blob: ecc6eb878294df1d9e3276de484af6f1628f6084 [file] [log] [blame] [view]
David Tolnay06faaca2018-09-01 20:25:04 -07001Parser for Rust source code
2===========================
David Tolnay35161ff2016-09-03 11:33:15 -07003
David Tolnayac9953b2016-09-07 08:37:12 -07004[![Build Status](https://api.travis-ci.org/dtolnay/syn.svg?branch=master)](https://travis-ci.org/dtolnay/syn)
5[![Latest Version](https://img.shields.io/crates/v/syn.svg)](https://crates.io/crates/syn)
David Tolnayb28acf32018-09-06 09:01:40 -07006[![Rust Documentation](https://img.shields.io/badge/api-rustdoc-blue.svg)](https://docs.rs/syn/0.15/syn/)
David Tolnay8659a232018-03-31 22:54:03 +02007[![Rustc Version 1.15+](https://img.shields.io/badge/rustc-1.15+-lightgray.svg)](https://blog.rust-lang.org/2017/02/02/Rust-1.15.html)
David Tolnayac9953b2016-09-07 08:37:12 -07008
David Tolnayc088adb2018-01-01 00:26:05 -05009Syn is a parsing library for parsing a stream of Rust tokens into a syntax tree
10of Rust source code.
David Tolnay35161ff2016-09-03 11:33:15 -070011
David Tolnayf594f182018-09-01 20:10:08 -070012Currently this library is geared toward use in Rust procedural macros, but
13contains some APIs that may be useful more generally.
David Tolnayf939f352016-09-11 18:00:09 -070014
David Tolnayc088adb2018-01-01 00:26:05 -050015[custom derive]: https://github.com/rust-lang/rfcs/blob/master/text/1681-macros-1.1.md
David Tolnayf939f352016-09-11 18:00:09 -070016
David Tolnayc088adb2018-01-01 00:26:05 -050017- **Data structures** Syn provides a complete syntax tree that can represent
18 any valid Rust source code. The syntax tree is rooted at [`syn::File`] which
19 represents a full source file, but there are other entry points that may be
20 useful to procedural macros including [`syn::Item`], [`syn::Expr`] and
21 [`syn::Type`].
David Tolnayfb9f7042016-12-22 12:31:39 -050022
David Tolnayc088adb2018-01-01 00:26:05 -050023- **Custom derives** Of particular interest to custom derives is
24 [`syn::DeriveInput`] which is any of the three legal input items to a derive
25 macro. An example below shows using this type in a library that can derive
26 implementations of a trait of your own.
27
David Tolnayf594f182018-09-01 20:10:08 -070028- **Parsing** Parsing in Syn is built around [parser functions] with the
29 signature `fn(ParseStream) -> Result<T>`. Every syntax tree node defined by
30 Syn is individually parsable and may be used as a building block for custom
31 syntaxes, or you may dream up your own brand new syntax without involving any
32 of our syntax tree types.
David Tolnayc088adb2018-01-01 00:26:05 -050033
34- **Location information** Every token parsed by Syn is associated with a
35 `Span` that tracks line and column information back to the source of that
36 token. These spans allow a procedural macro to display detailed error messages
37 pointing to all the right places in the user's code. There is an example of
38 this below.
39
40- **Feature flags** — Functionality is aggressively feature gated so your
41 procedural macros enable only what they need, and do not pay in compile time
42 for all the rest.
43
David Tolnayb28acf32018-09-06 09:01:40 -070044[`syn::File`]: https://docs.rs/syn/0.15/syn/struct.File.html
45[`syn::Item`]: https://docs.rs/syn/0.15/syn/enum.Item.html
46[`syn::Expr`]: https://docs.rs/syn/0.15/syn/enum.Expr.html
47[`syn::Type`]: https://docs.rs/syn/0.15/syn/enum.Type.html
48[`syn::DeriveInput`]: https://docs.rs/syn/0.15/syn/struct.DeriveInput.html
49[parser functions]: https://docs.rs/syn/0.15/syn/parse/index.html
David Tolnayc088adb2018-01-01 00:26:05 -050050
51If you get stuck with anything involving procedural macros in Rust I am happy to
52provide help even if the issue is not related to Syn. Please file a ticket in
53this repo.
54
55*Version requirement: Syn supports any compiler version back to Rust's very
56first support for procedural macros in Rust 1.15.0. Some features especially
57around error reporting are only available in newer compilers or on the nightly
58channel.*
59
David Tolnay86031562018-05-20 22:39:09 -070060[*Release notes*](https://github.com/dtolnay/syn/releases)
61
David Tolnayc088adb2018-01-01 00:26:05 -050062## Example of a custom derive
63
64The canonical custom derive using Syn looks like this. We write an ordinary Rust
65function tagged with a `proc_macro_derive` attribute and the name of the trait
66we are deriving. Any time that derive appears in the user's code, the Rust
67compiler passes their data structure as tokens into our macro. We get to execute
68arbitrary Rust code to figure out what to do with those tokens, then hand some
69tokens back to the compiler to compile into the user's crate.
70
71[`TokenStream`]: https://doc.rust-lang.org/proc_macro/struct.TokenStream.html
David Tolnay6c9f5b62016-09-13 15:19:22 -070072
David Tolnay69b538e2016-09-23 19:59:48 -070073```toml
74[dependencies]
David Tolnayb28acf32018-09-06 09:01:40 -070075syn = "0.15"
David Tolnay87003d02018-05-20 19:45:13 -070076quote = "0.6"
David Tolnay69b538e2016-09-23 19:59:48 -070077
78[lib]
David Tolnay45cc8492016-10-08 20:52:03 -070079proc-macro = true
David Tolnay69b538e2016-09-23 19:59:48 -070080```
81
White-Oak82d0db72016-09-13 21:45:58 +030082```rust
David Tolnay45cc8492016-10-08 20:52:03 -070083extern crate proc_macro;
White-Oak82d0db72016-09-13 21:45:58 +030084
David Tolnayc088adb2018-01-01 00:26:05 -050085use proc_macro::TokenStream;
David Tolnayfd5b1172018-12-31 17:54:36 -050086use quote::quote;
David Tolnayf594f182018-09-01 20:10:08 -070087use syn::{parse_macro_input, DeriveInput};
David Tolnayc088adb2018-01-01 00:26:05 -050088
David Tolnay45cc8492016-10-08 20:52:03 -070089#[proc_macro_derive(MyMacro)]
David Tolnayb4c63262016-09-23 20:03:06 -070090pub fn my_macro(input: TokenStream) -> TokenStream {
David Tolnayc088adb2018-01-01 00:26:05 -050091 // Parse the input tokens into a syntax tree
David Tolnayf594f182018-09-01 20:10:08 -070092 let input = parse_macro_input!(input as DeriveInput);
White-Oak82d0db72016-09-13 21:45:58 +030093
David Tolnay6c9f5b62016-09-13 15:19:22 -070094 // Build the output, possibly using quasi-quotation
95 let expanded = quote! {
96 // ...
97 };
98
David Tolnayc088adb2018-01-01 00:26:05 -050099 // Hand the output tokens back to the compiler
David Tolnay35b498e2018-09-01 20:10:40 -0700100 TokenStream::from(expanded)
White-Oak82d0db72016-09-13 21:45:58 +0300101}
102```
David Tolnay6c9f5b62016-09-13 15:19:22 -0700103
David Tolnayc088adb2018-01-01 00:26:05 -0500104The [`heapsize`] example directory shows a complete working Macros 1.1
David Tolnayb4f57242018-10-28 17:57:08 -0700105implementation of a custom derive. It works on any Rust compiler 1.15+. The
David Tolnayc088adb2018-01-01 00:26:05 -0500106example derives a `HeapSize` trait which computes an estimate of the amount of
107heap memory owned by a value.
David Tolnayb988b6d2016-10-05 00:12:37 -0700108
David Tolnayc088adb2018-01-01 00:26:05 -0500109[`heapsize`]: examples/heapsize
David Tolnayb988b6d2016-10-05 00:12:37 -0700110
111```rust
David Tolnayc088adb2018-01-01 00:26:05 -0500112pub trait HeapSize {
113 /// Total number of bytes of heap memory owned by `self`.
114 fn heap_size_of_children(&self) -> usize;
David Tolnayb988b6d2016-10-05 00:12:37 -0700115}
116```
117
David Tolnayc088adb2018-01-01 00:26:05 -0500118The custom derive allows users to write `#[derive(HeapSize)]` on data structures
119in their program.
David Tolnayb988b6d2016-10-05 00:12:37 -0700120
121```rust
David Tolnayc088adb2018-01-01 00:26:05 -0500122#[derive(HeapSize)]
123struct Demo<'a, T: ?Sized> {
124 a: Box<T>,
125 b: u8,
126 c: &'a str,
127 d: String,
David Tolnayb988b6d2016-10-05 00:12:37 -0700128}
129```
130
David Tolnayc088adb2018-01-01 00:26:05 -0500131## Spans and error reporting
David Tolnayc2263f32017-03-09 19:20:52 -0800132
David Tolnayc088adb2018-01-01 00:26:05 -0500133The token-based procedural macro API provides great control over where the
134compiler's error messages are displayed in user code. Consider the error the
135user sees if one of their field types does not implement `HeapSize`.
David Tolnay736829a2016-12-22 15:55:53 -0500136
137```rust
David Tolnayc088adb2018-01-01 00:26:05 -0500138#[derive(HeapSize)]
139struct Broken {
140 ok: String,
141 bad: std::thread::Thread,
David Tolnay736829a2016-12-22 15:55:53 -0500142}
143```
144
David Tolnayc088adb2018-01-01 00:26:05 -0500145By tracking span information all the way through the expansion of a procedural
David Tolnay3be1b782018-10-28 17:41:41 -0700146macro as shown in the `heapsize` example, token-based macros in Syn are able to
David Tolnayc088adb2018-01-01 00:26:05 -0500147trigger errors that directly pinpoint the source of the problem.
148
149```
150error[E0277]: the trait bound `std::thread::Thread: HeapSize` is not satisfied
151 --> src/main.rs:7:5
152 |
1537 | bad: std::thread::Thread,
154 | ^^^^^^^^^^^^^^^^^^^^^^^^ the trait `HeapSize` is not implemented for `std::thread::Thread`
155```
156
David Tolnayf594f182018-09-01 20:10:08 -0700157## Parsing a custom syntax
David Tolnayc088adb2018-01-01 00:26:05 -0500158
159The [`lazy-static`] example directory shows the implementation of a
160`functionlike!(...)` procedural macro in which the input tokens are parsed using
David Tolnayf594f182018-09-01 20:10:08 -0700161Syn's parsing API.
David Tolnayc088adb2018-01-01 00:26:05 -0500162
163[`lazy-static`]: examples/lazy-static
David Tolnayc088adb2018-01-01 00:26:05 -0500164
165The example reimplements the popular `lazy_static` crate from crates.io as a
166procedural macro.
167
168```
169lazy_static! {
170 static ref USERNAME: Regex = Regex::new("^[a-z0-9_-]{3,16}$").unwrap();
171}
172```
173
174The implementation shows how to trigger custom warnings and error messages on
175the macro input.
176
177```
178warning: come on, pick a more creative name
179 --> src/main.rs:10:16
180 |
18110 | static ref FOO: String = "lazy_static".to_owned();
182 | ^^^
183```
184
David Tolnay941c0922016-12-22 16:06:27 -0500185## Debugging
186
187When developing a procedural macro it can be helpful to look at what the
188generated code looks like. Use `cargo rustc -- -Zunstable-options
David Tolnayc088adb2018-01-01 00:26:05 -0500189--pretty=expanded` or the [`cargo expand`] subcommand.
190
191[`cargo expand`]: https://github.com/dtolnay/cargo-expand
David Tolnay941c0922016-12-22 16:06:27 -0500192
193To show the expanded code for some crate that uses your procedural macro, run
194`cargo expand` from that crate. To show the expanded code for one of your own
195test cases, run `cargo expand --test the_test_case` where the last argument is
196the name of the test file without the `.rs` extension.
197
David Tolnay3bfbd542017-01-16 14:57:53 -0800198This write-up by Brandon W Maister discusses debugging in more detail:
David Tolnayc088adb2018-01-01 00:26:05 -0500199[Debugging Rust's new Custom Derive system][debugging].
200
201[debugging]: https://quodlibetor.github.io/posts/debugging-rusts-new-custom-derive-system/
David Tolnay3bfbd542017-01-16 14:57:53 -0800202
David Tolnay686f5042016-10-30 19:24:51 -0700203## Optional features
204
205Syn puts a lot of functionality behind optional features in order to optimize
David Tolnayc088adb2018-01-01 00:26:05 -0500206compile time for the most common use cases. The following features are
207available.
David Tolnay686f5042016-10-30 19:24:51 -0700208
David Tolnayc088adb2018-01-01 00:26:05 -0500209- **`derive`** *(enabled by default)* — Data structures for representing the
210 possible input to a custom derive, including structs and enums and types.
211- **`full`** — Data structures for representing the syntax tree of all valid
212 Rust source code, including items and expressions.
213- **`parsing`** *(enabled by default)* — Ability to parse input tokens into a
214 syntax tree node of a chosen type.
215- **`printing`** *(enabled by default)* — Ability to print a syntax tree node as
216 tokens of Rust source code.
217- **`visit`** — Trait for traversing a syntax tree.
218- **`visit-mut`** — Trait for traversing and mutating in place a syntax tree.
219- **`fold`** — Trait for transforming an owned syntax tree.
220- **`clone-impls`** *(enabled by default)* — Clone impls for all syntax tree
221 types.
222- **`extra-traits`** — Debug, Eq, PartialEq, Hash impls for all syntax tree
223 types.
David Tolnayd236b552018-04-08 09:02:55 -0700224- **`proc-macro`** *(enabled by default)* — Runtime dependency on the dynamic
225 library libproc_macro from rustc toolchain.
David Tolnayc088adb2018-01-01 00:26:05 -0500226
David Tolnay1e064fe2018-10-28 17:28:41 -0700227## Proc macro shim
David Tolnayc088adb2018-01-01 00:26:05 -0500228
David Tolnay1e064fe2018-10-28 17:28:41 -0700229Syn uses the [proc-macro2] crate to emulate the compiler's procedural macro API
230in a stable way that works all the way back to Rust 1.15.0. This shim makes it
231possible to write code without regard for whether the current compiler version
232supports the features we use.
David Tolnayc088adb2018-01-01 00:26:05 -0500233
David Tolnay1e064fe2018-10-28 17:28:41 -0700234In general all of your code should be written against proc-macro2 rather than
235proc-macro. The one exception is in the signatures of procedural macro entry
236points, which are required by the language to use `proc_macro::TokenStream`.
David Tolnayc088adb2018-01-01 00:26:05 -0500237
David Tolnay1e064fe2018-10-28 17:28:41 -0700238The proc-macro2 crate will automatically detect and use the compiler's data
239structures on sufficiently new compilers.
David Tolnayed7a5082016-10-30 20:06:29 -0700240
David Tolnay807ecff2018-11-21 01:04:53 -0800241[proc-macro2]: https://github.com/alexcrichton/proc-macro2
242
David Tolnay35161ff2016-09-03 11:33:15 -0700243## License
244
245Licensed under either of
246
247 * Apache License, Version 2.0 ([LICENSE-APACHE](LICENSE-APACHE) or http://www.apache.org/licenses/LICENSE-2.0)
248 * MIT license ([LICENSE-MIT](LICENSE-MIT) or http://opensource.org/licenses/MIT)
249
250at your option.
251
252### Contribution
253
254Unless you explicitly state otherwise, any contribution intentionally submitted
255for inclusion in this crate by you, as defined in the Apache-2.0 license, shall
256be dual licensed as above, without any additional terms or conditions.