Pinocchio

<h1 align="center"> <code>pinocchio</code> </h1> <p align="center"> <img width="400" alt="Limestone" src="https://github.com/user-attachments/assets/3a1894b4-403f-4c35-90aa-548e7672fe90" /> </p> <p align="center"> Create Solana programs with no external dependencies attached. </p> <p align="center"> <a href="https://github.com/anza-xyz/pinocchio/actions/workflows/main.yml"><img src="https://img.shields.io/github/actions/workflow/status/anza-xyz/pinocchio/main.yml?logo=GitHub" /></a> <a href="https://crates.io/crates/pinocchio"><img src="https://img.shields.io/crates/v/pinocchio?logo=rust" /></a> <a href="https://docs.rs/pinocchio"><img src="https://img.shields.io/docsrs/pinocchio?logo=docsdotrs" /></a> </p> <p align="right"> <i>I've got no dependencies</i><br /> <i>To hold me down</i><br /> <i>To make me fret</i><br /> <i>Or make me frown</i><br /> <i>I had dependencies</i><br /> <i>But now I'm free</i><br /> <i>There are no dependencies on me</i> </p>

Overview

Pinocchio is a no external dependencies library to create Solana programs in Rust. The only dependencies are types from the Solana SDK specifically designed for on-chain programs. This mitigates dependency issues and offers an efficient zero-copy library to write programs, optimized in terms of both compute units consumption and binary size.

Features

• no_std crate
• only dependencies to Solana SDK types
• efficient program_entrypoint! macro with no copies or allocations
• lightweight lazy_program_entrypoint providing more control over how the input is parsed

Getting started

From your project folder:

cargo add pinocchio

This will add pinocchio as a dependency to your project.

Defining the program entrypoint

A Solana program needs to define an entrypoint, which will be called by the runtime to begin the program execution. The entrypoint! macro emits the common boilerplate to set up the program entrypoint. The macro will also set up global allocator and custom panic hook using the default_allocator! and default_panic_handler! macros.

The entrypoint! is a convenience macro that invokes three other macros to set all components required for a program execution:

• program_entrypoint!: declares the program entrypoint
• default_allocator!: declares the default (bump) global allocator
• default_panic_handler!: declares the default panic "hook" that works in combination with the std panic handler

When all dependencies are no_std, you should use nostd_panic_handler! instead of default_panic_handler! to declare a rust runtime panic handler. There's no need to do this when any dependency is std since rust compiler will emit a panic handler.

To use the entrypoint! macro, use the following in your entrypoint definition:

use pinocchio::{
  AccountView,
  Address,
  entrypoint,
  ProgramResult
};
use solana_program_log::log;

entrypoint!(process_instruction);

pub fn process_instruction(
  program_id: &Address,
  accounts: &mut [AccountView],
  instruction_data: &[u8],
) -> ProgramResult {
  log("Hello from my pinocchio program!");
  Ok(())
}

The information from the input is parsed into their own entities:

• program_id: the ID of the program being called
• accounts: the accounts received
• instruction_data: data for the instruction

pinocchio also offers variations of the program entrypoint (lazy_program_entrypoint) and global allocator (no_allocator). In order to use these, the program needs to specify the program entrypoint, global allocator and panic handler individually. The entrypoint! macro is equivalent to writing:

program_entrypoint!(process_instruction);
default_allocator!();
default_panic_handler!();

Any of these macros can be replaced by alternative implementations.

📌 Custom entrypoints with process_entrypoint

For programs that need maximum control over the entrypoint, pinocchio exposes the process_entrypoint function. This function is the same deserialization logic used internally by the program_entrypoint! macro, exposed as a public API and can be called directly from a custom entrypoint, allowing you to implement fast-path optimizations or custom pre-processing logic before falling back to standard input parsing.

To use process_entrypoint in a custom entrypoint:

use pinocchio::{
  AccountView,
  Address,
  default_panic_handler,
  entrypoint::process_entrypoint,
  MAX_TX_ACCOUNTS,
  no_allocator,
  ProgramResult,
};
use solana_program_log::log;

no_allocator!();
default_panic_handler!();

#[no_mangle]
pub unsafe extern "C" fn entrypoint(input: *mut u8) -> u64 {
    // Fast path: check the number of accounts
    let num_accounts = unsafe { *(input as *const u64) };
    if num_accounts == 0 {
        log("Fast path - no accounts!");
        return 0;
    }

    // Standard path: delegate to `process_entrypoint`
    unsafe { process_entrypoint::<MAX_TX_ACCOUNTS>(input, process_instruction) }
}

pub fn process_instruction(
  program_id: &Address,
  accounts: &mut [AccountView],
  instruction_data: &[u8],
) -> ProgramResult {
  log("Standard path");
  Ok(())
}

📌 lazy_program_entrypoint!

The entrypoint! macro looks similar to the "standard" one found in solana-program-entrypoint. It parses the whole input and provides the program_id, accounts and instruction_data separately. This consumes compute units before the program begins its execution. In some cases, it is beneficial for a program to have more control over when input parsing happens, including whether parsing is needed at all — this is the purpose of the lazy_program_entrypoint! macro. This macro only wraps the program input and provides methods to parse the input on demand.

The lazy_entrypoint is suitable for programs that have a single or very few instructions, since it requires the program to handle the parsing, which can become complex as the number of instructions increases. For larger programs, the program_entrypoint! will likely be easier and more efficient to use.

To use the lazy_program_entrypoint! macro, use the following in your entrypoint definition:

use pinocchio::{
  default_allocator,
  default_panic_handler,
  entrypoint::InstructionContext,
  lazy_program_entrypoint,
  ProgramResult
};

lazy_program_entrypoint!(process_instruction);
default_allocator!();
default_panic_handler!();

pub fn process_instruction(
  mut context: InstructionContext
) -> ProgramResult {
    Ok(())
}

The InstructionContext provides on-demand access to the information of the input:

• remaining(): number of remaining accounts to be parsed
• next_account(): parses the next available account (can be used as many times as accounts remaining)
• instruction_data(): parses the instruction data
• program_id(): parses the program id

⚠️ Note: The lazy_program_entrypoint! does not set up a global allocator nor a panic handler. A program should explicitly use one of the provided macros to set them up or include its own implementation.

📌 no_allocator!

When writing programs, it can be useful to make sure the program does not attempt to make any allocations. In these cases, pinocchio includes a no_allocator! macro that sets a global allocator that panics on any attempt to allocate memory.

To use the no_allocator! macro, use the following in your entrypoint definition:

use pinocchio::{
  AccountView,
  default_panic_handler,
  no_allocator,
  program_entrypoint,
  ProgramResult,
  Address
};

program_entrypoint!(process_instruction);
default_panic_handler!();
no_allocator!();

pub fn process_instruction(
  program_id: &Address,
  accounts: &mut [AccountView],
  instruction_data: &[u8],
) -> ProgramResult {
  Ok(())
}

⚠️ Note: The no_allocator! macro can also be used in combination with the lazy_program_entrypoint!.

Since the no_allocator! macro does not allocate memory, the 32kb memory region reserved for the heap remains unused. To take advantage of this, the no_allocator! macro emits an allocate_unchecked helper function that allows you to manually reserve memory for a type at compile time.

// static allocation:
//    - 0 is the offset when the type will be allocated
//    - `allocate_unchecked` returns a mutable reference to the allocated type
let lamports = allocate_unchecked::<u64>(0);
*lamports = 1_000_000_000;

Note that it is the developer's responsibility to ensure that types do not overlap in memory — the offset + <size of type> of different types must not overlap.

📌 [nostd_panic_handler!](https