prebindgen-flat 0.5.0

//! The elements: one variant per structure the source language allows.
//!
//! Every node — an item, a parameter, a field, a variant, a type — carries one
//! [`Origin`], so generated Rust names the source by re-emitting what the source
//! wrote, nothing re-parses a whole item to find a part of it, and no level has
//! to copy a piece of provenance down from the level above.
//!
//! **Structure only.** Everything that turns an element back into Rust tokens
//! lives in [`spell`](super::spell), so the shape of an element says nothing
//! about the language it came from.

use prebindgen::SourceLocation;

use super::{origin::Origin, ty::TypeRef};

/// One member of the flat API.
///
/// Three modelled kinds — a function, a type, a constant — plus
/// [`Element::Guard`] for an anonymous const and [`Element::Unsupported`] for
/// anything the language cannot express. No *marked*
/// item passes through verbatim: a `#[prebindgen]` crate marks the items that
/// cross the boundary, and the supporting code around them is the consumer
/// crate's job — the proc-macro enforces that already, refusing to mark a `use`,
/// `mod`, `impl` or `macro_rules!` at all.
#[derive(Clone, Debug)]
pub enum Element {
    Function(Function),
    /// A type declaration: a struct, either enum shape, or an opaque handle.
    Type(Type),
    Constant(Constant),
    /// An anonymous const — infrastructure re-emitted verbatim. Carries no API
    /// surface: nothing can name it, declare it, or cross it.
    Guard(Guard),
    /// An item the language cannot express — a parameter type outside the
    /// grammar, a `self` receiver, a reference to a type the flat API never
    /// declares, or a whole item kind it does not model such as a `union`.
    ///
    /// Indexed under its name so nothing else can claim it, with the diagnosis
    /// riding along. Parsing carries it; building a
    /// `Registry` from a model holding one fails, reporting
    /// every offender at once. See the [module docs](super) on where acceptance
    /// is enforced.
    Unsupported(Unsupported),
}

impl Element {
    /// The item's name, which is also its address: `#[prebindgen]` names live
    /// in one flat namespace across every ingested source crate.
    ///
    /// `None` when the item has no address — a [`Guard`], or an item kind with
    /// no identifier at all.
    pub fn name(&self) -> Option<&syn::Ident> {
        match self {
            Element::Function(f) => Some(&f.name),
            Element::Type(t) => Some(t.name()),
            Element::Constant(c) => Some(&c.name),
            Element::Guard(_) => None,
            Element::Unsupported(u) => u.name.as_ref(),
        }
    }

    /// Where the item was captured, including the crate that marked it.
    ///
    /// The same location every component of this item carries — they share one
    /// [`Origin::location`], because one captured record is one item.
    pub fn location(&self) -> &SourceLocation {
        match self {
            Element::Function(f) => &f.origin.location,
            Element::Type(t) => t.location(),
            Element::Constant(c) => &c.origin.location,
            Element::Guard(g) => &g.origin.location,
            Element::Unsupported(u) => &u.origin.location,
        }
    }

    /// The whole item as `syn` — **the escape**, at the item level. See
    /// [`Origin::as_syn`](super::Origin::as_syn).
    ///
    /// It builds a `syn::Item` rather than borrowing one, because each variant
    /// keeps the item kind it was parsed as. That makes it the natural route for
    /// an emitter re-stating a whole item, and the ledger's **item** bucket is
    /// where those land.
    pub(crate) fn as_syn(&self) -> syn::Item {
        match self {
            Element::Function(f) => syn::Item::Fn(f.origin.as_syn().clone()),
            Element::Type(t) => t.as_syn(),
            Element::Constant(c) => syn::Item::Const(c.origin.as_syn().clone()),
            Element::Guard(g) => syn::Item::Const(g.origin.as_syn().clone()),
            Element::Unsupported(u) => u.origin.as_syn().clone(),
        }
    }
}

/// A type the flat API declares.
///
/// Four shapes, and the classification is what a destination language acts on: a
/// product of fields, a sum, a named set of integers, or a handle whose contents
/// do not cross.
#[derive(Clone, Debug)]
pub enum Type {
    Struct(Struct),
    /// An enum whose alternatives carry payloads — a sum type.
    Variant(Variant),
    /// An enum whose every alternative is fieldless — a named set of integers.
    Enum(Enum),
    Extern(Extern),
}

impl Type {
    pub fn name(&self) -> &syn::Ident {
        match self {
            Type::Struct(s) => &s.name,
            Type::Variant(v) => &v.name,
            Type::Enum(e) => &e.name,
            Type::Extern(e) => &e.name,
        }
    }

    pub fn location(&self) -> &SourceLocation {
        self.location_rc()
    }

    /// The shared location itself, for building a sibling node's [`Origin`].
    pub(super) fn location_rc(&self) -> &std::rc::Rc<SourceLocation> {
        match self {
            Type::Struct(s) => &s.origin.location,
            Type::Variant(v) => &v.origin.location,
            Type::Enum(e) => &e.origin.location,
            Type::Extern(e) => &e.origin.location,
        }
    }

    /// The whole item as `syn` — **the escape**. See [`Element::as_syn`].
    pub(crate) fn as_syn(&self) -> syn::Item {
        match self {
            Type::Struct(s) => syn::Item::Struct(s.origin.as_syn().clone()),
            Type::Variant(v) => syn::Item::Enum(v.origin.as_syn().clone()),
            Type::Enum(e) => syn::Item::Enum(e.origin.as_syn().clone()),
            Type::Extern(e) => e.origin.as_syn().clone(),
        }
    }
}

/// A type the flat API **names** but whose contents it does not model.
///
/// Two spellings declare one thing, because what the frontend records is the fact
/// rather than the Rust shape that carried it:
///
/// * `#[prebindgen] pub type X = path::To<Thing>;` — how a foreign or
///   crate-private type gets a name here. A **one-way road**: the name is
///   thereafter the only way to spell that type inside the flat API, and the
///   qualified path stays refused. This declares a name; it is not an equivalence
///   between spellings: the normalization that makes `std::vec::Vec<T>` and
///   `Vec<T>` one key covers the names the *language* predeclares, and a crate's
///   own alias is never one of those — treating it as one is a category error.
/// * `#[prebindgen] pub struct X(..);` — a tuple struct, whose fields no adapter
///   has ever crossed.
///
/// Not necessarily a *handle*: `#[prebindgen] pub type Duration =
/// std::time::Duration;` crosses by value through a `convert!`, erased to a plain
/// integer. What it becomes — an opaque pointer, a `ptr_class`, a conversion — is
/// the adapter's decision, and this says only that the frontend does not model the
/// contents.
#[derive(Clone, Debug)]
pub struct Extern {
    pub name: syn::Ident,
    /// What the declaration points at, for an alias — `std::time::Duration`,
    /// `zenoh::Session`, `handles::Storage`. `None` for a tuple struct, which is
    /// itself the definition.
    ///
    /// Informational, and deliberately **not** classified. `Error` is
    /// `Box<dyn std::error::Error + Send + Sync>` behind a `zenoh::` alias in one
    /// crate and spelled openly in another, so being "a std type" is a property of
    /// the spelling, not of the type. An adapter that wants to recognise a target
    /// may; the frontend does not decide for it.
    pub target: Option<String>,
    /// The declaring item — a type alias or a tuple struct.
    pub origin: Origin<syn::Item>,
}

/// A `#[prebindgen]` free function.
#[derive(Clone, Debug)]
pub struct Function {
    pub name: syn::Ident,
    /// Parameters in declaration order.
    pub params: Vec<Param>,
    /// What the function returns. An elided return is
    /// [`TypeKind::Unit`](super::TypeKind), exactly as a written `-> ()` is:
    /// they mean the same thing, differ only in spelling, and every consumer
    /// today already normalizes one to the other on the spot.
    pub ret: TypeRef,
    /// The whole item: attributes, `cfg`, doc comments, body.
    pub origin: Origin<syn::ItemFn>,
}

impl Function {
    /// A synthesized **nullary getter**: `pub fn <ident>() -> <ret>`.
    ///
    /// The model's own constructor for the one element an adapter legitimately
    /// needs to invent — a declared `const`'s accessor, whose type flows through
    /// the ordinary output-converter machinery and so has to arrive as a
    /// `Function` like any other.
    ///
    /// It lives here because building one means **spelling** `ret`, and #280
    /// says a `TypeRef` is the model's to mint. An adapter that built this
    /// itself needed a spelling for a model element — which dragged the
    /// emission capability into validation and Kotlin rendering, both of which
    /// only wanted the resulting `Function`.
    ///
    /// The body is `unimplemented!()` and is never emitted: only the signature
    /// is read.
    pub fn synthetic_getter(ident: syn::Ident, ret: TypeRef) -> Self {
        let ret_syntax = ret.spell();
        let item: syn::ItemFn = syn::parse_quote! {
            pub fn #ident() -> #ret_syntax {
                unimplemented!()
            }
        };
        Self {
            name: ident,
            params: Vec::new(),
            origin: ret.origin_with(item),
            ret,
        }
    }
}

/// One parameter of a [`Function`].
#[derive(Clone, Debug)]
pub struct Param {
    pub name: syn::Ident,
    pub ty: TypeRef,
    /// The parameter as written — `mode: Mode`.
    pub origin: Origin<syn::PatType>,
}

/// A `#[prebindgen]` struct: a product of fields that cross the boundary.
///
/// A struct whose contents do *not* cross is an [`Extern`], not a `Struct` with
/// nothing in it — so `fields` is a plain list, and empty means the source wrote
/// a struct with no fields.
///
/// Whether the fields are named or positional is not recorded: a [`Field`]
/// already knows its own address, and the delimiters are spelling, read off the
/// syntax when the struct is spelled.
#[derive(Clone, Debug)]
pub struct Struct {
    pub name: syn::Ident,
    pub fields: Vec<Field>,
    pub origin: Origin<syn::ItemStruct>,
    /// This struct **as a type**, taken at parse time — the twin of
    /// [`Variant::reading`], stored and `pub(super)` for the same two reasons.
    pub(super) reading: TypeRef,
}

impl Struct {
    /// This struct as a type reference — what the **declaration** answers when
    /// something needs a reading naming it.
    ///
    /// The alternative is composing one from the name at the call site, which
    /// an adapter cannot do (minting is sealed to this crate) and which would
    /// be phase-dependent if routed through the registry instead: a
    /// decomposition is declared before anything is interned. The declaration
    /// is the one thing that can always say. Same reasoning as
    /// [`Variant::type_ref`].
    pub fn type_ref(&self) -> &TypeRef {
        &self.reading
    }
}

/// A `#[prebindgen]` enum whose alternatives carry payloads — a sum type.
///
/// Distinct from [`Enum`], which is the fieldless shape, because the two are
/// consumed as different constructs and **numbered differently**. A sum's
/// alternatives are identified by position: the mirror an adapter builds carries
/// no `repr` and numbers its own arms, so a Rust discriminant would be the wrong
/// answer here — which is why there is no slot for one.
///
/// Both shapes are spelled `enum` in Rust and both keep a `syn::ItemEnum` in
/// their origin. Which one an item *is* is the classification, and it is decided
/// once: any alternative with a field makes it a `Variant`.
#[derive(Clone, Debug)]
pub struct Variant {
    pub name: syn::Ident,
    /// Alternatives in declaration order; `alternatives[i].index == i`.
    pub alternatives: Vec<Alternative>,
    pub origin: Origin<syn::ItemEnum>,
    /// This sum **as a type**, taken at parse time — see [`Self::type_ref`].
    ///
    /// **Stored, not computed**, and that is what makes the accessor safe: a
    /// method composing `TypeRef::named(&self.name)` would answer for whatever
    /// name a caller put in the struct, so a `Variant` named `String` would
    /// yield `Named` over the spelling `String` — which the model reads as
    /// `Str`. A stored reading cannot disagree with the model, because the
    /// model is what put it there, and an assembler has no way to mint a
    /// different one.
    ///
    /// `pub(super)` is the second line, not the first: it also stops a
    /// `Variant` being assembled at all outside `flat` (`E0451`), so `name` and
    /// `reading` cannot be paired inconsistently with *each other*.
    pub(super) reading: TypeRef,
}

impl Variant {
    /// A reference to this sum **as a type** — what a consumer needs when it
    /// has to name the sum rather than walk it (jnigen's `SumTag` selector,
    /// which carries *which* sum it chooses between).
    ///
    /// The **declaration** answers, so no consumer has to mint a reading from
    /// the name and hope it matches what the model would have said.
    ///
    /// This returns state the parser took, **not** a fresh composition, and the
    /// difference is the difference between sealing and appearing to.
    ///
    /// A version that composed `TypeRef::named(&self.name)` would hand a
    /// `Variant` assembled with the name `String` a
    /// [`Named`](super::TypeKind::Named) over the spelling `String` — which the
    /// model reads as [`Str`](super::TypeKind::Str). That is the `kind`/`syntax`
    /// disagreement [`TypeRef`]'s private fields exist to prevent, and it was
    /// reachable from outside the crate while being invisible to every doctest
    /// there, because assembling the *element* is not minting the *type*.
    ///
    /// Reading a stored value closes it: whatever a caller does with the other
    /// fields, the reading here is the one the model made, and no caller can
    /// mint a different one to put in its place.
    ///
    /// The `Variant` is sealed as well — its `reading` field is `pub(super)` —
    /// so the two cannot even be paired inconsistently:
    ///
    /// ```compile_fail
    /// # use prebindgen_flat::flat::{Origin, Variant};
    /// let assembled = Variant {
    ///     name: syn::parse_str("String").unwrap(),
    ///     alternatives: vec![],
    ///     origin: Origin::new(
    ///         syn::parse_str("enum String { A(u8) }").unwrap(),
    ///         std::rc::Rc::new(Default::default()),
    ///     ),
    /// };
    /// let mismatched = assembled.type_ref();
    /// ```
    ///
    /// That doctest pins *"a consumer cannot assemble a `Variant`"* and nothing
    /// finer: measured, it still fails with the field made `pub` — as `E0063`
    /// (missing field) rather than `E0451` (private field), since a consumer
    /// cannot produce a `TypeRef` to supply either way. The visibility itself
    /// is the check the compiler runs on every build.
    pub fn type_ref(&self) -> &TypeRef {
        &self.reading
    }
}

/// One alternative of a [`Variant`].
#[derive(Clone, Debug)]
pub struct Alternative {
    pub name: syn::Ident,
    /// Position within its sum, `0..N-1` — the same fact a [`Field`] carries,
    /// for the same reason: a node handed out on its own still knows where it
    /// sits.
    ///
    /// This is the *only* numbering a sum has. What a destination language does
    /// with it is its own business: one may transmit it to say which alternative
    /// is live, another may send a name instead.
    pub index: usize,
    /// The alternative's payload, in declaration order. May be empty — a sum can
    /// mix payload-carrying and payload-free alternatives, and only the presence
    /// of *some* payload makes the type a `Variant`.
    pub fields: Vec<Field>,
    /// The alternative as written: delimiters, attributes, doc comments.
    pub origin: Origin<syn::Variant>,
}

impl Alternative {
    /// True when this alternative carries no payload.
    ///
    /// The *group* question, not the syntax one: `B`, `B()` and `B {}` are all
    /// empty by this test; what keeps their delimiters apart is the spelling,
    /// not this.
    pub fn is_empty(&self) -> bool {
        self.fields.is_empty()
    }
}

/// A `#[prebindgen]` enum whose every alternative is fieldless — the C-style
/// shape, a named set of integers.
///
/// Distinct from [`Variant`] because the identity of a member here is the value
/// Rust **assigns** it, not where it sits: a C header re-states each `= expr`
/// and a Kotlin `enum class` entry is `NAME(7)`. A sum has no such value, which
/// is why the two are separate entities rather than one with a dead field each.
#[derive(Clone, Debug)]
pub struct Enum {
    pub name: syn::Ident,
    /// This enum **as a type**, taken at parse time — the twin of
    /// [`Variant::reading`] and [`Struct::reading`], stored and `pub(super)`
    /// for the same two reasons.
    pub(super) reading: TypeRef,
    /// Values in declaration order; `values[i].index == i`.
    pub values: Vec<EnumValue>,
    pub origin: Origin<syn::ItemEnum>,
}

impl Enum {
    /// Every value paired with the number Rust assigns it, or the first value
    /// whose discriminant could not be evaluated.
    ///
    /// This is the numbering a destination language needs when it has no way to
    /// reference a Rust constant: a Kotlin `enum class` entry is `NAME(3)`, and
    /// the generated `int → value` decode matches on the same numbers, so both
    /// come from here and cannot drift. An `Err` is a refusal for *that*
    /// consumer only — one that re-emits the source spelling never asks.
    pub fn discriminant_values(&self) -> Result<Vec<(&syn::Ident, i64)>, &syn::Ident> {
        self.values
            .iter()
            .map(|v| match v.discriminant {
                Some(n) => Ok((&v.name, n)),
                None => Err(&v.name),
            })
            .collect()
    }
}

impl Enum {
    /// This enum as a type reference — what the **declaration** answers.
    /// See [`Variant::type_ref`].
    pub fn type_ref(&self) -> &TypeRef {
        &self.reading
    }
}

/// One named value of an [`Enum`].
#[derive(Clone, Debug)]

pub struct EnumValue {
    pub name: syn::Ident,
    /// Position within its enum, `0..N-1`. Not the identity — see
    /// [`Self::discriminant`] — but the same "where it sits" fact every node in
    /// an ordered list carries, and what a consumer falls back to when a
    /// discriminant cannot be evaluated.
    pub index: usize,
    /// The value Rust assigns — an explicit `= N` sets it, an implicit value
    /// takes the previous plus one, starting at 0. **This shape's identity.**
    ///
    /// `None` once a spelling the frontend cannot evaluate (a `const`, a `cfg`,
    /// arithmetic) has broken the chain, or once the chain has run out of `i64`.
    /// That is not a failure: only a consumer that needs the *number* is
    /// affected, and one that re-emits the *spelling* reads
    /// [`Self::origin`]`.syntax.discriminant` instead.
    pub discriminant: Option<i64>,
    /// The value as written: `= 0x07`, attributes, doc comments — and its
    /// delimiters, since `B` and `B()` are both fieldless and still spelled
    /// differently.
    pub origin: Origin<syn::Variant>,
}

/// One field of a [`Struct`] or of an [`Alternative`].
#[derive(Clone, Debug)]
pub struct Field {
    /// The field's name, or `None` for a positional one.
    pub name: Option<syn::Ident>,
    /// Position within its struct or alternative, `0..N-1` — the same fact an
    /// [`Alternative`] carries.
    ///
    /// The address of a positional field. A named field has one too, and simply
    /// does not need it: it is addressed by name, so this is available rather
    /// than used — the same way it carries its item's location.
    pub index: usize,
    pub ty: TypeRef,
    /// The field as written — `pub id: u64`, attributes and docs included.
    pub origin: Origin<syn::Field>,
}

/// A `#[prebindgen]` constant.
///
/// Always named: an unnamed `const _` is a [`Guard`], not a constant with no
/// address.
#[derive(Clone, Debug)]
pub struct Constant {
    pub name: syn::Ident,
    pub ty: TypeRef,
    /// The whole item — the initializer expression included, which is where a
    /// consumer that re-emits the value reads it from.
    pub origin: Origin<syn::ItemConst>,
}

/// An **anonymous const**: `const _: T = ..`, whatever produced it.
///
/// The definition is the shape, not the origin. A const with no name has no
/// address, so nothing can declare it, reference it, or emit it as an alias —
/// which is what puts it outside the flat API rather than in it, and that holds
/// however the item arrived: synthesized, hand-fed to [`FlatBuilder`](super::FlatBuilder), or written
/// as `#[prebindgen] const _: () = ..` in a source crate.
///
/// **Today's producer** is [`Source`](prebindgen::Source)'s cfg filter, which
/// synthesizes `const _: () = { konst::assertc_eq!(..) }` to assert that a source
/// crate's `FEATURES` match what the build script asked for. Nothing *marked*
/// that one — it is prebindgen's own item riding the same stream — but it is not
/// the only thing that can land here, and the model does not claim otherwise.
///
/// **Cardinality is zero or more.** `enable_feature_filtering(None)` produces
/// none, and each item iterator taken from a `Source` emits its own, so composing
/// two iterators from one crate yields two.
///
/// It carries no type, unlike a [`Constant`]. The item is emitted verbatim, so
/// what its types mean is the consumer crate's business; modelling them would let
/// a guard that names something undeclared refuse the whole element.
#[derive(Clone, Debug)]
pub struct Guard {
    /// Emitted verbatim, so the item is all there is.
    pub origin: Origin<syn::ItemConst>,
}

/// An item the language cannot express.
#[derive(Clone, Debug)]
pub struct Unsupported {
    /// The item's identifier, or `None` for an item kind that has none.
    pub name: Option<syn::Ident>,
    /// What could not be expressed, ready to be raised by whatever declares
    /// this item. Boxed: it is the size outlier among the elements, and this
    /// one is the rare variant.
    pub error: Box<super::ItemError>,
    /// The item as written, so a diagnosis can quote the source.
    pub origin: Origin<syn::Item>,
}

/// An item's `///` documentation, read off the attributes it was captured
/// with: `#[doc = " …"]` lines in order, one leading space stripped per line,
/// joined with `\n`; `None` when there are none. `*/` is defanged so the text
/// is safe inside a `/** … */` block, which is what every destination that
/// re-emits prose needs.
///
/// **Here rather than in an adapter.** A doc comment is something the *source*
/// said, so it is the model's to report — and reading it was the last common
/// reason an emitter reached for a captured item's node. Two adapters wanting
/// the same prose is one function, not a copy each.
fn docs_from(attrs: &[syn::Attribute]) -> Option<String> {
    let mut lines: Vec<String> = Vec::new();
    for attr in attrs {
        if !attr.path().is_ident("doc") {
            continue;
        }
        let syn::Meta::NameValue(nv) = &attr.meta else {
            continue;
        };
        let syn::Expr::Lit(syn::ExprLit {
            lit: syn::Lit::Str(s),
            ..
        }) = &nv.value
        else {
            continue;
        };
        let raw = s.value();
        let line = raw.strip_prefix(' ').unwrap_or(&raw);
        lines.push(line.replace("*/", "*\u{200B}/"));
    }
    (!lines.is_empty()).then(|| lines.join("\n"))
}

macro_rules! docs_accessor {
    ($($ty:ident),+ $(,)?) => {$(
        impl $ty {
            /// This item's `///` documentation.
            pub fn docs(&self) -> Option<String> {
                docs_from(&self.origin.syntax.attrs)
            }
        }
    )+};
}

docs_accessor!(
    Function,
    Struct,
    Enum,
    Variant,
    Constant,
    Field,
    Alternative,
    EnumValue
);