I'm trying to use syn to create an AST from a Rust file and then using quote to write it to another. However, when I write it, it puts extra spaces between everything.
Note that the example below is just to demonstrate the minimum reproducible problem I'm having. I realize that if I just wanted to copy the code over I could copy the file but it doesn't fit my case and I need to use an AST.
pub fn build_file() {
let current_dir = std::env::current_dir().expect("Unable to get current directory");
let rust_file = std::fs::read_to_string(current_dir.join("src").join("lib.rs")).expect("Unable to read rust file");
let ast = syn::parse_file(&rust_file).expect("Unable to create AST from rust file");
match std::fs::write("src/utils.rs", quote::quote!(#ast).to_string());
}
The file that it creates an AST of is this:
#[macro_use]
extern crate foo;
mod test;
fn init(handle: foo::InitHandle) {
handle.add_class::<Test::test>();
}
What it outputs is this:
# [macro_use] extern crate foo ; mod test ; fn init (handle : foo :: InitHandle) { handle . add_class :: < Test :: test > () ; }
I've even tried running it through rustfmt after writing it to the file like so:
utils::write_file("src/utils.rs", quote::quote!(#ast).to_string());
match std::process::Command::new("cargo").arg("fmt").output() {
Ok(_v) => (),
Err(e) => std::process::exit(1),
}
But it doesn't seem to make any difference.
The quote crate is not really concerned with pretty printing the generated code. You can run it through rustfmt, you just have to execute rustfmt src/utils.rs or cargo fmt -- src/utils.rs.
use std::fs;
use std::io;
use std::path::Path;
use std::process::Command;
fn write_and_fmt<P: AsRef<Path>, S: ToString>(path: P, code: S) -> io::Result<()> {
fs::write(&path, code.to_string())?;
Command::new("rustfmt")
.arg(path.as_ref())
.spawn()?
.wait()?;
Ok(())
}
Now you can just execute:
write_and_fmt("src/utils.rs", quote::quote!(#ast)).expect("unable to save or format");
See also "Any interest in a pretty-printing crate for Syn?" on the Rust forum.
As Martin mentioned in his answer, prettyplease can be used to format code fragments, which can be quite useful when testing proc macro where the standard to_string() on proc_macro2::TokenStream is rather hard to read.
Here a code sample to pretty print a proc_macro2::TokenStream parsable as a syn::Item:
fn pretty_print_item(item: proc_macro2::TokenStream) -> String {
let item = syn::parse2(item).unwrap();
let file = syn::File {
attrs: vec![],
items: vec![item],
shebang: None,
};
prettyplease::unparse(&file)
}
I used this in my tests to help me understand where is the wrong generated code:
assert_eq!(
expected.to_string(),
generate_event().to_string(),
"\n\nActual:\n {}",
pretty_print_item(generate_event())
);
Please see the new prettyplease crate. Advantages:
It can be used directly as a library.
It can handle code fragments while rustfmt only handles full files.
It is fast because it uses a simpler algorithm.
Similar to other answers, I also use prettyplease.
I use this little trick to pretty-print a proc_macro2::TokenStream (e.g. what you get from calling quote::quote!):
fn pretty_print(ts: &proc_macro2::TokenStream) -> String {
let file = syn::parse_file(&ts.to_string()).unwrap();
prettyplease::unparse(&file)
}
Basically, I convert the token stream to an unformatted String, then parse that String into a syn::File, and then pass that to prettyplease package.
Usage:
#[test]
fn it_works() {
let tokens = quote::quote! {
struct Foo {
bar: String,
baz: u64,
}
};
let formatted = pretty_print(&tokens);
let expected = "struct Foo {\n bar: String,\n baz: u64,\n}\n";
assert_eq!(formatted, expected);
}
Related
I have a function foo that can't be modified and contains println! and eprintln! code in it.
fn foo() {
println!("hello");
}
After I call the function, I have to test what it printed so I want to capture the stdout/stderr into a variable.
I strongly recommend against doing this, but if you are using nightly and don't mind using a feature that seems unlikely to ever be stabilized, you can directly capture stdout and stderr using hidden functionality of the standard library:
#![feature(internal_output_capture)]
use std::sync::Arc;
fn foo() {
println!("hello");
eprintln!("world");
}
fn main() {
std::io::set_output_capture(Some(Default::default()));
foo();
let captured = std::io::set_output_capture(None);
let captured = captured.unwrap();
let captured = Arc::try_unwrap(captured).unwrap();
let captured = captured.into_inner().unwrap();
let captured = String::from_utf8(captured).unwrap();
assert_eq!(captured, "hello\nworld\n");
}
It's very rare that a function "cannot be changed", so I'd encourage you to do so and use dependency injection instead. For example, if you are able to edit foo but do not want to change its signature, move all the code to a new function with generics which you can test directly:
use std::io::{self, Write};
fn foo() {
foo_inner(io::stdout(), io::stderr()).unwrap()
}
fn foo_inner(mut out: impl Write, mut err: impl Write) -> io::Result<()> {
writeln!(out, "hello")?;
writeln!(err, "world")?;
Ok(())
}
See also:
How can I test stdin and stdout?
How to take ownership of T from Arc<Mutex<T>>?
How do I convert a Vector of bytes (u8) to a string?
Not sure if this would work on windows, but should work on unix like systems. You should replace the file descriptor to something you can read later. I don't think it is really easy.
I would suggest to use stdio_override which already does that for you using files. You can redirect it, then execute the function and the read the file content.
From the example:
use stdio_override::StdoutOverride;
use std::fs;
let file_name = "./test.txt";
let guard = StdoutOverride::override_file(file_name)?;
println!("Isan to Stdout!");
let contents = fs::read_to_string(file_name)?;
assert_eq!("Isan to Stdout!\n", contents);
drop(guard);
println!("Outside!");
The library also support anything that implements AsRawFd, through the override_raw call. Confirming that it will probably just work on unix.
Otherwise, you can check on the implementation on how it is done internally, and maybe you could bypass a writer instead of a file somehow.
Shadow println!:
use std::{fs::File, io::Write, mem::MaybeUninit, sync::Mutex};
static mut FILE: MaybeUninit<Mutex<File>> = MaybeUninit::uninit();
macro_rules! println {
($($tt:tt)*) => {{
unsafe { writeln!(&mut FILE.assume_init_mut().lock().unwrap(), $($tt)*).unwrap(); }
}}
}
fn foo() {
println!("hello");
}
fn main() {
unsafe {
FILE.write(Mutex::new(File::create("out").unwrap()));
}
foo();
}
I'm currently using syn following an example to create an AST that can be mutated. I understand that I can modify the node I'm travesing (as shown below in my current code) but
I'm curious if I can add some code in between the current node and the next node. Is the syn crate capable of this?
use syn::visit_mut::{self, VisitMut};
use syn::Expr;
#[derive(Debug)]
struct MyStruct;
impl VisitMut for MyStruct {
fn visit_expr_mut(&mut self, node: &mut Expr) {
if let Expr::MethodCall(expr) = &node.to_owned() {
// I can modify the existing node like so:
*node = parse_quote!("// Hello World");
// How could I add something after this node and before the next?
}
}
}
pub fn create() {
let current_dir = std::env::current_dir().expect("Unable to get current directory");
let rust_file = std::fs::read_to_string(current_dir.join("src").join("lib.rs")).expect("Unable to read rust file");
let ast = syn::parse_file(&rust_file).expect("Unable to create AST from rust file");
MyStruct.visit_file_mut(&mut ast);
}
Edit to show use case:
The file I'm currently parsing looks like:
#[macro_use]
extern crate foo;
mod test;
fn init(handle: foo::InitHandle) {
handle.add_class::<Test::test>();
}
Let's say that when I read the AST, I want to add another mod and another handle for it like so:
#[macro_use]
extern crate foo;
mod test;
mod store;
fn init(handle: foo::InitHandle) {
handle.add_class::<Test::test>();
handle.add_class::<Store::store>();
}
As I commented, it highly depends on what you want to insert. Because you can't just insert anything before or after node easily.
For your specific case, you could use parse_quote! to produce an ExprBlock.
*node = parse_quote!(
{
#expr;
handle.add_class::<Store::store>();
}
);
Which with the following input:
fn init(handle: foo::InitHandle) {
handle.add_class::<Test::test>();
}
Would produce this output:
fn init(handle: foo::InitHandle) {
{
handle.add_class::<Test::test>();
handle.add_class::<Store::store>();
};
}
(Note I have reformatted the output, to be prettier)
Alternatively, you could override visit_block_mut() instead. That way you'd have access to stmts: Vec<Stmt>, and would be able to insert before and after a Stmt. The downside is that by doing it that way, you wouldn't be able to easily visit all Exprs, as by using visit_expr_mut().
I am implementing a function-like procedural macro which takes a single string literal as an argument, but I don't know how to get the value of the string literal.
If I print the variable, it shows a bunch of fields, which includes both the type and the value. They are clearly there, somewhere. How do I get them?
extern crate proc_macro;
use proc_macro::{TokenStream,TokenTree};
#[proc_macro]
pub fn my_macro(input: TokenStream) -> TokenStream {
let input: Vec<TokenTree> = input.into_iter().collect();
let literal = match &input.get(0) {
Some(TokenTree::Literal(literal)) => literal,
_ => panic!()
};
// can't do anything with "literal"
// println!("{:?}", literal.lit.symbol); says "unknown field"
format!("{:?}", format!("{:?}", literal)).parse().unwrap()
}
#![feature(proc_macro_hygiene)]
extern crate macros;
fn main() {
let value = macros::my_macro!("hahaha");
println!("it is {}", value);
// prints "it is Literal { lit: Lit { kind: Str, symbol: "hahaha", suffix: None }, span: Span { lo: BytePos(100), hi: BytePos(108), ctxt: #0 } }"
}
After running into the same problem countless times already, I finally wrote a library to help with this: litrs on crates.io. It compiles faster than syn and lets you inspect your literals.
use std::convert::TryFrom;
use litrs::StringLit;
use proc_macro::TokenStream;
use quote::quote;
#[proc_macro]
pub fn my_macro(input: TokenStream) -> TokenStream {
let input = input.into_iter().collect::<Vec<_>>();
if input.len() != 1 {
let msg = format!("expected exactly one input token, got {}", input.len());
return quote! { compile_error!(#msg) }.into();
}
let string_lit = match StringLit::try_from(&input[0]) {
// Error if the token is not a string literal
Err(e) => return e.to_compile_error(),
Ok(lit) => lit,
};
// `StringLit::value` returns the actual string value represented by the
// literal. Quotes are removed and escape sequences replaced with the
// corresponding value.
let v = string_lit.value();
// TODO: implement your logic here
}
See the documentation of litrs for more information.
To obtain more information about a literal, litrs uses the Display impl of Literal to obtain a string representation (as it would be written in source code) and then parses that string. For example, if the string starts with 0x one knows it has to be an integer literal, if it starts with r#" one knows it is a raw string literal. The crate syn does exactly the same.
Of course, it seems a bit wasteful to write and run a second parser given that rustc already parsed the literal. Yes, that's unfortunate and having a better API in proc_literal would be preferable. But right now, I think litrs (or syn if you are using syn anyway) are the best solutions.
(PS: I'm usually not a fan of promoting one's own libraries on Stack Overflow, but I am very familiar with the problem OP is having and I very much think litrs is the best tool for the job right now.)
If you're writing procedural macros, I'd recommend that you look into using the crates syn (for parsing) and quote (for code generation) instead of using proc-macro directly, since those are generally easier to deal with.
In this case, you can use syn::parse_macro_input to parse a token stream into any syntatic element of Rust (such as literals, expressions, functions), and will also take care of error messages in case parsing fails.
You can use LitStr to represent a string literal, if that's exactly what you need. The .value() function will give you a String with the contents of that literal.
You can use quote::quote to generate the output of the macro, and use # to insert the contents of a variable into the generated code.
use proc_macro::TokenStream;
use syn::{parse_macro_input, LitStr};
use quote::quote;
#[proc_macro]
pub fn my_macro(input: TokenStream) -> TokenStream {
// macro input must be `LitStr`, which is a string literal.
// if not, a relevant error message will be generated.
let input = parse_macro_input!(input as LitStr);
// get value of the string literal.
let str_value = input.value();
// do something with value...
let str_value = str_value.to_uppercase();
// generate code, include `str_value` variable (automatically encodes
// `String` as a string literal in the generated code)
(quote!{
#str_value
}).into()
}
I always want a string literal, so I found this solution that is good enough. Literal implements ToString, which I can then use with .parse().
#[proc_macro]
pub fn my_macro(input: TokenStream) -> TokenStream {
let input: Vec<TokenTree> = input.into_iter().collect();
let value = match &input.get(0) {
Some(TokenTree::Literal(literal)) => literal.to_string(),
_ => panic!()
};
let str_value: String = value.parse().unwrap();
// do whatever
format!("{:?}", str_value).parse().unwrap()
}
I had similar problem for parsing doc attribute. It is also represented as a TokenStream. This is not exact answer but maybe will guide in a proper direction:
fn from(value: &Vec<Attribute>) -> Vec<String> {
let mut lines = Vec::new();
for attr in value {
if !attr.path.is_ident("doc") {
continue;
}
if let Ok(Meta::NameValue(nv)) = attr.parse_meta() {
if let Lit::Str(lit) = nv.lit {
lines.push(lit.value());
}
}
}
lines
}
I'm looking for the equivalent of file!() & module_path!() in a procedural macro context.
For example, the following doesn't work:
file.rs:
#[some_attribute]
const A: bool = true;
macro.rs:
#[proc_macro_attribute]
pub fn some_attribute(attr: TokenStream, input: TokenStream) -> TokenStream {
println!("{}", file!());
input
}
This prints macro.rs which makes sense, but what I want is file.rs. Is there a way to achieve this? Is there also a similar way for module_path!()?
A requirement of this is that has to happen at compile-time.
I'm trying to create a file in the OUT_DIR containing constant values where the attribute is added with the module and the file that they are in.
I had the same problem and found out that Rust added a new experimential API to Rust macros (#54725) which allows exaclty what you want:
#![feature(proc_macro_span)]
#[proc_macro]
pub(crate) fn do_something(item: TokenStream) -> TokenStream {
let span = Span::call_site();
let source = span.source_file();
format!("println!(r#\"Path: {}\"#)", source.path().to_str().unwrap())
.parse()
.unwrap()
}
use my_macro_crate::*;
fn main() {
println!("Hello, world!");
do_something!();
}
Will output:
Hello, world!
Path: src\main.rs
Important
Apart from this API being experimential, the path might not be a real OS path. This can be the case if the Span was generated by a macro. Visit the documentation here.
The problem here is that println!("{}", file!()); is executed at compile time and not at runtime. Similar to an answer that was recently given here, you can edit the original function and insert your code at the beginning of it, which will be executed at runtime this time. You can still use the procedural macros file!() and module_path!(). Here is a macro.rs with this approach:
#[proc_macro_attribute]
pub fn some_attribute(_attr: TokenStream, input: TokenStream) -> TokenStream {
// prefix to be added to the function's body
let mut prefix: TokenStream = "
println!(\"Called from {:?} inside module path {:?}\",
file!(), module_path!());
".parse().unwrap();
// edit TokenStream
input.into_iter().map(|tt| {
match tt {
TokenTree::Group(ref g) // match function body
if g.delimiter() == proc_macro::Delimiter::Brace => {
// add logic before function body
prefix.extend(g.stream());
// return new function body as TokenTree
TokenTree::Group(proc_macro::Group::new(
proc_macro::Delimiter::Brace, prefix.clone()))
},
other => other, // else just forward
}
}).collect()
}
You can use it like this in your main.rs:
use mylib::some_attribute;
#[some_attribute]
fn yo() -> () { println!("yo"); }
fn main() { yo(); }
Note that the code is added before what's inside of the function's body. We could have inserted it at the end, but this would break the possibility of returning a value without a semicolon.
EDIT: Later realized that the OP wants it to run at compile time.
One uses this to send output to stdout:
println!("some output")
I think there is no corresponding macro to do the same for stderr.
After Rust 1.19
As of Rust 1.19, you can use the eprint and eprintln macros:
fn main() {
eprintln!("This is going to standard error!, {}", "awesome");
}
This was originally proposed in RFC 1896.
Before Rust 1.19
You can see the implementation of println! to dive into exactly how it works, but it was a bit overwhelming when I first read it.
You can format stuff to stderr using similar macros though:
use std::io::Write;
let name = "world";
writeln!(&mut std::io::stderr(), "Hello {}!", name);
This will give you a unused result which must be used warning though, as printing to IO can fail (this is not something we usually think about when printing!). We can see that the existing methods simply panic in this case, so we can update our code to do the same:
use std::io::Write;
let name = "world";
let r = writeln!(&mut std::io::stderr(), "Hello {}!", name);
r.expect("failed printing to stderr");
This is a bit much, so let's wrap it back in a macro:
use std::io::Write;
macro_rules! println_stderr(
($($arg:tt)*) => { {
let r = writeln!(&mut ::std::io::stderr(), $($arg)*);
r.expect("failed printing to stderr");
} }
);
fn main() {
let name = "world";
println_stderr!("Hello {}!", name)
}
print! and println! are convenience methods for writing to standard output. There are other macros with the same formatting features available for different tasks:
write! and writeln! to write a formatted string to a &mut Writer
format! to just generate a formatted String
To write to the standard error stream, you can use e.g. writeln! like this:
use std::io::Write;
fn main() {
let mut stderr = std::io::stderr();
writeln!(&mut stderr, "Error!").unwrap();
}
It's done so:
use std::io::Write;
fn main() {
std::io::stderr().write(b"some output\n");
}
You can test it by sending the program output to /dev/null to ensure it works (I ignore the warning):
$ rustc foo.rs && ./foo > /dev/null
foo.rs:4:5: 4:42 warning: unused result which must be used, #[warn(unused_must_use)] on by default
foo.rs:4 io::stderr().write(b"some output\n");
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
some output
Similarly, one can do the following for stdout:
use std::io::Write;
fn main() {
std::io::stdout().write(b"some output\n");
}
I think this means println! is just a convenience: it's shorter and it also allows some formatting. As an example of the latter, the following displays 0x400:
println!("0x{:x}", 1024u)
While not answering the precise question, maybe it’s of interest that there’s a log crate which specifies an interface for leveled logging that other crates (e.g. env_logger) can fulfill.
The output of such logging will be sent to stderr, and there are additional benefits for users, such as specifying the log level.
This is how using such a logger could look like:
#[macro_use]
extern crate log;
extern crate env_logger;
fn main() {
env_logger::init().unwrap();
error!("this is printed by default");
}
(Example adapted from http://burntsushi.net/rustdoc/env_logger/index.html#example)
Goal
stderr!("Code {}: Danger, Will Robinson! Danger!", 42);
Notes
The other answers generate an unused import warning with the latest nightly, so here's a modern macro that Just Works TM.
Code
macro_rules! stderr {
($($arg:tt)*) => (
use std::io::Write;
match writeln!(&mut ::std::io::stderr(), $($arg)* ) {
Ok(_) => {},
Err(x) => panic!("Unable to write to stderr (file handle closed?): {}", x),
}
)
}