Merge pull request #4573 from RalfJung/rustup

Rustup

4 files changed, 45 insertions, 23 deletions

diff --git a/compiler/rustc_parse/src/lexer/mod.rs b/compiler/rustc_parse/src/lexer/mod.rs
index e3bd6a9a327..f5f081efc49 100644
--- a/compiler/rustc_parse/src/lexer/mod.rs
+++ b/compiler/rustc_parse/src/lexer/mod.rs
@@ -44,19 +44,44 @@ pub(crate) struct UnmatchedDelim {
     pub candidate_span: Option<Span>,
 }
 
+/// Which tokens should be stripped before lexing the tokens.
+pub(crate) enum StripTokens {
+    /// Strip both shebang and frontmatter.
+    ShebangAndFrontmatter,
+    /// Strip the shebang but not frontmatter.
+    ///
+    /// That means that char sequences looking like frontmatter are simply
+    /// interpreted as regular Rust lexemes.
+    Shebang,
+    /// Strip nothing.
+    ///
+    /// In other words, char sequences looking like a shebang or frontmatter
+    /// are simply interpreted as regular Rust lexemes.
+    Nothing,
+}
+
 pub(crate) fn lex_token_trees<'psess, 'src>(
     psess: &'psess ParseSess,
     mut src: &'src str,
     mut start_pos: BytePos,
     override_span: Option<Span>,
-    frontmatter_allowed: FrontmatterAllowed,
+    strip_tokens: StripTokens,
 ) -> Result<TokenStream, Vec<Diag<'psess>>> {
-    // Skip `#!`, if present.
-    if let Some(shebang_len) = rustc_lexer::strip_shebang(src) {
-        src = &src[shebang_len..];
-        start_pos = start_pos + BytePos::from_usize(shebang_len);
+    match strip_tokens {
+        StripTokens::Shebang | StripTokens::ShebangAndFrontmatter => {
+            if let Some(shebang_len) = rustc_lexer::strip_shebang(src) {
+                src = &src[shebang_len..];
+                start_pos = start_pos + BytePos::from_usize(shebang_len);
+            }
+        }
+        StripTokens::Nothing => {}
     }
 
+    let frontmatter_allowed = match strip_tokens {
+        StripTokens::ShebangAndFrontmatter => FrontmatterAllowed::Yes,
+        StripTokens::Shebang | StripTokens::Nothing => FrontmatterAllowed::No,
+    };
+
     let cursor = Cursor::new(src, frontmatter_allowed);
     let mut lexer = Lexer {
         psess,
diff --git a/compiler/rustc_parse/src/lib.rs b/compiler/rustc_parse/src/lib.rs
index b790966acfd..d8792d7af4c 100644
--- a/compiler/rustc_parse/src/lib.rs
+++ b/compiler/rustc_parse/src/lib.rs
@@ -21,7 +21,6 @@ use rustc_ast::tokenstream::{DelimSpan, TokenStream};
 use rustc_ast::{AttrItem, Attribute, MetaItemInner, token};
 use rustc_ast_pretty::pprust;
 use rustc_errors::{Diag, EmissionGuarantee, FatalError, PResult, pluralize};
-use rustc_lexer::FrontmatterAllowed;
 use rustc_session::parse::ParseSess;
 use rustc_span::source_map::SourceMap;
 use rustc_span::{FileName, SourceFile, Span};
@@ -34,6 +33,8 @@ pub mod parser;
 use parser::Parser;
 use rustc_ast::token::Delimiter;
 
+use crate::lexer::StripTokens;
+
 pub mod lexer;
 
 mod errors;
@@ -62,10 +63,10 @@ pub fn new_parser_from_source_str(
     source: String,
 ) -> Result<Parser<'_>, Vec<Diag<'_>>> {
     let source_file = psess.source_map().new_source_file(name, source);
-    new_parser_from_source_file(psess, source_file, FrontmatterAllowed::Yes)
+    new_parser_from_source_file(psess, source_file, StripTokens::ShebangAndFrontmatter)
 }
 
-/// Creates a new parser from a simple (no frontmatter) source string.
+/// Creates a new parser from a simple (no shebang, no frontmatter) source string.
 ///
 /// On failure, the errors must be consumed via `unwrap_or_emit_fatal`, `emit`, `cancel`,
 /// etc., otherwise a panic will occur when they are dropped.
@@ -75,7 +76,7 @@ pub fn new_parser_from_simple_source_str(
     source: String,
 ) -> Result<Parser<'_>, Vec<Diag<'_>>> {
     let source_file = psess.source_map().new_source_file(name, source);
-    new_parser_from_source_file(psess, source_file, FrontmatterAllowed::No)
+    new_parser_from_source_file(psess, source_file, StripTokens::Nothing)
 }
 
 /// Creates a new parser from a filename. On failure, the errors must be consumed via
@@ -109,7 +110,7 @@ pub fn new_parser_from_file<'a>(
         }
         err.emit();
     });
-    new_parser_from_source_file(psess, source_file, FrontmatterAllowed::Yes)
+    new_parser_from_source_file(psess, source_file, StripTokens::ShebangAndFrontmatter)
 }
 
 pub fn utf8_error<E: EmissionGuarantee>(
@@ -160,10 +161,10 @@ pub fn utf8_error<E: EmissionGuarantee>(
 fn new_parser_from_source_file(
     psess: &ParseSess,
     source_file: Arc<SourceFile>,
-    frontmatter_allowed: FrontmatterAllowed,
+    strip_tokens: StripTokens,
 ) -> Result<Parser<'_>, Vec<Diag<'_>>> {
     let end_pos = source_file.end_position();
-    let stream = source_file_to_stream(psess, source_file, None, frontmatter_allowed)?;
+    let stream = source_file_to_stream(psess, source_file, None, strip_tokens)?;
     let mut parser = Parser::new(psess, stream, None);
     if parser.token == token::Eof {
         parser.token.span = Span::new(end_pos, end_pos, parser.token.span.ctxt(), None);
@@ -179,8 +180,8 @@ pub fn source_str_to_stream(
 ) -> Result<TokenStream, Vec<Diag<'_>>> {
     let source_file = psess.source_map().new_source_file(name, source);
     // used mainly for `proc_macro` and the likes, not for our parsing purposes, so don't parse
-    // frontmatters as frontmatters.
-    source_file_to_stream(psess, source_file, override_span, FrontmatterAllowed::No)
+    // frontmatters as frontmatters, but for compatibility reason still strip the shebang
+    source_file_to_stream(psess, source_file, override_span, StripTokens::Shebang)
 }
 
 /// Given a source file, produces a sequence of token trees. Returns any buffered errors from
@@ -189,7 +190,7 @@ fn source_file_to_stream<'psess>(
     psess: &'psess ParseSess,
     source_file: Arc<SourceFile>,
     override_span: Option<Span>,
-    frontmatter_allowed: FrontmatterAllowed,
+    strip_tokens: StripTokens,
 ) -> Result<TokenStream, Vec<Diag<'psess>>> {
     let src = source_file.src.as_ref().unwrap_or_else(|| {
         psess.dcx().bug(format!(
@@ -198,13 +199,7 @@ fn source_file_to_stream<'psess>(
         ));
     });
 
-    lexer::lex_token_trees(
-        psess,
-        src.as_str(),
-        source_file.start_pos,
-        override_span,
-        frontmatter_allowed,
-    )
+    lexer::lex_token_trees(psess, src.as_str(), source_file.start_pos, override_span, strip_tokens)
 }
 
 /// Runs the given subparser `f` on the tokens of the given `attr`'s item.
diff --git a/compiler/rustc_parse/src/parser/expr.rs b/compiler/rustc_parse/src/parser/expr.rs
index 9b09cbba7af..4c02547357e 100644
--- a/compiler/rustc_parse/src/parser/expr.rs
+++ b/compiler/rustc_parse/src/parser/expr.rs
@@ -2742,7 +2742,7 @@ impl<'a> Parser<'a> {
     /// The specified `edition` in `let_chains_policy` should be that of the whole `if` construct,
     /// i.e. the same span we use to later decide whether the drop behaviour should be that of
     /// edition `..=2021` or that of `2024..`.
-    // Public because it is used in rustfmt forks such as https://github.com/tucant/rustfmt/blob/30c83df9e1db10007bdd16dafce8a86b404329b2/src/parse/macros/html.rs#L57 for custom if expressions.
+    // Public to use it for custom `if` expressions in rustfmt forks like https://github.com/tucant/rustfmt
     pub fn parse_expr_cond(
         &mut self,
         let_chains_policy: LetChainsPolicy,
diff --git a/compiler/rustc_parse/src/parser/mod.rs b/compiler/rustc_parse/src/parser/mod.rs
index 15598f32429..3bbca622975 100644
--- a/compiler/rustc_parse/src/parser/mod.rs
+++ b/compiler/rustc_parse/src/parser/mod.rs
@@ -22,6 +22,8 @@ use std::{fmt, mem, slice};
 use attr_wrapper::{AttrWrapper, UsePreAttrPos};
 pub use diagnostics::AttemptLocalParseRecovery;
 pub(crate) use expr::ForbiddenLetReason;
+// Public to use it for custom `if` expressions in rustfmt forks like https://github.com/tucant/rustfmt
+pub use expr::LetChainsPolicy;
 pub(crate) use item::{FnContext, FnParseMode};
 pub use pat::{CommaRecoveryMode, RecoverColon, RecoverComma};
 pub use path::PathStyle;