Skip to content

Latest commit

 

History

History
 
 

tutorials

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
 
 
 
 

substrate-api-client tutorial

This little guide elaborates on the usage of the client we wrote to interact with substrate based blockchains. We will show from scratch how to setup interaction with a substrate blockchain.

First, we need to get a node running. We will use a custom substrate node found in our substrate-test-nodes repository.

  • Download and build the node:

    git clone https://github.com/scs/substrate-test-nodes
    cd substrate-test-nodes/
    git checkout api-M1.1
    cargo build --release
  • Run the node. The dev flag allows to run the node on its own without performing consensus with other nodes.

    ./target/release/substrate-test-node --dev

Now we will start to write client code that interacts with the node. The test-node includes a minimalistic Kitty runtime module derived from the Substrate Collectables Workshop. Let's start from scratch by making a new rust project called api-client-tutorial.

    cd $HOME
    cargo new api-client-tutorial

The Cargo.toml has to have one sole dependency and should look like:

[package]
name = "api-client-tutorial"
version = "0.1.0"
authors = ["Supercomputing Systems AG <info@scs.ch>"]
edition = "2018"

[dependencies]
substrate-api-client = { git = "https://github.com/scs/substrate-api-client.git" }

If we don't now exactly what our blockchain node features or what the runtime module is called we want to interact with, we can query the node metadata with our client. In the src/main.rs we will do two things:

  • First, we instantiate an Api that connects to a given url.
  • Second, we query the node metadata with the api.get_metadata() and print it in pretty json format afterwards.

src/main.rs:

use substrate_api_client::{Api, node_metadata};

fn main() {
    // instantiate an Api that connects to the given address
    let url = "127.0.0.1:9944";
    // if no signer is set in the whole program, we need to give to Api a specific type instead of an associated type
    // as during compilation the type needs to be defined.
    let api = Api::<sr25519::Pair>::new(format!("ws://{}", url));

    let meta = api.get_metadata();
    println!("Metadata:\n {}", node_metadata::pretty_format(&meta).unwrap());
}

If we now run the binary with cargo run, the metadata is printed to the terminal. The following exempt will be found along the metadata, which tells us that there is a KittyModule followed by

  • storage: data that is stored on chain
  • calls: runtime functions that can be called from the outside and
  • events: callbacks that are fired from the runtime that a client can subscribe to
    ...
    {
     "name": "KittyModule",
     "storage": {
      "prefix": "Kitty",
      "entries": [
       {
        "name": "Kitties",
        "modifier": "Default",
        "ty": {
         "Map": {
          "hasher": "Blake2_256",
          "key": "u64",
          "value": "Kitty<T::Hash, T::Balance>",
          "is_linked": false
         }
        },
        "default": [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
         0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0],
        "documentation": []
       },
       {
        "name": "KittyCount",
        "modifier": "Default",
        "ty": {
         "Plain": "u64"
        },
        "default": [0,0,0,0,0,0,0,0],
        "documentation": []
       },
       {
        "name": "KittyIndex",
        "modifier": "Default",
        "ty": {
         "Map": {
          "hasher": "Blake2_256",
          "key": "T::AccountId",
          "value": "u64",
          "is_linked": false
         }
        },
        "default": [0,0,0,0,0,0,0,0],
        "documentation": []
       }
      ]
     },
     "calls": [
      {
       "name": "create_kitty",
       "arguments": [
        {
         "name": "price",
         "ty": "T::Balance"
        }
       ],
       "documentation": []
      },
      {
       "name": "update_kitty",
       "arguments": [
        {
         "name": "price",
         "ty": "T::Balance"
        }
       ],
       "documentation": []
      }
     ],
     "event": [
      {
       "name": "StoredKitty",
       "arguments": [
        "AccountId",
        "u64"
       ],
       "documentation": []
      },
      {
       "name": "UpdatedKitty",
       "arguments": [
        "AccountId",
        "u64"
       ],
       "documentation": []
      }
     ],
     "constants": []
    }
    ...

As we have just created our node, no Kitty is stored on our substrate blockchain yet. But we find in the calls that there is a call named create_kitty, which presumably creates a Kitty with a price supplied as argument. We will call that function in order to create a Kitty. Calling a runtime function in substrate is done via an extrinsic, which is somewhat like a transaction in general blockchain jargon but not exactly, see Extrinsics. An extrinsic is always signed by the account that submits the extrinsic. Therefore, we set the signer field of the Api with an account, that is then used to sign the extrinsic.

let signer = AccountKeyring::Alice.pair();

let api = Api::new(format!("ws://{}", url)).set_signer(signer);

AccountKeyring belongs to the substrate crate keyring that offers some predefined keys facilitating smooth developer experience. Now we are ready to create an extrinsic for our KittyModule, which is performed via the compose_extrinsic! macro.

let xt: UncheckedExtrinsicV3<_, sr25519::Pair> = compose_extrinsic!(
    api.clone(),
    "KittyModule",
    "create_kitty",
    10 as u128
);

The first three arguments are always the Api, the runtime module name, and then the function name as defined in the metadata. Subsequently, the arguments of the runtime function that is called are supplied. Taking a look at the metadata, we see that the create_kitty call takes one argument, which is price of type T::Balance, which is a type alias defined in substrate for a u128. Therefore, we are free to use u128 as both encode to the same bytes. The macro does then query the account nonce of the sender from the node and creates a signed extrinsic ready to be encoded and sent. We have to explicitly put a type annotation for the UncheckedExtrinsic as ed25519 could also be used and the macro is not able to infer the type as macro expansion happens before names are resolved and types are inferred.

Note: The signing process is not straight forward. Additional information is included in the signing payload which is more than the extrinsic payload that only consists of the prepared call statements. All the details can be found in the code.

The following call submits an extrinsic to the node and waits for the transaction hash that is returned upon block inclusion of the extrinsic.

let tx_hash = api.send_extrinsic(xt.hex_encode()).unwrap();

Having received the transaction hash, we can check if the Kitty belonging to Alice's account has successfully been created. Again, looking at the metadata unveils that the there are two storage maps that are of interest to us. there is KittyIndex, which maps an AccountId to an index (u64) and there is Kitties, which maps this index to a Kitty. This indirect approach is overall computationally more efficient than a mapping from AccountId to Kitty directly. Thus, we need to lookup the index before we can access our Kitty.

We can query the storage values of a runtime module via the Api's get_storage method.

let res_str = api.get_storage("Kitty",
                              "KittyIndex",
                              Some(signer.public().encode())).unwrap();

The first argument of the get_storage call is always the storage Prefix followed by the entry name. If the storage value is a map, the key needs to be suplied as argument. The RPC interface of substrate does return a hex encoded string. In the substrate_api_client::util module reside several functions to handle those values. The following fits our needs:

let index = hexstr_to_u64(res_str).unwrap();

Now that we now the index of our Kitty in the Kitties map we can finally have a look at our Kitty.

let res_str = api.get_storage("Kitty",
                              "Kitties",
                              Some(index.encode())).unwrap();

let res_vec = Vec::from_hex(res_str).unwrap();

Naturally, the utils module does not have a hexstr_to_kitty function, instead we can transform it into a byte vector. Now we must decode this vector into a Kitty, but neither Rust nor substrate does know the structure of our Kitty. But luckily, the metadata does! In the metadata we find that our Kitty looks as follows: Kitty<T::Hash, T::Balance>. Hence, we can define the structure on the client side and tell rust what to decode into. T::Hash is again a substrate type alias for a 32 byte array aka [u8; 32].

#[derive(Encode, Decode, Debug)]
struct Kitty {
    id: [u8; 32],
    price: u128,
}

// the part after the colon explicitly tells rust the expected type.
let kitty: Kitty = Decode::decode(&mut res_vec.as_slice()).unwrap();
println!("[+] Cute decoded Kitty: {:?}\n", kitty);

In order to use the Decode::decode function we need to add the Parity's Codec crate as a dependency to the Cargo.toml. This is a basic Rust thing, and shall be left as an exercise to the reader.

This concludes this little tutorial. We went through important features of the substrate-api-client, namely creating extrinsics for custom runtime modules, reading storage values and decoding custom storage values. However, there are more features to our substrate api client such as predefined extrinsics for the srml-contract module, examples on how to listen to runtime events and more. The full code for this tutorial is found in the tutorials folder of the substrate-api-client.