Add support for raw-idents in cfgs

Author: Urgau

crates/cargo-platform/src/cfg.rs

@@ -16,16 +16,28 @@ pub enum CfgExpr {
 #[derive(Eq, PartialEq, Hash, Ord, PartialOrd, Clone, Debug)]
 pub enum Cfg {
     /// A named cfg value, like `unix`.
-    Name(String),
+    Name(Ident),
     /// A key/value cfg pair, like `target_os = "linux"`.
-    KeyPair(String, String),
+    KeyPair(Ident, String),
+}
+
+/// A identifier
+#[derive(Hash, Ord, PartialOrd, Clone, Debug)]
+pub struct Ident {
+    /// The identifier
+    pub name: String,
+    /// Is this a raw ident: `r#async`
+    ///
+    /// It's mainly used for display and doesn't
+    /// take part in the `PartialEq` as `foo` == `r#foo`.
+    pub raw: bool,
 }
 
 #[derive(PartialEq)]
 enum Token<'a> {
     LeftParen,
     RightParen,
-    Ident(&'a str),
+    Ident(bool, &'a str),
     Comma,
     Equals,
     String(&'a str),
@@ -49,6 +61,41 @@ struct Parser<'a> {
     t: Tokenizer<'a>,
 }
 
+impl Ident {
+    pub fn as_str(&self) -> &str {
+        &self.name
+    }
+}
+
+impl Eq for Ident {}
+
+impl PartialEq<str> for Ident {
+    fn eq(&self, other: &str) -> bool {
+        self.name == other
+    }
+}
+
+impl PartialEq<&str> for Ident {
+    fn eq(&self, other: &&str) -> bool {
+        self.name == *other
+    }
+}
+
+impl PartialEq<Ident> for Ident {
+    fn eq(&self, other: &Ident) -> bool {
+        self.name == other.name
+    }
+}
+
+impl fmt::Display for Ident {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        if self.raw {
+            f.write_str("r#")?;
+        }
+        f.write_str(&*self.name)
+    }
+}
+
 impl FromStr for Cfg {
     type Err = ParseError;
 
@@ -152,7 +199,8 @@ impl<'a> Parser<'a> {
 
     fn expr(&mut self) -> Result<CfgExpr, ParseError> {
         match self.peek() {
-            Some(Ok(Token::Ident(op @ "all"))) | Some(Ok(Token::Ident(op @ "any"))) => {
+            Some(Ok(Token::Ident(false, op @ "all")))
+            | Some(Ok(Token::Ident(false, op @ "any"))) => {
                 self.t.next();
                 let mut e = Vec::new();
                 self.eat(&Token::LeftParen)?;
@@ -169,7 +217,7 @@ impl<'a> Parser<'a> {
                     Ok(CfgExpr::Any(e))
                 }
             }
-            Some(Ok(Token::Ident("not"))) => {
+            Some(Ok(Token::Ident(false, "not"))) => {
                 self.t.next();
                 self.eat(&Token::LeftParen)?;
                 let e = self.expr()?;
@@ -187,7 +235,7 @@ impl<'a> Parser<'a> {
 
     fn cfg(&mut self) -> Result<Cfg, ParseError> {
         match self.t.next() {
-            Some(Ok(Token::Ident(name))) => {
+            Some(Ok(Token::Ident(raw, name))) => {
                 let e = if self.r#try(&Token::Equals) {
                     let val = match self.t.next() {
                         Some(Ok(Token::String(s))) => s,
@@ -205,9 +253,18 @@ impl<'a> Parser<'a> {
                             return Err(ParseError::new(self.t.orig, IncompleteExpr("a string")))
                         }
                     };
-                    Cfg::KeyPair(name.to_string(), val.to_string())
+                    Cfg::KeyPair(
+                        Ident {
+                            name: name.to_string(),
+                            raw,
+                        },
+                        val.to_string(),
+                    )
                 } else {
-                    Cfg::Name(name.to_string())
+                    Cfg::Name(Ident {
+                        name: name.to_string(),
+                        raw,
+                    })
                 };
                 Ok(e)
             }
@@ -287,14 +344,44 @@ impl<'a> Iterator for Tokenizer<'a> {
                     return Some(Err(ParseError::new(self.orig, UnterminatedString)));
                 }
                 Some((start, ch)) if is_ident_start(ch) => {
+                    let (start, raw) = if ch == 'r' {
+                        if let Some(&(_pos, '#')) = self.s.peek() {
+                            // starts with `r#` is a raw ident
+                            self.s.next();
+                            if let Some((start, ch)) = self.s.next() {
+                                if is_ident_start(ch) {
+                                    (start, true)
+                                } else {
+                                    // not a starting ident character
+                                    return Some(Err(ParseError::new(
+                                        self.orig,
+                                        UnexpectedChar(ch),
+                                    )));
+                                }
+                            } else {
+                                // not followed by a ident, error out
+                                return Some(Err(ParseError::new(
+                                    self.orig,
+                                    IncompleteExpr("identifier"),
+                                )));
+                            }
+                        } else {
+                            // starts with `r` but not does continue with `#`
+                            // cannot be a raw ident
+                            (start, false)
+                        }
+                    } else {
+                        // do not start with `r`, cannot be a raw ident
+                        (start, false)
+                    };
                     while let Some(&(end, ch)) = self.s.peek() {
                         if !is_ident_rest(ch) {
-                            return Some(Ok(Token::Ident(&self.orig[start..end])));
+                            return Some(Ok(Token::Ident(raw, &self.orig[start..end])));
                         } else {
                             self.s.next();
                         }
                     }
-                    return Some(Ok(Token::Ident(&self.orig[start..])));
+                    return Some(Ok(Token::Ident(raw, &self.orig[start..])));
                 }
                 Some((_, ch)) => {
                     return Some(Err(ParseError::new(self.orig, UnexpectedChar(ch))));

crates/cargo-platform/src/lib.rs

@@ -18,7 +18,7 @@ mod cfg;
 mod error;
 
 use cfg::KEYWORDS;
-pub use cfg::{Cfg, CfgExpr};
+pub use cfg::{Cfg, CfgExpr, Ident};
 pub use error::{ParseError, ParseErrorKind};
 
 /// Platform definition.

crates/cargo-platform/tests/test_cfg.rs

@@ -1,13 +1,37 @@
-use cargo_platform::{Cfg, CfgExpr, Platform};
+use cargo_platform::{Cfg, CfgExpr, Ident, Platform};
 use std::fmt;
 use std::str::FromStr;
 
 macro_rules! c {
     ($a:ident) => {
-        Cfg::Name(stringify!($a).to_string())
+        Cfg::Name(Ident {
+            name: stringify!($a).to_string(),
+            raw: false,
+        })
+    };
+    (r # $a:ident) => {
+        Cfg::Name(Ident {
+            name: stringify!($a).to_string(),
+            raw: true,
+        })
     };
     ($a:ident = $e:expr) => {
-        Cfg::KeyPair(stringify!($a).to_string(), $e.to_string())
+        Cfg::KeyPair(
+            Ident {
+                name: stringify!($a).to_string(),
+                raw: false,
+            },
+            $e.to_string(),
+        )
+    };
+    (r # $a:ident = $e:expr) => {
+        Cfg::KeyPair(
+            Ident {
+                name: stringify!($a).to_string(),
+                raw: true,
+            },
+            $e.to_string(),
+        )
     };
 }
 
@@ -56,10 +80,13 @@ fn cfg_syntax() {
     good("_bar", c!(_bar));
     good(" foo", c!(foo));
     good(" foo  ", c!(foo));
+    good("r#foo", c!(r # foo));
     good(" foo  = \"bar\"", c!(foo = "bar"));
     good("foo=\"\"", c!(foo = ""));
+    good("r#foo=\"\"", c!(r # foo = ""));
     good(" foo=\"3\"      ", c!(foo = "3"));
     good("foo = \"3 e\"", c!(foo = "3 e"));
+    good(" r#foo = \"3 e\"", c!(r # foo = "3 e"));
 }
 
 #[test]
@@ -78,6 +105,10 @@ fn cfg_syntax_bad() {
         "foo, bar",
         "unexpected content `, bar` found after cfg expression",
     );
+    bad::<Cfg>("r# foo", "unexpected character");
+    bad::<Cfg>("r #foo", "unexpected content");
+    bad::<Cfg>("r#\"foo\"", "unexpected character");
+    bad::<Cfg>("foo = r#\"\"", "unexpected character");
 }
 
 #[test]
@@ -126,6 +157,9 @@ fn cfg_matches() {
     assert!(e!(not(foo)).matches(&[]));
     assert!(e!(any((not(foo)), (all(foo, bar)))).matches(&[c!(bar)]));
     assert!(e!(any((not(foo)), (all(foo, bar)))).matches(&[c!(foo), c!(bar)]));
+    assert!(e!(foo).matches(&[c!(r # foo)]));
+    assert!(e!(r # foo).matches(&[c!(foo)]));
+    assert!(e!(r # foo).matches(&[c!(r # foo)]));
 
     assert!(!e!(foo).matches(&[]));
     assert!(!e!(foo).matches(&[c!(bar)]));

src/cargo/core/compiler/custom_build.rs

@@ -353,7 +353,9 @@ fn build_work(build_runner: &mut BuildRunner<'_, '_>, unit: &Unit) -> CargoResul
             // That is because Cargo queries rustc without any profile settings.
             continue;
         }
-        let k = format!("CARGO_CFG_{}", super::envify(&k));
+        // FIXME: We should handle raw-idents somehow instead of predenting they
+        // don't exist here
+        let k = format!("CARGO_CFG_{}", super::envify(k.as_str()));
         cmd.env(&k, v.join(","));
     }
 

tests/testsuite/cfg.rs

@@ -543,12 +543,15 @@ fn cfg_raw_idents() {
         .build();
 
     p.cargo("check")
-        .with_status(101)
         .with_stderr_data(str![[r#"
-[ERROR] failed to parse manifest at `[ROOT]/foo/Cargo.toml`
-
-Caused by:
-  failed to parse `any(r#true, r#all, r#target_os = "<>")` as a cfg expression: unexpected character `#` in cfg, expected parens, a comma, an identifier, or a string
+[WARNING] [[ROOT]/foo/Cargo.toml] future-incompatibility: the meaning of `cfg(r#true)` will change in the future
+ | Cargo is erroneously allowing `cfg(true)` and `cfg(false)`, but both forms are interpreted as false unless manually overridden with `--cfg`.
+ | In the future these will be built-in defines that will have the corresponding true/false value.
+ | It is recommended to avoid using these configs until they are properly supported.
+ | See <https://github.com/rust-lang/rust/issues/131204> for more information.
+[LOCKING] 1 package to latest compatible version
+[CHECKING] foo v0.1.0 ([ROOT]/foo)
+[FINISHED] `dev` profile [unoptimized + debuginfo] target(s) in [ELAPSED]s
 
 "#]])
         .run();
@@ -577,7 +580,7 @@ fn cfg_raw_idents_empty() {
 [ERROR] failed to parse manifest at `[ROOT]/foo/Cargo.toml`
 
 Caused by:
-  failed to parse `r#)` as a cfg expression: unexpected content `#)` found after cfg expression
+  failed to parse `r#)` as a cfg expression: unexpected character `)` in cfg, expected parens, a comma, an identifier, or a string
 
 "#]])
         .run();
@@ -606,7 +609,7 @@ fn cfg_raw_idents_not_really() {
 [ERROR] failed to parse manifest at `[ROOT]/foo/Cargo.toml`
 
 Caused by:
-  failed to parse `r#11)` as a cfg expression: unexpected content `#11)` found after cfg expression
+  failed to parse `r#11)` as a cfg expression: unexpected character `1` in cfg, expected parens, a comma, an identifier, or a string
 
 "#]])
         .run();