summaryrefslogtreecommitdiff
path: root/Test/VSI-Benchmarks/b4.dfy
diff options
context:
space:
mode:
authorGravatar Rustan Leino <unknown>2013-03-05 17:01:25 -0800
committerGravatar Rustan Leino <unknown>2013-03-05 17:01:25 -0800
commit9f7c11c0a12a9b802f23d613abeee91a2fb47743 (patch)
treeff0c5ffd982094c5d03828f1332be03a4d66b4ea /Test/VSI-Benchmarks/b4.dfy
parentc819fabbb8da669952cb7e2e5937c73ff6dcfabe (diff)
Removed Dafny, Jennisys, Chalice, and BCT, which now live in different Codeplex repositories.
Diffstat (limited to 'Test/VSI-Benchmarks/b4.dfy')
-rw-r--r--Test/VSI-Benchmarks/b4.dfy158
1 files changed, 0 insertions, 158 deletions
diff --git a/Test/VSI-Benchmarks/b4.dfy b/Test/VSI-Benchmarks/b4.dfy
deleted file mode 100644
index d5a56df4..00000000
--- a/Test/VSI-Benchmarks/b4.dfy
+++ /dev/null
@@ -1,158 +0,0 @@
-/*
- This test works with Z3 3.2 and Z3 4.1 (on Win7 x64). Other versions ... who knows.
-*/
-
-// Note: We are using the built-in equality to compare keys.
-// Note: The abstract specifications would do well with a map from keys
-// to values. However, Dafny does not support maps. Instead, such a
-// map is modeled by two sequences, one for the keys and one for the values.
-// The indices in these sequences correspond, so the dictionary maps
-// Keys[i] to Values[i].
-// The implementation uses a linked list, which isn't the most efficient
-// representation of a dictionary. However, for the benchmark, it shows
-// that the specification can use mathematical sequences while the
-// implementation uses a linked list.
-
-class Map<Key(==),Value> {
- ghost var Keys: seq<Key>;
- ghost var Values: seq<Value>;
- ghost var Repr: set<object>;
-
- var head: Node<Key,Value>;
- ghost var nodes: seq<Node<Key,Value>>;
-
- function Valid(): bool
- reads this, Repr;
- {
- this in Repr &&
- |Keys| == |Values| && |nodes| == |Keys| + 1 &&
- head == nodes[0] &&
- (forall i :: 0 <= i && i < |Keys| ==>
- nodes[i] != null &&
- nodes[i] in Repr &&
- nodes[i].key == Keys[i] && nodes[i].key !in Keys[i+1..] &&
- nodes[i].val == Values[i] &&
- nodes[i].next == nodes[i+1]) &&
- nodes[|nodes|-1] == null
- }
-
- method Init()
- modifies this;
- ensures Valid() && fresh(Repr - {this});
- ensures |Keys| == 0;
- {
- Keys := [];
- Values := [];
- Repr := {this};
- head := null;
- nodes := [null];
- }
-
- method Find(key: Key) returns (present: bool, val: Value)
- requires Valid();
- ensures !present ==> key !in Keys;
- ensures present ==> (exists i :: 0 <= i && i < |Keys| && Keys[i] == key && Values[i] == val);
- {
- var p, n, prev := FindIndex(key);
- if (p == null) {
- present := false;
- } else {
- val := p.val;
- present := true;
- }
- }
-
- method Add(key: Key, val: Value)
- requires Valid();
- modifies Repr;
- ensures Valid() && fresh(Repr - old(Repr));
- ensures (forall i :: 0 <= i && i < |old(Keys)| && old(Keys)[i] == key ==>
- |Keys| == |old(Keys)| &&
- Keys[i] == key && Values[i] == val &&
- (forall j :: 0 <= j && j < |Values| && i != j ==>
- Keys[j] == old(Keys)[j] && Values[j] == old(Values)[j]));
- ensures key !in old(Keys) ==> Keys == [key] + old(Keys) && Values == [val] + old(Values);
- {
- var p, n, prev := FindIndex(key);
- if (p == null) {
- var h := new Node<Key,Value>;
- h.key := key; h.val := val; h.next := head;
- head := h;
- Keys := [key] + Keys; Values := [val] + Values;
- nodes := [h] + nodes;
- Repr := Repr + {h};
- } else {
- p.val := val;
- Values := Values[n := val];
- }
- }
-
- method Remove(key: Key)// returns (ghost h: int)
- requires Valid();
- modifies Repr;
- ensures Valid() && fresh(Repr - old(Repr));
- // no key is introduced:
- ensures (forall k :: k in Keys ==> k in old(Keys));
- // at most one key is removed:
- ensures (forall k :: k in old(Keys) ==> k in Keys || k == key);
- // other values don't change:
- ensures key !in old(Keys) ==> Keys == old(Keys) && Values == old(Values);
- ensures key in old(Keys) ==>
- |Keys| == |old(Keys)| - 1 && key !in Keys &&
- (exists h ::
- 0 <= h && h < |old(Keys)| &&
- Keys[..h] == old(Keys)[..h] &&
- Values[..h] == old(Values)[..h] &&
- Keys[h..] == old(Keys)[h+1..] &&
- Values[h..] == old(Values)[h+1..]);
- {
- var p, n, prev := FindIndex(key);
- if (p != null) {
- Keys := Keys[..n] + Keys[n+1..];
- Values := Values[..n] + Values[n+1..];
-
- nodes := nodes[..n] + nodes[n+1..];
- if (prev == null) {
- head := head.next;
- } else {
- prev.next := p.next;
- }
- assert Keys[n..] == old(Keys)[n+1..];
- assert Values[n..] == old(Values)[n+1..];
- }
- }
-
- /*private*/ method FindIndex(key: Key) returns (p: Node<Key,Value>, ghost n: int, prev: Node<Key,Value>)
- requires Valid();
- ensures p == null ==> key !in Keys;
- ensures p != null ==>
- 0 <= n && n < |Keys| && Keys[n] == key &&
- key !in Keys[..n] && key !in Keys[n+1..] &&
- p == nodes[n] &&
- ((n == 0 && prev == null) || (0 < n && prev == nodes[n-1]));
- {
- n := 0;
- prev := null;
- p := head;
- while (p != null)
- invariant n <= |Keys| && p == nodes[n];
- invariant key !in Keys[..n];
- invariant (n == 0 && prev == null) || (0 < n && prev == nodes[n-1]);
- decreases |Keys| - n;
- {
- if (p.key == key) {
- return;
- } else {
- n := n + 1;
- prev := p;
- p := p.next;
- }
- }
- }
-}
-
-class Node<Key,Value> {
- var key: Key;
- var val: Value;
- var next: Node<Key,Value>;
-}