Init.Prelude

This is the first file in the Lean import hierarchy. It is responsible for setting up basic definitions, most of which Lean already has "built in knowledge" about, so it is important that they be set up in exactly this way. (For example, Lean will use PUnit in the desugaring of do notation, or in the pattern match compiler.)

@[inline]

def id {α : Sort u} (a : α) : α

The identity function. id takes an implicit argument α : Sort u (a type in any universe), and an argument a : α, and returns a.

Although this may look like a useless function, one application of the identity function is to explicitly put a type on an expression. If e has type T, and T' is definitionally equal to T, then @id T' e typechecks, and Lean knows that this expression has type T' rather than T. This can make a difference for typeclass inference, since T and T' may have different typeclass instances on them. show T' from e is sugar for an @id T' e expression.

Equations

• id a = a

@[inline]

def Function.comp {α : Sort u} {β : Sort v} {δ : Sort w} (f : β → δ) (g : α → β) : α → δ

Function composition is the act of pipelining the result of one function, to the input of another, creating an entirely new function. Example:

#eval Function.comp List.reverse (List.drop 2) [3, 2, 4, 1]
-- [1, 4]

You can use the notation f ∘ g as shorthand for Function.comp f g.

#eval (List.reverse ∘ List.drop 2) [3, 2, 4, 1]
-- [1, 4]

A simpler way of thinking about it, is that List.reverse ∘ List.drop 2 is equivalent to fun xs => List.reverse (List.drop 2 xs), the benefit is that the meaning of composition is obvious, and the representation is compact.

Equations

• (f ∘ g) x = f (g x)

@[inline]

def Function.const {α : Sort u} (β : Sort v) (a : α) : β → α

The constant function. If a : α, then Function.const β a : β → α is the "constant function with value a", that is, Function.const β a b = a.

example (b : Bool) : Function.const Bool 10 b = 10 :=
  rfl

#check Function.const Bool 10
-- Bool → Nat

Equations

• Function.const β a x✝ = a

@[irreducible]

def letFun {α : Sort u} {β : α → Sort v} (v : α) (f : (x : α) → β x) : β v

The encoding of let_fun x := v; b is letFun v (fun x => b). This is equal to (fun x => b) v, so the value of x is not accessible to b. This is in contrast to let x := v; b, where the value of x is accessible to b.

There is special support for letFun. Both WHNF and simp are aware of letFun and can reduce it when zeta reduction is enabled, despite the fact it is marked irreducible. For metaprogramming, the function Lean.Expr.letFun? can be used to recognize a let_fun expression to extract its parts as if it were a let expression.

Equations

• letFun v f = f v

@[reducible, inline]

abbrev inferInstance {α : Sort u} [i : α] : α

inferInstance synthesizes a value of any target type by typeclass inference. This function has the same type signature as the identity function, but the square brackets on the [i : α] argument means that it will attempt to construct this argument by typeclass inference. (This will fail if α is not a class.) Example:

#check (inferInstance : Inhabited Nat) -- Inhabited Nat

def foo : Inhabited (Nat × Nat) :=
  inferInstance

example : foo.default = (default, default) :=
  rfl

Equations

• inferInstance = i

@[reducible, inline]

abbrev inferInstanceAs (α : Sort u) [i : α] : α

inferInstanceAs α synthesizes a value of any target type by typeclass inference. This is just like inferInstance except that α is given explicitly instead of being inferred from the target type. It is especially useful when the target type is some α' which is definitionally equal to α, but the instance we are looking for is only registered for α (because typeclass search does not unfold most definitions, but definitional equality does.) Example:

#check inferInstanceAs (Inhabited Nat) -- Inhabited Nat

Equations

• inferInstanceAs α = i

inductive PUnit : Sort u

The unit type, the canonical type with one element, named unit or (). This is the universe-polymorphic version of Unit; it is preferred to use Unit instead where applicable. For more information about universe levels: Types as objects

• unit : PUnit

  PUnit.unit : PUnit is the canonical element of the unit type.

@[reducible, inline]

abbrev Unit : Type

The unit type, the canonical type with one element, named unit or (). In other words, it describes only a single value, which consists of said constructor applied to no arguments whatsoever. The Unit type is similar to void in languages derived from C.

Unit is actually defined as PUnit.{1} where PUnit is the universe polymorphic version. The Unit should be preferred over PUnit where possible to avoid unnecessary universe parameters.

In functional programming, Unit is the return type of things that "return nothing", since a type with one element conveys no additional information. When programming with monads, the type m Unit represents an action that has some side effects but does not return a value, while m α would be an action that has side effects and returns a value of type α.

Equations

• Unit = PUnit

@[reducible, match_pattern, inline]

abbrev Unit.unit : Unit

Unit.unit : Unit is the canonical element of the unit type. It can also be written as ().

Equations

• () = PUnit.unit

unsafe axiom lcErased : Type

Marker for information that has been erased by the code generator.

unsafe axiom lcProof {α : Prop} : α

Auxiliary unsafe constant used by the Compiler when erasing proofs from code.

It may look strange to have an axiom that says "every proposition is true", since this is obviously unsound, but the unsafe marker ensures that the kernel will not let this through into regular proofs. The lower levels of the code generator don't need proofs in terms, so this is used to stub the proofs out.

unsafe axiom lcCast {α : Sort u} {β : Sort v} (a : α) : β

Auxiliary unsafe constant used by the Compiler when erasing casts.

unsafe axiom lcUnreachable {α : Sort u} : α

Auxiliary unsafe constant used by the Compiler to mark unreachable code.

Like lcProof, this is an unsafe axiom, which means that even though it is not sound, the kernel will not let us use it for regular proofs.

Executing this expression to actually synthesize a value of type α causes immediate undefined behavior, and the compiler does take advantage of this to optimize the code assuming that it is not called. If it is not optimized out, it is likely to appear as a print message saying "unreachable code", but this behavior is not guaranteed or stable in any way.

inductive True : Prop

True is a proposition and has only an introduction rule, True.intro : True. In other words, True is simply true, and has a canonical proof, True.intro For more information: Propositional Logic

• intro : True

  True is true, and True.intro (or more commonly, trivial) is the proof.

inductive False : Prop

False is the empty proposition. Thus, it has no introduction rules. It represents a contradiction. False elimination rule, False.rec, expresses the fact that anything follows from a contradiction. This rule is sometimes called ex falso (short for ex falso sequitur quodlibet), or the principle of explosion. For more information: Propositional Logic

inductive Empty : Type

The empty type. It has no constructors. The Empty.rec eliminator expresses the fact that anything follows from the empty type.

inductive PEmpty : Sort u

The universe-polymorphic empty type. Prefer Empty or False where possible.

def Not (a : Prop) : Prop

Not p, or ¬p, is the negation of p. It is defined to be p → False, so if your goal is ¬p you can use intro h to turn the goal into h : p ⊢ False, and if you have hn : ¬p and h : p then hn h : False and (hn h).elim will prove anything. For more information: Propositional Logic

Equations

• (¬a) = (a → False)

@[macro_inline]

def False.elim {C : Sort u} (h : False) : C

False.elim : False → C says that from False, any desired proposition C holds. Also known as ex falso quodlibet (EFQ) or the principle of explosion.

The target type is actually C : Sort u which means it works for both propositions and types. When executed, this acts like an "unreachable" instruction: it is undefined behavior to run, but it will probably print "unreachable code". (You would need to construct a proof of false to run it anyway, which you can only do using sorry or unsound axioms.)

Equations

• h.elim = False.rec (fun (x : False) => C) h

@[macro_inline]

def absurd {a : Prop} {b : Sort v} (h₁ : a) (h₂ : ¬a) : b

Anything follows from two contradictory hypotheses. Example:

example (hp : p) (hnp : ¬p) : q := absurd hp hnp

For more information: Propositional Logic

Equations

• absurd h₁ h₂ = False.rec (fun (x : False) => b) ⋯

inductive Eq {α : Sort u_1} : α → α → Prop

The equality relation. It has one introduction rule, Eq.refl. We use a = b as notation for Eq a b. A fundamental property of equality is that it is an equivalence relation.

variable (α : Type) (a b c d : α)
variable (hab : a = b) (hcb : c = b) (hcd : c = d)

example : a = d :=
  Eq.trans (Eq.trans hab (Eq.symm hcb)) hcd

Equality is much more than an equivalence relation, however. It has the important property that every assertion respects the equivalence, in the sense that we can substitute equal expressions without changing the truth value. That is, given h1 : a = b and h2 : p a, we can construct a proof for p b using substitution: Eq.subst h1 h2. Example:

example (α : Type) (a b : α) (p : α → Prop)
        (h1 : a = b) (h2 : p a) : p b :=
  Eq.subst h1 h2

example (α : Type) (a b : α) (p : α → Prop)
    (h1 : a = b) (h2 : p a) : p b :=
  h1 ▸ h2

The triangle in the second presentation is a macro built on top of Eq.subst and Eq.symm, and you can enter it by typing \t. For more information: Equality

• refl {α : Sort u_1} (a : α) : a = a

  Eq.refl a : a = a is reflexivity, the unique constructor of the equality type. See also rfl, which is usually used instead.

@[match_pattern]

def rfl {α : Sort u} {a : α} : a = a

rfl : a = a is the unique constructor of the equality type. This is the same as Eq.refl except that it takes a implicitly instead of explicitly.

This is a more powerful theorem than it may appear at first, because although the statement of the theorem is a = a, Lean will allow anything that is definitionally equal to that type. So, for instance, 2 + 2 = 4 is proven in Lean by rfl, because both sides are the same up to definitional equality.

Equations

• ⋯ = ⋯

@[simp]

theorem id_eq {α : Sort u_1} (a : α) : id a = a

id x = x, as a @[simp] lemma.

theorem Eq.subst {α : Sort u} {motive : α → Prop} {a b : α} (h₁ : a = b) (h₂ : motive a) : motive b

The substitution principle for equality. If a = b and P a holds, then P b also holds. We conventionally use the name motive for P here, so that you can specify it explicitly using e.g. Eq.subst (motive := fun x => x < 5) if it is not otherwise inferred correctly.

This theorem is the underlying mechanism behind the rw tactic, which is essentially a fancy algorithm for finding good motive arguments to usefully apply this theorem to replace occurrences of a with b in the goal or hypotheses.

For more information: Equality

theorem Eq.symm {α : Sort u} {a b : α} (h : a = b) : b = a

Equality is symmetric: if a = b then b = a.

Because this is in the Eq namespace, if you have a variable h : a = b, h.symm can be used as shorthand for Eq.symm h as a proof of b = a.

For more information: Equality

theorem Eq.trans {α : Sort u} {a b c : α} (h₁ : a = b) (h₂ : b = c) : a = c

Equality is transitive: if a = b and b = c then a = c.

Because this is in the Eq namespace, if you have variables or expressions h₁ : a = b and h₂ : b = c, you can use h₁.trans h₂ : a = c as shorthand for Eq.trans h₁ h₂.

For more information: Equality

@[macro_inline]

def cast {α β : Sort u} (h : α = β) (a : α) : β

Cast across a type equality. If h : α = β is an equality of types, and a : α, then a : β will usually not typecheck directly, but this function will allow you to work around this and embed a in type β as cast h a : β.

It is best to avoid this function if you can, because it is more complicated to reason about terms containing casts, but if the types don't match up definitionally sometimes there isn't anything better you can do.

For more information: Equality

Equations

• cast h a = h ▸ a

theorem congrArg {α : Sort u} {β : Sort v} {a₁ a₂ : α} (f : α → β) (h : a₁ = a₂) : f a₁ = f a₂

Congruence in the function argument: if a₁ = a₂ then f a₁ = f a₂ for any (nondependent) function f. This is more powerful than it might look at first, because you can also use a lambda expression for f to prove that <something containing a₁> = <something containing a₂>. This function is used internally by tactics like congr and simp to apply equalities inside subterms.

For more information: Equality

theorem congr {α : Sort u} {β : Sort v} {f₁ f₂ : α → β} {a₁ a₂ : α} (h₁ : f₁ = f₂) (h₂ : a₁ = a₂) : f₁ a₁ = f₂ a₂

Congruence in both function and argument. If f₁ = f₂ and a₁ = a₂ then f₁ a₁ = f₂ a₂. This only works for nondependent functions; the theorem statement is more complex in the dependent case.

For more information: Equality

theorem congrFun {α : Sort u} {β : α → Sort v} {f g : (x : α) → β x} (h : f = g) (a : α) : f a = g a

Congruence in the function part of an application: If f = g then f a = g a.

Initialize the Quotient Module, which effectively adds the following definitions:

opaque Quot {α : Sort u} (r : α → α → Prop) : Sort u

opaque Quot.mk {α : Sort u} (r : α → α → Prop) (a : α) : Quot r

opaque Quot.lift {α : Sort u} {r : α → α → Prop} {β : Sort v} (f : α → β) :
  (∀ a b : α, r a b → Eq (f a) (f b)) → Quot r → β

opaque Quot.ind {α : Sort u} {r : α → α → Prop} {β : Quot r → Prop} :
  (∀ a : α, β (Quot.mk r a)) → ∀ q : Quot r, β q

opaque Quot {α : Sort u} (r : α → α → Prop) : Sort u

Let α be any type, and let r be an equivalence relation on α. It is mathematically common to form the "quotient" α / r, that is, the type of elements of α "modulo" r. Set theoretically, one can view α / r as the set of equivalence classes of α modulo r. If f : α → β is any function that respects the equivalence relation in the sense that for every x y : α, r x y implies f x = f y, then f "lifts" to a function f' : α / r → β defined on each equivalence class ⟦x⟧ by f' ⟦x⟧ = f x. Lean extends the Calculus of Constructions with additional constants that perform exactly these constructions, and installs this last equation as a definitional reduction rule.

Given a type α and any binary relation r on α, Quot r is a type. Note that r is not required to be an equivalence relation. Quot is the basic building block used to construct later the type Quotient.

opaque Quot.mk {α : Sort u} (r : α → α → Prop) (a : α) : Quot r

Given a type α and any binary relation r on α, Quot.mk maps α to Quot r. So that if r : α → α → Prop and a : α, then Quot.mk r a is an element of Quot r.

See Quot.

opaque Quot.ind {α : Sort u} {r : α → α → Prop} {β : Quot r → Prop} (mk : ∀ (a : α), β (Quot.mk r a)) (q : Quot r) : β q

Given a type α and any binary relation r on α, Quot.ind says that every element of Quot r is of the form Quot.mk r a.

See Quot and Quot.lift.

opaque Quot.lift {α : Sort u} {r : α → α → Prop} {β : Sort v} (f : α → β) (a : ∀ (a b : α), r a b → f a = f b) : Quot r → β

Given a type α, any binary relation r on α, a function f : α → β, and a proof h that f respects the relation r, then Quot.lift f h is the corresponding function on Quot r.

The idea is that for each element a in α, the function Quot.lift f h maps Quot.mk r a (the r-class containing a) to f a, wherein h shows that this function is well defined. In fact, the computation principle is declared as a reduction rule.

unsafe axiom Quot.lcInv {α : Sort u} {r : α → α → Prop} (q : Quot r) : α

Unsafe auxiliary constant used by the compiler to erase Quot.lift.

inductive HEq {α : Sort u} : α → {β : Sort u} → β → Prop

Heterogeneous equality. HEq a b asserts that a and b have the same type, and casting a across the equality yields b, and vice versa.

You should avoid using this type if you can. Heterogeneous equality does not have all the same properties as Eq, because the assumption that the types of a and b are equal is often too weak to prove theorems of interest. One important non-theorem is the analogue of congr: If HEq f g and HEq x y and f x and g y are well typed it does not follow that HEq (f x) (g y). (This does follow if you have f = g instead.) However if a and b have the same type then a = b and HEq a b are equivalent.

• refl {α : Sort u} (a : α) : HEq a a

  Reflexivity of heterogeneous equality.

@[match_pattern]

def HEq.rfl {α : Sort u} {a : α} : HEq a a

A version of HEq.refl with an implicit argument.

Equations

• ⋯ = ⋯

theorem eq_of_heq {α : Sort u} {a a' : α} (h : HEq a a') : a = a'

@[unbox]

structure Prod (α : Type u) (β : Type v) : Type (max u v)

Product type (aka pair). You can use α × β as notation for Prod α β. Given a : α and b : β, Prod.mk a b : Prod α β. You can use (a, b) as notation for Prod.mk a b. Moreover, (a, b, c) is notation for Prod.mk a (Prod.mk b c). Given p : Prod α β, p.1 : α and p.2 : β. They are short for Prod.fst p and Prod.snd p respectively. You can also write p.fst and p.snd. For more information: Constructors with Arguments

• fst : α

  The first projection out of a pair. if p : α × β then p.1 : α.

• snd : β

  The second projection out of a pair. if p : α × β then p.2 : β.

structure PProd (α : Sort u) (β : Sort v) : Sort (max (max 1 u) v)

Similar to Prod, but α and β can be propositions. You can use α ×' β as notation for PProd α β. We use this type internally to automatically generate the brecOn recursor.

• fst : α

  The first projection out of a pair. if p : PProd α β then p.1 : α.

• snd : β

  The second projection out of a pair. if p : PProd α β then p.2 : β.

structure MProd (α β : Type u) : Type u

Similar to Prod, but α and β are in the same universe. We say MProd is the universe monomorphic product type.

• fst : α

  The first projection out of a pair. if p : MProd α β then p.1 : α.

• snd : β

  The second projection out of a pair. if p : MProd α β then p.2 : β.

structure And (a b : Prop) : Prop

And a b, or a ∧ b, is the conjunction of propositions. It can be constructed and destructed like a pair: if ha : a and hb : b then ⟨ha, hb⟩ : a ∧ b, and if h : a ∧ b then h.left : a and h.right : b.

• intro :: (
• left : a

  Extract the left conjunct from a conjunction. h : a ∧ b then h.left, also notated as h.1, is a proof of a.

• right : b

  Extract the right conjunct from a conjunction. h : a ∧ b then h.right, also notated as h.2, is a proof of b.

• )

inductive Or (a b : Prop) : Prop

Or a b, or a ∨ b, is the disjunction of propositions. There are two constructors for Or, called Or.inl : a → a ∨ b and Or.inr : b → a ∨ b, and you can use match or cases to destruct an Or assumption into the two cases.

• inl {a b : Prop} (h : a) : a ∨ b

  Or.inl is "left injection" into an Or. If h : a then Or.inl h : a ∨ b.

• inr {a b : Prop} (h : b) : a ∨ b

  Or.inr is "right injection" into an Or. If h : b then Or.inr h : a ∨ b.

theorem Or.intro_left {a : Prop} (b : Prop) (h : a) : a ∨ b

Alias for Or.inl.

theorem Or.intro_right {b : Prop} (a : Prop) (h : b) : a ∨ b

Alias for Or.inr.

theorem Or.elim {a b c : Prop} (h : a ∨ b) (left : a → c) (right : b → c) : c

Proof by cases on an Or. If a ∨ b, and both a and b imply proposition c, then c is true.

theorem Or.resolve_left {a b : Prop} (h : a ∨ b) (na : ¬a) : b

theorem Or.resolve_right {a b : Prop} (h : a ∨ b) (nb : ¬b) : a

theorem Or.neg_resolve_left {a b : Prop} (h : ¬a ∨ b) (ha : a) : b

theorem Or.neg_resolve_right {a b : Prop} (h : a ∨ ¬b) (nb : b) : a

inductive Bool : Type

Bool is the type of boolean values, true and false. Classically, this is equivalent to Prop (the type of propositions), but the distinction is important for programming, because values of type Prop are erased in the code generator, while Bool corresponds to the type called bool or boolean in most programming languages.

• false : Bool

  The boolean value false, not to be confused with the proposition False.

• true : Bool

  The boolean value true, not to be confused with the proposition True.

structure Subtype {α : Sort u} (p : α → Prop) : Sort (max 1 u)

Subtype p, usually written as {x : α // p x}, is a type which represents all the elements x : α for which p x is true. It is structurally a pair-like type, so if you have x : α and h : p x then ⟨x, h⟩ : {x // p x}. An element s : {x // p x} will coerce to α but you can also make it explicit using s.1 or s.val.

• val : α

  If s : {x // p x} then s.val : α is the underlying element in the base type. You can also write this as s.1, or simply as s when the type is known from context.

• property : p self.val

  If s : {x // p x} then s.2 or s.property is the assertion that p s.1, that is, that s is in fact an element for which p holds.

@[reducible]

def optParam (α : Sort u) (default : α) : Sort u

Gadget for optional parameter support.

A binder like (x : α := default) in a declaration is syntax sugar for x : optParam α default, and triggers the elaborator to attempt to use default to supply the argument if it is not supplied.

Equations

• optParam α default = α

@[reducible]

def outParam (α : Sort u) : Sort u

Gadget for marking output parameters in type classes.

For example, the Membership class is defined as:

class Membership (α : outParam (Type u)) (γ : Type v)

This means that whenever a typeclass goal of the form Membership ?α ?γ comes up, Lean will wait to solve it until ?γ is known, but then it will run typeclass inference, and take the first solution it finds, for any value of ?α, which thereby determines what ?α should be.

This expresses that in a term like a ∈ s, s might be a Set α or List α or some other type with a membership operation, and in each case the "member" type α is determined by looking at the container type.

Equations

• outParam α = α

@[reducible]

def semiOutParam (α : Sort u) : Sort u

Gadget for marking semi output parameters in type classes.

Semi-output parameters influence the order in which arguments to type class instances are processed. Lean determines an order where all non-(semi-)output parameters to the instance argument have to be figured out before attempting to synthesize an argument (that is, they do not contain assignable metavariables created during TC synthesis). This rules out instances such as [Mul β] : Add α (because β could be anything). Marking a parameter as semi-output is a promise that instances of the type class will always fill in a value for that parameter.

For example, the Coe class is defined as:

class Coe (α : semiOutParam (Sort u)) (β : Sort v)

This means that all Coe instances should provide a concrete value for α (i.e., not an assignable metavariable). An instance like Coe Nat Int or Coe α (Option α) is fine, but Coe α Nat is not since it does not provide a value for α.

Equations

• semiOutParam α = α

@[reducible]

def namedPattern {α : Sort u} (x a : α) (h : x = a) : α

Auxiliary declaration used to implement named patterns like x@h:p.

Equations

• a = a

@[never_extract, extern lean_sorry]

axiom sorryAx (α : Sort u) (synthetic : Bool := false) : α

Auxiliary axiom used to implement sorry.

The sorry term/tactic expands to sorryAx _ (synthetic := false). This is a proof of anything, which is intended for stubbing out incomplete parts of a proof while still having a syntactically correct proof skeleton. Lean will give a warning whenever a proof uses sorry, so you aren't likely to miss it, but you can double check if a theorem depends on sorry by using #print axioms my_thm and looking for sorryAx in the axiom list.

The synthetic flag is false when written explicitly by the user, but it is set to true when a tactic fails to prove a goal, or if there is a type error in the expression. A synthetic sorry acts like a regular one, except that it suppresses follow-up errors in order to prevent one error from causing a cascade of other errors because the desired term was not constructed.

theorem eq_false_of_ne_true {b : Bool} : ¬b = true → b = false

theorem eq_true_of_ne_false {b : Bool} : ¬b = false → b = true

theorem ne_false_of_eq_true {b : Bool} : b = true → ¬b = false

theorem ne_true_of_eq_false {b : Bool} : b = false → ¬b = true

class Inhabited (α : Sort u) : Sort (max 1 u)

Inhabited α is a typeclass that says that α has a designated element, called (default : α). This is sometimes referred to as a "pointed type".

This class is used by functions that need to return a value of the type when called "out of domain". For example, Array.get! arr i : α returns a value of type α when arr : Array α, but if i is not in range of the array, it reports a panic message, but this does not halt the program, so it must still return a value of type α (and in fact this is required for logical consistency), so in this case it returns default.

• default : α

  default is a function that produces a "default" element of any Inhabited type. This element does not have any particular specified properties, but it is often an all-zeroes value.

class inductive Nonempty (α : Sort u) : Prop

Nonempty α is a typeclass that says that α is not an empty type, that is, there exists an element in the type. It differs from Inhabited α in that Nonempty α is a Prop, which means that it does not actually carry an element of α, only a proof that there exists such an element. Given Nonempty α, you can construct an element of α nonconstructively using Classical.choice.

• intro {α : Sort u} (val : α) : Nonempty α

  If val : α, then α is nonempty.

axiom Classical.choice {α : Sort u} : Nonempty α → α

The axiom of choice. Nonempty α is a proof that α has an element, but the element itself is erased. The axiom choice supplies a particular element of α given only this proof.

The textbook axiom of choice normally makes a family of choices all at once, but that is implied from this formulation, because if α : ι → Type is a family of types and h : ∀ i, Nonempty (α i) is a proof that they are all nonempty, then fun i => Classical.choice (h i) : ∀ i, α i is a family of chosen elements. This is actually a bit stronger than the ZFC choice axiom; this is sometimes called "global choice".

In Lean, we use the axiom of choice to derive the law of excluded middle (see Classical.em), so it will often show up in axiom listings where you may not expect. You can use #print axioms my_thm to find out if a given theorem depends on this or other axioms.

This axiom can be used to construct "data", but obviously there is no algorithm to compute it, so Lean will require you to mark any definition that would involve executing Classical.choice or other axioms as noncomputable, and will not produce any executable code for such definitions.

theorem Nonempty.elim {α : Sort u} {p : Prop} (h₁ : Nonempty α) (h₂ : α → p) : p

The elimination principle for Nonempty α. If Nonempty α, and we can prove p given any element x : α, then p holds. Note that it is essential that p is a Prop here; the version with p being a Sort u is equivalent to Classical.choice.

instance instNonemptyOfInhabited {α : Sort u} [Inhabited α] : Nonempty α

noncomputable def Classical.ofNonempty {α : Sort u} [Nonempty α] : α

A variation on Classical.choice that uses typeclass inference to infer the proof of Nonempty α.

Equations

• Classical.ofNonempty = Classical.choice ⋯

instance instNonemptyForall {α : Sort u} {β : Sort v} [Nonempty β] : Nonempty (α → β)

instance Pi.instNonempty {α : Sort u} {β : α → Sort v} [∀ (a : α), Nonempty (β a)] : Nonempty ((a : α) → β a)

instance instInhabitedSort : Inhabited (Sort u)

Equations

• instInhabitedSort = { default := PUnit }

instance instInhabitedForall (α : Sort u) {β : Sort v} [Inhabited β] : Inhabited (α → β)

Equations

• instInhabitedForall α = { default := fun (x : α) => default }

instance Pi.instInhabited {α : Sort u} {β : α → Sort v} [(a : α) → Inhabited (β a)] : Inhabited ((a : α) → β a)

Equations

• Pi.instInhabited = { default := fun (x : α) => default }

instance instInhabitedBool : Inhabited Bool

Equations

• instInhabitedBool = { default := false }

structure PLift (α : Sort u) : Type u

Universe lifting operation from Sort u to Type u.

• up :: (
• down : α

  Extract a value from PLift α

• )

theorem PLift.up_down {α : Sort u} (b : PLift α) : { down := b.down } = b

Bijection between α and PLift α

theorem PLift.down_up {α : Sort u} (a : α) : { down := a }.down = a

Bijection between α and PLift α

def NonemptyType : Type (u + 1)

NonemptyType.{u} is the type of nonempty types in universe u. It is mainly used in constant declarations where we wish to introduce a type and simultaneously assert that it is nonempty, but otherwise make the type opaque.

Equations

• NonemptyType = { α : Type ?u.4 // Nonempty α }

@[reducible, inline]

abbrev NonemptyType.type (type : NonemptyType) : Type u

The underlying type of a NonemptyType.

Equations

• type.type = type.val

instance instInhabitedNonemptyType : Inhabited NonemptyType

NonemptyType is inhabited, because PUnit is a nonempty type.

Equations

• instInhabitedNonemptyType = { default := ⟨PUnit, ⋯⟩ }

structure ULift (α : Type s) : Type (max s r)

Universe lifting operation from a lower Type universe to a higher one. To express this using level variables, the input is Type s and the output is Type (max s r), so if s ≤ r then the latter is (definitionally) Type r.

The universe variable r is written first so that ULift.{r} α can be used when s can be inferred from the type of α.

• up :: (
• down : α

  Extract a value from ULift α

• )

theorem ULift.up_down {α : Type u} (b : ULift α) : { down := b.down } = b

Bijection between α and ULift.{v} α

theorem ULift.down_up {α : Type u} (a : α) : { down := a }.down = a

Bijection between α and ULift.{v} α

class inductive Decidable (p : Prop) : Type

Decidable p is a data-carrying class that supplies a proof that p is either true or false. It is equivalent to Bool (and in fact it has the same code generation as Bool) together with a proof that the Bool is true iff p is.

Decidable instances are used to infer "computation strategies" for propositions, so that you can have the convenience of writing propositions inside if statements and executing them (which actually executes the inferred decidability instance instead of the proposition, which has no code).

If a proposition p is Decidable, then (by decide : p) will prove it by evaluating the decidability instance to isTrue h and returning h.

Because Decidable carries data, when writing @[simp] lemmas which include a Decidable instance on the LHS, it is best to use {_ : Decidable p} rather than [Decidable p] so that non-canonical instances can be found via unification rather than typeclass search.

• isFalse {p : Prop} (h : ¬p) : Decidable p

  Prove that p is decidable by supplying a proof of ¬p

• isTrue {p : Prop} (h : p) : Decidable p

  Prove that p is decidable by supplying a proof of p

@[inline_if_reduce]

def Decidable.decide (p : Prop) [h : Decidable p] : Bool

Convert a decidable proposition into a boolean value.

If p : Prop is decidable, then decide p : Bool is the boolean value which is true if p is true and false if p is false.

Equations

• decide p = Decidable.casesOn h (fun (x : ¬p) => false) fun (x : p) => true

@[reducible, inline]

abbrev DecidablePred {α : Sort u} (r : α → Prop) : Sort (max 1 u)

A decidable predicate. See Decidable.

Equations

• DecidablePred r = ((a : α) → Decidable (r a))

@[reducible, inline]

abbrev DecidableRel {α : Sort u} (r : α → α → Prop) : Sort (max 1 u)

A decidable relation. See Decidable.

Equations

• DecidableRel r = ((a b : α) → Decidable (r a b))

@[reducible, inline]

abbrev DecidableEq (α : Sort u) : Sort (max 1 u)

Asserts that α has decidable equality, that is, a = b is decidable for all a b : α. See Decidable.

Equations

• DecidableEq α = ((a b : α) → Decidable (a = b))

def decEq {α : Sort u} [inst : DecidableEq α] (a b : α) : Decidable (a = b)

Proves that a = b is decidable given DecidableEq α.

Equations

• decEq a b = inst a b

theorem decide_eq_true {p : Prop} [inst : Decidable p] : p → decide p = true

theorem decide_eq_false {p : Prop} [Decidable p] : ¬p → decide p = false

theorem of_decide_eq_true {p : Prop} [inst : Decidable p] : decide p = true → p

theorem of_decide_eq_false {p : Prop} [inst : Decidable p] : decide p = false → ¬p

theorem of_decide_eq_self_eq_true {α : Sort u_1} [inst : DecidableEq α] (a : α) : decide (a = a) = true

@[inline]

def Bool.decEq (a b : Bool) : Decidable (a = b)

Decidable equality for Bool

Equations

• false.decEq false = isTrue ⋯
• false.decEq true = isFalse ⋯
• true.decEq false = isFalse ⋯
• true.decEq true = isTrue ⋯

@[inline]

instance instDecidableEqBool : DecidableEq Bool

Equations

• instDecidableEqBool = Bool.decEq

class BEq (α : Type u) : Type u

BEq α is a typeclass for supplying a boolean-valued equality relation on α, notated as a == b. Unlike DecidableEq α (which uses a = b), this is Bool valued instead of Prop valued, and it also does not have any axioms like being reflexive or agreeing with =. It is mainly intended for programming applications. See LawfulBEq for a version that requires that == and = coincide.

Typically we prefer to put the "more variable" term on the left, and the "more constant" term on the right.

• beq : α → α → Bool

  Boolean equality, notated as a == b.

instance instBEqOfDecidableEq {α : Type u_1} [DecidableEq α] : BEq α

Equations

• instBEqOfDecidableEq = { beq := fun (a b : α) => decide (a = b) }

@[macro_inline]

def dite {α : Sort u} (c : Prop) [h : Decidable c] (t : c → α) (e : ¬c → α) : α

"Dependent" if-then-else, normally written via the notation if h : c then t(h) else e(h), is sugar for dite c (fun h => t(h)) (fun h => e(h)), and it is the same as if c then t else e except that t is allowed to depend on a proof h : c, and e can depend on h : ¬c. (Both branches use the same name for the hypothesis, even though it has different types in the two cases.)

We use this to be able to communicate the if-then-else condition to the branches. For example, Array.get arr i h expects a proof h : i < arr.size in order to avoid a bounds check, so you can write if h : i < arr.size then arr.get i h else ... to avoid the bounds check inside the if branch. (Of course in this case we have only lifted the check into an explicit if, but we could also use this proof multiple times or derive i < arr.size from some other proposition that we are checking in the if.)

Equations

• dite c t e = Decidable.casesOn h e t

if-then-else

@[macro_inline]

def ite {α : Sort u} (c : Prop) [h : Decidable c] (t e : α) : α

if c then t else e is notation for ite c t e, "if-then-else", which decides to return t or e depending on whether c is true or false. The explicit argument c : Prop does not have any actual computational content, but there is an additional [Decidable c] argument synthesized by typeclass inference which actually determines how to evaluate c to true or false. Write if h : c then t else e instead for a "dependent if-then-else" dite, which allows t/e to use the fact that c is true/false.

Equations

• (if c then t else e) = Decidable.casesOn h (fun (x : ¬c) => e) fun (x : c) => t

@[macro_inline]

instance instDecidableAnd {p q : Prop} [dp : Decidable p] [dq : Decidable q] : Decidable (p ∧ q)

Equations

• instDecidableAnd = isTrue ⋯
• instDecidableAnd = isFalse ⋯
• instDecidableAnd = isFalse ⋯

@[macro_inline]

instance instDecidableOr {p q : Prop} [dp : Decidable p] [dq : Decidable q] : Decidable (p ∨ q)

Equations

• instDecidableOr = isTrue ⋯
• instDecidableOr = isTrue ⋯
• instDecidableOr = isFalse ⋯

instance instDecidableNot {p : Prop} [dp : Decidable p] : Decidable ¬p

Equations

• instDecidableNot = isFalse ⋯
• instDecidableNot = isTrue hq

Boolean operators

@[macro_inline]

def cond {α : Type u} (c : Bool) (x y : α) : α

cond b x y is the same as if b then x else y, but optimized for a boolean condition. It can also be written as bif b then x else y. This is @[macro_inline] because x and y should not be eagerly evaluated (see ite).

Equations

• (bif true then x else y) = x
• (bif false then x else y) = y

@[macro_inline]

def Bool.or (x y : Bool) : Bool

or x y, or x || y, is the boolean "or" operation (not to be confused with Or : Prop → Prop → Prop, which is the propositional connective). It is @[macro_inline] because it has C-like short-circuiting behavior: if x is true then y is not evaluated.

Equations

• (true || y) = true
• (false || y) = y

@[macro_inline]

def Bool.and (x y : Bool) : Bool

and x y, or x && y, is the boolean "and" operation (not to be confused with And : Prop → Prop → Prop, which is the propositional connective). It is @[macro_inline] because it has C-like short-circuiting behavior: if x is false then y is not evaluated.

Equations

• (false && y) = false
• (true && y) = y

@[inline]

def Bool.not : Bool → Bool

not x, or !x, is the boolean "not" operation (not to be confused with Not : Prop → Prop, which is the propositional connective).

Equations

• (!true) = false
• (!false) = true

inductive Nat : Type

The type of natural numbers, starting at zero. It is defined as an inductive type freely generated by "zero is a natural number" and "the successor of a natural number is a natural number".

You can prove a theorem P n about n : Nat by induction n, which will expect a proof of the theorem for P 0, and a proof of P (succ i) assuming a proof of P i. The same method also works to define functions by recursion on natural numbers: induction and recursion are two expressions of the same operation from Lean's point of view.

open Nat
example (n : Nat) : n < succ n := by
  induction n with
  | zero =>
    show 0 < 1
    decide
  | succ i ih => -- ih : i < succ i
    show succ i < succ (succ i)
    exact Nat.succ_lt_succ ih

This type is special-cased by both the kernel and the compiler:

• The type of expressions contains "Nat literals" as a primitive constructor, and the kernel knows how to reduce zero/succ expressions to nat literals.
• If implemented naively, this type would represent a numeral n in unary as a linked list with n links, which is horribly inefficient. Instead, the runtime itself has a special representation for Nat which stores numbers up to 2^63 directly and larger numbers use an arbitrary precision "bignum" library (usually GMP).

• zero : Nat

  Nat.zero, is the smallest natural number. This is one of the two constructors of Nat. Using Nat.zero should usually be avoided in favor of 0 : Nat or simply 0, in order to remain compatible with the simp normal form defined by Nat.zero_eq.

• succ (n : Nat) : Nat

  The successor function on natural numbers, succ n = n + 1. This is one of the two constructors of Nat. Using succ n should usually be avoided in favor of n + 1, in order to remain compatible with the simp normal form defined by Nat.succ_eq_add_one.

instance instInhabitedNat : Inhabited Nat

Equations

• instInhabitedNat = { default := Nat.zero }

class OfNat (α : Type u) : Nat → Type u

The class OfNat α n powers the numeric literal parser. If you write 37 : α, Lean will attempt to synthesize OfNat α 37, and will generate the term (OfNat.ofNat 37 : α).

There is a bit of infinite regress here since the desugaring apparently still contains a literal 37 in it. The type of expressions contains a primitive constructor for "raw natural number literals", which you can directly access using the macro nat_lit 37. Raw number literals are always of type Nat. So it would be more correct to say that Lean looks for an instance of OfNat α (nat_lit 37), and it generates the term (OfNat.ofNat (nat_lit 37) : α).

• ofNat : α

  The OfNat.ofNat function is automatically inserted by the parser when the user writes a numeric literal like 1 : α. Implementations of this typeclass can therefore customize the behavior of n : α based on n and α.

@[defaultInstance 100]

instance instOfNatNat (n : Nat) : OfNat Nat n

Equations

• instOfNatNat n = { ofNat := n }

class LE (α : Type u) : Type u

LE α is the typeclass which supports the notation x ≤ y where x y : α.

• le : α → α → Prop

  The less-equal relation: x ≤ y

class LT (α : Type u) : Type u

LT α is the typeclass which supports the notation x < y where x y : α.

• lt : α → α → Prop

  The less-than relation: x < y

@[reducible]

def GE.ge {α : Type u} [LE α] (a b : α) : Prop

a ≥ b is an abbreviation for b ≤ a.

Equations

• (a ≥ b) = (b ≤ a)

@[reducible]

def GT.gt {α : Type u} [LT α] (a b : α) : Prop

a > b is an abbreviation for b < a.

Equations

• (a > b) = (b < a)

class Max (α : Type u) : Type u

Max α is the typeclass which supports the operation max x y where x y : α.

• max : α → α → α

  The maximum operation: max x y.

@[inline]

def maxOfLe {α : Type u_1} [LE α] [DecidableRel LE.le] : Max α

Implementation of the max operation using ≤.

Equations

• maxOfLe = { max := fun (x y : α) => if x ≤ y then y else x }

class Min (α : Type u) : Type u

Min α is the typeclass which supports the operation min x y where x y : α.

• min : α → α → α

  The minimum operation: min x y.

@[inline]

def minOfLe {α : Type u_1} [LE α] [DecidableRel LE.le] : Min α

Implementation of the min operation using ≤.

Equations

• minOfLe = { min := fun (x y : α) => if x ≤ y then x else y }

class Trans {α : Sort u_1} {β : Sort u_2} {γ : Sort u_3} (r : α → β → Sort u) (s : β → γ → Sort v) (t : outParam (α → γ → Sort w)) : Sort (max (max (max (max (max (max 1 u) u_1) u_2) u_3) v) w)

Transitive chaining of proofs, used e.g. by calc.

It takes two relations r and s as "input", and produces an "output" relation t, with the property that r a b and s b c implies t a c. The calc tactic uses this so that when it sees a chain with a ≤ b and b < c it knows that this should be a proof of a < c because there is an instance Trans (·≤·) (·<·) (·<·).

• trans {a : α} {b : β} {c : γ} : r a b → s b c → t a c

  Compose two proofs by transitivity, generalized over the relations involved.

instance instTransEq {α : Sort u_1} {γ : Sort u_2} (r : α → γ → Sort u) : Trans Eq r r

Equations

• instTransEq r = { trans := fun {a b : α} {c : γ} (heq : a = b) (h' : r b c) => ⋯ ▸ h' }

instance instTransEq_1 {α : Sort u_1} {β : Sort u_2} (r : α → β → Sort u) : Trans r Eq r

Equations

• instTransEq_1 r = { trans := fun {a : α} {b c : β} (h' : r a b) (heq : b = c) => heq ▸ h' }

class HAdd (α : Type u) (β : Type v) (γ : outParam (Type w)) : Type (max (max u v) w)

The notation typeclass for heterogeneous addition. This enables the notation a + b : γ where a : α, b : β.

• hAdd : α → β → γ

  a + b computes the sum of a and b. The meaning of this notation is type-dependent.

class HSub (α : Type u) (β : Type v) (γ : outParam (Type w)) : Type (max (max u v) w)

The notation typeclass for heterogeneous subtraction. This enables the notation a - b : γ where a : α, b : β.

• hSub : α → β → γ

  a - b computes the difference of a and b. The meaning of this notation is type-dependent.

  • For natural numbers, this operator saturates at 0: a - b = 0 when a ≤ b.

class HMul (α : Type u) (β : Type v) (γ : outParam (Type w)) : Type (max (max u v) w)

The notation typeclass for heterogeneous multiplication. This enables the notation a * b : γ where a : α, b : β.

• hMul : α → β → γ

  a * b computes the product of a and b. The meaning of this notation is type-dependent.

class HDiv (α : Type u) (β : Type v) (γ : outParam (Type w)) : Type (max (max u v) w)

The notation typeclass for heterogeneous division. This enables the notation a / b : γ where a : α, b : β.

• hDiv : α → β → γ

  a / b computes the result of dividing a by b. The meaning of this notation is type-dependent.

  • For most types like Nat, Int, Rat, Real, a / 0 is defined to be 0.
  • For Nat, a / b rounds downwards.
  • For Int, a / b rounds downwards if b is positive or upwards if b is negative. It is implemented as Int.ediv, the unique function satisfying a % b + b * (a / b) = a and 0 ≤ a % b < natAbs b for b ≠ 0. Other rounding conventions are available using the functions Int.fdiv (floor rounding) and Int.div (truncation rounding).
  • For Float, a / 0 follows the IEEE 754 semantics for division, usually resulting in inf or nan.

class HMod (α : Type u) (β : Type v) (γ : outParam (Type w)) : Type (max (max u v) w)

The notation typeclass for heterogeneous modulo / remainder. This enables the notation a % b : γ where a : α, b : β.

• hMod : α → β → γ

  a % b computes the remainder upon dividing a by b. The meaning of this notation is type-dependent.

  • For Nat and Int it satisfies a % b + b * (a / b) = a, and a % 0 is defined to be a.

class HPow (α : Type u) (β : Type v) (γ : outParam (Type w)) : Type (max (max u v) w)

The notation typeclass for heterogeneous exponentiation. This enables the notation a ^ b : γ where a : α, b : β.

• hPow : α → β → γ

  a ^ b computes a to the power of b. The meaning of this notation is type-dependent.

class HAppend (α : Type u) (β : Type v) (γ : outParam (Type w)) : Type (max (max u v) w)

The notation typeclass for heterogeneous append. This enables the notation a ++ b : γ where a : α, b : β.

• hAppend : α → β → γ

  a ++ b is the result of concatenation of a and b, usually read "append". The meaning of this notation is type-dependent.

class HOrElse (α : Type u) (β : Type v) (γ : outParam (Type w)) : Type (max (max u v) w)

The typeclass behind the notation a <|> b : γ where a : α, b : β. Because b is "lazy" in this notation, it is passed as Unit → β to the implementation so it can decide when to evaluate it.

• hOrElse : α → (Unit → β) → γ

  a <|> b executes a and returns the result, unless it fails in which case it executes and returns b. Because b is not always executed, it is passed as a thunk so it can be forced only when needed. The meaning of this notation is type-dependent.

class HAndThen (α : Type u) (β : Type v) (γ : outParam (Type w)) : Type (max (max u v) w)

The typeclass behind the notation a >> b : γ where a : α, b : β. Because b is "lazy" in this notation, it is passed as Unit → β to the implementation so it can decide when to evaluate it.

• hAndThen : α → (Unit → β) → γ

  a >> b executes a, ignores the result, and then executes b. If a fails then b is not executed. Because b is not always executed, it is passed as a thunk so it can be forced only when needed. The meaning of this notation is type-dependent.

class HAnd (α : Type u) (β : Type v) (γ : outParam (Type w)) : Type (max (max u v) w)

The typeclass behind the notation a &&& b : γ where a : α, b : β.

• hAnd : α → β → γ

  a &&& b computes the bitwise AND of a and b. The meaning of this notation is type-dependent.

class HXor (α : Type u) (β : Type v) (γ : outParam (Type w)) : Type (max (max u v) w)

The typeclass behind the notation a ^^^ b : γ where a : α, b : β.

• hXor : α → β → γ

  a ^^^ b computes the bitwise XOR of a and b. The meaning of this notation is type-dependent.

class HOr (α : Type u) (β : Type v) (γ : outParam (Type w)) : Type (max (max u v) w)

The typeclass behind the notation a ||| b : γ where a : α, b : β.

• hOr : α → β → γ

  a ||| b computes the bitwise OR of a and b. The meaning of this notation is type-dependent.

class HShiftLeft (α : Type u) (β : Type v) (γ : outParam (Type w)) : Type (max (max u v) w)

The typeclass behind the notation a <<< b : γ where a : α, b : β.

• hShiftLeft : α → β → γ

  a <<< b computes a shifted to the left by b places. The meaning of this notation is type-dependent.

  • On Nat, this is equivalent to a * 2 ^ b.
  • On UInt8 and other fixed width unsigned types, this is the same but truncated to the bit width.

class HShiftRight (α : Type u) (β : Type v) (γ : outParam (Type w)) : Type (max (max u v) w)

The typeclass behind the notation a >>> b : γ where a : α, b : β.

• hShiftRight : α → β → γ

  a >>> b computes a shifted to the right by b places. The meaning of this notation is type-dependent.

  • On Nat and fixed width unsigned types like UInt8, this is equivalent to a / 2 ^ b.

class Zero (α : Type u) : Type u

A type with a zero element.

• zero : α

  The zero element of the type.

class Add (α : Type u) : Type u

The homogeneous version of HAdd: a + b : α where a b : α.

• add : α → α → α

  a + b computes the sum of a and b. See HAdd.

class Sub (α : Type u) : Type u

The homogeneous version of HSub: a - b : α where a b : α.

• sub : α → α → α

  a - b computes the difference of a and b. See HSub.

class Mul (α : Type u) : Type u

The homogeneous version of HMul: a * b : α where a b : α.

• mul : α → α → α

  a * b computes the product of a and b. See HMul.

class Neg (α : Type u) : Type u

The notation typeclass for negation. This enables the notation -a : α where a : α.

• neg : α → α

  -a computes the negative or opposite of a. The meaning of this notation is type-dependent.

class Div (α : Type u) : Type u

The homogeneous version of HDiv: a / b : α where a b : α.

• div : α → α → α

  a / b computes the result of dividing a by b. See HDiv.

class Mod (α : Type u) : Type u

The homogeneous version of HMod: a % b : α where a b : α.

• mod : α → α → α

  a % b computes the remainder upon dividing a by b. See HMod.

class Dvd (α : Type u_1) : Type u_1

Notation typeclass for the ∣ operation (typed as \|), which represents divisibility.

• dvd : α → α → Prop

  Divisibility. a ∣ b (typed as \|) means that there is some c such that b = a * c.

class Pow (α : Type u) (β : Type v) : Type (max u v)

The homogeneous version of HPow: a ^ b : α where a : α, b : β. (The right argument is not the same as the left since we often want this even in the homogeneous case.)

Types can choose to subscribe to particular defaulting behavior by providing an instance to either NatPow or HomogeneousPow:

• NatPow is for types whose exponents is preferentially a Nat.
• HomogeneousPow is for types whose base and exponent are preferentially the same.

• pow : α → β → α

  a ^ b computes a to the power of b. See HPow.

class NatPow (α : Type u) : Type u

The homogeneous version of Pow where the exponent is a Nat. The purpose of this class is that it provides a default Pow instance, which can be used to specialize the exponent to Nat during elaboration.

For example, if x ^ 2 should preferentially elaborate with 2 : Nat then x's type should provide an instance for this class.

• pow : α → Nat → α

  a ^ n computes a to the power of n where n : Nat. See Pow.

class HomogeneousPow (α : Type u) : Type u

The completely homogeneous version of Pow where the exponent has the same type as the base. The purpose of this class is that it provides a default Pow instance, which can be used to specialize the exponent to have the same type as the base's type during elaboration. This is to say, a type should provide an instance for this class in case x ^ y should be elaborated with both x and y having the same type.

For example, the Float type provides an instance of this class, which causes expressions such as (2.2 ^ 2.2 : Float) to elaborate.

• pow : α → α → α

  a ^ b computes a to the power of b where a and b both have the same type.

class Append (α : Type u) : Type u

The homogeneous version of HAppend: a ++ b : α where a b : α.

• append : α → α → α

  a ++ b is the result of concatenation of a and b. See HAppend.

class OrElse (α : Type u) : Type u

The homogeneous version of HOrElse: a <|> b : α where a b : α. Because b is "lazy" in this notation, it is passed as Unit → α to the implementation so it can decide when to evaluate it.

• orElse : α → (Unit → α) → α

  The implementation of a <|> b : α. See HOrElse.

class AndThen (α : Type u) : Type u

The homogeneous version of HAndThen: a >> b : α where a b : α. Because b is "lazy" in this notation, it is passed as Unit → α to the implementation so it can decide when to evaluate it.

• andThen : α → (Unit → α) → α

  The implementation of a >> b : α. See HAndThen.

class AndOp (α : Type u) : Type u

The homogeneous version of HAnd: a &&& b : α where a b : α. (It is called AndOp because And is taken for the propositional connective.)

• and : α → α → α

  The implementation of a &&& b : α. See HAnd.

class Xor (α : Type u) : Type u

The homogeneous version of HXor: a ^^^ b : α where a b : α.

• xor : α → α → α

  The implementation of a ^^^ b : α. See HXor.

class OrOp (α : Type u) : Type u

The homogeneous version of HOr: a ||| b : α where a b : α. (It is called OrOp because Or is taken for the propositional connective.)

• or : α → α → α

  The implementation of a ||| b : α. See HOr.

class Complement (α : Type u) : Type u

The typeclass behind the notation ~~~a : α where a : α.

• complement : α → α

  The implementation of ~~~a : α.

class ShiftLeft (α : Type u) : Type u

The homogeneous version of HShiftLeft: a <<< b : α where a b : α.

• shiftLeft : α → α → α

  The implementation of a <<< b : α. See HShiftLeft.

class ShiftRight (α : Type u) : Type u

The homogeneous version of HShiftRight: a >>> b : α where a b : α.

• shiftRight : α → α → α

  The implementation of a >>> b : α. See HShiftRight.

@[defaultInstance 1000]

instance instHAdd {α : Type u_1} [Add α] : HAdd α α α

Equations

• instHAdd = { hAdd := fun (a b : α) => Add.add a b }

@[defaultInstance 1000]

instance instHSub {α : Type u_1} [Sub α] : HSub α α α

Equations

• instHSub = { hSub := fun (a b : α) => Sub.sub a b }

@[defaultInstance 1000]

instance instHMul {α : Type u_1} [Mul α] : HMul α α α

Equations

• instHMul = { hMul := fun (a b : α) => Mul.mul a b }

@[defaultInstance 1000]

instance instHDiv {α : Type u_1} [Div α] : HDiv α α α

Equations

• instHDiv = { hDiv := fun (a b : α) => Div.div a b }

@[defaultInstance 1000]

instance instHMod {α : Type u_1} [Mod α] : HMod α α α

Equations

• instHMod = { hMod := fun (a b : α) => Mod.mod a b }

@[defaultInstance 1000]

instance instHPow {α : Type u_1} {β : Type u_2} [Pow α β] : HPow α β α

Equations

• instHPow = { hPow := fun (a : α) (b : β) => Pow.pow a b }

@[defaultInstance 1000]

instance instPowNat {α : Type u_1} [NatPow α] : Pow α Nat

Equations

• instPowNat = { pow := fun (a : α) (n : Nat) => NatPow.pow a n }

@[defaultInstance 1000]

instance instPowOfHomogeneousPow {α : Type u_1} [HomogeneousPow α] : Pow α α

Equations

• instPowOfHomogeneousPow = { pow := fun (a b : α) => HomogeneousPow.pow a b }

@[defaultInstance 1000]

instance instHAppendOfAppend {α : Type u_1} [Append α] : HAppend α α α

Equations

• instHAppendOfAppend = { hAppend := fun (a b : α) => Append.append a b }

@[defaultInstance 1000]

instance instHOrElseOfOrElse {α : Type u_1} [OrElse α] : HOrElse α α α

Equations

• instHOrElseOfOrElse = { hOrElse := fun (a : α) (b : Unit → α) => OrElse.orElse a b }

@[defaultInstance 1000]

instance instHAndThenOfAndThen {α : Type u_1} [AndThen α] : HAndThen α α α

Equations

• instHAndThenOfAndThen = { hAndThen := fun (a : α) (b : Unit → α) => AndThen.andThen a b }

@[defaultInstance 1000]

instance instHAndOfAndOp {α : Type u_1} [AndOp α] : HAnd α α α

Equations

• instHAndOfAndOp = { hAnd := fun (a b : α) => AndOp.and a b }

@[defaultInstance 1000]

instance instHXorOfXor {α : Type u_1} [Xor α] : HXor α α α

Equations

• instHXorOfXor = { hXor := fun (a b : α) => Xor.xor a b }

@[defaultInstance 1000]

instance instHOrOfOrOp {α : Type u_1} [OrOp α] : HOr α α α

Equations

• instHOrOfOrOp = { hOr := fun (a b : α) => OrOp.or a b }

@[defaultInstance 1000]

instance instHShiftLeftOfShiftLeft {α : Type u_1} [ShiftLeft α] : HShiftLeft α α α

Equations

• instHShiftLeftOfShiftLeft = { hShiftLeft := fun (a b : α) => ShiftLeft.shiftLeft a b }

@[defaultInstance 1000]

instance instHShiftRightOfShiftRight {α : Type u_1} [ShiftRight α] : HShiftRight α α α

Equations

• instHShiftRightOfShiftRight = { hShiftRight := fun (a b : α) => ShiftRight.shiftRight a b }

class Membership (α : outParam (Type u)) (γ : Type v) : Type (max u v)

The typeclass behind the notation a ∈ s : Prop where a : α, s : γ. Because α is an outParam, the "container type" γ determines the type of the elements of the container.

• mem : γ → α → Prop

  The membership relation a ∈ s : Prop where a : α, s : γ.

@[match_pattern, extern lean_nat_add]

def Nat.add : Nat → Nat → Nat

Addition of natural numbers.

This definition is overridden in both the kernel and the compiler to efficiently evaluate using the "bignum" representation (see Nat). The definition provided here is the logical model (and it is soundness-critical that they coincide).

Equations

• x✝.add Nat.zero = x✝
• x✝.add b.succ = (x✝.add b).succ

instance instAddNat : Add Nat

Equations

• instAddNat = { add := Nat.add }

@[extern lean_nat_mul]

def Nat.mul : Nat → Nat → Nat

Multiplication of natural numbers.

This definition is overridden in both the kernel and the compiler to efficiently evaluate using the "bignum" representation (see Nat). The definition provided here is the logical model (and it is soundness-critical that they coincide).

Equations

• x✝.mul 0 = 0
• x✝.mul b.succ = (x✝.mul b).add x✝

instance instMulNat : Mul Nat

Equations

• instMulNat = { mul := Nat.mul }

@[extern lean_nat_pow]

def Nat.pow (m : Nat) : Nat → Nat

The power operation on natural numbers.

This definition is overridden in both the kernel and the compiler to efficiently evaluate using the "bignum" representation (see Nat). The definition provided here is the logical model.

Equations

• m.pow 0 = 1
• m.pow n.succ = (m.pow n).mul m

instance instNatPowNat : NatPow Nat

Equations

• instNatPowNat = { pow := Nat.pow }

@[extern lean_nat_dec_eq]

def Nat.beq : Nat → Nat → Bool

(Boolean) equality of natural numbers.

This definition is overridden in both the kernel and the compiler to efficiently evaluate using the "bignum" representation (see Nat). The definition provided here is the logical model (and it is soundness-critical that they coincide).

Equations

• Nat.zero.beq Nat.zero = true
• Nat.zero.beq n.succ = false
• n.succ.beq Nat.zero = false
• n.succ.beq m.succ = n.beq m

theorem Nat.eq_of_beq_eq_true {n m : Nat} : n.beq m = true → n = m

theorem Nat.ne_of_beq_eq_false {n m : Nat} : n.beq m = false → ¬n = m

@[reducible, extern lean_nat_dec_eq]

def Nat.decEq (n m : Nat) : Decidable (n = m)

A decision procedure for equality of natural numbers.

This definition is overridden in the compiler to efficiently evaluate using the "bignum" representation (see Nat). The definition provided here is the logical model.

Equations

• n.decEq m = match h : n.beq m with | true => isTrue ⋯ | false => isFalse ⋯

@[inline]

instance instDecidableEqNat : DecidableEq Nat

Equations

• instDecidableEqNat = Nat.decEq

@[extern lean_nat_dec_le]

def Nat.ble : Nat → Nat → Bool

The (Boolean) less-equal relation on natural numbers.

This definition is overridden in both the kernel and the compiler to efficiently evaluate using the "bignum" representation (see Nat). The definition provided here is the logical model (and it is soundness-critical that they coincide).

Equations

• Nat.zero.ble Nat.zero = true
• Nat.zero.ble n.succ = true
• n.succ.ble Nat.zero = false
• n.succ.ble m.succ = n.ble m

inductive Nat.le (n : Nat) : Nat → Prop

An inductive definition of the less-equal relation on natural numbers, characterized as the least relation ≤ such that n ≤ n and n ≤ m → n ≤ m + 1.

• refl {n : Nat} : n.le n

  Less-equal is reflexive: n ≤ n

• step {n m : Nat} : n.le m → n.le m.succ

  If n ≤ m, then n ≤ m + 1.

instance instLENat : LE Nat

Equations

• instLENat = { le := Nat.le }

def Nat.lt (n m : Nat) : Prop

The strict less than relation on natural numbers is defined as n < m := n + 1 ≤ m.

Equations

• n.lt m = n.succ.le m

instance instLTNat : LT Nat

Equations

• instLTNat = { lt := Nat.lt }

theorem Nat.not_succ_le_zero (n : Nat) : n.succ ≤ 0 → False

@[simp]

theorem Nat.not_lt_zero (n : Nat) : ¬n < 0

@[simp]

theorem Nat.zero_le (n : Nat) : 0 ≤ n

theorem Nat.succ_le_succ {n m : Nat} : n ≤ m → n.succ ≤ m.succ

@[simp]

theorem Nat.zero_lt_succ (n : Nat) : 0 < n.succ

theorem Nat.le_step {n m : Nat} (h : n ≤ m) : n ≤ m.succ

theorem Nat.le_trans {n m k : Nat} : n ≤ m → m ≤ k → n ≤ k

theorem Nat.lt_trans {n m k : Nat} (h₁ : n < m) : m < k → n < k

theorem Nat.le_succ (n : Nat) : n ≤ n.succ

theorem Nat.le_succ_of_le {n m : Nat} (h : n ≤ m) : n ≤ m.succ

@[simp]

theorem Nat.le_refl (n : Nat) : n ≤ n

theorem Nat.succ_pos (n : Nat) : 0 < n.succ

@[extern lean_nat_pred]

def Nat.pred : Nat → Nat

The predecessor function on natural numbers.

This definition is overridden in the compiler to use n - 1 instead. The definition provided here is the logical model.

Equations

• Nat.pred 0 = 0
• n.succ.pred = n

theorem Nat.pred_le_pred {n m : Nat} : n ≤ m → n.pred ≤ m.pred

theorem Nat.le_of_succ_le_succ {n m : Nat} : n.succ ≤ m.succ → n ≤ m

theorem Nat.le_of_lt_succ {m n : Nat} : m < n.succ → m ≤ n

theorem Nat.eq_or_lt_of_le {n m : Nat} : n ≤ m → n = m ∨ n < m

theorem Nat.lt_or_ge (n m : Nat) : n < m ∨ n ≥ m

theorem Nat.not_succ_le_self (n : Nat) : ¬n.succ ≤ n

@[simp]

theorem Nat.lt_irrefl (n : Nat) : ¬n < n

theorem Nat.lt_of_le_of_lt {n m k : Nat} (h₁ : n ≤ m) (h₂ : m < k) : n < k

theorem Nat.le_antisymm {n m : Nat} (h₁ : n ≤ m) (h₂ : m ≤ n) : n = m

theorem Nat.lt_of_le_of_ne {n m : Nat} (h₁ : n ≤ m) (h₂ : ¬n = m) : n < m

theorem Nat.le_of_ble_eq_true {n m : Nat} (h : n.ble m = true) : n ≤ m

theorem Nat.ble_self_eq_true (n : Nat) : n.ble n = true

theorem Nat.ble_succ_eq_true {n m : Nat} : n.ble m = true → n.ble m.succ = true

theorem Nat.ble_eq_true_of_le {n m : Nat} (h : n ≤ m) : n.ble m = true

theorem Nat.not_le_of_not_ble_eq_true {n m : Nat} (h : ¬n.ble m = true) : ¬n ≤ m

@[extern lean_nat_dec_le]

instance Nat.decLe (n m : Nat) : Decidable (n ≤ m)

Equations

• n.decLe m = if h : n.ble m = true then isTrue ⋯ else isFalse ⋯

@[extern lean_nat_dec_lt]

instance Nat.decLt (n m : Nat) : Decidable (n < m)

Equations

• n.decLt m = n.succ.decLe m

instance instMinNat : Min Nat

Equations

• instMinNat = minOfLe

@[extern lean_nat_sub]

def Nat.sub : Nat → Nat → Nat

(Truncated) subtraction of natural numbers. Because natural numbers are not closed under subtraction, we define n - m to be 0 when n < m.

This definition is overridden in both the kernel and the compiler to efficiently evaluate using the "bignum" representation (see Nat). The definition provided here is the logical model (and it is soundness-critical that they coincide).

Equations

• x✝.sub 0 = x✝
• x✝.sub b.succ = (x✝.sub b).pred

instance instSubNat : Sub Nat

Equations

• instSubNat = { sub := Nat.sub }

@[extern lean_system_platform_nbits]

opaque System.Platform.getNumBits : Unit → { n : Nat // n = 32 ∨ n = 64 }

Gets the word size of the platform. That is, whether the platform is 64 or 32 bits.

This function is opaque because we cannot guarantee at compile time that the target will have the same size as the host, and also because we would like to avoid typechecking being architecture-dependent. Nevertheless, Lean only works on 64 and 32 bit systems so we can encode this as a fact available for proof purposes.

def System.Platform.numBits : Nat

Gets the word size of the platform. That is, whether the platform is 64 or 32 bits.

Equations

• System.Platform.numBits = (System.Platform.getNumBits ()).val

theorem System.Platform.numBits_eq : System.Platform.numBits = 32 ∨ System.Platform.numBits = 64

structure Fin (n : Nat) : Type

Fin n is a natural number i with the constraint that 0 ≤ i < n. It is the "canonical type with n elements".

• val : Nat

  If i : Fin n, then i.val : ℕ is the described number. It can also be written as i.1 or just i when the target type is known.

• isLt : ↑self < n

  If i : Fin n, then i.2 is a proof that i.1 < n.

theorem Fin.eq_of_val_eq {n : Nat} {i j : Fin n} : ↑i = ↑j → i = j

theorem Fin.val_eq_of_eq {n : Nat} {i j : Fin n} (h : i = j) : ↑i = ↑j

instance instDecidableEqFin (n : Nat) : DecidableEq (Fin n)

Equations

• instDecidableEqFin n i j = match decEq ↑i ↑j with | isTrue h => isTrue ⋯ | isFalse h => isFalse ⋯

instance instLTFin {n : Nat} : LT (Fin n)

Equations

• instLTFin = { lt := fun (a b : Fin n) => ↑a < ↑b }

instance instLEFin {n : Nat} : LE (Fin n)

Equations

• instLEFin = { le := fun (a b : Fin n) => ↑a ≤ ↑b }

instance Fin.decLt {n : Nat} (a b : Fin n) : Decidable (a < b)

Equations

• a.decLt b = (↑a).decLt ↑b

instance Fin.decLe {n : Nat} (a b : Fin n) : Decidable (a ≤ b)

Equations

• a.decLe b = (↑a).decLe ↑b

structure BitVec (w : Nat) : Type

A bitvector of the specified width.

This is represented as the underlying Nat number in both the runtime and the kernel, inheriting all the special support for Nat.

• ofFin :: (
• toFin : Fin (2 ^ w)

  Interpret a bitvector as a number less than 2^w. O(1), because we use Fin as the internal representation of a bitvector.

• )

def BitVec.decEq {n : Nat} (x y : BitVec n) : Decidable (x = y)

Bitvectors have decidable equality. This should be used via the instance DecidableEq (BitVec n).

Equations

• { toFin := n_1 }.decEq { toFin := m } = if h : n_1 = m then isTrue ⋯ else isFalse ⋯

instance instDecidableEqBitVec {n : Nat} : DecidableEq (BitVec n)

Equations

• instDecidableEqBitVec = BitVec.decEq

@[match_pattern]

def BitVec.ofNatLt {n : Nat} (i : Nat) (p : i < 2 ^ n) : BitVec n

The BitVec with value i, given a proof that i < 2^n.

Equations

• i#'p = { toFin := ⟨i, p⟩ }

def BitVec.toNat {n : Nat} (x : BitVec n) : Nat

Given a bitvector x, return the underlying Nat. This is O(1) because BitVec is a (zero-cost) wrapper around a Nat.

Equations

• x.toNat = ↑x.toFin

instance instLTBitVec {n : Nat} : LT (BitVec n)

Equations

• instLTBitVec = { lt := fun (x1 x2 : BitVec n) => x1.toNat < x2.toNat }

instance instDecidableLtBitVec {n : Nat} (x y : BitVec n) : Decidable (x < y)

Equations

• instDecidableLtBitVec x y = inferInstanceAs (Decidable (x.toNat < y.toNat))

instance instLEBitVec {n : Nat} : LE (BitVec n)

Equations

• instLEBitVec = { le := fun (x1 x2 : BitVec n) => x1.toNat ≤ x2.toNat }

instance instDecidableLeBitVec {n : Nat} (x y : BitVec n) : Decidable (x ≤ y)

Equations

• instDecidableLeBitVec x y = inferInstanceAs (Decidable (x.toNat ≤ y.toNat))

@[reducible, inline]

abbrev UInt8.size : Nat

The size of type UInt8, that is, 2^8 = 256.

Equations

• UInt8.size = 256

structure UInt8 : Type

The type of unsigned 8-bit integers. This type has special support in the compiler to make it actually 8 bits rather than wrapping a Nat.

• toBitVec : BitVec 8

  Unpack a UInt8 as a BitVec 8. This function is overridden with a native implementation.

@[extern lean_uint8_of_nat]

def UInt8.ofNatCore (n : Nat) (h : n < UInt8.size) : UInt8

Pack a Nat less than 2^8 into a UInt8. This function is overridden with a native implementation.

Equations

• UInt8.ofNatCore n h = { toBitVec := n#'h }

@[extern lean_uint8_dec_eq]

def UInt8.decEq (a b : UInt8) : Decidable (a = b)

Decides equality on UInt8. This function is overridden with a native implementation.

Equations

• { toBitVec := n }.decEq { toBitVec := m } = if h : n = m then isTrue ⋯ else isFalse ⋯

instance instDecidableEqUInt8 : DecidableEq UInt8

Equations

• instDecidableEqUInt8 = UInt8.decEq

instance instInhabitedUInt8 : Inhabited UInt8

Equations

• instInhabitedUInt8 = { default := UInt8.ofNatCore 0 instInhabitedUInt8.proof_1 }

@[reducible, inline]

abbrev UInt16.size : Nat

The size of type UInt16, that is, 2^16 = 65536.

Equations

• UInt16.size = 65536

structure UInt16 : Type

The type of unsigned 16-bit integers. This type has special support in the compiler to make it actually 16 bits rather than wrapping a Nat.

• toBitVec : BitVec 16

  Unpack a UInt16 as a BitVec 16. This function is overridden with a native implementation.

@[extern lean_uint16_of_nat]

def UInt16.ofNatCore (n : Nat) (h : n < UInt16.size) : UInt16

Pack a Nat less than 2^16 into a UInt16. This function is overridden with a native implementation.

Equations

• UInt16.ofNatCore n h = { toBitVec := n#'h }

@[extern lean_uint16_dec_eq]

def UInt16.decEq (a b : UInt16) : Decidable (a = b)

Decides equality on UInt16. This function is overridden with a native implementation.

Equations

• { toBitVec := n }.decEq { toBitVec := m } = if h : n = m then isTrue ⋯ else isFalse ⋯

instance instDecidableEqUInt16 : DecidableEq UInt16

Equations

• instDecidableEqUInt16 = UInt16.decEq

instance instInhabitedUInt16 : Inhabited UInt16

Equations

• instInhabitedUInt16 = { default := UInt16.ofNatCore 0 instInhabitedUInt16.proof_1 }

@[reducible, inline]

abbrev UInt32.size : Nat

The size of type UInt32, that is, 2^32 = 4294967296.

Equations

• UInt32.size = 4294967296

structure UInt32 : Type

The type of unsigned 32-bit integers. This type has special support in the compiler to make it actually 32 bits rather than wrapping a Nat.

• toBitVec : BitVec 32

  Unpack a UInt32 as a `BitVec 32. This function is overridden with a native implementation.

@[extern lean_uint32_of_nat]

def UInt32.ofNatCore (n : Nat) (h : n < UInt32.size) : UInt32

Pack a Nat less than 2^32 into a UInt32. This function is overridden with a native implementation.

Equations

• UInt32.ofNatCore n h = { toBitVec := n#'h }

@[extern lean_uint32_to_nat]

def UInt32.toNat (n : UInt32) : Nat

Unpack a UInt32 as a Nat. This function is overridden with a native implementation.

Equations

• n.toNat = n.toBitVec.toNat

@[extern lean_uint32_dec_eq]

def UInt32.decEq (a b : UInt32) : Decidable (a = b)

Decides equality on UInt32. This function is overridden with a native implementation.

Equations

• { toBitVec := n }.decEq { toBitVec := m } = if h : n = m then isTrue ⋯ else isFalse ⋯

instance instDecidableEqUInt32 : DecidableEq UInt32

Equations

• instDecidableEqUInt32 = UInt32.decEq

instance instInhabitedUInt32 : Inhabited UInt32

Equations

• instInhabitedUInt32 = { default := UInt32.ofNatCore 0 instInhabitedUInt32.proof_1 }

instance instLTUInt32 : LT UInt32

Equations

• instLTUInt32 = { lt := fun (a b : UInt32) => a.toBitVec < b.toBitVec }

instance instLEUInt32 : LE UInt32

Equations

• instLEUInt32 = { le := fun (a b : UInt32) => a.toBitVec ≤ b.toBitVec }

@[extern lean_uint32_dec_lt]

def UInt32.decLt (a b : UInt32) : Decidable (a < b)

Decides less-equal on UInt32. This function is overridden with a native implementation.

Equations

• a.decLt b = inferInstanceAs (Decidable (a.toBitVec < b.toBitVec))

@[extern lean_uint32_dec_le]

def UInt32.decLe (a b : UInt32) : Decidable (a ≤ b)

Decides less-than on UInt32. This function is overridden with a native implementation.

Equations

• a.decLe b = inferInstanceAs (Decidable (a.toBitVec ≤ b.toBitVec))

instance instDecidableLtUInt32 (a b : UInt32) : Decidable (a < b)

Equations

• instDecidableLtUInt32 a b = a.decLt b

instance instDecidableLeUInt32 (a b : UInt32) : Decidable (a ≤ b)

Equations

• instDecidableLeUInt32 a b = a.decLe b

instance instMaxUInt32 : Max UInt32

Equations

• instMaxUInt32 = maxOfLe

instance instMinUInt32 : Min UInt32

Equations

• instMinUInt32 = minOfLe

@[reducible, inline]

abbrev UInt64.size : Nat

The size of type UInt64, that is, 2^64 = 18446744073709551616.

Equations

• UInt64.size = 18446744073709551616

structure UInt64 : Type

The type of unsigned 64-bit integers. This type has special support in the compiler to make it actually 64 bits rather than wrapping a Nat.

• toBitVec : BitVec 64

  Unpack a UInt64 as a BitVec 64. This function is overridden with a native implementation.

@[extern lean_uint64_of_nat]

def UInt64.ofNatCore (n : Nat) (h : n < UInt64.size) : UInt64

Pack a Nat less than 2^64 into a UInt64. This function is overridden with a native implementation.

Equations

• UInt64.ofNatCore n h = { toBitVec := n#'h }

@[extern lean_uint64_dec_eq]

def UInt64.decEq (a b : UInt64) : Decidable (a = b)

Decides equality on UInt64. This function is overridden with a native implementation.

Equations

• { toBitVec := n }.decEq { toBitVec := m } = if h : n = m then isTrue ⋯ else isFalse ⋯

instance instDecidableEqUInt64 : DecidableEq UInt64

Equations

• instDecidableEqUInt64 = UInt64.decEq

instance instInhabitedUInt64 : Inhabited UInt64

Equations

• instInhabitedUInt64 = { default := UInt64.ofNatCore 0 instInhabitedUInt64.proof_1 }

@[reducible, inline]

abbrev USize.size : Nat

The size of type USize, that is, 2^System.Platform.numBits.

Equations

• USize.size = 2 ^ System.Platform.numBits

theorem usize_size_eq : USize.size = 4294967296 ∨ USize.size = 18446744073709551616

theorem usize_size_pos : 0 < USize.size

structure USize : Type

A USize is an unsigned integer with the size of a word for the platform's architecture.

For example, if running on a 32-bit machine, USize is equivalent to UInt32. Or on a 64-bit machine, UInt64.

• toBitVec : BitVec System.Platform.numBits

  Unpack a USize as a BitVec System.Platform.numBits. This function is overridden with a native implementation.

@[extern lean_usize_of_nat]

def USize.ofNatCore (n : Nat) (h : n < USize.size) : USize

Pack a Nat less than USize.size into a USize. This function is overridden with a native implementation.

Equations

• USize.ofNatCore n h = { toBitVec := n#'h }

@[extern lean_usize_dec_eq]

def USize.decEq (a b : USize) : Decidable (a = b)

Decides equality on USize. This function is overridden with a native implementation.

Equations

• { toBitVec := n }.decEq { toBitVec := m } = if h : n = m then isTrue ⋯ else isFalse ⋯

instance instDecidableEqUSize : DecidableEq USize

Equations

• instDecidableEqUSize = USize.decEq

instance instInhabitedUSize : Inhabited USize

Equations

• instInhabitedUSize = { default := USize.ofNatCore 0 usize_size_pos }

@[reducible, inline]

abbrev Nat.isValidChar (n : Nat) : Prop

A Nat denotes a valid unicode codepoint if it is less than 0x110000, and it is also not a "surrogate" character (the range 0xd800 to 0xdfff inclusive).

Equations

• n.isValidChar = (n < 55296 ∨ 57343 < n ∧ n < 1114112)

@[reducible, inline]

abbrev UInt32.isValidChar (n : UInt32) : Prop

A UInt32 denotes a valid unicode codepoint if it is less than 0x110000, and it is also not a "surrogate" character (the range 0xd800 to 0xdfff inclusive).

Equations

• n.isValidChar = n.toNat.isValidChar

structure Char : Type

The Char Type represents an unicode scalar value. See http://www.unicode.org/glossary/#unicode_scalar_value).

• val : UInt32

  The underlying unicode scalar value as a UInt32.

• valid : self.val.isValidChar

  The value must be a legal codepoint.

@[extern lean_uint32_of_nat]

def Char.ofNatAux (n : Nat) (h : n.isValidChar) : Char

Pack a Nat encoding a valid codepoint into a Char. This function is overridden with a native implementation.

Equations

• Char.ofNatAux n h = { val := { toBitVec := n#'⋯ }, valid := h }

@[match_pattern, noinline]

def Char.ofNat (n : Nat) : Char

Convert a Nat into a Char. If the Nat does not encode a valid unicode scalar value, '\0' is returned instead.

Equations

• Char.ofNat n = if h : n.isValidChar then Char.ofNatAux n h else { val := { toBitVec := 0#'Char.ofNat.proof_1 }, valid := Char.ofNat.proof_2 }

theorem Char.eq_of_val_eq {c d : Char} : c.val = d.val → c = d

theorem Char.val_eq_of_eq {c d : Char} : c = d → c.val = d.val

theorem Char.ne_of_val_ne {c d : Char} (h : ¬c.val = d.val) : ¬c = d

theorem Char.val_ne_of_ne {c d : Char} (h : ¬c = d) : ¬c.val = d.val

instance instDecidableEqChar : DecidableEq Char

Equations

• instDecidableEqChar c d = match decEq c.val d.val with | isTrue h => isTrue ⋯ | isFalse h => isFalse ⋯

def Char.utf8Size (c : Char) : Nat

Returns the number of bytes required to encode this Char in UTF-8.

Equations

• One or more equations did not get rendered due to their size.

@[unbox]

inductive Option (α : Type u) : Type u

Option α is the type of values which are either some a for some a : α, or none. In functional programming languages, this type is used to represent the possibility of failure, or sometimes nullability.

For example, the function HashMap.get? : HashMap α β → α → Option β looks up a specified key a : α inside the map. Because we do not know in advance whether the key is actually in the map, the return type is Option β, where none means the value was not in the map, and some b means that the value was found and b is the value retrieved.

The xs[i] syntax, which is used to index into collections, has a variant xs[i]? that returns an optional value depending on whether the given index is valid. For example, if m : HashMap α β and a : α, then m[a]? is equivalent to HashMap.get? m a.

To extract a value from an Option α, we use pattern matching:

def map (f : α → β) (x : Option α) : Option β :=
  match x with
  | some a => some (f a)
  | none => none

We can also use if let to pattern match on Option and get the value in the branch:

def map (f : α → β) (x : Option α) : Option β :=
  if let some a := x then
    some (f a)
  else
    none

• none {α : Type u} : Option α

  No value.

• some {α : Type u} (val : α) : Option α

  Some value of type α.

instance instInhabitedOption {α : Type u_1} : Inhabited (Option α)

Equations

• instInhabitedOption = { default := none }

@[macro_inline]

def Option.getD {α : Type u_1} (opt : Option α) (dflt : α) : α

Get with default. If opt : Option α and dflt : α, then opt.getD dflt returns a if opt = some a and dflt otherwise.

This function is @[macro_inline], so dflt will not be evaluated unless opt turns out to be none.

Equations

• (some x).getD dflt = x
• none.getD dflt = dflt

@[inline]

def Option.map {α : Type u_1} {β : Type u_2} (f : α → β) : Option α → Option β

Map a function over an Option by applying the function to the contained value if present.

Equations

• Option.map f (some x) = some (f x)
• Option.map f none = none

inductive List (α : Type u) : Type u

List α is the type of ordered lists with elements of type α. It is implemented as a linked list.

List α is isomorphic to Array α, but they are useful for different things:

• List α is easier for reasoning, and Array α is modeled as a wrapper around List α
• List α works well as a persistent data structure, when many copies of the tail are shared. When the value is not shared, Array α will have better performance because it can do destructive updates.

• nil {α : Type u} : List α

  [] is the empty list.

• cons {α : Type u} (head : α) (tail : List α) : List α

  If a : α and l : List α, then cons a l, or a :: l, is the list whose first element is a and with l as the rest of the list.

instance instInhabitedList {α : Type u_1} : Inhabited (List α)

Equations

• instInhabitedList = { default := [] }

def List.hasDecEq {α : Type u} [DecidableEq α] (a b : List α) : Decidable (a = b)

Implements decidable equality for List α, assuming α has decidable equality.

Equations

• [].hasDecEq [] = isTrue ⋯
• (head :: tail).hasDecEq [] = isFalse ⋯
• [].hasDecEq (head :: tail) = isFalse ⋯
• (a :: as).hasDecEq (b :: bs) = match decEq a b with | isTrue hab => match as.hasDecEq bs with | isTrue habs => isTrue ⋯ | isFalse nabs => isFalse ⋯ | isFalse nab => isFalse ⋯

instance instDecidableEqList {α : Type u} [DecidableEq α] : DecidableEq (List α)

Equations

• instDecidableEqList = List.hasDecEq

def List.length {α : Type u_1} : List α → Nat

The length of a list: [].length = 0 and (a :: l).length = l.length + 1.

This function is overridden in the compiler to lengthTR, which uses constant stack space, while leaving this function to use the "naive" recursion which is easier for reasoning.

Equations

• [].length = 0
• (head :: as).length = as.length + 1

def List.lengthTRAux {α : Type u_1} : List α → Nat → Nat

Auxiliary function for List.lengthTR.

Equations

• [].lengthTRAux x✝ = x✝
• (head :: as).lengthTRAux x✝ = as.lengthTRAux x✝.succ

def List.lengthTR {α : Type u_1} (as : List α) : Nat

A tail-recursive version of List.length, used to implement List.length without running out of stack space.

Equations

• as.lengthTR = as.lengthTRAux 0

def List.get {α : Type u} (as : List α) : Fin as.length → α

as.get i returns the i'th element of the list as. This version of the function uses i : Fin as.length to ensure that it will not index out of bounds.

Equations

• (a :: tail).get ⟨0, isLt⟩ = a
• (head :: as).get ⟨i.succ, h⟩ = as.get ⟨i, ⋯⟩

def List.set {α : Type u_1} : List α → Nat → α → List α

l.set n a sets the value of list l at (zero-based) index n to a: [a, b, c, d].set 1 b' = [a, b', c, d]

Equations

• (head :: as).set 0 x✝ = x✝ :: as
• (a :: as).set n.succ x✝ = a :: as.set n x✝
• [].set x✝¹ x✝ = []

@[specialize #[]]

def List.foldl {α : Type u} {β : Type v} (f : α → β → α) (init : α) : List β → α

Folds a function over a list from the left: foldl f z [a, b, c] = f (f (f z a) b) c

Equations

• List.foldl f x✝ [] = x✝
• List.foldl f x✝ (b :: l) = List.foldl f (f x✝ b) l

def List.concat {α : Type u} : List α → α → List α

l.concat a appends a at the end of l, that is, l ++ [a].

Equations

• [].concat x✝ = [x✝]
• (a :: as).concat x✝ = a :: as.concat x✝

structure String : Type

String is the type of (UTF-8 encoded) strings.

The compiler overrides the data representation of this type to a byte sequence, and both String.utf8ByteSize and String.length are cached and O(1).

• data : List Char

  Unpack String into a List Char. This function is overridden by the compiler and is O(n) in the length of the list.

@[extern lean_string_dec_eq]

def String.decEq (s₁ s₂ : String) : Decidable (s₁ = s₂)

Decides equality on String. This function is overridden with a native implementation.

Equations

• { data := s₁_2 }.decEq { data := s₂_2 } = if h : s₁_2 = s₂_2 then isTrue ⋯ else isFalse ⋯

instance instDecidableEqString : DecidableEq String

Equations

• instDecidableEqString = String.decEq

structure String.Pos : Type

A byte position in a String. Internally, Strings are UTF-8 encoded. Codepoint positions (counting the Unicode codepoints rather than bytes) are represented by plain Nats instead. Indexing a String by a byte position is constant-time, while codepoint positions need to be translated internally to byte positions in linear-time.

A byte position p is valid for a string s if 0 ≤ p ≤ s.endPos and p lies on a UTF8 byte boundary.

• byteIdx : Nat

  Get the underlying byte index of a String.Pos

instance instInhabitedPos : Inhabited String.Pos

Equations

• instInhabitedPos = { default := { byteIdx := 0 } }

instance instDecidableEqPos : DecidableEq String.Pos

Equations

• instDecidableEqPos { byteIdx := b } { byteIdx := b_1 } = match decEq b b_1 with | isTrue h => isTrue ⋯ | isFalse h => isFalse ⋯

structure Substring : Type

A Substring is a view into some subslice of a String. The actual string slicing is deferred because this would require copying the string; here we only store a reference to the original string for garbage collection purposes.

• str : String

  The underlying string to slice.

• startPos : String.Pos

  The byte position of the start of the string slice.

• stopPos : String.Pos

  The byte position of the end of the string slice.

instance instInhabitedSubstring : Inhabited Substring

Equations

• instInhabitedSubstring = { default := { str := "", startPos := { byteIdx := 0 }, stopPos := { byteIdx := 0 } } }

@[inline]

def Substring.bsize : Substring → Nat

The byte length of the substring.

Equations

• { str := str, startPos := b, stopPos := e }.bsize = e.byteIdx.sub b.byteIdx

@[extern lean_string_utf8_byte_size]

def String.utf8ByteSize : String → Nat

The UTF-8 byte length of this string. This is overridden by the compiler to be cached and O(1).

Equations

• { data := s }.utf8ByteSize = String.utf8ByteSize.go s

def String.utf8ByteSize.go : List Char → Nat

Equations

• String.utf8ByteSize.go [] = 0
• String.utf8ByteSize.go (c :: cs) = String.utf8ByteSize.go cs + c.utf8Size

instance instHAddPos : HAdd String.Pos String.Pos String.Pos

Equations

• instHAddPos = { hAdd := fun (p₁ p₂ : String.Pos) => { byteIdx := p₁.byteIdx + p₂.byteIdx } }

instance instHSubPos : HSub String.Pos String.Pos String.Pos

Equations

• instHSubPos = { hSub := fun (p₁ p₂ : String.Pos) => { byteIdx := p₁.byteIdx - p₂.byteIdx } }

instance instHAddPosChar : HAdd String.Pos Char String.Pos

Equations

• instHAddPosChar = { hAdd := fun (p : String.Pos) (c : Char) => { byteIdx := p.byteIdx + c.utf8Size } }

instance instHAddPosString : HAdd String.Pos String String.Pos

Equations

• instHAddPosString = { hAdd := fun (p : String.Pos) (s : String) => { byteIdx := p.byteIdx + s.utf8ByteSize } }

instance instLEPos : LE String.Pos

Equations

• instLEPos = { le := fun (p₁ p₂ : String.Pos) => p₁.byteIdx ≤ p₂.byteIdx }

instance instLTPos : LT String.Pos

Equations

• instLTPos = { lt := fun (p₁ p₂ : String.Pos) => p₁.byteIdx < p₂.byteIdx }

instance instDecidableLePos (p₁ p₂ : String.Pos) : Decidable (p₁ ≤ p₂)

Equations

• instDecidableLePos p₁ p₂ = inferInstanceAs (Decidable (p₁.byteIdx ≤ p₂.byteIdx))

instance instDecidableLtPos (p₁ p₂ : String.Pos) : Decidable (p₁ < p₂)

Equations

• instDecidableLtPos p₁ p₂ = inferInstanceAs (Decidable (p₁.byteIdx < p₂.byteIdx))

@[inline]

def String.endPos (s : String) : String.Pos

A String.Pos pointing at the end of this string.

Equations

• s.endPos = { byteIdx := s.utf8ByteSize }

@[inline]

def String.toSubstring (s : String) : Substring

Convert a String into a Substring denoting the entire string.

Equations

• s.toSubstring = { str := s, startPos := { byteIdx := 0 }, stopPos := s.endPos }

def String.toSubstring' (s : String) : Substring

String.toSubstring without [inline] annotation.

Equations

• s.toSubstring' = s.toSubstring

unsafe def unsafeCast {α : Sort u} {β : Sort v} (a : α) : β

This function will cast a value of type α to type β, and is a no-op in the compiler. This function is extremely dangerous because there is no guarantee that types α and β have the same data representation, and this can lead to memory unsafety. It is also logically unsound, since you could just cast True to False. For all those reasons this function is marked as unsafe.

It is implemented by lifting both α and β into a common universe, and then using cast (lcProof : ULift (PLift α) = ULift (PLift β)) to actually perform the cast. All these operations are no-ops in the compiler.

Using this function correctly requires some knowledge of the data representation of the source and target types. Some general classes of casts which are safe in the current runtime:

• Array α to Array β where α and β have compatible representations, or more generally for other inductive types.
• Quot α r and α.
• @Subtype α p and α, or generally any structure containing only one non-Prop field of type α.
• Casting α to/from NonScalar when α is a boxed generic type (i.e. a function that accepts an arbitrary type α and is not specialized to a scalar type like UInt8).

Equations

• unsafeCast a = (cast ⋯ { down := { down := a } }).down.down

@[never_extract, extern lean_panic_fn]

def panicCore {α : Type u} [Inhabited α] (msg : String) : α

Auxiliary definition for panic.

Equations

• panicCore msg = default

@[never_extract, noinline]

def panic {α : Type u} [Inhabited α] (msg : String) : α

(panic "msg" : α) has a built-in implementation which prints msg to the error buffer. It does not terminate execution, and because it is a safe function, it still has to return an element of α, so it takes [Inhabited α] and returns default. It is primarily intended for debugging in pure contexts, and assertion failures.

Because this is a pure function with side effects, it is marked as @[never_extract] so that the compiler will not perform common sub-expression elimination and other optimizations that assume that the expression is pure.

Equations

• panic msg = panicCore msg

structure Array (α : Type u) : Type u

Array α is the type of dynamic arrays with elements from α. This type has special support in the runtime.

An array has a size and a capacity; the size is Array.size but the capacity is not observable from Lean code. Arrays perform best when unshared; as long as they are used "linearly" all updates will be performed destructively on the array, so it has comparable performance to mutable arrays in imperative programming languages.

From the point of view of proofs Array α is just a wrapper around List α.

• toList : List α

  Converts a Array α into an List α.

  At runtime, this projection is implemented by Array.toListImpl and is O(n) in the length of the array.

@[reducible, match_pattern, inline]

abbrev List.toArray {α : Type u_1} (xs : List α) : Array α

Converts a List α into an Array α.

You can also use the synonym List.toArray when dot notation is convenient.

At runtime, this constructor is implemented by List.toArrayImpl and is O(n) in the length of the list.

Equations

• xs.toArray = { toList := xs }

@[extern lean_mk_empty_array_with_capacity]

def Array.mkEmpty {α : Type u} (c : Nat) : Array α

Construct a new empty array with initial capacity c.

Equations

• Array.mkEmpty c = { toList := [] }

def Array.empty {α : Type u} : Array α

Construct a new empty array.

Equations

• #[] = Array.mkEmpty 0

@[reducible, extern lean_array_get_size]

def Array.size {α : Type u} (a : Array α) : Nat

Get the size of an array. This is a cached value, so it is O(1) to access.

Equations

• a.size = a.toList.length

@[extern lean_array_fget]

def Array.get {α : Type u} (a : Array α) (i : Nat) (h : i < a.size) : α

Access an element from an array without needing a runtime bounds checks, using a Nat index and a proof that it is in bounds.

This function does not use get_elem_tactic to automatically find the proof that the index is in bounds. This is because the tactic itself needs to look up values in arrays. Use the indexing notation a[i] instead.

Equations

• a.get i h = a.toList.get ⟨i, h⟩

@[reducible, inline]

abbrev Array.getD {α : Type u_1} (a : Array α) (i : Nat) (v₀ : α) : α

Access an element from an array, or return v₀ if the index is out of bounds.

Equations

• a.getD i v₀ = if h : i < a.size then a.get i h else v₀

@[extern lean_array_get]

def Array.get! {α : Type u} [Inhabited α] (a : Array α) (i : Nat) : α

Access an element from an array, or panic if the index is out of bounds.

Equations

• a.get! i = a.getD i default

@[extern lean_array_push]

def Array.push {α : Type u} (a : Array α) (v : α) : Array α

Push an element onto the end of an array. This is amortized O(1) because Array α is internally a dynamic array.

Equations

• a.push v = { toList := a.toList.concat v }

def Array.mkArray0 {α : Type u} : Array α

Create array #[]

Equations

• #[] = Array.mkEmpty 0

def Array.mkArray1 {α : Type u} (a₁ : α) : Array α

Create array #[a₁]

Equations

• #[a₁] = (Array.mkEmpty 1).push a₁

def Array.mkArray2 {α : Type u} (a₁ a₂ : α) : Array α

Create array #[a₁, a₂]

Equations

• #[a₁, a₂] = ((Array.mkEmpty 2).push a₁).push a₂

def Array.mkArray3 {α : Type u} (a₁ a₂ a₃ : α) : Array α

Create array #[a₁, a₂, a₃]

Equations

• #[a₁, a₂, a₃] = (((Array.mkEmpty 3).push a₁).push a₂).push a₃

def Array.mkArray4 {α : Type u} (a₁ a₂ a₃ a₄ : α) : Array α

Create array #[a₁, a₂, a₃, a₄]

Equations

• #[a₁, a₂, a₃, a₄] = ((((Array.mkEmpty 4).push a₁).push a₂).push a₃).push a₄

def Array.mkArray5 {α : Type u} (a₁ a₂ a₃ a₄ a₅ : α) : Array α

Create array #[a₁, a₂, a₃, a₄, a₅]

Equations

• #[a₁, a₂, a₃, a₄, a₅] = (((((Array.mkEmpty 5).push a₁).push a₂).push a₃).push a₄).push a₅

def Array.mkArray6 {α : Type u} (a₁ a₂ a₃ a₄ a₅ a₆ : α) : Array α

Create array #[a₁, a₂, a₃, a₄, a₅, a₆]

Equations

• #[a₁, a₂, a₃, a₄, a₅, a₆] = ((((((Array.mkEmpty 6).push a₁).push a₂).push a₃).push a₄).push a₅).push a₆

def Array.mkArray7 {α : Type u} (a₁ a₂ a₃ a₄ a₅ a₆ a₇ : α) : Array α

Create array #[a₁, a₂, a₃, a₄, a₅, a₆, a₇]

Equations

• #[a₁, a₂, a₃, a₄, a₅, a₆, a₇] = (((((((Array.mkEmpty 7).push a₁).push a₂).push a₃).push a₄).push a₅).push a₆).push a₇

def Array.mkArray8 {α : Type u} (a₁ a₂ a₃ a₄ a₅ a₆ a₇ a₈ : α) : Array α

Create array #[a₁, a₂, a₃, a₄, a₅, a₆, a₇, a₈]

Equations

• #[a₁, a₂, a₃, a₄, a₅, a₆, a₇, a₈] = ((((((((Array.mkEmpty 8).push a₁).push a₂).push a₃).push a₄).push a₅).push a₆).push a₇).push a₈

def Array.appendCore {α : Type u} (as bs : Array α) : Array α

Slower Array.append used in quotations.

Equations

• as.appendCore bs = Array.appendCore.loop bs bs.size 0 as

def Array.appendCore.loop {α : Type u} (bs : Array α) (i j : Nat) (as : Array α) : Array α

Equations

• Array.appendCore.loop bs i j as = if hlt : j < bs.size then match i with | 0 => as | i'.succ => Array.appendCore.loop bs i' (j + 1) (as.push (bs.get j hlt)) else as

def Array.extract {α : Type u_1} (as : Array α) (start stop : Nat) : Array α

Returns the slice of as from indices start to stop (exclusive). If start is greater or equal to stop, the result is empty. If stop is greater than the length of as, the length is used instead.

Equations

• as.extract start stop = Array.extract.loop as ((min stop as.size).sub start) start (Array.mkEmpty ((min stop as.size).sub start))

def Array.extract.loop {α : Type u_1} (as : Array α) (i j : Nat) (bs : Array α) : Array α

Equations

• Array.extract.loop as i j bs = if hlt : j < as.size then match i with | 0 => bs | i'.succ => Array.extract.loop as i' (j + 1) (bs.push (as.get j hlt)) else bs

class Bind (m : Type u → Type v) : Type (max (u + 1) v)

The typeclass which supplies the >>= "bind" function. See Monad.

• bind {α β : Type u} : m α → (α → m β) → m β

  If x : m α and f : α → m β, then x >>= f : m β represents the result of executing x to get a value of type α and then passing it to f.

class Pure (f : Type u → Type v) : Type (max (u + 1) v)

The typeclass which supplies the pure function. See Monad.

• pure {α : Type u} : α → f α

  If a : α, then pure a : f α represents a monadic action that does nothing and returns a.

class Functor (f : Type u → Type v) : Type (max (u + 1) v)

In functional programming, a "functor" is a function on types F : Type u → Type v equipped with an operator called map or <$> such that if f : α → β then map f : F α → F β, so f <$> x : F β if x : F α. This corresponds to the category-theory notion of functor in the special case where the category is the category of types and functions between them, except that this class supplies only the operations and not the laws (see LawfulFunctor).

• map {α β : Type u} : (α → β) → f α → f β

  If f : α → β and x : F α then f <$> x : F β.

• mapConst {α β : Type u} : α → f β → f α

  The special case const a <$> x, which can sometimes be implemented more efficiently.

class Seq (f : Type u → Type v) : Type (max (u + 1) v)

The typeclass which supplies the <*> "seq" function. See Applicative.

• seq {α β : Type u} : f (α → β) → (Unit → f α) → f β

  If mf : F (α → β) and mx : F α, then mf <*> mx : F β. In a monad this is the same as do let f ← mf; x ← mx; pure (f x): it evaluates first the function, then the argument, and applies one to the other.

  To avoid surprising evaluation semantics, mx is taken "lazily", using a Unit → f α function.

class SeqLeft (f : Type u → Type v) : Type (max (u + 1) v)

The typeclass which supplies the <* "seqLeft" function. See Applicative.

• seqLeft {α β : Type u} : f α → (Unit → f β) → f α

  If x : F α and y : F β, then x <* y evaluates x, then y, and returns the result of x.

  To avoid surprising evaluation semantics, y is taken "lazily", using a Unit → f β function.

class SeqRight (f : Type u → Type v) : Type (max (u + 1) v)

The typeclass which supplies the *> "seqRight" function. See Applicative.

• seqRight {α β : Type u} : f α → (Unit → f β) → f β

  If x : F α and y : F β, then x *> y evaluates x, then y, and returns the result of y.

  To avoid surprising evaluation semantics, y is taken "lazily", using a Unit → f β function.

class Applicative (f : Type u → Type v) extends Functor f, Pure f, Seq f, SeqLeft f, SeqRight f : Type (max (u + 1) v)

An applicative functor is an intermediate structure between Functor and Monad. It mainly consists of two operations:

• pure : α → F α
• seq : F (α → β) → F α → F β (written as <*>)

The seq operator gives a notion of evaluation order to the effects, where the first argument is executed before the second, but unlike a monad the results of earlier computations cannot be used to define later actions.

• map {α β : Type u} : (α → β) → f α → f β
• mapConst {α β : Type u} : α → f β → f α
• pure {α : Type u} : α → f α
• seq {α β : Type u} : f (α → β) → (Unit → f α) → f β
• seqLeft {α β : Type u} : f α → (Unit → f β) → f α
• seqRight {α β : Type u} : f α → (Unit → f β) → f β

class Monad (m : Type u → Type v) extends Applicative m, Bind m : Type (max (u + 1) v)

A monad is a structure which abstracts the concept of sequential control flow. It mainly consists of two operations:

• pure : α → F α
• bind : F α → (α → F β) → F β (written as >>=)

Like many functional programming languages, Lean makes extensive use of monads for structuring programs. In particular, the do notation is a very powerful syntax over monad operations, and it depends on a Monad instance.

See the do notation chapter of the manual for details.

• map {α β : Type u} : (α → β) → m α → m β
• mapConst {α β : Type u} : α → m β → m α
• pure {α : Type u} : α → m α
• seq {α β : Type u} : m (α → β) → (Unit → m α) → m β
• seqLeft {α β : Type u} : m α → (Unit → m β) → m α
• seqRight {α β : Type u} : m α → (Unit → m β) → m β
• bind {α β : Type u} : m α → (α → m β) → m β

instance instInhabitedForallOfMonad {α : Type u} {m : Type u → Type v} [Monad m] : Inhabited (α → m α)

Equations

• instInhabitedForallOfMonad = { default := pure }

instance instInhabitedOfMonad {α : Type u} {m : Type u → Type v} [Monad m] [Inhabited α] : Inhabited (m α)

Equations

• instInhabitedOfMonad = { default := pure default }

instance instNonemptyOfMonad {m : Type u_1 → Type u_2} {α : Type u_1} [Monad m] [Nonempty α] : Nonempty (m α)

class MonadLift (m : semiOutParam (Type u → Type v)) (n : Type u → Type w) : Type (max (max (u + 1) v) w)

A function for lifting a computation from an inner Monad to an outer Monad. Like Haskell's MonadTrans, but n does not have to be a monad transformer. Alternatively, an implementation of MonadLayer without layerInvmap (so far).

• monadLift {α : Type u} : m α → n α

  Lifts a value from monad m into monad n.

class MonadLiftT (m : Type u → Type v) (n : Type u → Type w) : Type (max (max (u + 1) v) w)

The reflexive-transitive closure of MonadLift. monadLift is used to transitively lift monadic computations such as StateT.get or StateT.put s. Corresponds to Haskell's MonadLift.

• monadLift {α : Type u} : m α → n α

  Lifts a value from monad m into monad n.

@[reducible, inline]

abbrev liftM {m : Type u_1 → Type u_2} {n : Type u_1 → Type u_3} [self : MonadLiftT m n] {α : Type u_1} : m α → n α

Lifts a value from monad m into monad n.

Equations

• @liftM = @monadLift

@[always_inline]

instance instMonadLiftTOfMonadLift (m : Type u_1 → Type u_2) (n : Type u_1 → Type u_3) (o : Type u_1 → Type u_4) [MonadLift n o] [MonadLiftT m n] : MonadLiftT m o

Equations

• instMonadLiftTOfMonadLift m n o = { monadLift := fun {α : Type ?u.38} (x : m α) => MonadLift.monadLift (monadLift x) }

instance instMonadLiftT (m : Type u_1 → Type u_2) : MonadLiftT m m

Equations

• instMonadLiftT m = { monadLift := fun {α : Type ?u.14} (x : m α) => x }

class MonadEval (m : semiOutParam (Type u → Type v)) (n : Type u → Type w) : Type (max (max (u + 1) v) w)

Typeclass used for adapting monads. This is similar to MonadLift, but instances are allowed to make use of default state for the purpose of synthesizing such an instance, if necessary. Every MonadLift instance gives a MonadEval instance.

The purpose of this class is for the #eval command, which looks for a MonadEval m CommandElabM or MonadEval m IO instance.

• monadEval {α : Type u} : m α → n α

  Evaluates a value from monad m into monad n.

instance instMonadEvalOfMonadLift {m : Type u_1 → Type u_2} {n : Type u_1 → Type u_3} [MonadLift m n] : MonadEval m n

Equations

• instMonadEvalOfMonadLift = { monadEval := fun {α : Type ?u.25} => MonadLift.monadLift }

class MonadEvalT (m : Type u → Type v) (n : Type u → Type w) : Type (max (max (u + 1) v) w)

The transitive closure of MonadEval.

• monadEval {α : Type u} : m α → n α

  Evaluates a value from monad m into monad n.

instance instMonadEvalTOfMonadEval (m : Type u_1 → Type u_2) (n : Type u_1 → Type u_3) (o : Type u_1 → Type u_4) [MonadEval n o] [MonadEvalT m n] : MonadEvalT m o

Equations

• instMonadEvalTOfMonadEval m n o = { monadEval := fun {α : Type ?u.38} (x : m α) => MonadEval.monadEval (MonadEvalT.monadEval x) }

instance instMonadEvalT (m : Type u_1 → Type u_2) : MonadEvalT m m

Equations

• instMonadEvalT m = { monadEval := fun {α : Type ?u.14} (x : m α) => x }

class MonadFunctor (m : semiOutParam (Type u → Type v)) (n : Type u → Type w) : Type (max (max (u + 1) v) w)

A functor in the category of monads. Can be used to lift monad-transforming functions. Based on MFunctor from the pipes Haskell package, but not restricted to monad transformers. Alternatively, an implementation of MonadTransFunctor.

• monadMap {α : Type u} : ({β : Type u} → m β → m β) → n α → n α

  Lifts a monad morphism f : {β : Type u} → m β → m β to monadMap f : {α : Type u} → n α → n α.

class MonadFunctorT (m : Type u → Type v) (n : Type u → Type w) : Type (max (max (u + 1) v) w)

The reflexive-transitive closure of MonadFunctor. monadMap is used to transitively lift Monad morphisms.

• monadMap {α : Type u} : ({β : Type u} → m β → m β) → n α → n α

  Lifts a monad morphism f : {β : Type u} → m β → m β to monadMap f : {α : Type u} → n α → n α.

@[always_inline]

instance instMonadFunctorTOfMonadFunctor (m : Type u_1 → Type u_2) (n : Type u_1 → Type u_3) (o : Type u_1 → Type u_4) [MonadFunctor n o] [MonadFunctorT m n] : MonadFunctorT m o

Equations

• One or more equations did not get rendered due to their size.

instance monadFunctorRefl (m : Type u_1 → Type u_2) : MonadFunctorT m m

Equations

• monadFunctorRefl m = { monadMap := fun {α : Type ?u.16} (f : {β : Type ?u.16} → m β → m β) => f }

@[unbox]

inductive Except (ε : Type u) (α : Type v) : Type (max u v)

Except ε α is a type which represents either an error of type ε, or an "ok" value of type α. The error type is listed first because Except ε : Type → Type is a Monad: the pure operation is ok and the bind operation returns the first encountered error.

• error {ε : Type u} {α : Type v} : ε → Except ε α

  A failure value of type ε

• ok {ε : Type u} {α : Type v} : α → Except ε α

  A success value of type α

instance instInhabitedExcept {ε : Type u} {α : Type v} [Inhabited ε] : Inhabited (Except ε α)

Equations

• instInhabitedExcept = { default := Except.error default }

class MonadExceptOf (ε : semiOutParam (Type u)) (m : Type v → Type w) : Type (max (max u (v + 1)) w)

An implementation of Haskell's MonadError class. A MonadError ε m is a monad m with two operations:

• throw : ε → m α "throws an error" of type ε to the nearest enclosing catch block
• tryCatch (body : m α) (handler : ε → m α) : m α will catch any errors in body and pass the resulting error to handler. Errors in handler will not be caught.

The try ... catch e => ... syntax inside do blocks is sugar for the tryCatch operation.

• throw {α : Type v} : ε → m α

  throw : ε → m α "throws an error" of type ε to the nearest enclosing catch block.

• tryCatch {α : Type v} (body : m α) (handler : ε → m α) : m α

  tryCatch (body : m α) (handler : ε → m α) : m α will catch any errors in body and pass the resulting error to handler. Errors in handler will not be caught.

@[reducible, inline]

abbrev throwThe (ε : Type u) {m : Type v → Type w} [MonadExceptOf ε m] {α : Type v} (e : ε) : m α

This is the same as throw, but allows specifying the particular error type in case the monad supports throwing more than one type of error.

Equations

• throwThe ε e = MonadExceptOf.throw e

@[reducible, inline]

abbrev tryCatchThe (ε : Type u) {m : Type v → Type w} [MonadExceptOf ε m] {α : Type v} (x : m α) (handle : ε → m α) : m α

This is the same as tryCatch, but allows specifying the particular error type in case the monad supports throwing more than one type of error.

Equations

• tryCatchThe ε x handle = MonadExceptOf.tryCatch x handle

class MonadExcept (ε : outParam (Type u)) (m : Type v → Type w) : Type (max (max u (v + 1)) w)

Similar to MonadExceptOf, but ε is an outParam for convenience.

• throw {α : Type v} : ε → m α

  throw : ε → m α "throws an error" of type ε to the nearest enclosing catch block.

• tryCatch {α : Type v} : m α → (ε → m α) → m α

  tryCatch (body : m α) (handler : ε → m α) : m α will catch any errors in body and pass the resulting error to handler. Errors in handler will not be caught.

def MonadExcept.ofExcept {m : Type u_1 → Type u_2} {ε : Type u_3} {α : Type u_1} [Monad m] [MonadExcept ε m] : Except ε α → m α

"Unwraps" an Except ε α to get the α, or throws the exception otherwise.

Equations

• ofExcept (Except.ok a) = pure a
• ofExcept (Except.error e) = throw e

instance instMonadExceptOfMonadExceptOf (ε : Type u) (m : Type v → Type w) [MonadExceptOf ε m] : MonadExcept ε m

Equations

• instMonadExceptOfMonadExceptOf ε m = { throw := fun {α : Type ?u.24} => throwThe ε, tryCatch := fun {α : Type ?u.24} => tryCatchThe ε }

@[inline]

def MonadExcept.orElse {ε : Type u} {m : Type v → Type w} [MonadExcept ε m] {α : Type v} (t₁ : m α) (t₂ : Unit → m α) : m α

A MonadExcept can implement t₁ <|> t₂ as try t₁ catch _ => t₂.

Equations

• MonadExcept.orElse t₁ t₂ = tryCatch t₁ fun (x : ε) => t₂ ()

instance MonadExcept.instOrElse {ε : Type u} {m : Type v → Type w} [MonadExcept ε m] {α : Type v} : OrElse (m α)

Equations

• MonadExcept.instOrElse = { orElse := MonadExcept.orElse }

def ReaderT (ρ : Type u) (m : Type u → Type v) (α : Type u) : Type (max u v)

An implementation of Haskell's ReaderT. This is a monad transformer which equips a monad with additional read-only state, of type ρ.

Equations

• ReaderT ρ m α = (ρ → m α)

instance instInhabitedReaderT (ρ : Type u) (m : Type u → Type v) (α : Type u) [Inhabited (m α)] : Inhabited (ReaderT ρ m α)

Equations

• instInhabitedReaderT ρ m α = { default := fun (x : ρ) => default }

@[inline]

def ReaderT.run {ρ : Type u} {m : Type u → Type v} {α : Type u} (x : ReaderT ρ m α) (r : ρ) : m α

If x : ReaderT ρ m α and r : ρ, then x.run r : ρ runs the monad with the given reader state.

Equations

• x.run r = x r

instance ReaderT.instMonadLift {ρ : Type u} {m : Type u → Type v} : MonadLift m (ReaderT ρ m)

Equations

• ReaderT.instMonadLift = { monadLift := fun {α : Type ?u.21} (x : m α) (x_1 : ρ) => x }

@[always_inline]

instance ReaderT.instMonadExceptOf {ρ : Type u} {m : Type u → Type v} (ε : Type u_1) [MonadExceptOf ε m] : MonadExceptOf ε (ReaderT ρ m)

Equations

• One or more equations did not get rendered due to their size.

@[inline]

def ReaderT.read {ρ : Type u} {m : Type u → Type v} [Monad m] : ReaderT ρ m ρ

(← read) : ρ gets the read-only state of a ReaderT ρ.

Equations

• ReaderT.read = pure

@[inline]

def ReaderT.pure {ρ : Type u} {m : Type u → Type v} [Monad m] {α : Type u} (a : α) : ReaderT ρ m α

The pure operation of the ReaderT monad.

Equations

• ReaderT.pure a x✝ = pure a

@[inline]

def ReaderT.bind {ρ : Type u} {m : Type u → Type v} [Monad m] {α β : Type u} (x : ReaderT ρ m α) (f : α → ReaderT ρ m β) : ReaderT ρ m β

The bind operation of the ReaderT monad.

Equations

• x.bind f r = do let a ← x r f a r

@[always_inline]

instance ReaderT.instFunctorOfMonad {ρ : Type u} {m : Type u → Type v} [Monad m] : Functor (ReaderT ρ m)

Equations

• One or more equations did not get rendered due to their size.

@[always_inline]

instance ReaderT.instApplicativeOfMonad {ρ : Type u} {m : Type u → Type v} [Monad m] : Applicative (ReaderT ρ m)

Equations

• ReaderT.instApplicativeOfMonad = Applicative.mk

instance ReaderT.instMonad {ρ : Type u} {m : Type u → Type v} [Monad m] : Monad (ReaderT ρ m)

Equations

• ReaderT.instMonad = Monad.mk

instance ReaderT.instMonadFunctor (ρ : Type u_1) (m : Type u_1 → Type u_2) : MonadFunctor m (ReaderT ρ m)

Equations

• ReaderT.instMonadFunctor ρ m = { monadMap := fun {α : Type ?u.24} (f : {β : Type ?u.24} → m β → m β) (x : ReaderT ρ m α) (ctx : ρ) => f (x ctx) }

@[inline]

def ReaderT.adapt {ρ : Type u} {m : Type u → Type v} {ρ' α : Type u} (f : ρ' → ρ) : ReaderT ρ m α → ReaderT ρ' m α

adapt (f : ρ' → ρ) precomposes function f on the reader state of a ReaderT ρ, yielding a ReaderT ρ'.

Equations

• ReaderT.adapt f x r = x (f r)

class MonadReaderOf (ρ : semiOutParam (Type u)) (m : Type u → Type v) : Type v

An implementation of Haskell's MonadReader (sans functional dependency; see also MonadReader in this module). It does not contain local because this function cannot be lifted using monadLift. local is instead provided by the MonadWithReader class as withReader.

Note: This class can be seen as a simplification of the more "principled" definition

class MonadReaderOf (ρ : Type u) (n : Type u → Type u) where
  lift {α : Type u} : ({m : Type u → Type u} → [Monad m] → ReaderT ρ m α) → n α

• read : m ρ

  (← read) : ρ reads the state out of monad m.

@[inline]

def readThe (ρ : Type u) {m : Type u → Type v} [MonadReaderOf ρ m] : m ρ

Like read, but with ρ explicit. This is useful if a monad supports MonadReaderOf for multiple different types ρ.

Equations

• readThe ρ = MonadReaderOf.read

class MonadReader (ρ : outParam (Type u)) (m : Type u → Type v) : Type v

Similar to MonadReaderOf, but ρ is an outParam for convenience.

• read : m ρ

  (← read) : ρ reads the state out of monad m.

instance instMonadReaderOfMonadReaderOf (ρ : Type u) (m : Type u → Type v) [MonadReaderOf ρ m] : MonadReader ρ m

Equations

• instMonadReaderOfMonadReaderOf ρ m = { read := readThe ρ }

instance instMonadReaderOfOfMonadLift {ρ : Type u} {m : Type u → Type v} {n : Type u → Type w} [MonadLift m n] [MonadReaderOf ρ m] : MonadReaderOf ρ n

Equations

• instMonadReaderOfOfMonadLift = { read := liftM read }

instance instMonadReaderOfReaderTOfMonad {ρ : Type u} {m : Type u → Type v} [Monad m] : MonadReaderOf ρ (ReaderT ρ m)

Equations

• instMonadReaderOfReaderTOfMonad = { read := ReaderT.read }

class MonadWithReaderOf (ρ : semiOutParam (Type u)) (m : Type u → Type v) : Type (max (u + 1) v)

MonadWithReaderOf ρ adds the operation withReader : (ρ → ρ) → m α → m α. This runs the inner x : m α inside a modified context after applying the function f : ρ → ρ. In addition to ReaderT itself, this operation lifts over most monad transformers, so it allows us to apply withReader to monads deeper in the stack.

• withReader {α : Type u} : (ρ → ρ) → m α → m α

  withReader (f : ρ → ρ) (x : m α) : m α runs the inner x : m α inside a modified context after applying the function f : ρ → ρ.

@[inline]

def withTheReader (ρ : Type u) {m : Type u → Type v} [MonadWithReaderOf ρ m] {α : Type u} (f : ρ → ρ) (x : m α) : m α

Like withReader, but with ρ explicit. This is useful if a monad supports MonadWithReaderOf for multiple different types ρ.

Equations

• withTheReader ρ f x = MonadWithReaderOf.withReader f x

class MonadWithReader (ρ : outParam (Type u)) (m : Type u → Type v) : Type (max (u + 1) v)

Similar to MonadWithReaderOf, but ρ is an outParam for convenience.

• withReader {α : Type u} : (ρ → ρ) → m α → m α

  withReader (f : ρ → ρ) (x : m α) : m α runs the inner x : m α inside a modified context after applying the function f : ρ → ρ.

instance instMonadWithReaderOfMonadWithReaderOf (ρ : Type u) (m : Type u → Type v) [MonadWithReaderOf ρ m] : MonadWithReader ρ m

Equations

• instMonadWithReaderOfMonadWithReaderOf ρ m = { withReader := fun {α : Type ?u.23} => withTheReader ρ }

instance instMonadWithReaderOfOfMonadFunctor {ρ : Type u} {m n : Type u → Type v} [MonadFunctor m n] [MonadWithReaderOf ρ m] : MonadWithReaderOf ρ n

Equations

• instMonadWithReaderOfOfMonadFunctor = { withReader := fun {α : Type ?u.36} (f : ρ → ρ) => monadMap fun {β : Type ?u.36} => withTheReader ρ f }

instance instMonadWithReaderOfReaderT {ρ : Type u} {m : Type u → Type v} : MonadWithReaderOf ρ (ReaderT ρ m)

Equations

• instMonadWithReaderOfReaderT = { withReader := fun {α : Type ?u.23} (f : ρ → ρ) (x : ReaderT ρ m α) (ctx : ρ) => x (f ctx) }

class MonadStateOf (σ : semiOutParam (Type u)) (m : Type u → Type v) : Type (max (u + 1) v)

An implementation of MonadState. In contrast to the Haskell implementation, we use overlapping instances to derive instances automatically from monadLift.

• get : m σ

  (← get) : σ gets the state out of a monad m.

• set : σ → m PUnit

  set (s : σ) replaces the state with value s.

• modifyGet {α : Type u} : (σ → α × σ) → m α

  modifyGet (f : σ → α × σ) applies f to the current state, replaces the state with the return value, and returns a computed value.

  It is equivalent to do let (a, s) := f (← get); put s; pure a, but modifyGet f may be preferable because the former does not use the state linearly (without sufficient inlining).

@[reducible, inline]

abbrev getThe (σ : Type u) {m : Type u → Type v} [MonadStateOf σ m] : m σ

Like get, but with σ explicit. This is useful if a monad supports MonadStateOf for multiple different types σ.

Equations

• getThe σ = MonadStateOf.get

@[reducible, inline]

abbrev modifyThe (σ : Type u) {m : Type u → Type v} [MonadStateOf σ m] (f : σ → σ) : m PUnit

Like modify, but with σ explicit. This is useful if a monad supports MonadStateOf for multiple different types σ.

Equations

• modifyThe σ f = MonadStateOf.modifyGet fun (s : σ) => (PUnit.unit, f s)

@[reducible, inline]

abbrev modifyGetThe {α : Type u} (σ : Type u) {m : Type u → Type v} [MonadStateOf σ m] (f : σ → α × σ) : m α

Like modifyGet, but with σ explicit. This is useful if a monad supports MonadStateOf for multiple different types σ.

Equations

• modifyGetThe σ f = MonadStateOf.modifyGet f

class MonadState (σ : outParam (Type u)) (m : Type u → Type v) : Type (max (u + 1) v)

Similar to MonadStateOf, but σ is an outParam for convenience.

• get : m σ

  (← get) : σ gets the state out of a monad m.

• set : σ → m PUnit

  set (s : σ) replaces the state with value s.

• modifyGet {α : Type u} : (σ → α × σ) → m α

  modifyGet (f : σ → α × σ) applies f to the current state, replaces the state with the return value, and returns a computed value.

  It is equivalent to do let (a, s) := f (← get); put s; pure a, but modifyGet f may be preferable because the former does not use the state linearly (without sufficient inlining).

instance instMonadStateOfMonadStateOf (σ : Type u) (m : Type u → Type v) [MonadStateOf σ m] : MonadState σ m

Equations

• instMonadStateOfMonadStateOf σ m = { get := getThe σ, set := set, modifyGet := fun {α : Type ?u.25} (f : σ → α × σ) => MonadStateOf.modifyGet f }

@[inline]

def modify {σ : Type u} {m : Type u → Type v} [MonadState σ m] (f : σ → σ) : m PUnit

modify (f : σ → σ) applies the function f to the state.

It is equivalent to do set (f (← get)), but modify f may be preferable because the former does not use the state linearly (without sufficient inlining).

Equations

• modify f = modifyGet fun (s : σ) => (PUnit.unit, f s)

@[inline]

def getModify {σ : Type u} {m : Type u → Type v} [MonadState σ m] (f : σ → σ) : m σ

getModify f gets the state, applies function f, and returns the old value of the state. It is equivalent to get <* modify f but may be more efficient.

Equations

• getModify f = modifyGet fun (s : σ) => (s, f s)

@[always_inline]

instance instMonadStateOfOfMonadLift {σ : Type u} {m : Type u → Type v} {n : Type u → Type w} [MonadLift m n] [MonadStateOf σ m] : MonadStateOf σ n

Equations

• instMonadStateOfOfMonadLift = { get := liftM MonadStateOf.get, set := fun (s : σ) => liftM (set s), modifyGet := fun {α : Type ?u.38} (f : σ → α × σ) => monadLift (modifyGet f) }

inductive EStateM.Result (ε σ α : Type u) : Type u

Result ε σ α is equivalent to Except ε α × σ, but using a single combined inductive yields a more efficient data representation.

• ok {ε σ α : Type u} : α → σ → EStateM.Result ε σ α

  A success value of type α, and a new state σ.

• error {ε σ α : Type u} : ε → σ → EStateM.Result ε σ α

  A failure value of type ε, and a new state σ.

instance EStateM.instInhabitedResult {ε σ α : Type u} [Inhabited ε] [Inhabited σ] : Inhabited (EStateM.Result ε σ α)

Equations

• EStateM.instInhabitedResult = { default := EStateM.Result.error default default }

def EStateM (ε σ α : Type u) : Type u

EStateM ε σ is a combined error and state monad, equivalent to ExceptT ε (StateM σ) but more efficient.

Equations

• EStateM ε σ α = (σ → EStateM.Result ε σ α)

instance EStateM.instInhabited {ε σ α : Type u} [Inhabited ε] : Inhabited (EStateM ε σ α)

Equations

• EStateM.instInhabited = { default := fun (s : σ) => EStateM.Result.error default s }

@[inline]

def EStateM.pure {ε σ α : Type u} (a : α) : EStateM ε σ α

The pure operation of the EStateM monad.

Equations

• EStateM.pure a s = EStateM.Result.ok a s

@[inline]

def EStateM.set {ε σ : Type u} (s : σ) : EStateM ε σ PUnit

The set operation of the EStateM monad.

Equations

• EStateM.set s x✝ = EStateM.Result.ok PUnit.unit s

@[inline]

def EStateM.get {ε σ : Type u} : EStateM ε σ σ

The get operation of the EStateM monad.

Equations

• EStateM.get s = EStateM.Result.ok s s

@[inline]

def EStateM.modifyGet {ε σ α : Type u} (f : σ → α × σ) : EStateM ε σ α

The modifyGet operation of the EStateM monad.

Equations

• EStateM.modifyGet f s = match f s with | (a, s) => EStateM.Result.ok a s

@[inline]

def EStateM.throw {ε σ α : Type u} (e : ε) : EStateM ε σ α

The throw operation of the EStateM monad.

Equations

• EStateM.throw e s = EStateM.Result.error e s

class EStateM.Backtrackable (δ : outParam (Type u)) (σ : Type u) : Type u

Auxiliary instance for saving/restoring the "backtrackable" part of the state. Here σ is the state, and δ is some subpart of it, and we have a getter and setter for it (a "lens" in the Haskell terminology).

• save : σ → δ

  save s : δ retrieves a copy of the backtracking state out of the state.

• restore : σ → δ → σ

  restore (s : σ) (x : δ) : σ applies the old backtracking state x to the state s to get a backtracked state s'.

@[inline]

def EStateM.tryCatch {ε σ δ : Type u} [EStateM.Backtrackable δ σ] {α : Type u} (x : EStateM ε σ α) (handle : ε → EStateM ε σ α) : EStateM ε σ α

Implementation of tryCatch for EStateM where the state is Backtrackable.

Equations

• x.tryCatch handle s = match x s with | EStateM.Result.error e s_1 => handle e (EStateM.Backtrackable.restore s_1 (EStateM.Backtrackable.save s)) | ok => ok

@[inline]

def EStateM.orElse {ε σ α δ : Type u} [EStateM.Backtrackable δ σ] (x₁ : EStateM ε σ α) (x₂ : Unit → EStateM ε σ α) : EStateM ε σ α

Implementation of orElse for EStateM where the state is Backtrackable.

Equations

• x₁.orElse x₂ s = match x₁ s with | EStateM.Result.error a s_1 => x₂ () (EStateM.Backtrackable.restore s_1 (EStateM.Backtrackable.save s)) | ok => ok

@[inline]

def EStateM.adaptExcept {ε σ α ε' : Type u} (f : ε → ε') (x : EStateM ε σ α) : EStateM ε' σ α

Map the exception type of a EStateM ε σ α by a function f : ε → ε'.

Equations

• EStateM.adaptExcept f x s = match x s with | EStateM.Result.error e s => EStateM.Result.error (f e) s | EStateM.Result.ok a s => EStateM.Result.ok a s

@[inline]

def EStateM.bind {ε σ α β : Type u} (x : EStateM ε σ α) (f : α → EStateM ε σ β) : EStateM ε σ β

The bind operation of the EStateM monad.

Equations

• x.bind f s = match x s with | EStateM.Result.ok a s => f a s | EStateM.Result.error e s => EStateM.Result.error e s

@[inline]

def EStateM.map {ε σ α β : Type u} (f : α → β) (x : EStateM ε σ α) : EStateM ε σ β

The map operation of the EStateM monad.

Equations

• EStateM.map f x s = match x s with | EStateM.Result.ok a s => EStateM.Result.ok (f a) s | EStateM.Result.error e s => EStateM.Result.error e s

@[inline]

def EStateM.seqRight {ε σ α β : Type u} (x : EStateM ε σ α) (y : Unit → EStateM ε σ β) : EStateM ε σ β

The seqRight operation of the EStateM monad.

Equations

• x.seqRight y s = match x s with | EStateM.Result.ok a s => y () s | EStateM.Result.error e s => EStateM.Result.error e s

@[always_inline]

instance EStateM.instMonad {ε σ : Type u} : Monad (EStateM ε σ)

Equations

• EStateM.instMonad = Monad.mk

instance EStateM.instOrElseOfBacktrackable {ε σ α δ : Type u} [EStateM.Backtrackable δ σ] : OrElse (EStateM ε σ α)

Equations

• EStateM.instOrElseOfBacktrackable = { orElse := EStateM.orElse }

instance EStateM.instMonadStateOf {ε σ : Type u} : MonadStateOf σ (EStateM ε σ)

Equations

• EStateM.instMonadStateOf = { get := EStateM.get, set := EStateM.set, modifyGet := fun {α : Type ?u.17} => EStateM.modifyGet }

instance EStateM.instMonadExceptOfOfBacktrackable {ε σ δ : Type u} [EStateM.Backtrackable δ σ] : MonadExceptOf ε (EStateM ε σ)

Equations

• EStateM.instMonadExceptOfOfBacktrackable = { throw := fun {α : Type ?u.30} => EStateM.throw, tryCatch := fun {α : Type ?u.30} => EStateM.tryCatch }

@[inline]

def EStateM.run {ε σ α : Type u} (x : EStateM ε σ α) (s : σ) : EStateM.Result ε σ α

Execute an EStateM on initial state s to get a Result.

Equations

• x.run s = x s

@[inline]

def EStateM.run' {ε σ α : Type u} (x : EStateM ε σ α) (s : σ) : Option α

Execute an EStateM on initial state s for the returned value α. If the monadic action throws an exception, returns none instead.

Equations

• x.run' s = match x.run s with | EStateM.Result.ok v a => some v | EStateM.Result.error a a_1 => none

@[inline]

def EStateM.dummySave {σ : Type u} : σ → PUnit

The save implementation for Backtrackable PUnit σ.

Equations

• EStateM.dummySave x✝ = PUnit.unit

@[inline]

def EStateM.dummyRestore {σ : Type u} : σ → PUnit → σ

The restore implementation for Backtrackable PUnit σ.

Equations

• EStateM.dummyRestore s x✝ = s

instance EStateM.nonBacktrackable {σ : Type u} : EStateM.Backtrackable PUnit σ

Dummy default instance. This makes every σ trivially "backtrackable" by doing nothing on backtrack. Because this is the first declared instance of Backtrackable _ σ, it will be picked only if there are no other Backtrackable _ σ instances registered.

Equations

• EStateM.nonBacktrackable = { save := EStateM.dummySave, restore := EStateM.dummyRestore }

class Hashable (α : Sort u) : Sort (max 1 u)

A class for types that can be hashed into a UInt64.

• hash : α → UInt64

  Hashes the value a : α into a UInt64.

@[extern lean_uint64_mix_hash]

opaque mixHash (u₁ u₂ : UInt64) : UInt64

An opaque hash mixing operation, used to implement hashing for tuples.

instance instHashableSubtype {α : Sort u_1} [Hashable α] {p : α → Prop} : Hashable (Subtype p)

Equations

• instHashableSubtype = { hash := fun (a : Subtype p) => hash a.val }

@[extern lean_string_hash]

opaque String.hash (s : String) : UInt64

An opaque string hash function.

instance instHashableString : Hashable String

Equations

• instHashableString = { hash := String.hash }

@[implemented_by Lean.Name.hash._override]

noncomputable def Lean.Name.hash : Lean.Name → UInt64

A hash function for names, which is stored inside the name itself as a computed field.

Equations

• Lean.Name.anonymous.hash = UInt64.ofNatCore 1723 Lean.Name.hash.proof_3
• (p.str s).hash = mixHash p.hash s.hash
• (p.num v).hash = mixHash p.hash (if h : v < UInt64.size then UInt64.ofNatCore v h else UInt64.ofNatCore 17 Lean.Name.hash.proof_4)

inductive Lean.Name : Type

Hierarchical names consist of a sequence of components, each of which is either a string or numeric, that are written separated by dots (.).

Hierarchical names are used to name declarations and for creating unique identifiers for free variables and metavariables.

You can create hierarchical names using a backtick:

`Lean.Meta.whnf

It is short for .str (.str (.str .anonymous "Lean") "Meta") "whnf".

You can use double backticks to request Lean to statically check whether the name corresponds to a Lean declaration in scope.

``Lean.Meta.whnf

If the name is not in scope, Lean will report an error.

There are two ways to convert a String to a Name:

1. Name.mkSimple creates a name with a single string component.

2. String.toName first splits the string into its dot-separated components, and then creates a hierarchical name.

• anonymous : Lean.Name

  The "anonymous" name.

• str (pre : Lean.Name) (str : String) : Lean.Name

  A string name. The name Lean.Meta.run is represented at

  .str (.str (.str .anonymous "Lean") "Meta") "run"
  

• num (pre : Lean.Name) (i : Nat) : Lean.Name

  A numerical name. This kind of name is used, for example, to create hierarchical names for free variables and metavariables. The identifier _uniq.231 is represented as

  .num (.str .anonymous "_uniq") 231
  

instance Lean.instInhabitedName : Inhabited Lean.Name

Equations

• Lean.instInhabitedName = { default := Lean.Name.anonymous }

instance Lean.instHashableName : Hashable Lean.Name

Equations

• Lean.instHashableName = { hash := Lean.Name.hash }

@[reducible, inline, export lean_name_mk_string]

abbrev Lean.Name.mkStr (p : Lean.Name) (s : String) : Lean.Name

.str p s is now the preferred form.

Equations

• p.mkStr s = p.str s

@[reducible, inline, export lean_name_mk_numeral]

abbrev Lean.Name.mkNum (p : Lean.Name) (v : Nat) : Lean.Name

.num p v is now the preferred form.

Equations

• p.mkNum v = p.num v

@[reducible, inline]

abbrev Lean.Name.mkSimple (s : String) : Lean.Name

Converts a String to a Name without performing any parsing. mkSimple s is short for .str .anonymous s.

This means that mkSimple "a.b" is the name «a.b», not a.b.

Equations

• Lean.Name.mkSimple s = Lean.Name.anonymous.str s

@[reducible]

def Lean.Name.mkStr1 (s₁ : String) : Lean.Name

Make name s₁

Equations

• Lean.Name.mkStr1 s₁ = Lean.Name.anonymous.str s₁

@[reducible]

def Lean.Name.mkStr2 (s₁ s₂ : String) : Lean.Name

Make name s₁.s₂

Equations

• Lean.Name.mkStr2 s₁ s₂ = (Lean.Name.anonymous.str s₁).str s₂

@[reducible]

def Lean.Name.mkStr3 (s₁ s₂ s₃ : String) : Lean.Name

Make name s₁.s₂.s₃

Equations

• Lean.Name.mkStr3 s₁ s₂ s₃ = ((Lean.Name.anonymous.str s₁).str s₂).str s₃

@[reducible]

def Lean.Name.mkStr4 (s₁ s₂ s₃ s₄ : String) : Lean.Name

Make name s₁.s₂.s₃.s₄

Equations

• Lean.Name.mkStr4 s₁ s₂ s₃ s₄ = (((Lean.Name.anonymous.str s₁).str s₂).str s₃).str s₄

@[reducible]

def Lean.Name.mkStr5 (s₁ s₂ s₃ s₄ s₅ : String) : Lean.Name

Make name s₁.s₂.s₃.s₄.s₅

Equations

• Lean.Name.mkStr5 s₁ s₂ s₃ s₄ s₅ = ((((Lean.Name.anonymous.str s₁).str s₂).str s₃).str s₄).str s₅

@[reducible]

def Lean.Name.mkStr6 (s₁ s₂ s₃ s₄ s₅ s₆ : String) : Lean.Name

Make name s₁.s₂.s₃.s₄.s₅.s₆

Equations

• Lean.Name.mkStr6 s₁ s₂ s₃ s₄ s₅ s₆ = (((((Lean.Name.anonymous.str s₁).str s₂).str s₃).str s₄).str s₅).str s₆

@[reducible]

def Lean.Name.mkStr7 (s₁ s₂ s₃ s₄ s₅ s₆ s₇ : String) : Lean.Name

Make name s₁.s₂.s₃.s₄.s₅.s₆.s₇

Equations

• Lean.Name.mkStr7 s₁ s₂ s₃ s₄ s₅ s₆ s₇ = ((((((Lean.Name.anonymous.str s₁).str s₂).str s₃).str s₄).str s₅).str s₆).str s₇

@[reducible]

def Lean.Name.mkStr8 (s₁ s₂ s₃ s₄ s₅ s₆ s₇ s₈ : String) : Lean.Name

Make name s₁.s₂.s₃.s₄.s₅.s₆.s₇.s₈

Equations

• Lean.Name.mkStr8 s₁ s₂ s₃ s₄ s₅ s₆ s₇ s₈ = (((((((Lean.Name.anonymous.str s₁).str s₂).str s₃).str s₄).str s₅).str s₆).str s₇).str s₈

@[extern lean_name_eq]

def Lean.Name.beq : Lean.Name → Lean.Name → Bool

(Boolean) equality comparator for names.

Equations

• Lean.Name.anonymous.beq Lean.Name.anonymous = true
• (p₁.str s₁).beq (p₂.str s₂) = (s₁ == s₂ && p₁.beq p₂)
• (p₁.num n₁).beq (p₂.num n₂) = (n₁ == n₂ && p₁.beq p₂)
• x✝¹.beq x✝ = false

instance Lean.Name.instBEq : BEq Lean.Name

Equations

• Lean.Name.instBEq = { beq := Lean.Name.beq }

def Lean.Name.appendCore : Lean.Name → Lean.Name → Lean.Name

This function does not have special support for macro scopes. See Name.append.

Equations

• x✝.appendCore Lean.Name.anonymous = x✝
• x✝.appendCore (p.str s) = (x✝.appendCore p).str s
• x✝.appendCore (p.num d) = (x✝.appendCore p).num d

def Lean.defaultMaxRecDepth : Nat

The default maximum recursion depth. This is adjustable using the maxRecDepth option.

Equations

• Lean.defaultMaxRecDepth = 512

def Lean.maxRecDepthErrorMessage : String

The message to display on stack overflow.

Equations

• Lean.maxRecDepthErrorMessage = "maximum recursion depth has been reached\nuse `set_option maxRecDepth <num>` to increase limit\nuse `set_option diagnostics true` to get diagnostic information"

Syntax

inductive Lean.SourceInfo : Type

Source information of tokens.

• original (leading : Substring) (pos : String.Pos) (trailing : Substring) (endPos : String.Pos) : Lean.SourceInfo

  Token from original input with whitespace and position information. leading will be inferred after parsing by Syntax.updateLeading. During parsing, it is not at all clear what the preceding token was, especially with backtracking.

• synthetic (pos endPos : String.Pos) (canonical : Bool := false) : Lean.SourceInfo

  Synthesized syntax (e.g. from a quotation) annotated with a span from the original source. In the delaborator, we "misuse" this constructor to store synthetic positions identifying subterms.

  The canonical flag on synthetic syntax is enabled for syntax that is not literally part of the original input syntax but should be treated "as if" the user really wrote it for the purpose of hovers and error messages. This is usually used on identifiers, to connect the binding site to the user's original syntax even if the name of the identifier changes during expansion, as well as on tokens where we will attach targeted messages.

  The syntax token%$stx in a syntax quotation will annotate the token token with the span from stx and also mark it as canonical.

  As a rough guide, a macro expansion should only use a given piece of input syntax in a single canonical token, although this is sometimes violated when the same identifier is used to declare two binders, as in the macro expansion for dependent if:

  `(if $h : $cond then $t else $e) ~>
  `(dite $cond (fun $h => $t) (fun $h => $t))
  

  In these cases if the user hovers over h they will see information about both binding sites.

• none : Lean.SourceInfo

  Synthesized token without position information.

instance Lean.instInhabitedSourceInfo : Inhabited Lean.SourceInfo

Equations

• Lean.instInhabitedSourceInfo = { default := Lean.SourceInfo.none }

def Lean.SourceInfo.getPos? (info : Lean.SourceInfo) (canonicalOnly : Bool := false) : Option String.Pos

Gets the position information from a SourceInfo, if available. If canonicalOnly is true, then .synthetic syntax with canonical := false will also return none.

Equations

• (Lean.SourceInfo.original leading pos trailing endPos).getPos? canonicalOnly = some pos
• (Lean.SourceInfo.synthetic pos endPos true).getPos? canonicalOnly = some pos
• (Lean.SourceInfo.synthetic pos endPos canonical).getPos? = some pos
• info.getPos? canonicalOnly = none

def Lean.SourceInfo.getTailPos? (info : Lean.SourceInfo) (canonicalOnly : Bool := false) : Option String.Pos

Gets the end position information from a SourceInfo, if available. If canonicalOnly is true, then .synthetic syntax with canonical := false will also return none.

Equations

• (Lean.SourceInfo.original leading pos trailing endPos).getTailPos? canonicalOnly = some endPos
• (Lean.SourceInfo.synthetic pos endPos true).getTailPos? canonicalOnly = some endPos
• (Lean.SourceInfo.synthetic pos endPos canonical).getTailPos? = some endPos
• info.getTailPos? canonicalOnly = none

def Lean.SourceInfo.getTrailing? (info : Lean.SourceInfo) : Option Substring

Gets the substring representing the trailing whitespace of a SourceInfo, if available.

Equations

• (Lean.SourceInfo.original leading pos trailing endPos).getTrailing? = some trailing
• info.getTrailing? = none

def Lean.SourceInfo.getTrailingTailPos? (info : Lean.SourceInfo) (canonicalOnly : Bool := false) : Option String.Pos

Gets the end position information of the trailing whitespace of a SourceInfo, if available. If canonicalOnly is true, then .synthetic syntax with canonical := false will also return none.

Equations

• info.getTrailingTailPos? canonicalOnly = match info.getTrailing? with | some trailing => some trailing.stopPos | none => info.getTailPos? canonicalOnly

@[reducible, inline]

abbrev Lean.SyntaxNodeKind : Type

A SyntaxNodeKind classifies Syntax.node values. It is an abbreviation for Name, and you can use name literals to construct SyntaxNodeKinds, but they need not refer to declarations in the environment. Conventionally, a SyntaxNodeKind will correspond to the Parser or ParserDesc declaration that parses it.

Equations

• Lean.SyntaxNodeKind = Lean.Name

Syntax AST

inductive Lean.Syntax.Preresolved : Type

Binding information resolved and stored at compile time of a syntax quotation. Note: We do not statically know whether a syntax expects a namespace or term name, so a Syntax.ident may contain both preresolution kinds.

• namespace (ns : Lean.Name) : Lean.Syntax.Preresolved

  A potential namespace reference

• decl (n : Lean.Name) (fields : List String) : Lean.Syntax.Preresolved

  A potential global constant or section variable reference, with additional field accesses

inductive Lean.Syntax : Type

Syntax objects used by the parser, macro expander, delaborator, etc.

• missing : Lean.Syntax

  A missing syntax corresponds to a portion of the syntax tree that is missing because of a parse error. The indexing operator on Syntax also returns missing for indexing out of bounds.

• node (info : Lean.SourceInfo) (kind : Lean.SyntaxNodeKind) (args : Array Lean.Syntax) : Lean.Syntax

  Node in the syntax tree.

  The info field is used by the delaborator to store the position of the subexpression corresponding to this node. The parser sets the info field to none. The parser sets the info field to none, with position retrieval continuing recursively. Nodes created by quotations use the result from SourceInfo.fromRef so that they are marked as synthetic even when the leading/trailing token is not. The delaborator uses the info field to store the position of the subexpression corresponding to this node.

  (Remark: the node constructor did not have an info field in previous versions. This caused a bug in the interactive widgets, where the popup for a + b was the same as for a. The delaborator used to associate subexpressions with pretty-printed syntax by setting the (string) position of the first atom/identifier to the (expression) position of the subexpression. For example, both a and a + b have the same first identifier, and so their infos got mixed up.)

• atom (info : Lean.SourceInfo) (val : String) : Lean.Syntax

  An atom corresponds to a keyword or piece of literal unquoted syntax. These correspond to quoted strings inside syntax declarations. For example, in (x + y), "(", "+" and ")" are atom and x and y are ident.

• ident (info : Lean.SourceInfo) (rawVal : Substring) (val : Lean.Name) (preresolved : List Lean.Syntax.Preresolved) : Lean.Syntax

  An ident corresponds to an identifier as parsed by the ident or rawIdent parsers.

  • rawVal is the literal substring from the input file
  • val is the parsed identifier (with hygiene)
  • preresolved is the list of possible declarations this could refer to

def Lean.Syntax.node1 (info : Lean.SourceInfo) (kind : Lean.SyntaxNodeKind) (a₁ : Lean.Syntax) : Lean.Syntax

Create syntax node with 1 child

Equations

• Lean.Syntax.node1 info kind a₁ = Lean.Syntax.node info kind #[a₁]

def Lean.Syntax.node2 (info : Lean.SourceInfo) (kind : Lean.SyntaxNodeKind) (a₁ a₂ : Lean.Syntax) : Lean.Syntax

Create syntax node with 2 children

Equations

• Lean.Syntax.node2 info kind a₁ a₂ = Lean.Syntax.node info kind #[a₁, a₂]

def Lean.Syntax.node3 (info : Lean.SourceInfo) (kind : Lean.SyntaxNodeKind) (a₁ a₂ a₃ : Lean.Syntax) : Lean.Syntax

Create syntax node with 3 children

Equations

• Lean.Syntax.node3 info kind a₁ a₂ a₃ = Lean.Syntax.node info kind #[a₁, a₂, a₃]

def Lean.Syntax.node4 (info : Lean.SourceInfo) (kind : Lean.SyntaxNodeKind) (a₁ a₂ a₃ a₄ : Lean.Syntax) : Lean.Syntax

Create syntax node with 4 children

Equations

• Lean.Syntax.node4 info kind a₁ a₂ a₃ a₄ = Lean.Syntax.node info kind #[a₁, a₂, a₃, a₄]

def Lean.Syntax.node5 (info : Lean.SourceInfo) (kind : Lean.SyntaxNodeKind) (a₁ a₂ a₃ a₄ a₅ : Lean.Syntax) : Lean.Syntax

Create syntax node with 5 children

Equations

• Lean.Syntax.node5 info kind a₁ a₂ a₃ a₄ a₅ = Lean.Syntax.node info kind #[a₁, a₂, a₃, a₄, a₅]

def Lean.Syntax.node6 (info : Lean.SourceInfo) (kind : Lean.SyntaxNodeKind) (a₁ a₂ a₃ a₄ a₅ a₆ : Lean.Syntax) : Lean.Syntax

Create syntax node with 6 children

Equations

• Lean.Syntax.node6 info kind a₁ a₂ a₃ a₄ a₅ a₆ = Lean.Syntax.node info kind #[a₁, a₂, a₃, a₄, a₅, a₆]

def Lean.Syntax.node7 (info : Lean.SourceInfo) (kind : Lean.SyntaxNodeKind) (a₁ a₂ a₃ a₄ a₅ a₆ a₇ : Lean.Syntax) : Lean.Syntax

Create syntax node with 7 children

Equations

• Lean.Syntax.node7 info kind a₁ a₂ a₃ a₄ a₅ a₆ a₇ = Lean.Syntax.node info kind #[a₁, a₂, a₃, a₄, a₅, a₆, a₇]

def Lean.Syntax.node8 (info : Lean.SourceInfo) (kind : Lean.SyntaxNodeKind) (a₁ a₂ a₃ a₄ a₅ a₆ a₇ a₈ : Lean.Syntax) : Lean.Syntax

Create syntax node with 8 children

Equations

• Lean.Syntax.node8 info kind a₁ a₂ a₃ a₄ a₅ a₆ a₇ a₈ = Lean.Syntax.node info kind #[a₁, a₂, a₃, a₄, a₅, a₆, a₇, a₈]

def Lean.SyntaxNodeKinds : Type

SyntaxNodeKinds is a set of SyntaxNodeKind (implemented as a list).

Equations

• Lean.SyntaxNodeKinds = List Lean.SyntaxNodeKind

structure Lean.TSyntax (ks : Lean.SyntaxNodeKinds) : Type

A Syntax value of one of the given syntax kinds. Note that while syntax quotations produce/expect TSyntax values of the correct kinds, this is not otherwise enforced and can easily be circumvented by direct use of the constructor. The namespace TSyntax.Compat can be opened to expose a general coercion from Syntax to any TSyntax ks for porting older code.

• raw : Lean.Syntax

  The underlying Syntax value.

instance Lean.instInhabitedSyntax : Inhabited Lean.Syntax

Equations

• Lean.instInhabitedSyntax = { default := Lean.Syntax.missing }

instance Lean.instInhabitedTSyntax {ks : Lean.SyntaxNodeKinds} : Inhabited (Lean.TSyntax ks)

Equations

• Lean.instInhabitedTSyntax = { default := { raw := default } }

Builtin kinds

@[reducible, inline]

abbrev Lean.choiceKind : Lean.SyntaxNodeKind

The choice kind is used when a piece of syntax has multiple parses, and the determination of which to use is deferred until typing information is available.

Equations

• Lean.choiceKind = `choice

@[reducible, inline]

abbrev Lean.nullKind : Lean.SyntaxNodeKind

The null kind is used for raw list parsers like many.

Equations

• Lean.nullKind = `null

@[reducible, inline]

abbrev Lean.groupKind : Lean.SyntaxNodeKind

The group kind is by the group parser, to avoid confusing with the null kind when used inside optional.

Equations

• Lean.groupKind = `group

@[reducible, inline]

abbrev Lean.identKind : Lean.SyntaxNodeKind

ident is not actually used as a node kind, but it is returned by getKind in the ident case so that things that handle different node kinds can also handle ident.

Equations

• Lean.identKind = `ident

@[reducible, inline]

abbrev Lean.strLitKind : Lean.SyntaxNodeKind

str is the node kind of string literals like "foo".

Equations

• Lean.strLitKind = `str

@[reducible, inline]

abbrev Lean.charLitKind : Lean.SyntaxNodeKind

char is the node kind of character literals like 'A'.

Equations

• Lean.charLitKind = `char

@[reducible, inline]

abbrev Lean.numLitKind : Lean.SyntaxNodeKind

num is the node kind of number literals like 42.

Equations

• Lean.numLitKind = `num

@[reducible, inline]

abbrev Lean.scientificLitKind : Lean.SyntaxNodeKind

scientific is the node kind of floating point literals like 1.23e-3.

Equations

• Lean.scientificLitKind = `scientific

@[reducible, inline]

abbrev Lean.nameLitKind : Lean.SyntaxNodeKind

name is the node kind of name literals like `foo.

Equations

• Lean.nameLitKind = `name

@[reducible, inline]

abbrev Lean.fieldIdxKind : Lean.SyntaxNodeKind

fieldIdx is the node kind of projection indices like the 2 in x.2.

Equations

• Lean.fieldIdxKind = `fieldIdx

@[reducible, inline]

abbrev Lean.hygieneInfoKind : Lean.SyntaxNodeKind

hygieneInfo is the node kind of the hygieneInfo parser, which is an "invisible token" which captures the hygiene information at the current point without parsing anything.

They can be used to generate identifiers (with Lean.HygieneInfo.mkIdent) as if they were introduced by the calling context, not the called macro.

Equations

• Lean.hygieneInfoKind = `hygieneInfo

@[reducible, inline]

abbrev Lean.interpolatedStrLitKind : Lean.SyntaxNodeKind

interpolatedStrLitKind is the node kind of interpolated string literal fragments like "value = { and }" in s!"value = {x}".

Equations

• Lean.interpolatedStrLitKind = `interpolatedStrLitKind

@[reducible, inline]

abbrev Lean.interpolatedStrKind : Lean.SyntaxNodeKind

interpolatedStrKind is the node kind of an interpolated string literal like "value = {x}" in s!"value = {x}".

Equations

• Lean.interpolatedStrKind = `interpolatedStrKind

@[inline]

def Lean.mkNode (k : Lean.SyntaxNodeKind) (args : Array Lean.Syntax) : Lean.TSyntax k

Creates an info-less node of the given kind and children.

Equations

• Lean.mkNode k args = { raw := Lean.Syntax.node Lean.SourceInfo.none k args }

@[inline]

def Lean.mkNullNode (args : Array Lean.Syntax := #[]) : Lean.Syntax

Creates an info-less nullKind node with the given children, if any.

Equations

• Lean.mkNullNode args = (Lean.mkNode Lean.nullKind args).raw

def Lean.Syntax.getKind (stx : Lean.Syntax) : Lean.SyntaxNodeKind

Gets the kind of a Syntax node. For non-node syntax, we use "pseudo kinds": identKind for ident, missing for missing, and the atom's string literal for atoms.

Equations

• (Lean.Syntax.node info k args).getKind = k
• Lean.Syntax.missing.getKind = `missing
• (Lean.Syntax.atom info v).getKind = Lean.Name.mkSimple v
• (Lean.Syntax.ident info rawVal val preresolved).getKind = Lean.identKind

def Lean.Syntax.setKind (stx : Lean.Syntax) (k : Lean.SyntaxNodeKind) : Lean.Syntax

Changes the kind at the root of a Syntax node to k. Does nothing for non-node nodes.

Equations

• (Lean.Syntax.node info kind args).setKind k = Lean.Syntax.node info k args
• stx.setKind k = stx

def Lean.Syntax.isOfKind (stx : Lean.Syntax) (k : Lean.SyntaxNodeKind) : Bool

Is this a syntax with node kind k?

Equations

• stx.isOfKind k = (stx.getKind == k)

def Lean.Syntax.getArg (stx : Lean.Syntax) (i : Nat) : Lean.Syntax

Gets the i'th argument of the syntax node. This can also be written stx[i]. Returns missing if i is out of range.

Equations

• (Lean.Syntax.node info kind args).getArg i = args.getD i Lean.Syntax.missing
• stx.getArg i = Lean.Syntax.missing

def Lean.Syntax.getArgs (stx : Lean.Syntax) : Array Lean.Syntax

Gets the list of arguments of the syntax node, or #[] if it's not a node.

Equations

• (Lean.Syntax.node info kind args).getArgs = args
• stx.getArgs = #[]

def Lean.Syntax.getNumArgs (stx : Lean.Syntax) : Nat

Gets the number of arguments of the syntax node, or 0 if it's not a node.

Equations

• (Lean.Syntax.node info kind args).getNumArgs = args.size
• stx.getNumArgs = 0

def Lean.Syntax.getOptional? (stx : Lean.Syntax) : Option Lean.Syntax

Assuming stx was parsed by optional, returns the enclosed syntax if it parsed something and none otherwise.

Equations

• (Lean.Syntax.node info kind args).getOptional? = match kind == Lean.nullKind && args.size == 1 with | true => some (args.get! 0) | false => none
• stx.getOptional? = none

def Lean.Syntax.isMissing : Lean.Syntax → Bool

Is this syntax .missing?

Equations

• Lean.Syntax.missing.isMissing = true
• x✝.isMissing = false

def Lean.Syntax.isNodeOf (stx : Lean.Syntax) (k : Lean.SyntaxNodeKind) (n : Nat) : Bool

Is this syntax a node with kind k?

Equations

• stx.isNodeOf k n = (stx.isOfKind k && stx.getNumArgs == n)

def Lean.Syntax.isIdent : Lean.Syntax → Bool

stx.isIdent is true iff stx is an identifier.

Equations

• (Lean.Syntax.ident info rawVal val preresolved).isIdent = true
• x✝.isIdent = false

def Lean.Syntax.getId : Lean.Syntax → Lean.Name

If this is an ident, return the parsed value, else .anonymous.

Equations

• (Lean.Syntax.ident info rawVal val preresolved).getId = val
• x✝.getId = Lean.Name.anonymous

partial def Lean.Syntax.getHeadInfo? : Lean.Syntax → Option Lean.SourceInfo

Retrieve the left-most node or leaf's info in the Syntax tree.

partial def Lean.Syntax.getHeadInfo?.loop (args : Array Lean.Syntax) (i : Nat) : Option Lean.SourceInfo

def Lean.Syntax.getHeadInfo (stx : Lean.Syntax) : Lean.SourceInfo

Retrieve the left-most leaf's info in the Syntax tree, or none if there is no token.

Equations

• stx.getHeadInfo = match stx.getHeadInfo? with | some info => info | none => Lean.SourceInfo.none

def Lean.Syntax.getPos? (stx : Lean.Syntax) (canonicalOnly : Bool := false) : Option String.Pos

Get the starting position of the syntax, if possible. If canonicalOnly is true, non-canonical synthetic nodes are treated as not carrying position information.

Equations

• stx.getPos? canonicalOnly = stx.getHeadInfo.getPos? canonicalOnly

partial def Lean.Syntax.getTailPos? (stx : Lean.Syntax) (canonicalOnly : Bool := false) : Option String.Pos

Get the ending position of the syntax, if possible. If canonicalOnly is true, non-canonical synthetic nodes are treated as not carrying position information.

partial def Lean.Syntax.getTailPos?.loop (canonicalOnly : Bool := false) (args : Array Lean.Syntax) (i : Nat) : Option String.Pos

structure Lean.Syntax.SepArray (sep : String) : Type

An array of syntax elements interspersed with separators. Can be coerced to/from Array Syntax to automatically remove/insert the separators.

• elemsAndSeps : Array Lean.Syntax

  The array of elements and separators, ordered like #[el1, sep1, el2, sep2, el3].

structure Lean.Syntax.TSepArray (ks : Lean.SyntaxNodeKinds) (sep : String) : Type

A typed version of SepArray.

• elemsAndSeps : Array Lean.Syntax

  The array of elements and separators, ordered like #[el1, sep1, el2, sep2, el3].

@[reducible, inline]

abbrev Lean.TSyntaxArray (ks : Lean.SyntaxNodeKinds) : Type

An array of syntaxes of kind ks.

Equations

• Lean.TSyntaxArray ks = Array (Lean.TSyntax ks)

unsafe def Lean.TSyntaxArray.rawImpl {ks : Lean.SyntaxNodeKinds} : Lean.TSyntaxArray ks → Array Lean.Syntax

Implementation of TSyntaxArray.raw.

Equations

• Lean.TSyntaxArray.rawImpl = unsafeCast

@[implemented_by Lean.TSyntaxArray.rawImpl]

opaque Lean.TSyntaxArray.raw {ks : Lean.SyntaxNodeKinds} (as : Lean.TSyntaxArray ks) : Array Lean.Syntax

Converts a TSyntaxArray to an Array Syntax, without reallocation.

unsafe def Lean.TSyntaxArray.mkImpl {ks : Lean.SyntaxNodeKinds} : Array Lean.Syntax → Lean.TSyntaxArray ks

Implementation of TSyntaxArray.mk.

Equations

• Lean.TSyntaxArray.mkImpl = unsafeCast

@[implemented_by Lean.TSyntaxArray.mkImpl]

opaque Lean.TSyntaxArray.mk {ks : Lean.SyntaxNodeKinds} (as : Array Lean.Syntax) : Lean.TSyntaxArray ks

Converts an Array Syntax to a TSyntaxArray, without reallocation.

def Lean.SourceInfo.fromRef (ref : Lean.Syntax) (canonical : Bool := false) : Lean.SourceInfo

Constructs a synthetic SourceInfo using a ref : Syntax for the span.

Equations

• One or more equations did not get rendered due to their size.

def Lean.mkAtom (val : String) : Lean.Syntax

Constructs a synthetic atom with no source info.

Equations

• Lean.mkAtom val = Lean.Syntax.atom Lean.SourceInfo.none val

def Lean.mkAtomFrom (src : Lean.Syntax) (val : String) (canonical : Bool := false) : Lean.Syntax

Constructs a synthetic atom with source info coming from src.

Equations

• Lean.mkAtomFrom src val canonical = Lean.Syntax.atom (Lean.SourceInfo.fromRef src canonical) val

Parser descriptions

inductive Lean.ParserDescr : Type

A ParserDescr is a grammar for parsers. This is used by the syntax command to produce parsers without having to import Lean.

• const (name : Lean.Name) : Lean.ParserDescr

  A (named) nullary parser, like ppSpace

• unary (name : Lean.Name) (p : Lean.ParserDescr) : Lean.ParserDescr

  A (named) unary parser, like group(p)

• binary (name : Lean.Name) (p₁ p₂ : Lean.ParserDescr) : Lean.ParserDescr

  A (named) binary parser, like orelse or andthen (written as p1 <|> p2 and p1 p2 respectively in syntax)

• node (kind : Lean.SyntaxNodeKind) (prec : Nat) (p : Lean.ParserDescr) : Lean.ParserDescr

  Parses using p, then pops the stack to create a new node with kind kind. The precedence prec is used to determine whether the parser should apply given the current precedence level.

• trailingNode (kind : Lean.SyntaxNodeKind) (prec lhsPrec : Nat) (p : Lean.ParserDescr) : Lean.ParserDescr

  Like node but for trailing parsers (which start with a nonterminal). Assumes the lhs is already on the stack, and parses using p, then pops the stack including the lhs to create a new node with kind kind. The precedence prec and lhsPrec are used to determine whether the parser should apply.

• symbol (val : String) : Lean.ParserDescr

  A literal symbol parser: parses val as a literal. This parser does not work on identifiers, so symbol arguments are declared as "keywords" and cannot be used as identifiers anywhere in the file.

• nonReservedSymbol (val : String) (includeIdent : Bool) : Lean.ParserDescr

  Like symbol, but without reserving val as a keyword. If includeIdent is true then ident will be reinterpreted as atom if it matches.

• cat (catName : Lean.Name) (rbp : Nat) : Lean.ParserDescr

  Parses using the category parser catName with right binding power (i.e. precedence) rbp.

• parser (declName : Lean.Name) : Lean.ParserDescr

  Parses using another parser declName, which can be either a Parser or ParserDescr.

• nodeWithAntiquot (name : String) (kind : Lean.SyntaxNodeKind) (p : Lean.ParserDescr) : Lean.ParserDescr

  Like node, but also declares that the body can be matched using an antiquotation with name name. For example, def $id:declId := 1 uses an antiquotation with name declId in the place where a declId is expected.

• sepBy (p : Lean.ParserDescr) (sep : String) (psep : Lean.ParserDescr) (allowTrailingSep : Bool := false) : Lean.ParserDescr

  A sepBy(p, sep) parses 0 or more occurrences of p separated by sep. psep is usually the same as symbol sep, but it can be overridden. sep is only used in the antiquot syntax: $x;* would match if sep is ";". allowTrailingSep is true if e.g. a, b, is also allowed to match.

• sepBy1 (p : Lean.ParserDescr) (sep : String) (psep : Lean.ParserDescr) (allowTrailingSep : Bool := false) : Lean.ParserDescr

  sepBy1 is just like sepBy, except it takes 1 or more instead of 0 or more occurrences of p.

instance Lean.instInhabitedParserDescr : Inhabited Lean.ParserDescr

Equations

• Lean.instInhabitedParserDescr = { default := Lean.ParserDescr.symbol "" }

@[reducible, inline]

abbrev Lean.TrailingParserDescr : Type

Although TrailingParserDescr is an abbreviation for ParserDescr, Lean will look at the declared type in order to determine whether to add the parser to the leading or trailing parser table. The determination is done automatically by the syntax command.

Equations

• Lean.TrailingParserDescr = Lean.ParserDescr

Runtime support for making quotation terms auto-hygienic, by mangling identifiers introduced by them with a "macro scope" supplied by the context. Details to appear in a paper soon.

@[reducible, inline]

abbrev Lean.MacroScope : Type

A macro scope identifier is just a Nat that gets bumped every time we enter a new macro scope. Within a macro scope, all occurrences of identifier x parse to the same thing, but x parsed from different macro scopes will produce different identifiers.

Equations

• Lean.MacroScope = Nat

def Lean.reservedMacroScope : Nat

Macro scope used internally. It is not available for our frontend.

Equations

• Lean.reservedMacroScope = 0

def Lean.firstFrontendMacroScope : Nat

First macro scope available for our frontend

Equations

• Lean.firstFrontendMacroScope = Lean.reservedMacroScope + 1

class Lean.MonadRef (m : Type → Type) : Type 1

A MonadRef is a monad that has a ref : Syntax in the read-only state. This is used to keep track of the location where we are working; if an exception is thrown, the ref gives the location where the error will be reported, assuming no more specific location is provided.

• getRef : m Lean.Syntax

  Get the current value of the ref

• withRef {α : Type} : Lean.Syntax → m α → m α

  Run x : m α with a modified value for the ref

instance Lean.instMonadRefOfMonadLiftOfMonadFunctor (m n : Type → Type) [MonadLift m n] [MonadFunctor m n] [Lean.MonadRef m] : Lean.MonadRef n

Equations

• One or more equations did not get rendered due to their size.

def Lean.replaceRef (ref oldRef : Lean.Syntax) : Lean.Syntax

Replaces oldRef with ref, unless ref has no position info. This biases us to having a valid span to report an error on.

Equations

• Lean.replaceRef ref oldRef = match ref.getPos? with | some val => ref | x => oldRef

@[inline]

def Lean.withRef {m : Type → Type} [Monad m] [Lean.MonadRef m] {α : Type} (ref : Lean.Syntax) (x : m α) : m α

Run x : m α with a modified value for the ref. This is not exactly the same as MonadRef.withRef, because it uses replaceRef to avoid putting syntax with bad spans in the state.

Equations

• Lean.withRef ref x = do let oldRef ← Lean.getRef let ref : Lean.Syntax := Lean.replaceRef ref oldRef Lean.MonadRef.withRef ref x

class Lean.MonadQuotation (m : Type → Type) extends Lean.MonadRef m : Type 1

A monad that supports syntax quotations. Syntax quotations (in term position) are monadic values that when executed retrieve the current "macro scope" from the monad and apply it to every identifier they introduce (independent of whether this identifier turns out to be a reference to an existing declaration, or an actually fresh binding during further elaboration). We also apply the position of the result of getRef to each introduced symbol, which results in better error positions than not applying any position.

• getRef : m Lean.Syntax
• withRef {α : Type} : Lean.Syntax → m α → m α
• getCurrMacroScope : m Lean.MacroScope

  Get the fresh scope of the current macro invocation

• getMainModule : m Lean.Name

  Get the module name of the current file. This is used to ensure that hygienic names don't clash across multiple files.

• withFreshMacroScope {α : Type} : m α → m α

  Execute action in a new macro invocation context. This transformer should be used at all places that morally qualify as the beginning of a "macro call", e.g. elabCommand and elabTerm in the case of the elaborator. However, it can also be used internally inside a "macro" if identifiers introduced by e.g. different recursive calls should be independent and not collide. While returning an intermediate syntax tree that will recursively be expanded by the elaborator can be used for the same effect, doing direct recursion inside the macro guarded by this transformer is often easier because one is not restricted to passing a single syntax tree. Modelling this helper as a transformer and not just a monadic action ensures that the current macro scope before the recursive call is restored after it, as expected.

@[inline]

def Lean.MonadRef.mkInfoFromRefPos {m : Type → Type} [Monad m] [Lean.MonadRef m] : m Lean.SourceInfo

Construct a synthetic SourceInfo from the ref in the monad state.

Equations

• Lean.MonadRef.mkInfoFromRefPos = do let __do_lift ← Lean.getRef pure (Lean.SourceInfo.fromRef __do_lift)

instance Lean.instMonadQuotationOfMonadFunctorOfMonadLift {m n : Type → Type} [MonadFunctor m n] [MonadLift m n] [Lean.MonadQuotation m] : Lean.MonadQuotation n

Equations

• One or more equations did not get rendered due to their size.

We represent a name with macro scopes as

<actual name>._@.(<module_name>.<scopes>)*.<module_name>._hyg.<scopes>

Example: suppose the module name is Init.Data.List.Basic, and name is foo.bla, and macroscopes [2, 5]

foo.bla._@.Init.Data.List.Basic._hyg.2.5

We may have to combine scopes from different files/modules. The main modules being processed is always the right-most one. This situation may happen when we execute a macro generated in an imported file in the current file.

foo.bla._@.Init.Data.List.Basic.2.1.Init.Lean.Expr._hyg.4

The delimiter _hyg is used just to improve the hasMacroScopes performance.

def Lean.Name.hasMacroScopes : Lean.Name → Bool

Does this name have hygienic macro scopes?

Equations

• (pre.str s).hasMacroScopes = (s == "_hyg")
• (p.num i).hasMacroScopes = p.hasMacroScopes
• x✝.hasMacroScopes = false

@[export lean_erase_macro_scopes]

def Lean.Name.eraseMacroScopes (n : Lean.Name) : Lean.Name

Remove the macro scopes from the name.

Equations

• n.eraseMacroScopes = match n.hasMacroScopes with | true => Lean.eraseMacroScopesAux✝ n | false => n

@[export lean_simp_macro_scopes]

def Lean.Name.simpMacroScopes (n : Lean.Name) : Lean.Name

Helper function we use to create binder names that do not need to be unique.

Equations

• n.simpMacroScopes = match n.hasMacroScopes with | true => Lean.simpMacroScopesAux✝ n | false => n

structure Lean.MacroScopesView : Type

A MacroScopesView represents a parsed hygienic name. extractMacroScopes will decode it from a Name, and .review will re-encode it. The grammar of a hygienic name is:

<name>._@.(<module_name>.<scopes>)*.<mainModule>._hyg.<scopes>

• name : Lean.Name

  The original (unhygienic) name.

• imported : Lean.Name

  All the name components (<module_name>.<scopes>)* from the imports concatenated together.

• mainModule : Lean.Name

  The main module in which this identifier was parsed.

• scopes : List Lean.MacroScope

  The list of macro scopes.

instance Lean.instInhabitedMacroScopesView : Inhabited Lean.MacroScopesView

Equations

• Lean.instInhabitedMacroScopesView = { default := { name := default, imported := default, mainModule := default, scopes := default } }

def Lean.MacroScopesView.review (view : Lean.MacroScopesView) : Lean.Name

Encode a hygienic name from the parsed pieces.

Equations

• One or more equations did not get rendered due to their size.

def Lean.extractMacroScopes (n : Lean.Name) : Lean.MacroScopesView

Revert all addMacroScope calls. v = extractMacroScopes n → n = v.review. This operation is useful for analyzing/transforming the original identifiers, then adding back the scopes (via MacroScopesView.review).

Equations

• One or more equations did not get rendered due to their size.

def Lean.addMacroScope (mainModule n : Lean.Name) (scp : Lean.MacroScope) : Lean.Name

Add a new macro scope onto the name n, in the given mainModule.

Equations

• One or more equations did not get rendered due to their size.

def Lean.Name.append (a b : Lean.Name) : Lean.Name

Appends two names a and b, propagating macro scopes from a or b, if any, to the result. Panics if both a and b have macro scopes.

This function is used for the Append Name instance.

See also Lean.Name.appendCore, which appends names without any consideration for macro scopes. Also consider Lean.Name.eraseMacroScopes to erase macro scopes before appending, if appropriate.

Equations

• One or more equations did not get rendered due to their size.

instance Lean.instAppendName : Append Lean.Name

Equations

• Lean.instAppendName = { append := Lean.Name.append }

@[inline]

def Lean.MonadQuotation.addMacroScope {m : Type → Type} [Lean.MonadQuotation m] [Monad m] (n : Lean.Name) : m Lean.Name

Add a new macro scope onto the name n, using the monad state to supply the main module and current macro scope.

Equations

• Lean.MonadQuotation.addMacroScope n = do let mainModule ← Lean.getMainModule let scp ← Lean.getCurrMacroScope pure (Lean.addMacroScope mainModule n scp)

def Lean.Syntax.matchesNull (stx : Lean.Syntax) (n : Nat) : Bool

Is this syntax a null node?

Equations

• stx.matchesNull n = stx.isNodeOf Lean.nullKind n

def Lean.Syntax.matchesIdent (stx : Lean.Syntax) (id : Lean.Name) : Bool

Function used for determining whether a syntax pattern `(id) is matched. There are various conceivable notions of when two syntactic identifiers should be regarded as identical, but semantic definitions like whether they refer to the same global name cannot be implemented without context information (i.e. MonadResolveName). Thus in patterns we default to the structural solution of comparing the identifiers' Name values, though we at least do so modulo macro scopes so that identifiers that "look" the same match. This is particularly useful when dealing with identifiers that do not actually refer to Lean bindings, e.g. in the stx pattern `(many($p)).

Equations

• stx.matchesIdent id = (stx.isIdent && stx.getId.eraseMacroScopes == id.eraseMacroScopes)

def Lean.Syntax.matchesLit (stx : Lean.Syntax) (k : Lean.SyntaxNodeKind) (val : String) : Bool

Is this syntax a node kind k wrapping an atom _ val?

Equations

• (Lean.Syntax.node info kind args).matchesLit k val = (k == kind && match args.getD 0 Lean.Syntax.missing with | Lean.Syntax.atom info val' => val == val' | x => false)
• stx.matchesLit k val = false

instance Lean.Macro.instNonemptyMethodsRef : Nonempty Lean.Macro.MethodsRef✝

structure Lean.Macro.Context : Type

The read-only context for the MacroM monad.

• methods : Lean.Macro.MethodsRef✝

  An opaque reference to the Methods object. This is done to break a dependency cycle: the Methods involve MacroM which has not been defined yet.

• mainModule : Lean.Name

  The currently parsing module.

• currMacroScope : Lean.MacroScope

  The current macro scope.

• currRecDepth : Nat

  The current recursion depth.

• maxRecDepth : Nat

  The maximum recursion depth.

• ref : Lean.Syntax

  The syntax which supplies the position of error messages.

inductive Lean.Macro.Exception : Type

An exception in the MacroM monad.

• error : Lean.Syntax → String → Lean.Macro.Exception

  A general error, given a message and a span (expressed as a Syntax).

• unsupportedSyntax : Lean.Macro.Exception

  An unsupported syntax exception. We keep this separate because it is used for control flow: if one macro does not support a syntax then we try the next one.

structure Lean.Macro.State : Type

The mutable state for the MacroM monad.

• macroScope : Lean.MacroScope

  The global macro scope counter, used for producing fresh scope names.

• traceMsgs : List (Lean.Name × String)

  The list of trace messages that have been produced, each with a trace class and a message.

instance Lean.Macro.instInhabitedState : Inhabited Lean.Macro.State

Equations

• Lean.Macro.instInhabitedState = { default := { macroScope := default, traceMsgs := default } }

@[reducible, inline]

abbrev Lean.MacroM (α : Type) : Type

The MacroM monad is the main monad for macro expansion. It has the information needed to handle hygienic name generation, and is the monad that macro definitions live in.

Notably, this is a (relatively) pure monad: there is no IO and no access to the Environment. That means that things like declaration lookup are impossible here, as well as IO.Ref or other side-effecting operations. For more capabilities, macros can instead be written as elab using adaptExpander.

Equations

• Lean.MacroM = ReaderT Lean.Macro.Context (EStateM Lean.Macro.Exception Lean.Macro.State)

@[reducible, inline]

abbrev Lean.Macro : Type

A macro has type Macro, which is a Syntax → MacroM Syntax: it receives an input syntax and is supposed to "expand" it into another piece of syntax.

Equations

• Lean.Macro = (Lean.Syntax → Lean.MacroM Lean.Syntax)

instance Lean.Macro.instMonadRefMacroM : Lean.MonadRef Lean.MacroM

Equations

• One or more equations did not get rendered due to their size.

def Lean.Macro.addMacroScope (n : Lean.Name) : Lean.MacroM Lean.Name

Add a new macro scope to the name n.

Equations

• Lean.Macro.addMacroScope n = do let ctx ← read pure (Lean.addMacroScope ctx.mainModule n ctx.currMacroScope)

def Lean.Macro.throwUnsupported {α : Type} : Lean.MacroM α

Throw an unsupportedSyntax exception.

Equations

• Lean.Macro.throwUnsupported = throw Lean.Macro.Exception.unsupportedSyntax

def Lean.Macro.throwError {α : Type} (msg : String) : Lean.MacroM α

Throw an error with the given message, using the ref for the location information.

Equations

• Lean.Macro.throwError msg = do let ref ← Lean.getRef throw (Lean.Macro.Exception.error ref msg)

def Lean.Macro.throwErrorAt {α : Type} (ref : Lean.Syntax) (msg : String) : Lean.MacroM α

Throw an error with the given message and location information.

Equations

• Lean.Macro.throwErrorAt ref msg = Lean.withRef ref (Lean.Macro.throwError msg)

@[inline]

def Lean.Macro.withFreshMacroScope {α : Type} (x : Lean.MacroM α) : Lean.MacroM α

Increments the macro scope counter so that inside the body of x the macro scope is fresh.

Equations

• One or more equations did not get rendered due to their size.

@[inline]

def Lean.Macro.withIncRecDepth {α : Type} (ref : Lean.Syntax) (x : Lean.MacroM α) : Lean.MacroM α

Run x with an incremented recursion depth counter.

Equations

• One or more equations did not get rendered due to their size.

instance Lean.Macro.instMonadQuotationMacroM : Lean.MonadQuotation Lean.MacroM

Equations

• One or more equations did not get rendered due to their size.

structure Lean.Macro.Methods : Type

The opaque methods that are available to MacroM.

• expandMacro? : Lean.Syntax → Lean.MacroM (Option Lean.Syntax)

  Expands macros in the given syntax. A return value of none means there was nothing to expand.

• getCurrNamespace : Lean.MacroM Lean.Name

  Get the current namespace in the file.

• hasDecl : Lean.Name → Lean.MacroM Bool

  Check if a given name refers to a declaration.

• resolveNamespace : Lean.Name → Lean.MacroM (List Lean.Name)

  Resolves the given name to an overload list of namespaces.

• resolveGlobalName : Lean.Name → Lean.MacroM (List (Lean.Name × List String))

  Resolves the given name to an overload list of global definitions. The List String in each alternative is the deduced list of projections (which are ambiguous with name components).

instance Lean.Macro.instInhabitedMethods : Inhabited Lean.Macro.Methods

Equations

• One or more equations did not get rendered due to their size.

unsafe def Lean.Macro.mkMethodsImp (methods : Lean.Macro.Methods) : Lean.Macro.MethodsRef✝

Implementation of mkMethods.

Equations

• Lean.Macro.mkMethodsImp methods = unsafeCast methods

@[implemented_by Lean.Macro.mkMethodsImp]

opaque Lean.Macro.mkMethods (methods : Lean.Macro.Methods) : Lean.Macro.MethodsRef✝

Make an opaque reference to a Methods.

instance Lean.Macro.instInhabitedMethodsRef : Inhabited Lean.Macro.MethodsRef✝

Equations

• Lean.Macro.instInhabitedMethodsRef = { default := Lean.Macro.mkMethods default }

unsafe def Lean.Macro.getMethodsImp : Lean.MacroM Lean.Macro.Methods

Implementation of getMethods.

Equations

• Lean.Macro.getMethodsImp = do let ctx ← read pure (unsafeCast ctx.methods)

@[implemented_by Lean.Macro.getMethodsImp]

opaque Lean.Macro.getMethods : Lean.MacroM Lean.Macro.Methods

Extract the methods list from the MacroM state.

def Lean.Macro.expandMacro? (stx : Lean.Syntax) : Lean.MacroM (Option Lean.Syntax)

expandMacro? stx returns some stxNew if stx is a macro, and stxNew is its expansion.

Equations

• Lean.expandMacro? stx = do let __do_lift ← Lean.Macro.getMethods __do_lift.expandMacro? stx

def Lean.Macro.hasDecl (declName : Lean.Name) : Lean.MacroM Bool

Returns true if the environment contains a declaration with name declName

Equations

• Lean.Macro.hasDecl declName = do let __do_lift ← Lean.Macro.getMethods __do_lift.hasDecl declName

def Lean.Macro.getCurrNamespace : Lean.MacroM Lean.Name

Gets the current namespace given the position in the file.

Equations

• Lean.Macro.getCurrNamespace = do let __do_lift ← Lean.Macro.getMethods __do_lift.getCurrNamespace

def Lean.Macro.resolveNamespace (n : Lean.Name) : Lean.MacroM (List Lean.Name)

Resolves the given name to an overload list of namespaces.

Equations

• Lean.Macro.resolveNamespace n = do let __do_lift ← Lean.Macro.getMethods __do_lift.resolveNamespace n

def Lean.Macro.resolveGlobalName (n : Lean.Name) : Lean.MacroM (List (Lean.Name × List String))

Resolves the given name to an overload list of global definitions. The List String in each alternative is the deduced list of projections (which are ambiguous with name components).

Remark: it will not trigger actions associated with reserved names. Recall that Lean has reserved names. For example, a definition foo has a reserved name foo.def for theorem containing stating that foo is equal to its definition. The action associated with foo.def automatically proves the theorem. At the macro level, the name is resolved, but the action is not executed. The actions are executed by the elaborator when converting Syntax into Expr.

Equations

• Lean.Macro.resolveGlobalName n = do let __do_lift ← Lean.Macro.getMethods __do_lift.resolveGlobalName n

def Lean.Macro.trace (clsName : Lean.Name) (msg : String) : Lean.MacroM Unit

Add a new trace message, with the given trace class and message.

Equations

• Lean.Macro.trace clsName msg = modify fun (s : Lean.Macro.State) => { macroScope := s.macroScope, traceMsgs := (clsName, msg) :: s.traceMsgs }

@[reducible, inline]

abbrev Lean.PrettyPrinter.UnexpandM (α : Type) : Type

The unexpander monad, essentially Syntax → Option α. The Syntax is the ref, and it has the possibility of failure without an error message.

Equations

• Lean.PrettyPrinter.UnexpandM = ReaderT Lean.Syntax (EStateM Unit Unit)

@[reducible, inline]

abbrev Lean.PrettyPrinter.Unexpander : Type

Function that tries to reverse macro expansions as a post-processing step of delaboration. While less general than an arbitrary delaborator, it can be declared without importing Lean. Used by the [app_unexpander] attribute.

Equations

• Lean.PrettyPrinter.Unexpander = (Lean.Syntax → Lean.PrettyPrinter.UnexpandM Lean.Syntax)

instance Lean.PrettyPrinter.instMonadQuotationUnexpandM : Lean.MonadQuotation Lean.PrettyPrinter.UnexpandM

Equations

• Lean.PrettyPrinter.instMonadQuotationUnexpandM = Lean.MonadQuotation.mk (pure 0) (pure `_fakeMod) fun {α : Type} => id