WL#13094 Support new utf8mb4 collations

Also added unit test that checks if all collations reported by server are covered
by our logic.

Author: rsomla1

cdk/include/mysql/cdk/protocol/mysqlx/collations.h

@@ -302,6 +302,12 @@
   X(utf8mb4,247,vietnamese,ci) \
   X(utf8mb4,277,vi_0900,ai_ci) \
   X(utf8mb4,300,vi_0900,as_cs) \
+  X(utf8mb4,304,ja_0900,as_cs_ks) \
+  X(utf8mb4,305,uca0900,as_ci) \
+  X(utf8mb4,306,ru_0900,ai_ci) \
+  X(utf8mb4,307,ru_0900,as_cs) \
+  X(utf8mb4,308,zh_0900,as_cs) \
+  X(utf8mb4,309,uca0900,bin) \
   COLLATIONS_utf8mb4_EXTRA
 
 #define COLLATIONS_utf8mb4_EXTRA

devapi/result.cc

@@ -281,8 +281,10 @@ struct CollationInfo::Access
   enum coll_case {
     case_ci = CollationInfo::case_ci,
     case_ai_ci = case_ci,
+    case_as_ci = case_ci,
     case_cs = CollationInfo::case_cs,
     case_as_cs = case_cs,
+    case_as_cs_ks = case_cs,
     case_bin = CollationInfo::case_bin
   };
 
@@ -317,11 +319,6 @@ coll_name(
 {
   static std::vector<std::string> special;
 
-  // Note: special exception for "binary" collation (no _bin suffix)
-
-  if (sensitivity == "bin")
-    return cs == "binary" ? "binary" : name_bin;
-
   /*
     For generic UCA collations, such as uca0900, the "uca" prefix is
     not present in the MySQL collation name. For example, for the uca0900
@@ -335,7 +332,14 @@ coll_name(
     special.push_back(cs + "_" + coll.substr(3) + "_" + sensitivity);
     return special.back().c_str();
   }
-  return name;
+
+  if (sensitivity == "bin")
+  {
+    // Note: special exception for "binary" collation (no _bin suffix)
+    return cs == "binary" ? "binary" : name_bin;
+  }
+  else
+    return name;
 }
 
 

devapi/tests/types-t.cc

@@ -497,6 +497,56 @@ TEST_F(Types, string)
 }
 
 
+inline
+const CollationInfo* get_collation(unsigned id)
+{
+#define COLL_FIND(COL) COLLATIONS_##COL(COLL_FIND1)
+#define COLL_FIND1(CS,ID,COLL,CASE) \
+  case ID: return &Collation<CharacterSet::CS>::COLL_CONST_NAME(COLL,CASE);
+
+  switch (id)
+  {
+    CDK_CS_LIST(COLL_FIND)
+  default:
+    return nullptr;
+  }
+}
+
+
+TEST_F(Types, collations)
+{
+  using col_data = std::pair<unsigned, string>;
+  std::vector<col_data> unknown;
+
+  Table t = getSchema("information_schema").getTable("collations");
+
+  for (Row r : t.select("id", "collation_name").execute())
+  {
+    col_data col = { r[0], r[1] };
+
+    const CollationInfo *info = get_collation(col.first);
+
+    if (!info)
+      unknown.push_back(col);
+    else
+    {
+      EXPECT_EQ(std::string{ col.second }, std::string{ info->getName() })
+        << "bad collation name";
+    }
+  }
+
+  if (!unknown.empty())
+  {
+    cout << "Unknown collations:" << endl;
+    for (col_data col : unknown)
+    {
+      cout << " -" << col.first << ": " << col.second << endl;
+    }
+    FAIL() << "There are unknown collations";
+  }
+}
+
+
 TEST_F(Types, blob)
 {
   SKIP_IF_NO_XPLUGIN;

include/mysqlx/devapi/collations.h

@@ -187,6 +187,8 @@ static PUBLIC_API const CollationInfo COLL_CONST_NAME(COLL,CASE);
 #define COLL_CONST_NAME_ai_ci(COLL) COLL##_ai_ci
 #define COLL_CONST_NAME_cs(COLL)    COLL##_cs
 #define COLL_CONST_NAME_as_cs(COLL) COLL##_as_cs
+#define COLL_CONST_NAME_as_ci(COLL) COLL##_as_ci
+#define COLL_CONST_NAME_as_cs_ks(COLL) COLL##_as_cs_ks
 
 // Add utf8mb4 alias for bin collation for compatibility
 

include/mysqlx/devapi/mysql_collations.h

@@ -302,6 +302,12 @@
   X(utf8mb4,247,vietnamese,ci) \
   X(utf8mb4,277,vi_0900,ai_ci) \
   X(utf8mb4,300,vi_0900,as_cs) \
+  X(utf8mb4,304,ja_0900,as_cs_ks) \
+  X(utf8mb4,305,uca0900,as_ci) \
+  X(utf8mb4,306,ru_0900,ai_ci) \
+  X(utf8mb4,307,ru_0900,as_cs) \
+  X(utf8mb4,308,zh_0900,as_cs) \
+  X(utf8mb4,309,uca0900,bin) \
   COLLATIONS_utf8mb4_EXTRA
 
 #define COLLATIONS_utf8mb4_EXTRA