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(
table_folder=self._table_folder)
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 __init__(self, table_folder, process_num=None):
"""
@summary:
---------
@param table_folder:
@param process_num: 进程编号
---------
@result:
"""
super(Collector, self).__init__()
self._db = RedisDB()
self._thread_stop = False
self._todo_requests = collections.deque()
self._tab_requests = setting.TAB_REQUSETS.format(
table_folder=table_folder)
self._tab_spider_status = setting.TAB_SPIDER_STATUS.format(
table_folder=table_folder)
self._spider_mark = LOCAL_HOST_IP + ("_%s" % process_num
if process_num else "_0")
self._interval = setting.COLLECTOR_SLEEP_TIME
self._request_count = setting.COLLECTOR_TASK_COUNT
self._is_collector_task = False
self._db.clear(self._tab_spider_status)
def __init__(self, table_folder):
if not hasattr(self, "_table_item"):
super(ItemBuffer, self).__init__()
self._thread_stop = False
self._is_adding_to_db = False
self._table_folder = table_folder
self._items_queue = Queue(maxsize=MAX_ITEM_COUNT)
self._db = RedisDB()
self._table_item = setting.TAB_ITEM
self._table_request = setting.TAB_REQUSETS.format(
table_folder=table_folder)
self._item_tables = {
# 'xxx_item': {'tab_item': 'xxx:xxx_item'} # 记录item名与redis中item名对应关系
}
self._item_update_keys = {
# 'xxx:xxx_item': ['id', 'name'...] # 记录redis中item名与需要更新的key对应关系
}
self._export_data = ExportData(
) if setting.ADD_ITEM_TO_MYSQL else None
self.db_tip()
def __init__(self, table_folder):
super(HandleFailedRequests, self).__init__()
self._table_folder = table_folder
self._redisdb = RedisDB()
self._request_buffer = RequestBuffer(self._table_folder)
self._table_failed_request = setting.TAB_FAILED_REQUSETS.format(
table_folder=table_folder)
def delete_tables(self, delete_tables_list):
if isinstance(delete_tables_list, bool):
delete_tables_list = [self._table_folder + "*"]
elif not isinstance(delete_tables_list, (list, tuple)):
delete_tables_list = [delete_tables_list]
redis = RedisDB()
for delete_tab in delete_tables_list:
if delete_tab == "*":
delete_tab = self._table_folder + "*"
tables = redis.getkeys(delete_tab)
for table in tables:
log.info("正在删除表 %s" % table)
redis.clear(table)
def __init__(self, table_folder):
if not hasattr(self, "_requests_deque"):
super(RequestBuffer, self).__init__()
self._thread_stop = False
self._is_adding_to_db = False
self._requests_deque = collections.deque()
self._del_requests_deque = collections.deque()
self._db = RedisDB()
self._table_request = setting.TAB_REQUSETS.format(
table_folder=table_folder)
self._table_failed_request = setting.TAB_FAILED_REQUSETS.format(
table_folder=table_folder)
def _start(self):
# 启动request_buffer
self._request_buffer.start()
# 启动item_buffer
self._item_buffer.start()
# 启动collector
self._collector.start()
# 启动parser control
for i in range(self._parser_count):
parser_control = self._parser_control_obj(
self._collector,
self._table_folder,
self._request_buffer,
self._item_buffer,
)
for parser in self._parsers:
parser_control.add_parser(parser)
parser_control.start()
self._parser_controls.append(parser_control)
# 下发任务 因为时间可能比较长，放到最后面
if setting.RETRY_FAILED_REQUESTS:
# 重设失败的任务, 不用加锁，原子性操作
handle_failed_requests = HandleFailedRequests(self._table_folder)
handle_failed_requests.reput_failed_requests_to_requests()
# 下发新任务
if self._auto_start_requests: # 自动下发
if self.wait_lock:
# 将添加任务处加锁，防止多进程之间添加重复的任务
with RedisLock(
key=self._spider_name,
timeout=3600,
wait_timeout=60,
redis_cli=RedisDB().get_redis_obj(),
) as lock:
if lock.locked:
self.__add_task()
else:
self.__add_task()
def __init__(self,
name: str,
expire_time: int,
expire_time_record_key=None):
if not name:
raise ValueError("name cant't be None")
if not expire_time:
raise ValueError("please set expire time, units is seconds")
if not self.__class__.redis_db:
self.__class__.redis_db = RedisDB()
self.name = name
self.expire_time = expire_time
self.expire_time_record_key = expire_time_record_key
self.record_expire_time()
self.del_expire_key()
def check_filter_capacity(self):
"""
检测filter状态，如果已满，加载新的filter
@return:
"""
if (not self._check_capacity_time
or time.time() - self._check_capacity_time > 1800):
# with self._thread_lock:
with RedisLock(
key="ScalableBloomFilter",
timeout=300,
wait_timeout=300,
redis_cli=RedisDB().get_redis_obj(),
) as lock: # 全局锁 同一时间只有一个进程在真正的创建新的filter，等这个进程创建完，其他进程只是把刚创建的filter append进来
if lock.locked:
while True:
if self.filters[-1].is_at_capacity:
self.filters.append(self.create_filter())
else:
break
self._check_capacity_time = time.time()
def __init__(self, table_folder):
if not hasattr(self, "_requests_deque"):
super(RequestBuffer, self).__init__()
self._thread_stop = False
self._is_adding_to_db = False
self._requests_deque = collections.deque()
self._del_requests_deque = collections.deque()
self._db = RedisDB()
self._table_request = setting.TAB_REQUSETS.format(
table_folder=table_folder)
self._table_failed_request = setting.TAB_FAILED_REQUSETS.format(
table_folder=table_folder)
if not self.__class__.dedup and setting.REQUEST_FILTER_ENABLE:
self.__class__.dedup = Dedup(
filter_type=Dedup.ExpireFilter,
name=table_folder,
expire_time=2592000,
to_md5=False,
) # 过期时间为一个月
def __init__(
self,
table_folder=None,
parser_count=None,
begin_callback=None,
end_callback=None,
delete_tabs=(),
process_num=None,
auto_stop_when_spider_done=None,
auto_start_requests=None,
send_run_time=True,
batch_interval=0,
*parser_args,
**parser_kwargs
):
"""
@summary: 调度器
---------
@param table_folder: 爬虫request及item存放reis中的文件夹
@param parser_count: 线程数，默认为配置文件中的线程数
@param begin_callback: 爬虫开始回调函数
@param end_callback: 爬虫结束回调函数
@param delete_tabs: 爬虫启动时删除的表，类型: 元组/bool/string。 支持正则
@param process_num: 进程数
@param auto_stop_when_spider_done: 爬虫抓取完毕后是否自动结束或等待任务，默认自动结束
@param auto_start_requests: 爬虫是否自动添加任务
@param send_run_time: 发送运行时间
@param batch_interval: 抓取时间间隔 默认为0 天为单位 多次启动时，只有当前时间与第一次抓取结束的时间间隔大于指定的时间间隔时，爬虫才启动
@param *parser_args: 传给parser下start_requests的参数, tuple()
@param **parser_kwargs: 传给parser下start_requests的参数, dict()
---------
@result:
"""
super(Scheduler, self).__init__()
for key, value in self.__class__.__custom_setting__.items():
setattr(setting, key, value)
self._table_folder = table_folder or setting.TABLE_FOLDER
if not self._table_folder:
raise Exception(
"""
table_folder 为redis中存放request与item的目录。不能为空，
可在setting中配置，如 TABLE_FOLDER = 'test'
或spider初始化时传参, 如 TestSpider(table_folder='test')
"""
)
self._request_buffer = RequestBuffer(table_folder)
self._item_buffer = ItemBuffer(table_folder)
self._collector = Collector(table_folder, process_num)
self._parsers = []
self._parser_controls = []
self._parser_control_obj = PaserControl
self._parser_args = parser_args
self._parser_kwargs = parser_kwargs
self._auto_stop_when_spider_done = (
auto_stop_when_spider_done
if auto_stop_when_spider_done is not None
else setting.AUTO_STOP_WHEN_SPIDER_DONE
)
self._auto_start_requests = (
auto_start_requests
if auto_start_requests is not None
else setting.PARSER_AUTO_START_REQUESTS
)
self._send_run_time = send_run_time
self._batch_interval = batch_interval
self._begin_callback = (
begin_callback
if begin_callback
else lambda: log.info("\n********** spider begin **********")
)
self._end_callback = (
end_callback
if end_callback
else lambda: log.info("\n********** spider end **********")
)
self._parser_count = setting.PARSER_COUNT if not parser_count else parser_count
self._spider_name = table_folder
self._project_name = table_folder.split(":")[0]
self._tab_spider_time = setting.TAB_SPIDER_TIME.format(
table_folder=table_folder
)
self._tab_spider_status = setting.TAB_SPIDER_STATUS.format(
table_folder=table_folder
)
self._tab_requests = setting.TAB_REQUSETS.format(table_folder=table_folder)
self._tab_failed_requests = setting.TAB_FAILED_REQUSETS.format(
table_folder=table_folder
)
self._is_notify_end = False # 是否已经通知结束
self._last_task_count = 0 # 最近一次任务数量
self._redisdb = RedisDB()
self._project_total_state_table = "{}_total_state".format(self._project_name)
self._is_exist_project_total_state_table = False
# Request 缓存设置
Request.cached_table_folder = table_folder
Request.cached_expire_time = setting.RESPONSE_CACHED_EXPIRE_TIME
delete_tabs = delete_tabs or setting.DELETE_TABS
if delete_tabs:
self.delete_tables(delete_tabs)
self._last_check_task_status_time = 0
def __init__(self,
task_table,
batch_record_table,
batch_name,
batch_interval,
task_keys,
task_state="state",
min_task_count=10000,
check_task_interval=5,
task_limit=10000,
related_table_folder=None,
related_batch_record=None,
task_condition="",
task_order_by="",
table_folder=None,
parser_count=None,
begin_callback=None,
end_callback=None,
delete_tabs=(),
process_num=None,
auto_stop_when_spider_done=None,
send_run_time=False,
*parser_args,
**parser_kwargs):
"""
@summary: 批次爬虫
必要条件
1、需有任务表
任务表中必须有id 及 任务状态字段 如 state。如指定parser_name字段，则任务会自动下发到对应的parser下, 否则会下发到所有的parser下。其他字段可根据爬虫需要的参数自行扩充
参考建表语句如下：
CREATE TABLE `table_name` (
`id` int(11) NOT NULL AUTO_INCREMENT,
`param` varchar(1000) DEFAULT NULL COMMENT '爬虫需要的抓取数据需要的参数',
`state` int(11) DEFAULT NULL COMMENT '任务状态',
`parser_name` varchar(255) DEFAULT NULL COMMENT '任务解析器的脚本类名',
PRIMARY KEY (`id`),
UNIQUE KEY `nui` (`param`) USING BTREE
) ENGINE=InnoDB AUTO_INCREMENT=2 DEFAULT CHARSET=utf8;
2、需有批次记录表 不存在自动创建
此表节结构固定，参考建表语句如下：
CREATE TABLE `xxx_batch_record` (
`id` int(11) NOT NULL AUTO_INCREMENT,
`batch_date` date DEFAULT NULL,
`done_count` int(11) DEFAULT NULL,
`total_count` int(11) DEFAULT NULL,
PRIMARY KEY (`id`)
) ENGINE=InnoDB AUTO_INCREMENT=1 DEFAULT CHARSET=utf8;
---------
@param task_table: mysql中的任务表
@param batch_record_table: mysql 中的批次记录表
@param batch_name: 批次采集程序名称
@param batch_interval: 批次间隔 天为单位。 如想一小时一批次，可写成1/24
@param task_keys: 需要获取的任务字段 列表 [] 如需指定解析的parser，则需将parser_name字段取出来。
@param task_state: mysql中任务表的state字段名
@param min_task_count: redis 中最少任务数, 少于这个数量会从mysql的任务表取任务
@param check_task_interval: 检查是否还有任务的时间间隔；
@param task_limit: 数据库中取任务的数量
@param table_folder: 爬虫request及item存放reis中的文件夹
@param parser_count: 线程数，默认为配置文件中的线程数
@param begin_callback: 爬虫开始回调函数
@param end_callback: 爬虫结束回调函数
@param delete_tabs: 爬虫启动时删除的表，元组类型。 支持正则
@param process_num: 进程数
@param auto_stop_when_spider_done: 爬虫抓取完毕后是否自动结束或等待任务，默认自动结束
@param send_run_time: 发送运行时间
@param related_table_folder: 有关联的其他爬虫任务表（redis）
@param related_batch_record: 有关联的其他爬虫批次表（mysql）注意：要避免环路 如 A -> B & B -> A 。 环路可用related_table_folder指定
related_table_folder 与 related_batch_record 选其一配置即可。
若相关连的爬虫为批次爬虫，推荐以related_batch_record配置，
若相关连的爬虫为普通爬虫，无批次表，可以以related_table_folder配置
@param task_condition: 任务条件 用于从一个大任务表中挑选出数据自己爬虫的任务，及where后的条件语句
@param task_order_by: 取任务时的排序条件 如 id desc
@param *parser_args: 传给parser下start_requests的参数, tuple()
@param **parser_kwargs: 传给parser下start_requests的参数, dict()
---------
@result:
"""
Scheduler.__init__(
self,
table_folder=table_folder,
parser_count=parser_count,
begin_callback=begin_callback,
end_callback=end_callback,
delete_tabs=delete_tabs,
process_num=process_num,
auto_stop_when_spider_done=auto_stop_when_spider_done,
auto_start_requests=False,
send_run_time=send_run_time,
batch_interval=batch_interval,
*parser_args,
**parser_kwargs)
self._redisdb = RedisDB()
self._mysqldb = MysqlDB()
self._request_buffer = RequestBuffer(self._table_folder)
self._task_table = task_table # mysql中的任务表
self._batch_record_table = batch_record_table # mysql 中的批次记录表
self._batch_name = batch_name # 批次采集程序名称
self._task_keys = task_keys # 需要获取的任务字段
self._task_state = task_state # mysql中任务表的state字段名
self._min_task_count = min_task_count # redis 中最少任务数
self._check_task_interval = check_task_interval
self._task_limit = task_limit # mysql中一次取的任务数量
self._related_task_tables = [
setting.TAB_REQUSETS.format(table_folder=table_folder)
] # 自己的task表也需要检查是否有任务
if related_table_folder:
self._related_task_tables.append(
setting.TAB_REQUSETS.format(table_folder=related_table_folder))
self._related_batch_record = related_batch_record
self._task_condition_prefix_and = task_condition and " and {}".format(
task_condition)
self._task_condition_prefix_where = task_condition and " where {}".format(
task_condition)
self._task_order_by = task_order_by and " order by {}".format(
task_order_by)
self._batch_date_cache = None
if self._batch_interval >= 1:
self._date_format = "%Y-%m-%d"
elif self._batch_interval < 1 and self._batch_interval >= 1 / 24:
self._date_format = "%Y-%m-%d %H"
else:
self._date_format = "%Y-%m-%d %H:%M"
# 报警相关
self._send_msg_interval = datetime.timedelta(hours=1) # 每隔1小时发送一次报警
self._last_send_msg_time = None
self._spider_last_done_time = None # 爬虫最近已做任务数量时间
self._spider_last_done_count = 0 # 爬虫最近已做任务数量
self._spider_deal_speed_cached = None
self._is_more_parsers = True # 多模版类爬虫
def task_is_done(self):
"""
@summary: 检查任务状态 是否做完 同时更新批次时间 (不能挂 挂了批次时间就不更新了)
---------
---------
@result: True / False （做完 / 未做完）
"""
is_done = False
# 查看批次记录表任务状态
sql = 'select date_format(batch_date, "{date_format}"), total_count, done_count, is_done from {batch_record_table} order by id desc limit 1'.format(
date_format=self._date_format.replace(":%M", ":%i"),
batch_record_table=self._batch_record_table,
)
batch_info = self._mysqldb.find(sql)
if batch_info is None:
raise Exception("查询批次信息失败")
if batch_info:
self._batch_date_cache, total_count, done_count, is_done = batch_info[
0] # 更新self._batch_date_cache, 防止新批次已经开始了，但self._batch_date_cache还是原来的批次时间
log.info("《%s》 批次时间%s 批次进度 %s/%s 完成状态 %d" % (
self._batch_name,
self._batch_date_cache,
done_count,
total_count,
is_done,
))
os.environ[
"batch_date"] = self._batch_date_cache # 更新BatchParser里边的批次时间
if is_done: # 检查任务表中是否有没做的任务 若有则is_done 为 False
# 比较耗时 加锁防止多进程同时查询
with RedisLock(
key=self._spider_name,
timeout=3600,
wait_timeout=0,
redis_cli=RedisDB().get_redis_obj(),
) as lock:
if lock.locked:
log.info("批次表标记已完成，正在检查任务表是否有未完成的任务")
sql = "select 1 from %s where (%s = 0 or %s=2)%s limit 1" % (
self._task_table,
self._task_state,
self._task_state,
self._task_condition_prefix_and,
)
tasks = self._mysqldb.find(sql) # [(1,)] / []
if tasks:
log.info("检测到任务表中有未完成任务，等待任务下发")
is_done = False
# 更新batch_record 表的is_done 状态，减少查询任务表的次数
sql = 'update {batch_record_table} set is_done = 0 where batch_date = "{batch_date}"'.format(
batch_record_table=self._batch_record_table,
batch_date=self._batch_date_cache,
)
self._mysqldb.update(sql)
else:
log.info("任务表中任务均已完成，爬虫结束")
else:
log.info("批次表标记已完成，其他爬虫进程正在检查任务表是否有未完成的任务，本进程跳过检查，继续等待")
is_done = False
return is_done
def __init__(self, name):
self.name = name
self.count_cached_name = name + "_count_cached"
if not self.__class__.redis_db:
self.__class__.redis_db = RedisDB()
def _cache_db(self):
if not self.__class__.cache_db:
self.__class__.cache_db = RedisDB(
) # .from_url(setting.pika_spider_1_uri)
return self.__class__.cache_db
def __init__(self):
self._redisdb = RedisDB()
self._to_db = MysqlDB()