Reading Blobs

Using an agent? Try this prompt

Read a blob from Walrus using curl, first by blob ID and then by Sui object ID. Show me how to get the correct content-type header and how to request a consistency check.

You can read blobs using HTTP GET requests with their blob ID or object ID. Set $AGGREGATOR to an aggregator endpoint from the Network Reference.

Reading a blob right after upload?

When you read through a CDN-fronted aggregator immediately after certification, the CDN might briefly cache a 404 from before the blob propagated. If your app knows the blob was just certified, retry with backoff. See Reading Blobs Right After Upload.

Reading by blob ID

The following curl command reads a blob and writes it to an output file:

$ curl "$AGGREGATOR/v1/blobs/<BLOB_ID>" -o <FILE_NAME>

To print the contents of a blob directly in the terminal:

$ curl "$AGGREGATOR/v1/blobs/<BLOB_ID>"

tip

Modern browsers attempt to sniff the content type for these resources and generally do a good job of inferring content types for media. The aggregator intentionally prevents sniffing from inferring dangerous executable types such as JavaScript or style sheet types.

Reading by object ID

You can also read blobs by using the object ID of a Sui blob object or a shared blob. The following curl command downloads the blob corresponding to a Sui object ID:

$ curl "$AGGREGATOR/v1/blobs/by-object-id/<OBJECT_ID>" -o <FILE_NAME>

Downloading blobs by object ID allows setting HTTP headers. The aggregator recognizes the following attribute keys and returns the values in the corresponding HTTP headers when present: content-disposition, content-encoding, content-language, content-location, content-type, and link.

Consistency checks

The consistency checks performed by the aggregator are the same as those performed by the CLI. For special use cases, you can enable the strict consistency check by adding a query parameter strict_consistency_check=true (starting with v1.35). If the writer of the blob is known and trusted, you can disable the consistency check by adding a query parameter skip_consistency_check=true (starting with v1.36).