openshot.Fraction(30000, 1000), 1280, 720,
        openshot.Fraction(1, 1), False, False, 3000000)
    return w

def video_gen(name_hash,name,text_content, image_urls,multiLang,avatar):
    file_prepare_long(name, name_hash, text_content,image_urls,multiLang)
    
    for fname in range(len(text_content)):
        call_anchor(name_hash+"/"+str(fname),avatar)
    print('called............................................')
    ck=cKey(0,254,0,270)
    ck_anchor=cKey(0,255,1,320)
    t = openshot.Timeline(1280, 720, openshot.Fraction(30000, 1000), 44100, 2, openshot.LAYOUT_STEREO)
    t.Open()
    main_timer = 0
    LOGO_OP = openshot.FFmpegReader(dir_video+"LOGO_OP_4.mp4")
    LOGO_OP.Open()         # Open the reader
    head_duration = LOGO_OP.info.duration
    LOGO_OP_clip = video_photo_clip(vid=LOGO_OP,layer=4,position=0,end=head_duration
                    ,location_y=-0.03,scale_x=0.8,scale_y=0.704)
    t.AddClip(LOGO_OP_clip)
    bg_head = openshot.FFmpegReader(dir_video+"complete_head_aispokesgirl.mp4")
    bg_head.Open()
    bg_head_clip = video_photo_clip(vid=bg_head,layer=2,position=0,end=LOGO_OP.info.duration,ck=ck)
    t.AddClip(bg_head_clip)
    main_timer += head_duration
    bg_head.Close()
    LOGO_OP.Close()
    
    anchor = openshot.FFmpegReader(dir_anchor+name_hash+"/0.mp4")
    anchor.Open()
    #anchor_clip = video_photo_clip(vid=anchor,layer=4,scale_x=0.65,scale_y=0.65,
    #        location_x=0.35,location_y=0.25,position=main_timer, end=anchor.info.duration,ck=ck_anchor,audio=False)
    #t.AddClip(anchor_clip)

    speech = openshot.FFmpegReader(dir_sound+name_hash+"/0.mp3")
    speech.Open()
    speech_clip = openshot.Clip(speech)
    speech_clip.Position(main_timer)
    speech_clip.End(anchor.info.duration)
    t.AddClip(speech_clip)
    main_timer += anchor.info.duration
    anchor.Close()
    speech.Close()
    
    LOGO_ED = openshot.FFmpegReader(dir_video+"LOGO_ED.avi")
    LOGO_ED.Open()
    LOGO_ED_clip = video_photo_clip(vid=LOGO_ED,layer=4,position=main_timer,end=LOGO_ED.info.duration
                    ,location_x=0.005,location_y=-0.031, scale_x=0.8,scale_y=0.6825)
    t.AddClip(LOGO_ED_clip)
    main_timer += LOGO_ED.info.duration
    LOGO_ED.Close()
    
    bg = openshot.FFmpegReader(dir_video+"complete_double_aispokesgirl.mp4")
    bg.Open()
    bg_times = math.floor(main_timer/bg.info.duration)
    left_time = (main_timer) % bg.info.duration
    bg_clip_list = [None] * bg_times
    bg_list = [None] * bg_times
    bg.Close()
    bg_timer = head_duration
    for idx in range(bg_times):
        bg_list[idx] = openshot.FFmpegReader(dir_video+"complete_double_aispokesgirl.mp4")
        bg_list[idx].Open()
        bg_clip_list[idx] = video_photo_clip(bg_list[idx],layer=2,position=bg_timer,end=bg_list[idx].info.duration,ck=ck)
        t.AddClip(bg_clip_list[idx])
        bg_timer += bg_list[idx].info.duration
        bg_list[idx].Close()
    bg_left = openshot.FFmpegReader(dir_video+"complete_double_aispokesgirl.mp4")
    bg_left.Open()
    bg_left_clip = video_photo_clip(bg_left,layer=2,position=bg_timer,end=left_time,ck=ck)
    t.AddClip(bg_left_clip)
    bg_left.Close()


    title = openshot.QtImageReader(dir_title+name_hash+".png")
    title.Open()         # Open the reader
    title_clip = video_photo_clip(vid=title, layer=4,location_x=-0.047, location_y=0.801,position=0,end=head_duration+main_timer)
    t.AddClip(title_clip)

    w = video_writer_init(tmp_video_dir+name_hash+"raw.mp4")
    w.Open()
    frames = int(t.info.fps)*int(main_timer)
    for n in range(frames):
        f=t.GetFrame(n)
        w.WriteFrame(f)
    t.Close()
    w.Close()
    print(name+"RAW DONE : www.choozmo.com:8168/"+tmp_video_dir+name_hash+"raw.mp4")
    #start adding sub
    
    #add sub
    Ctr_Autosub.init()
    Ctr_Autosub.generate_subtitles(tmp_video_dir+name_hash+"raw.mp4",'zh',listener_progress,output=tmp_video_dir+name_hash+"script.txt",concurrency=DEFAULT_CONCURRENCY,subtitle_file_format=DEFAULT_SUBTITLE_FORMAT)
    
    sub_dict = parse_script(tmp_video_dir+name_hash+"script.txt",split_by_pun(text_content[0]))
    for subd in sub_dict:
        print(subd)
    
    generate_subtitle_image_from_dict(name_hash, sub_dict)

    #sv_path = dir_subtitle + name_hash + '/' + str(script['index'])+'.png'

    t = openshot.Timeline(1280, 720, openshot.Fraction(30000, 1000), 44100, 2, openshot.LAYOUT_STEREO)
    t.Open()

    raw = openshot.FFmpegReader(tmp_video_dir+name_hash+"raw.mp4")
    raw.Open()
    raw_clip = video_photo_clip(vid=raw,layer=2,position=0, end=raw.info.duration)
    t.AddClip(raw_clip)
    


    sub_img_list = [None] * len(sub_dict)
    sub_clip_list = [None] * len(sub_dict)
    for sub_obj in sub_dict:
        idx = int(sub_obj['index'])
        sub_img_list[idx] = openshot.QtImageReader(dir_subtitle + name_hash + '/' + str(idx)+'.png')
        sub_img_list[idx].Open()
        #if sub_obj['duration']>3:
        #    print('warning')
        #print('start:',sub_obj['start'],', duration :', sub_obj['duration'],'  content',sub_obj['content'],'idx:',sub_obj['index'])
        
        sub_clip_list[idx] = video_photo_clip(vid=sub_img_list[idx], layer=6,location_x=0.069, location_y=0.89,position=sub_obj['start'],end=math.ceil(sub_obj['duration']))
        t.AddClip(sub_clip_list[idx])
        sub_img_list[idx].Close()


    tp = parser()
    img_dict_ls = tp.image_clip_info(sub_dict)
    img_clip_list = [None]*len(listdir(dir_photo+name_hash))
    img_list = [None]*len(img_clip_list)

    img_file_ls = listdir(dir_photo+name_hash)

    for img_idx in range(len(img_file_ls)):
        img_list[img_idx] = openshot.FFmpegReader(dir_photo+name_hash+'/'+img_file_ls[img_idx])
        img_list[img_idx].Open()
        img_clip_list[img_idx] = video_photo_clip(vid=img_list[img_idx],layer=3
                ,scale_x=0.81,scale_y=0.68,location_y=-0.03,position=img_dict_ls[img_idx]['start'],end=img_dict_ls[img_idx]['duration'],audio=False)
        t.AddClip(img_clip_list[img_idx])
        img_list[img_idx].Close()

    anchor = openshot.FFmpegReader(dir_anchor+name_hash+"/0.mp4")
    anchor.Open()
    anchor_clip = video_photo_clip(vid=anchor,layer=4,scale_x=0.65,scale_y=0.65,
            location_x=0.35,location_y=0.25,position=head_duration, end=anchor.info.duration,ck=ck_anchor,audio=False)
    t.AddClip(anchor_clip)


    w = video_writer_init(tmp_video_dir+name_hash+".mp4")
    w.Open()
    frames = int(t.info.fps)*int(main_timer)
    for n in range(frames):
        f=t.GetFrame(n)
        w.WriteFrame(f)
    t.Close()
    w.Close()


    os.remove(tmp_video_dir+name_hash+"raw.mp4")
    os.remove(tmp_video_dir+name_hash+"script.txt")
    print(name+"ALL DONE : www.choozmo.com:8168/"+video_sub_folder+name_hash+"raw.mp4")


def anchor_video_v2(name_hash,name,text_content, image_urls,multiLang,avatar):

    file_prepare(name, name_hash, text_content,image_urls,multiLang)
    sub_list=generate_subtitle_image(name_hash,text_content)
    
    for fname in range(len(text_content)):
        call_anchor(name_hash+"/"+str(fname),avatar)
        print('step finish')
    print('called............................................')

    ck=cKey(0,254,0,270)
    ck_anchor=cKey(0,255,1,320)
    duration = 0
    #average layer level is 3
    t = openshot.Timeline(1280, 720, openshot.Fraction(30000, 1000), 44100, 2, openshot.LAYOUT_STEREO)
    t.Open()

    main_timer = 0
    
    LOGO_OP = openshot.FFmpegReader(dir_video+"LOGO_OP_4.mp4")
    LOGO_OP.Open()         # Open the reader
    LOGO_OP_clip = video_photo_clip(vid=LOGO_OP,layer=4,position=0,end=LOGO_OP.info.duration
                    ,location_y=-0.03,scale_x=0.8,scale_y=0.704)
    t.AddClip(LOGO_OP_clip)
    bg_head = openshot.FFmpegReader(dir_video+"complete_head_aispokesgirl.mp4")
    bg_head.Open()
    bg_head_clip = video_photo_clip(vid=bg_head,layer=2,position=0,end=LOGO_OP.info.duration,ck=ck)
    t.AddClip(bg_head_clip)
    main_timer += LOGO_OP.info.duration
    head_duration = LOGO_OP.info.duration
    bg_head.Close()
    LOGO_OP.Close()

    
    clip_duration=0
    photo_clip_list = [None]*len(text_content)
    img_list = [None]*len(text_content)
    anchor_clip_list = [None] * len(text_content)
    anchor_list = [None] * len(text_content)
    audio_clip_list = [None] * len(text_content)
    audio_list = [None] * len(text_content)
    sub_clip_list = [None] * len(text_content)
    sub_img_list = [None] * len(text_content)
    
    idx = 0
    for p in listdir(dir_photo+name_hash):
        
        anchor_list[idx] = openshot.FFmpegReader(dir_anchor+name_hash+"/"+str(idx)+".mp4")
        clip_duration = anchor_list[idx].info.duration
        anchor_list[idx].Open()
        anchor_clip_list[idx] = video_photo_clip(vid=anchor_list[idx],layer=4,scale_x=0.65,scale_y=0.65,
                location_x=0.35,location_y=0.25,position=main_timer, end=clip_duration,ck=ck_anchor,audio=False)
        t.AddClip(anchor_clip_list[idx])

        img_list[idx] = openshot.FFmpegReader(dir_photo+name_hash+'/'+p)
        img_list[idx].Open()
        photo_clip_list[idx] = video_photo_clip(vid=img_list[idx],layer=3
                ,scale_x=0.8,scale_y=0.704,location_y=-0.03,position=main_timer,end=clip_duration,audio=False)
        t.AddClip(photo_clip_list[idx])
        img_list[idx].Close()

        audio_list[idx] = openshot.FFmpegReader(dir_sound+name_hash+"/"+str(idx)+".mp3")
        audio_list[idx].Open()
        audio_clip_list[idx] = openshot.Clip(audio_list[idx])
        audio_clip_list[idx].Position(main_timer)
        audio_clip_list[idx].End(clip_duration)
        t.AddClip(audio_clip_list[idx])

        img_list[idx].Close()
        anchor_list[idx].Close()
        audio_list[idx].Close()
            
        sub_img_list[idx] = [None] * len(sub_list[idx])
        sub_clip_list[idx] = [None] * len(sub_list[idx])
        sub_timer = 0
        for sub_idx in range(len(sub_list[idx])):
            sub_img_list[idx][sub_idx] = openshot.QtImageReader(sub_list[idx][sub_idx]['path'])
            sub_img_list[idx][sub_idx].Open()
            sub_duration = 0.205*sub_list[idx][sub_idx]['count']
            sub_clip_list[idx][sub_idx] = video_photo_clip(vid=sub_img_list[idx][sub_idx], layer=6,location_x=0.069, location_y=0.89,position=main_timer+sub_timer,end=sub_duration)
            t.AddClip(sub_clip_list[idx][sub_idx])
            sub_img_list[idx][sub_idx].Close()
            sub_timer += sub_duration
            print(sub_list[idx][sub_idx]['path'])
        main_timer += clip_duration
        idx+=1
    
    LOGO_ED = openshot.FFmpegReader(dir_video+"LOGO_ED.avi")
    LOGO_ED.Open()
    LOGO_ED_clip = video_photo_clip(vid=LOGO_ED,layer=4,position=main_timer,end=LOGO_ED.info.duration+2
                    ,location_x=0.005,location_y=-0.031
                    ,scale_x=0.8,scale_y=0.6825)
    t.AddClip(LOGO_ED_clip)
    ED_duration = LOGO_ED.info.duration
    LOGO_ED.Close()
    

    bg = openshot.FFmpegReader(dir_video+"complete_double_aispokesgirl.mp4")
    bg.Open()
    bg_times = math.floor(main_timer+ED_duration/bg.info.duration)
    left_time = (main_timer+ED_duration) % bg.info.duration
    bg_clip_list = [None] * bg_times
    bg_list = [None] * bg_times
    bg.Close()
    bg_timer = head_duration
    for idx in range(bg_times):
        bg_list[idx] = openshot.FFmpegReader(dir_video+"complete_double_aispokesgirl.mp4")
        bg_list[idx].Open()
        bg_clip_list[idx] = video_photo_clip(bg_list[idx],layer=2,position=bg_timer
                ,end=bg_list[idx].info.duration,ck=ck)
        t.AddClip(bg_clip_list[idx])
        bg_timer += bg_list[idx].info.duration
        bg_list[idx].Close()
    bg_left = openshot.FFmpegReader(dir_video+"complete_double_aispokesgirl.mp4")
    bg_left.Open()
    bg_left_clip = video_photo_clip(bg_left,layer=2,position=bg_timer,end=left_time,ck=ck)
    t.AddClip(bg_left_clip)
    bg_left.Close()

    title = openshot.QtImageReader(dir_title+name_hash+".png")
    title.Open()         # Open the reader
    title_clip = video_photo_clip(vid=title, layer=4,location_x=-0.047, location_y=0.801,position=0,end=head_duration+main_timer)
    t.AddClip(title_clip)

    ####start building
    w = openshot.FFmpegWriter(tmp_video_dir+name_hash+".mp4")
    w.SetAudioOptions(True, "aac", 44100, 2, openshot.LAYOUT_STEREO, 3000000)
    w.SetVideoOptions(True, "libx264", openshot.Fraction(30000, 1000), 1280, 720,
        openshot.Fraction(1, 1), False, False, 3000000)
    w.Open()
    
    #may change duration into t.info.duration
    frames = int(t.info.fps)*int(head_duration+main_timer+ED_duration)
    for n in range(frames):
        f=t.GetFrame(n)
        w.WriteFrame(f)
        
    #notify_group(name+"的影片已經產生完成囉! www.choozmo.com:8168/"+video_sub_folder+name_hash+".mp4")
    t.Close()
    w.Close()
    print("video at : www.choozmo.com:8168/"+video_sub_folder+name_hash+".mp4")


def anchor_video_eng(name_hash,name,text_content, image_urls,sub_titles,avatar):
    file_prepare(name, name_hash, text_content,image_urls,'eng')
    sub_list=generate_subtitle_image_ENG(name_hash,sub_titles)
    
    for fname in range(len(text_content)):
        call_anchor(name_hash+"/"+str(fname),avatar)
        print('step finish')
    print('called............................................')

    ck=cKey(0,254,0,270)
    ck_anchor=cKey(0,255,1,320)
    duration = 0
    #average layer level is 3
    t = openshot.Timeline(1280, 720, openshot.Fraction(30000, 1000), 44100, 2, openshot.LAYOUT_STEREO)
    t.Open()

    main_timer = 0
    #add logo
    LOGO_OP = openshot.FFmpegReader(dir_video+"LOGO_OP_4.mp4")
    LOGO_OP.Open()         # Open the reader
    LOGO_OP_clip = video_photo_clip(vid=LOGO_OP,layer=4,position=0,end=LOGO_OP.info.duration
                    ,location_y=-0.03,scale_x=0.8,scale_y=0.704)
    t.AddClip(LOGO_OP_clip)
    #add background video  (head is different)
    bg_head = openshot.FFmpegReader(dir_video+"complete_head_aispokesgirl.mp4")
    bg_head.Open()
    bg_head_clip = video_photo_clip(vid=bg_head,layer=2,position=0,end=LOGO_OP.info.duration,ck=ck)
    t.AddClip(bg_head_clip)
    
    main_timer += LOGO_OP.info.duration
    head_duration = LOGO_OP.info.duration
    bg_head.Close()
    LOGO_OP.Close()

    #prepare empty list 
    clip_duration=0
    photo_clip_list = [None]*len(text_content)
    img_list = [None]*len(text_content)
    anchor_clip_list = [None] * len(text_content)
    anchor_list = [None] * len(text_content)
    audio_clip_list = [None] * len(text_content)
    audio_list = [None] * len(text_content)
    sub_clip_list = [None] * len(text_content)
    #openshot image holder
    sub_img_list = [None] * len(text_content)
    
    idx = 0
    for p in listdir(dir_photo+name_hash):
        
        anchor_list[idx] = openshot.FFmpegReader(dir_anchor+name_hash+"/"+str(idx)+".mp4")
        clip_duration = anchor_list[idx].info.duration
        anchor_list[idx].Open()
        anchor_clip_list[idx] = video_photo_clip(vid=anchor_list[idx],layer=4,scale_x=0.65,scale_y=0.65,
                location_x=0.35,location_y=0.25,position=main_timer, end=clip_duration,ck=ck_anchor,audio=False)
        t.AddClip(anchor_clip_list[idx])
        #insert image 
        img_list[idx] = openshot.FFmpegReader(dir_photo+name_hash+'/'+p)
        img_list[idx].Open()
        photo_clip_list[idx] = video_photo_clip(vid=img_list[idx],layer=3
                ,scale_x=0.81,scale_y=0.68,location_y=-0.03,position=main_timer,end=clip_duration,audio=False)
        t.AddClip(photo_clip_list[idx])
        img_list[idx].Close()
        #insert audio (speech)
        audio_list[idx] = openshot.FFmpegReader(dir_sound+name_hash+"/"+str(idx)+".mp3")
        audio_list[idx].Open()
        audio_clip_list[idx] = openshot.Clip(audio_list[idx])
        audio_clip_list[idx].Position(main_timer)
        audio_clip_list[idx].End(clip_duration)
        t.AddClip(audio_clip_list[idx])
        #insert subtitle
        sub_img_list[idx] = openshot.QtImageReader(sub_list[idx])
        sub_img_list[idx].Open()
        sub_clip_list[idx] = video_photo_clip(vid=sub_img_list[idx], layer=6,location_x=0.069, location_y=0.89,position=main_timer,end=clip_duration)
        t.AddClip(sub_clip_list[idx])

        img_list[idx].Close()
        anchor_list[idx].Close()
        audio_list[idx].Close()
        sub_img_list[idx].Close()
            
        main_timer += clip_duration
        idx+=1
    
    LOGO_ED = openshot.FFmpegReader(dir_video+"ED_ENG.mp4")
    LOGO_ED.Open()
    LOGO_ED_clip = video_photo_clip(vid=LOGO_ED,layer=4,position=main_timer,end=LOGO_ED.info.duration+2
                    ,location_x=0.005,location_y=-0.031
                    ,scale_x=0.8,scale_y=0.6825)
    t.AddClip(LOGO_ED_clip)
    ED_duration = LOGO_ED.info.duration
    LOGO_ED.Close()
    

    bg = openshot.FFmpegReader(dir_video+"complete_double_aispokesgirl.mp4")
    bg.Open()
    bg_times = math.floor(main_timer+ED_duration/bg.info.duration)
    left_time = (main_timer+ED_duration) % bg.info.duration
    bg_clip_list = [None] * bg_times
    bg_list = [None] * bg_times
    bg.Close()
    bg_timer = head_duration
    for idx in range(bg_times):
        bg_list[idx] = openshot.FFmpegReader(dir_video+"complete_double_aispokesgirl.mp4")
        bg_list[idx].Open()
        bg_clip_list[idx] = video_photo_clip(bg_list[idx],layer=2,position=bg_timer
                ,end=bg_list[idx].info.duration,ck=ck)
        t.AddClip(bg_clip_list[idx])
        bg_timer += bg_list[idx].info.duration
        bg_list[idx].Close()
    bg_left = openshot.FFmpegReader(dir_video+"complete_double_aispokesgirl.mp4")
    bg_left.Open()
    bg_left_clip = video_photo_clip(bg_left,layer=2,position=bg_timer,end=left_time,ck=ck)
    t.AddClip(bg_left_clip)
    bg_left.Close()

    title = openshot.QtImageReader(dir_title+name_hash+".png")
    title.Open()         # Open the reader
    title_clip = video_photo_clip(vid=title, layer=4,location_x=-0.047, location_y=0.801,position=0,end=head_duration+main_timer)
    t.AddClip(title_clip)

    ####start building
    w = openshot.FFmpegWriter(tmp_video_dir+name_hash+".mp4")
    w.SetAudioOptions(True, "aac", 44100, 2, openshot.LAYOUT_STEREO, 3000000)
    w.SetVideoOptions(True, "libx264", openshot.Fraction(30000, 1000), 1280, 720,
        openshot.Fraction(1, 1), False, False, 3000000)
    w.Open()
    
    #may change duration into t.info.duration
    frames = int(t.info.fps)*int(head_duration+main_timer+ED_duration)
    for n in range(frames):
        f=t.GetFrame(n)
        w.WriteFrame(f)
        
    #notify_group(name+"(ENG)的影片已經產生完成囉! www.choozmo.com:8168/"+video_sub_folder+name_hash+".mp4")
    t.Close()
    w.Close()
    print("video at : www.choozmo.com:8168/"+video_sub_folder+name_hash+".mp4")
    
    #line notifs

import pyttsx3
def make_speech(text):
    engine = pyttsx3.init()
    #voices = engine.getProperty('voices')
    engine.setProperty('voice', 'Mandarin')
    engine.save_to_file(text, '/app/speech.mp3')
    engine.runAndWait()