pub struct Function { /* private fields */ }
Expand description
An encoder for a function body within the code section.
§Example
use wasm_encoder::{CodeSection, Function, Instruction};
// Define the function body for:
//
// (func (param i32 i32) (result i32)
// local.get 0
// local.get 1
// i32.add)
let locals = vec![];
let mut func = Function::new(locals);
func.instruction(&Instruction::LocalGet(0));
func.instruction(&Instruction::LocalGet(1));
func.instruction(&Instruction::I32Add);
// Add our function to the code section.
let mut code = CodeSection::new();
code.function(&func);
Implementations§
Source§impl Function
impl Function
Sourcepub fn new<L>(locals: L) -> Self
pub fn new<L>(locals: L) -> Self
Create a new function body with the given locals.
The argument is an iterator over (N, Ty)
, which defines
that the next N
locals will be of type Ty
.
For example, a function with locals 0 and 1 of type I32 and local 2 of type F32 would be created as:
let f = Function::new([(2, ValType::I32), (1, ValType::F32)]);
For more information about the code section (and function definition) in the WASM binary format see the WebAssembly spec
Sourcepub fn new_with_locals_types<L>(locals: L) -> Selfwhere
L: IntoIterator<Item = ValType>,
pub fn new_with_locals_types<L>(locals: L) -> Selfwhere
L: IntoIterator<Item = ValType>,
Create a function from a list of locals’ types.
Unlike Function::new
, this constructor simply takes a list of types
which are in order associated with locals.
For example:
let f = Function::new([(2, ValType::I32), (1, ValType::F32)]);
let g = Function::new_with_locals_types([
ValType::I32, ValType::I32, ValType::F32
]);
assert_eq!(f, g)
Sourcepub fn instruction(&mut self, instruction: &Instruction<'_>) -> &mut Self
pub fn instruction(&mut self, instruction: &Instruction<'_>) -> &mut Self
Write an instruction into this function body.
Sourcepub fn raw<B>(&mut self, bytes: B) -> &mut Selfwhere
B: IntoIterator<Item = u8>,
pub fn raw<B>(&mut self, bytes: B) -> &mut Selfwhere
B: IntoIterator<Item = u8>,
Add raw bytes to this function’s body.
Sourcepub fn byte_len(&self) -> usize
pub fn byte_len(&self) -> usize
The number of bytes already added to this function.
This number doesn’t include the variable-width size field that encode
will write before the added bytes, since the size of that field isn’t
known until all the instructions are added to this function.
Sourcepub fn into_raw_body(self) -> Vec<u8>
pub fn into_raw_body(self) -> Vec<u8>
Unwraps and returns the raw byte encoding of this function.
This encoding doesn’t include the variable-width size field
that encode
will write before the added bytes. As such, its
length will match the return value of [byte_len
].
§Use Case
This raw byte form is suitable for later using with
CodeSection::raw
. Note that it differs from what results
from Function::encode
precisely due to the lack of the
length prefix; CodeSection::raw
will use this. Using
Function::encode
instead produces bytes that cannot be fed
into other wasm-encoder types without stripping off the length
prefix, which is awkward and error-prone.
This method combined with CodeSection::raw
may be useful
together if one wants to save the result of function encoding
and use it later: for example, caching the result of some code
generation process.
For example:
let mut f = Function::new([]);
f.instruction(&Instruction::End);
let bytes = f.into_raw_body();
// (save `bytes` somewhere for later use)
let mut code = CodeSection::new();
code.raw(&bytes[..]);
assert_eq!(2, bytes.len()); // Locals count, then `end`
assert_eq!(3, code.byte_len()); // Function length byte, function body