app.py

import time
import random
import os
import gradio as gr
import pandas as pd
import textwrap

import traceback

# 爬虫-------------
from save_cookie import save_cookie, get_cookie, cookie_f
from scrap_util import getDriver, titleLocInfo, find_key_paragrap, extract_from_driver, table_record_doc
import helium as hm
from postDouyin import senDouyin
# 模型-------------
from transformers import *

import pandas as pd
import os
import random
# import IPython.display as ipd
import numpy as np
# 视频-------------
import moviepy.video.io.ImageSequenceClip
from moviepy.editor import AudioFileClip,TextClip,CompositeVideoClip

import inspect, math
from PIL import Image, ImageDraw, ImageFont
from txtImgPost import  myPost, reshape_texts, generatePost
# 发布-------------


VIPtitle = "👑潜龙在渊输入内容生成视频✍️👒"
print(f">>>{VIPtitle}")
# print(f"秘钥文件路径:{cookie_f}")
mv_path = "./movie_output"
img_path = "./imgpost"
font_path = "./fonts"
movies = [os.path.join(mv_path,i) for i in os.listdir(mv_path) if i.endswith("mp4")]
exam_video = movies[0]
imgs_cur = [os.path.join(img_path,i) for i in os.listdir(mv_path) if i.endswith("png")]
templates_path = "./templates/"
templates = [os.path.join(templates_path,i) for i in os.listdir(templates_path) if i.endswith("csv")]
templates_name = [i.strip(templates_path).strip(".csv") for i in templates]
preview_templates = [os.path.join(templates_path,i) for i in os.listdir(templates_path) if i.endswith("jpg")]
# 字体------
font_list = [os.path.join(font_path,i) for i in os.listdir(font_path) if i.split(".")[1] in ["ttc",  "ttf", "otf"]]
background_img = "kaobianBottem.jpeg"
background_img = "./templates/zf_board_temp.jpeg"

a_ = Image.open(background_img)
WIDTH = 900
HEIGHT = 1400
WIDTH,HEIGHT = a_.size

# should load from files and build new from file
cookie_fns = ["抖音北京人事考试","抖音广东人事考试","抖音四川人事考试","抖音浙江人事考试","抖音江苏人事考试","抖音山东人事考试","抖音河南人事考试"]
cookie_fns = os.listdir("./cookie_list/")
cookie_fn = cookie_fns[0]
description = "URL--> 爬取-->解析--> 音频--> 图片--> 视频"

driver = getDriver()
# sub_url ="https://www.js.msa.gov.cn/art/2023/2/24/art_11436_1391666.html"
# hm.set_driver(driver)  # 给它一个selnuim driver
# hm.go_to(sub_url)
# html = driver.page_source
html = "<None>"

examples = [
    ["朝天区2023年上半年面向社会公开考试招聘事业单位工作人员公告-“四川•朝天”门户网",
     "报名时间。本次公开考试招聘采取网络报名方式进行，不组织现场报名。报名时间：2023年3月10日至3月17日24:00。报名网站：广元人事考试网（http://gypta.e21cn.com/）",
     "缴费时间。2023年05月27日 在 网上缴费",
     "考试时间-笔试时间  2.笔试分别于2023年4月8日、9日举行（具体时间、地点见《准考证》）。\
其中，4月8日笔试科目为《教育公共基础》，4月9日笔试科目为《卫生专业技术岗位人员公共科目笔试》、《综合知识》",
     "准考证领取。网上报名且缴费成功的报考者，凭身份证号、姓名，于2023年4月3日至4月7日24:00前登录报名网站打印准考证"
    ],
    ]

# 预测函数
# qa = pipeline("question-answering", model="uer/roberta-base-chinese-extractive-qa")
def custom_predict(context, question):
    answer_result = qa(context=context, question=question)
    answer = question + ": " + answer_result["answer"]
    score = answer_result["score"]
    return answer, score

def get_content_from_driver(task_link, content=None):
    driver = getDriver()
    hm.set_driver(driver)  # 给它一个selnuim driver
    if content:
        pass
    hm.go_to(task_link)
    print("hm.go_to(task_link)")
    time.sleep(1)
    items_ = driver.find_elements_by_xpath("//p")
    items_ = [i.text for i in items_ if i.text != ""]
    context_to_label =  "\n".join(items_)
    doc = extract_from_driver(driver)
    doc["url"] = task_link
    doc["content"] = context_to_label
    n_bm_sj,_ = custom_predict(context = context_to_label, question="报名时间")
    n_fee_sj,_ = custom_predict(context = context_to_label, question="缴费时间")
    n_ks_sj,_ = custom_predict(context = context_to_label, question="考试时间")
    n_zkz_sj,_ = custom_predict(context = context_to_label, question="准考证时间")
    need = [doc["title"],doc["zwlx"], doc["zwk_sheng"], doc["zwk_diqu"],
            doc["tidy_bm_sj"], doc["tidy_fee_sj"], doc["tidy_ks_sj"], doc["tidy_zkz_sj"],context_to_label,
            n_bm_sj, n_fee_sj, n_ks_sj, n_zkz_sj
           ]
    return need

def image_preview(orimage=None, text="Hello Ai", x=10, y=20, w=500, h=100, bac_color = "#FFbbFF",
                  txt_color = "#000000",front="simsun.ttc", size = 50):
    if orimage is None:
        return None
    size = int(size)
    image = Image.fromarray(orimage)
    draw = ImageDraw.Draw(image)
    draw.rectangle((x, y, x + w, y + h), fill=bac_color)
    # Add text to the text box
    font = ImageFont.truetype(front, size)
    text_x = x + size
    text_y = y + size
    words_perline = int(w / size)
    draw.text((text_x, text_y), text, font=font, fill=txt_color)
    return image

# 位置暂时由系统决定
def generate_image(bac_img, title,a,b,c,d,e,f,   #-- 当前文本行内字数可调，宽度可调 words_curline or 文本宽度
                   font_title, sz1, title_x, title_y, color1,   # for title
                   font_subtt, sz2, subtt_x, subtt_y, color2,
                   font_txt,   sz3, txt_x,  color3, bac_color):
    if bac_img is None:
        bac_img = "./kaobianBottem.jpeg"
    bing_post = "bingpost/"
    spacing = 20
    imgs_ = os.listdir(bing_post)
    imgs_ = [i for i in imgs_ if i.endswith("png")]
    ch_img = random.choice(imgs_)
    front_img = os.path.join(bing_post, ch_img)    
    postcard=myPost(front_img=front_img, img = bac_img)
    width, height = postcard.get_width_height()
    print(f"width, height :{width}  {height}")
    #  前景图， 待开放参数控制
    # postcard.drawFrontground() 
    # 放置标题  + 字体字号
    spacing=20
    # --理想总行字数
    words, rows = postcard.getLinesCount(sz=sz1, spacing=spacing) 
    words = (width - title_x) / ( sz1 + spacing) 
    words = int(words) + 1
    title_text = textwrap.wrap(title, width=words)
    title_text = [i.center(words) for i in title_text]
    postcard.postBoxText(title_text, font=font_title, sz = sz1,
                      x=title_x, 
                          # x=0, 
                     y=title_y, color = color1, spacing=spacing,  ali = "center",
                         bac_color=None)

    # 子标题 + 内容的处理  + 每行字数
    # x=40
    y = subtt_y  #  标题与剩下内容高度
    #- -- 扣除起始位置 能放字数 ---
    words = int((width) / sz3)
    for k,contents in enumerate([a,b,c,d,e,f]):
        if len(contents) < 1:
            continue
        lines = contents.split("。")
        if len(lines) < 2:
            continue
        sub_title = lines[0]
        sub_texts = "。".join(lines[1:])
        words = int((width) / sz3) + 2
        sub_text = textwrap.wrap(sub_texts, width=words)
        # sub_text = [i.center(words) for i in sub_text]
        # -----文本框效果--------
        # draw.rectangle((x, y, x + w, y + h), fill=bac_color)
        postcard.postBoxText([sub_title], font=font_subtt, sz = sz2,
                      x=subtt_x, y=y, color = color2, spacing=spacing,
                ali = "left" ,bac_color=bac_color)
        y += int(sz2*2.5)
        postcard.postBoxText(sub_text, font=font_txt,  sz = sz3,
                      x=txt_x, y=y, color = color3, spacing=spacing,
                             ali = "left", bac_color=None)
        y += (sz3 + spacing) * len(sub_text) + 50
    return postcard.get_res()

def generate_template( pre_img, input_image,  mt_name,
                    font_title, sz1, title_x, title_y, color1,
                   font_subtt, sz2, subtt_x, subtt_y, color2,
                   font_txt,   sz3, txt_x,  color3, bac_color):
    if pre_img is None:
        print("没调试好，不能保存")
        return None
    if type(pre_img) is np.ndarray:
        template_img = f"./templates/{mt_name}.png"
        template_img_preview = f"./templates/{mt_name}.jpg"
        print("保存图片", template_img)
        pil = Image.fromarray(input_image)
        pil_2 = Image.fromarray(pre_img)
        pil_2.save(template_img_preview)
        pil.save(template_img)
    print("开始保存模板")
    # 保存模板预览效果
    template_name = f"./templates/{mt_name}.csv"
    name = ["template_img","font_title", "sz1", "title_x", "title_y", "color1",
        "font_subtt", "sz2", "subtt_x", "subtt_y", "color2",
        "font_txt", "sz3", "txt_x", "color3", "bac_color"]
    value = [[template_img, font_title, sz1, title_x, title_y, color1,
                   font_subtt, sz2, subtt_x, subtt_y, color2,
                   font_txt,   sz3, txt_x,  color3, bac_color]]
    record = pd.DataFrame(data=value, columns=name)
    record.to_csv(template_name, index=False)
    return [os.path.join(templates_path,i) for i in os.listdir(templates_path) if i.endswith("jpg")]

def load_template(mt_name):
    template_name = f"./templates/{mt_name}.csv"
    template_img = f"./templates/{mt_name}.png"
    if not os.path.exists(template_name):
        print("error on dealing this template name :", mt_name)
        return None
    template_df = pd.read_csv(template_name)
    # print(template_df)
    # return ["font_title", 12, 1, 1, "color1",
    #     "font_subtt", 12, 123, 1, "color2",
    #     "font_txt", 123, 12, "color3"]
    # res = [template_img]
    # res.extend()
    res = list(template_df.values[0])
    return res
   
# def change_textbox(choice):
#     #根据不同输入对输出控件进行更新
#     if choice == "short":
#         return gr.update(lines=2, visible=True, value="Short story: ")
#     elif choice == "long":
#         return gr.update(lines=8, visible=True, value="Long story...")
#     else:
#         return gr.update(visible=False)
# with gr.Blocks() as demo:
#     radio = gr.Radio(
#         ["short", "long", "none"], label="Essay Length to Write?"
#     )
#     text = gr.Textbox(lines=2, interactive=True)
#     radio.change(fn=change_textbox, inputs=radio, outputs=text)
# demo.launch()

    
key_index =  ["bm_sj","fee_sj","ks_sj","zkz_sj"]
def generate_mv(bac_img, title, a,b,c,d,e,f, sz1, sz2, sz3, spacing, color1, color2):
    global movies, imgs_cur
    bing_dir = "./bingpost"
    front_imgs =[os.path.join(bing_dir,i) for i in os.listdir("./bingpost") if i.endswith(".png")]
    front_img = random.choice(front_imgs)
    if bac_img is None:
        bac_img = "./kaobianBottem.jpeg"
    imge_dir = "imgpost"
    fps = 1
    image_dur = 3
    movie_dir = "movie_output"

    msg = []
    new_title = title.replace('\n', "").replace('”', "").replace('“', "").replace(' ', "")
    output_wav = f"audio_output/{new_title}.wav"
    tts_cmd = f'edge-tts  --voice zh-CN-XiaoxiaoNeural  --rate=+10% --text "{title}" --write-media {output_wav} 2>&1 >/dev/null'
    _m = os.popen(tts_cmd).read()
    msg.append(_m)
    # image
    audio_file = AudioFileClip(output_wav)
    movie_file = f"{movie_dir}/{new_title}.mp4"
    img_files = []
    mv_image_files = []
    
    for k,lines in enumerate([title,a,b,c,d]):
        k = key_index[k]
        if not lines:
            # 这个关键信息没有，跳过
            continue
        if len(lines) < 2:
            continue
        lines = lines.split("\n")
        filename = f"{k}_{new_title}.png"
        filename = os.path.join(imge_dir, filename)
        img_files.append(filename)
        # --- how long a image last display
        for i in range(image_dur):   
            mv_image_files.append(filename)
        img = generatePost(lines=lines,front_img = front_img, bac_img = bac_img, fn_name = filename)

    # clip_img = moviepy.video.io.ImageSequenceClip.ImageSequenceClip(img_files, fps=fps) #durations=audio_file.duration ) #
    if len(img_files) < 1:
        print("error, this url not get a valid key info:{new_title},url:{url_}")
        return "","",""
    clip_img = moviepy.video.io.ImageSequenceClip.ImageSequenceClip(mv_image_files, fps=fps, durations=audio_file.duration ) #
    clip_img = clip_img.set_audio(audio_file)
    clip_img.write_videofile(movie_file, fps=fps)#, audio_codec="aac")
    print(movie_file, ":done")
    movies = [os.path.join(mv_path,i) for i in os.listdir(mv_path) if i.endswith("mp4")]
    # imgs = [os.path.join(img_path,i) for i in os.listdir(mv_path) if i.endswith("png")]
    imgs_cur = img_files
    return movie_file, img_files

def exit_func():
    driver.close()
    time.sleep(3)
    exit(1)

# def refresh_template():
#     global templates_name
#     templates_name = [i.strip(templates_path).strip(".csv") for i in templates]
#     return templates_name
    
# 登录抖音
def loginDouyin():
    url = 'https://creator.douyin.com/'
    hm.set_driver(driver)  # 给它一个selnuim driver
    hm.go_to(url)
    # driver.get_screenshot_as_file("1.png")
    print("-------------------请扫描二维码登录抖音创作者中心-------------------")
    hm.click(hm.Text("登录"))
    time.sleep(1)
    hm.click(hm.Text("确认"))
    time.sleep(1)
    qr_element = driver.find_element_by_class_name("qrcode-image")
    qr_element.click()
    qr_element.screenshot("qr.png")
    return "qr.png"

def run_save_cookie(account_name):
    cookie_fns.append(account_name)
    cookie_file_name = cookie_f.format(account_name)
    save_cookie(driver, cookie_file_name)
    gr.Dropdown.update(choices=account_fn, value=cookie_fns)
    return "succ"
    
def mySendDouyin(cookie_fn, movie_file):
    driver = getDriver()
    # new_title = title.replace('\n', "").replace('”', "").replace('“', "").replace(' ', "")
    # movie_file = f"{movie_dir}/{new_title}.mp4"
    print(f"in mySendDouyin func: try send {movie_file}")
    if not os.path.exists(movie_file):
        raise f"movie_file not exists"
    print(movie_file)
    # senDouyin(account_name = cookie_fn, title = title, movie_path = movie_file, loc = "北京")
    for i in range(5):
        try:
            senDouyin(driver = driver, account_name = cookie_fn, title = title, movie_path = movie_file, loc = "北京")
            print(f"accunt {cookie_fn} 发布 {movie_file} succ")
            break
        except Exception as e:
            traceback.print_exc()
            driver.get_screenshot_as_file("1.png")
            print("---mySendDouyin faild, continue try--")
            continue
    return "succ"

# 构建Blocks上下文 =======================================================================
with gr.Blocks() as demo:
    with gr.Tabs():
        # --------------generate movie -------------      
        with gr.TabItem("生成 & 预览"):
            gr.Markdown(f"# {VIPtitle}")
            gr.Markdown(f"{description}")
            with gr.Row():       # 行排列        
                with gr.Column():    # 列排列
                    title = gr.Textbox("朝天区2023招聘事业单位工作人员公告",label="标题",interactive=True)
                    a = gr.Textbox("xxxx",label="a")
                    b = gr.Textbox(lines=2, value = "报名时间。2023年3月10日至3月17日24:00。", label="b",interactive=True)
                    c = gr.Textbox(lines=2, value = "缴费时间。2023年05月27日 ", label="c",interactive=True)
                    d = gr.Textbox(lines=2, value = "子标题1。笔试分别于2023年4月8日", label="d",interactive=True)
                    e = gr.Textbox(lines=2, value = "子标题2。广元人事考试网打印本人准考证。", label="e",interactive=True)
                    f = gr.Textbox(lines=2, value =  "子标题3。 见准考证。", label="f",interactive=True)
                    
                # movie_file = gr.Video(exam_video,label="movie")
                with gr.Column():
                    # -- 输入背景图底图 作为画布 --
                    input_image = gr.Image(background_img, label = "背景图， input",interactive=True)
                    title_x = gr.Slider(1, WIDTH, label = "标题左右移动", value=60, step = 5)
                    title_y = gr.Slider(1, HEIGHT, label = "标题上下移动", value=320, step = 5)
                    font_title = gr.Dropdown(choices = font_list, label = "标题字体", 
                                    value = "fonts/特创易新恒LOGO美黑.ttf")
                    sz1 = gr.Slider(1, 100, label = "标题大小", value=80, step = 5)
                    color1 = gr.Textbox("blue", label = "标题颜色")
                    bac_color = gr.Textbox("blue", label = "文本框颜色 空不设置文本框")

                with gr.Column():
                    # -- 子标题
                    font_subtt = gr.Dropdown(choices = font_list, label = "子标题字体", value = "fonts/吉页颉润体简.ttf")
                    sz2 = gr.Slider(1, 100, label = "subTitleSZ", value=55, step = 5)
                    subtt_x = gr.Slider(1, WIDTH, label = "子标题左右-->", value=80, step = 5)
                    subtt_y = gr.Slider(1, HEIGHT, label = "子标题 上下 ^  v", value=780, step = 5)
                    color2 = gr.Textbox("#007000", label = "子标题颜色")
                    font_txt = gr.Dropdown(choices = font_list, label = "文本字体", value = "fonts/simsun.ttc")
                    sz3 = gr.Slider(1, 100, label = "wordSize", value=40, step = 5)
                    txt_x = gr.Slider(1, WIDTH, label = "内容位置_x", value=80, step = 5)
                    color3 = gr.Textbox("#101010", label = "文字颜色")
            
            with gr.Row():
                with gr.Column():       # ----按钮控件------
                    pre_img_bt = gr.Button("预览")
                    record_template = gr.Button("导出模板")
                    mt_name = gr.Textbox("模板_v1",label="导出模板必填名字")
                    # font_txt = gr.Dropdown(choices = font_list, label = "文本字体", value = "fonts/simsun.ttc")
                    mt_selected = gr.Dropdown(choices = templates_name, label = "选择模板载入")
                    load_mt = gr.Button("载入模板参数")
                # submit = gr.Button("生成")
                # account_fn = gr.Dropdown(choices=cookie_fns, label = "账号选择", value = "抖音广东人事考试")
                # with gr.Column():
                pre_img2 = gr.Image(label="生成结果预览", interactive=False)
                with gr.Row():
                    gallery = gr.Gallery(
                        value = preview_templates,
                        label="preview", show_label=True, elem_id="gallery"
                        ).style(columns=[3], rows=[2], object_fit="contain", height="auto")
            
                    # mv_files = gr.Files(movies, label="movies")
                    # img_files = gr.Files(imgs_cur, label="movie_imgs")
        with gr.TabItem("生成&发布"):
            gr.Markdown(f"{description}")
            with gr.Row():       # 行排列        
                with gr.Column():    # 列排列
                    title1 = gr.Textbox("朝天区2023招聘事业单位工作人员公告",label="标题")
                    a1 = gr.Textbox("xxxx",label="a")
                    b1 = gr.Textbox(lines=2, value = "报名时间。2023年3月10日至3月17日24:00。", label="b")
                    c1 = gr.Textbox(lines=2, value = "缴费时间。2023年05月27日 ", label="c")
                    d1 = gr.Textbox(lines=2, value = "子标题1。笔试分别于2023年4月8日", label="d")
                    e1 = gr.Textbox(lines=2, value = "子标题2。广元人事考试网打印本人准考证。", label="e")
                    f1 = gr.Textbox(lines=2, value =  "子标题3。 见准考证。", label="f")
                    
                with gr.Column():
                    # 上传做好的图进行生成
                    input_image_2 = gr.Image(background_img, label = "背景图， input",interactive=True)
                    
            with gr.Row():
                with gr.Column():       # ----按钮控件------
                    post_img_bt = gr.Button("test_post")
                    refresh_template_bt = gr.Button("刷新模板")
                    # mt_name = gr.Textbox("模板_v1",label="导出模板必填名字")
                    mt_name_selected = gr.Dropdown(choices = templates_name, label = "选择模板进行生产")
                # pre_img2 = gr.Image(label="生成结果预览", interactive=False)

        with gr.TabItem("登录账号"):
            with gr.Row():
                login_ = gr.Button("登录抖音")
                login_save = gr.Button("登录完成保存")                
            with gr.Row():
                account_name_new = gr.Textbox(label="新建账号")
                login_qr = gr.Image("qr.png", shape=(200, 200), label="登录账号")
                # gr.Dropdown.update(choices=cookie_fns, value=cookie_fn)
        with gr.TabItem("todo内嵌浏览器pyqt html"):
            mp = gr.HTML(html, elem_id="coords", visible=True)
            pass

    pre_img_bt.click(fn=generate_image, inputs=[input_image, title,a,b,c,d,e,f, 
                   font_title, sz1, title_x, title_y, color1,
                   font_subtt, sz2, subtt_x, subtt_y, color2,
                   font_txt,   sz3, txt_x,  color3, bac_color],
                  outputs=[pre_img2])
    record_template.click(fn=generate_template, inputs=[pre_img2, input_image, mt_name,
                   font_title, sz1, title_x, title_y, color1,
                   font_subtt, sz2, subtt_x, subtt_y, color2,
                   font_txt,   sz3, txt_x,  color3, bac_color],
                  outputs=[gallery])
    load_mt.click(fn=load_template, 
                  inputs = [mt_selected],
                  outputs=[input_image, font_title, sz1, title_x, title_y, color1,
                   font_subtt, sz2, subtt_x, subtt_y, color2,
                   font_txt,   sz3, txt_x,  color3, bac_color])
    
    # refresh_template_bt.click(fn=refresh_template, inputs=None, outputs=[templates_name])
    # template_params = []
    # post_img_bt.click(inputs = [title, a,b,c,d,e,f])  # 用于追加模板参数，简洁版本
    # submit.click(fn=generate_mv, inputs=[title, a,b,c,d], 
    #              outputs=[img_files])
                 # outputs=[movie_file, img_files])
    # exit_.click(fn=exit_func)
    login_.click(fn=loginDouyin, outputs=[login_qr])
    login_save.click(fn=run_save_cookie, inputs = [account_name_new])
    # post_.click(fn=mySendDouyin, inputs = [account_fn, movie_file])
    # 绑定clear点击函数
    # clear.click(fn=clear_input, inputs=[], outputs=[context, question, answer, score])

if __name__ == "__main__":
    demo.queue().launch(share=True)
    print("run from current ")