sql: add plpgsql parser for telemetry purposes

This commit  removes portions of the plpgparser
that arent in use or could be shared with the sql
parser/lexer.

Release note: None

pkg/BUILD.bazel

@@ -1797,7 +1797,8 @@ GO_TARGETS = [
     "//pkg/sql/physicalplan/replicaoracle:replicaoracle_test",
     "//pkg/sql/physicalplan:physicalplan",
     "//pkg/sql/physicalplan:physicalplan_test",
-    "//pkg/sql/plpgsql/parser/lexbase:lexbase",
+    "//pkg/sql/plpgsql/parser/plpgsqllexbase/allkeywords:allkeywords",
+    "//pkg/sql/plpgsql/parser/plpgsqllexbase/allkeywords:allkeywords_lib",
     "//pkg/sql/plpgsql/parser/plpgsqllexbase:plpgsqllexbase",
     "//pkg/sql/plpgsql/parser:parser",
     "//pkg/sql/plpgsql/parser:parser_test",
@@ -3021,8 +3022,8 @@ GET_X_DATA_TARGETS = [
     "//pkg/sql/physicalplan:get_x_data",
     "//pkg/sql/physicalplan/replicaoracle:get_x_data",
     "//pkg/sql/plpgsql/parser:get_x_data",
-    "//pkg/sql/plpgsql/parser/lexbase:get_x_data",
     "//pkg/sql/plpgsql/parser/plpgsqllexbase:get_x_data",
+    "//pkg/sql/plpgsql/parser/plpgsqllexbase/allkeywords:get_x_data",
     "//pkg/sql/privilege:get_x_data",
     "//pkg/sql/protoreflect:get_x_data",
     "//pkg/sql/protoreflect/test:get_x_data",

pkg/gen/misc.bzl

@@ -9,9 +9,6 @@ MISC_SRCS = [
     "//pkg/roachprod/vm/aws:terraform/main.tf",
     "//pkg/spanconfig/spanconfigstore:entry_interval_btree.go",
     "//pkg/spanconfig/spanconfigstore:entry_interval_btree_test.go",
-    "//pkg/sql/plpgsql/parser/lexbase:keywords.go",
-    "//pkg/sql/plpgsql/parser/lexbase:reserved_keywords.go",
-    "//pkg/sql/plpgsql/parser/lexbase:tokens.go",
     "//pkg/sql/plpgsql/parser:plpgsql.go",
     "//pkg/sql/scanner:token_names_test.go",
     "//pkg/sql:txnstatetransitions_diagram.gv",

pkg/sql/plpgsql/parser/BUILD.bazel

@@ -14,19 +14,13 @@ genrule(
     ],
     outs = ["plpgsql.go"],
     cmd = """
-    GO_REL_PATH=`dirname $(location @go_sdk//:bin/go)`
-    GO_ABS_PATH=`cd $$GO_REL_PATH && pwd`
-    export PATH=$$GO_ABS_PATH:$$PATH
-    export HOME=$(GENDIR)
-    export GOPATH=/nonexist-gopath
       $(location :plpgsql-gen) $(location plpgsql.y) \
           $(location plpgsql.go) $(location @org_golang_x_tools//cmd/goyacc) \
           $(location @com_github_cockroachdb_gostdlib//x/tools/cmd/goimports)
     """,
     exec_tools = [
         ":plpgsql-gen",
         "@com_github_cockroachdb_gostdlib//x/tools/cmd/goimports",
-        "@go_sdk//:bin/go",
         "@org_golang_x_tools//cmd/goyacc",
     ],
     visibility = ["//visibility:public"],
@@ -48,6 +42,7 @@ go_library(
         "//pkg/sql/pgwire/pgcode",
         "//pkg/sql/pgwire/pgerror",
         "//pkg/sql/plpgsql/parser/plpgsqllexbase",
+        "//pkg/sql/scanner",
         "//pkg/sql/sem/plpgsqltree",
         "//pkg/sql/sem/tree",
         "//pkg/sql/types",
@@ -71,9 +66,8 @@ go_test(
     data = glob(["testdata/**"]),
     deps = [
         ":parser",
-        "//pkg/testutils",
+        "//pkg/testutils/datapathutils",
         "@com_github_cockroachdb_datadriven//:datadriven",
-        "@com_github_stretchr_testify//require",
     ],
 )
 

pkg/sql/plpgsql/parser/lexer.go

@@ -43,8 +43,6 @@ type lexer struct {
 	numAnnotations  tree.AnnotationIdx
 
 	lastError error
-
-	env *plpgsqltree.PLpgSQLEnv
 }
 
 func (l *lexer) init(sql string, tokens []plpgsqlSymType, nakedIntType *types.T) {
@@ -326,9 +324,7 @@ func (l *lexer) ReadSqlConstruct(
 	return strings.Join(exprTokenStrs, " "), terminatorMet
 }
 
-func (l *lexer) ProcessQueryForCursorWithoutExplicitExpr(
-	openStmt *plpgsqltree.PLpgSQLStmtOpen,
-) {
+func (l *lexer) ProcessQueryForCursorWithoutExplicitExpr(openStmt *plpgsqltree.PLpgSQLStmtOpen) {
 	l.lastPos++
 	if int(l.Peek().id) == EXECUTE {
 		dynamicQuery, endToken := l.ReadSqlExpressionStr2(USING, ';')
@@ -378,24 +374,11 @@ func (l *lexer) lastToken() plpgsqlSymType {
 	return l.tokens[l.lastPos]
 }
 
-// NewAnnotation returns a new annotation index.
-func (l *lexer) NewAnnotation() tree.AnnotationIdx {
-	l.numAnnotations++
-	return l.numAnnotations
-}
-
 // SetStmt is called from the parser when the statement is constructed.
 func (l *lexer) SetStmt(stmt plpgsqltree.PLpgSQLStatement) {
 	l.stmt = stmt.(*plpgsqltree.PLpgSQLStmtBlock)
 }
 
-// UpdateNumPlaceholders is called from the parser when a placeholder is constructed.
-func (l *lexer) UpdateNumPlaceholders(p *tree.Placeholder) {
-	if n := int(p.Idx) + 1; l.numPlaceholders < n {
-		l.numPlaceholders = n
-	}
-}
-
 // PurposelyUnimplemented wraps Error, setting lastUnimplementedError.
 func (l *lexer) PurposelyUnimplemented(feature string, reason string) {
 	// We purposely do not use unimp here, as it appends hints to suggest that
@@ -495,12 +478,3 @@ func (l *lexer) populateErrorDetails() {
 	fmt.Fprintf(&buf, "%s^", strings.Repeat(" ", int(lastTok.pos)-j))
 	l.lastError = errors.WithDetail(l.lastError, buf.String())
 }
-
-// specialHelpErrorPrefix is a special prefix that must be present at
-// the start of an error message to be considered a valid help
-// response payload by the CLI shell.
-const specialHelpErrorPrefix = "help token in input"
-
-func (l *lexer) populateHelpMsg(msg string) {
-	l.lastError = errors.WithHint(errors.Wrap(l.lastError, specialHelpErrorPrefix), msg)
-}

pkg/sql/plpgsql/parser/parse.go

@@ -11,9 +11,9 @@
 package parser
 
 import (
-	"github.com/cockroachdb/cockroach/pkg/sql/scanner"
 	"go/constant"
 
+	"github.com/cockroachdb/cockroach/pkg/sql/scanner"
 	"github.com/cockroachdb/cockroach/pkg/sql/sem/plpgsqltree"
 	"github.com/cockroachdb/cockroach/pkg/sql/sem/tree"
 	"github.com/cockroachdb/cockroach/pkg/sql/types"
@@ -70,7 +70,6 @@ type Parser struct {
 	lexer      lexer
 	parserImpl plpgsqlParserImpl
 	tokBuf     [8]plpgsqlSymType
-	stmtBuf    [1]Statement
 }
 
 // INT8 is the historical interpretation of INT. This should be left
@@ -84,20 +83,6 @@ func (p *Parser) Parse(sql string) (Statement, error) {
 	return p.parseWithDepth(1, sql, defaultNakedIntType)
 }
 
-// ParseWithInt parses a sql statement string and returns a list of
-// Statements. The INT token will result in the specified TInt type.
-func (p *Parser) ParseWithInt(sql string, nakedIntType *types.T) (Statement, error) {
-	return p.parseWithDepth(1, sql, nakedIntType)
-}
-
-func (p *Parser) parseOneWithInt(sql string, nakedIntType *types.T) (Statement, error) {
-	stmt, err := p.parseWithDepth(1, sql, nakedIntType)
-	if err != nil {
-		return Statement{}, err
-	}
-	return stmt, nil
-}
-
 func (p *Parser) scanOneStmt() (sql string, tokens []plpgsqlSymType, done bool) {
 	var lval plpgsqlSymType
 	tokens = p.tokBuf[:0]

pkg/sql/plpgsql/parser/parser_test.go

@@ -11,33 +11,13 @@
 package parser_test
 
 import (
-	"github.com/cockroachdb/cockroach/pkg/testutils/datapathutils"
 	"testing"
 
 	"github.com/cockroachdb/cockroach/pkg/sql/plpgsql/parser"
+	"github.com/cockroachdb/cockroach/pkg/testutils/datapathutils"
 	"github.com/cockroachdb/datadriven"
-	"github.com/stretchr/testify/require"
 )
 
-func TestParseDeclareSection(t *testing.T) {
-	fn := `
-DECLARE
-  order_cnt integer := 10;
-BEGIN
-  CASE
-  WHEN order_cnt BETWEEN 0 AND 100 THEN
-    CALL fn(1);
-  WHEN order_cnt > 100 THEN
-    CALL fn(2);
-  ELSE
-    CALL fn(3);
-END CASE;
-END`
-	stmt, err := parser.Parse(fn)
-	require.NoError(t, err)
-	require.Equal(t, "DECLARE\nBEGIN\nEND\n", stmt.String())
-}
-
 func TestParseDataDriver(t *testing.T) {
 	datadriven.Walk(t, datapathutils.TestDataPath(t), func(t *testing.T, path string) {
 		datadriven.RunTest(t, path, func(t *testing.T, d *datadriven.TestData) string {

pkg/sql/plpgsql/parser/plpgsql.y

@@ -61,14 +61,6 @@ func (u *plpgsqlSymUnion) plpgsqlStmtBlock() *plpgsqltree.PLpgSQLStmtBlock {
     return u.val.(*plpgsqltree.PLpgSQLStmtBlock)
 }
 
-func (u *plpgsqlSymUnion) plpgsqlExpr() plpgsqltree.PLpgSQLExpr {
-    return u.val.(plpgsqltree.PLpgSQLExpr)
-}
-
-func (u *plpgsqlSymUnion) plpgsqlExprs() []plpgsqltree.PLpgSQLExpr {
-    return u.val.([]plpgsqltree.PLpgSQLExpr)
-}
-
 func (u *plpgsqlSymUnion) plpgsqlStmtCaseWhenArm() *plpgsqltree.PLpgSQLStmtCaseWhenArm {
     return u.val.(*plpgsqltree.PLpgSQLStmtCaseWhenArm)
 }
@@ -89,19 +81,10 @@ func (u *plpgsqlSymUnion) plpgsqlStatements() []plpgsqltree.PLpgSQLStatement {
     return u.val.([]plpgsqltree.PLpgSQLStatement)
 }
 
-func (u *plpgsqlSymUnion) plWDatum() *plWDatum {
-    return u.val.(*plWDatum)
-}
-
 func (u *plpgsqlSymUnion) int32() int32 {
     return u.val.(int32)
 }
 
-func (u *plpgsqlSymUnion) uint32() uint32 {
-    return u.val.(uint32)
-}
-
-
 func (u *plpgsqlSymUnion) bool() bool {
     return u.val.(bool)
 }

pkg/sql/plpgsql/parser/plpgsqllexbase/.gitignore

@@ -0,0 +1,4 @@
+# generated by ../parser/Makefile
+keywords.go
+reserved_keywords.go
+tokens.go

pkg/sql/plpgsql/parser/plpgsqllexbase/allkeywords/BUILD.bazel

@@ -0,0 +1,17 @@
+load("//build/bazelutil/unused_checker:unused.bzl", "get_x_data")
+load("@io_bazel_rules_go//go:def.bzl", "go_binary", "go_library")
+
+go_library(
+    name = "allkeywords_lib",
+    srcs = ["main.go"],
+    importpath = "github.com/cockroachdb/cockroach/pkg/sql/plpgsql/parser/plpgsqllexbase/allkeywords",
+    visibility = ["//visibility:private"],
+)
+
+go_binary(
+    name = "allkeywords",
+    embed = [":allkeywords_lib"],
+    visibility = ["//visibility:public"],
+)
+
+get_x_data(name = "get_x_data")