Skip to content

Commit

Permalink
expression: add builtin func charset() (pingcap#32470)
Browse files Browse the repository at this point in the history 
close pingcap#3931
  • Loading branch information
@Defined2014
Defined2014 authored Feb 21, 2022
1 parent 1b04c3b commit 6aa0b20
Show file tree
Hide file tree
Showing 3 changed files with 71 additions and 43 deletions.
30 changes: 29 additions & 1 deletion expression/builtin_info.go
View file
Original file line number Diff line number Diff line change
Expand Up @@ -616,7 +616,35 @@ type charsetFunctionClass struct {
}

func (c *charsetFunctionClass) getFunction(ctx sessionctx.Context, args []Expression) (builtinFunc, error) {
return nil, errFunctionNotExists.GenWithStackByArgs("FUNCTION", "CHARSET")
if err := c.verifyArgs(args); err != nil {
return nil, err
}
argsTps := make([]types.EvalType, 0, len(args))
for _, arg := range args {
argsTps = append(argsTps, arg.GetType().EvalType())
}
bf, err := newBaseBuiltinFuncWithTp(ctx, c.funcName, args, types.ETString, argsTps...)
if err != nil {
return nil, err
}
bf.tp.Charset, bf.tp.Collate = ctx.GetSessionVars().GetCharsetInfo()
bf.tp.Flen = 64
sig := &builtinCharsetSig{bf}
return sig, nil
}

type builtinCharsetSig struct {
baseBuiltinFunc
}

func (b *builtinCharsetSig) Clone() builtinFunc {
newSig := &builtinCharsetSig{}
newSig.cloneFrom(&b.baseBuiltinFunc)
return newSig
}

func (b *builtinCharsetSig) evalString(_ chunk.Row) (string, bool, error) {
return b.args[0].GetType().Charset, false, nil
}

type coercibilityFunctionClass struct {
Expand Down
6 changes: 3 additions & 3 deletions expression/builtin_info_test.go
View file
Original file line number Diff line number Diff line change
Expand Up @@ -192,9 +192,9 @@ func TestCharset(t *testing.T) {
ctx := createContext(t)
fc := funcs[ast.Charset]
f, err := fc.getFunction(ctx, datumsToConstants(types.MakeDatums(nil)))
require.Nil(t, f)
require.Error(t, err)
require.Regexp(t, "FUNCTION CHARSET does not exist$", err.Error())
require.NotNil(t, f)
require.NoError(t, err)
require.Equal(t, 64, f.getRetTp().Flen)
}

func TestCoercibility(t *testing.T) {
Expand Down
78 changes: 39 additions & 39 deletions expression/integration_test.go
View file
Original file line number Diff line number Diff line change
Expand Up @@ -4176,7 +4176,7 @@ func TestSelectLimitPlanCache(t *testing.T) {
tk.MustQuery("execute stmt").Check(testkit.Rows("1", "2"))
}

func TestCollation(t *testing.T) {
func TestCollationAndCharset(t *testing.T) {
store, clean := testkit.CreateMockStore(t)
defer clean()

Expand All @@ -4186,51 +4186,51 @@ func TestCollation(t *testing.T) {
tk.MustExec("create table t (utf8_bin_c varchar(10) charset utf8 collate utf8_bin, utf8_gen_c varchar(10) charset utf8 collate utf8_general_ci, bin_c binary, num_c int, " +
"abin char collate ascii_bin, lbin char collate latin1_bin, u4bin char collate utf8mb4_bin, u4ci char collate utf8mb4_general_ci)")
tk.MustExec("insert into t values ('a', 'b', 'c', 4, 'a', 'a', 'a', 'a')")
tk.MustQuery("select collation(null)").Check(testkit.Rows("binary"))
tk.MustQuery("select collation(2)").Check(testkit.Rows("binary"))
tk.MustQuery("select collation(2 + 'a')").Check(testkit.Rows("binary"))
tk.MustQuery("select collation(2 + utf8_gen_c) from t").Check(testkit.Rows("binary"))
tk.MustQuery("select collation(2 + utf8_bin_c) from t").Check(testkit.Rows("binary"))
tk.MustQuery("select collation(concat(utf8_bin_c, 2)) from t").Check(testkit.Rows("utf8_bin"))
tk.MustQuery("select collation(concat(utf8_gen_c, 'abc')) from t").Check(testkit.Rows("utf8_general_ci"))
tk.MustQuery("select collation(concat(utf8_gen_c, null)) from t").Check(testkit.Rows("utf8_general_ci"))
tk.MustQuery("select collation(concat(utf8_gen_c, num_c)) from t").Check(testkit.Rows("utf8_general_ci"))
tk.MustQuery("select collation(concat(utf8_bin_c, utf8_gen_c)) from t").Check(testkit.Rows("utf8_bin"))
tk.MustQuery("select collation(upper(utf8_bin_c)) from t").Check(testkit.Rows("utf8_bin"))
tk.MustQuery("select collation(upper(utf8_gen_c)) from t").Check(testkit.Rows("utf8_general_ci"))
tk.MustQuery("select collation(upper(bin_c)) from t").Check(testkit.Rows("binary"))
tk.MustQuery("select collation(concat(abin, bin_c)) from t").Check(testkit.Rows("binary"))
tk.MustQuery("select collation(concat(lbin, bin_c)) from t").Check(testkit.Rows("binary"))
tk.MustQuery("select collation(concat(utf8_bin_c, bin_c)) from t").Check(testkit.Rows("binary"))
tk.MustQuery("select collation(concat(utf8_gen_c, bin_c)) from t").Check(testkit.Rows("binary"))
tk.MustQuery("select collation(concat(u4bin, bin_c)) from t").Check(testkit.Rows("binary"))
tk.MustQuery("select collation(concat(u4ci, bin_c)) from t").Check(testkit.Rows("binary"))
tk.MustQuery("select collation(concat(abin, u4bin)) from t").Check(testkit.Rows("utf8mb4_bin"))
tk.MustQuery("select collation(concat(lbin, u4bin)) from t").Check(testkit.Rows("utf8mb4_bin"))
tk.MustQuery("select collation(concat(utf8_bin_c, u4bin)) from t").Check(testkit.Rows("utf8mb4_bin"))
tk.MustQuery("select collation(concat(utf8_gen_c, u4bin)) from t").Check(testkit.Rows("utf8mb4_bin"))
tk.MustQuery("select collation(concat(u4ci, u4bin)) from t").Check(testkit.Rows("utf8mb4_bin"))
tk.MustQuery("select collation(concat(abin, u4ci)) from t").Check(testkit.Rows("utf8mb4_general_ci"))
tk.MustQuery("select collation(concat(lbin, u4ci)) from t").Check(testkit.Rows("utf8mb4_general_ci"))
tk.MustQuery("select collation(concat(utf8_bin_c, u4ci)) from t").Check(testkit.Rows("utf8mb4_general_ci"))
tk.MustQuery("select collation(concat(utf8_gen_c, u4ci)) from t").Check(testkit.Rows("utf8mb4_general_ci"))
tk.MustQuery("select collation(concat(abin, utf8_bin_c)) from t").Check(testkit.Rows("utf8_bin"))
tk.MustQuery("select collation(concat(lbin, utf8_bin_c)) from t").Check(testkit.Rows("utf8_bin"))
tk.MustQuery("select collation(concat(utf8_gen_c, utf8_bin_c)) from t").Check(testkit.Rows("utf8_bin"))
tk.MustQuery("select collation(concat(abin, utf8_gen_c)) from t").Check(testkit.Rows("utf8_general_ci"))
tk.MustQuery("select collation(concat(lbin, utf8_gen_c)) from t").Check(testkit.Rows("utf8_general_ci"))
tk.MustQuery("select collation(concat(abin, lbin)) from t").Check(testkit.Rows("latin1_bin"))
tk.MustQuery("select collation(null), charset(null)").Check(testkit.Rows("binary binary"))
tk.MustQuery("select collation(2), charset(2)").Check(testkit.Rows("binary binary"))
tk.MustQuery("select collation(2 + 'a'), charset(2 + 'a')").Check(testkit.Rows("binary binary"))
tk.MustQuery("select collation(2 + utf8_gen_c), charset(2 + utf8_gen_c) from t").Check(testkit.Rows("binary binary"))
tk.MustQuery("select collation(2 + utf8_bin_c), charset(2 + utf8_bin_c) from t").Check(testkit.Rows("binary binary"))
tk.MustQuery("select collation(concat(utf8_bin_c, 2)), charset(concat(utf8_bin_c, 2)) from t").Check(testkit.Rows("utf8_bin utf8"))
tk.MustQuery("select collation(concat(utf8_gen_c, 'abc')), charset(concat(utf8_gen_c, 'abc')) from t").Check(testkit.Rows("utf8_general_ci utf8"))
tk.MustQuery("select collation(concat(utf8_gen_c, null)), charset(concat(utf8_gen_c, null)) from t").Check(testkit.Rows("utf8_general_ci utf8"))
tk.MustQuery("select collation(concat(utf8_gen_c, num_c)), charset(concat(utf8_gen_c, num_c)) from t").Check(testkit.Rows("utf8_general_ci utf8"))
tk.MustQuery("select collation(concat(utf8_bin_c, utf8_gen_c)), charset(concat(utf8_bin_c, utf8_gen_c)) from t").Check(testkit.Rows("utf8_bin utf8"))
tk.MustQuery("select collation(upper(utf8_bin_c)), charset(upper(utf8_bin_c)) from t").Check(testkit.Rows("utf8_bin utf8"))
tk.MustQuery("select collation(upper(utf8_gen_c)), charset(upper(utf8_gen_c)) from t").Check(testkit.Rows("utf8_general_ci utf8"))
tk.MustQuery("select collation(upper(bin_c)), charset(upper(bin_c)) from t").Check(testkit.Rows("binary binary"))
tk.MustQuery("select collation(concat(abin, bin_c)), charset(concat(abin, bin_c)) from t").Check(testkit.Rows("binary binary"))
tk.MustQuery("select collation(concat(lbin, bin_c)), charset(concat(lbin, bin_c)) from t").Check(testkit.Rows("binary binary"))
tk.MustQuery("select collation(concat(utf8_bin_c, bin_c)), charset(concat(utf8_bin_c, bin_c)) from t").Check(testkit.Rows("binary binary"))
tk.MustQuery("select collation(concat(utf8_gen_c, bin_c)), charset(concat(utf8_gen_c, bin_c)) from t").Check(testkit.Rows("binary binary"))
tk.MustQuery("select collation(concat(u4bin, bin_c)), charset(concat(u4bin, bin_c)) from t").Check(testkit.Rows("binary binary"))
tk.MustQuery("select collation(concat(u4ci, bin_c)), charset(concat(u4ci, bin_c)) from t").Check(testkit.Rows("binary binary"))
tk.MustQuery("select collation(concat(abin, u4bin)), charset(concat(abin, u4bin)) from t").Check(testkit.Rows("utf8mb4_bin utf8mb4"))
tk.MustQuery("select collation(concat(lbin, u4bin)), charset(concat(lbin, u4bin)) from t").Check(testkit.Rows("utf8mb4_bin utf8mb4"))
tk.MustQuery("select collation(concat(utf8_bin_c, u4bin)), charset(concat(utf8_bin_c, u4bin)) from t").Check(testkit.Rows("utf8mb4_bin utf8mb4"))
tk.MustQuery("select collation(concat(utf8_gen_c, u4bin)), charset(concat(utf8_gen_c, u4bin)) from t").Check(testkit.Rows("utf8mb4_bin utf8mb4"))
tk.MustQuery("select collation(concat(u4ci, u4bin)), charset(concat(u4ci, u4bin)) from t").Check(testkit.Rows("utf8mb4_bin utf8mb4"))
tk.MustQuery("select collation(concat(abin, u4ci)), charset(concat(abin, u4ci)) from t").Check(testkit.Rows("utf8mb4_general_ci utf8mb4"))
tk.MustQuery("select collation(concat(lbin, u4ci)), charset(concat(lbin, u4ci)) from t").Check(testkit.Rows("utf8mb4_general_ci utf8mb4"))
tk.MustQuery("select collation(concat(utf8_bin_c, u4ci)), charset(concat(utf8_bin_c, u4ci)) from t").Check(testkit.Rows("utf8mb4_general_ci utf8mb4"))
tk.MustQuery("select collation(concat(utf8_gen_c, u4ci)), charset(concat(utf8_gen_c, u4ci)) from t").Check(testkit.Rows("utf8mb4_general_ci utf8mb4"))
tk.MustQuery("select collation(concat(abin, utf8_bin_c)), charset(concat(abin, utf8_bin_c)) from t").Check(testkit.Rows("utf8_bin utf8"))
tk.MustQuery("select collation(concat(lbin, utf8_bin_c)), charset(concat(lbin, utf8_bin_c)) from t").Check(testkit.Rows("utf8_bin utf8"))
tk.MustQuery("select collation(concat(utf8_gen_c, utf8_bin_c)), charset(concat(utf8_gen_c, utf8_bin_c)) from t").Check(testkit.Rows("utf8_bin utf8"))
tk.MustQuery("select collation(concat(abin, utf8_gen_c)), charset(concat(abin, utf8_gen_c)) from t").Check(testkit.Rows("utf8_general_ci utf8"))
tk.MustQuery("select collation(concat(lbin, utf8_gen_c)), charset(concat(lbin, utf8_gen_c)) from t").Check(testkit.Rows("utf8_general_ci utf8"))
tk.MustQuery("select collation(concat(abin, lbin)), charset(concat(abin, lbin)) from t").Check(testkit.Rows("latin1_bin latin1"))

tk.MustExec("set names utf8mb4 collate utf8mb4_bin")
tk.MustQuery("select collation('a')").Check(testkit.Rows("utf8mb4_bin"))
tk.MustQuery("select collation('a'), charset('a')").Check(testkit.Rows("utf8mb4_bin utf8mb4"))
tk.MustExec("set names utf8mb4 collate utf8mb4_general_ci")
tk.MustQuery("select collation('a')").Check(testkit.Rows("utf8mb4_general_ci"))
tk.MustQuery("select collation('a'), charset('a')").Check(testkit.Rows("utf8mb4_general_ci utf8mb4"))

tk.MustExec("set names utf8mb4 collate utf8mb4_general_ci")
tk.MustExec("set @test_collate_var = 'a'")
tk.MustQuery("select collation(@test_collate_var)").Check(testkit.Rows("utf8mb4_general_ci"))
tk.MustQuery("select collation(@test_collate_var), charset(@test_collate_var)").Check(testkit.Rows("utf8mb4_general_ci utf8mb4"))
tk.MustExec("set @test_collate_var = concat(\"a\", \"b\" collate utf8mb4_bin)")
tk.MustQuery("select collation(@test_collate_var)").Check(testkit.Rows("utf8mb4_bin"))
tk.MustQuery("select collation(@test_collate_var), charset(@test_collate_var)").Check(testkit.Rows("utf8mb4_bin utf8mb4"))

tk.MustQuery("select locate('1', '123' collate utf8mb4_bin, 2 collate `binary`);").Check(testkit.Rows("0"))
tk.MustQuery("select 1 in ('a' collate utf8mb4_bin, 'b' collate utf8mb4_general_ci);").Check(testkit.Rows("0"))
Expand Down

0 comments on commit 6aa0b20

Please sign in to comment.