refactor

fix-ir/Cargo.toml

@@ -0,0 +1,17 @@
+[package]
+name = "fix-ir"
+version = "0.1.0"
+edition = "2024"
+
+[dependencies]
+bumpalo = { workspace = true }
+ghost-cell = { workspace = true }
+rnix = { workspace = true }
+rowan = { workspace = true }
+string-interner = { workspace = true }
+hashbrown = { workspace = true }
+num_enum = { workspace = true }
+
+fix-builtins = { path = "../fix-builtins" }
+fix-common = { path = "../fix-common" }
+fix-error = { path = "../fix-error" }

fix-ir/src/lib.rs

@@ -0,0 +1,301 @@
+use std::hash::Hash;
+use std::ops::Deref;
+
+use bumpalo::Bump;
+use bumpalo::boxed::Box;
+use bumpalo::collections::Vec;
+use fix_builtins::{BUILTINS, BuiltinId};
+use fix_common::StringId;
+use ghost_cell::{GhostCell, GhostToken};
+use num_enum::TryFromPrimitive as _;
+use rnix::{TextRange, ast};
+use string_interner::DefaultStringInterner;
+
+pub mod downgrade;
+
+pub type HashMap<'ir, K, V> = hashbrown::HashMap<K, V, hashbrown::DefaultHashBuilder, &'ir Bump>;
+
+#[repr(transparent)]
+#[derive(Clone, Copy)]
+pub struct IrRef<'id, 'ir>(&'ir GhostCell<'id, Ir<'ir, Self>>);
+
+impl<'id, 'ir> IrRef<'id, 'ir> {
+    pub fn new(ir: &'ir GhostCell<'id, Ir<'ir, Self>>) -> Self {
+        Self(ir)
+    }
+
+    pub fn alloc(bump: &'ir Bump, ir: Ir<'ir, Self>) -> Self {
+        Self(bump.alloc(GhostCell::new(ir)))
+    }
+
+    pub fn borrow<'a>(&'a self, token: &'a GhostToken<'id>) -> &'a Ir<'ir, Self> {
+        self.0.borrow(token)
+    }
+
+    /// Freeze a mutable IR reference into a read-only one, consuming the
+    /// `GhostToken` to prevent any further mutation.
+    ///
+    /// # Safety
+    /// The transmute is sound because:
+    /// - `GhostCell<'id, T>` is `#[repr(transparent)]` over `T`
+    /// - `IrRef<'id, 'ir>` is `#[repr(transparent)]` over
+    ///   `&'ir GhostCell<'id, Ir<'ir, Self>>`
+    /// - `RawIrRef<'ir>` is `#[repr(transparent)]` over `&'ir Ir<'ir, Self>`
+    /// - `Ir<'ir, Ref>` is `#[repr(C)]` and both ref types are pointer-sized
+    ///
+    /// Consuming the `GhostToken` guarantees no `borrow_mut` calls can occur
+    /// afterwards, so the shared `&Ir` references from `RawIrRef::Deref` can
+    /// never alias with mutable references.
+    pub fn freeze(self, _token: GhostToken<'id>) -> RawIrRef<'ir> {
+        unsafe { std::mem::transmute(self) }
+    }
+}
+
+#[repr(transparent)]
+#[derive(Clone, Copy, Debug)]
+pub struct RawIrRef<'ir>(pub &'ir Ir<'ir, Self>);
+
+impl<'ir> Deref for RawIrRef<'ir> {
+    type Target = Ir<'ir, RawIrRef<'ir>>;
+    fn deref(&self) -> &Self::Target {
+        self.0
+    }
+}
+
+#[repr(C)]
+#[derive(Debug)]
+pub enum Ir<'ir, Ref> {
+    Int(i64),
+    Float(f64),
+    Bool(bool),
+    Null,
+    Str(Box<'ir, String>),
+    AttrSet {
+        stcs: HashMap<'ir, StringId, (Ref, TextRange)>,
+        dyns: Vec<'ir, (Ref, Ref, TextRange)>,
+    },
+    List {
+        items: Vec<'ir, Ref>,
+    },
+    Path(Ref),
+    ConcatStrings {
+        parts: Vec<'ir, Ref>,
+        force_string: bool,
+    },
+
+    // OPs
+    UnOp {
+        rhs: Ref,
+        kind: UnOpKind,
+    },
+    BinOp {
+        lhs: Ref,
+        rhs: Ref,
+        kind: BinOpKind,
+    },
+    HasAttr {
+        lhs: Ref,
+        rhs: Vec<'ir, Attr<Ref>>,
+    },
+    Select {
+        expr: Ref,
+        attrpath: Vec<'ir, Attr<Ref>>,
+        default: Option<Ref>,
+        span: TextRange,
+    },
+
+    // Conditionals
+    If {
+        cond: Ref,
+        consq: Ref,
+        alter: Ref,
+    },
+    Assert {
+        assertion: Ref,
+        expr: Ref,
+        assertion_raw: String,
+        span: TextRange,
+    },
+
+    With {
+        namespace: Ref,
+        body: Ref,
+        thunks: Vec<'ir, (ThunkId, Ref)>,
+    },
+    WithLookup(StringId),
+
+    // Function related
+    Func {
+        body: Ref,
+        param: Option<Param<'ir>>,
+        thunks: Vec<'ir, (ThunkId, Ref)>,
+    },
+    Arg {
+        layer: usize,
+    },
+    Call {
+        func: Ref,
+        arg: Ref,
+        span: TextRange,
+    },
+
+    // Builtins
+    Builtins,
+    Builtin(BuiltinId),
+    BuiltinConst(StringId),
+
+    // Misc
+    TopLevel {
+        body: Ref,
+        thunks: Vec<'ir, (ThunkId, Ref)>,
+    },
+    Thunk(ThunkId),
+    CurPos(TextRange),
+    ReplBinding(StringId),
+    ScopedImportBinding(StringId),
+}
+
+#[repr(transparent)]
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, PartialOrd, Ord)]
+pub struct ThunkId(pub usize);
+
+#[repr(transparent)]
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, PartialOrd, Ord)]
+pub struct SpanId(pub u32);
+
+/// Represents a key in an attribute path.
+#[allow(unused)]
+#[derive(Debug)]
+pub enum Attr<Ref> {
+    /// A dynamic attribute key, which is an expression that must evaluate to a string.
+    /// Example: `attrs.${key}`
+    Dynamic(Ref, TextRange),
+    /// A static attribute key.
+    /// Example: `attrs.key`
+    Str(StringId, TextRange),
+}
+
+/// The kinds of binary operations supported in Nix.
+#[derive(Clone, Copy, Debug, Hash, PartialEq, Eq)]
+pub enum BinOpKind {
+    // Arithmetic
+    Add,
+    Sub,
+    Div,
+    Mul,
+
+    // Comparison
+    Eq,
+    Neq,
+    Lt,
+    Gt,
+    Leq,
+    Geq,
+
+    // Logical
+    And,
+    Or,
+    Impl,
+
+    // Set/String/Path operations
+    Con, // List concatenation (`++`)
+    Upd, // AttrSet update (`//`)
+
+    // Not standard, but part of rnix AST
+    PipeL,
+    PipeR,
+}
+
+impl From<ast::BinOpKind> for BinOpKind {
+    fn from(op: ast::BinOpKind) -> Self {
+        use BinOpKind::*;
+        use ast::BinOpKind as kind;
+        match op {
+            kind::Concat => Con,
+            kind::Update => Upd,
+            kind::Add => Add,
+            kind::Sub => Sub,
+            kind::Mul => Mul,
+            kind::Div => Div,
+            kind::And => And,
+            kind::Equal => Eq,
+            kind::Implication => Impl,
+            kind::Less => Lt,
+            kind::LessOrEq => Leq,
+            kind::More => Gt,
+            kind::MoreOrEq => Geq,
+            kind::NotEqual => Neq,
+            kind::Or => Or,
+            kind::PipeLeft => PipeL,
+            kind::PipeRight => PipeR,
+        }
+    }
+}
+
+/// The kinds of unary operations.
+#[derive(Clone, Copy, Debug, Hash, PartialEq, Eq)]
+pub enum UnOpKind {
+    Neg, // Negation (`-`)
+    Not, // Logical not (`!`)
+}
+
+impl From<ast::UnaryOpKind> for UnOpKind {
+    fn from(value: ast::UnaryOpKind) -> Self {
+        match value {
+            ast::UnaryOpKind::Invert => UnOpKind::Not,
+            ast::UnaryOpKind::Negate => UnOpKind::Neg,
+        }
+    }
+}
+
+/// Describes the parameters of a function.
+#[derive(Debug)]
+pub struct Param<'ir> {
+    pub required: Vec<'ir, (StringId, TextRange)>,
+    pub optional: Vec<'ir, (StringId, TextRange)>,
+    pub ellipsis: bool,
+}
+
+pub fn new_global_env(
+    strings: &mut DefaultStringInterner,
+) -> hashbrown::HashMap<StringId, Ir<'static, RawIrRef<'static>>> {
+    let mut global_env = hashbrown::HashMap::new();
+    let builtins_sym = StringId(strings.get_or_intern("builtins"));
+    global_env.insert(builtins_sym, Ir::Builtins);
+
+    for (idx, &(name, _)) in BUILTINS.iter().enumerate() {
+        let id = BuiltinId::try_from_primitive(idx as u8).expect("infallible");
+        let name = StringId(strings.get_or_intern(name));
+        global_env.insert(name, Ir::Builtin(id));
+    }
+
+    let consts = [
+        (
+            "__currentSystem",
+            Ir::BuiltinConst(StringId(strings.get_or_intern("currentSystem"))),
+        ),
+        ("__langVersion", Ir::Int(6)),
+        (
+            "__nixVersion",
+            Ir::BuiltinConst(StringId(strings.get_or_intern("nixVersion"))),
+        ),
+        (
+            "__storeDir",
+            Ir::BuiltinConst(StringId(strings.get_or_intern("storeDir"))),
+        ),
+        (
+            "__nixPath",
+            Ir::BuiltinConst(StringId(strings.get_or_intern("nixPath"))),
+        ),
+        ("null", Ir::Null),
+        ("true", Ir::Bool(true)),
+        ("false", Ir::Bool(false)),
+    ];
+
+    for (name, ir) in consts {
+        let name = StringId(strings.get_or_intern(name));
+        global_env.insert(name, ir);
+    }
+
+    global_env
+}