Merge remote-tracking branch 'origin/main'

# Conflicts:
#	DB.py
This commit is contained in:
晓丰 2025-07-17 10:54:26 +08:00
commit 5bb852ac5c
2 changed files with 183 additions and 192 deletions

369
DB.py
View File

@ -29,51 +29,51 @@ _engine = create_engine(
_meta = MetaData() _meta = MetaData()
# 操作记录表 # 操作记录表
video_op = Table("sh_dm_video_op_v3", _meta, video_op = Table("sh_dm_video_op_v3", _meta,
Column("id", Integer, primary_key=True, autoincrement=True), Column("id", Integer, primary_key=True, autoincrement=True),
Column("v_id", String(64)), Column("v_id", String(64)),
Column("v_xid", String(64)), Column("v_xid", String(64)),
Column("a_id", Integer, default=0), Column("a_id", Integer, default=0),
Column("level", Integer), Column("level", Integer),
Column("name_title", String(100)), Column("name_title", String(100)),
Column("keyword", String(100)), Column("keyword", String(100)),
Column("rn", String(50)), Column("rn", String(50)),
Column("history_status", String(100)), Column("history_status", String(100)),
Column("is_repeat", Integer), Column("is_repeat", Integer),
Column("is_piracy", String(2), default='3'), Column("is_piracy", String(2), default='3'),
Column("sort", Integer), Column("sort", Integer),
Column("createtime", Integer), Column("createtime", Integer),
Column("updatetime", Integer), Column("updatetime", Integer),
Column("operatetime", Integer), Column("operatetime", Integer),
Column("batch", Integer), Column("batch", Integer),
Column("machine", Integer), Column("machine", Integer),
Column("ts_status", Integer, default=1), Column("ts_status", Integer, default=1),
) )
# 视频表 # 视频表
video = Table("sh_dm_video_v3", _meta, video = Table("sh_dm_video_v3", _meta,
Column("id", Integer, primary_key=True, autoincrement=True), Column("id", Integer, primary_key=True, autoincrement=True),
Column("v_id", String(64)), Column("v_id", String(64)),
Column("v_name", String(100), nullable=False), Column("v_name", String(100), nullable=False),
Column("v_xid", String(64), nullable=False), Column("v_xid", String(64), nullable=False),
Column("title", String(255), nullable=False), Column("title", String(255), nullable=False),
Column("link", String(255), nullable=False), Column("link", String(255), nullable=False),
Column("edition", String(255), default=''), Column("edition", String(255), default=''),
Column("duration", String(11), default='0'), Column("duration", String(11), default='0'),
Column("watch_number", Integer, default=0), Column("watch_number", Integer, default=0),
Column("follow_number", Integer, default=0), Column("follow_number", Integer, default=0),
Column("video_number", Integer, default=0), Column("video_number", Integer, default=0),
Column("public_time", DateTime), Column("public_time", DateTime),
Column("cover_pic", String(255)), Column("cover_pic", String(255)),
Column("sort", Integer), Column("sort", Integer),
Column("u_xid", String(64)), Column("u_xid", String(64)),
Column("u_id", String(100)), Column("u_id", String(100)),
Column("u_pic", String(255)), Column("u_pic", String(255)),
Column("u_name", String(255)), Column("u_name", String(255)),
Column("status", Integer, default=1), Column("status", Integer, default=1),
Column("createtime", Integer, default=0), Column("createtime", Integer, default=0),
Column("updatetime", Integer, default=0), Column("updatetime", Integer, default=0),
Column("operatetime", Integer, default=0), Column("operatetime", Integer, default=0),
) )
# 作者表 # 作者表
video_author = Table( video_author = Table(
"sh_dm_video_author", "sh_dm_video_author",
@ -116,8 +116,7 @@ def mysql_retry(max_retries: int = 3, base_delay: float = 1.0):
}.get(errno, f"MySQL错误{errno}") }.get(errno, f"MySQL错误{errno}")
wait = base_delay * (2 ** (attempt - 1)) wait = base_delay * (2 ** (attempt - 1))
logger.warning( logger.warning(f"[MySQL][{fn.__name__}] 第{attempt}次重试({errno} {reason}{e},等待 {wait:.1f}s...")
f"[MySQL][{fn.__name__}] 第{attempt}次重试({errno} {reason}{e},等待 {wait:.1f}s...")
# 仅对断连类错误尝试重连 # 仅对断连类错误尝试重连
if errno in {2013, 2006}: if errno in {2013, 2006}:
@ -129,10 +128,8 @@ def mysql_retry(max_retries: int = 3, base_delay: float = 1.0):
raise raise
return wrapper return wrapper
return decorator return decorator
def redis_retry(max_retries: int = 3): def redis_retry(max_retries: int = 3):
""" """
装饰器工厂指定最大重试次数 装饰器工厂指定最大重试次数
@ -178,7 +175,6 @@ class DBVidcon:
self.l0_list_key = "video_l0_queue" self.l0_list_key = "video_l0_queue"
self.l1_list_key = "video_l1_queue" self.l1_list_key = "video_l1_queue"
self.l2_list_key = "video_l2_queue" self.l2_list_key = "video_l2_queue"
self.web_list_key = "video_web_queue"
self.report_list = "report_queue" self.report_list = "report_queue"
self.error_list_key = "error_save_queue" self.error_list_key = "error_save_queue"
self.conn = pymysql.connect(**self._MYSQL_CONF) self.conn = pymysql.connect(**self._MYSQL_CONF)
@ -261,13 +257,6 @@ class DBVidcon:
self.redis.lpush(self.l2_list_key, *raws) self.redis.lpush(self.l2_list_key, *raws)
logger.info(f"[写入l2] 已推入 {len(raws)}") logger.info(f"[写入l2] 已推入 {len(raws)}")
@redis_retry(max_retries=3)
def push_web(self, raws):
if isinstance(raws, str):
raws = [raws]
self.redis.lpush(self.web_list_key, *raws)
logger.info(f"[写入web] 已推入 {len(raws)}")
@redis_retry(max_retries=3) @redis_retry(max_retries=3)
def push_report(self, raws): def push_report(self, raws):
"""原子操作:清空列表并写入新数据""" """原子操作:清空列表并写入新数据"""
@ -329,17 +318,6 @@ class DBVidcon:
items = [] items = []
return items return items
@redis_retry(max_retries=3)
def get_web_items(self, count: int = 1):
try:
items = self.fetch_from_redis(count, list_key=self.web_list_key)
except Exception as e:
logger.info("[Redis web pop error]", e)
self.reconnect_redis()
items = []
return items
@redis_retry(max_retries=3) @redis_retry(max_retries=3)
def item_keyword(self, count: int = 20): def item_keyword(self, count: int = 20):
try: try:
@ -395,48 +373,53 @@ class DBVidcon:
@mysql_retry() @mysql_retry()
def get_report_video(self): def get_report_video(self):
sql = """ sql = """
SELECT id, SELECT
name_title, id,
link name_title,
FROM sh_dm_fight_records link
WHERE status = 1 \ FROM
""" sh_dm_fight_records
WHERE
status = 1
"""
self.cursor.execute(sql) self.cursor.execute(sql)
return self.cursor.fetchall() return self.cursor.fetchall()
@mysql_retry() @mysql_retry()
def get_subsequent_report_video(self, did: int): def get_subsequent_report_video(self,did: int):
sql = """ sql = """
SELECT DISTINCT report_id SELECT DISTINCT report_id
FROM sh_dm_fight_records FROM sh_dm_fight_records
WHERE status = 2 WHERE
AND subsequent_status = 1 status = 2
AND report_time != '' AND subsequent_status in (1, 3)
AND mid = %s \ AND report_time != ''
""" AND mid = %s
self.cursor.execute(sql, did) """
self.cursor.execute(sql,did)
return self.cursor.fetchall() return self.cursor.fetchall()
@mysql_retry() @mysql_retry()
def getreport_video(self): def getreport_video(self):
sql = """ sql = """
SELECT id, SELECT
v_xid id,
FROM sh_dm_fight_records v_xid
WHERE is_removed = '' FROM
or is_removed IS NULL sh_dm_fight_records
or is_removed = 0 \ WHERE
""" is_removed = '' or is_removed IS NULL or is_removed = 0
"""
self.cursor.execute(sql) self.cursor.execute(sql)
return self.cursor.fetchall() return self.cursor.fetchall()
@mysql_retry() @mysql_retry()
def mark_video_removed(self, d_id: int, removed_flag: int = 1): def mark_video_removed(self, d_id: int, removed_flag: int = 1):
sql = """ sql = """
UPDATE sh_dm_fight_records UPDATE sh_dm_fight_records
SET is_removed = %s SET is_removed = %s
WHERE id = %s \ WHERE id = %s
""" """
self.cursor.execute(sql, (removed_flag, d_id)) self.cursor.execute(sql, (removed_flag, d_id))
self.flush() self.flush()
@ -481,12 +464,11 @@ class DBVidcon:
@mysql_retry() @mysql_retry()
def update_video_ts_status(self): def update_video_ts_status(self):
sql = """ sql = """
UPDATE sh_dm_video_v2 v UPDATE sh_dm_video_v2 v
JOIN sh_dm_video_author a JOIN sh_dm_video_author a ON v.u_xid = a.u_xid
ON v.u_xid = a.u_xid SET v.ts_status = 3
SET v.ts_status = 3 WHERE a.white_status = 1;
WHERE a.white_status = 1; \ """
"""
self.cursor.execute(sql) self.cursor.execute(sql)
self.flush() self.flush()
logger.info("[更新视频举报状态] 已执行完毕") logger.info("[更新视频举报状态] 已执行完毕")
@ -506,32 +488,38 @@ class DBVidcon:
while True: while True:
try: try:
sql_op = """ sql_op = """
INSERT INTO sh_dm_video_op_v2 (v_id, v_xid, a_id, level, name_title, INSERT INTO sh_dm_video_op_v2 (
keyword, rn, history_status, is_repeat, v_id, v_xid, a_id, level, name_title,
sort, createtime, updatetime, batch, machine) keyword, rn, history_status, is_repeat,
VALUES (%(v_id)s, %(v_xid)s, %(a_id)s, %(level)s, %(v_name)s, sort, createtime, updatetime, batch, machine
%(keyword)s, %(rn)s, %(history_status)s, %(is_repeat)s, ) VALUES (
%(sort)s, UNIX_TIMESTAMP(), UNIX_TIMESTAMP(), %(batch)s, %(machine_id)s) \ %(v_id)s, %(v_xid)s, %(a_id)s, %(level)s, %(v_name)s,
""" %(keyword)s, %(rn)s, %(history_status)s, %(is_repeat)s,
%(sort)s, UNIX_TIMESTAMP(), UNIX_TIMESTAMP(), %(batch)s, %(machine_id)s
)
"""
self.cursor.execute(sql_op, data) self.cursor.execute(sql_op, data)
sql_update = """ sql_update = """
INSERT INTO sh_dm_video_v2 (v_id, v_xid, rn, v_name, title, link, INSERT INTO sh_dm_video_v2 (
edition, duration, v_id, v_xid, rn, v_name, title, link,
public_time, cover_pic, sort, edition, duration,
u_xid, u_id, u_pic, u_name, public_time, cover_pic, sort,
status, createtime, updatetime) u_xid, u_id, u_pic, u_name,
VALUES (%(v_id)s, %(v_xid)s, %(rn)s, %(v_name)s, %(title)s, %(link)s, status, createtime, updatetime
'', %(duration)s, ) VALUES (
%(create_time)s, %(cover_pic)s, %(sort)s, %(v_id)s, %(v_xid)s, %(rn)s, %(v_name)s, %(title)s, %(link)s,
%(u_xid)s, %(u_id)s, %(u_pic)s, %(u_name)s, '', %(duration)s,
1, UNIX_TIMESTAMP(), UNIX_TIMESTAMP()) ON DUPLICATE KEY %(create_time)s, %(cover_pic)s, %(sort)s,
UPDATE %(u_xid)s, %(u_id)s, %(u_pic)s, %(u_name)s,
title = 1, UNIX_TIMESTAMP(), UNIX_TIMESTAMP()
VALUES (title), duration = )
VALUES (duration), cover_pic = ON DUPLICATE KEY UPDATE
VALUES (cover_pic), sort = title = VALUES(title),
VALUES (sort), updatetime = UNIX_TIMESTAMP(); \ duration = VALUES(duration),
""" cover_pic = VALUES(cover_pic),
sort = VALUES(sort),
updatetime = UNIX_TIMESTAMP();
"""
self.cursor.execute(sql_update, data) self.cursor.execute(sql_update, data)
break # 成功跳出重试循环 break # 成功跳出重试循环
@ -557,26 +545,33 @@ class DBVidcon:
@mysql_retry() @mysql_retry()
def fetch_keyword_title(self, level: int = 99): def fetch_keyword_title(self, level: int = 99):
sql = """ sql = """
SELECT k.keyword, SELECT
k.rn, k.keyword,
t.title AS v_name, k.rn,
ANY_VALUE(t.level) AS level t.title AS v_name,
FROM sh_dm_keyword k ANY_VALUE(t.level) AS level
LEFT JOIN FROM
sh_dm_title t ON k.title = t.title sh_dm_keyword k
WHERE k.status = 1 LEFT JOIN
AND t.status = 1 sh_dm_title t ON k.title = t.title
AND NOT EXISTS (SELECT 1 WHERE
FROM sh_dm_black_keyword b k.status = 1
WHERE ( AND t.status = 1
(b.keyword IS NULL OR b.keyword = '') AND b.title = t.title AND NOT EXISTS (
) SELECT 1
OR ( FROM sh_dm_black_keyword b
b.keyword IS NOT NULL AND b.keyword != '' AND b.keyword = k.keyword WHERE
)) (
AND t.level = %s (b.keyword IS NULL OR b.keyword = '') AND b.title = t.title
GROUP BY k.keyword, k.rn; \ )
""" OR (
b.keyword IS NOT NULL AND b.keyword != '' AND b.keyword = k.keyword
)
)
AND t.level = %s
GROUP BY
k.keyword, k.rn;
"""
self.cursor.execute(sql, (level,)) self.cursor.execute(sql, (level,))
return self.cursor.fetchall() return self.cursor.fetchall()
@ -597,10 +592,10 @@ class DBVidcon:
start_ts = int(time.time()) start_ts = int(time.time())
sql = """ sql = """
INSERT INTO sh_dm_batch_log INSERT INTO sh_dm_batch_log
(batch, info, t0, t1, t2, starttime) (batch, info, t0, t1, t2, starttime)
VALUES (%s, %s, %s, %s, %s, %s) \ VALUES (%s, %s, %s, %s, %s, %s)
""" """
try: try:
self.cursor.execute(sql, (batch, level, t0, t1, t2, start_ts)) self.cursor.execute(sql, (batch, level, t0, t1, t2, start_ts))
self.conn.commit() self.conn.commit()
@ -671,10 +666,6 @@ class DBVidcon:
def l2_empty(self) -> bool: def l2_empty(self) -> bool:
return self.redis.llen(self.l2_list_key) == 0 return self.redis.llen(self.l2_list_key) == 0
@redis_retry(max_retries=3)
def web_empty(self) -> bool:
return self.redis.llen(self.web_list_key) == 0
@redis_retry(max_retries=3) @redis_retry(max_retries=3)
def pop_error_item(self): def pop_error_item(self):
""" """
@ -688,30 +679,30 @@ class DBVidcon:
class DBSA: class DBSA:
# ======= 可调参数 ======= # ======= 可调参数 =======
FLUSH_EVERY_ROWS = 100 # 行数阈值 FLUSH_EVERY_ROWS = 100 # 行数阈值
FLUSH_INTERVAL = 30 # 秒阈值 FLUSH_INTERVAL = 30 # 秒阈值
MAX_SQL_RETRY = 3 # SQL 死锁自旋 MAX_SQL_RETRY = 3 # SQL 死锁自旋
SQL_RETRY_BASE_SLEEP = 0.5 # 自旋退避基数 SQL_RETRY_BASE_SLEEP = 0.5 # 自旋退避基数
FLUSH_RETRY = 3 # flush 整体轮次 FLUSH_RETRY = 3 # flush 整体轮次
DELAY_ON_FAIL = 10 # flush 失败等待 DELAY_ON_FAIL = 10 # flush 失败等待
DEADLOCK_ERRNO = 1213 # MySQL 死锁码 DEADLOCK_ERRNO = 1213 # MySQL 死锁码
LOCK_TIMEOUT = 3 # 互斥锁超时 LOCK_TIMEOUT = 3 # 互斥锁超时
# ======================== # ========================
# ----- 缓冲区 ----- # ----- 缓冲区 -----
_buf_op = [] _buf_op = []
_buf_vid = [] _buf_vid = []
_buf_payload = [] _buf_payload = []
_last_flush = time.time() _last_flush = time.time()
# ----- 并发控制 ----- # ----- 并发控制 -----
_lock = threading.Lock() _lock = threading.Lock()
_existing_op_keys = set() _existing_op_keys = set()
_existing_vid_keys = set() _existing_vid_keys = set()
# ----- queue / 后台线程模式 ----- # ----- queue / 后台线程模式 -----
_queue_mode = False _queue_mode = False
_queue = Queue() _queue = Queue()
# ================== 退回 Redis 模拟 ================== # ================== 退回 Redis 模拟 ==================
@staticmethod @staticmethod
@ -735,7 +726,7 @@ class DBSA:
data["sort"] = data.get("index", 0) data["sort"] = data.get("index", 0)
now_ts = int(time.time()) now_ts = int(time.time())
op_index_key = (data["v_xid"] or "", data["keyword"] or "", now_ts) op_index_key = (data["v_xid"] or "", data["keyword"] or "", now_ts)
vid_index_key = (data["v_xid"] or "", data["v_name"] or "") vid_index_key = (data["v_xid"] or "", data["v_name"] or "")
# ---------- ① 获取互斥锁 ---------- # ---------- ① 获取互斥锁 ----------
@ -757,8 +748,8 @@ class DBSA:
keyword=data["keyword"], keyword=data["keyword"],
is_repeat=data["is_repeat"], is_repeat=data["is_repeat"],
sort=data["sort"], sort=data["sort"],
createtime=now_ts, # 首次插入 createtime=now_ts, # 首次插入
updatetime=now_ts, # 后续更新只改这一列 updatetime=now_ts, # 后续更新只改这一列
batch=data.get("batch", 0), batch=data.get("batch", 0),
machine=data.get("machine_id", 0), machine=data.get("machine_id", 0),
is_piracy=data.get("is_piracy", '3'), is_piracy=data.get("is_piracy", '3'),
@ -859,9 +850,9 @@ class DBSA:
if not cls._lock.acquire(timeout=cls.LOCK_TIMEOUT): if not cls._lock.acquire(timeout=cls.LOCK_TIMEOUT):
raise RuntimeError("flush 未取得 cls._lock可能死锁") raise RuntimeError("flush 未取得 cls._lock可能死锁")
try: try:
op_rows = cls._buf_op[:] op_rows = cls._buf_op[:]
vid_rows = cls._buf_vid[:] vid_rows = cls._buf_vid[:]
payloads = cls._buf_payload[:] payloads = cls._buf_payload[:]
cls._clear_buffers() cls._clear_buffers()
finally: finally:
cls._lock.release() cls._lock.release()
@ -903,9 +894,9 @@ class DBSA:
if op_rows: if op_rows:
stmt_op = mysql_insert(video_op).values(op_rows) stmt_op = mysql_insert(video_op).values(op_rows)
ondup_op = stmt_op.on_duplicate_key_update( ondup_op = stmt_op.on_duplicate_key_update(
updatetime=stmt_op.inserted.updatetime, updatetime = stmt_op.inserted.updatetime,
# ts_status = stmt_op.inserted.ts_status, # ts_status = stmt_op.inserted.ts_status,
is_repeat=stmt_op.inserted.is_repeat, is_repeat = stmt_op.inserted.is_repeat,
) )
cls._safe_execute(ondup_op, desc="video_op") cls._safe_execute(ondup_op, desc="video_op")
logger.info("落表:操作记录 %d", len(op_rows)) logger.info("落表:操作记录 %d", len(op_rows))
@ -914,23 +905,23 @@ class DBSA:
if vid_rows: if vid_rows:
stmt_vid = mysql_insert(video).values(vid_rows) stmt_vid = mysql_insert(video).values(vid_rows)
ondup_vid = stmt_vid.on_duplicate_key_update( ondup_vid = stmt_vid.on_duplicate_key_update(
title=stmt_vid.inserted.title, title = stmt_vid.inserted.title,
v_name=stmt_vid.inserted.v_name, v_name = stmt_vid.inserted.v_name,
link=stmt_vid.inserted.link, link = stmt_vid.inserted.link,
edition=stmt_vid.inserted.edition, edition = stmt_vid.inserted.edition,
duration=stmt_vid.inserted.duration, duration = stmt_vid.inserted.duration,
watch_number=stmt_vid.inserted.watch_number, watch_number = stmt_vid.inserted.watch_number,
follow_number=stmt_vid.inserted.follow_number, follow_number = stmt_vid.inserted.follow_number,
video_number=stmt_vid.inserted.video_number, video_number = stmt_vid.inserted.video_number,
public_time=stmt_vid.inserted.public_time, public_time = stmt_vid.inserted.public_time,
cover_pic=stmt_vid.inserted.cover_pic, cover_pic = stmt_vid.inserted.cover_pic,
sort=stmt_vid.inserted.sort, sort = stmt_vid.inserted.sort,
u_xid=stmt_vid.inserted.u_xid, u_xid = stmt_vid.inserted.u_xid,
u_id=stmt_vid.inserted.u_id, u_id = stmt_vid.inserted.u_id,
u_pic=stmt_vid.inserted.u_pic, u_pic = stmt_vid.inserted.u_pic,
u_name=stmt_vid.inserted.u_name, u_name = stmt_vid.inserted.u_name,
status=stmt_vid.inserted.status, status = stmt_vid.inserted.status,
updatetime=stmt_vid.inserted.updatetime, updatetime = stmt_vid.inserted.updatetime,
) )
cls._safe_execute(ondup_vid, desc="video") cls._safe_execute(ondup_vid, desc="video")
logger.info("落表:视频记录 %d", len(vid_rows)) logger.info("落表:视频记录 %d", len(vid_rows))
@ -994,7 +985,7 @@ class DBSA:
data["sort"] = data.get("index", 0) data["sort"] = data.get("index", 0)
now_ts = int(time.time()) now_ts = int(time.time())
op_key = (data["v_xid"] or "", data["keyword"] or "", now_ts) op_key = (data["v_xid"] or "", data["keyword"] or "", now_ts)
vid_key = (data["v_xid"] or "", data["v_name"] or "") vid_key = (data["v_xid"] or "", data["v_name"] or "")
if op_key in cls._existing_op_keys or vid_key in cls._existing_vid_keys: if op_key in cls._existing_op_keys or vid_key in cls._existing_vid_keys:
return return

View File

@ -59,11 +59,11 @@ def get_searchInfo(keyword, level, headers, proxy_name, r=2):
if r == 2: if r == 2:
logger.info(f"NET处理->{keyword},\trn->{proxy_name},\tlevel->{level}") logger.info(f"NET处理->{keyword},\trn->{proxy_name},\tlevel->{level}")
video_list = [] video_list = []
max_page = 2 max_page = 3
limit = 30 limit = 100
endpoint = 'https://api.dailymotion.com/videos' endpoint = 'https://api.dailymotion.com/videos'
if level == 0 or level == 1: if level == 0 or level == 1:
max_page = 3 max_page = 4
limit = 100 limit = 100
for j in range(1, max_page): for j in range(1, max_page):
params = { params = {