flan/lib/ast.ml
Joseph Ferano 9a820d86cd Sweep every field label from the colon spelling to the dot
The script is in tools/ rather than thrown away, because two lanes are
writing Flan in the old spelling right now and their files need the same
pass at merge.

It works on forms, not on text: a keyword becomes a dot only where it sits
in a field-label position inside a brace, so an enum member in value
position, a map key inside an EDN string and a type-position {K V} are all
left alone. :keys keeps its colon -- it names no field.
2026-09-12 14:47:54 +07:00

152 lines
6.9 KiB
OCaml

(** The AST: syntax with special forms recognised, before typing.
Sugar is gone by this point. [when], [unless], [cond] and [and]/[or] are
desugared into [If] and [Do]; they are compiler special forms until macros
arrive at milestone 5, so there is nothing to preserve for a macroexpander
to see yet.
Types here are *surface* type expressions, not resolved types. [Ptr] and
[Option] are still just names; the checker resolves them. *)
(* ── Type expressions ──────────────────────────────────────────────── *)
type texpr = { t : texpr_kind; tloc : Loc.t }
and texpr_kind =
| Tname of string (* i32 bool Cursor string *)
| Tslice of texpr (* [u8] ptr+len *)
| Tarray of len * texpr (* [4 f32] [rows [cols u32]] *)
| Tmap of texpr * texpr (* {string i32} *)
| Tapp of string * texpr list (* (Ptr Cursor) (Option f64) *)
| Tfn of texpr list * texpr (* (Fn [a a] bool) *)
(* An array length is an integer or a compile-time constant's name. *)
and len =
| Lint of int64
| Lname of string
(* ── Expressions ───────────────────────────────────────────────────── *)
type expr = { e : expr_kind; loc : Loc.t }
and expr_kind =
| Int of int64
| Float of float
| Byte of int
| Str of string
| Kw of string (* :space — coerced at typed call sites *)
| Quote of string (* 'skip-form — restart names *)
| Var of string
| Do of expr list
| Let of binding list * expr list
| If of expr * expr * expr option
| While of expr * expr list
| Return of expr option
| Set of place * expr
| Field of expr * string (* (.pos c) — auto-derefs one level *)
| Call of expr * expr list
| Match of expr * arm list
| Struct of string * (string * expr) list (* (Cursor {.src s}) *)
| Arr of expr list (* [0xE6B800FF ...] — a fixed array value *)
(* These bind names or alter control flow, so none of them can be a call. *)
| Fn of string list * expr list (* (fn [x y] ...) — non-escaping *)
| Dotimes of string * expr * expr list (* (dotimes [i n] ...) *)
| Defer of expr list (* runs on scope exit *)
| Unwrap of unwrap * expr (* (some x) / (try x) *)
(* (handler-bind [(Type [c] body ...) ...] body ...) — spec-conditions.md.
A clause binds a name for the condition, so this cannot be a call. *)
| HandlerBind of hclause list * expr list
| Signal of sigkind * expr (* (signal c) / (error c) *)
(* (restart-case body (name [p T] body ...) ...) and
(invoke-restart 'name arg ...). Both alter control flow, so neither can be
a call, and a clause binds its parameters — §3. *)
| RestartCase of expr * rclause list
| InvokeRestart of string * expr list
(* Two ways to signal, because they are two different things — §1 and §2.
[signal] returns Unit whatever it finds; [error] has type Never and, with
nothing transferring, the program stops. *)
and sigkind = Ssignal | Serror
and hclause = { hty : texpr; hname : string; hbody : expr list; hloc : Loc.t }
(* [rparams] are §3's inline annotations, the same name/type pairs a [defn]
takes. They are bound in the clause body and filled in by whatever invoked
the restart, which is why their count and types are checked at run time
(§3): a restart is found by name on a dynamic stack. *)
and rclause =
{ rname : string; rparams : field list; rbody : expr list; rloc : Loc.t }
(* Inline name/type pairs, as in [defn], [let] and [defstruct]. Here because a
restart clause's parameters are one, and a clause is part of an expression. *)
and field = { fname : string; fty : texpr; floc : Loc.t }
(* Two unwrap operators, because they are two different things — plan.org. *)
and unwrap = Usome | Utry
and binding = { bname : string; bty : texpr option; bval : expr; bloc : Loc.t }
(* The fixed list of assignable forms — spec-memory.md. Not setf. *)
and place =
| Pvar of string
| Pfield of expr * string (* (set (.hp e) v) *)
| Pindex of expr * expr list (* (set (at grid r c) v) *)
| Pderef of expr (* (set (deref p) v) *)
and arm = { pat : pattern; body : expr list; aloc : Loc.t }
and pattern =
| Pctor of string * string list (* (Some e) (Rect w h) None *)
| Pwild (* _ :else *)
(* ── Declarations ──────────────────────────────────────────────────── *)
type fn = {
name : string;
params : field list;
ret : texpr option; (* None means Unit *)
fbody : expr list;
nloc : Loc.t;
}
type decl = { d : decl_kind; dloc : Loc.t }
and decl_kind =
| Package of string
| Import of string * string (* alias, path *)
| Defalias of string * texpr
| Defstruct of string * field list
| Defunion of string * variant list
| Defn of fn
(* No body, so no [defn]: a foreign function, and the string is the C symbol
it is actually called by (plan.org, Types — [declare] is kept only where
there is no body). *)
| Declare of fn * string
(* The same, but written in the C library's own terms — structs by value,
strings as strings. [Shim] generates the C that flattens it and rewrites
this into a [Declare] plus an ordinary [Defn], so nothing downstream sees
one. Two forms and not one because [(declare f [p string] ...)] already
means "the symbol takes ptr+len", which is the opposite of what this
means. *)
| DeclareC of fn * string
(* Inline name/value pairs, as everywhere else. The members are what a
keyword at a call site resolves against. *)
| Defenum of string * (string * int64) list
(* value is optional: ZII. `uninit` opts out and is recorded as Uninit. *)
| Defvar of string * texpr option * init
| Defconst of string * texpr option * expr
and variant = { vname : string; vfields : field list; vloc : Loc.t }
and init = Zeroed | Uninit | Init of expr
(* Every top-level name a declaration introduces, whatever kind it is. There is
one top-level namespace, so this is both the set [Load] renames on an import
and the set [Check] refuses to see twice — one definition, so the two cannot
drift apart. *)
let declared_name (d : decl) =
match d.d with
| Defenum (n, _) | Defalias (n, _) | Defstruct (n, _) | Defunion (n, _)
| Defvar (n, _, _) | Defconst (n, _, _) -> Some n
| Declare (fn, _) | DeclareC (fn, _) | Defn fn -> Some fn.name
| Package _ | Import _ -> None