| // RUN: %verify "%s" | |
| /******************************************************************************* | |
| * Copyright by the contributors to the Dafny Project | |
| * SPDX-License-Identifier: MIT | |
| *******************************************************************************/ | |
| include "../Wrappers.dfy" | |
| include "../Functions.dfy" | |
| include "../Collections/Sequences/Seq.dfy" | |
| include "Unicode.dfy" | |
| /** | |
| * A Unicode encoding form assigns each Unicode scalar value to a unique code unit sequence. | |
| * | |
| * A concrete `EncodingForm` MUST define the following: | |
| * - The type `CodeUnit`. | |
| * - The predicate `IsMinimalWellFormedCodeUnitSubsequence`, which defines the set of encodings of scalar values, | |
| * known as "minimal well-formed code unit subsequences". | |
| * - The function `SplitPrefixMinimalWellFormedCodeUnitSubsequence`, which defines the algorithm by which to parse | |
| * a minimal well-formed code unit subsequence from the beginning of a code unit sequence. | |
| * - The function `EncodeScalarValue`, which defines the mapping from scalar values to minimal well-formed code unit | |
| * subsequences. | |
| * - The function `DecodeMinimalWellFormedCodeUnitSubsequence`, which defines the mapping from minimal well-formed | |
| * code unit subsequences to scalar values. | |
| */ | |
| abstract module {:options "-functionSyntax:4"} UnicodeEncodingForm { | |
| import opened Wrappers | |
| import Functions | |
| import Seq | |
| import Unicode | |
| type CodeUnitSeq = seq<CodeUnit> | |
| type WellFormedCodeUnitSeq = s: CodeUnitSeq | |
| | IsWellFormedCodeUnitSequence(s) | |
| witness [] | |
| type MinimalWellFormedCodeUnitSeq = s: CodeUnitSeq | |
| | IsMinimalWellFormedCodeUnitSubsequence(s) | |
| witness * | |
| // | |
| // Begin abstract items. | |
| // | |
| /** | |
| * A code unit is the minimal bit combination that can represent a unit of encoded text for processing or | |
| * interchange. (Section 3.9 D77.) | |
| */ | |
| type CodeUnit | |
| /** | |
| * A well-formed Unicode code unit sequence that maps to a single Unicode scalar value. | |
| * (Section 3.9 D85a.) | |
| */ | |
| function IsMinimalWellFormedCodeUnitSubsequence(s: CodeUnitSeq): (b: bool) | |
| ensures b ==> | |
| && |s| > 0 | |
| && forall i | 0 < i < |s| :: !IsMinimalWellFormedCodeUnitSubsequence(s[..i]) | |
| decreases |s| | |
| /** | |
| * Returns the shortest prefix of `s` that is a minimal well-formed code unit subsequence, | |
| * or None if there is no such prefix. | |
| */ | |
| function SplitPrefixMinimalWellFormedCodeUnitSubsequence(s: CodeUnitSeq): (maybePrefix: Option<MinimalWellFormedCodeUnitSeq>) | |
| ensures |s| == 0 ==> maybePrefix.None? | |
| ensures (exists i | 0 < i <= |s| :: IsMinimalWellFormedCodeUnitSubsequence(s[..i])) <==> | |
| && maybePrefix.Some? | |
| ensures maybePrefix.Some? ==> | |
| && var prefix := maybePrefix.Extract(); | |
| && 0 < |prefix| <= |s| | |
| && prefix == s[..|prefix|] | |
| && forall i | 0 < i < |prefix| :: !IsMinimalWellFormedCodeUnitSubsequence(s[..i]) | |
| /** | |
| * Returns the minimal well-formed code unit subsequence that this encoding form assigns to the given scalar value. | |
| * The Unicode standard requires that this is injective. | |
| * | |
| * TODO: enforce that implementations satisfy Functions.Injective | |
| */ | |
| function EncodeScalarValue(v: Unicode.ScalarValue): (m: MinimalWellFormedCodeUnitSeq) | |
| /** | |
| * Returns the scalar value that this encoding form assigns to the given minimal well-formed code unit subsequence. | |
| */ | |
| function DecodeMinimalWellFormedCodeUnitSubsequence(m: MinimalWellFormedCodeUnitSeq): (v: Unicode.ScalarValue) | |
| ensures EncodeScalarValue(v) == m | |
| // | |
| // End abstract items. | |
| // | |
| /** | |
| * If `ms` is the concatenation of a minimal well-formed code unit subsequence `m` and a code unit sequence `s`, | |
| * then the shortest minimal well-formed code unit subsequence prefix of `ms` is simply `m`. | |
| */ | |
| lemma LemmaSplitPrefixMinimalWellFormedCodeUnitSubsequenceInvertsPrepend(m: MinimalWellFormedCodeUnitSeq, s: CodeUnitSeq) | |
| ensures SplitPrefixMinimalWellFormedCodeUnitSubsequence(m + s) == Some(m) | |
| { | |
| var ms := m + s; | |
| assert IsMinimalWellFormedCodeUnitSubsequence(ms[..|m|]); | |
| var prefix := SplitPrefixMinimalWellFormedCodeUnitSubsequence(ms).Extract(); | |
| calc ==> { | |
| IsMinimalWellFormedCodeUnitSubsequence(m); | |
| |prefix| <= |m|; | |
| prefix == ms[..|prefix|] == m[..|prefix|] == m; | |
| } | |
| } | |
| /** | |
| * Returns the unique partition of the given code unit sequence into minimal well-formed code unit subsequences, | |
| * or None if no such partition exists. | |
| */ | |
| function PartitionCodeUnitSequenceChecked(s: CodeUnitSeq): (maybeParts: Option<seq<MinimalWellFormedCodeUnitSeq>>) | |
| ensures maybeParts.Some? ==> Seq.Flatten(maybeParts.Extract()) == s | |
| decreases |s| | |
| { | |
| if s == [] then Some([]) | |
| else | |
| var prefix :- SplitPrefixMinimalWellFormedCodeUnitSubsequence(s); | |
| // Recursing/iterating on subsequences leads to quadratic running time in most/all Dafny runtimes as of this writing. | |
| // This definition (and others in the Unicode modules) emphasizes clarity and correctness, | |
| // and while the by-method implementation avoids stack overflows due to non-tail-recursive recursion, | |
| // it doesn't yet avoid the subsequences. | |
| // The best solution would be to ensure all Dafny runtimes implement subsequences as an O(1) | |
| // operation, so this implementation would become linear. | |
| var restParts :- PartitionCodeUnitSequenceChecked(s[|prefix|..]); | |
| Some([prefix] + restParts) | |
| } by method { | |
| if s == [] { | |
| return Some([]); | |
| } | |
| var result: seq<MinimalWellFormedCodeUnitSeq> := []; | |
| var rest := s; | |
| while |rest| > 0 | |
| invariant PartitionCodeUnitSequenceChecked(s).Some? | |
| <==> PartitionCodeUnitSequenceChecked(rest).Some? | |
| invariant | |
| PartitionCodeUnitSequenceChecked(s).Some? ==> | |
| && PartitionCodeUnitSequenceChecked(s).value | |
| == result + PartitionCodeUnitSequenceChecked(rest).value | |
| { | |
| var prefix :- SplitPrefixMinimalWellFormedCodeUnitSubsequence(rest); | |
| result := result + [prefix]; | |
| rest := rest[|prefix|..]; | |
| } | |
| assert result + [] == result; | |
| return Some(result); | |
| } | |
| /** | |
| * Returns the unique partition of the given well-formed code unit sequence into minimal well-formed code unit | |
| * subsequences. | |
| */ | |
| function PartitionCodeUnitSequence(s: WellFormedCodeUnitSeq): (parts: seq<MinimalWellFormedCodeUnitSeq>) | |
| ensures Seq.Flatten(parts) == s | |
| { | |
| PartitionCodeUnitSequenceChecked(s).Extract() | |
| } | |
| /** | |
| * The partitioning of a minimal well-formed code unit subsequence is the singleton sequence | |
| * containing exactly the minimal well-formed code unit subsequence. | |
| */ | |
| lemma LemmaPartitionMinimalWellFormedCodeUnitSubsequence(m: MinimalWellFormedCodeUnitSeq) | |
| ensures PartitionCodeUnitSequenceChecked(m) == Some([m]) | |
| { | |
| LemmaSplitPrefixMinimalWellFormedCodeUnitSubsequenceInvertsPrepend(m, []); | |
| calc == { | |
| Some(m); | |
| SplitPrefixMinimalWellFormedCodeUnitSubsequence(m + []); | |
| { assert m + [] == m; } | |
| SplitPrefixMinimalWellFormedCodeUnitSubsequence(m); | |
| } | |
| calc == { | |
| PartitionCodeUnitSequenceChecked(m); | |
| Some([m] + []); | |
| { assert [m] + [] == [m]; } | |
| Some([m]); | |
| } | |
| } | |
| /** | |
| * A code unit sequence is well-formed iff it can be partitioned into a sequence of minimal well-formed code unit subsequences. | |
| */ | |
| function IsWellFormedCodeUnitSequence(s: CodeUnitSeq): (b: bool) | |
| { | |
| PartitionCodeUnitSequenceChecked(s).Some? | |
| } | |
| /** | |
| * A minimal well-formed code unit subsequence is a well-formed code unit sequence. | |
| */ | |
| lemma LemmaMinimalWellFormedCodeUnitSubsequenceIsWellFormedSequence(m: MinimalWellFormedCodeUnitSeq) | |
| ensures IsWellFormedCodeUnitSequence(m) | |
| { | |
| LemmaPartitionMinimalWellFormedCodeUnitSubsequence(m); | |
| } | |
| /** | |
| * The concatenation of a minimal well-formed code unit subsequence and a well-formed code unit sequence | |
| * is itself a well-formed code unit sequence. | |
| */ | |
| lemma LemmaPrependMinimalWellFormedCodeUnitSubsequence(m: MinimalWellFormedCodeUnitSeq, s: WellFormedCodeUnitSeq) | |
| ensures IsWellFormedCodeUnitSequence(m + s) | |
| { | |
| LemmaPartitionMinimalWellFormedCodeUnitSubsequence(m); | |
| LemmaSplitPrefixMinimalWellFormedCodeUnitSubsequenceInvertsPrepend(m, s); | |
| assert PartitionCodeUnitSequenceChecked(m + s).Some?; | |
| } | |
| /** | |
| * The concatenation of minimal well-formed code unit subsequences is itself a well-formed code unit sequence. | |
| */ | |
| lemma LemmaFlattenMinimalWellFormedCodeUnitSubsequences(ms: seq<MinimalWellFormedCodeUnitSeq>) | |
| ensures IsWellFormedCodeUnitSequence(Seq.Flatten(ms)) | |
| { | |
| if |ms| == 0 { | |
| assert IsWellFormedCodeUnitSequence(Seq.Flatten(ms)); | |
| } | |
| else { | |
| var head := ms[0]; | |
| var tail := ms[1..]; | |
| LemmaFlattenMinimalWellFormedCodeUnitSubsequences(tail); | |
| var flatTail := Seq.Flatten(tail); | |
| LemmaPrependMinimalWellFormedCodeUnitSubsequence(head, flatTail); | |
| assert IsWellFormedCodeUnitSequence(head + flatTail); | |
| } | |
| } | |
| /** | |
| * The concatenation of well-formed code unit sequences is itself a well-formed code unit sequence. | |
| */ | |
| lemma LemmaConcatWellFormedCodeUnitSubsequences(s: WellFormedCodeUnitSeq, t: WellFormedCodeUnitSeq) | |
| ensures IsWellFormedCodeUnitSequence(s + t) | |
| { | |
| var partsS := PartitionCodeUnitSequence(s); | |
| var partsT := PartitionCodeUnitSequence(t); | |
| var partsST := partsS + partsT; | |
| Seq.LemmaFlattenConcat(partsS, partsT); | |
| assert s + t == Seq.Flatten(partsST); | |
| assert forall part | part in partsST :: | |
| && |part| > 0 | |
| && IsMinimalWellFormedCodeUnitSubsequence(part); | |
| LemmaFlattenMinimalWellFormedCodeUnitSubsequences(partsST); | |
| } | |
| /** | |
| * Returns the well-formed Unicode string that is the encoding of the given scalar value sequence. | |
| */ | |
| function EncodeScalarSequence(vs: seq<Unicode.ScalarValue>): (s: WellFormedCodeUnitSeq) | |
| { | |
| var ms := Seq.Map(EncodeScalarValue, vs); | |
| LemmaFlattenMinimalWellFormedCodeUnitSubsequences(ms); | |
| Seq.Flatten(ms) | |
| } | |
| by method { | |
| // Optimize to to avoid allocating the intermediate unflattened sequence. | |
| // We can't quite use Seq.FlatMap easily because we need to prove the result | |
| // is not just a seq<CodeUnit> but a WellFormedCodeUnitSeq. | |
| // TODO: We can be even more efficient by using a JSON.Utils.Vectors.Vector instead. | |
| s := []; | |
| ghost var unflattened: seq<MinimalWellFormedCodeUnitSeq> := []; | |
| for i := |vs| downto 0 | |
| invariant unflattened == Seq.Map(EncodeScalarValue, vs[i..]) | |
| invariant s == Seq.Flatten(unflattened) | |
| { | |
| var next: MinimalWellFormedCodeUnitSeq := EncodeScalarValue(vs[i]); | |
| unflattened := [next] + unflattened; | |
| LemmaPrependMinimalWellFormedCodeUnitSubsequence(next, s); | |
| s := next + s; | |
| } | |
| } | |
| /** | |
| * Returns the scalar value sequence encoded by the given well-formed Unicode string. | |
| */ | |
| function DecodeCodeUnitSequence(s: WellFormedCodeUnitSeq): (vs: seq<Unicode.ScalarValue>) | |
| ensures EncodeScalarSequence(vs) == s | |
| { | |
| var parts := PartitionCodeUnitSequence(s); | |
| var vs := Seq.Map(DecodeMinimalWellFormedCodeUnitSubsequence, parts); | |
| calc == { | |
| s; | |
| Seq.Flatten(parts); | |
| { assert parts == Seq.Map(EncodeScalarValue, vs); } | |
| Seq.Flatten(Seq.Map(EncodeScalarValue, vs)); | |
| EncodeScalarSequence(vs); | |
| } | |
| vs | |
| } | |
| /** | |
| * Returns the scalar value sequence encoded by the given code unit sequence, or None if the given Unicode string | |
| * is not well-formed. | |
| */ | |
| function DecodeCodeUnitSequenceChecked(s: CodeUnitSeq): (maybeVs: Option<seq<Unicode.ScalarValue>>) | |
| ensures IsWellFormedCodeUnitSequence(s) ==> | |
| && maybeVs.Some? | |
| && maybeVs.Extract() == DecodeCodeUnitSequence(s) | |
| ensures !IsWellFormedCodeUnitSequence(s) ==> && maybeVs.None? | |
| { | |
| // IsWellFormedCodeUnitSequence and DecodeCodeUnitSequence each call PartitionCodeUnitSequence, | |
| // so for efficiency we avoid recomputing the partition in the by-method. | |
| if IsWellFormedCodeUnitSequence(s) then Some(DecodeCodeUnitSequence(s)) | |
| else None | |
| } | |
| by method { | |
| var maybeParts := PartitionCodeUnitSequenceChecked(s); | |
| if maybeParts.None? { | |
| return None; | |
| } | |
| var parts := maybeParts.value; | |
| var vs := Seq.Map(DecodeMinimalWellFormedCodeUnitSubsequence, parts); | |
| calc == { | |
| s; | |
| Seq.Flatten(parts); | |
| { assert parts == Seq.Map(EncodeScalarValue, vs); } | |
| Seq.Flatten(Seq.Map(EncodeScalarValue, vs)); | |
| EncodeScalarSequence(vs); | |
| } | |
| return Some(vs); | |
| } | |
| } | |