sync

Author: Lucas McDonald

DynamoDbEncryption/dafny/DynamoDbEncryption/test/DynamoToStruct.dfy

@@ -376,14 +376,15 @@ module DynamoToStructTest {
   //= type=test
   //# Entries in a String Set MUST be ordered in ascending [UTF-16 binary order](./string-ordering.md#utf-16-binary-order).
   method {:test} TestSortSSAttr() {
+    // "\ud800\udc02" <-> "𐀂"
     var stringSetValue := AttributeValue.SS(["&","｡","\ud800\udc02"]);
     // Note that string values are UTF-8 encoded, but sorted by UTF-16 encoding.
     var encodedStringSetData := StructuredDataTerminal(value := [
                                                          0,0,0,3, // 3 entries in set
                                                          0,0,0,1, // 1st entry is 1 byte
                                                          0x26, // "&" in UTF-8 encoding
                                                          0,0,0,4, // 2nd entry is 4 bytes
-                                                         0xF0,0x90,0x80,0x82, // "\ud800\udc02" in UTF-8 encoding
+                                                         0xF0,0x90,0x80,0x82, // "𐀂" in UTF-8 encoding
                                                          0,0,0,3, // 3rd entry is 3 bytes
                                                          0xEF,0xBD,0xA1 // "｡" in UTF-8 encoding
                                                        ],
@@ -395,6 +396,7 @@ module DynamoToStructTest {
 
     var newStringSetValue := StructuredToAttr(encodedStringSetData);
     expect newStringSetValue.Success?;
+    // "\ud800\udc02" <-> "𐀂"
     expect newStringSetValue.value == AttributeValue.SS(["&","\ud800\udc02","｡"]);
   }
 
@@ -415,10 +417,12 @@ module DynamoToStructTest {
 
   method {:test} TestSetsInListAreSorted() {
     var nSetValue := AttributeValue.NS(["2","1","10"]);
+    // "\ud800\udc02" <-> "𐀂"
     var sSetValue := AttributeValue.SS(["&","｡","\ud800\udc02"]);
     var bSetValue := AttributeValue.BS([[1,0],[1],[2]]);
 
     var sortedNSetValue := AttributeValue.NS(["1","10","2"]);
+    // "\ud800\udc02" <-> "𐀂"
     var sortedSSetValue := AttributeValue.SS(["&","\ud800\udc02","｡"]);
     var sortedBSetValue := AttributeValue.BS([[1],[1,0],[2]]);
 
@@ -444,10 +448,12 @@ module DynamoToStructTest {
 
   method {:test} TestSetsInMapAreSorted() {
     var nSetValue := AttributeValue.NS(["2","1","10"]);
+    // "\ud800\udc02" <-> "𐀂"
     var sSetValue := AttributeValue.SS(["&","｡","\ud800\udc02"]);
     var bSetValue := AttributeValue.BS([[1,0],[1],[2]]);
 
     var sortedNSetValue := AttributeValue.NS(["1","10","2"]);
+    // "\ud800\udc02" <-> "𐀂"
     var sortedSSetValue := AttributeValue.SS(["&","\ud800\udc02","｡"]);
     var sortedBSetValue := AttributeValue.BS([[1],[1,0],[2]]);
 
@@ -490,6 +496,7 @@ module DynamoToStructTest {
   method {:test} TestSortMapKeys() {
     var nullValue := AttributeValue.NULL(true);
 
+    // "\ud800\udc02" <-> "𐀂"
     var mapValue := AttributeValue.M(map["&" := nullValue, "｡" := nullValue, "\ud800\udc02" := nullValue]);
 
     // Note that the string values are encoded as UTF-8, but are sorted according to UTF-16 encoding.
@@ -500,7 +507,7 @@ module DynamoToStructTest {
         0x26, // "&" UTF-8 encoded
         0,0, 0,0,0,0, // null value
         0,1, 0,0,0,4, // 2nd key is a string 4 bytes long
-        0xF0, 0x90, 0x80, 0x82, // "\ud800\udc02" UTF-8 encoded
+        0xF0, 0x90, 0x80, 0x82, // "𐀂" UTF-8 encoded
         0,0, 0,0,0,0, // null value
         0,1, 0,0,0,3, // 3rd key is a string 3 bytes long
         0xEF, 0xBD, 0xA1, // "｡"

TestVectors/dafny/DDBEncryption/src/WriteManifest.dfy

@@ -218,21 +218,20 @@ module {:options "-functionSyntax:4"} WriteManifest {
   const DoNothing : CryptoAction := 3
 
   const A : string := "A"
-  const B : string := "\ud000" // Ud000"
-  const C : string := "\ufe4c" // Ufe4c"
-  const D : string := "\u100001" // U10001
-  const E : string := "\u100002" // U10002 - same high surrogate as D
-  const F : string := "\u200002" // U20002 - different high surrogate as D
+  const B : string := "\ud000" // "Ud000" <-> "퀀"
+  const C : string := "\ufe4c" // "Ufe4c" <-> "﹌"
+  const D : string := "\u10001" // "U10001" <-> "𐀁" (surrogate pair: "\uD800\uDC01")
+  const E : string := "\u10002" // "U10002" <-> "𐀂" (same high surrogate as D: "\uD800\uDC02")
+  const F : string := "\u20002" // "U20002" <-> "𠀂"  (different high surrogate as D: "\D840\uDC02")
 
-  // Dafny doesn't handle unicode surrogates correctly.
-  // lemma CheckLengths()
-  //   ensures |A| == 1
-  //   ensures |B| == 1
-  //   ensures |C| == 1
-  //   ensures |D| == 2
-  //   ensures |E| == 2
-  //   ensures |F| == 2
-  // {}
+  lemma CheckLengths()
+    ensures |A| == 1
+    ensures |B| == 1
+    ensures |C| == 1
+    ensures |D| == 2
+    ensures |E| == 2
+    ensures |F| == 2
+  {}
 
   // Let's make attribute names with complex characters.
   // It shouldn't matter, but let's make sure