将嵌套字典转换为数据框

Question

我得到了嵌套了dict的数据。

我正在尝试使用json_normalize转换为数据帧，但出现错误。

d = ['{"appMetadata": {"index": "cfs_ccs_eddi_35725", "host": "iaasn00009634", "job": "splunk_scraper"}, "timestampEpochSecond": 1545172308711, "metricTags": {"source": "/local/apps/eddi/presentment/logs/eddi-http-presentment-requests.log"}, "metricName": "splunk.logs.tstats.count.per.min", "metricValue": 5, "metricType": "count"}']

df = pd.DataFrame.from_dict({(i,j): d[i][j] 
                             for i in d.keys() 
                             for j in d[i].keys()},
                            orient='index')

错误是：

  

AttributeError：“列表”对象没有属性“键”

Answer 1

将d列表转换为字典

d = dict(itertools.zip_longest(*[iter(l)] * 2, fillvalue=""))

Answer 2

您有一个list，里面有str，而不是dict或json：

• 请注意，所示数据的dict部分由引号['{bunch_of_stuff}']
包围。
• 可以通过ast.literal_eval
来解决

数据：

d = ['{"appMetadata": {"index": "cfs_ccs_eddi_35725", "host": "iaasn00009634", "job": "splunk_scraper"}, "timestampEpochSecond": 1545172308711, "metricTags": {"source": "/local/apps/eddi/presentment/logs/eddi-http-presentment-requests.log"}, "metricName": "splunk.logs.tstats.count.per.min", "metricValue": 5, "metricType": "count"}']

代码：

from ast import literal_eval

d = literal_eval(d[0])

print(d)

固定d的输出：

{'appMetadata': {'index': 'cfs_ccs_eddi_35725',
  'host': 'iaasn00009634',
  'job': 'splunk_scraper'},
 'timestampEpochSecond': 1545172308711,
 'metricTags': {'source': '/local/apps/eddi/presentment/logs/eddi-http-presentment-requests.log'},
 'metricName': 'splunk.logs.tstats.count.per.min',
 'metricValue': 5,
 'metricType': 'count'}

pandas.io.json.json_normalize：

from pandas.io.json import json_normalize
import pandas as pd

df = json_normalize(d)

print(df)

df的输出：

 timestampEpochSecond                        metricName  metricValue metricType   appMetadata.index appMetadata.host appMetadata.job                                                     metricTags.source
        1545172308711  splunk.logs.tstats.count.per.min            5      count  cfs_ccs_eddi_35725    iaasn00009634  splunk_scraper  /local/apps/eddi/presentment/logs/eddi-http-presentment-requests.log