rustc_proc_macro/bridge/
symbol.rs

1//! Client-side interner used for symbols.
2//!
3//! This is roughly based on the symbol interner from `rustc_span` and the
4//! DroplessArena from `rustc_arena`. It is unfortunately a complete
5//! copy/re-implementation rather than a dependency as it is difficult to depend
6//! on crates from within `proc_macro`, due to it being built at the same time
7//! as `std`.
8//!
9//! If at some point in the future it becomes easier to add dependencies to
10//! proc_macro, this module should probably be removed or simplified.
11
12use std::cell::RefCell;
13use std::num::NonZero;
14
15use super::*;
16
17/// Handle for a symbol string stored within the Interner.
18#[derive(Copy, Clone, PartialEq, Eq, Hash)]
19pub struct Symbol(NonZero<u32>);
20
21impl !Send for Symbol {}
22impl !Sync for Symbol {}
23
24impl Symbol {
25    /// Intern a new `Symbol`
26    pub(crate) fn new(string: &str) -> Self {
27        INTERNER.with_borrow_mut(|i| i.intern(string))
28    }
29
30    /// Creates a new `Symbol` for an identifier.
31    ///
32    /// Validates and normalizes before converting it to a symbol.
33    pub(crate) fn new_ident(string: &str, is_raw: bool) -> Self {
34        // Fast-path: check if this is a valid ASCII identifier
35        if Self::is_valid_ascii_ident(string.as_bytes()) || string == "$crate" {
36            if is_raw && !Self::can_be_raw(string) {
37                panic!("`{}` cannot be a raw identifier", string);
38            }
39            return Self::new(string);
40        }
41
42        // Slow-path: If the string is already ASCII we're done, otherwise ask
43        // our server to do this for us over RPC.
44        // We don't need to check for identifiers which can't be raw here,
45        // because all of them are ASCII.
46        if string.is_ascii() {
47            Err(())
48        } else {
49            client::Symbol::normalize_and_validate_ident(string)
50        }
51        .unwrap_or_else(|_| panic!("`{:?}` is not a valid identifier", string))
52    }
53
54    /// Run a callback with the symbol's string value.
55    pub(crate) fn with<R>(self, f: impl FnOnce(&str) -> R) -> R {
56        INTERNER.with_borrow(|i| f(i.get(self)))
57    }
58
59    /// Clear out the thread-local symbol interner, making all previously
60    /// created symbols invalid such that `with` will panic when called on them.
61    pub(crate) fn invalidate_all() {
62        INTERNER.with_borrow_mut(|i| i.clear());
63    }
64
65    /// Checks if the ident is a valid ASCII identifier.
66    ///
67    /// This is a short-circuit which is cheap to implement within the
68    /// proc-macro client to avoid RPC when creating simple idents, but may
69    /// return `false` for a valid identifier if it contains non-ASCII
70    /// characters.
71    fn is_valid_ascii_ident(bytes: &[u8]) -> bool {
72        matches!(bytes.first(), Some(b'_' | b'a'..=b'z' | b'A'..=b'Z'))
73            && bytes[1..]
74                .iter()
75                .all(|b| matches!(b, b'_' | b'a'..=b'z' | b'A'..=b'Z' | b'0'..=b'9'))
76    }
77
78    // Mimics the behavior of `Symbol::can_be_raw` from `rustc_span`
79    fn can_be_raw(string: &str) -> bool {
80        match string {
81            "_" | "super" | "self" | "Self" | "crate" | "$crate" => false,
82            _ => true,
83        }
84    }
85}
86
87impl fmt::Debug for Symbol {
88    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
89        self.with(|s| fmt::Debug::fmt(s, f))
90    }
91}
92
93impl fmt::Display for Symbol {
94    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
95        self.with(|s| fmt::Display::fmt(s, f))
96    }
97}
98
99impl<S> Encode<S> for Symbol {
100    fn encode(self, w: &mut Writer, s: &mut S) {
101        self.with(|sym| sym.encode(w, s))
102    }
103}
104
105impl<S: server::Server> Decode<'_, '_, server::HandleStore<server::MarkedTypes<S>>>
106    for Marked<S::Symbol, Symbol>
107{
108    fn decode(r: &mut Reader<'_>, s: &mut server::HandleStore<server::MarkedTypes<S>>) -> Self {
109        Mark::mark(S::intern_symbol(<&str>::decode(r, s)))
110    }
111}
112
113impl<S: server::Server> Encode<server::HandleStore<server::MarkedTypes<S>>>
114    for Marked<S::Symbol, Symbol>
115{
116    fn encode(self, w: &mut Writer, s: &mut server::HandleStore<server::MarkedTypes<S>>) {
117        S::with_symbol_string(&self.unmark(), |sym| sym.encode(w, s))
118    }
119}
120
121impl<S> Decode<'_, '_, S> for Symbol {
122    fn decode(r: &mut Reader<'_>, s: &mut S) -> Self {
123        Symbol::new(<&str>::decode(r, s))
124    }
125}
126
127thread_local! {
128    static INTERNER: RefCell<Interner> = RefCell::new(Interner {
129        arena: arena::Arena::new(),
130        names: fxhash::FxHashMap::default(),
131        strings: Vec::new(),
132        // Start with a base of 1 to make sure that `NonZero<u32>` works.
133        sym_base: NonZero::new(1).unwrap(),
134    });
135}
136
137/// Basic interner for a `Symbol`, inspired by the one in `rustc_span`.
138struct Interner {
139    arena: arena::Arena,
140    // SAFETY: These `'static` lifetimes are actually references to data owned
141    // by the Arena. This is safe, as we never return them as static references
142    // from `Interner`.
143    names: fxhash::FxHashMap<&'static str, Symbol>,
144    strings: Vec<&'static str>,
145    // The offset to apply to symbol names stored in the interner. This is used
146    // to ensure that symbol names are not re-used after the interner is
147    // cleared.
148    sym_base: NonZero<u32>,
149}
150
151impl Interner {
152    fn intern(&mut self, string: &str) -> Symbol {
153        if let Some(&name) = self.names.get(string) {
154            return name;
155        }
156
157        let name = Symbol(
158            self.sym_base
159                .checked_add(self.strings.len() as u32)
160                .expect("`proc_macro` symbol name overflow"),
161        );
162
163        let string: &str = self.arena.alloc_str(string);
164
165        // SAFETY: we can extend the arena allocation to `'static` because we
166        // only access these while the arena is still alive.
167        let string: &'static str = unsafe { &*(string as *const str) };
168        self.strings.push(string);
169        self.names.insert(string, name);
170        name
171    }
172
173    /// Reads a symbol's value from the store while it is held.
174    fn get(&self, symbol: Symbol) -> &str {
175        // NOTE: Subtract out the offset which was added to make the symbol
176        // nonzero and prevent symbol name re-use.
177        let name = symbol
178            .0
179            .get()
180            .checked_sub(self.sym_base.get())
181            .expect("use-after-free of `proc_macro` symbol");
182        self.strings[name as usize]
183    }
184
185    /// Clear all symbols from the store, invalidating them such that `get` will
186    /// panic if they are accessed in the future.
187    fn clear(&mut self) {
188        // NOTE: Be careful not to panic here, as we may be called on the client
189        // when a `catch_unwind` isn't installed.
190        self.sym_base = self.sym_base.saturating_add(self.strings.len() as u32);
191        self.names.clear();
192        self.strings.clear();
193
194        // SAFETY: This is cleared after the names and strings tables are
195        // cleared out, so no references into the arena should remain.
196        self.arena = arena::Arena::new();
197    }
198}
rustc_proc_macro/bridge/symbol.rs

rustc_proc_macro/bridge/
symbol.rs
