Add the first 31 bytes to the hash of long strings

Author: michaelmior

src/core/json/include/sourcemeta/core/json_hash.h

@@ -49,7 +49,6 @@ template <typename T> struct PropertyHashJSON {
       -> hash_type {
     hash_type result;
     assert(!value.empty());
-    assert(value.size() <= 31);
     // Copy starting a byte 2
     std::memcpy(reinterpret_cast<char *>(&result) + 1, value.data(), size);
     return result;
@@ -126,17 +125,19 @@ template <typename T> struct PropertyHashJSON {
         // This case is specifically designed to be constant with regards to
         // string length, and to exploit the fact that most JSON objects don't
         // have a lot of entries, so hash collision is not as common
-        return {(size + static_cast<typename hash_type::type>(value.front()) +
-                 static_cast<typename hash_type::type>(value.back())) %
-                // Make sure the property hash can never exceed 8 bits
-                256};
+        auto hash = this->perfect(value, 31);
+        hash.a |= (size + static_cast<typename hash_type::type>(value.front()) +
+                   static_cast<typename hash_type::type>(value.back())) %
+                  // Make sure the property hash can never exceed 8 bits
+                  256;
+        return hash;
     }
   }
 
   inline auto is_perfect(const hash_type &hash) const noexcept -> bool {
     // If there is anything written past the first byte,
     // then it is a perfect hash
-    return hash.a > 255;
+    return (hash.a & 255) == 0;
   }
 };
 

test/json/json_hash_test.cc

@@ -7,7 +7,7 @@ TEST(JSON_key_hash, hash_empty) {
       hasher;
   const sourcemeta::core::JSON::String value{""};
   const auto hash{hasher(value)};
-  EXPECT_FALSE(hasher.is_perfect(hash));
+  EXPECT_TRUE(hasher.is_perfect(hash));
 #if defined(__SIZEOF_INT128__)
   EXPECT_EQ(hash.a,
             (__uint128_t{0x0000000000000000} << 64) | 0x0000000000000000);
@@ -619,14 +619,14 @@ TEST(JSON_key_hash, hash_fooooooooooooooooooooooooooooooo) {
   EXPECT_FALSE(hasher.is_perfect(hash));
 #if defined(__SIZEOF_INT128__)
   EXPECT_EQ(hash.a,
-            (__uint128_t{0x0000000000000000} << 64) | 0x00000000000000f5);
+            (__uint128_t{0x6f6f6f6f6f6f6f6f} << 64) | 0x6f6f6f6f6f6f66f5);
   EXPECT_EQ(hash.b,
-            (__uint128_t{0x0000000000000000} << 64) | 0x0000000000000000);
+            (__uint128_t{0x6f6f6f6f6f6f6f6f} << 64) | 0x6f6f6f6f6f6f6f6f);
 #else
   // 0x20 (length) + 0x66 (f) + 0x6f (o)
-  EXPECT_EQ(hash.a, 0x00000000000000f5);
-  EXPECT_EQ(hash.b, 0x0000000000000000);
-  EXPECT_EQ(hash.c, 0x0000000000000000);
-  EXPECT_EQ(hash.d, 0x0000000000000000);
+  EXPECT_EQ(hash.a, 0x6f6f6f6f6f6f66f5);
+  EXPECT_EQ(hash.b, 0x6f6f6f6f6f6f6f6f);
+  EXPECT_EQ(hash.c, 0x6f6f6f6f6f6f6f6f);
+  EXPECT_EQ(hash.d, 0x6f6f6f6f6f6f6f6f);
 #endif
 }