Sync parser up to 26.2.

go.mod

@@ -1,17 +1,18 @@
 module github.com/cockroachdb/cockroachdb-parser
 
-go 1.23.8
+go 1.24
 
 require (
 	github.com/bazelbuild/rules_go v0.46.0
 	github.com/biogo/store v0.0.0-20201120204734-aad293a2328f
 	github.com/blevesearch/snowballstem v0.9.0
 	github.com/cockroachdb/apd/v3 v3.1.0
+	github.com/cockroachdb/crlib v0.0.0-20251122031428-fe658a2dbda1
 	github.com/cockroachdb/errors v1.11.3
 	github.com/cockroachdb/gostdlib v1.19.0
 	github.com/cockroachdb/logtags v0.0.0-20241215232642-bb51bb14a506
 	github.com/cockroachdb/redact v1.1.5
-	github.com/cockroachdb/version v0.0.0-20250314144055-3860cd14adf2
+	github.com/cockroachdb/version v0.0.0-20250509181251-54dac3003410
 	github.com/dave/dst v0.27.2
 	github.com/dustin/go-humanize v1.0.0
 	github.com/gogo/protobuf v1.3.2

go.sum

@@ -60,6 +60,8 @@ github.com/client9/misspell v0.3.4/go.mod h1:qj6jICC3Q7zFZvVWo7KLAzC3yx5G7kyvSDk
 github.com/cncf/udpa/go v0.0.0-20191209042840-269d4d468f6f/go.mod h1:M8M6+tZqaGXZJjfX53e64911xZQV5JYwmTeXPW+k8Sc=
 github.com/cockroachdb/apd/v3 v3.1.0 h1:MK3Ow7LH0W8zkd5GMKA1PvS9qG3bWFI95WaVNfyZJ/w=
 github.com/cockroachdb/apd/v3 v3.1.0/go.mod h1:6qgPBMXjATAdD/VefbRP9NoSLKjbB4LCoA7gN4LpHs4=
+github.com/cockroachdb/crlib v0.0.0-20251122031428-fe658a2dbda1 h1:iX0YCYC5Jbt2/g7zNTP/QxhrV8Syp5kkzNiERKeN1uE=
+github.com/cockroachdb/crlib v0.0.0-20251122031428-fe658a2dbda1/go.mod h1:NjNuToN/FbhwH1cCyM9G4Rhtxx+ZaOgtoqFR+thng7w=
 github.com/cockroachdb/errors v1.11.3 h1:5bA+k2Y6r+oz/6Z/RFlNeVCesGARKuC6YymtcDrbC/I=
 github.com/cockroachdb/errors v1.11.3/go.mod h1:m4UIW4CDjx+R5cybPsNrRbreomiFqt8o1h1wUVazSd8=
 github.com/cockroachdb/gostdlib v1.19.0 h1:cSISxkVnTlWhTkyple/T6NXzOi5659FkhxvUgZv+Eb0=
@@ -68,8 +70,8 @@ github.com/cockroachdb/logtags v0.0.0-20241215232642-bb51bb14a506 h1:ASDL+UJcILM
 github.com/cockroachdb/logtags v0.0.0-20241215232642-bb51bb14a506/go.mod h1:Mw7HqKr2kdtu6aYGn3tPmAftiP3QPX63LdK/zcariIo=
 github.com/cockroachdb/redact v1.1.5 h1:u1PMllDkdFfPWaNGMyLD1+so+aq3uUItthCFqzwPJ30=
 github.com/cockroachdb/redact v1.1.5/go.mod h1:BVNblN9mBWFyMyqK1k3AAiSxhvhfK2oOZZ2lK+dpvRg=
-github.com/cockroachdb/version v0.0.0-20250314144055-3860cd14adf2 h1:8Vfw2iNEpYIV6aLtMwT5UOGuPmp9MKlEKWKFTuB+MPU=
-github.com/cockroachdb/version v0.0.0-20250314144055-3860cd14adf2/go.mod h1:P9WiZOdQ1R/ZZDL0WzF5wlyRvrjtfhNOwMZymFpBwjE=
+github.com/cockroachdb/version v0.0.0-20250509181251-54dac3003410 h1:GuIAxxl30gmd9m7ct6fgYb49GUv8GAxICU5VfU9GRvc=
+github.com/cockroachdb/version v0.0.0-20250509181251-54dac3003410/go.mod h1:P9WiZOdQ1R/ZZDL0WzF5wlyRvrjtfhNOwMZymFpBwjE=
 github.com/codahale/hdrhistogram v0.0.0-20161010025455-3a0bb77429bd/go.mod h1:sE/e/2PUdi/liOCUjSTXgM1o87ZssimdTWN964YiIeI=
 github.com/containerd/continuity v0.0.0-20190827140505-75bee3e2ccb6/go.mod h1:GL3xCUCBDV3CZiTSEKksMWbLE66hEyuu9qyDOOqM47Y=
 github.com/coreos/bbolt v1.3.2/go.mod h1:iRUV2dpdMOn7Bo10OQBFzIJO9kkE559Wcmn+qkEiiKk=

patches/0007-update-functions-parser.patch

@@ -5,7 +5,7 @@ index 4f7bc03..5b31fd0 100644
 @@ -51,6 +51,13 @@ const (
  	// as Oracle is case insensitive if object name is not quoted.
  	EncAlwaysQuoted
- 
+
 +	// EncSkipEscapeString indicates that the string should not be escaped,
 +	// and non-ASCII characters are allowed.
 +	// More specifically, it means that the tree.DString won't be wrapped
@@ -27,7 +27,7 @@ index 4f7bc03..5b31fd0 100644
 @@ -165,7 +173,13 @@ func EncodeSQLStringWithFlags(buf *bytes.Buffer, in string, flags EncodeFlags) {
  			}
  		}
- 
+
 -		if !escapedString {
 +		// If non-ASCII characters are allowed,
 +		// skip escaping and write the original UTF-8 character.
@@ -48,7 +48,7 @@ index 4f7bc03..5b31fd0 100644
 +		// quote char, so we will  write the quote char directly.
 +		stringencoding.EncodeEscapedChar(buf, in, r, ch, i, '\'', !skipEscape)
  	}
- 
+
 -	quote := !escapedString && !bareStrings
 +	quote := !escapedString && !bareStrings && !skipEscape
  	if quote {
@@ -69,7 +69,7 @@ index 05d64e1..dabff21 100644
 @@ -1364,6 +1364,19 @@ const (
  	OrderedSetAgg
  )
- 
+
 +// onlyNameFunc is the list of function who can be compiled with only the
 +// name. This is for PG compatibility, where examples such as `CURRENT_TIMESTAMP()`
 +// is not allowed, but `CURRENT_TIMESTAMP` is allowed.
@@ -89,7 +89,7 @@ index 05d64e1..dabff21 100644
 @@ -1385,41 +1398,43 @@ func (node *FuncExpr) Format(ctx *FmtCtx) {
  		ctx.FormatNode(&node.Func)
  	})
- 
+
 -	ctx.WriteByte('(')
 -	ctx.WriteString(typ)
 -	ctx.FormatNode(&node.Exprs)
@@ -168,22 +168,22 @@ index 6832c7e..df15c15 100644
 --- a/pkg/sql/sem/tree/format.go
 +++ b/pkg/sql/sem/tree/format.go
 @@ -195,6 +195,10 @@ const (
- 	// FmtSkipAsOfSystemTimeClauses prevents the formatter from printing AS OF
- 	// SYSTEM TIME clauses.
- 	FmtSkipAsOfSystemTimeClauses
+ 	// FmtPLpgSQLParen will wrap some expressions in parenthesis when in PLpgSQL
+ 	// context. This should only be used in tests.
+ 	FmtPLpgSQLParen
 +
 +	// FmtFuncOnlyName instructs the formating for a function only print
 +	// its name, without printing the body of the function nor with the brackets.
 +	FmtFuncOnlyName
  )
- 
+
  const genericArityIndicator = "__more__"
 diff --git a/pkg/util/stringencoding/string_encoding.go b/pkg/util/stringencoding/string_encoding.go
 index fcb7fac..afa0090 100644
 --- a/pkg/util/stringencoding/string_encoding.go
 +++ b/pkg/util/stringencoding/string_encoding.go
 @@ -73,6 +73,7 @@ func init() {
- 
+
  // EncodeEscapedChar is used internally to write out a character from a larger
  // string that needs to be escaped to a buffer.
 +// If slashedQuoteChar is true, it will write a backslash before the quoteChar.

pkg/build/bazel/util/tinystringer/main.go

@@ -332,7 +332,7 @@ func (s tinyStringer) computeConstantValues(
 			if genDecl.Tok != token.CONST {
 				continue
 			}
-			var inferAppropriateType, inIota bool
+			var inferAppropriateType, inIota, inShift bool
 			var iotaVal int
 			for _, spec := range genDecl.Specs {
 				valueSpec, ok := spec.(*ast.ValueSpec)
@@ -360,7 +360,11 @@ func (s tinyStringer) computeConstantValues(
 				if valueSpec.Values == nil {
 					if inIota {
 						nameToInt[constName] = iotaVal
-						iotaVal += 1
+						if inShift {
+							iotaVal <<= 1
+						} else {
+							iotaVal += 1
+						}
 					} else {
 						nameToInt[constName] = 0
 					}
@@ -401,37 +405,55 @@ func (s tinyStringer) computeConstantValues(
 						inIota = false
 					}
 				} else if binExpr, ok := valueSpec.Values[0].(*ast.BinaryExpr); ok {
-					// Handle iota + N or iota - N.
-					iotaIdent, ok := binExpr.X.(*ast.Ident)
+					inShift = false
+					xArg, yArg := binExpr.X, binExpr.Y
+					if binExpr.Op == token.SHL {
+						// Handle 1 << iota - the order of arguments is switched.
+						inShift = true
+						xArg, yArg = yArg, xArg
+					}
+					// Handle iota + N, iota - N, 1 << iota.
+					iotaIdent, ok := xArg.(*ast.Ident)
 					if !ok || iotaIdent.Name != "iota" {
-						err = fmt.Errorf("expected 'iota' in binary expression %+v; found %+v", binExpr, binExpr.X)
+						err = fmt.Errorf("expected 'iota' in binary expression %+v; found %+v", binExpr, xArg)
 						return
 					}
 					var otherNumParsed int64
-					if otherNum, ok := binExpr.Y.(*ast.BasicLit); ok && otherNum.Kind == token.INT {
+					if otherNum, ok := yArg.(*ast.BasicLit); ok && otherNum.Kind == token.INT {
 						otherNumParsed, err = strconv.ParseInt(otherNum.Value, 0, 0)
 						if err != nil {
 							return
 						}
-					} else if otherRef, ok := binExpr.Y.(*ast.Ident); ok {
+					} else if otherRef, ok := yArg.(*ast.Ident); ok {
 						otherNum, ok := nameToInt[otherRef.Name]
 						if !ok {
 							err = fmt.Errorf("could not find value of %s", otherRef.Name)
 							return
 						}
 						otherNumParsed = int64(otherNum)
 					} else {
-						err = fmt.Errorf("couldn't parse second argument of binary expression %+v; found %+v", binExpr, binExpr.Y)
+						err = fmt.Errorf("couldn't parse second argument of binary expression %+v; found %+v", binExpr, yArg)
 						return
 					}
 					if binExpr.Op == token.ADD {
 						iotaVal = iotaVal + int(otherNumParsed)
 					} else if binExpr.Op == token.SUB {
 						iotaVal = iotaVal - int(otherNumParsed)
+					} else if binExpr.Op == token.SHL {
+						// We currently assume 1 << iota and not N << iota.
+						if otherNumParsed != 1 {
+							err = fmt.Errorf("only 1 << iota is supported; found %d << iota", otherNumParsed)
+							return
+						}
+						iotaVal = int(otherNumParsed)
 					}
 					inIota = true
 					nameToInt[constName] = iotaVal
-					iotaVal += 1
+					if inShift {
+						iotaVal <<= 1
+					} else {
+						iotaVal += 1
+					}
 				} else {
 					err = fmt.Errorf("don't know how to process %+v", valueSpec.Values[0])
 					return

pkg/build/info.go

@@ -7,6 +7,7 @@ package build
 
 import (
 	"bytes"
+	"crypto/fips140"
 	_ "embed"
 	"fmt"
 	"runtime"
@@ -35,8 +36,9 @@ var (
 	channel          string
 )
 
-// Distribution is changed by the CCL init-time hook in non-APL builds.
-var Distribution = "OSS"
+// Distribution is a historical artifact from when we had OSS and non-OSS (CCL)
+// builds.
+const Distribution = "CCL"
 
 var (
 	cgoCompiler       = cgoVersion()
@@ -107,16 +109,16 @@ func BinaryVersion() string {
 // It returns "vX.Y" for all release versions, and all prerelease versions >= "alpha.1".
 // X and Y are the major and minor, respectively, of the version specified in version.txt.
 // For all other prerelease versions, it returns "dev".
-// N.B. new public-facing doc URLs are expected to be up beginning with the "alpha.1" prerelease. Otherwise, "dev" will
+// N.B. new public-facing doc URLs are expected to be up once the "alpha.1" prerelease is shipped. Otherwise, "dev" will
 // cause the url mapper to redirect to the latest stable release.
 func VersionForURLs() string {
 	if parsedVersionTxt.IsPrerelease() {
 		phaseAndOrdinal := parsedVersionTxt.Format("%P.%o")
-		// builds prior to "alpha.1" use 'dev' in their URLs
-		if phaseAndOrdinal < "alpha.1" {
+		// builds use 'dev' in their URLs until "alpha.1" is shipped
+		if phaseAndOrdinal <= "alpha.1" {
 			return "dev"
 		}
-	} else if parsedVersionTxt.IsCustomOrNightlyBuild() {
+	} else if parsedVersionTxt.IsCustomOrAdhocBuild() {
 		return "dev"
 	}
 	return parsedVersionTxt.Major().String()
@@ -168,6 +170,10 @@ func (b Info) Long() string {
 	fmt.Fprintf(tw, "C Compiler:       %s\n", b.CgoCompiler)
 	fmt.Fprintf(tw, "Build Commit ID:  %s\n", b.Revision)
 	fmt.Fprintf(tw, "Build Type:       %s\n", b.Type)
+	if fips140.Enabled() {
+		fmt.Fprintf(tw, "FIPS enabled:     true\n")
+	}
+
 	fmt.Fprintf(tw, "Enabled Assertions: %t", b.EnabledAssertions) // No final newline: cobra prints one for us.
 	_ = tw.Flush()
 	return buf.String()

pkg/col/coldata/json.go

@@ -15,17 +15,21 @@ import (
 
 // JSONs is a representation of columnar JSON data. It's simply a wrapper around
 // the flat Bytes structure. To pull a JSON out of the structure, we construct
-// a new "encodedJSON" object from scratch on demand.
+// a new JSONEncoded object from scratch on demand.
 type JSONs struct {
 	Bytes
+	// jsonScratch batches allocations of JSONEncoded objects (under the
+	// assumption that the vector doesn't have NULLs).
+	jsonScratch []json.JSONEncoded
 	// scratch is a scratch space for encoding a JSON object on demand.
 	scratch []byte
 }
 
 // NewJSONs returns a new JSONs presized to n elements.
 func NewJSONs(n int) *JSONs {
 	return &JSONs{
-		Bytes: *NewBytes(n),
+		Bytes:       *NewBytes(n),
+		jsonScratch: make([]json.JSONEncoded, n),
 	}
 }
 
@@ -38,11 +42,12 @@ func (js *JSONs) Get(i int) json.JSON {
 	if len(bytes) == 0 {
 		return json.NullJSONValue
 	}
-	ret, err := json.FromEncoding(bytes)
+	j := &js.jsonScratch[i]
+	err := json.FromEncodingInto(bytes, j)
 	if err != nil {
 		colexecerror.ExpectedError(err)
 	}
-	return ret
+	return j
 }
 
 // Set sets the ith JSON in JSONs.
@@ -62,6 +67,13 @@ func (js *JSONs) Set(i int, j json.JSON) {
 func (js *JSONs) Window(start, end int) *JSONs {
 	return &JSONs{
 		Bytes: *js.Bytes.Window(start, end),
+		// TODO(yuzefovich): in the general case, if we simply reused
+		// js.jsonScratch here, it could lead to problems down the line (because
+		// Get calls on js and on the window into js would be reusing the same
+		// JSONEncoded object). However, in practice the windowed batch should
+		// be consumed soon after it's created, without the overlap with Gets on
+		// the original batch. Think through this.
+		jsonScratch: make([]json.JSONEncoded, end-start),
 	}
 }
 
@@ -81,12 +93,14 @@ func (js *JSONs) CopySlice(src *JSONs, destIdx, srcStartIdx, srcEndIdx int) {
 // values from src into the receiver starting at destIdx.
 func (js *JSONs) AppendSlice(src *JSONs, destIdx, srcStartIdx, srcEndIdx int) {
 	js.Bytes.AppendSlice(&src.Bytes, destIdx, srcStartIdx, srcEndIdx)
+	js.ensureJSONScratch()
 }
 
 // appendSliceWithSel appends all values specified in sel from the source into
 // the receiver starting at position destIdx.
 func (js *JSONs) appendSliceWithSel(src *JSONs, destIdx int, sel []int) {
 	js.Bytes.appendSliceWithSel(&src.Bytes, destIdx, sel)
+	js.ensureJSONScratch()
 }
 
 // String is used for debugging purposes.
@@ -99,3 +113,18 @@ func (js *JSONs) String() string {
 	}
 	return builder.String()
 }
+
+// Deserialize updates b according to the "arrow-like" format that was produced
+// by Serialize.
+func (js *JSONs) Deserialize(data []byte, offsets []int32) {
+	js.Bytes.Deserialize(data, offsets)
+	js.ensureJSONScratch()
+}
+
+func (js *JSONs) ensureJSONScratch() {
+	if cap(js.jsonScratch) < js.Bytes.Len() {
+		js.jsonScratch = make([]json.JSONEncoded, js.Bytes.Len())
+	} else {
+		js.jsonScratch = js.jsonScratch[:js.Bytes.Len()]
+	}
+}

pkg/col/coldata/vec.go

@@ -11,6 +11,7 @@ import (
 
 	"github.com/cockroachdb/cockroachdb-parser/pkg/col/typeconv"
 	"github.com/cockroachdb/cockroachdb-parser/pkg/sql/types"
+	"github.com/cockroachdb/cockroachdb-parser/pkg/util/json"
 )
 
 // Column is an interface that represents a raw array of a Go native type.
@@ -160,11 +161,14 @@ func (cf *defaultColumnFactory) MakeColumns(columns []Column, t *types.T, length
 		}
 	case types.JsonFamily:
 		alloc := make([]element, allocLength)
+		jsonScratchAlloc := make([]json.JSONEncoded, allocLength)
 		wrapperAlloc := make([]JSONs, len(columns))
 		for i := range columns {
 			wrapperAlloc[i].elements = alloc[:length:length]
+			wrapperAlloc[i].jsonScratch = jsonScratchAlloc[:length:length]
 			columns[i] = &wrapperAlloc[i]
 			alloc = alloc[length:]
+			jsonScratchAlloc = jsonScratchAlloc[length:]
 		}
 	default:
 		panic(fmt.Sprintf("StandardColumnFactory doesn't support %s", t))

pkg/geo/geomfn/distance.go

@@ -220,7 +220,7 @@ func distanceInternal(
 	aIt := geo.NewGeomTIterator(aGeomT, emptyBehavior)
 	aGeom, aNext, aErr := aIt.Next()
 	if aErr != nil {
-		return 0, err
+		return 0, aErr
 	}
 	for aNext {
 		aGeodist, err := geomToGeodist(aGeom)
@@ -231,7 +231,7 @@ func distanceInternal(
 		bIt := geo.NewGeomTIterator(bGeomT, emptyBehavior)
 		bGeom, bNext, bErr := bIt.Next()
 		if bErr != nil {
-			return 0, err
+			return 0, bErr
 		}
 		for bNext {
 			bGeodist, err := geomToGeodist(bGeom)
@@ -248,13 +248,13 @@ func distanceInternal(
 
 			bGeom, bNext, bErr = bIt.Next()
 			if bErr != nil {
-				return 0, err
+				return 0, bErr
 			}
 		}
 
 		aGeom, aNext, aErr = aIt.Next()
 		if aErr != nil {
-			return 0, err
+			return 0, aErr
 		}
 	}
 	return c.DistanceUpdater().Distance(), nil