以下代码的python版本是2.7.15
工作中需要对平台指标的数据进行二次汇聚，故需要写部分python脚本，故在此记录一下。
先看一下Base类。
由于URL涉及公司业务，所以暂时删除掉了；该类主要封装了formatURL主要是对url进行一个format，request就是一个http请求工具，getMetricData是对指标数据进行汇聚，dictTransformToList是将dict转化为列表

import requests
import re
import json
class Base(object):
  def __init__(self):
    self.name = 'Base'
    self.URL = ''
  def formatUrl(self, url, _dict):
    keys = re.findall(r'{\w+}', url)
    for key in keys:
      if _dict.has_key(key[1:-1]):
        url = url.replace('$' + key, _dict[key[1:-1]])
    return url
  def request(self, *args):
    url = self.formatUrl(self.URL, {'metricName': args[0] if len(args) > 0 else self.metricName, 'appID': self.appID})
    response = None
    try:
      response = {"data": requests.get(url, timeout=5000), "errorCode": 0}
    except requests.exceptions.RequestException:
      response = {"data": {}, "errorCode": -1, "errorMsg": "Request exception"}
    except requests.exceptions.HTTPError:
      response = {"data": {}, "errorCode": -2, "errorMsg": "HHTP exception"}
    except requests.exceptions.ConnectionError:
      response = {"data": {}, "errorCode": -3, "errorMsg": "Connection exception"}
    except requests.exceptions.Timeout:
      response = {"data": {}, "errorCode": -4, "errorMsg": "Timeout exception"}
    except requests.exceptions.Exception:
      response = {"data": {}, "errorCode": -5, "errorMsg": "Unknown exception"}
    finally:
      print type(response['data'])
      if response["errorCode"] == 0:
        return response['data']['text']
      else:
        return {'metrics': []}
  def getMetricsData(self, metricList):
    metricData = {}
    for item in metricList:
      metricNameslist = item['metricname'].split('#')
      parentKeyName = metricNameslist[self.metricKeysIndics[0]]
      childKeyName = metricNameslist[self.metricKeysIndics[1]] if len(self.metricKeysIndics) == 2 else ''
      metricData[parentKeyName] = metricData[parentKeyName] if metricData.has_key(parentKeyName) else {}
      for metricNameIndex in self.metricNameIndics:
        metricName = metricNameslist[metricNameIndex]
        metricData[parentKeyName][metricName] = item['metrics'].values().pop()
    return metricData
  def dictTransformToList(self, _dict):
    return map(lambda key:dict({'_key': key}, **_dict[key]), _dict.keys())

抽出来一个指标服务进行举例说明：
从ambari中获取指标数据，然后进行整理，比如计算集群的某个值，该集群有20个节点，那么将20个节点的数据进行统计，进行汇总出来集群的数据，如下所示：

from base import Base
class Zookeeper(Base):
  def __init__(self):
    super(Zookeeper, self).__init__()
    self.appID = 'zookeeper'
    self.metricName = 'zkInfo%23%25'
    self.metricPrefix = '_zkInfo#'
    self.metricKeysIndics = [2]
    self.metricNameIndics = [1]
    self.ret = {'errorcode' : 0, 'errormsg' : '' }
  def run(self):
    data = self.request()
    if len(data['metrics']) == 0:
      return dict(self.ret, **{
        'errorcode': -1
      })
    response = self.getMetricsData(data['metrics'])
    response = self.dictTransformToList(response)
    response = self.metricReduce(response)
    res = {}
    for key in response.keys():
      res[self.metricPrefix + '_' + key] = response[key]
    print dict(self.ret, ** {
      'data': res
    })
  def metricReduce(self, metricDataList):
    def reduceFn(prev, now):
      return {
       此处指标名称涉及公司业务，故删除，主要是进行reduce计算
      }
    return reduce(reduceFn, metricDataList, {})

完。