Auto merge of rust-lang#14084 - Veykril:float-parse, r=Veykril

fix: Fix parsing of nested tuple field accesses in a cursed way

This is absolutely terrible but seems to work. Macro fragment parsing comes next.

Author: bors

crates/hir-def/src/item_tree.rs

@@ -111,7 +111,8 @@ impl ItemTree {
             Some(node) => node,
             None => return Default::default(),
         };
-        if never!(syntax.kind() == SyntaxKind::ERROR) {
+        if never!(syntax.kind() == SyntaxKind::ERROR, "{:?} from {:?} {}", file_id, syntax, syntax)
+        {
             // FIXME: not 100% sure why these crop up, but return an empty tree to avoid a panic
             return Default::default();
         }
@@ -133,7 +134,7 @@ impl ItemTree {
                     ctx.lower_macro_stmts(stmts)
                 },
                 _ => {
-                    panic!("cannot create item tree from {syntax:?} {syntax}");
+                    panic!("cannot create item tree for file {file_id:?} from {syntax:?} {syntax}");
                 },
             }
         };

crates/hir-def/src/macro_expansion_tests/mbe.rs

@@ -97,6 +97,41 @@ fn#19 main#20(#21)#21 {#22
 "##]],
     );
 }
+#[test]
+fn float_field_acces_macro_input() {
+    check(
+        r#"
+macro_rules! foo {
+    ($expr:expr) => {
+        fn foo() {
+            $expr;
+        }
+    };
+}
+foo!(x .0.1);
+foo!(x .2. 3);
+foo!(x .4 .5);
+"#,
+        expect![[r#"
+macro_rules! foo {
+    ($expr:expr) => {
+        fn foo() {
+            $expr;
+        }
+    };
+}
+fn foo() {
+    (x.0.1);
+}
+fn foo() {
+    (x.2.3);
+}
+fn foo() {
+    (x.4.5);
+}
+"#]],
+    );
+}
 
 #[test]
 fn mbe_smoke_test() {

crates/hir-def/src/macro_expansion_tests/proc_macros.rs

@@ -104,7 +104,7 @@ macro_rules! id {
         $($t)*
     };
 }
-id /*+errors*/! {
+id! {
     #[proc_macros::identity]
     impl Foo for WrapBj {
         async fn foo(&self) {
@@ -113,18 +113,17 @@ id /*+errors*/! {
     }
 }
 "#,
-        expect![[r##"
+        expect![[r#"
 macro_rules! id {
     ($($t:tt)*) => {
         $($t)*
     };
 }
-/* parse error: expected SEMICOLON */
 #[proc_macros::identity] impl Foo for WrapBj {
     async fn foo(&self ) {
         self .0.id().await ;
     }
 }
-"##]],
+"#]],
     );
 }

crates/mbe/src/syntax_bridge.rs

@@ -95,6 +95,9 @@ pub fn token_tree_to_syntax_node(
             parser::Step::Token { kind, n_input_tokens: n_raw_tokens } => {
                 tree_sink.token(kind, n_raw_tokens)
             }
+            parser::Step::FloatSplit { ends_in_dot: has_pseudo_dot } => {
+                tree_sink.float_split(has_pseudo_dot)
+            }
             parser::Step::Enter { kind } => tree_sink.start_node(kind),
             parser::Step::Exit => tree_sink.finish_node(),
             parser::Step::Error { msg } => tree_sink.error(msg.to_string()),
@@ -796,6 +799,43 @@ fn delim_to_str(d: tt::DelimiterKind, closing: bool) -> Option<&'static str> {
 }
 
 impl<'a> TtTreeSink<'a> {
+    /// Parses a float literal as if it was a one to two name ref nodes with a dot inbetween.
+    /// This occurs when a float literal is used as a field access.
+    fn float_split(&mut self, has_pseudo_dot: bool) {
+        let (text, _span) = match self.cursor.token_tree() {
+            Some(tt::buffer::TokenTreeRef::Leaf(tt::Leaf::Literal(lit), _)) => {
+                (lit.text.as_str(), lit.span)
+            }
+            _ => unreachable!(),
+        };
+        match text.split_once('.') {
+            Some((left, right)) => {
+                assert!(!left.is_empty());
+                self.inner.start_node(SyntaxKind::NAME_REF);
+                self.inner.token(SyntaxKind::INT_NUMBER, left);
+                self.inner.finish_node();
+
+                // here we move the exit up, the original exit has been deleted in process
+                self.inner.finish_node();
+
+                self.inner.token(SyntaxKind::DOT, ".");
+
+                if has_pseudo_dot {
+                    assert!(right.is_empty(), "{left}.{right}");
+                } else {
+                    self.inner.start_node(SyntaxKind::NAME_REF);
+                    self.inner.token(SyntaxKind::INT_NUMBER, right);
+                    self.inner.finish_node();
+
+                    // the parser creates an unbalanced start node, we are required to close it here
+                    self.inner.finish_node();
+                }
+            }
+            None => unreachable!(),
+        }
+        self.cursor = self.cursor.bump();
+    }
+
     fn token(&mut self, kind: SyntaxKind, mut n_tokens: u8) {
         if kind == LIFETIME_IDENT {
             n_tokens = 2;

crates/mbe/src/to_parser_input.rs

@@ -45,6 +45,13 @@ pub(crate) fn to_parser_input(buffer: &TokenBuffer<'_>) -> parser::Input {
                             .unwrap_or_else(|| panic!("Fail to convert given literal {:#?}", &lit));
 
                         res.push(kind);
+
+                        if kind == FLOAT_NUMBER && !inner_text.ends_with('.') {
+                            // Tag the token as joint if it is float with a fractional part
+                            // we use this jointness to inform the parser about what token split
+                            // event to emit when we encounter a float literal in a field access
+                            res.was_joint();
+                        }
                     }
                     tt::Leaf::Ident(ident) => match ident.text.as_ref() {
                         "_" => res.push(T![_]),

crates/mbe/src/tt_iter.rs

@@ -150,6 +150,11 @@ impl<'a> TtIter<'a> {
                         cursor = cursor.bump_subtree();
                     }
                 }
+                parser::Step::FloatSplit { .. } => {
+                    // FIXME: We need to split the tree properly here, but mutating the token trees
+                    // in the buffer is somewhat tricky to pull off.
+                    cursor = cursor.bump_subtree();
+                }
                 parser::Step::Enter { .. } | parser::Step::Exit => (),
                 parser::Step::Error { .. } => error = true,
             }
@@ -166,19 +171,18 @@ impl<'a> TtIter<'a> {
 
         if cursor.is_root() {
             while curr != cursor {
-                if let Some(token) = curr.token_tree() {
-                    res.push(token);
-                }
+                let Some(token) = curr.token_tree() else { break };
+                res.push(token.cloned());
                 curr = curr.bump();
             }
         }
+
         self.inner = self.inner.as_slice()[res.len()..].iter();
         let res = match res.len() {
-            1 => Some(res[0].cloned()),
-            0 => None,
+            0 | 1 => res.pop(),
             _ => Some(tt::TokenTree::Subtree(tt::Subtree {
                 delimiter: tt::Delimiter::unspecified(),
-                token_trees: res.into_iter().map(|it| it.cloned()).collect(),
+                token_trees: res,
             })),
         };
         ExpandResult { value: res, err }

crates/parser/src/event.rs

@@ -74,7 +74,13 @@ pub(crate) enum Event {
         kind: SyntaxKind,
         n_raw_tokens: u8,
     },
-
+    /// When we parse `foo.0.0` or `foo. 0. 0` the lexer will hand us a float literal
+    /// instead of an integer literal followed by a dot as the lexer has no contextual knowledge.
+    /// This event instructs whatever consumes the events to split the float literal into
+    /// the corresponding parts.
+    FloatSplitHack {
+        ends_in_dot: bool,
+    },
     Error {
         msg: String,
     },
@@ -125,6 +131,11 @@ pub(super) fn process(mut events: Vec<Event>) -> Output {
             Event::Token { kind, n_raw_tokens } => {
                 res.token(kind, n_raw_tokens);
             }
+            Event::FloatSplitHack { ends_in_dot } => {
+                res.float_split_hack(ends_in_dot);
+                let ev = mem::replace(&mut events[i + 1], Event::tombstone());
+                assert!(matches!(ev, Event::Finish), "{ev:?}");
+            }
             Event::Error { msg } => res.error(msg),
         }
     }

crates/parser/src/grammar/expressions.rs

@@ -379,7 +379,7 @@ fn postfix_expr(
             // }
             T!['('] if allow_calls => call_expr(p, lhs),
             T!['['] if allow_calls => index_expr(p, lhs),
-            T![.] => match postfix_dot_expr(p, lhs) {
+            T![.] => match postfix_dot_expr::<false>(p, lhs) {
                 Ok(it) => it,
                 Err(it) => {
                     lhs = it;
@@ -393,35 +393,44 @@ fn postfix_expr(
         block_like = BlockLike::NotBlock;
     }
     return (lhs, block_like);
+}
 
-    fn postfix_dot_expr(
-        p: &mut Parser<'_>,
-        lhs: CompletedMarker,
-    ) -> Result<CompletedMarker, CompletedMarker> {
+fn postfix_dot_expr<const FLOAT_RECOVERY: bool>(
+    p: &mut Parser<'_>,
+    lhs: CompletedMarker,
+) -> Result<CompletedMarker, CompletedMarker> {
+    if !FLOAT_RECOVERY {
         assert!(p.at(T![.]));
-        if p.nth(1) == IDENT && (p.nth(2) == T!['('] || p.nth_at(2, T![::])) {
-            return Ok(method_call_expr(p, lhs));
-        }
+    }
+    let nth1 = if FLOAT_RECOVERY { 0 } else { 1 };
+    let nth2 = if FLOAT_RECOVERY { 1 } else { 2 };
 
-        // test await_expr
-        // fn foo() {
-        //     x.await;
-        //     x.0.await;
-        //     x.0().await?.hello();
-        // }
-        if p.nth(1) == T![await] {
-            let m = lhs.precede(p);
-            p.bump(T![.]);
-            p.bump(T![await]);
-            return Ok(m.complete(p, AWAIT_EXPR));
-        }
+    if p.nth(nth1) == IDENT && (p.nth(nth2) == T!['('] || p.nth_at(nth2, T![::])) {
+        return Ok(method_call_expr::<FLOAT_RECOVERY>(p, lhs));
+    }
 
-        if p.at(T![..=]) || p.at(T![..]) {
-            return Err(lhs);
+    // test await_expr
+    // fn foo() {
+    //     x.await;
+    //     x.0.await;
+    //     x.0().await?.hello();
+    //     x.0.0.await;
+    //     x.0. await;
+    // }
+    if p.nth(nth1) == T![await] {
+        let m = lhs.precede(p);
+        if !FLOAT_RECOVERY {
+            p.bump(T![.]);
         }
+        p.bump(T![await]);
+        return Ok(m.complete(p, AWAIT_EXPR));
+    }
 
-        Ok(field_expr(p, lhs))
+    if p.at(T![..=]) || p.at(T![..]) {
+        return Err(lhs);
     }
+
+    field_expr::<FLOAT_RECOVERY>(p, lhs)
 }
 
 // test call_expr
@@ -455,11 +464,22 @@ fn index_expr(p: &mut Parser<'_>, lhs: CompletedMarker) -> CompletedMarker {
 // fn foo() {
 //     x.foo();
 //     y.bar::<T>(1, 2,);
+//     x.0.0.call();
+//     x.0. call();
 // }
-fn method_call_expr(p: &mut Parser<'_>, lhs: CompletedMarker) -> CompletedMarker {
-    assert!(p.at(T![.]) && p.nth(1) == IDENT && (p.nth(2) == T!['('] || p.nth_at(2, T![::])));
+fn method_call_expr<const FLOAT_RECOVERY: bool>(
+    p: &mut Parser<'_>,
+    lhs: CompletedMarker,
+) -> CompletedMarker {
+    if FLOAT_RECOVERY {
+        assert!(p.nth(0) == IDENT && (p.nth(1) == T!['('] || p.nth_at(1, T![::])));
+    } else {
+        assert!(p.at(T![.]) && p.nth(1) == IDENT && (p.nth(2) == T!['('] || p.nth_at(2, T![::])));
+    }
     let m = lhs.precede(p);
-    p.bump_any();
+    if !FLOAT_RECOVERY {
+        p.bump(T![.]);
+    }
     name_ref(p);
     generic_args::opt_generic_arg_list(p, true);
     if p.at(T!['(']) {
@@ -472,21 +492,35 @@ fn method_call_expr(p: &mut Parser<'_>, lhs: CompletedMarker) -> CompletedMarker
 // fn foo() {
 //     x.foo;
 //     x.0.bar;
+//     x.0.1;
+//     x.0. bar;
 //     x.0();
 // }
-fn field_expr(p: &mut Parser<'_>, lhs: CompletedMarker) -> CompletedMarker {
-    assert!(p.at(T![.]));
+fn field_expr<const FLOAT_RECOVERY: bool>(
+    p: &mut Parser<'_>,
+    lhs: CompletedMarker,
+) -> Result<CompletedMarker, CompletedMarker> {
+    if !FLOAT_RECOVERY {
+        assert!(p.at(T![.]));
+    }
     let m = lhs.precede(p);
-    p.bump(T![.]);
+    if !FLOAT_RECOVERY {
+        p.bump(T![.]);
+    }
     if p.at(IDENT) || p.at(INT_NUMBER) {
         name_ref_or_index(p);
     } else if p.at(FLOAT_NUMBER) {
-        // FIXME: How to recover and instead parse INT + T![.]?
-        p.bump_any();
+        return match p.split_float(m) {
+            (true, m) => {
+                let lhs = m.complete(p, FIELD_EXPR);
+                postfix_dot_expr::<true>(p, lhs)
+            }
+            (false, m) => Ok(m.complete(p, FIELD_EXPR)),
+        };
     } else {
         p.error("expected field name or number");
     }
-    m.complete(p, FIELD_EXPR)
+    Ok(m.complete(p, FIELD_EXPR))
 }
 
 // test try_expr

crates/parser/src/lib.rs

@@ -102,10 +102,14 @@ impl TopEntryPoint {
                 match step {
                     Step::Enter { .. } => depth += 1,
                     Step::Exit => depth -= 1,
+                    Step::FloatSplit { ends_in_dot: has_pseudo_dot } => {
+                        depth -= 1 + !has_pseudo_dot as usize
+                    }
                     Step::Token { .. } | Step::Error { .. } => (),
                 }
             }
             assert!(!first, "no tree at all");
+            assert_eq!(depth, 0, "unbalanced tree");
         }
 
         res