def singalCheck(taskResultId):
# 根据任务编号查询所有任务是否已经完成
executeCount = TaskInfo.select().where(
(TaskInfo.taskResultId == taskResultId) & (TaskInfo.state != 1) & (TaskInfo.state != 6)).count()
subTaskResult = TaskInfo.select().where(TaskInfo.taskResultId == taskResultId).count()
print 'subTaskResult', subTaskResult
print 'executeCount', executeCount
if executeCount == subTaskResult and executeCount != 0:
#将本次任务检查结果明细生成到result目录
checkTaskResult = TaskResult.getOne(TaskResult.taskResultId == taskResultId)
taskId = checkTaskResult.taskId.taskId
result_taskresultid = checkTaskResult.taskResultId
taskCount = TaskInfo.select().where(
(TaskInfo.taskResultId == taskResultId)).count()
onceCount = config.getTaskResultCount
packCount = int(math.ceil(float(taskCount)/float(onceCount)))
#生成检查结果明细文件
print '开始生成检查结果明细文件'
print 'packCount:',packCount
genTaskResultFile(taskId,result_taskresultid,packCount)
print '单次任务检查完毕结束当前任务,开始发送邮件通知'
# 说明该任务结果已经发送完毕，从apscheduler任务调度中删除该任务
scheduler.remove_job(str(taskResultId))
bigTaskId = checkTaskResult.taskId.taskId
taskResultId = checkTaskResult.taskResultId
# 更新上一次子任务的状态
oTime = format(datetime.datetime.now(), '%Y-%m-%d %H:%M:%S')
q = TaskResult.update(state="2", overTime=oTime).where(
TaskResult.taskResultId == taskResultId)
q.execute()
#修改任务结果对应的大任务状态为已完成
q = Task.update(state="2").where(Task.taskId==taskId)
q.execute()
# 任务执行完毕后发送邮件通知
mutil = MailUtil()
# 获取当前任务绑定的邮箱账号
toEmail = TaskResult.getOne(TaskResult.taskResultId == taskResultId).taskId.userId.email
if toEmail is not None:
from_addr = config.SEND_EMAIL
password = config.SEND_EMAIL_PASSWORD
to_addr = toEmail
smtp_server = config.SMTP_SERVER
msg = str.format(config.MAIL_NOTICE, bigTaskId, bigTaskId, str(taskResultId))
subject = config.MAIL_SUBJECT
mutil.sendMail.delay(from_addr, password, to_addr, smtp_server, msg, subject)
else:
print "该任务未绑定接收邮箱,任务结果编号:",taskResultId
def checkAllLz(filePath, taskResultId):
taskResult = TaskResult.getOne(TaskResult.taskResultId == taskResultId)
delayDay = taskResult.taskId.intervalDay
#根据当前任务编号目录获取该目录下的所有xml文件
print '目录路径:',filePath
fnames=os.listdir(filePath)
for name in fnames:
xmlName = filePath+'/'+name
print 'xmlName:',xmlName
data = parseXMLFile(xmlName, 'CheckItem')
print 'data:',data
# 将需要检查的信息入库
for d in data:
cname = d.get('cname')
url = d.get('url')
if url != '':
if url[-1] == '/':
url = url.replace('http://', '')[0:-1].replace(' ','')
if cname!='':
cname = cname.replace(' ','')
webArea = d.get('area')
webtype = d.get('WebType')
# 检查更新company
if Company.getOne(Company.coname == cname) is None:
c = Company()
c.coname = cname
c.save(force_insert=True)
# 检查更新website
if Website.getOne(Website.domain == url) is not None:
q = Website.update(domain=url, type=webtype, area=webArea).where(Website.domain == url)
q.execute()
else:
com = Company.getOne(Company.coname == cname)
w = Website()
w.regID = com
w.domain = url
w.area = webArea
w.type = webtype
w.save(force_insert=True)
updateWeb = Website.getOne(Website.domain == url)
subTask = TaskInfo()
subTask.taskResultId = taskResult
subTask.webId = updateWeb
subTask.state = '1'
subTask.save(force_insert=True)
taskResultId = str(taskResultId)
if delayDay > 0:
# 需要周期执行的任务
executeMultiTaskInfo(taskResultId)
else:
#logger.debug("开始调用单次任务")
# 单次执行的任务
executeSingleTaskInfo(taskResultId)
def fetchCycle(subtaskId, taskResultId, delayTag):
if delayTag:
checkTaskResult = TaskResult.getOne(TaskResult.taskResultId == taskResultId)
bigTaskId = checkTaskResult.taskId.taskId
# 根据任务结果编号获取本次需要检查的任务记录
subTask = TaskInfo.getOne(TaskInfo.id == subtaskId)
if subTask is not None:
companyName = subTask.cname
url = subTask.url
# 抓取检测
fetchWebsite(companyName, bigTaskId, subTask.id, url)
else:
print 'taskinfo记录为空:', subTask
def intervalDelayTask(taskResultId):
# 获取celery中当前已经正在进行的任务数
nowCount = TaskInfo.select().order_by(TaskInfo.id).where((TaskInfo.state == '6'))
if nowCount >= 0:
# 当celery中正在做的任务数量少于指定的数量时，向celery添加需要执行的任务
if nowCount <= config.celeryMaxCount:
taskCount = TaskInfo.select().where(
(TaskInfo.state == '1') & (TaskInfo.taskResultId == taskResultId)).count()
print 'taskCount:', taskCount
if taskCount == 0:
singalCheck(taskResultId)
# 查询该任务设置的延迟时间开启下一次需要检查的任务
interval = Task.getOne(Task.taskId == (
TaskResult.select(TaskResult.taskId).where(TaskResult.taskResultId == taskResultId))).intervalDay
print 'interval:', interval
if interval != "":
# 生成需要轮巡的新主任务结果记录
taskResult = TaskResult()
lastTaskResult = TaskResult.getOne(TaskResult.taskResultId == taskResultId)
taskResult.taskId = lastTaskResult.taskId
taskResult.state = '1'
taskResult.save()
#将上一次的任务结果编号所对应的webId指定给新的任务结果
psql_db.transaction()
try:
query = (TaskInfo
.insert_from(
fields=[TaskInfo.webId],
query=TaskInfo.select(TaskInfo.webId).where(TaskInfo.taskResultId == lastTaskResult)))
query.execute()
q = TaskInfo.update(taskResultId=taskResult).where(TaskInfo.taskResultId.is_null())
q.execute()
except Exception, e:
print e
psql_db.rollback()
# 获取当前时间
ctime = datetime.datetime.now()
delay_time = int(interval)
stime = ctime + datetime.timedelta(seconds=delay_time)
scheduler.add_job(intervalDelayTask, "date", next_run_time=stime, args=[taskResult.taskResultId],
jobstore="default", id=taskResult.taskResultId)
else:
tasks = TaskInfo.select().order_by(TaskInfo.id).paginate(0, config.sendCeleryCount).where(
(TaskInfo.taskResultId == taskResultId) & (TaskInfo.state == '1'))
for subTask in tasks:
subtaskId = subTask.id
fetchCycle.apply_async((subtaskId,), queue="celery")
# 更新taskinfo状态为已发送
q = TaskInfo.update(state='6').where(TaskInfo.id == subtaskId)
q.execute()
