Turn cfg_match into a builtin

Author: c410-f3r

compiler/rustc_ast/src/token.rs

@@ -229,35 +229,61 @@ fn ident_can_begin_type(name: Symbol, span: Span, is_raw: bool) -> bool {
 #[derive(PartialEq, Encodable, Decodable, Debug, HashStable_Generic)]
 pub enum TokenKind {
     /* Expression-operator symbols. */
+    /// `=`
     Eq,
+    /// `>`
     Lt,
+    /// `<=`
     Le,
+    /// `==`
     EqEq,
+    /// `!=`
     Ne,
+    /// `>`
     Ge,
+    /// `>=`
     Gt,
+    /// `&&`
     AndAnd,
+    /// `||`
     OrOr,
+    /// `!`
     Not,
+    /// `~`
     Tilde,
     BinOp(BinOpToken),
     BinOpEq(BinOpToken),
 
     /* Structural symbols */
+    /// `@`
     At,
+    /// `.`
     Dot,
+    /// `..`
     DotDot,
+    /// `...`
     DotDotDot,
+    /// `..=`
     DotDotEq,
+    /// `,`
     Comma,
+    /// `;`
     Semi,
+    /// `:`
     Colon,
+    /// `::`
     ModSep,
+    /// `->`
     RArrow,
+    /// `<-`
     LArrow,
+    /// `=>`
     FatArrow,
+    /// `#`
     Pound,
+    /// `$`
     Dollar,
+    /// `?`
     Question,
     /// Used by proc macros for representing lifetimes, not generated by lexer right now.
     SingleQuote,
@@ -296,6 +322,7 @@ pub enum TokenKind {
     /// similarly to symbols in string literal tokens.
     DocComment(CommentKind, ast::AttrStyle, Symbol),
 
+    /// End Of File
     Eof,
 }
 

compiler/rustc_builtin_macros/messages.ftl

@@ -69,6 +69,9 @@ builtin_macros_cfg_accessible_indeterminate = cannot determine whether the path
 builtin_macros_cfg_accessible_literal_path = `cfg_accessible` path cannot be a literal
 builtin_macros_cfg_accessible_multiple_paths = multiple `cfg_accessible` paths are specified
 builtin_macros_cfg_accessible_unspecified_path = `cfg_accessible` path is not specified
+builtin_macros_cfg_match_bad_arm = conditional arm must be declared with a trailing `=>`
+builtin_macros_cfg_match_bad_wildcard = the last arm is expected to be a wildcard
+builtin_macros_cfg_match_missing_comma = conditional arms with a single element must end with a comma
 builtin_macros_concat_bytes_array = cannot concatenate doubly nested array
     .note = byte strings are treated as arrays of bytes
     .help = try flattening the array

compiler/rustc_builtin_macros/src/cfg_match.rs

@@ -0,0 +1,264 @@
+use crate::errors;
+use rustc_ast::{
+    attr::mk_attr,
+    ptr::P,
+    token,
+    tokenstream::{DelimSpan, TokenStream, TokenTree},
+    AssocItem, AssocItemKind, AttrArgs, AttrStyle, Attribute, DelimArgs, Item, ItemKind, Path,
+    Stmt, StmtKind,
+};
+use rustc_errors::PResult;
+use rustc_expand::base::{DummyResult, ExtCtxt, MacResult};
+use rustc_parse::parser::{ForceCollect, Parser};
+use rustc_span::{
+    symbol::{kw, sym, Ident},
+    Span,
+};
+use smallvec::SmallVec;
+
+// ```rust
+// cfg_match! {
+//     cfg(unix) => { fn foo() -> i32 { 1 } },
+//     _ => { fn foo() -> i32 { 2 } },
+// }
+// ```
+//
+// The above `cfg_match!` is expanded to the following elements:
+//
+// ```rust
+// #[cfg(all(unix, not(any())))]
+// fn foo() -> i32 { 1 }
+//
+// #[cfg(all(not(any(unix))))]
+// fn foo() -> i32 { 2 }
+// ```
+pub fn expand_cfg_match(
+    cx: &mut ExtCtxt<'_>,
+    sp: Span,
+    tts: TokenStream,
+) -> Box<dyn MacResult + 'static> {
+    let (does_not_have_wildcard, mut items, mut items_offsets) = match parse(cx, tts) {
+        Err(mut err) => {
+            err.emit();
+            return DummyResult::any(sp);
+        }
+        Ok(assert) => assert,
+    };
+    let mut no = Vec::with_capacity(items_offsets.len());
+    iter(
+        (cx, &mut no),
+        &mut items,
+        &mut items_offsets,
+        |(local_cx, local_no), yes, items| {
+            let attr = mk_cfg(local_cx, local_no, sp, Some(yes));
+            items.iter_mut().for_each(|item| item.attrs.push(attr.clone()));
+            local_no.push(yes);
+        },
+        |(local_cx, local_no), yes, items| {
+            let attr = mk_cfg(local_cx, local_no, sp, does_not_have_wildcard.then_some(yes));
+            items.iter_mut().for_each(|item| item.attrs.push(attr.clone()));
+        },
+    );
+    Box::new(CfgMatchOutput(items))
+}
+
+fn iter<'io, A>(
+    mut aux: A,
+    items: &mut [P<Item>],
+    items_offsets: &'io [(TokenStream, usize)],
+    mut cb: impl FnMut(&mut A, &'io TokenStream, &mut [P<Item>]),
+    mut cb_last: impl FnMut(&mut A, &'io TokenStream, &mut [P<Item>]),
+) {
+    match items_offsets {
+        [] => return,
+        [first] => {
+            cb_last(&mut aux, &first.0, items.get_mut(..first.1).unwrap_or_default());
+        }
+        [first, rest @ .., last] => {
+            let mut offset = first.1;
+            let mut prev_offset = 0;
+            cb(&mut aux, &first.0, items.get_mut(prev_offset..offset).unwrap_or_default());
+            for elem in rest {
+                prev_offset = offset;
+                offset = elem.1;
+                cb(&mut aux, &elem.0, items.get_mut(prev_offset..offset).unwrap_or_default());
+            }
+            prev_offset = offset;
+            offset = last.1;
+            cb_last(&mut aux, &last.0, items.get_mut(prev_offset..offset).unwrap_or_default());
+        }
+    }
+}
+
+// #[cfg(all(** YES **, not(any(** NO **, ** NO **, ..))))]
+fn mk_cfg(
+    cx: &mut ExtCtxt<'_>,
+    no: &[&TokenStream],
+    sp: Span,
+    yes: Option<&TokenStream>,
+) -> Attribute {
+    let mut any_tokens = TokenStream::new(Vec::with_capacity(4));
+    if let &[first, ref rest @ ..] = no {
+        any_tokens.push_stream(first.clone());
+        for elem in rest.iter().copied() {
+            any_tokens.push_tree(TokenTree::token_alone(token::Comma, sp));
+            any_tokens.push_stream(elem.clone());
+        }
+    }
+
+    let mut not_tokens = TokenStream::new(Vec::with_capacity(2));
+    not_tokens.push_tree(TokenTree::token_alone(token::Ident(sym::any, false), sp));
+    not_tokens.push_tree(TokenTree::Delimited(
+        DelimSpan::from_single(sp),
+        token::Delimiter::Parenthesis,
+        any_tokens,
+    ));
+
+    let mut all_tokens = TokenStream::new(Vec::with_capacity(4));
+    if let Some(elem) = yes {
+        all_tokens.push_stream(elem.clone());
+        all_tokens.push_tree(TokenTree::token_alone(token::Comma, sp));
+    }
+    all_tokens.push_tree(TokenTree::token_alone(token::Ident(sym::not, false), sp));
+    all_tokens.push_tree(TokenTree::Delimited(
+        DelimSpan::from_single(sp),
+        token::Delimiter::Parenthesis,
+        not_tokens,
+    ));
+
+    let mut tokens = TokenStream::new(Vec::with_capacity(2));
+    tokens.push_tree(TokenTree::token_alone(token::Ident(sym::all, false), sp));
+    tokens.push_tree(TokenTree::Delimited(
+        DelimSpan::from_single(sp),
+        token::Delimiter::Parenthesis,
+        all_tokens,
+    ));
+
+    mk_attr(
+        &cx.sess.parse_sess.attr_id_generator,
+        AttrStyle::Outer,
+        Path::from_ident(Ident::new(sym::cfg, sp)),
+        AttrArgs::Delimited(DelimArgs {
+            dspan: DelimSpan::from_single(sp),
+            delim: token::Delimiter::Parenthesis,
+            tokens,
+        }),
+        sp,
+    )
+}
+
+fn parse<'a>(
+    cx: &mut ExtCtxt<'a>,
+    tts: TokenStream,
+) -> PResult<'a, (bool, Vec<P<Item>>, Vec<(TokenStream, usize)>)> {
+    let mut parser = cx.new_parser_from_tts(tts);
+    if parser.token == token::Eof {
+        return parser.unexpected();
+    }
+    let mut does_not_have_wildcard = true;
+    let mut items = Vec::with_capacity(4);
+    let mut items_offsets = Vec::with_capacity(4);
+    loop {
+        match parse_cfg_arm(&mut items, &mut parser)? {
+            None => break,
+            Some(attr) => {
+                items_offsets.push((attr, items.len()));
+            }
+        }
+    }
+    if parser.token != token::Eof {
+        parse_wildcard_arm(&mut items, &mut parser)?;
+        does_not_have_wildcard = false;
+        items_offsets.push((TokenStream::new(vec![]), items.len()));
+    }
+    if parser.token != token::Eof {
+        return parser.unexpected();
+    }
+    Ok((does_not_have_wildcard, items, items_offsets))
+}
+
+fn parse_arbitrary_arm_block<'a>(
+    items: &mut Vec<P<Item>>,
+    parser: &mut Parser<'a>,
+) -> PResult<'a, ()> {
+    if parser.eat(&token::OpenDelim(token::Delimiter::Brace)) {
+        loop {
+            let item = match parser.parse_item(ForceCollect::No) {
+                Ok(Some(elem)) => elem,
+                _ => break,
+            };
+            items.push(item);
+        }
+        parser.expect(&token::CloseDelim(token::Delimiter::Brace))?;
+    } else {
+        let Ok(Some(item)) = parser.parse_item(ForceCollect::No) else {
+            return parser.unexpected();
+        };
+        if !parser.eat(&token::Comma) {
+            return Err(parser
+                .sess
+                .create_err(errors::CfgMatchMissingComma { span: parser.token.span }));
+        }
+        items.push(item);
+    }
+    Ok(())
+}
+
+fn parse_cfg_arm<'a>(
+    items: &mut Vec<P<Item>>,
+    parser: &mut Parser<'a>,
+) -> PResult<'a, Option<TokenStream>> {
+    if !parser.eat_keyword(sym::cfg) {
+        return Ok(None);
+    }
+    let TokenTree::Delimited(_, delim, tokens) = parser.parse_token_tree() else {
+        return parser.unexpected();
+    };
+    if delim != token::Delimiter::Parenthesis || !parser.eat(&token::FatArrow) {
+        return Err(parser
+            .sess
+            .create_err(errors::CfgMatchBadArm { span: parser.token.span }));
+    }
+    parse_arbitrary_arm_block(items, parser)?;
+    Ok(Some(tokens))
+}
+
+fn parse_wildcard_arm<'a>(items: &mut Vec<P<Item>>, parser: &mut Parser<'a>) -> PResult<'a, ()> {
+    if !parser.eat_keyword(kw::Underscore) || !parser.eat(&token::FatArrow) {
+        return Err(parser
+            .sess
+            .create_err(errors::CfgMatchBadWildcard { span: parser.token.span }));
+    }
+    parse_arbitrary_arm_block(items, parser)?;
+    Ok(())
+}
+
+struct CfgMatchOutput(Vec<P<Item>>);
+
+impl MacResult for CfgMatchOutput {
+    fn make_impl_items(self: Box<Self>) -> Option<SmallVec<[P<AssocItem>; 1]>> {
+        let mut rslt = SmallVec::with_capacity(self.0.len());
+        rslt.extend(self.0.into_iter().filter_map(|item| {
+            let Item { attrs, id, span, vis, ident, kind, tokens } = item.into_inner();
+            let ItemKind::Fn(fun) = kind else {
+                return None;
+            };
+            Some(P(Item { attrs, id, ident, kind: AssocItemKind::Fn(fun), span, tokens, vis }))
+        }));
+        Some(rslt)
+    }
+
+    fn make_items(self: Box<Self>) -> Option<SmallVec<[P<Item>; 1]>> {
+        Some(<_>::from(self.0))
+    }
+
+    fn make_stmts(self: Box<Self>) -> Option<SmallVec<[Stmt; 1]>> {
+        let mut rslt = SmallVec::with_capacity(self.0.len());
+        rslt.extend(self.0.into_iter().map(|item| {
+            let id = item.id;
+            let span = item.span;
+            Stmt { id, kind: StmtKind::Item(item), span }
+        }));
+        Some(rslt)
+    }
+}

compiler/rustc_builtin_macros/src/errors.rs

@@ -831,3 +831,24 @@ pub(crate) struct ExpectedRegisterClassOrExplicitRegister {
     #[primary_span]
     pub(crate) span: Span,
 }
+
+#[derive(Diagnostic)]
+#[diag(builtin_macros_cfg_match_bad_arm)]
+pub(crate) struct CfgMatchBadArm {
+    #[primary_span]
+    pub(crate) span: Span,
+}
+
+#[derive(Diagnostic)]
+#[diag(builtin_macros_cfg_match_bad_wildcard)]
+pub(crate) struct CfgMatchBadWildcard {
+    #[primary_span]
+    pub(crate) span: Span,
+}
+
+#[derive(Diagnostic)]
+#[diag(builtin_macros_cfg_match_missing_comma)]
+pub(crate) struct CfgMatchMissingComma {
+    #[primary_span]
+    pub(crate) span: Span,
+}