feat: the C AST used by the C codegen backend (#6)

Author: Wybxc

bootstrap/src/test.rs

@@ -22,6 +22,12 @@ impl Run for TestCommand {
             cprintln!("<r,s>Test failed</r,s>: {}", info);
         }));
 
+        cprintln!("<b>[TEST]</b> running cargo test");
+        let mut command = std::process::Command::new("cargo");
+        command.args(["test", "--manifest-path", "crates/Cargo.toml"]);
+        log::debug!("running {:?}", command);
+        assert!(command.status().unwrap().success(), "failed to run {:?}", command);
+
         let testcases = self.collect_testcases(manifest);
         cprintln!("<b>[TEST]</b> found {} testcases", testcases.len());
 

crates/Cargo.lock

@@ -1,6 +1,6 @@
 [workspace]
 resolver = "2"
-members = ["rustc_codegen_c"]
+members = ["rustc_codegen_c", "rustc_codegen_c_ast"]
 
 [workspace.package]
 version = "0.1.0"

crates/Cargo.toml

@@ -0,0 +1,10 @@
+[package]
+name = "rustc_codegen_c_ast"
+edition = "2021"
+version.workspace = true
+
+[dependencies]
+
+# This package uses rustc crates.
+[package.metadata.rust-analyzer]
+rustc_private = true

crates/rustc_codegen_c_ast/Cargo.toml

@@ -0,0 +1,15 @@
+//! This module defines the memory arena for C AST nodes.
+
+use crate::decl::CDeclKind;
+use crate::expr::CExprKind;
+use crate::func::CFuncKind;
+use crate::stmt::CStmtKind;
+use crate::ty::CTyKind;
+
+rustc_arena::declare_arena!([
+    [] decl: CDeclKind<'tcx>,
+    [] expr: CExprKind<'tcx>,
+    [] func: CFuncKind<'tcx>,
+    [] stmt: CStmtKind<'tcx>,
+    [] ty: CTyKind<'tcx>,
+]);

crates/rustc_codegen_c_ast/src/arena.rs

@@ -0,0 +1,50 @@
+//! This module defines AST nodes for C declarations.
+
+use crate::expr::{CExpr, CValue};
+use crate::pretty::{Print, PrinterCtx, INDENT};
+use crate::ty::{print_declarator, CTy};
+use crate::ModuleCtx;
+
+/// C declarations.
+pub type CDecl<'mx> = &'mx CDeclKind<'mx>;
+
+/// C declaration kinds.
+#[derive(Debug, Clone)]
+pub enum CDeclKind<'mx> {
+    /// Variable declaration consisting of a name, type, and optional initializer.
+    ///
+    /// Example:
+    /// - `int foo;` // `ty val`
+    /// - `int foo = bar` `ty val = expr`
+    Var { name: CValue<'mx>, ty: CTy<'mx>, init: Option<CExpr<'mx>> },
+}
+
+impl<'mx> ModuleCtx<'mx> {
+    /// Create a new declaration.
+    pub fn decl(self, decl: CDeclKind<'mx>) -> CDecl<'mx> {
+        self.arena().alloc(decl)
+    }
+
+    /// Create a new variable declaration.
+    pub fn var(self, name: CValue<'mx>, ty: CTy<'mx>, init: Option<CExpr<'mx>>) -> CDecl<'mx> {
+        self.decl(CDeclKind::Var { name, ty, init })
+    }
+}
+
+impl Print for CDecl<'_> {
+    fn print_to(&self, ctx: &mut PrinterCtx) {
+        match self {
+            CDeclKind::Var { name, ty, init } => {
+                ctx.ibox(INDENT, |ctx| {
+                    print_declarator(*ty, Some(*name), ctx);
+                    if let Some(init) = init {
+                        ctx.word(" =");
+                        ctx.softbreak();
+                        init.print_to(ctx);
+                    }
+                    ctx.word(";");
+                });
+            }
+        }
+    }
+}

crates/rustc_codegen_c_ast/src/decl.rs

@@ -0,0 +1,133 @@
+//! This module defines the AST nodes for C expressions.
+
+use crate::pretty::{Print, PrinterCtx, INDENT};
+use crate::ty::{print_declarator, CTy};
+use crate::ModuleCtx;
+
+/// Represents the values of C variables, parameters, and scalars.
+///
+/// There are two variants to distinguish between constants and variables,
+/// as is done in LLVM IR. We follow the `rustc_codegen_ssa` convention for this representation.
+#[derive(Clone, Copy, Debug, Hash, PartialEq, Eq)]
+pub enum CValue<'mx> {
+    /// A constant scalar
+    Scalar(i128),
+    /// A local variable indexed by a number, in the form `_0`, `_1`, etc.
+    Local(usize),
+    /// A function name
+    Func(&'mx str),
+}
+
+/// C expressions.
+pub type CExpr<'mx> = &'mx CExprKind<'mx>;
+
+/// C expressions.
+#[derive(Debug, Clone)]
+pub enum CExprKind<'mx> {
+    /// A "raw" C expression, simply a string of C code, which is printed as-is.
+    Raw(&'static str),
+    /// A value, such as a constant, variable, or function name.
+    Value(CValue<'mx>),
+    /// A binary operation expression, e.g. `lhs + rhs`.
+    Binary { lhs: CExpr<'mx>, rhs: CExpr<'mx>, op: &'static str },
+    /// A type cast expression, e.g. `(int) x`.
+    Cast { ty: CTy<'mx>, expr: CExpr<'mx> },
+    /// A function call expression, e.g. `foo(x, y)`.
+    Call { callee: CExpr<'mx>, args: Vec<CExpr<'mx>> },
+    /// A member access expression, e.g. `foo.bar` or `foo->bar`.
+    Member {
+        expr: CExpr<'mx>,
+        /// Whether to use the `->` operator instead of `.`.
+        arrow: bool,
+        field: &'mx str,
+    },
+}
+
+impl<'mx> ModuleCtx<'mx> {
+    /// Create a new expression.
+    pub fn expr(&self, expr: CExprKind<'mx>) -> CExpr<'mx> {
+        self.arena().alloc(expr)
+    }
+
+    /// Create a new raw expression.
+    pub fn raw(&self, raw: &'static str) -> CExpr<'mx> {
+        self.expr(CExprKind::Raw(raw))
+    }
+
+    /// Create a new value expression.
+    pub fn value(&self, value: CValue<'mx>) -> CExpr<'mx> {
+        self.expr(CExprKind::Value(value))
+    }
+
+    /// Create a new binary expression.
+    pub fn binary(&self, lhs: CExpr<'mx>, rhs: CExpr<'mx>, op: &'static str) -> CExpr<'mx> {
+        self.expr(CExprKind::Binary { lhs, rhs, op })
+    }
+
+    /// Create a new cast expression.
+    pub fn cast(&self, ty: CTy<'mx>, expr: CExpr<'mx>) -> CExpr<'mx> {
+        self.expr(CExprKind::Cast { ty, expr })
+    }
+
+    /// Create a new function call expression.
+    pub fn call(&self, callee: CExpr<'mx>, args: Vec<CExpr<'mx>>) -> CExpr<'mx> {
+        self.expr(CExprKind::Call { callee, args })
+    }
+
+    /// Create a new member access expression.
+    pub fn member(&self, expr: CExpr<'mx>, field: &'mx str) -> CExpr<'mx> {
+        self.expr(CExprKind::Member { expr, field, arrow: false })
+    }
+}
+
+impl Print for CValue<'_> {
+    fn print_to(&self, ctx: &mut PrinterCtx) {
+        match self {
+            CValue::Scalar(i) => ctx.word(i.to_string()),
+            CValue::Local(i) => ctx.word(format!("_{}", i)),
+            CValue::Func(name) => ctx.word(name.to_string()),
+        }
+    }
+}
+
+impl Print for CExpr<'_> {
+    fn print_to(&self, ctx: &mut PrinterCtx) {
+        match self {
+            CExprKind::Raw(raw) => ctx.word(*raw),
+            CExprKind::Value(value) => value.print_to(ctx),
+            CExprKind::Binary { lhs, rhs, op } => ctx.ibox_delim(INDENT, ("(", ")"), 0, |ctx| {
+                ctx.ibox(-INDENT, |ctx| lhs.print_to(ctx));
+
+                ctx.softbreak();
+                ctx.word(*op);
+                ctx.nbsp();
+
+                rhs.print_to(ctx);
+            }),
+            CExprKind::Cast { ty, expr } => ctx.ibox(INDENT, |ctx| {
+                ctx.word("(");
+                print_declarator(*ty, None, ctx);
+                ctx.word(")");
+
+                ctx.nbsp();
+                expr.print_to(ctx);
+            }),
+            CExprKind::Call { callee, args } => ctx.ibox(INDENT, |ctx| {
+                callee.print_to(ctx);
+                ctx.cbox_delim(INDENT, ("(", ")"), 0, |ctx| {
+                    ctx.seperated(",", args, |ctx, arg| arg.print_to(ctx));
+                });
+            }),
+            CExprKind::Member { expr, arrow, field } => ctx.cbox(INDENT, |ctx| {
+                expr.print_to(ctx);
+                ctx.zerobreak();
+                if *arrow {
+                    ctx.word("->");
+                } else {
+                    ctx.word(".");
+                }
+                ctx.word(field.to_string());
+            }),
+        }
+    }
+}

crates/rustc_codegen_c_ast/src/expr.rs

@@ -0,0 +1,91 @@
+//! This module defines AST nodes for C functions.
+
+use std::cell::{Cell, RefCell};
+
+use rustc_data_structures::intern::Interned;
+
+use crate::expr::CValue;
+use crate::pretty::{Print, PrinterCtx};
+use crate::stmt::{print_compound, CStmt};
+use crate::ty::{print_declarator, CTy};
+use crate::ModuleCtx;
+
+/// C functions definition.
+pub type CFunc<'mx> = Interned<'mx, CFuncKind<'mx>>;
+
+/// C function definition.
+#[derive(Debug, Clone)]
+pub struct CFuncKind<'mx> {
+    /// Function name.
+    pub name: &'mx str,
+    /// Return type.
+    pub ty: CTy<'mx>,
+    /// Function parameters.
+    pub params: Vec<(CTy<'mx>, CValue<'mx>)>,
+    /// Function body.
+    pub body: RefCell<Vec<CStmt<'mx>>>,
+    /// A counter for local variables, for generating unique names.
+    local_var_counter: Cell<usize>,
+}
+
+impl<'mx> CFuncKind<'mx> {
+    /// Make a new function definition.
+    pub fn new(name: &'mx str, ty: CTy<'mx>, params: impl IntoIterator<Item = CTy<'mx>>) -> Self {
+        let params = params
+            .into_iter()
+            .enumerate()
+            .map(|(i, ty)| (ty, CValue::Local(i)))
+            .collect::<Vec<_>>();
+        let local_var_counter = Cell::new(params.len());
+
+        Self { name, ty, params, body: RefCell::new(Vec::new()), local_var_counter }
+    }
+
+    /// Push a statement to the end of the function body.
+    pub fn push_stmt(&self, stmt: CStmt<'mx>) {
+        self.body.borrow_mut().push(stmt);
+    }
+
+    /// Get a new unique local variable.
+    pub fn next_local_var(&self) -> CValue {
+        let val = CValue::Local(self.local_var_counter.get());
+        self.local_var_counter.set(self.local_var_counter.get() + 1);
+        val
+    }
+}
+
+impl<'mx> ModuleCtx<'mx> {
+    /// Create a new function definition.
+    pub fn func(&self, func: CFuncKind<'mx>) -> &'mx CFuncKind<'mx> {
+        self.arena().alloc(func)
+    }
+}
+
+impl Print for CFunc<'_> {
+    fn print_to(&self, ctx: &mut PrinterCtx) {
+        ctx.ibox(0, |ctx| {
+            print_signature(*self, ctx);
+            ctx.softbreak(); // I don't know how to avoid a newline here
+            print_compound(&self.0.body.borrow(), ctx);
+        })
+    }
+}
+
+pub(crate) fn print_func_decl(func: CFunc, ctx: &mut PrinterCtx) {
+    print_signature(func, ctx);
+    ctx.word(";");
+}
+
+fn print_signature(func: CFunc, ctx: &mut PrinterCtx) {
+    ctx.ibox(0, |ctx| {
+        print_declarator(func.0.ty, Some(CValue::Func(func.0.name)), ctx);
+
+        ctx.valign_delim(("(", ")"), |ctx| {
+            ctx.seperated(",", &func.0.params, |ctx, (ty, name)| {
+                ctx.ibox(0, |ctx| {
+                    print_declarator(*ty, Some(*name), ctx);
+                })
+            })
+        });
+    });
+}