Crate syn_solidity

source ·
Expand description

§syn-solidity

syn-powered parser for Solidity-like TokenStreams.

The parsed root element is the File, which contains a list of Items. Items also support outer attributes, as shown below.

⚠️ Work in progress ⚠️

§Design

This parser is specifically designed for Rust procedural macros. It aims to mimic the behavior of the official Solidity compiler (Solc) when it comes to parsing valid Solidity code. This means that all valid Solidity code, as recognized by Solc v0.5.*1 and above, will also be recognized and parsed correctly by syn-solidity.

However, syn-solidity is more permissive and lenient compared to the official Solidity compiler and grammar specifications. Some examples of code patterns that are valid in syn-solidity but not in the official compiler include:

  • identifiers are Rust identifiers (syn::Ident), and as such cannot contain the dollar sign ($), but can contain unicode characters
  • trailing punctuation, like commas (,) in function arguments or enums definitions
  • certain variable and function attributes in certain contexts, like internal functions or functions with implementations ({ ... }) in interfaces
  • parameter storage locations in item definitions, like uint256[] memory in a struct or error definition
  • the tuple type (T, U, ..) is allowed wherever a type is expected, and can optionally be preceded by the tuple keyword. This is the same as ethers.js’s Human-Readable ABI

This lenient behavior is intentionally designed to facilitate usage within procedural macros, and to reduce general code complexity in the parser and AST.

§Known limitations

This parser is limited to only valid Rust tokens, meaning that certain Solidity constructs are not supported. Some examples include, but are not limited to:

  • dollar signs ($) inside of identifiers
  • single quote strings
  • hex and unicode string literal prefixes. Literal prefixes are reserved in Rust edition 2021 and above.
  • "\uXXXX" unicode escapes. Rust uses "\u{XXXX}" for unicode codepoints
  • invalid nested block comments. For example, /*/*/ does not parse.

For the most part, you can copy-paste Solidity code and expect it to parse correctly most of the time. You can see a few examples of Solidity code that parses correctly (after some very light patching) in the tests directory.

§Examples

Basic usage:

use quote::quote;
use syn_solidity::{Expr, File, Item, Lit, Stmt};

// Create a Solidity `TokenStream`
let tokens = quote! {
    /// @name HelloWorld
    /// @notice A hello world example in Solidity.
    contract HelloWorld {
        /// @notice Returns the string "Hello, World!".
        function helloWorld() external pure returns (string memory) {
            return "Hello, World!";
        }
    }
};

// Parse the tokens into a `File`
let ast: File = syn_solidity::parse2(tokens)?;

let items: &[Item] = &ast.items;
let Some(Item::Contract(contract)) = items.first() else {
    unreachable!()
};
assert_eq!(contract.name, "HelloWorld");
assert_eq!(contract.attrs.len(), 2); // doc comments

let body: &[Item] = &contract.body;
let Some(Item::Function(function)) = body.first() else {
    unreachable!()
};
assert_eq!(function.attrs.len(), 1); // doc comment
assert_eq!(function.name.as_ref().unwrap(), "helloWorld");
assert!(function.parameters.is_empty()); // ()
assert_eq!(function.attributes.len(), 2); // external pure
assert!(function.returns.is_some());

let Some([Stmt::Return(ret)]) = function.body() else {
    unreachable!()
};
let Some(Expr::Lit(Lit::Str(s))) = &ret.expr else {
    unreachable!()
};
assert_eq!(s.value(), "Hello, World!");

  1. Older versions may still parse successfully, but this is not guaranteed. 

Re-exports§

Modules§

  • Solidity keywords.
  • Helper trait and methods to manipulate syntax tree nodes’ spans.
  • visitvisit
    Syntax tree traversal to walk a shared borrow of a syntax tree.
  • visit_mutvisit-mut
    Syntax tree traversal to mutate an exclusive borrow of a syntax tree in place.

Macros§

Structs§

  • A list of named or unnamed arguments: { foo: 42, bar: 64 } or (42, 64).
  • A list of flags of an assembly statement.
  • A curly-braced block of statements: { ... }.
  • A catch clause of a StmtTry: catch { ... }.
  • An event parameter.
  • An array literal expression: [a, b, c, d].
  • A binary operation: a + b, a += b.
  • A function call expression: foo(42) or foo({ bar: 42 }).
  • Function call options: foo.bar{ value: 1, gas: 2 }.
  • A unary delete expression: delete vector.
  • A square bracketed indexing expression: vector[2].
  • Access of a named member: obj.k.
  • A new expression: new Contract.
  • A payable expression: payable(address(0x...)).
  • A postfix unary expression: foo++.
  • A ternary (AKA conditional) expression: foo ? bar : baz.
  • A tuple expression: (a, b, c, d).
  • A type() expression: type(uint256)
  • A unary operation: !x, -x.
  • A Solidity file. The root of the AST.
  • A list of unique function attributes. Used in ItemFunction.
  • A hex string.
  • An import alias.
  • A list of import aliases: { Foo as Bar, Baz } from "foo.sol".
  • An import directive: import "foo.sol";.
  • A glob import directive: * as Foo from "foo.sol".
  • A plain import directive: import "foo.sol" as Foo;.
  • A list of inheritance specifiers of an ItemContract: is ERC20("Token", "TKN"), Ownable.
  • A contract, abstract contract, interface, or library definition: contract Foo is Bar("foo"), Baz { ... }.
  • An enum definition: enum Foo { A, B, C }.
  • An error definition: error Foo(uint256 a, uint256 b);.
  • A function, constructor, fallback, receive, or modifier definition: function helloWorld() external pure returns(string memory);.
  • A struct definition: struct Foo { uint256 bar; }.
  • A user-defined value type definition: type Foo is uint256;.
  • A modifier invocation, or an inheritance specifier.
  • A named argument in an argument list: foo: uint256(42).
  • A named argument list: { foo: uint256(42), bar: true }.
  • The override attribute.
  • A list of VariableDeclarations, separated by P.
  • A pragma directive: pragma solidity ^0.8.0;
  • The returns attribute of a function.
  • A Solidity identifier.
  • A list of identifiers, separated by dots.
  • An assembly block, with optional flags: assembly "evmasm" { ... }.
  • A break statement: break;.
  • A continue statement: continue;.
  • A do-while statement: do { ... } while (condition);.
  • An emit statement: emit FooBar(42);.
  • An expression with a trailing semicolon.
  • A for statement: for (uint256 i; i < 42; ++i) { ... }.
  • An if statement with an optional else block: if (expr) { ... } else { ... }.
  • A return statement: return 42;.
  • A revert statement: revert("error");.
  • A try statement: try fooBar(42) catch { ... }.
  • A variable declaration statement: uint256 foo = 42;.
  • A while statement: while (i < 42) { ... }.
  • An array type.
  • A function type: function() returns (string memory).
  • A mapping type: mapping(uint key => string value)
  • A tuple type.
  • An unchecked block: unchecked { ... }.
  • A unicode string.
  • A using directive: using { A, B.mul as * } for uint256 global;.
  • A declaration of variables in a tuple: (,,uint256 foo,string memory bar).
  • A list of unique variable attributes.
  • A variable declaration: string memory hello.
  • An enum variant.
  • Represents the walrus operator :=.
  • A Yul block contains YulStmt between curly braces.
  • Represents a non-default case of a Yul switch statement.
  • Yul function call.
  • Yul for loop e.g for {let i := 0} lt(i,10) {i := add(i,1)} {mstore(i,7)}.
  • Yul function definition: function f() -> a, b { ... }.
  • A Yul identifier.
  • A Yul if statement: if lt(a, b) { sstore(0, 1) }.
  • In inline assembly, only dot-less identifiers can be declared, but dotted paths can reference declarations made outside the assembly block.
  • The return attribute of a Yul function definition.
  • A Yul switch statement can consist of only a default-case or one or more non-default cases optionally followed by a default-case.
  • Represents the default case of a Yul switch statement.
  • Yul variable assignment. x := 0 or x, y := foo(). Assigning values to multiple variables requires a function call.
  • Declares Yul variables, which may or may not have initial values. E.x. let x := 0 let x let x, y := foo() let x, y, z

Enums§

Functions§

Type Aliases§