extract json in array in AWS Athena

问题

I have sent logs from kubernetes to an S3 bucket and want to query it using Athena

The log looks like this

[{      "date":1589895855.077230,
      "log":"192.168.85.35 - - [19/May/2020:13:44:15 +0000] \"GET /healthz HTTP/1.1\" 200 3284 \"-\" \"ELB-HealthChecker/2.0\" \"-\"",
      "stream":"stdout",
      "time":"2020-05-19T13:44:15.077230187Z",
      "kubernetes":{
         "pod_name":"myapp-deployment-cd984ffb-kjfbm",
         "namespace_name":"master",
         "pod_id":"eace0175-99cd-11ea-95e4-0aee746ae5d6",
         "labels":{
            "app":"myapp",
            "pod-template-hash":"cd984ffb"
          },
         "annotations":{
            "cluster-autoscaler.kubernetes.io/safe-to-evict":"false",
            "kubernetes.io/psp":"eks.privileged"
          },
         "host":"ip-1-1-1-1.eu-north-1.compute.internal",
         "container_name":"myapp",
         "docker_id":"cb2cda1ed46c5f09d15090fc3f654b1de35970001e366923287cfbd4a4abf4a1"
      }
},
{      "date":1589995860.077230,
      "log":"192.168.1.40 - - [20/May/2020:17:31:00 +0000] \"GET /healthz HTTP/1.1\" 200 3284 \"-\" \"ELB-HealthChecker/2.0\" \"-\"",
      "stream":"stdout",
      "time":"2020-05-20T17:31:00.077230187Z",
      "kubernetes":{
         "pod_name":"myapp-deployment-cd984ffb-kjfbm",
         "namespace_name":"master",
         "pod_id":"eace0175-99cd-11ea-95e4-0aee746ae5d6",
         "labels":{
            "app":"myapp",
            "pod-template-hash":"cd984ffb"
          },
         "annotations":{
            "cluster-autoscaler.kubernetes.io/safe-to-evict":"false",
            "kubernetes.io/psp":"eks.privileged"
          },
         "host":"ip-1-1-1-1.eu-north-1.compute.internal",
         "container_name":"myapp",
         "docker_id":"cb2cda1ed46c5f09d15090fc3f654b1de35970001e366923287cfbd4a4abf4a1"
      }
},]

So an array with json object in it basically.

I am using an CREATE EXTERNAL TABLE query in Athena to create the table. What I have tried is:

CREATE EXTERNAL TABLE IF NOT EXISTS athenadb.mytable (
                   `data` string
                 )
                 ROW FORMAT SERDE 'org.openx.data.jsonserde.JsonSerDe'
                 LOCATION 's3://mybucket/testlog'
                 TBLPROPERTIES ('has_encrypted_data'='false');

This only read the first item in the array into the table, unless I specified more rows such as

data1 string

data2 string

data3 string

However since I don't know how many items is in the array I need something more dynamic.

Then I tried this

CREATE EXTERNAL TABLE IF NOT EXISTS athenadb.mytable (
                   `data` string
                 )
                 LOCATION 's3://mybucket/testlog'
                 TBLPROPERTIES ('has_encrypted_data'='false');

Now I get the entire log (both entries) in one row in the table.

From here I have tried to use UNNEST but I get errors that "cannot unnest type: varchar"

What would be the simplest way to get each {} into its own row in the table? Maybe done from the CREATE EXTERNAL TABLE without needing any extra queries afterwards?

Edit:

Tried this as well now

SELECT data
FROM mytable
CROSS JOIN UNNEST(CAST(json_parse(data) AS array)) AS data2

But I get "Unknown type: array"

I found a similar question here: How do I import an array of data into separate rows in a hive table?

But there didn't seem to be any suggested solution that created the wanted result.

回答1:

Combine unnest with casting json to array(json):

SELECT data, e
FROM mytable
CROSS JOIN UNNEST(CAST(json_parse(data) AS array(json))) t(e)

Note: array<json> is a legacy version of array(json) type definition. The latter is SQL standard compliant.

来源：https://stackoverflow.com/questions/61969182/extract-json-in-array-in-aws-athena

标签

presto

amazon-athena