class Pipeline():
'''
存储结果管道
（不负责压next task）
'''
def __init__(self, task):
self.task = task
self.taskUtil = TaskUtil()
self.mongoUtil = MongoUtil()
pass
def run(self):
'''
分发
:return: 无
'''
Log.i('Pipeline.run()')
if self.task['results'] is not None and len(self.task['results']) > 0:
#下次任务入队列
if self.task['next_tasks'] is not None:
for next_task in self.task['next_tasks']:
self.taskUtil.insert_one(next_task)
#本次解析结果入库
# 利用反射机制自动执行pipeline_<parser名>（）函数，如果找不到则执行默认的pipeline_default()函数
if hasattr(self, 'pipeline_' + self.task['parser']):
func = getattr(self, 'pipeline_' + self.task['parser'])
func(self.task['table'])
else:
self.pipeline_default(self.task['table'])
#将完整task存入mongo，并将本条task
self.task['state'] = 'done'
self.taskUtil.replace_one(self.task['_id'], self.task)
else:
#没有解析出结果，则表示中间出错了，等待下次再启动
pass
Log.i('this task is finished')
def pipeline_default(self, collection_name):
'''
存储demo
demo_info
{
'_id':'http://tieba.baidu.com',
'name':'百度贴吧'
}
:return:
'''
if self.task['parent'] is None:
self.task['parent'] = {}
if self.task['results'] is not None:
for result in self.task['results']:
insert_data = dict(self.task['parent'], **result)
self.mongoUtil.insert(collection_name=collection_name,
insert_data=insert_data)
pass
def __del__(self):
self.mongoUtil.close_conn()
class TaskUtil:
'''
Task操作工具
tasks
{
"parser": "phase"
"request": "http://so.eduyun.cn/synResource",
"response": "<html>...</html>"
"parent": {},
"state": "done",
"uptime":
}
'''
def __init__(self, request):
self.mongoUtil = MongoUtil()
count = self.mongoUtil.count(collection_name='tasks')
if count == 0:
task = None
task_file = os.path.join(
os.path.dirname(os.path.abspath(__file__)), '..', 'conf',
'task.json')
f = open(task_file, encoding='utf-8')
task = f.read()
f.close()
insert_data = json.loads(task)
self.mongoUtil.insert(collection_name='tasks',
insert_data=insert_data)
def get_ready(self):
'''
获取一条待执行的任务（准备状态），并置为doing状态
:return: dict 单条任务
'''
# 过滤条件，不存在state字段或state=ready
filter_dict = {
'$or': [{
'state': {
'$exists': False
}
}, {
'state': 'ready'
}]
}
# 更新条件，将state=doing
update_dict = {'$set': {'state': 'doing'}}
# 执行mongo操作
task = self.mongoUtil.find_one_and_update(collection_name='tasks',
filter_dict=filter_dict,
update_dict=update_dict)
return task
def set_state(self, id, state):
'''
设置任务状态(ready，doing，done)
:param id: str 主键id
:param state: str 更新状态值
:return: 无
'''
filter_dict = {'_id': id}
update_dict = {'$set': {'state': state}}
self.mongoUtil.update(collection_name='tasks',
filter_dict=filter_dict,
update_dict=update_dict)
def replace_one(self, id, task):
'''
更新整个任务
:param id: str
:param task: dict
:return: 无
'''
filter_dict = {'_id': id}
r = self.mongoUtil.find_one_and_replace(collection_name='tasks',
filter_dict=filter_dict,
replace_dict=task)
return r
def insert_one(self, task):
'''
插入一条task
:param parser: str 解析器
:param request: str 请求的url
:return: 无
'''
r = self.mongoUtil.find_one(collection_name='tasks', filter_dict=task)
if r is None:
task['state'] = 'ready'
r = self.mongoUtil.insert(collection_name='tasks',
insert_data=task)
def __del__(self):
self.mongoUtil.close_conn()
class TaskUtil:
'''
Task操作工具
tasks
{
"parser": "phase"
"request": "http://so.eduyun.cn/synResource",
"response": "<html>...</html>"
"parent": {},
"state": "done",
"uptime":
}
'''
def __init__(self):
self.mongoUtil = MongoUtil()
count = self.mongoUtil.count(collection_name='tasks')
if count == 0:
first_task_parser = Setting.FIRST_TASK_PARSER
first_task_url = Setting.FIRST_TASK_URL
first_task_table = Setting.FIRST_TASK_TABLE
insert_data = {
"parser": first_task_parser,
"request": first_task_url,
"table": first_task_table,
"parent": {},
"state": "ready"
}
self.mongoUtil.insert(collection_name='tasks',
insert_data=insert_data)
def get_ready(self):
'''
获取一条待执行的任务（准备状态），并置为doing状态
:return: dict 单条任务
'''
# 过滤条件，不存在state字段或state=ready
filter_dict = {
'$or': [{
'state': {
'$exists': False
}
}, {
'state': 'ready'
}]
}
# 更新条件，将state=doing
update_dict = {'$set': {'state': 'doing'}}
# 执行mongo操作
task = self.mongoUtil.find_one_and_update(collection_name='tasks',
filter_dict=filter_dict,
update_dict=update_dict)
return task
def set_state(self, id, state):
'''
设置任务状态(ready，doing，done)
:param id: str 主键id
:param state: str 更新状态值
:return: 无
'''
filter_dict = {'_id': id}
update_dict = {'$set': {'state': state}}
self.mongoUtil.update(collection_name='tasks',
filter_dict=filter_dict,
update_dict=update_dict)
def replace_one(self, id, task):
'''
更新整个任务
:param id: str
:param task: dict
:return: 无
'''
filter_dict = {'_id': id}
r = self.mongoUtil.find_one_and_replace(collection_name='tasks',
filter_dict=filter_dict,
replace_dict=task)
return r
def insert_one(self, task):
'''
插入一条task
:param parser: str 解析器
:param request: str 请求的url
:return: 无
'''
r = self.mongoUtil.find_one(collection_name='tasks', filter_dict=task)
if r is None:
task['state'] = 'ready'
r = self.mongoUtil.insert(collection_name='tasks',
insert_data=task)
def __del__(self):
self.mongoUtil.close_conn()