Auto merge of #100996 - m-ou-se:format-args-2, r=estebank

Rewrite and refactor format_args!() builtin macro.

This is a near complete rewrite of `compiler/rustc_builtin_macros/src/format.rs`.

This gets rid of the massive unmaintanable [`Context` struct](https://github.com/rust-lang/rust/blob/76531befc4b0352247ada67bd225e8cf71ee5686/compiler/rustc_builtin_macros/src/format.rs#L176-L263), and splits the macro expansion into three parts:

1. First, `parse_args` will parse the `(literal, arg, arg, name=arg, name=arg)` syntax, but doesn't parse the template (the literal) itself.
2. Second, `make_format_args` will parse the template, the format options, resolve argument references, produce diagnostics, and turn the whole thing into a `FormatArgs` structure.
3. Finally, `expand_parsed_format_args` will turn that `FormatArgs` structure into the expression that the macro expands to.

In other words, the `format_args` builtin macro used to be a hard-to-maintain 'single pass compiler', which I've split into a three phase compiler with a parser/tokenizer (step 1), semantic analysis (step 2), and backend (step 3). (It's compilers all the way down. ^^)

This can serve as a great starting point for #99012, which will only need to change the implementation of 3, while leaving step 1 and 2 unchanged.

It also makes rust-lang/compiler-team#541 easier, which could then upgrade the new `FormatArgs` struct to an `ast` node and remove step 3, moving that step to later in the compilation process.

It also fixes a few diagnostics bugs.

This also [significantly reduces](https://gist.github.com/m-ou-se/b67b2d54172c4837a5ab1b26fa3e5284) the amount of generated code for cases with arguments in non-default order without formatting options, like `"{1} {0}"` or `"{a} {}"`, etc.

Author: bors

compiler/rustc_builtin_macros/src/format.rs

@@ -0,0 +1,240 @@
+use rustc_ast::ptr::P;
+use rustc_ast::Expr;
+use rustc_data_structures::fx::FxHashMap;
+use rustc_span::symbol::{Ident, Symbol};
+use rustc_span::Span;
+
+// Definitions:
+//
+// format_args!("hello {abc:.xyz$}!!", abc="world");
+// └──────────────────────────────────────────────┘
+//                     FormatArgs
+//
+// format_args!("hello {abc:.xyz$}!!", abc="world");
+//                                     └─────────┘
+//                                      argument
+//
+// format_args!("hello {abc:.xyz$}!!", abc="world");
+//              └───────────────────┘
+//                     template
+//
+// format_args!("hello {abc:.xyz$}!!", abc="world");
+//               └────┘└─────────┘└┘
+//                      pieces
+//
+// format_args!("hello {abc:.xyz$}!!", abc="world");
+//               └────┘           └┘
+//                   literal pieces
+//
+// format_args!("hello {abc:.xyz$}!!", abc="world");
+//                     └─────────┘
+//                     placeholder
+//
+// format_args!("hello {abc:.xyz$}!!", abc="world");
+//                      └─┘  └─┘
+//                      positions (could be names, numbers, empty, or `*`)
+
+/// (Parsed) format args.
+///
+/// Basically the "AST" for a complete `format_args!()`.
+///
+/// E.g., `format_args!("hello {name}");`.
+#[derive(Clone, Debug)]
+pub struct FormatArgs {
+    pub span: Span,
+    pub template: Vec<FormatArgsPiece>,
+    pub arguments: FormatArguments,
+}
+
+/// A piece of a format template string.
+///
+/// E.g. "hello" or "{name}".
+#[derive(Clone, Debug)]
+pub enum FormatArgsPiece {
+    Literal(Symbol),
+    Placeholder(FormatPlaceholder),
+}
+
+/// The arguments to format_args!().
+///
+/// E.g. `1, 2, name="ferris", n=3`,
+/// but also implicit captured arguments like `x` in `format_args!("{x}")`.
+#[derive(Clone, Debug)]
+pub struct FormatArguments {
+    arguments: Vec<FormatArgument>,
+    num_unnamed_args: usize,
+    num_explicit_args: usize,
+    names: FxHashMap<Symbol, usize>,
+}
+
+impl FormatArguments {
+    pub fn new() -> Self {
+        Self {
+            arguments: Vec::new(),
+            names: FxHashMap::default(),
+            num_unnamed_args: 0,
+            num_explicit_args: 0,
+        }
+    }
+
+    pub fn add(&mut self, arg: FormatArgument) -> usize {
+        let index = self.arguments.len();
+        if let Some(name) = arg.kind.ident() {
+            self.names.insert(name.name, index);
+        } else if self.names.is_empty() {
+            // Only count the unnamed args before the first named arg.
+            // (Any later ones are errors.)
+            self.num_unnamed_args += 1;
+        }
+        if !matches!(arg.kind, FormatArgumentKind::Captured(..)) {
+            // This is an explicit argument.
+            // Make sure that all arguments so far are explcit.
+            assert_eq!(
+                self.num_explicit_args,
+                self.arguments.len(),
+                "captured arguments must be added last"
+            );
+            self.num_explicit_args += 1;
+        }
+        self.arguments.push(arg);
+        index
+    }
+
+    pub fn by_name(&self, name: Symbol) -> Option<(usize, &FormatArgument)> {
+        let i = *self.names.get(&name)?;
+        Some((i, &self.arguments[i]))
+    }
+
+    pub fn by_index(&self, i: usize) -> Option<&FormatArgument> {
+        (i < self.num_explicit_args).then(|| &self.arguments[i])
+    }
+
+    pub fn unnamed_args(&self) -> &[FormatArgument] {
+        &self.arguments[..self.num_unnamed_args]
+    }
+
+    pub fn named_args(&self) -> &[FormatArgument] {
+        &self.arguments[self.num_unnamed_args..self.num_explicit_args]
+    }
+
+    pub fn explicit_args(&self) -> &[FormatArgument] {
+        &self.arguments[..self.num_explicit_args]
+    }
+
+    pub fn into_vec(self) -> Vec<FormatArgument> {
+        self.arguments
+    }
+}
+
+#[derive(Clone, Debug)]
+pub struct FormatArgument {
+    pub kind: FormatArgumentKind,
+    pub expr: P<Expr>,
+}
+
+#[derive(Clone, Debug)]
+pub enum FormatArgumentKind {
+    /// `format_args(…, arg)`
+    Normal,
+    /// `format_args(…, arg = 1)`
+    Named(Ident),
+    /// `format_args("… {arg} …")`
+    Captured(Ident),
+}
+
+impl FormatArgumentKind {
+    pub fn ident(&self) -> Option<Ident> {
+        match self {
+            &Self::Normal => None,
+            &Self::Named(id) => Some(id),
+            &Self::Captured(id) => Some(id),
+        }
+    }
+}
+
+#[derive(Clone, Debug, PartialEq, Eq)]
+pub struct FormatPlaceholder {
+    /// Index into [`FormatArgs::arguments`].
+    pub argument: FormatArgPosition,
+    /// The span inside the format string for the full `{…}` placeholder.
+    pub span: Option<Span>,
+    /// `{}`, `{:?}`, or `{:x}`, etc.
+    pub format_trait: FormatTrait,
+    /// `{}` or `{:.5}` or `{:-^20}`, etc.
+    pub format_options: FormatOptions,
+}
+
+#[derive(Clone, Debug, PartialEq, Eq)]
+pub struct FormatArgPosition {
+    /// Which argument this position refers to (Ok),
+    /// or would've referred to if it existed (Err).
+    pub index: Result<usize, usize>,
+    /// What kind of position this is. See [`FormatArgPositionKind`].
+    pub kind: FormatArgPositionKind,
+    /// The span of the name or number.
+    pub span: Option<Span>,
+}
+
+#[derive(Copy, Clone, Debug, PartialEq, Eq)]
+pub enum FormatArgPositionKind {
+    /// `{}` or `{:.*}`
+    Implicit,
+    /// `{1}` or `{:1$}` or `{:.1$}`
+    Number,
+    /// `{a}` or `{:a$}` or `{:.a$}`
+    Named,
+}
+
+#[derive(Copy, Clone, Debug, Hash, PartialEq, Eq)]
+pub enum FormatTrait {
+    /// `{}`
+    Display,
+    /// `{:?}`
+    Debug,
+    /// `{:e}`
+    LowerExp,
+    /// `{:E}`
+    UpperExp,
+    /// `{:o}`
+    Octal,
+    /// `{:p}`
+    Pointer,
+    /// `{:b}`
+    Binary,
+    /// `{:x}`
+    LowerHex,
+    /// `{:X}`
+    UpperHex,
+}
+
+#[derive(Clone, Debug, Default, PartialEq, Eq)]
+pub struct FormatOptions {
+    /// The width. E.g. `{:5}` or `{:width$}`.
+    pub width: Option<FormatCount>,
+    /// The precision. E.g. `{:.5}` or `{:.precision$}`.
+    pub precision: Option<FormatCount>,
+    /// The alignment. E.g. `{:>}` or `{:<}` or `{:^}`.
+    pub alignment: Option<FormatAlignment>,
+    /// The fill character. E.g. the `.` in `{:.>10}`.
+    pub fill: Option<char>,
+    /// The `+`, `-`, `0`, `#`, `x?` and `X?` flags.
+    pub flags: u32,
+}
+
+#[derive(Clone, Debug, PartialEq, Eq)]
+pub enum FormatAlignment {
+    /// `{:<}`
+    Left,
+    /// `{:>}`
+    Right,
+    /// `{:^}`
+    Center,
+}
+
+#[derive(Clone, Debug, PartialEq, Eq)]
+pub enum FormatCount {
+    /// `{:5}` or `{:.5}`
+    Literal(usize),
+    /// `{:.*}`, `{:.5$}`, or `{:a$}`, etc.
+    Argument(FormatArgPosition),
+}