ai-spokesgirl/openshot_word.py

import openshot
import os
import re
from PIL import Image,ImageDraw,ImageFont

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_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_photo_clip(video=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(video)
    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 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 txt2image(content, save_target,lang='zh',size=26,fon="font/DFT_B7.ttc"):
    unicode_text = trim_punctuation(content)
    font = ''
    if lang=='zh':
        font = ImageFont.truetype(font=fon, size=size)
    else :
        font = ImageFont.truetype(font="font/arial.ttf", size=size)
    text_width, text_height = font.getsize(unicode_text)
    W, H = (1280,500)
    canvas = Image.new('RGBA', (W, H), (255, 255, 255, 0) )
    draw = ImageDraw.Draw(canvas)
    w, h = draw.textsize(content,font = font)
    text= unicode_text
    draw.text(((W-w)/2,0), text,'black', font)
    canvas.save(save_target, "PNG")

def text_to_short_vedio(bg,text_form,vedio_time):
    t = openshot.Timeline(1280, 720, openshot.Fraction(30000, 1000), 44100, 2, openshot.LAYOUT_STEREO)
    t.Open()
    
    # 去背參數
    ck = cKey(0, 254, 0, 270)
    ck_anchor = cKey(0, 255, 0, 320)

    anchor = openshot.FFmpegReader(bg)
    anchor.Open()
    anchor_clip = video_photo_clip(video=anchor,layer=2,scale_x=1,scale_y=1,
            location_x=0,location_y=0,position=0, end=vedio_time,ck=ck_anchor,audio=True)
    t.AddClip(anchor_clip)
    anchor.Close()

    for text_tmp in text_form:
        print(text_tmp['text'])
        time = float(text_tmp['end']) - float(text_tmp['start'])
        file_name = "tmp/save_target_" + text_tmp['text'] + ".png"
        txt2image(text_tmp['text'], file_name,lang='zh',size = text_tmp['size'])
        exec('text_anchor_{} = openshot.QtImageReader("tmp/save_target_{}.png")'.format(text_tmp['text'],text_tmp['text']))
        exec('text_anchor_{}.Open()'.format(text_tmp['text']))
        exec('text_anchor_{}.Open()'.format(text_tmp['text']))
        exec('text_anchor_clip_{} = video_photo_clip(video=text_anchor_{},layer=4,scale_x=1,scale_y=1,\
                location_x=0,location_y=0.7,position=text_tmp["start"], end=time,ck=ck_anchor,audio=True)'.format(text_tmp['text'],text_tmp['text']))
        exec('t.AddClip(text_anchor_clip_{})'.format(text_tmp['text']))
        exec('text_anchor_{}.Close()'.format(text_tmp['text']))
        exec('os.remove("tmp/save_target_{}.png")'.format(text_tmp['text']))
        


    w = video_writer_init("output/test2.mp4")
    w.Open()

    frames = int(t.info.fps)*int(vedio_time)
    for n in range(frames):
        f=t.GetFrame(n)
        w.WriteFrame(f)

    t.Close()
    w.Close()


if __name__ == '__main__':
    text_form = [{'text':"texttexttext",'start':0,'end':3,'size':26},{'text':"test22222222",'start':4,'end':6,'size':26}]
    text_to_short_vedio("input/bg/dog.mp4",text_form,6)