stdlib: Don't hash in the length of the string

Some characters are not considered when comparing strings. Hashing in the length
would cause a difference of hash values in equal strings.

Swift SVN r31468

Author: aschwaighofer

stdlib/public/runtime/UnicodeNormalization.cpp

@@ -138,6 +138,10 @@ static intptr_t hashChunk(const UCollator *Collator, intptr_t HashState,
     Collator, Str, Length, ErrorCode);
   while (U_SUCCESS(*ErrorCode)) {
     intptr_t Elem = ucol_next(CollationIterator, ErrorCode);
+    // Ignore zero valued collation elements. They don't participate in the
+    // ordering relation.
+    if (Elem == 0)
+      continue;
     if (Elem != UCOL_NULLORDER) {
       Elem *= HASH_M;
       Elem ^= Elem >> HASH_R;
@@ -162,7 +166,7 @@ static intptr_t hashFinish(intptr_t HashState) {
 extern "C"
 intptr_t _swift_stdlib_unicode_hash(const uint16_t *Str, int32_t Length) {
   UErrorCode ErrorCode = U_ZERO_ERROR;
-  intptr_t HashState = HASH_SEED ^ (Length * HASH_M);
+  intptr_t HashState = HASH_SEED;
   HashState = hashChunk(GetRootCollator(), HashState, Str, Length, &ErrorCode);
 
   if (U_FAILURE(ErrorCode)) {
@@ -176,7 +180,7 @@ extern "C"
 intptr_t _swift_stdlib_unicode_hash_ascii(const char *Str, int32_t Length) {
   UErrorCode ErrorCode = U_ZERO_ERROR;
   const UCollator *Collator = GetRootCollator();
-  intptr_t HashState = HASH_SEED ^ (Length * HASH_M);
+  intptr_t HashState = HASH_SEED;
   uint16_t HashBuffer[ASCII_HASH_BUFFER_SIZE];
 
   int32_t Pos = 0;

validation-test/stdlib/HashingICU.swift

@@ -7,18 +7,14 @@
 // use StdlibUnittest because that doesn't work on linux yet. May go away in
 // favour of the more comprehensive tests that already exist once it does.
 
-// ASCII strings
-// CHECK: 5308980208032766932
-print("boom".hashValue)
-// CHECK-NEXT: 6894346571320922064
-print("zoom".hashValue)
-
-// Unicode strings
-// CHECK-NEXT: 3514641426931780352
-print("ZOO≪M".hashValue)
-// CHECK-NEXT: 7349636929305805742
-print("moo≪m".hashValue)
-
 // Let's not crash on changing case.
 let upper = "\u{00df}".uppercaseString
 let lower = "\u{0130}".lowercaseString
+
+// ASCII strings
+// CHECK: true
+print("abc".hashValue == "\0abc".hashValue)
+
+// Unicode strings
+// CHECK-NEXT: true
+print("abc\u{0130}".hashValue == "\0abc\u{0130}".hashValue)