What does the JSON for batch invalidation of a Cloudfront distribution look like?

Question

I'm trying to invalidate a single file on a AWS Cloudfront distribution using the (as of now) experimental aws cloudfront CLI tool. I cannot figure out how to format the JSON it expects for passing to the --invalidation-batch parameter.

The only documentation I found only explains the XML it expects, yet I am having a hard time translating that into JSON: http://docs.aws.amazon.com/AmazonCloudFront/latest/APIReference/CreateInvalidation.html

I tried:

{
   "Paths" : {
       "Quantity" : 1,
       "Items" : ["/foobar.js"]
   },
   "CallerReference" : "foo-bar-baz"
}

Has anyone yet used this and figured out to format the JSON?

Answer 1

You can generate sample JSON using the following command.

$ aws cloudfront create-invalidation --generate-cli-skeleton | vi -

Output:

{
"DistributionId": "",
"InvalidationBatch": {
    "Paths": {
        "Quantity": 0,
        "Items": [
            ""
        ]
    },
    "CallerReference": ""
}
}

Answer 2

Here's a complete, working example of doing it from a Bash script, including handling the need to create a unique id for each invalidation (unique enough here for my purposes):

INVALIDATION_ID=$(date +"%S")
INVALIDATION_JSON="{
    \"DistributionId\": \"YOUR_ID\",
    \"InvalidationBatch\": {
        \"Paths\": {
            \"Quantity\": 2,
            \"Items\": [
                \"/foo.png\",
                \"/bar.jpg\"
            ]
        },
        \"CallerReference\": \"$INVALIDATION_ID\"
    }
}"

aws cloudfront create-invalidation --cli-input-json "$INVALIDATION_JSON"