Re: [PATCH 2/4] rust: macros: refactor generics parsing of `#[pin_data]` into its own function

[Gary Guo <gary@garyguo.net>]

On Mon, 24 Apr 2023 08:11:38 +0000
Benno Lossin <benno.lossin@proton.me> wrote:

> Other macros might also want to parse generics. Additionally this makes
> the code easier to read, as the next commit will introduce more code in
> `#[pin_data]`. Also add more comments to explain how parsing generics
> work.
> 
> Signed-off-by: Benno Lossin <benno.lossin@proton.me>

Reviewed-by: Gary Guo <gary@garyguo.net>

> ---
>  rust/macros/helpers.rs  | 86 ++++++++++++++++++++++++++++++++++++++++-
>  rust/macros/pin_data.rs | 70 +++++----------------------------
>  2 files changed, 94 insertions(+), 62 deletions(-)
> 
> diff --git a/rust/macros/helpers.rs b/rust/macros/helpers.rs
> index b2bdd4d8c958..afb0f2e3a36a 100644
> --- a/rust/macros/helpers.rs
> +++ b/rust/macros/helpers.rs
> @@ -1,6 +1,6 @@
>  // SPDX-License-Identifier: GPL-2.0
> 
> -use proc_macro::{token_stream, Group, TokenTree};
> +use proc_macro::{token_stream, Group, Punct, Spacing, TokenStream, TokenTree};
> 
>  pub(crate) fn try_ident(it: &mut token_stream::IntoIter) -> Option<String> {
>      if let Some(TokenTree::Ident(ident)) = it.next() {
> @@ -69,3 +69,87 @@ pub(crate) fn expect_end(it: &mut token_stream::IntoIter) {
>          panic!("Expected end");
>      }
>  }
> +
> +pub(crate) struct Generics {
> +    pub(crate) impl_generics: Vec<TokenTree>,
> +    pub(crate) ty_generics: Vec<TokenTree>,
> +}
> +
> +/// Parses the given `TokenStream` into `Generics` and the rest.
> +///
> +/// The generics are not present in the rest, but a where clause might remain.
> +pub(crate) fn parse_generics(input: TokenStream) -> (Generics, Vec<TokenTree>) {
> +    // `impl_generics`, the declared generics with their bounds.
> +    let mut impl_generics = vec![];
> +    // Only the names of the generics, without any bounds.
> +    let mut ty_generics = vec![];
> +    // Tokens not related to the generics e.g. the `where` token and definition.
> +    let mut rest = vec![];
> +    // The current level of `<`.
> +    let mut nesting = 0;
> +    let mut toks = input.into_iter();
> +    // If we are at the beginning of a generic parameter.
> +    let mut at_start = true;
> +    for tt in &mut toks {
> +        match tt.clone() {
> +            TokenTree::Punct(p) if p.as_char() == '<' => {
> +                if nesting >= 1 {
> +                    // This is inside of the generics and part of some bound.
> +                    impl_generics.push(tt);
> +                }
> +                nesting += 1;
> +            }
> +            TokenTree::Punct(p) if p.as_char() == '>' => {
> +                // This is a parsing error, so we just end it here.
> +                if nesting == 0 {
> +                    break;
> +                } else {
> +                    nesting -= 1;
> +                    if nesting >= 1 {
> +                        // We are still inside of the generics and part of some bound.
> +                        impl_generics.push(tt);
> +                    }
> +                    if nesting == 0 {
> +                        break;
> +                    }
> +                }
> +            }
> +            tt => {
> +                if nesting == 1 {
> +                    // Here depending on the token, it might be a generic variable name.
> +                    match &tt {
> +                        // Ignore const.
> +                        TokenTree::Ident(i) if i.to_string() == "const" => {}
> +                        TokenTree::Ident(_) if at_start => {
> +                            ty_generics.push(tt.clone());
> +                            // We also already push the `,` token, this makes it easier to append
> +                            // generics.
> +                            ty_generics.push(TokenTree::Punct(Punct::new(',', Spacing::Alone)));
> +                            at_start = false;
> +                        }
> +                        TokenTree::Punct(p) if p.as_char() == ',' => at_start = true,
> +                        // Lifetimes begin with `'`.
> +                        TokenTree::Punct(p) if p.as_char() == '\'' && at_start => {
> +                            ty_generics.push(tt.clone());
> +                        }
> +                        _ => {}
> +                    }
> +                }
> +                if nesting >= 1 {
> +                    impl_generics.push(tt);
> +                } else if nesting == 0 {
> +                    // If we haven't entered the generics yet, we still want to keep these tokens.
> +                    rest.push(tt);
> +                }
> +            }
> +        }
> +    }
> +    rest.extend(toks);
> +    (
> +        Generics {
> +            impl_generics,
> +            ty_generics,
> +        },
> +        rest,
> +    )
> +}
> diff --git a/rust/macros/pin_data.rs b/rust/macros/pin_data.rs
> index 954149d77181..c593b05d9e8c 100644
> --- a/rust/macros/pin_data.rs
> +++ b/rust/macros/pin_data.rs
> @@ -1,71 +1,19 @@
>  // SPDX-License-Identifier: Apache-2.0 OR MIT
> 
> -use proc_macro::{Punct, Spacing, TokenStream, TokenTree};
> +use crate::helpers::{parse_generics, Generics};
> +use proc_macro::TokenStream;
> 
>  pub(crate) fn pin_data(args: TokenStream, input: TokenStream) -> TokenStream {
>      // This proc-macro only does some pre-parsing and then delegates the actual parsing to
>      // `kernel::__pin_data!`.
> -    //
> -    // In here we only collect the generics, since parsing them in declarative macros is very
> -    // elaborate. We also do not need to analyse their structure, we only need to collect them.
> 
> -    // `impl_generics`, the declared generics with their bounds.
> -    let mut impl_generics = vec![];
> -    // Only the names of the generics, without any bounds.
> -    let mut ty_generics = vec![];
> -    // Tokens not related to the generics e.g. the `impl` token.
> -    let mut rest = vec![];
> -    // The current level of `<`.
> -    let mut nesting = 0;
> -    let mut toks = input.into_iter();
> -    // If we are at the beginning of a generic parameter.
> -    let mut at_start = true;
> -    for tt in &mut toks {
> -        match tt.clone() {
> -            TokenTree::Punct(p) if p.as_char() == '<' => {
> -                if nesting >= 1 {
> -                    impl_generics.push(tt);
> -                }
> -                nesting += 1;
> -            }
> -            TokenTree::Punct(p) if p.as_char() == '>' => {
> -                if nesting == 0 {
> -                    break;
> -                } else {
> -                    nesting -= 1;
> -                    if nesting >= 1 {
> -                        impl_generics.push(tt);
> -                    }
> -                    if nesting == 0 {
> -                        break;
> -                    }
> -                }
> -            }
> -            tt => {
> -                if nesting == 1 {
> -                    match &tt {
> -                        TokenTree::Ident(i) if i.to_string() == "const" => {}
> -                        TokenTree::Ident(_) if at_start => {
> -                            ty_generics.push(tt.clone());
> -                            ty_generics.push(TokenTree::Punct(Punct::new(',', Spacing::Alone)));
> -                            at_start = false;
> -                        }
> -                        TokenTree::Punct(p) if p.as_char() == ',' => at_start = true,
> -                        TokenTree::Punct(p) if p.as_char() == '\'' && at_start => {
> -                            ty_generics.push(tt.clone());
> -                        }
> -                        _ => {}
> -                    }
> -                }
> -                if nesting >= 1 {
> -                    impl_generics.push(tt);
> -                } else if nesting == 0 {
> -                    rest.push(tt);
> -                }
> -            }
> -        }
> -    }
> -    rest.extend(toks);
> +    let (
> +        Generics {
> +            impl_generics,
> +            ty_generics,
> +        },
> +        mut rest,
> +    ) = parse_generics(input);
>      // This should be the body of the struct `{...}`.
>      let last = rest.pop();
>      quote!(::kernel::__pin_data! {
> --
> 2.40.0
> 
>