Merge cockroachdb#59690

59690: geo/wkt: implement parser for points with Z and M dimensions r=otan a=andyyang890

This patch adds a parser that is capable of parsing WKT
representations of points with Z and M dimensions.

Refs: cockroachdb#53091

Release note: None

Co-authored-by: Andy Yang <[email protected]>

Makefile

@@ -816,6 +816,9 @@ SQLPARSER_TARGETS = \
 	pkg/sql/lex/keywords.go \
 	pkg/sql/lexbase/reserved_keywords.go
 
+WKTPARSER_TARGETS = \
+	pkg/geo/wkt/wkt.go
+
 PROTOBUF_TARGETS := bin/.go_protobuf_sources bin/.gw_protobuf_sources
 
 DOCGEN_TARGETS := \
@@ -1135,7 +1138,7 @@ dupl: bin/.bootstrap
 
 .PHONY: generate
 generate: ## Regenerate generated code.
-generate: protobuf $(DOCGEN_TARGETS) $(OPTGEN_TARGETS) $(LOG_TARGETS) $(SQLPARSER_TARGETS) $(SETTINGS_DOC_PAGE) bin/langgen bin/terraformgen
+generate: protobuf $(DOCGEN_TARGETS) $(OPTGEN_TARGETS) $(LOG_TARGETS) $(SQLPARSER_TARGETS) $(WKTPARSER_TARGETS) $(SETTINGS_DOC_PAGE) bin/langgen bin/terraformgen
 	$(GO) generate $(GOFLAGS) $(GOMODVENDORFLAGS) -tags '$(TAGS)' -ldflags '$(LINKFLAGS)' $(PKG)
 	$(MAKE) execgen
 

build/variables.mk

@@ -154,6 +154,7 @@ define VALID_VARS
   WEBPACK
   WEBPACK_DASHBOARD
   WEBPACK_DEV_SERVER
+  WKTPARSER_TARGETS
   XCC
   XCMAKE_SYSTEM_NAME
   XCXX

pkg/geo/wkt/.gitignore

@@ -0,0 +1 @@
+y.output

pkg/geo/wkt/BUILD.bazel

@@ -0,0 +1,23 @@
+load("@io_bazel_rules_go//go:def.bzl", "go_library", "go_test")
+
+go_library(
+    name = "wkt",
+    srcs = [
+        "lex.go",
+        "wkt.go",
+        "wkt_generated.go",
+    ],
+    importpath = "github.com/cockroachdb/cockroach/pkg/geo/wkt",
+    visibility = ["//visibility:public"],
+    deps = ["@com_github_twpayne_go_geom//:go-geom"],
+)
+
+go_test(
+    name = "wkt_test",
+    srcs = ["wkt_test.go"],
+    embed = [":wkt"],
+    deps = [
+        "@com_github_stretchr_testify//require",
+        "@com_github_twpayne_go_geom//:go-geom",
+    ],
+)

pkg/geo/wkt/lex.go

@@ -0,0 +1,195 @@
+// Copyright 2021 The Cockroach Authors.
+//
+// Use of this software is governed by the Business Source License
+// included in the file licenses/BSL.txt.
+//
+// As of the Change Date specified in that file, in accordance with
+// the Business Source License, use of this software will be governed
+// by the Apache License, Version 2.0, included in the file
+// licenses/APL.txt.
+
+package wkt
+
+import (
+	"fmt"
+	"strconv"
+	"strings"
+	"unicode"
+
+	"github.com/twpayne/go-geom"
+)
+
+// LexError is an error that occurs during lexing.
+type LexError struct {
+	problem string
+	pos     int
+}
+
+func (e *LexError) Error() string {
+	return fmt.Sprintf("lex error: %s at pos %d", e.problem, e.pos)
+}
+
+// ParseError is an error that occurs during parsing, which happens after lexing.
+type ParseError struct {
+	line string
+}
+
+func (e *ParseError) Error() string {
+	return fmt.Sprintf("parse error: could not parse %q", e.line)
+}
+
+// Constant expected by parser when lexer reaches EOF.
+const eof = 0
+
+type wktLex struct {
+	line    string
+	pos     int
+	ret     geom.T
+	lastErr error
+}
+
+// Lex lexes a token from the input.
+func (l *wktLex) Lex(yylval *wktSymType) int {
+	// Skip leading spaces.
+	l.trimLeft()
+
+	// Lex a token.
+	switch c := l.peek(); c {
+	case eof:
+		return eof
+	case '(', ')', ',':
+		return int(l.next())
+	default:
+		if unicode.IsLetter(c) {
+			return l.keyword()
+		} else if isNumRune(c) {
+			return l.num(yylval)
+		} else {
+			l.lastErr = &LexError{
+				problem: "unrecognized character",
+				pos:     l.pos,
+			}
+			return eof
+		}
+	}
+}
+
+func getKeywordToken(tokStr string) int {
+	switch tokStr {
+	case "EMPTY":
+		return EMPTY
+	case "POINT":
+		return POINT
+	case "POINTZ":
+		return POINTZ
+	case "POINTM":
+		return POINTM
+	case "POINTZM":
+		return POINTZM
+	default:
+		return eof
+	}
+}
+
+// keyword lexes a string keyword.
+func (l *wktLex) keyword() int {
+	startPos := l.pos
+	var b strings.Builder
+
+	for {
+		c := l.peek()
+		if !unicode.IsLetter(c) {
+			break
+		}
+		// Add the uppercase letter to the string builder.
+		b.WriteRune(unicode.ToUpper(l.next()))
+	}
+
+	// Check for extra dimensions for geometry types.
+	if b.String() != "EMPTY" {
+		l.trimLeft()
+		if unicode.ToUpper(l.peek()) == 'Z' {
+			l.next()
+			b.WriteRune('Z')
+		}
+		if unicode.ToUpper(l.peek()) == 'M' {
+			l.next()
+			b.WriteRune('M')
+		}
+	}
+
+	ret := getKeywordToken(b.String())
+	if ret == eof {
+		l.lastErr = &LexError{
+			problem: "invalid keyword",
+			pos:     startPos,
+		}
+	}
+
+	return ret
+}
+
+func isNumRune(r rune) bool {
+	switch r {
+	case '-', '.':
+		return true
+	default:
+		return unicode.IsDigit(r)
+	}
+}
+
+// num lexes a number.
+func (l *wktLex) num(yylval *wktSymType) int {
+	startPos := l.pos
+	var b strings.Builder
+
+	for {
+		c := l.peek()
+		if !isNumRune(c) {
+			break
+		}
+		b.WriteRune(l.next())
+	}
+
+	fl, err := strconv.ParseFloat(b.String(), 64)
+	if err != nil {
+		l.lastErr = &LexError{
+			problem: "invalid number",
+			pos:     startPos,
+		}
+		return eof
+	}
+	yylval.coord = fl
+	return NUM
+}
+
+func (l *wktLex) peek() rune {
+	if l.pos == len(l.line) {
+		return eof
+	}
+	return rune(l.line[l.pos])
+}
+
+func (l *wktLex) next() rune {
+	c := l.peek()
+	if c != eof {
+		l.pos++
+	}
+	return c
+}
+
+func (l *wktLex) trimLeft() {
+	for {
+		c := l.peek()
+		if c == eof || !unicode.IsSpace(c) {
+			break
+		}
+		l.next()
+	}
+}
+
+func (l *wktLex) Error(s string) {
+	// Lex errors are set in the Lex function.
+	// todo (ayang) improve parse error messages
+	/* EMPTY */
+}

pkg/geo/wkt/wkt.go

@@ -0,0 +1,28 @@
+// Copyright 2021 The Cockroach Authors.
+//
+// Use of this software is governed by the Business Source License
+// included in the file licenses/BSL.txt.
+//
+// As of the Change Date specified in that file, in accordance with
+// the Business Source License, use of this software will be governed
+// by the Apache License, Version 2.0, included in the file
+// licenses/APL.txt.
+
+//go:generate goyacc -o wkt_generated.go -p "wkt" wkt.y
+
+package wkt
+
+import "github.com/twpayne/go-geom"
+
+// Unmarshal accepts a string and parses it to a geom.T.
+func Unmarshal(wkt string) (geom.T, error) {
+	wktlex := &wktLex{line: wkt}
+	ret := wktParse(wktlex)
+	if wktlex.lastErr != nil {
+		return nil, wktlex.lastErr
+	}
+	if ret != 0 {
+		return nil, &ParseError{line: wkt}
+	}
+	return wktlex.ret, nil
+}

pkg/geo/wkt/wkt.y

@@ -0,0 +1,104 @@
+// Copyright 2021 The Cockroach Authors.
+//
+// Use of this software is governed by the Business Source License
+// included in the file licenses/BSL.txt.
+//
+// As of the Change Date specified in that file, in accordance with
+// the Business Source License, use of this software will be governed
+// by the Apache License, Version 2.0, included in the file
+// licenses/APL.txt.
+
+%{
+
+package wkt
+
+import "github.com/twpayne/go-geom"
+
+%}
+
+%union {
+	str       string
+	geom      geom.T
+	coord     float64
+	coordList []float64
+}
+
+%token <str> POINT POINTZ POINTM POINTZM
+%token <str> EMPTY
+//%token <str> LINESTRING POLYGON MULTIPOINT MULTILINESTRING MULTIPOLYGON GEOMETRYCOLLECTION
+%token <coord> NUM
+
+%type <geom> geometry
+%type <geom> point
+%type <coordList> two_coords three_coords four_coords
+
+%%
+
+start:
+	geometry
+	{
+		wktlex.(*wktLex).ret = $1
+	}
+
+geometry:
+	point
+
+point:
+	POINT two_coords
+	{
+		$$ = geom.NewPointFlat(geom.XY, $2)
+	}
+| POINT three_coords
+	{
+		$$ = geom.NewPointFlat(geom.XYZ, $2)
+	}
+| POINT four_coords
+	{
+		$$ = geom.NewPointFlat(geom.XYZM, $2)
+	}
+| POINTZ three_coords
+	{
+		$$ = geom.NewPointFlat(geom.XYZ, $2)
+	}
+| POINTM three_coords
+	{
+		$$ = geom.NewPointFlat(geom.XYM, $2)
+	}
+| POINTZM four_coords
+	{
+		$$ = geom.NewPointFlat(geom.XYZM, $2)
+	}
+| POINT EMPTY
+	{
+		$$ = geom.NewPointEmpty(geom.XY)
+	}
+| POINTZ EMPTY
+	{
+		$$ = geom.NewPointEmpty(geom.XYZ)
+	}
+| POINTM EMPTY
+	{
+		$$ = geom.NewPointEmpty(geom.XYM)
+	}
+| POINTZM EMPTY
+	{
+		$$ = geom.NewPointEmpty(geom.XYZM)
+	}
+
+two_coords:
+	'(' NUM NUM ')'
+	{
+		$$ = []float64{$2, $3}
+	}
+
+three_coords:
+	'(' NUM NUM NUM ')'
+	{
+		$$ = []float64{$2, $3, $4}
+	}
+
+four_coords:
+	'(' NUM NUM NUM NUM ')'
+	{
+		$$ = []float64{$2, $3, $4, $5}
+	}