parser, infoschema, executor: Add information_schema.keywords (#48807)

.gitignore

@@ -38,3 +38,5 @@ bazel-tidb
 MODULE.bazel.lock
 .ijwb/
 /oom_record/
+*.log.json
+genkeyword

pkg/executor/builder.go

@@ -2097,7 +2097,8 @@ func (b *executorBuilder) buildMemTable(v *plannercore.PhysicalMemTable) exec.Ex
 			strings.ToLower(infoschema.ClusterTableMemoryUsageOpsHistory),
 			strings.ToLower(infoschema.TableResourceGroups),
 			strings.ToLower(infoschema.TableRunawayWatches),
-			strings.ToLower(infoschema.TableCheckConstraints):
+			strings.ToLower(infoschema.TableCheckConstraints),
+			strings.ToLower(infoschema.TableKeywords):
 			memTracker := memory.NewTracker(v.ID(), -1)
 			memTracker.AttachTo(b.ctx.GetSessionVars().StmtCtx.MemTracker)
 			return &MemTableReaderExec{

pkg/executor/infoschema_reader.go

@@ -43,6 +43,7 @@ import (
 	"github.com/pingcap/tidb/pkg/infoschema"
 	"github.com/pingcap/tidb/pkg/kv"
 	"github.com/pingcap/tidb/pkg/meta/autoid"
+	"github.com/pingcap/tidb/pkg/parser"
 	"github.com/pingcap/tidb/pkg/parser/charset"
 	"github.com/pingcap/tidb/pkg/parser/model"
 	"github.com/pingcap/tidb/pkg/parser/mysql"
@@ -197,6 +198,8 @@ func (e *memtableRetriever) retrieve(ctx context.Context, sctx sessionctx.Contex
 			err = e.setDataFromRunawayWatches(sctx)
 		case infoschema.TableCheckConstraints:
 			err = e.setDataFromCheckConstraints(sctx, dbs)
+		case infoschema.TableKeywords:
+			err = e.setDataFromKeywords()
 		}
 		if err != nil {
 			return nil, err
@@ -3390,6 +3393,16 @@ func (e *memtableRetriever) setDataFromResourceGroups() error {
 	return nil
 }
 
+func (e *memtableRetriever) setDataFromKeywords() error {
+	rows := make([][]types.Datum, 0, len(parser.Keywords))
+	for _, kw := range parser.Keywords {
+		row := types.MakeDatums(kw.Word, kw.Reserved)
+		rows = append(rows, row)
+	}
+	e.rows = rows
+	return nil
+}
+
 func checkRule(rule *label.Rule) (dbName, tableName string, partitionName string, err error) {
 	s := strings.Split(rule.ID, "/")
 	if len(s) < 3 {

pkg/executor/infoschema_reader_internal_test.go

@@ -71,3 +71,11 @@ func TestSetDataFromCheckConstraints(t *testing.T) {
 	require.Equal(t, types.NewStringDatum("t2_c1"), mt.rows[0][2])
 	require.Equal(t, types.NewStringDatum("(id<10)"), mt.rows[0][3])
 }
+
+func TestSetDataFromKeywords(t *testing.T) {
+	mt := memtableRetriever{}
+	err := mt.setDataFromKeywords()
+	require.NoError(t, err)
+	require.Equal(t, types.NewStringDatum("ADD"), mt.rows[0][0]) // Keyword: ADD
+	require.Equal(t, types.NewIntDatum(1), mt.rows[0][1])        // Reserved: true(1)
+}

pkg/infoschema/tables.go

@@ -209,6 +209,8 @@ const (
 	TableRunawayWatches = "RUNAWAY_WATCHES"
 	// TableCheckConstraints is the list of CHECK constraints.
 	TableCheckConstraints = "CHECK_CONSTRAINTS"
+	// TableKeywords is the list of keywords.
+	TableKeywords = "KEYWORDS"
 )
 
 const (
@@ -318,6 +320,7 @@ var tableIDMap = map[string]int64{
 	TableResourceGroups:                  autoid.InformationSchemaDBID + 88,
 	TableRunawayWatches:                  autoid.InformationSchemaDBID + 89,
 	TableCheckConstraints:                autoid.InformationSchemaDBID + 90,
+	TableKeywords:                        autoid.InformationSchemaDBID + 92,
 }
 
 // columnInfo represents the basic column information of all kinds of INFORMATION_SCHEMA tables
@@ -1648,6 +1651,11 @@ var tableCheckConstraintsCols = []columnInfo{
 	{name: "CHECK_CLAUSE", tp: mysql.TypeLongBlob, size: types.UnspecifiedLength, flag: mysql.NotNullFlag},
 }
 
+var tableKeywords = []columnInfo{
+	{name: "WORD", tp: mysql.TypeVarchar, size: 128},
+	{name: "RESERVED", tp: mysql.TypeLong, size: 11},
+}
+
 // GetShardingInfo returns a nil or description string for the sharding information of given TableInfo.
 // The returned description string may be:
 //   - "NOT_SHARDED": for tables that SHARD_ROW_ID_BITS is not specified.
@@ -2166,6 +2174,7 @@ var tableNameToColumns = map[string][]columnInfo{
 	TableResourceGroups:                     tableResourceGroupsCols,
 	TableRunawayWatches:                     tableRunawayWatchListCols,
 	TableCheckConstraints:                   tableCheckConstraintsCols,
+	TableKeywords:                           tableKeywords,
 }
 
 func createInfoSchemaTable(_ autoid.Allocators, meta *model.TableInfo) (table.Table, error) {

pkg/parser/BUILD.bazel

@@ -9,8 +9,10 @@ go_library(
     name = "parser",
     srcs = [
         "digester.go",
+        "generate.go",
         "hintparser.go",
         "hintparserimpl.go",
+        "keywords.go",
         "lexer.go",
         "misc.go",
         "parser.go",
@@ -41,6 +43,7 @@ go_test(
         "consistent_test.go",
         "digester_test.go",
         "hintparser_test.go",
+        "keywords_test.go",
         "lexer_test.go",
         "main_test.go",
         "parser_test.go",

pkg/parser/Makefile

@@ -1,12 +1,18 @@
 .PHONY: all parser clean
 
-all: fmt parser
+all: fmt parser generate
 
 test: fmt parser
 	sh test.sh
 
 parser: parser.go hintparser.go
 
+genkeyword: generate_keyword/genkeyword.go
+	go build -C generate_keyword -o ../genkeyword
+
+generate: genkeyword parser.y
+	go generate
+
 %arser.go: prefix = $(@:parser.go=)
 %arser.go: %arser.y bin/goyacc
 	@echo "bin/goyacc -o $@ -p yy$(prefix) -t $(prefix)Parser $<"

pkg/parser/generate.go

@@ -0,0 +1,4 @@
+package parser
+
+// Generate keywords.go
+//go:generate ./genkeyword

pkg/parser/generate_keyword/BUILD.bazel

@@ -1,4 +1,4 @@
-load("@io_bazel_rules_go//go:def.bzl", "go_binary", "go_library")
+load("@io_bazel_rules_go//go:def.bzl", "go_binary", "go_library", "go_test")
 
 go_library(
     name = "generate_keyword_lib",
@@ -12,3 +12,12 @@ go_binary(
     embed = [":generate_keyword_lib"],
     visibility = ["//visibility:public"],
 )
+
+go_test(
+    name = "generate_keyword_test",
+    timeout = "short",
+    srcs = ["genkeyword_test.go"],
+    embed = [":generate_keyword_lib"],
+    flaky = True,
+    deps = ["@com_github_stretchr_testify//require"],
+)

pkg/parser/generate_keyword/genkeyword_test.go

@@ -0,0 +1,15 @@
+package main
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/require"
+)
+
+func TestParseLine(t *testing.T) {
+	add := parseLine("	add               \"ADD\"")
+	require.Equal(t, add, "ADD")
+
+	tso := parseLine("	tidbCurrentTSO    \"TiDB_CURRENT_TSO\"")
+	require.Equal(t, tso, "TiDB_CURRENT_TSO")
+}