def run(self):
if not self.is_reach_next_spider_time():
return
self._start()
while True:
try:
if self.all_thread_is_done():
if not self._is_notify_end:
self.spider_end() # 跑完一轮
self.record_spider_state(
spider_type=1,
state=1,
spider_end_time=tools.get_current_date(),
batch_interval=self._batch_interval,
)
self._is_notify_end = True
if not self._keep_alive:
self._stop_all_thread()
break
else:
self._is_notify_end = False
self.check_task_status()
except Exception as e:
log.exception(e)
tools.delay_time(1) # 1秒钟检查一次爬虫状态
def __add_item_to_db(self, items, update_items, requests, callbacks,
items_fingerprints):
export_success = False
self._is_adding_to_db = True
# 去重
if setting.ITEM_FILTER_ENABLE:
items, items_fingerprints = self.__dedup_items(
items, items_fingerprints)
# 分捡
items_dict = self.__pick_items(items)
update_items_dict = self.__pick_items(update_items,
is_update_item=True)
# item批量入库
while items_dict:
tab_item, datas = items_dict.popitem()
log.debug("""
-------------- item 批量入库 --------------
表名: %s
datas: %s
""" % (tab_item, tools.dumps_json(datas, indent=16)))
export_success = self.__export_to_db(tab_item, datas)
# 执行批量update
while update_items_dict:
tab_item, datas = update_items_dict.popitem()
log.debug("""
-------------- item 批量更新 --------------
表名: %s
datas: %s
""" % (tab_item, tools.dumps_json(datas, indent=16)))
update_keys = self._item_update_keys.get(tab_item)
export_success = self.__export_to_db(tab_item,
datas,
is_update=True,
update_keys=update_keys)
# 执行回调
while callbacks:
try:
callback = callbacks.pop(0)
callback()
except Exception as e:
log.exception(e)
# 删除做过的request
if requests:
self._db.zrem(self._table_request, requests)
# 去重入库
if export_success and setting.ITEM_FILTER_ENABLE:
if items_fingerprints:
self.__class__.dedup.add(items_fingerprints, skip_check=True)
self._is_adding_to_db = False
def get_user(self, block=True) -> Optional[GuestUser]:
"""
Args:
block: 无用户时是否等待
Returns:
"""
while True:
try:
user_id = self._get_user_id()
user_str = None
if user_id:
user_str = self._redisdb.hget(self._tab_user_pool, user_id)
# 如果没取到user，可能是其他爬虫将此用户删除了，需要重刷新本地缓存的用户id
if not user_str:
self._load_users_id()
continue
if not user_id and block:
self._keep_alive = False
with RedisLock(key=self._tab_user_pool,
lock_timeout=3600,
wait_timeout=0) as _lock:
if _lock.locked:
self.run()
continue
return user_str and GuestUser(**eval(user_str))
except Exception as e:
log.exception(e)
tools.delay_time(1)
def put_failed_request(self, request, table=None):
try:
request_dict = request.to_dict
self._db.zadd(table or self._table_failed_request, request_dict,
request.priority)
except Exception as e:
log.exception(e)
def run(self):
while True:
try:
now_user_count = self._redisdb.hget_count(self._tab_user_pool)
need_user_count = self._min_users - now_user_count
if need_user_count > 0:
log.info("当前在线user数为 {} 小于 {}, 生产user".format(
now_user_count, self._min_users))
try:
user = self.login()
if user:
self.add_user(user)
except Exception as e:
log.exception(e)
else:
log.debug("当前user数为 {} 数量足够 暂不生产".format(now_user_count))
if self._keep_alive:
tools.delay_time(10)
else:
break
except Exception as e:
log.exception(e)
tools.delay_time(1)
def handel_exception(self, e):
"""
处理异常
@param e:
@return:
"""
log.exception(e)
def start_monitor_task(self, *args, **kws):
if not self.is_reach_next_spider_time():
return
self._auto_start_requests = False
redisdb = RedisDB()
if not self._parsers: # 不是add_parser 模式
self._parsers.append(self)
while True:
try:
# 检查redis中是否有任务
tab_requests = setting.TAB_REQUSETS.format(
redis_key=self._redis_key
)
todo_task_count = redisdb.zget_count(tab_requests)
if todo_task_count < self._min_task_count: # 添加任务
# make start requests
self.distribute_task(*args, **kws)
else:
log.info("redis 中尚有%s条积压任务，暂时不派发新任务" % todo_task_count)
except Exception as e:
log.exception(e)
if self._auto_stop_when_spider_done:
break
time.sleep(self._check_task_interval)
def __add_request_to_db(self):
request_list = []
prioritys = []
callbacks = []
while self._requests_deque:
request = self._requests_deque.popleft()
self._is_adding_to_db = True
if callable(request):
# 函数
# 注意：应该考虑闭包情况。闭包情况可写成
# def test(xxx = xxx):
# # TODO 业务逻辑 使用 xxx
# 这么写不会导致xxx为循环结束后的最后一个值
callbacks.append(request)
continue
priority = request.priority
# 如果需要去重并且库中已重复 则continue
if (request.filter_repeat and setting.REQUEST_FILTER_ENABLE
and not self.__class__.dedup.add(request.fingerprint)):
log.debug("request已存在 url = %s" % request.url)
continue
else:
request_list.append(str(request.to_dict))
prioritys.append(priority)
if len(request_list) > MAX_URL_COUNT:
self._db.zadd(self._table_request, request_list, prioritys)
request_list = []
prioritys = []
# 入库
if request_list:
self._db.zadd(self._table_request, request_list, prioritys)
# 执行回调
for callback in callbacks:
try:
callback()
except Exception as e:
log.exception(e)
# 删除已做任务
if self._del_requests_deque:
request_done_list = []
while self._del_requests_deque:
request_done_list.append(self._del_requests_deque.popleft())
# 去掉request_list中的requests， 否则可能会将刚添加的request删除
request_done_list = list(
set(request_done_list) - set(request_list))
if request_done_list:
self._db.zrem(self._table_request, request_done_list)
self._is_adding_to_db = False
def run(self):
while not self._thread_stop:
try:
self.__add_request_to_db()
except Exception as e:
log.exception(e)
tools.delay_time(1)
def run(self):
self._thread_stop = False
while not self._thread_stop:
try:
self.__input_data()
except Exception as e:
log.exception(e)
time.sleep(0.1)
self._is_collector_task = False
def run(self):
while not self._thread_stop:
try:
self.__report_node_heartbeat()
self.__input_data()
except Exception as e:
log.exception(e)
self._is_collector_task = False
time.sleep(self._interval)
def run(self):
while True:
try:
now_cookie_count = self._redisdb.lget_count(self._tab_cookie_pool)
need_cookie_count = self._min_cookies - now_cookie_count
if need_cookie_count > 0:
log.info(
"当前cookie数为 {} 小于 {}, 生产cookie".format(
now_cookie_count, self._min_cookies
)
)
try:
cookies = self.create_cookie()
if cookies:
self.add_cookies(cookies)
except Exception as e:
log.exception(e)
else:
log.info("当前cookie数为 {} 数量足够 暂不生产".format(now_cookie_count))
# 判断cookie池近一分钟数量是否有变化，无变化则认为爬虫不再用了，退出
last_count_info = self._redisdb.strget(
self._tab_cookie_pool_last_count
)
if not last_count_info:
self._redisdb.strset(
self._tab_cookie_pool_last_count,
"{}:{}".format(time.time(), now_cookie_count),
)
else:
last_time, last_count = last_count_info.split(":")
last_time = float(last_time)
last_count = int(last_count)
if time.time() - last_time > 60:
if now_cookie_count == last_count:
log.info("近一分钟，cookie池数量无变化，判定爬虫未使用，退出生产")
break
else:
self._redisdb.strset(
self._tab_cookie_pool_last_count,
"{}:{}".format(time.time(), now_cookie_count),
)
if self._keep_alive:
log.info("sleep 10")
tools.delay_time(10)
else:
break
except Exception as e:
log.exception(e)
tools.delay_time(1)
def run(self):
"""
@summary: 重写run方法 检查mysql中的任务是否做完， 做完停止
---------
---------
@result:
"""
try:
self.create_batch_record_table()
if not self._parsers: # 不是add_parser 模式
self._parsers.append(self)
self._start()
while True:
try:
self.heartbeat()
if (
self.task_is_done() and self.all_thread_is_done()
): # redis全部的任务已经做完 并且mysql中的任务已经做完（检查各个线程all_thread_is_done，防止任务没做完，就更新任务状态，导致程序结束的情况）
if not self._is_notify_end:
self.spider_end()
self.record_spider_state(
spider_type=2,
state=1,
batch_date=self._batch_date_cache,
spider_end_time=tools.get_current_date(),
batch_interval=self._batch_interval,
)
self._is_notify_end = True
if not self._keep_alive:
self._stop_all_thread()
break
else:
self._is_notify_end = False
self.check_task_status()
except Exception as e:
log.exception(e)
tools.delay_time(10) # 10秒钟检查一次爬虫状态
except Exception as e:
msg = "《%s》主线程异常 爬虫结束 exception: %s" % (self._batch_name, e)
log.error(msg)
self.send_msg(msg,
level="error",
message_prefix="《%s》爬虫异常结束".format(self._batch_name))
os._exit(137) # 使退出码为35072 方便爬虫管理器重启
def run(self):
while True:
try:
try:
with RedisLock(
key=self._tab_user_pool, lock_timeout=3600, wait_timeout=0
) as _lock:
if _lock.locked:
for user in self._load_user():
retry_times = 0
while retry_times <= self._login_retry_times:
try:
user = self.login(user)
if user:
self.add_user(user)
else:
self.handle_login_failed_user(user)
break
except NotImplementedError:
log.error(
f"{self.__class__.__name__} must be implementation login method！"
)
os._exit(0)
except Exception as e:
self.handel_exception(e)
log.debug(
f"login failed, user: {user} retry_times: {retry_times}"
)
retry_times += 1
else:
self.handle_login_failed_user(user)
now_user_count = self._redisdb.hget_count(
self._tab_user_pool
)
log.info("当前在线user数为 {}".format(now_user_count))
except Exception as e:
log.exception(e)
if self._keep_alive:
tools.delay_time(10)
else:
break
except Exception as e:
log.exception(e)
tools.delay_time(1)
def run(self):
self._thread_stop = False
while not self._thread_stop:
try:
request = self._collector.get_request()
if not request:
if not self.is_show_tip:
log.debug("等待任务...")
self.is_show_tip = True
continue
self.is_show_tip = False
self.deal_request(request)
except Exception as e:
log.exception(e)
def flush(self):
try:
items = []
update_items = []
requests = []
callbacks = []
items_fingerprints = []
data_count = 0
while not self._items_queue.empty():
data = self._items_queue.get_nowait()
data_count += 1
# data 分类
if callable(data):
callbacks.append(data)
elif isinstance(data, UpdateItem):
update_items.append(data)
elif isinstance(data, Item):
items.append(data)
if setting.ITEM_FILTER_ENABLE:
items_fingerprints.append(data.fingerprint)
else: # request-redis
requests.append(data)
if data_count >= UPLOAD_BATCH_MAX_SIZE:
self.__add_item_to_db(
items, update_items, requests, callbacks, items_fingerprints
)
items = []
update_items = []
requests = []
callbacks = []
items_fingerprints = []
data_count = 0
if data_count:
self.__add_item_to_db(
items, update_items, requests, callbacks, items_fingerprints
)
except Exception as e:
log.exception(e)
def get_cookie(self, wait_when_null=True) -> User:
while True:
try:
user_cookie = self._redisdb.rpoplpush(self._tab_cookie_pool)
if not user_cookie and wait_when_null:
log.info("暂无cookie 生产中...")
self.login()
continue
if user_cookie:
user_cookie = eval(user_cookie)
return User(**user_cookie)
return None
except Exception as e:
log.exception(e)
tools.delay_time(1)
def run(self):
if not self._parsers: # 不是add_parser 模式
self._parsers.append(self)
self._start()
while True:
try:
if self.all_thread_is_done():
self._stop_all_thread()
break
except Exception as e:
log.exception(e)
tools.delay_time(1) # 1秒钟检查一次爬虫状态
self.delete_tables([self._redis_key + "*"])
def __put_requests(self, requests_list):
for request in requests_list:
try:
request_dict = {
"request_obj": Request.from_dict(eval(request)),
"request_redis": request,
}
except Exception as e:
log.exception("""
error %s
request %s
""" % (e, request))
request_dict = None
if request_dict:
self._todo_requests.put(request_dict)
def run(self):
while not self._thread_stop:
try:
request = self._memory_db.get()
if not request:
if not self.is_show_tip:
log.debug("等待任务...")
self.is_show_tip = True
time.sleep(1)
continue
self.is_show_tip = False
self.deal_request(request)
except Exception as e:
log.exception(e)
time.sleep(3)
def run(self):
while not self._thread_stop:
try:
requests = self._memory_db.get()
if not requests:
if not self.is_show_tip:
log.debug("parser 等待任务 ...")
self.is_show_tip = True
time.sleep(1)
self._wait_task_time += 1
continue
self.is_show_tip = False
self.deal_requests([requests])
except Exception as e:
log.exception(e)
def get_cookie(self, wait_when_null=True):
while True:
try:
cookie_info = self._redisdb.rpoplpush(self._tab_cookie_pool)
if not cookie_info and wait_when_null:
log.info("暂无cookie 生产中...")
self._keep_alive = False
self._min_cookies = 1
with RedisLock(
key=self._tab_cookie_pool, lock_timeout=3600, wait_timeout=5
) as _lock:
if _lock.locked:
self.run()
continue
return eval(cookie_info) if cookie_info else {}
except Exception as e:
log.exception(e)
tools.delay_time(1)
def run(self):
while not self._thread_stop:
try:
requests = self._collector.get_requests(setting.SPIDER_TASK_COUNT)
if not requests:
if not self.is_show_tip:
log.info("parser 等待任务 ...")
self.is_show_tip = True
# log.info('parser 等待任务 {}...'.format(tools.format_seconds(self._wait_task_time)))
time.sleep(1)
self._wait_task_time += 1
continue
self.is_show_tip = False
self.deal_requests(requests)
except Exception as e:
log.exception(e)
def save_items(self, table, items: List[Dict]) -> bool:
"""
保存数据
Args:
table: 表名
items: 数据，[{},{},...]
Returns: 是否保存成功 True / False
若False，不会将本批数据入到去重库，以便再次入库
"""
try:
add_count = self.to_db.add_batch(coll_name=table, datas=items)
datas_size = len(items)
log.info("共导出 %s 条数据到 %s, 新增 %s条, 重复 %s 条" %
(datas_size, table, add_count, datas_size - add_count))
return True
except Exception as e:
log.exception(e)
return False
def run(self):
self.start_callback()
for i in range(self._thread_count):
parser_control = AirSpiderParserControl(self._memory_db,
self._item_buffer)
parser_control.add_parser(self)
parser_control.start()
self._parser_controls.append(parser_control)
self._item_buffer.start()
self.distribute_task()
while True:
try:
if self.all_thread_is_done():
# 停止 parser_controls
for parser_control in self._parser_controls:
parser_control.stop()
# 关闭item_buffer
self._item_buffer.stop()
# 关闭webdirver
if Request.webdriver_pool:
Request.webdriver_pool.close()
log.info("无任务，爬虫结束")
break
except Exception as e:
log.exception(e)
tools.delay_time(1) # 1秒钟检查一次爬虫状态
self.end_callback()
# 为了线程可重复start
self._started.clear()
# 关闭打点
metrics.close()
def get_user(self, block=True) -> Optional[NormalUser]:
while True:
try:
user_id = self._get_user_id()
user_str = None
if user_id:
user_str = self._redisdb.hget(self._tab_user_pool, user_id)
# 如果没取到user，可能是其他爬虫将此用户删除了，需要重刷新本地缓存的用户id
if not user_str:
self._load_users_id()
continue
if not user_id and block:
self._keep_alive = False
self.run()
continue
return user_str and NormalUser(**eval(user_str))
except Exception as e:
log.exception(e)
tools.delay_time(1)
def reput_failed_requests_to_requests(self):
log.debug("正在重置失败的requests...")
total_count = 0
while True:
try:
failed_requests = self.get_failed_requests()
if not failed_requests:
break
for request in failed_requests:
request["retry_times"] = 0
request_obj = Request.from_dict(request)
self._request_buffer.put_request(request_obj)
total_count += 1
except Exception as e:
log.exception(e)
self._request_buffer.flush()
log.debug("重置%s条失败requests为待抓取requests" % total_count)
def emit(self, point=None, force=False):
"""
1. 添加新点到 pending
2. 如果符合条件，尝试聚合并打点
3. 更新打点时间
:param point:
:param force: 强制提交所有点 默认False
:return:
"""
if point:
self.pending_points.put(point)
# 判断是否需要提交点 1、数量 2、间隔 3、强力打点
if not (force
or self.pending_points.qsize() >= self.max_points # noqa: W503
or time.time() - self.last_emit_ts >
self.emit_interval # noqa: W503
):
return
# 需要打点，读取可以打点的值, 确保只有一个线程在做点的压缩
with self.lock:
points = self._get_ready_emit(force=force)
if not points:
return
try:
self.influxdb.write_points(
points,
batch_size=self.batch_size,
time_precision=self.time_precision,
retention_policy=self.retention_policy,
)
except Exception:
log.exception("error writing points")
self.last_emit_ts = time.time()
def login(self):
"""
@return: 1 成功 0 失败
"""
try:
# 预检查
if not self.is_time_to_login():
log.info("此账号尚未到登陆时间: {}".format(self.username))
time.sleep(5)
return 0
cookies = self.create_cookie()
if not cookies:
raise Exception("登陆失败 未获取到合法cookie")
if not isinstance(cookies, dict):
raise Exception("cookie 必须为字典格式")
# 保存cookie
self.set_login_time()
self.set_cookies(cookies)
log.info("登录成功 {}".format(self.username))
self.record_user_status(LimitTimesUserStatus.LOGIN_SUCCESS)
return 1
except Exception as e:
log.exception(e)
send_msg(
msg=f"{self.SITE_NAME} {self.username} 账号登陆异常 exception: {str(e)}",
level="error",
message_prefix=f"{self.SITE_NAME} {self.username} 账号登陆异常",
)
log.info("登录失败 {}".format(self.username))
self.record_user_status(LimitTimesUserStatus.LOGIN_FALIED)
return 0
def update_items(self,
table,
items: List[Dict],
update_keys=Tuple) -> bool:
"""
更新数据
Args:
table: 表名
items: 数据，[{},{},...]
update_keys: 更新的字段, 如 ("title", "publish_time")
Returns: 是否更新成功 True / False
若False，不会将本批数据入到去重库，以便再次入库
"""
try:
add_count = self.to_db.add_batch(
coll_name=table,
datas=items,
update_columns=update_keys or list(items[0].keys()),
)
datas_size = len(items)
update_count = datas_size - add_count
msg = "共导出 %s 条数据到 %s, 新增 %s 条, 更新 %s 条" % (
datas_size,
table,
add_count,
update_count,
)
if update_keys:
msg += " 更新字段为 {}".format(update_keys)
log.info(msg)
return True
except Exception as e:
log.exception(e)
return False
