catala/compiler/shared_ast/scope.mli

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

145 lines
5.1 KiB
OCaml
Raw Normal View History

(* This file is part of the Catala compiler, a specification language for tax
and social benefits computation rules. Copyright (C) 2020-2022 Inria,
contributor: Denis Merigoux <denis.merigoux@inria.fr>, Alain Delaët-Tixeuil
<alain.delaet--tixeuil@inria.fr>, Louis Gesbert <louis.gesbert@inria.fr>
Licensed under the Apache License, Version 2.0 (the "License"); you may not
use this file except in compliance with the License. You may obtain a copy of
the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
License for the specific language governing permissions and limitations under
the License. *)
(** Functions handling the code item structures of [shared_ast], in particular
the scopes *)
2022-11-21 12:46:17 +03:00
open Catala_utils
open Definitions
(** {2 Traversal functions} *)
val fold_left_lets :
f:('a -> 'e scope_let -> 'e Var.t -> 'a) ->
init:'a ->
'e scope_body_expr ->
'a
(** Usage:
[fold_left_lets ~f:(fun acc scope_let scope_let_var -> ...) ~init scope_lets],
where [scope_let_var] is the variable bound to the scope let in the next
scope lets to be examined. *)
val fold_right_lets :
f:('expr1 scope_let -> 'expr1 Var.t -> 'a -> 'a) ->
init:('expr1 -> 'a) ->
'expr1 scope_body_expr ->
'a
(** Usage:
[fold_right_lets ~f:(fun scope_let scope_let_var acc -> ...) ~init scope_lets],
where [scope_let_var] is the variable bound to the scope let in the next
scope lets to be examined (which are before in the program order). *)
val map_exprs_in_lets :
2023-12-20 19:43:31 +03:00
?transform_types:(typ -> typ) ->
Swap boxing and annotations in expressions This was the only reasonable solution I found to the issue raised [here](https://github.com/CatalaLang/catala/pull/334#discussion_r987175884). This was a pretty tedious rewrite, but it should now ensure we are doing things correctly. As a bonus, the "smart" expression constructors are now used everywhere to build expressions (so another refactoring like this one should be much easier) and this makes the code overall feel more straightforward (`Bindlib.box_apply` or `let+` no longer need to be visible!) --- Basically, we were using values of type `gexpr box = naked_gexpr marked box` throughout when (re-)building expressions. This was done 99% of the time by using `Bindlib.box_apply add_mark naked_e` right after building `naked_e`. In lots of places, we needed to recover the annotation of this expression later on, typically to build its parent term (to inherit the position, or build the type). Since it wasn't always possible to wrap these uses within `box_apply` (esp. as bindlib boxes aren't a monad), here and there we had to call `Bindlib.unbox`, just to recover the position or type. This had the very unpleasant effect of forcing the resolution of the whole box (including applying any stored closures) to reach the top-level annotation which isn't even dependant on specific variable bindings. Then, generally, throwing away the result. Therefore, the change proposed here transforms - `naked_gexpr marked Bindlib.box` into - `naked_gexpr Bindlib.box marked` (aliased to `boxed_gexpr` or `gexpr boxed` for convenience) This means only 1. not fitting the mark into the box right away when building, and 2. accessing the top-level mark directly without unboxing The functions for building terms from module `Shared_ast.Expr` could be changed easily. But then they needed to be consistently used throughout, without manually building terms through `Bindlib.apply_box` -- which covers most of the changes in this patch. `Expr.Box.inj` is provided to swap back to a box, before binding for example. Additionally, this gives a 40% speedup on `make -C examples pass_all_tests`, which hints at the amount of unnecessary work we were doing --'
2022-10-06 20:13:45 +03:00
f:('expr1 -> 'expr2 boxed) ->
varf:('expr1 Var.t -> 'expr2 Var.t) ->
'expr1 scope_body_expr ->
Swap boxing and annotations in expressions This was the only reasonable solution I found to the issue raised [here](https://github.com/CatalaLang/catala/pull/334#discussion_r987175884). This was a pretty tedious rewrite, but it should now ensure we are doing things correctly. As a bonus, the "smart" expression constructors are now used everywhere to build expressions (so another refactoring like this one should be much easier) and this makes the code overall feel more straightforward (`Bindlib.box_apply` or `let+` no longer need to be visible!) --- Basically, we were using values of type `gexpr box = naked_gexpr marked box` throughout when (re-)building expressions. This was done 99% of the time by using `Bindlib.box_apply add_mark naked_e` right after building `naked_e`. In lots of places, we needed to recover the annotation of this expression later on, typically to build its parent term (to inherit the position, or build the type). Since it wasn't always possible to wrap these uses within `box_apply` (esp. as bindlib boxes aren't a monad), here and there we had to call `Bindlib.unbox`, just to recover the position or type. This had the very unpleasant effect of forcing the resolution of the whole box (including applying any stored closures) to reach the top-level annotation which isn't even dependant on specific variable bindings. Then, generally, throwing away the result. Therefore, the change proposed here transforms - `naked_gexpr marked Bindlib.box` into - `naked_gexpr Bindlib.box marked` (aliased to `boxed_gexpr` or `gexpr boxed` for convenience) This means only 1. not fitting the mark into the box right away when building, and 2. accessing the top-level mark directly without unboxing The functions for building terms from module `Shared_ast.Expr` could be changed easily. But then they needed to be consistently used throughout, without manually building terms through `Bindlib.apply_box` -- which covers most of the changes in this patch. `Expr.Box.inj` is provided to swap back to a box, before binding for example. Additionally, this gives a 40% speedup on `make -C examples pass_all_tests`, which hints at the amount of unnecessary work we were doing --'
2022-10-06 20:13:45 +03:00
'expr2 scope_body_expr Bindlib.box
2023-03-28 10:38:47 +03:00
(** Usage
[map_exprs_in_lets ~f:(fun e -> ...) ~varf:(fun var -> ...) scope_body_expr],
where [e] is the right-hand-side of a scope let or the result of the scope
body, and [var] represents the left-hand-side variable of a scope let.
[~varf] is usually the identity function or [Var.translate] when the map
sends the expression to a new flavor of the shared AST. If [~reset_types] is
activated, then the resulting types in the scope let left-hand-sides will be
reset to [TAny]. *)
val fold_left :
f:('a -> 'expr1 code_item -> 'expr1 Var.t -> 'a) ->
init:'a ->
'expr1 code_item_list ->
'a
(** Usage: [fold_left ~f:(fun acc code_def code_var -> ...) ~init code_def],
where [code_var] is the variable bound to the code item in the next code
items to be examined. *)
val fold_right :
f:('expr1 code_item -> 'expr1 Var.t -> 'a -> 'a) ->
init:'a ->
'expr1 code_item_list ->
'a
(** Usage:
[fold_right_scope ~f:(fun scope_def scope_var acc -> ...) ~init scope_def],
where [scope_var] is the variable bound to the scope in the next scopes to
be examined (which are before in the program order). *)
Swap boxing and annotations in expressions This was the only reasonable solution I found to the issue raised [here](https://github.com/CatalaLang/catala/pull/334#discussion_r987175884). This was a pretty tedious rewrite, but it should now ensure we are doing things correctly. As a bonus, the "smart" expression constructors are now used everywhere to build expressions (so another refactoring like this one should be much easier) and this makes the code overall feel more straightforward (`Bindlib.box_apply` or `let+` no longer need to be visible!) --- Basically, we were using values of type `gexpr box = naked_gexpr marked box` throughout when (re-)building expressions. This was done 99% of the time by using `Bindlib.box_apply add_mark naked_e` right after building `naked_e`. In lots of places, we needed to recover the annotation of this expression later on, typically to build its parent term (to inherit the position, or build the type). Since it wasn't always possible to wrap these uses within `box_apply` (esp. as bindlib boxes aren't a monad), here and there we had to call `Bindlib.unbox`, just to recover the position or type. This had the very unpleasant effect of forcing the resolution of the whole box (including applying any stored closures) to reach the top-level annotation which isn't even dependant on specific variable bindings. Then, generally, throwing away the result. Therefore, the change proposed here transforms - `naked_gexpr marked Bindlib.box` into - `naked_gexpr Bindlib.box marked` (aliased to `boxed_gexpr` or `gexpr boxed` for convenience) This means only 1. not fitting the mark into the box right away when building, and 2. accessing the top-level mark directly without unboxing The functions for building terms from module `Shared_ast.Expr` could be changed easily. But then they needed to be consistently used throughout, without manually building terms through `Bindlib.apply_box` -- which covers most of the changes in this patch. `Expr.Box.inj` is provided to swap back to a box, before binding for example. Additionally, this gives a 40% speedup on `make -C examples pass_all_tests`, which hints at the amount of unnecessary work we were doing --'
2022-10-06 20:13:45 +03:00
val map :
f:('e1 code_item -> 'e2 code_item Bindlib.box) ->
varf:('e1 Var.t -> 'e2 Var.t) ->
'e1 code_item_list ->
'e2 code_item_list Bindlib.box
val map_ctx :
f:('ctx -> 'e1 code_item -> 'ctx * 'e2 code_item Bindlib.box) ->
varf:('e1 Var.t -> 'e2 Var.t) ->
'ctx ->
'e1 code_item_list ->
'e2 code_item_list Bindlib.box
(** Similar to [map], but a context is passed left-to-right through the given
function *)
val fold_map :
f:('ctx -> 'e1 Var.t -> 'e1 code_item -> 'ctx * 'e2 code_item Bindlib.box) ->
varf:('e1 Var.t -> 'e2 Var.t) ->
'ctx ->
'e1 code_item_list ->
'ctx * 'e2 code_item_list Bindlib.box
val map_exprs :
Swap boxing and annotations in expressions This was the only reasonable solution I found to the issue raised [here](https://github.com/CatalaLang/catala/pull/334#discussion_r987175884). This was a pretty tedious rewrite, but it should now ensure we are doing things correctly. As a bonus, the "smart" expression constructors are now used everywhere to build expressions (so another refactoring like this one should be much easier) and this makes the code overall feel more straightforward (`Bindlib.box_apply` or `let+` no longer need to be visible!) --- Basically, we were using values of type `gexpr box = naked_gexpr marked box` throughout when (re-)building expressions. This was done 99% of the time by using `Bindlib.box_apply add_mark naked_e` right after building `naked_e`. In lots of places, we needed to recover the annotation of this expression later on, typically to build its parent term (to inherit the position, or build the type). Since it wasn't always possible to wrap these uses within `box_apply` (esp. as bindlib boxes aren't a monad), here and there we had to call `Bindlib.unbox`, just to recover the position or type. This had the very unpleasant effect of forcing the resolution of the whole box (including applying any stored closures) to reach the top-level annotation which isn't even dependant on specific variable bindings. Then, generally, throwing away the result. Therefore, the change proposed here transforms - `naked_gexpr marked Bindlib.box` into - `naked_gexpr Bindlib.box marked` (aliased to `boxed_gexpr` or `gexpr boxed` for convenience) This means only 1. not fitting the mark into the box right away when building, and 2. accessing the top-level mark directly without unboxing The functions for building terms from module `Shared_ast.Expr` could be changed easily. But then they needed to be consistently used throughout, without manually building terms through `Bindlib.apply_box` -- which covers most of the changes in this patch. `Expr.Box.inj` is provided to swap back to a box, before binding for example. Additionally, this gives a 40% speedup on `make -C examples pass_all_tests`, which hints at the amount of unnecessary work we were doing --'
2022-10-06 20:13:45 +03:00
f:('expr1 -> 'expr2 boxed) ->
varf:('expr1 Var.t -> 'expr2 Var.t) ->
'expr1 code_item_list ->
'expr2 code_item_list Bindlib.box
(** This is the main map visitor for all the expressions inside all the scopes
of the program. *)
val get_body_mark : (_, 'm) gexpr scope_body -> 'm mark
(** {2 Conversions} *)
val to_expr :
decl_ctx -> ('a any, 'm) gexpr scope_body -> 'm mark -> ('a, 'm) boxed_gexpr
(** Usage: [to_expr ctx body scope_position] where [scope_position] corresponds
to the line of the scope declaration for instance. *)
type 'e scope_name_or_var = ScopeName of ScopeName.t | ScopeVar of 'e Var.t
val unfold :
decl_ctx ->
((_, 'm) gexpr as 'e) code_item_list ->
'm mark ->
'e scope_name_or_var ->
Swap boxing and annotations in expressions This was the only reasonable solution I found to the issue raised [here](https://github.com/CatalaLang/catala/pull/334#discussion_r987175884). This was a pretty tedious rewrite, but it should now ensure we are doing things correctly. As a bonus, the "smart" expression constructors are now used everywhere to build expressions (so another refactoring like this one should be much easier) and this makes the code overall feel more straightforward (`Bindlib.box_apply` or `let+` no longer need to be visible!) --- Basically, we were using values of type `gexpr box = naked_gexpr marked box` throughout when (re-)building expressions. This was done 99% of the time by using `Bindlib.box_apply add_mark naked_e` right after building `naked_e`. In lots of places, we needed to recover the annotation of this expression later on, typically to build its parent term (to inherit the position, or build the type). Since it wasn't always possible to wrap these uses within `box_apply` (esp. as bindlib boxes aren't a monad), here and there we had to call `Bindlib.unbox`, just to recover the position or type. This had the very unpleasant effect of forcing the resolution of the whole box (including applying any stored closures) to reach the top-level annotation which isn't even dependant on specific variable bindings. Then, generally, throwing away the result. Therefore, the change proposed here transforms - `naked_gexpr marked Bindlib.box` into - `naked_gexpr Bindlib.box marked` (aliased to `boxed_gexpr` or `gexpr boxed` for convenience) This means only 1. not fitting the mark into the box right away when building, and 2. accessing the top-level mark directly without unboxing The functions for building terms from module `Shared_ast.Expr` could be changed easily. But then they needed to be consistently used throughout, without manually building terms through `Bindlib.apply_box` -- which covers most of the changes in this patch. `Expr.Box.inj` is provided to swap back to a box, before binding for example. Additionally, this gives a 40% speedup on `make -C examples pass_all_tests`, which hints at the amount of unnecessary work we were doing --'
2022-10-06 20:13:45 +03:00
'e boxed
val build_typ_from_sig :
decl_ctx -> StructName.t -> StructName.t -> Pos.t -> typ
(** [build_typ_from_sig ctx in_struct out_struct pos] builds the arrow type for
the specified scope *)
val input_type : typ -> Runtime.io_input Mark.pos -> typ
2023-11-07 20:25:57 +03:00
(** Returns the correct input type for scope input variables: this is [typ] for
non-reentrant variables, but for reentrant variables, it is nested in a
[TDefault], which only applies to the return type on functions. Note that
this doesn't take thunking into account (thunking is added during the
scopelang->dcalc translation) *)
(** {2 Analysis and tests} *)
val free_vars_body_expr : 'e scope_body_expr -> 'e Var.Set.t
val free_vars_item : 'e code_item -> 'e Var.Set.t
val free_vars : 'e code_item_list -> 'e Var.Set.t