fix for Issue emicklei#73 (emicklei#74)

* ignore illegal escape chars while parsing option constant

README.md

@@ -46,8 +46,5 @@ Package in Go for parsing Google Protocol Buffers [.proto files version 2 + 3] (
 
 See (https://github.com/emicklei/proto-contrib) for other contributions on top of this package such as protofmt, proto2xsd and proto2gql.
 
-#### known issues
-
-- the proto2 test file in (https://github.com/emicklei/proto-contrib/cmd/protofmt) folder contains character escape sequences that are currently not accepted by the scanner. See line 537 and 573.
 
 © 2017, [ernestmicklei.com](http://ernestmicklei.com).  MIT License. Contributions welcome.

option.go

@@ -82,18 +82,25 @@ func (o *Option) parse(p *Parser) error {
 		return p.unexpected(lit, "option value assignment =", o)
 	}
 	r := p.peekNonWhitespace()
-	if '{' == r {
-		p.next() // consume {
-		return o.parseAggregate(p)
-	}
-	// non aggregate
-	l := new(Literal)
-	l.Position = pos
-	if err := l.parse(p); err != nil {
-		return err
-	}
-	o.Constant = *l
-	return nil
+	var err error
+	// values of an option can have illegal escape sequences
+	// for the standard Go scanner used by this package.
+	p.ignoreIllegalEscapesWhile(func() {
+		if '{' == r {
+			// aggregate
+			p.next() // consume {
+			err = o.parseAggregate(p)
+		} else {
+			// non aggregate
+			l := new(Literal)
+			l.Position = pos
+			if e := l.parse(p); e != nil {
+				err = e
+			}
+			o.Constant = *l
+		}
+	})
+	return err
 }
 
 // inlineComment is part of commentInliner.

option_test.go

@@ -249,6 +249,41 @@ func TestFieldCustomOptions(t *testing.T) {
 	}
 }
 
+func TestIgnoreIllegalEscapeCharsInAggregatedConstants(t *testing.T) {
+	src := `syntax = "proto3";
+	message Person {
+	  string name  = 3 [(validate.rules).string = {
+						  pattern:   "^[^\d\s]+( [^\d\s]+)*$",
+						  max_bytes: 256,
+					   }];
+	}`
+	p := newParserOn(src)
+	d, err := p.Parse()
+	if err != nil {
+		t.Fatal(err)
+	}
+	f := d.Elements[1].(*Message).Elements[0].(*NormalField)
+	if got, want := f.Options[0].AggregatedConstants[0].Source, "^[^\\d\\s]+( [^\\d\\s]+)*$"; got != want {
+		t.Errorf("got [%v] want [%v]", got, want)
+	}
+}
+
+func TestIgnoreIllegalEscapeCharsInConstant(t *testing.T) {
+	src := `syntax = "proto2";
+	message Person {
+		optional string cpp_trigraph = 20 [default = "? \? ?? \?? \??? ??/ ?\?-"];
+	}`
+	p := newParserOn(src)
+	d, err := p.Parse()
+	if err != nil {
+		t.Fatal(err)
+	}
+	f := d.Elements[1].(*Message).Elements[0].(*NormalField)
+	if got, want := f.Options[0].Constant.Source, "? \\? ?? \\?? \\??? ??/ ?\\?-"; got != want {
+		t.Errorf("got [%v] want [%v]", got, want)
+	}
+}
+
 func TestFieldCustomOptionExtendedIdent(t *testing.T) {
 	proto := `Type field = 1 [(validate.rules).enum.defined_only = true];`
 	p := newParserOn(proto)

parser.go

@@ -30,6 +30,7 @@ import (
 	"io"
 	"runtime"
 	"strconv"
+	"strings"
 	"text/scanner"
 )
 
@@ -64,6 +65,22 @@ func (p *Parser) handleScanError(s *scanner.Scanner, msg string) {
 		fmt.Errorf("go scanner error at %v = %v", s.Position, msg))
 }
 
+// ignoreIllegalEscapesWhile is called for scanning constants of an option.
+// Such content can have a syntax that is not acceptable by the Go scanner.
+// This temporary installs a handler that ignores only one type of error: illegal char escape
+func (p *Parser) ignoreIllegalEscapesWhile(block func()) {
+	// during block call change error handler
+	p.scanner.Error = func(s *scanner.Scanner, msg string) {
+		if strings.Contains(msg, "illegal char escape") { // too bad there is no constant for this in scanner pkg
+			return
+		}
+		p.handleScanError(s, msg)
+	}
+	block()
+	// restore
+	p.scanner.Error = p.handleScanError
+}
+
 // Parse parses a proto definition. May return a parse or scanner error.
 func (p *Parser) Parse() (*Proto, error) {
 	proto := new(Proto)