123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219 |
- import json
- import os
- from urllib.parse import urlencode
- import pandas as pd
- import requests
- import yaml
- from common_func import get_db_engine, mysql_replace_into, NpEncoder
- import datetime
- from config.url import voice_to_script_task_submit_url, voice_to_script_task_result_url, get_material_info_from_ocean_engine_url, \
- get_video_info_from_ocean_engine_url
- import time
- def get_material_info(project_name, period_type):
- material_df = pd.DataFrame()
- has_more = True # 是否还存在分页数据, 初始化为 True
- limit = 10 # 每页获取30条
- page = 1 # 第几页
- while has_more:
- request_data = {'list_type': 1,
- 'material_type': 3,
- 'order_by': 'click_show_rate',
- 'period_type': period_type,
- 'aggr_app_code': 4,
- 'aggr_category_list': '[]',
- 'video_type': '[]',
- 'keywords': project_name,
- 'landing_type': '[]',
- 'limit': limit,
- 'page': page,
- 'video_duration_type': 5}
- request_path = get_material_info_from_ocean_engine_url + '?' + urlencode(request_data)
- request = requests.get(request_path)
- result = json.loads(request.text)
- material_page_df = pd.DataFrame(result['data']['materials'])
- material_df = material_df.append(material_page_df)
- if result.get('code') == 0 and result.get('data').get('has_more') is True:
- page += 1
- else:
- has_more = False
- # 数据类型的处理,方便入库
- # metrics dict to str
- # title list to str
- # video_type list to str
- # watermarks list to str
- material_df[['metrics', 'title', 'video_type', 'watermarks']] = \
- material_df[['metrics', 'title', 'video_type', 'watermarks']].astype(str)
- material_df.rename(columns={'vid': 'signature'}, inplace=True)
- # 添加项目名称和日期
- material_df['project_name'] = project_name
- material_df['stat_date'] = datetime.datetime.today().strftime('%Y-%m-%d')
- # 写入数据库
- material_df.to_sql(name="ctop_ai_material_info_from_ocean_engine",
- con=write_engine,
- if_exists='append',
- index=False,
- chunksize=chunk_size,
- method=mysql_replace_into)
- return material_df
- def get_video_info(vid, project_name):
- """
- 为了提高数据获取的完整性,每次只请求10条数据
- :param vid:
- :param project_name:
- :return:
- """
- video_df = pd.DataFrame()
- # 每次请求的视频个数
- cnt_per_request = 10
- # 总的视频个数
- total_cnt = len(vid)
- for i in range(0, total_cnt, cnt_per_request):
- if i + cnt_per_request < total_cnt:
- query_ids = vid[i: i + cnt_per_request]
- else:
- query_ids = vid[i:]
- request_data = {"query_ids": query_ids, "water_mark": "creative_center"}
- request = requests.post(url=get_video_info_from_ocean_engine_url,
- headers={'Content-Type': 'application/json'},
- data=json.dumps(request_data, cls=NpEncoder)
- )
- response_data = json.loads(request.text)
- if response_data.get('code') == 0 and response_data.get('data'):
- for key, value in response_data['data'].items():
- single_dict = value
- single_dict['signature'] = key
- single_df = pd.DataFrame([single_dict])
- video_df = video_df.append(single_df)
- # 数据类型的处理,方便入库
- # play_info list to str
- video_df['play_info'] = video_df['play_info'].astype(str)
- video_df.drop(labels='video_id', axis=1, inplace=True)
- # 添加项目名称和日期
- video_df['project_name'] = project_name
- video_df['stat_date'] = datetime.datetime.today().strftime('%Y-%m-%d')
- # 写入数据库
- video_df.to_sql(name="ctop_ai_video_info_from_ocean_engine",
- con=write_engine,
- if_exists='append',
- index=False,
- chunksize=chunk_size,
- method=mysql_replace_into)
- return video_df
- def submit_script_task(df):
- """
- 向腾讯云提交语音转脚本的任务
- :param df: DataFrame columns 包含 signature 和 url
- :return: task_ids
- """
- # 1 获取已经被提交过的任务
- sql = """select md5 signature from tb_asr_result """
- submitted_task_df = pd.read_sql(sql, read_engine)
- # 2 需要提交的任务,去掉历史被提交过的任务,防止重复提交浪费服务时长
- to_submit_task_df = df[~df.signature.isin(submitted_task_df.signature.values)]
- # 3 发送请求,提交任务
- for index, row in to_submit_task_df.iterrows():
- material_md5 = row['signature']
- material_url = row['video_url']
- request_data = {"md5": material_md5, "url": material_url}
- request_full_path = voice_to_script_task_submit_url + '?' + urlencode(request_data)
- request = requests.post(request_full_path)
- try:
- result = json.loads(request.text)
- print(result)
- except:
- print("error", request.text)
- # 4 获取素材对应的发送请求的 task_id
- sql = """
- select task_id from tb_asr_result where md5 in %s
- """ % (tuple(df.signature.values),)
- task_id_df = pd.read_sql(sql, read_engine)
- task_ids = task_id_df['task_id'].values
- return task_ids
- def get_result_from_tx(task_id_lst):
- """
- 从腾讯云获取脚本
- 每隔5分钟获取一次,直到没有 执行中或者等待执行 的任务为止
- Status Integer 任务状态码,0:任务等待,1:任务执行中,2:任务成功,3:任务失败。
- StatusStr String 任务状态,waiting:任务等待,doing:任务执行中,success:任务成功,failed:任务失败。
- ErrorMsg String 失败原因说明。
- """
- while True:
- print("sleep 5 mins")
- time.sleep(60 * 5)
- sql = """select task_id, task_status from tb_asr_result where task_id in %s and task_status in (0,1)""" % (tuple(task_id_lst),)
- task_status_df = pd.read_sql(sql, read_engine)
- if task_status_df.empty:
- break
- for task_id in task_status_df.task_id.values:
- request_data = {'task_id': task_id}
- request_full_path = voice_to_script_task_result_url + '?' + urlencode(request_data)
- request = requests.post(request_full_path)
- try:
- result = json.loads(request.text)
- print(task_id, result['status'])
- except:
- print("error", task_id, request.text)
- if __name__ == '__main__':
- # 1 读取配置文件
- with open('config/config.yaml', mode='r', encoding='utf-8') as f:
- config = yaml.load(f.read(), Loader=yaml.FullLoader)
- # 1-1 数据库连接引擎,依据开发环境/生产环境 进行切换
- # 读数据库引擎使用生产数据库,写数据库引擎依据系统环境进行切换(测试数据库/生产数据库)
- # 注意: 该项目的读和写 都使用测试数据库
- if os.getenv('LYY_DEV', 'unknown') == 'dev':
- write_engine = get_db_engine(config['devDB'])
- else:
- write_engine = get_db_engine(config['devDB'])
- read_engine = get_db_engine(config['devDB'])
- # 1-2 每次写入数据库的行数
- chunk_size = config['chunkSize']
- # 1-3 读取项目列表
- project_name_lst = config['projectName']
- # 2 分项目获取巨量引擎数据
- for project in project_name_lst:
- # 2-1 获取物料列表并写入数据库
- material_info_df = get_material_info(project, 7)
- # 2-2 根据 signature 获取 url 并写入数据库
- vid_lst = material_info_df['signature'].values
- video_info_df = get_video_info(vid_lst, project)
- # 2-3 向腾讯云提交语音转脚本的任务
- task_df = video_info_df[['signature', 'video_url']]
- task_ids = submit_script_task(task_df)
- # 2-4 向腾讯云获取已提交任务的脚本
- get_result_from_tx(task_ids)
|