Merge pull request #110 from github/patch-costs-are-proportional-to-the-size-of-their-trees

Patch costs are proportional to the size of their trees

prototype/Doubt/Algorithm.swift

@@ -52,6 +52,18 @@ public enum Algorithm<Term: TermType, B> {
 		let recurOrReplace = {
 			recur($0, $1) ?? .Pure(.Replace($0, $1))
 		}
+		func cost(diff: Diff) -> Int {
+			return diff.map { abs(($0.state.before?.size ?? 0) - ($0.state.after?.size ?? 0)) }.iterate { syntax in
+				switch syntax {
+				case .Leaf:
+					return 0
+				case let .Indexed(costs):
+					return costs.reduce(0, combine: +)
+				case let .Keyed(costs):
+					return costs.values.reduce(0, combine: +)
+				}
+			}
+		}
 		switch self {
 		case let .Pure(b):
 			return b
@@ -82,7 +94,7 @@ public enum Algorithm<Term: TermType, B> {
 			return f(Dictionary(elements: deleted + inserted + patched)).evaluate(equals, recur: recur)
 
 		case let .Roll(.ByIndex(a, b, f)):
-			return f(SES(a, b, recur: recur)).evaluate(equals, recur: recur)
+			return f(SES(a, b, cost: cost, recur: recur)).evaluate(equals, recur: recur)
 		}
 	}
 }

prototype/Doubt/Patch.swift

@@ -38,7 +38,7 @@ public enum Patch<A>: CustomDebugStringConvertible {
 			return ".Insert(\(String(reflecting: b)))"
 		case let .Delete(a):
 			return ".Delete(\(String(reflecting: a)))"
-		}
+	}
 	}
 }
 
@@ -85,6 +85,8 @@ extension Patch where A: CustomJSONConvertible {
 }
 
 
+// MARK: - PatchConvertible
+
 /// A hack to enable constrained extensions on `Free<A, Patch<Term: TermType where LeafType == A>`.
 public protocol PatchConvertible {
 	typealias Element

prototype/Doubt/SES.swift

@@ -1,25 +1,12 @@
 /// Computes the SES (shortest edit script), i.e. the shortest sequence of diffs (`Free<A, Patch<Term>>`) for two arrays of `Term`s which would suffice to transform `a` into `b`.
 ///
 /// This is computed w.r.t. an `equals` function, which computes the equality of leaf nodes within terms, and a `recur` function, which produces diffs representing matched-up terms.
-public func SES<Term, A>(a: [Term], _ b: [Term], recur: (Term, Term) -> Free<A, Patch<Term>>?) -> [Free<A, Patch<Term>>] {
+public func SES<Term, A>(a: [Term], _ b: [Term], cost: Free<A, Patch<Term>> -> Int, recur: (Term, Term) -> Free<A, Patch<Term>>?) -> [Free<A, Patch<Term>>] {
 	typealias Diff = Free<A, Patch<Term>>
 
 	if a.isEmpty { return b.map { Diff.Pure(Patch.Insert($0)) } }
 	if b.isEmpty { return a.map { Diff.Pure(Patch.Delete($0)) } }
 
-	func cost(diff: Diff) -> Int {
-		return diff.map(const(1)).iterate { syntax in
-			switch syntax {
-			case .Leaf:
-				return 0
-			case let .Indexed(costs):
-				return costs.reduce(0, combine: +)
-			case let .Keyed(costs):
-				return costs.values.reduce(0, combine: +)
-			}
-		}
-	}
-
 	func cons(diff: Diff, rest: Memo<Stream<(Diff, Int)>>) -> Stream<(Diff, Int)> {
 		return .Cons((diff, cost(diff) + costOfStream(rest)), rest)
 	}

prototype/Doubt/TermType.swift

@@ -6,6 +6,33 @@ public protocol TermType {
 }
 
 
+extension TermType {
+	/// Catamorphism over `TermType`s.
+	///
+	/// Folds the tree encoded by the receiver into a single value by recurring top-down through the tree, applying `transform` to leaves, then to branches, and so forth.
+	public func cata<Result>(transform: Syntax<Result, LeafType> -> Result) -> Result {
+		return self |> ({ $0.unwrap } >>> { $0.map { $0.cata(transform) } } >>> transform)
+	}
+
+
+	/// The count of nodes in the receiver.
+	///
+	/// This is used to compute the cost of patches, such that a patch inserting a very large tree will be charged approximately the same as a very large tree consisting of many small patches.
+	public var size: Int {
+		return cata {
+			switch $0 {
+			case .Leaf:
+				return 1
+			case let .Indexed(i):
+				return i.reduce(1, combine: +)
+			case let .Keyed(k):
+				return k.values.reduce(1, combine: +)
+			}
+		}
+	}
+}
+
+
 extension Cofree: TermType {}
 
 
@@ -16,3 +43,6 @@ extension TermType {
 		return Syntax.equals(ifLeaf: leaf, ifRecur: equals(leaf))(a.unwrap, b.unwrap)
 	}
 }
+
+
+import Prelude

prototype/DoubtTests/SESTests.swift

@@ -57,7 +57,7 @@ private let c = Term((), .Leaf("c"))
 private let d = Term((), .Leaf("d"))
 
 private func SES(a: [Term], _ b: [Term]) -> [Diff] {
-	return SES(a, b) { Cofree.equals(annotation: const(true), leaf: ==)($0, $1) ? Diff($1) : nil }
+	return SES(a, b, cost: const(1)) { Cofree.equals(annotation: const(true), leaf: ==)($0, $1) ? Diff($1) : nil }
 }
 
 private func == (a: [Diff], b: [Diff]) -> Bool {