123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211 |
- // Copyright (c) 2017 Ernest Micklei
- //
- // MIT License
- //
- // Permission is hereby granted, free of charge, to any person obtaining
- // a copy of this software and associated documentation files (the
- // "Software"), to deal in the Software without restriction, including
- // without limitation the rights to use, copy, modify, merge, publish,
- // distribute, sublicense, and/or sell copies of the Software, and to
- // permit persons to whom the Software is furnished to do so, subject to
- // the following conditions:
- //
- // The above copyright notice and this permission notice shall be
- // included in all copies or substantial portions of the Software.
- //
- // THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- // EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- // MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
- // NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
- // LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- // OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- // WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- package proto
- import (
- "bytes"
- "errors"
- "fmt"
- "io"
- "runtime"
- "strconv"
- "strings"
- "text/scanner"
- )
- // Parser represents a parser.
- type Parser struct {
- debug bool
- scanner *scanner.Scanner
- buf *nextValues
- scannerErrors []error
- }
- // nextValues is to capture the result of next()
- type nextValues struct {
- pos scanner.Position
- tok token
- lit string
- }
- // NewParser returns a new instance of Parser.
- func NewParser(r io.Reader) *Parser {
- s := new(scanner.Scanner)
- s.Init(r)
- s.Mode = scanner.ScanIdents | scanner.ScanFloats | scanner.ScanStrings | scanner.ScanRawStrings | scanner.ScanComments
- p := &Parser{scanner: s}
- s.Error = p.handleScanError
- return p
- }
- // handleScanError is called from the underlying Scanner
- func (p *Parser) handleScanError(s *scanner.Scanner, msg string) {
- p.scannerErrors = append(p.scannerErrors,
- 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)
- if p.scanner.Filename != "" {
- proto.Filename = p.scanner.Filename
- }
- pro, parseError := proto.parse(p)
- // see if it was a scanner error
- if len(p.scannerErrors) > 0 {
- buf := new(bytes.Buffer)
- for _, each := range p.scannerErrors {
- fmt.Fprintln(buf, each)
- }
- return proto, errors.New(buf.String())
- }
- return pro, parseError
- }
- // Filename is for reporting. Optional.
- func (p *Parser) Filename(f string) {
- p.scanner.Filename = f
- }
- // next returns the next token using the scanner or drain the buffer.
- func (p *Parser) next() (pos scanner.Position, tok token, lit string) {
- if p.buf != nil {
- // consume buf
- vals := *p.buf
- p.buf = nil
- return vals.pos, vals.tok, vals.lit
- }
- ch := p.scanner.Scan()
- if ch == scanner.EOF {
- return p.scanner.Position, tEOF, ""
- }
- lit = p.scanner.TokenText()
- return p.scanner.Position, asToken(lit), lit
- }
- // nextPut sets the buffer
- func (p *Parser) nextPut(pos scanner.Position, tok token, lit string) {
- p.buf = &nextValues{pos, tok, lit}
- }
- func (p *Parser) unexpected(found, expected string, obj interface{}) error {
- debug := ""
- if p.debug {
- _, file, line, _ := runtime.Caller(1)
- debug = fmt.Sprintf(" at %s:%d (with %#v)", file, line, obj)
- }
- return fmt.Errorf("%v: found %q but expected [%s]%s", p.scanner.Position, found, expected, debug)
- }
- func (p *Parser) nextInteger() (i int, err error) {
- _, tok, lit := p.next()
- if "-" == lit {
- i, err = p.nextInteger()
- return i * -1, err
- }
- if tok != tIDENT {
- return 0, errors.New("non integer")
- }
- if strings.HasPrefix(lit, "0x") {
- // hex decode
- var i64 int64
- i64, err = strconv.ParseInt(lit, 0, 64)
- return int(i64), err
- }
- i, err = strconv.Atoi(lit)
- return
- }
- // nextIdentifier consumes tokens which may have one or more dot separators (namespaced idents).
- func (p *Parser) nextIdentifier() (pos scanner.Position, tok token, lit string) {
- return p.nextIdent(false)
- }
- // nextTypeName implements the Packages and Name Resolution for finding the name of the type.
- func (p *Parser) nextTypeName() (pos scanner.Position, tok token, lit string) {
- pos, tok, lit = p.nextIdent(false)
- if tDOT == tok {
- // leading dot allowed
- pos, tok, lit = p.nextIdent(false)
- lit = "." + lit
- }
- return
- }
- func (p *Parser) nextIdent(keywordStartAllowed bool) (pos scanner.Position, tok token, lit string) {
- pos, tok, lit = p.next()
- if tIDENT != tok {
- // can be keyword
- if !(isKeyword(tok) && keywordStartAllowed) {
- return
- }
- // proceed with keyword as first literal
- }
- startPos := pos
- fullLit := lit
- // see if identifier is namespaced
- for {
- r := p.scanner.Peek()
- if '.' != r {
- break
- }
- p.next() // consume dot
- pos, tok, lit := p.next()
- if tIDENT != tok && !isKeyword(tok) {
- p.nextPut(pos, tok, lit)
- break
- }
- fullLit = fmt.Sprintf("%s.%s", fullLit, lit)
- }
- return startPos, tIDENT, fullLit
- }
- func (p *Parser) peekNonWhitespace() rune {
- r := p.scanner.Peek()
- if r == scanner.EOF {
- return r
- }
- if isWhitespace(r) {
- // consume it
- p.scanner.Next()
- return p.peekNonWhitespace()
- }
- return r
- }
|