sql_help.py 16 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373
  1. import os
  2. import sys
  3. from typing import Optional
  4. from common.mysql_db_aigc import AigcMysqlHelper
  5. sys.path.append(os.getcwd())
  6. from datetime import datetime
  7. from common.mysql_db import MysqlHelper
  8. class sqlCollect():
  9. """
  10. 视频信息写入库中
  11. """
  12. @classmethod
  13. def insert_task(cls, task_mark, video_id, mark, channel):
  14. current_time = datetime.now()
  15. formatted_time = current_time.strftime("%Y-%m-%d %H:%M:%S")
  16. insert_sql = f"""INSERT INTO pj_video_data (task_name, used_video_id, mark_name, data_time, channel) values ('{task_mark}' ,'{video_id}','{mark}', '{formatted_time}', '{channel}')"""
  17. MysqlHelper.update_values(
  18. sql=insert_sql
  19. )
  20. """
  21. 判断该任务id是否用过
  22. """
  23. @classmethod
  24. def is_used(cls, task_mark, video_id, mark_name, channel):
  25. sql = """
  26. SELECT used_video_id
  27. FROM pj_video_data
  28. WHERE used_video_id = %s AND task_name = %s AND mark_name = %s AND channel = %s
  29. ORDER BY data_time DESC
  30. LIMIT 1
  31. """
  32. data = MysqlHelper.get_values(sql, (str(video_id), task_mark, mark_name, channel))
  33. if len(data) == 0 or data == ():
  34. return False
  35. return True
  36. """
  37. 快手小程序判断该任务id是否用过
  38. """
  39. @classmethod
  40. def ks_is_used_xcx(cls, video_id, channel):
  41. sql = """SELECT used_video_id FROM pj_video_data WHERE used_video_id = %s AND channel = %s AND data_time >= DATE_SUB(NOW(), INTERVAL 7 DAY) ORDER BY data_time DESC LIMIT 1;"""
  42. data = MysqlHelper.get_values(sql, (str(video_id), channel))
  43. if len(data) == 0 or data == ():
  44. return False
  45. return True
  46. @classmethod
  47. def get_history_id(cls, channel, url):
  48. """
  49. 从数据库表中读取 id
  50. """
  51. sql = f"""select name_id from accounts where name = %s and platform = %s and useful = 1 limit 1"""
  52. data = MysqlHelper.get_values(sql, (url, channel))
  53. if data:
  54. return data[0][0]
  55. else:
  56. return False
  57. @classmethod
  58. def insert_history_id(cls, account_name, target, channel):
  59. insert_sql = f"""INSERT INTO accounts (name, name_id, platform, useful) values ("{account_name}", "{target}", "{channel}", 1 )"""
  60. MysqlHelper.update_values(
  61. sql=insert_sql
  62. )
  63. @classmethod
  64. def insert_machine_making_data(cls, name: str, task_mark: str, channel_id: str, url: str, v_id: str, piaoquan_id: str, new_title: str, code: str, formatted_time, old_title: str, oss_object_key: str):
  65. insert_sql = f"""INSERT INTO machine_making_data (name, task_mark, channel, user, v_id, pq_uid, title, pq_vid, data_time, old_title, oss_object_key) values ("{name}", "{task_mark}", "{channel_id}", "{url}", "{v_id}" , "{piaoquan_id}", "{new_title}", "{code}", "{formatted_time}", "{old_title}", "{oss_object_key}")"""
  66. MysqlHelper.update_values(
  67. sql=insert_sql
  68. )
  69. @classmethod
  70. def get_feed_count(cls, channel):
  71. sql = """SELECT count(0) FROM machine_making_data where channel = %s and DATE(data_time) = CURRENT_DATE;"""
  72. count = MysqlHelper.get_values(sql, (str(channel)))
  73. return count
  74. @classmethod
  75. def get_channel_count(cls, channel, name):
  76. sql = """SELECT count(0) FROM machine_making_data where channel = %s and name = %s and DATE(data_time) = CURRENT_DATE;"""
  77. count = MysqlHelper.get_values(sql, (str(channel),name))
  78. return count
  79. @classmethod
  80. def get_name_count(cls, name):
  81. sql = """SELECT count(0) FROM machine_making_data where name = %s and DATE(data_time) = CURRENT_DATE;"""
  82. count = MysqlHelper.get_values(sql, (str(name)))
  83. return count
  84. """
  85. 判断该任务id是否用过
  86. """
  87. @classmethod
  88. def ks_is_used(cls, photo_id):
  89. sql = """
  90. SELECT photo_id
  91. FROM ks_category_video
  92. WHERE photo_id = %s
  93. """
  94. data = MysqlHelper.get_values(sql, (str(photo_id)))
  95. if len(data) == 0 or data == ():
  96. return False
  97. return True
  98. """查询视频号user是否添加过"""
  99. @classmethod
  100. def sph_channel_user(cls, user_id):
  101. sql = """
  102. SELECT user_id
  103. FROM sph_channel_user
  104. WHERE user_id = %s
  105. """
  106. data = MysqlHelper.get_values(sql, (str(user_id)))
  107. if len(data) == 0 or data == ():
  108. return True
  109. return False
  110. """抓取视频号数据入库"""
  111. @classmethod
  112. def sph_data_info(cls, channel: str, objectId: str, video_url: str, cover: str, video_title: str, share_cnt: str, like_cnt: str, oss_video_key: str, oss_cover_key: str, nick_name: str, user_name: str, comment_count: str, fav_count: str, create_time: str, duration: str):
  113. insert_sql = f"""INSERT INTO sph_data_info (channel, video_id, video_url, video_cover, video_title, share_cnt, like_cnt, oss_url, oss_cover, nick_name, user_name, comment_count, fav_count, create_time, duration)
  114. values
  115. ("{channel}", "{objectId}", "{video_url}", "{cover}", "{video_title}", "{share_cnt}", "{like_cnt}", "{oss_video_key}", "{oss_cover_key}", "{nick_name}", "{user_name}", "{comment_count}", "{fav_count}", "{create_time}", "{duration}")"""
  116. res = MysqlHelper.update_values(
  117. sql=insert_sql
  118. )
  119. return res
  120. """查询是否有视频号是否插入过数据库"""
  121. @classmethod
  122. def sph_data_info_v_id(cls, video_id, channel):
  123. sql = """
  124. SELECT video_id
  125. FROM sph_data_info
  126. WHERE video_id = %s and channel = %s
  127. """
  128. data = MysqlHelper.get_values(sql, (str(video_id), channel))
  129. if data:
  130. return True
  131. return False
  132. @classmethod
  133. def sph_data_info_count(cls, user, channel):
  134. sql = """SELECT count(*) FROM sph_data_info WHERE nick_name = %s and channel = %s """
  135. data = MysqlHelper.get_values(sql, (user, channel))
  136. return data[0]
  137. """获取视频号历史数据"""
  138. @classmethod
  139. def sph_data_info_list(cls, user):
  140. sql = """SELECT video_id, video_title, share_cnt, like_cnt, oss_url, oss_cover, duration FROM sph_data_info WHERE nick_name = %s ORDER BY share_cnt DESC"""
  141. data = MysqlHelper.get_values(sql, (user))
  142. if data:
  143. return data
  144. return None
  145. """查询是否有视频号数据"""
  146. @classmethod
  147. def sph_channel_user_list(cls):
  148. sql = """
  149. SELECT user_id
  150. FROM sph_channel_user
  151. WHERE status = %s
  152. """
  153. data = MysqlHelper.get_values(sql, (0))
  154. if len(data) == 0 or data == ():
  155. return None
  156. return data
  157. @classmethod
  158. def update_sph_channel_user_status(cls, user):
  159. """
  160. 视频号账号状态改为1 代表已经抓取过
  161. """
  162. sql = f"""UPDATE sph_channel_user set status = 1 where user_id = '{user}'"""
  163. res = MysqlHelper.update_values(
  164. sql=sql
  165. )
  166. return res
  167. """插入视频号信息"""
  168. @classmethod
  169. def insert_sph_channel_user(cls, channel, user_id):
  170. insert_sql = f"""INSERT INTO sph_channel_user (channel, user_id, status) values ("{channel}", "{user_id}", 0)"""
  171. res = MysqlHelper.update_values(
  172. sql=insert_sql
  173. )
  174. @classmethod
  175. def insert_ks_data(cls, user_name: str, user_sex: str, time_data, caption: str, view_count: str, like_count: str,
  176. share_count: str, duration: str, main_mv_url: str, thumbnail_url: str, user_id: str, status: str,
  177. photo_id: str, category_name: str, age: str, oss_object: Optional[str] = None,
  178. video_uid: Optional[str] = None):
  179. current_time = datetime.now()
  180. formatted_time = current_time.strftime("%Y-%m-%d %H:%M:%S")
  181. insert_sql = f"""INSERT INTO ks_category_video (user_name, user_sex, time_data, caption, view_count, like_count, share_count, duration, main_mv_url, thumbnail_url, user_id, status, age_proportion, video_oss_path, pq_video_id, update_time, photo_id, category_name) values ("{user_name}", "{user_sex}", "{time_data}", "{caption}", "{view_count}", "{like_count}", "{share_count}", "{duration}", "{main_mv_url}", "{thumbnail_url}", "{user_id}", "{status}", "{age}", "{oss_object}", "{video_uid}", "{formatted_time}", "{photo_id}", "{category_name}")"""
  182. res = MysqlHelper.update_values(
  183. sql=insert_sql
  184. )
  185. @classmethod
  186. def get_shp_dd_data(cls, url):
  187. """
  188. 获取视频号单点内容
  189. """
  190. sql = f"""select video_id,title,author_id,author_name,cover_url,video_url,video_duration,from_user_id,from_user_name,from_group_id,from_group_name,source,wx_msg, is_encrypted, decode_key from dandian_content where from_user_name = %s and has_used = 0 ORDER BY create_timestamp DESC limit 2"""
  191. data = AigcMysqlHelper.get_values(sql, (url,))
  192. return data
  193. @classmethod
  194. def get_feed_data(cls, channel):
  195. """
  196. 获取feed流视频
  197. """
  198. sql = f"""select video_id,channel,video_url,cover_url,title from automator_feed_video where channel = %s and has_used = 0 and if_50_like = 1 ORDER BY create_time DESC limit 1"""
  199. data = AigcMysqlHelper.get_values(sql, (channel,))
  200. return data
  201. @classmethod
  202. def update_feed_vid(cls, vid):
  203. """
  204. 视频号单点修改状态为1
  205. """
  206. sql = f"""UPDATE automator_feed_video set has_used = 1 where video_id = '{vid}'"""
  207. res = AigcMysqlHelper.update_values(
  208. sql=sql
  209. )
  210. return res
  211. @classmethod
  212. def update_feed_vid_2(cls, vid):
  213. """
  214. 视频号单点修改状态为1
  215. """
  216. sql = f"""UPDATE automator_feed_video set has_used = 2 where video_id = '{vid}'"""
  217. res = AigcMysqlHelper.update_values(
  218. sql=sql
  219. )
  220. return res
  221. @classmethod
  222. def update_shp_dd_vid(cls, vid):
  223. """
  224. 视频号单点修改状态为1
  225. """
  226. sql = f"""UPDATE dandian_content set has_used = 1 where video_id = '{vid}'"""
  227. res = AigcMysqlHelper.update_values(
  228. sql=sql
  229. )
  230. return res
  231. @classmethod
  232. def update_shp_dd_vid_4(cls, vid):
  233. """
  234. 视频号单点修改状态为4
  235. """
  236. sql = f"""UPDATE dandian_content set has_used = 4 where video_id = '{vid}'"""
  237. res = AigcMysqlHelper.update_values(
  238. sql=sql
  239. )
  240. return res
  241. @classmethod
  242. def select_ks_star_data(cls, profile_id):
  243. sql = """SELECT * FROM ks_star_info where profile_id = %s"""
  244. data = MysqlHelper.get_values(sql, (profile_id))
  245. if data:
  246. return data
  247. return None
  248. @classmethod
  249. def insert_ks_star_data(cls, user_id: str, star_id: str, name, kwai_id: str, gender: str, fans_number: str,
  250. profile_id: str, star_tag_str: str, industry_tag_str: str, photo_expect_play: str, photo_expect_cpm: str, photo_interaction_rate: str,
  251. photo_complete_play_rate: str, fans_increase_num: str, fans_increase_rate: str,):
  252. insert_sql = f"""INSERT INTO ks_star_info (user_id, star_id, name, kwai_id, gender, fans_number, profile_id, star_tag_str, industry_tag_str, photo_expect_play, photo_expect_cpm, photo_interaction_rate, photo_complete_play_rate, fans_increase_num, fans_increase_rate) values ("{user_id}", "{star_id}", "{name}", "{kwai_id}", "{gender}", "{fans_number}", "{profile_id}", "{star_tag_str}", "{industry_tag_str}", "{photo_expect_play}", "{photo_expect_cpm}", "{photo_interaction_rate}", "{photo_complete_play_rate}", "{fans_increase_num}", "{fans_increase_rate}")"""
  253. res = MysqlHelper.update_values(
  254. sql=insert_sql
  255. )
  256. return res
  257. @classmethod
  258. def insert_ks_star_works_info(cls, star_id: str, profile_id: str, pplay_median_data_90: str, pplay_median_data_30: str, pphoto_cnt_data_90: str, pphoto_cnt_data_30: str, pavg_video_duration_data_90: str, pavg_video_duration_data_30: str, star_work_report_90: str, star_work_report_30: str, pavg_share_cnt_90: str, pavg_share_cnt_30: str):
  259. insert_sql = f"""INSERT INTO ks_star_works_info
  260. (star_id, profile_id, pplay_median_data_90, pplay_median_data_30, pphoto_cnt_data_90, pphoto_cnt_data_30, pavg_video_duration_data_90, pavg_video_duration_data_30, star_work_report_90, star_work_report_30, pavg_share_cnt_90, pavg_share_cnt_30)
  261. values
  262. ("{star_id}", "{profile_id}", "{pplay_median_data_90}", "{pplay_median_data_30}", "{pphoto_cnt_data_90}", "{pphoto_cnt_data_30}", "{pavg_video_duration_data_90}", "{pavg_video_duration_data_30}", "{star_work_report_90}", "{star_work_report_30}", "{pavg_share_cnt_90}", "{pavg_share_cnt_30}")"""
  263. res = MysqlHelper.update_values(
  264. sql=insert_sql
  265. )
  266. return res
  267. @classmethod
  268. def insert_ks_list_portrait_info(cls, star_id: str, profile_id: str, type: str,
  269. name: str, proportion: str):
  270. insert_sql = f"""INSERT INTO ks_list_portrait_info
  271. (star_id, profile_id, type, name, proportion)
  272. values
  273. ("{star_id}", "{profile_id}", "{type}", "{name}", "{proportion}")"""
  274. res = MysqlHelper.update_values(
  275. sql=insert_sql
  276. )
  277. return res
  278. """品类视频新条数记录"""
  279. @classmethod
  280. def insert_spider_supply_targetcnt(cls, channel_id: str, name: str, url: str,
  281. number: str, new_count: str, s_cnt: str):
  282. insert_sql = f"""INSERT INTO spider_supply_targetcnt
  283. (channel_id, name, url, o_cnt, n_cnt, s_cnt)
  284. values
  285. ("{channel_id}", "{name}", "{url}", "{number}", "{new_count}", "{s_cnt}")"""
  286. res = MysqlHelper.update_values(
  287. sql=insert_sql
  288. )
  289. return res
  290. """
  291. 获取相似溯源基础账号
  292. """
  293. @classmethod
  294. def get_machine_making_reflux(cls, channel, channel1, channel2, channel3, channel4):
  295. sql = """SELECT user FROM machine_making_reflux WHERE (channel = %s or channel = %s or channel = %s or channel = %s or channel = %s) and status = 0"""
  296. data = MysqlHelper.get_values(sql, (channel, channel1, channel2, channel3, channel4))
  297. if len(data) == 0 or data == ():
  298. return None
  299. return data
  300. """
  301. 修改相似溯源基础账号状态
  302. """
  303. @classmethod
  304. def update_machine_making_reflux(cls, user_id):
  305. sql = f"""UPDATE machine_making_reflux set status = 1 where user = '{user_id}'"""
  306. res = MysqlHelper.update_values(
  307. sql=sql
  308. )
  309. return res
  310. """插入符合规则的用户主页"""
  311. @classmethod
  312. def insert_user_data(cls, uid, name, account_id, channel, status):
  313. insert_sql = f"""INSERT INTO machine_making_traceability (old_account_id, name, new_account_id, channel, status) values ("{uid}", "{name}","{account_id}","{channel}", {status})"""
  314. res = MysqlHelper.update_values(
  315. sql=insert_sql
  316. )
  317. return res
  318. """相似溯源-视频号数据插入"""
  319. @classmethod
  320. def insert_xssy_sph_info(cls, account_user: str, traceable_user: str, channel: str, traceable_user_v2: str, has_used: str, appid:Optional[str] = None, pq_id:Optional[str] = None):
  321. insert_sql = f"""INSERT INTO xssy_sph (account_user, traceable_user, channel, traceable_user_v2, pq_id, has_used, appid) values ("{account_user}", "{traceable_user}","{channel}", "{traceable_user_v2}","{pq_id}", {has_used},"{appid}")"""
  322. res = MysqlHelper.update_values(
  323. sql=insert_sql
  324. )
  325. return res
  326. """查询该账号是否存在"""
  327. @classmethod
  328. def select_crawler_uesr_v3(cls, link: str):
  329. sql = """SELECT uid FROM crawler_user_v3 WHERE link = %s and source = 'jiqizidonggaizao'"""
  330. data = MysqlHelper.get_values(sql, (link))
  331. if data:
  332. return data
  333. return None