jmespath.adoc

JMESPath

Table of Contents

• Specification
  • literals
  • projections
    • list: id[*]
    • object: id.*
  • function evaluation: &
  • Current element: @
• Integrations
  • aws cli outputs
  • jmespath.search()

Specification

literals

literals (including numbers, e.g `5`) must be 'quoted' or `quoted`

projections

null values are not included in the final list (unlike multiselect)

list: id[*]

id[?…​] (filters):

$ aws ec2 describe-images --owners self --query 'Images[?Description != None && contains(Description,`ed`)].[ImageId,Description]' --output table

object: id.*

create a list with the objects' values:

"ops": {                        \         \    [
  "functionA": {"numArgs": 2},   \  ops.*  \     {"numArgs": 2},
  "functionB": {"numArgs": 3}    /         /     {"numArgs": 3}
}                               /         /    ]

function evaluation: &

arguments are evaluated before calling a function

sort_by(Contents, &Date)

 Date would return null (it's not in the outer hash)
&Date will be evaluated later

{
  "Contents": [
    {
      "Date": "...",
      ...
    },
    {
      "Date": "...",
      ...
    }
  ]
}

Current element: @

aws iam list-account-aliases --query 'AccountAliases[*].{Alias: @}'

[              \   [
    "alias1",   \      { "Alias": "alias1" },
    "alias2"    /      { "Alias": "alias2" }
]              /   ]

Integrations

aws cli outputs

$ aws kms list-aliases --query 'Aliases[*].AliasName'                        # list of scalars
$ aws kms list-aliases --query 'Aliases[*].[AliasName]'        --output text # list of lists   => newlines
$ aws kms list-aliases --query 'Aliases[*].{Alias: AliasName}' --output text # list of objects => newlines

jmespath.search()

>>> jmespath.search('ls[*][1]', {"ls":[[1,2],[1,2]]})
[2, 2]                          +-- dict
                                +-- json.load(file)