LongArticlesJob/coldStartTasks/multi_modal/generate_text_from_video.py

"""
@author: luojunhui
"""
import os
import time
import traceback

import requests

from pymysql.cursors import DictCursor
from tqdm import tqdm

from applications.api import GoogleAIAPI
from applications.const import VideoToTextConst
from applications.db import DatabaseConnector
from config import long_articles_config
from config import apolloConfig

# 办公室网络调试需要打开代理
# os.environ["HTTP_PROXY"] = "http://192.168.100.20:1087"
# os.environ["HTTPS_PROXY"] = "http://192.168.100.20:1087"

const = VideoToTextConst()
config = apolloConfig(env="prod")

# pool_size
POOL_SIZE = int(config.getConfigValue("video_extract_pool_size"))
# batch_size
BATCH_SIZE = int(config.getConfigValue("video_extract_batch_size"))


def download_file(pq_vid, video_url):
    """
    下载视频文件
    """
    file_name = "static/{}.mp4".format(pq_vid)
    if os.path.exists(file_name):
        return file_name

    proxies = {
        "http": None,
        "https": None
    }
    with open(file_name, 'wb') as f:
        response = requests.get(video_url, proxies=proxies)
        f.write(response.content)
    return file_name


class GenerateTextFromVideo(object):
    """
    从视频中生成文本
    """
    def __init__(self):
        self.google_ai_api = GoogleAIAPI()
        self.db = DatabaseConnector(db_config=long_articles_config)

    def connect_db(self):
        """
        连接数据库
        """
        self.db.connect()

    def input_task_list(self):
        """
        输入任务列表, 从single_video_pool中获取
        """
        sql = f"""
        select article_title, concat('https://rescdn.yishihui.com/', video_oss_path ) as video_url, audit_video_id
        from publish_single_video_source 
        where audit_status = {const.AUDIT_SUCCESS_STATUS} and bad_status = {const.ARTICLE_GOOD_STATUS} and extract_status = {const.EXTRACT_INIT_STATUS}
        order by id desc;
        """
        task_list = self.db.fetch(sql, cursor_type=DictCursor)
        insert_sql = f"""
        insert ignore into video_content_understanding
            (pq_vid, video_ori_title, video_oss_path)
        values (%s, %s, %s);
        """
        affected_rows = self.db.save_many(
            insert_sql,
            params_list=[(i['audit_video_id'], i['article_title'], i['video_url']) for i in task_list]
        )
        print(affected_rows)

    def update_video_status(self, ori_status, new_status, pq_vid):
        """
        更新视频状态
        """
        sql = f"""
            update video_content_understanding
            set status = %s, status_update_timestamp = %s
            WHERE pq_vid = %s and status = %s;
        """
        affected_rows = self.db.save(
            query=sql,
            params=(new_status, pq_vid, ori_status, int(time.time()))
        )
        return affected_rows

    def upload_video_to_google_ai(self, max_processing_video_count=POOL_SIZE):
        """
        上传视频到Google AI
        max_processing_video_count: 处理中的最大视频数量，默认20
        video_content_understanding 表status字段
        0: 未处理
        1: 处理中
        2: 处理完成
        """
        # 查询出在视频处于PROCESSING状态的视频数量
        select_sql = f"""
            select count(1) as processing_count 
            from video_content_understanding 
            where status = {const.VIDEO_UNDERSTAND_PROCESSING_STATUS};
        """
        count = self.db.fetch(select_sql, cursor_type=DictCursor)[0]['processing_count']
        rest_video_count = max_processing_video_count - count
        success_upload_count = 0
        if rest_video_count:
            sql = f"""
                select pq_vid, video_oss_path 
                from video_content_understanding 
                where status = {const.VIDEO_UNDERSTAND_INIT_STATUS} 
                limit {rest_video_count};
            """

            task_list = self.db.fetch(sql, cursor_type=DictCursor)
            for task in tqdm(task_list, desc="upload_video_task"):
                lock_rows = self.update_video_status(
                    ori_status=const.VIDEO_UNDERSTAND_INIT_STATUS,
                    new_status=const.VIDEO_LOCK,
                    pq_vid=task['pq_vid'],
                )
                if not lock_rows:
                    continue
                try:
                    file_path = download_file(task['pq_vid'], task['video_oss_path'])
                    google_upload_result = self.google_ai_api.upload_file(file_path)
                    if google_upload_result:
                        file_name, file_state, expire_time = google_upload_result
                        update_sql = f"""
                            update video_content_understanding
                            set status = %s, file_name = %s, file_state = %s, file_expire_time = %s
                            where pq_vid = %s and status = %s;
                        """
                        self.db.save(
                            update_sql,
                            params=(
                                const.VIDEO_UNDERSTAND_PROCESSING_STATUS,
                                file_name,
                                file_state,
                                expire_time,
                                task['pq_vid'],
                                const.VIDEO_LOCK
                            )
                        )
                        success_upload_count += 1

                except Exception as e:
                    print("task upload failed because of {}".format(e))
                    print("trace_back: ", traceback.format_exc())
                    # roll back status
                    self.update_video_status(
                        ori_status=const.VIDEO_LOCK,
                        new_status=const.VIDEO_UNDERSTAND_INIT_STATUS,
                        pq_vid=task['pq_vid'],
                    )

        return success_upload_count

    def delete_video_from_google(self, file_name):
        """
        删除视频文件
        """
        self.google_ai_api.delete_video(file_name)

    def get_tasks(self):
        """
        获取处理视频转文本任务
        """
        sql = f"""
            select pq_vid, file_name 
            from video_content_understanding 
            where status = {const.VIDEO_UNDERSTAND_PROCESSING_STATUS} 
            order by file_expire_time 
            limit {BATCH_SIZE};
        """
        task_list = self.db.fetch(sql, cursor_type=DictCursor)
        return task_list

    def convert_video_to_text_with_google_ai(self):
        """
        处理视频转文本任务
        """
        task_list = self.get_tasks()
        while task_list:
            for task in tqdm(task_list, desc="convert video to text"):
                print(task['pq_vid'], task['file_name'])
                # LOCK TASK
                lock_row = self.update_video_status(
                    ori_status=const.VIDEO_UNDERSTAND_PROCESSING_STATUS,
                    new_status=const.VIDEO_LOCK,
                    pq_vid=task['pq_vid'],
                )
                if not lock_row:
                    continue
                file_name = task['file_name']
                video_local_path = "static/{}.mp4".format(task['pq_vid'])
                google_file = self.google_ai_api.get_google_file(file_name)
                state = google_file.state.name
                match state:
                    case 'ACTIVE':
                        try:
                            video_text = self.google_ai_api.get_video_text(
                                prompt="分析我上传的视频的画面和音频，用叙述故事的风格将视频所描述的事件进行总结，需要保证视频内容的完整性，并且用中文进行输出，直接返回生成的文本。",
                                video_file=google_file
                            )
                            if video_text:
                                update_sql = f"""
                                    update video_content_understanding
                                    set status = %s, video_text = %s, file_state = %s
                                    where pq_vid = %s and status = %s;
                                """
                                self.db.save(
                                    update_sql,
                                    params=(
                                        const.VIDEO_UNDERSTAND_SUCCESS_STATUS,
                                        video_text,
                                        state,
                                        task['pq_vid'],
                                        const.VIDEO_LOCK
                                    )
                                )
                                # delete local file and google file
                                if os.path.exists(video_local_path):
                                    os.remove(video_local_path)

                                tqdm.write("video transform to text success, delete local file")
                                task_list.remove(task)

                                self.google_ai_api.delete_video(file_name)
                                tqdm.write("delete video from google success: {}".format(file_name))
                            else:
                                # roll back status
                                self.update_video_status(
                                    ori_status=const.VIDEO_LOCK,
                                    new_status=const.VIDEO_UNDERSTAND_PROCESSING_STATUS,
                                    pq_vid=task['pq_vid'],
                                )
                        except Exception as e:
                            # roll back status
                            self.update_video_status(
                                ori_status=const.VIDEO_LOCK,
                                new_status=const.VIDEO_UNDERSTAND_PROCESSING_STATUS,
                                pq_vid=task['pq_vid'],
                            )
                            tqdm.write(str(e))
                            continue

                    case 'PROCESSING':
                        tqdm.write("video is still processing")
                        # roll back status
                        self.update_video_status(
                            ori_status=const.VIDEO_LOCK,
                            new_status=const.VIDEO_UNDERSTAND_PROCESSING_STATUS,
                            pq_vid=task['pq_vid'],
                        )

                    case 'FAILED':
                        self.update_video_status(
                            ori_status=const.VIDEO_LOCK,
                            new_status=const.VIDEO_UNDERSTAND_FAIL_STATUS,
                            pq_vid=task['pq_vid']
                        )

                        if os.path.exists(video_local_path):
                            os.remove(video_local_path)

                        self.google_ai_api.delete_video(file_name)
                        tqdm.write("video process failed, delete local file")

                time.sleep(const.SLEEP_SECONDS)

            tqdm.write("执行完一轮任务，剩余数量：{}".format(len(task_list)))
            time.sleep(const.SLEEP_SECONDS)