将来自 Telegraf 的 JSON 数据扁平化，以便在 ThingsBoard 中使用

我连接了 ThingsBoard 和 Telegraf 以可视化 CPU 使用率，但收到的数据是嵌套 JSON 格式。我尝试了不同的方法，但无法以扁平化的 JSON 格式获取数据。

使用 Telegraf 1.30.0 版本，数据以以下格式返回：

[
{
  "fields":{
    "usage_guest":0、
    "usage_guest_nice":0、
    "usage_idle":91.40585349348385,
    "usage_iowait":0.1691761123329754,
    "usage_irq":0、
    "usage_nice":0.13534088986632262,
    "usage_softirq":0.08458805616646366,
    "usage_steal":0、
    "usage_system":2.317712738961436,
    "usage_user":5.887328709186236
    },
  "name":"yavar"、
  "tags":{}、
  "timestamp":1716900420
},
{
  "fields":{
    "usage_guest":0、
    "usage_guest_nice":0、
    "usage_idle":91.94207836457213,
    "usage_iowait":0.9028960817717189,
    "usage_irq":0、
    "usage_nice":0.30664395229985647,
    "usage_softirq":0.11925042589437249,
    "usage_steal":0、
    "usage_system":1.8568994889266988,
    "usage_user":4.872231686542196
    },
  "name":"yavar"、
  "tags":{}、
  "timestamp":1716900420
},
{
  "fields":{
    "usage_guest":0、
    "usage_guest_nice":0、
    "usage_idle":91.22272957889072,
    "usage_iowait":0.8963301200743917,
    "usage_irq":0、
    "usage_nice":0.35514967021814886,
    "usage_softirq":0.05073566717402127,
    "usage_steal":0、
    "usage_system":2.130898021308797,
    "usage_user":5.344156942329791
    },
  "name":"yavar"、
  "tags":{}、
  "timestamp":1716900420
},
{
  "fields":{
    "usage_guest":0、
    "usage_guest_nice":0、
    "usage_idle":92.3648648648699,
    "usage_iowait":0.4222972972973181,
    "usage_irq":0、
    "usage_nice":0.3209459459459579,
    "usage_softirq":0.05067567567566809,
    "usage_steal":0、
    "usage_system":1.2837837837838315,
    "usage_user":5.557432432432644
    },
  "name":"yavar"、
  "tags":{}、
  "timestamp":1716900420
},
{
  "fields":{
    "usage_guest":0、
    "usage_guest_nice":0、
    "usage_idle":92.40399255626502,
    "usage_iowait":0.16917611233301705,
    "usage_irq":0、
    "usage_nice":0.11842327863312155,
    "usage_softirq":0.033835222466601006,
    "usage_steal":0、
    "usage_system":1.3364912874308539,
    "usage_user":5.938081542887148
    },
  "name":"cpu4"、
  "tags":{}、
  "timestamp":1716900420
},
{
  "fields":{
    "usage_guest":0、
    "usage_guest_nice":0、
    "usage_idle":92.9369918699103,
    "usage_iowait":0.5928184281842386,
    "usage_irq":0、
    "usage_nice":0.16937669376690467,
    "usage_softirq":0.016937669376688058,
    "usage_steal":0、
    "usage_system":1.1856368563684772,
    "usage_user":5.09823848238449
    },
  "name":"yavar"、
  "tags":{}、
  "timestamp":1716900420
},
{
  "fields":{
    "usage_guest":0、
    "usage_guest_nice":0、
    "usage_idle":93.07888040712137,
    "usage_iowait":0.25445292620872184,
    "usage_irq":0、
    "usage_nice":0.15267175572520417,
    "usage_softirq":0.06785411365564227,
    "usage_steal":0、
    "usage_system":1.5436810856660335,
    "usage_user":4.902459711620043
    },
  "name":"yavar"、
  "tags":{}、
  "timestamp":1716900420
},
{
  "fields":{
    "usage_guest":0、
    "usage_guest_nice":0、
    "usage_idle":92.3622845555872,
    "usage_iowait":0.33795201081445,
    "usage_irq":0、
    "usage_nice":0.15207840486649526,
    "usage_softirq":0.03379520108144139,
    "usage_steal":0、
    "usage_system":1.824940858397751,
    "usage_user":5.288948969245875
    },
  "name":"yavar"、
  "tags":{}、
  "timestamp":1716900420
},
{
  "fields":{
    "usage_guest":0、
    "usage_guest_nice":0、
    "usage_idle":92.21845893310595,
    "usage_iowait":0.4678238780693211,
    "usage_irq":0、
    "usage_nice":0.21168501270109033,
    "usage_softirq":0.05503810330228156,
    "usage_steal":0、
    "usage_system":1.682895850973822,
    "usage_user":5.364098221845659
    },
  "name":"yavar"、
  "tags":{}、
  "timestamp":1716900420
}
]

可以使用 Telegraf 的 starlark 处理器来将嵌套的 JSON 数据扁平化为 ThingsBoard 可接受的格式。下面是一个配置示例：

# ... 其他 Telegraf 输入插件配置 ...

[[processors.starlark]]
  source = '''
  def apply(metric):
    fields = metric.fields.copy()
    for k, v in fields.items():
      metric.fields[f"{metric.name}_{k}"] = v
    metric.fields.pop("usage_guest", None)
    metric.fields.pop("usage_guest_nice", None)
    metric.fields.pop("usage_idle", None)
    metric.fields.pop("usage_iowait", None)
    metric.fields.pop("usage_irq", None)
    metric.fields.pop("usage_nice", None)
    metric.fields.pop("usage_softirq", None)
    metric.fields.pop("usage_steal", None)
    metric.fields.pop("usage_system", None)
    metric.fields.pop("usage_user", None)
    return metric
  '''

# ... ThingsBoard 输出插件配置 ...

解释:

1. [[processors.starlark]] : 这行定义了一个新的 Starlark 处理器。
2. source : 包含要执行的 Starlark 脚本。
3. apply(metric) : 这个函数接收一个 metric 对象，该对象代表 Telegraf 收集到的单个数据点。
4. fields = metric.fields.copy() : 复制 fields 字典以避免修改原始数据。
5. for k, v in fields.items(): : 遍历 fields 字典中的每个键值对。
6. metric.fields[f"{metric.name}_{k}"] = v : 使用 name 和原始键名创建一个新的扁平化键名，并将值赋给它。
7. metric.fields.pop("key", None) : 删除原始的嵌套键。
8. return metric : 返回修改后的 metric 对象。

使用此配置，Telegraf 将输出以下格式的扁平化 JSON 数据:

[
  {
    "name": "yavar",
    "tags": {},
    "timestamp": 1716900420,
    "yavar_usage_guest": 0,
    "yavar_usage_guest_nice": 0,
    "yavar_usage_idle": 91.40585349348385,
    "yavar_usage_iowait": 0.1691761123329754,
    "yavar_usage_irq": 0,
    "yavar_usage_nice": 0.13534088986632262,
    "yavar_usage_softirq": 0.08458805616646366,
    "yavar_usage_steal": 0,
    "yavar_usage_system": 2.317712738961436,
    "yavar_usage_user": 5.887328709186236
  },
  # ... 其他数据点 ...
]

请确保根据的实际需求调整 starlark 处理器中的代码，例如修改键名或添加其他逻辑。