path: root/vendor/github.com/hashicorp/hcl/json/scanner/scanner.go

package scanner

import (
	"bytes"
	"fmt"
	"os"
	"unicode"
	"unicode/utf8"

	"github.com/hashicorp/hcl/json/token"
)

// eof represents a marker rune for the end of the reader.
const eof = rune(0)

// Scanner defines a lexical scanner
type Scanner struct {
	buf *bytes.Buffer // Source buffer for advancing and scanning
	src []byte        // Source buffer for immutable access

	// Source Position
	srcPos  token.Pos // current position
	prevPos token.Pos // previous position, used for peek() method

	lastCharLen int // length of last character in bytes
	lastLineLen int // length of last line in characters (for correct column reporting)

	tokStart int // token text start position
	tokEnd   int // token text end  position

	// Error is called for each error encountered. If no Error
	// function is set, the error is reported to os.Stderr.
	Error func(pos token.Pos, msg string)

	// ErrorCount is incremented by one for each error encountered.
	ErrorCount int

	// tokPos is the start position of most recently scanned token; set by
	// Scan. The Filename field is always left untouched by the Scanner.  If
	// an error is reported (via Error) and Position is invalid, the scanner is
	// not inside a token.
	tokPos token.Pos
}

// New creates and initializes a new instance of Scanner using src as
// its source content.
func New(src []byte) *Scanner {
	// even though we accept a src, we read from a io.Reader compatible type
	// (*bytes.Buffer). So in the future we might easily change it to streaming
	// read.
	b := bytes.NewBuffer(src)
	s := &Scanner{
		buf: b,
		src: src,
	}

	// srcPosition always starts with 1
	s.srcPos.Line = 1
	return s
}

// next reads the next rune from the bufferred reader. Returns the rune(0) if
// an error occurs (or io.EOF is returned).
func (s *Scanner) next() rune {
	ch, size, err := s.buf.ReadRune()
	if err != nil {
		// advance for error reporting
		s.srcPos.Column++
		s.srcPos.Offset += size
		s.lastCharLen = size
		return eof
	}

	if ch == utf8.RuneError && size == 1 {
		s.srcPos.Column++
		s.srcPos.Offset += size
		s.lastCharLen = size
		s.err("illegal UTF-8 encoding")
		return ch
	}

	// remember last position
	s.prevPos = s.srcPos

	s.srcPos.Column++
	s.lastCharLen = size
	s.srcPos.Offset += size

	if ch == '\n' {
		s.srcPos.Line++
		s.lastLineLen = s.srcPos.Column
		s.srcPos.Column = 0
	}

	// debug
	// fmt.Printf("ch: %q, offset:column: %d:%d\n", ch, s.srcPos.Offset, s.srcPos.Column)
	return ch
}

// unread unreads the previous read Rune and updates the source position
func (s *Scanner) unread() {
	if err := s.buf.UnreadRune(); err != nil {
		panic(err) // this is user fault, we should catch it
	}
	s.srcPos = s.prevPos // put back last position
}

// peek returns the next rune without advancing the reader.
func (s *Scanner) peek() rune {
	peek, _, err := s.buf.ReadRune()
	if err != nil {
		return eof
	}

	s.buf.UnreadRune()
	return peek
}

// Scan scans the next token and returns the token.
func (s *Scanner) Scan() token.Token {
	ch := s.next()

	// skip white space
	for isWhitespace(ch) {
		ch = s.next()
	}

	var tok token.Type

	// token text markings
	s.tokStart = s.srcPos.Offset - s.lastCharLen

	// token position, initial next() is moving the offset by one(size of rune
	// actually), though we are interested with the starting point
	s.tokPos.Offset = s.srcPos.Offset - s.lastCharLen
	if s.srcPos.Column > 0 {
		// common case: last character was not a '\n'
		s.tokPos.Line = s.srcPos.Line
		s.tokPos.Column = s.srcPos.Column
	} else {
		// last character was a '\n'
		// (we cannot be at the beginning of the source
		// since we have called next() at least once)
		s.tokPos.Line = s.srcPos.Line - 1
		s.tokPos.Column = s.lastLineLen
	}

	switch {
	case isLetter(ch):
		lit := s.scanIdentifier()
		if lit == "true" || lit == "false" {
			tok = token.BOOL
		} else if lit == "null" {
			tok = token.NULL
		} else {
			s.err("illegal char")
		}
	case isDecimal(ch):
		tok = s.scanNumber(ch)
	default:
		switch ch {
		case eof:
			tok = token.EOF
		case '"':
			tok = token.STRING
			s.scanString()
		case '.':
			tok = token.PERIOD
			ch = s.peek()
			if isDecimal(ch) {
				tok = token.FLOAT
				ch = s.scanMantissa(ch)
				ch = s.scanExponent(ch)
			}
		case '[':
			tok = token.LBRACK
		case ']':
			tok = token.RBRACK
		case '{':
			tok = token.LBRACE
		case '}':
			tok = token.RBRACE
		case ',':
			tok = token.COMMA
		case ':':
			tok = token.COLON
		case '-':
			if isDecimal(s.peek()) {
				ch := s.next()
				tok = s.scanNumber(ch)
			} else {
				s.err("illegal char")
			}
		default:
			s.err("illegal char: " + string(ch))
		}
	}

	// finish token ending
	s.tokEnd = s.srcPos.Offset

	// create token literal
	var tokenText string
	if s.tokStart >= 0 {
		tokenText = string(s.src[s.tokStart:s.tokEnd])
	}
	s.tokStart = s.tokEnd // ensure idempotency of tokenText() call

	return token.Token{
		Type: tok,
		Pos:  s.tokPos,
		Text: tokenText,
	}
}

// scanNumber scans a HCL number definition starting with the given rune
func (s *Scanner) scanNumber(ch rune) token.Type {
	zero := ch == '0'
	pos := s.srcPos

	s.scanMantissa(ch)
	ch = s.next() // seek forward
	if ch == 'e' || ch == 'E' {
		ch = s.scanExponent(ch)
		return token.FLOAT
	}

	if ch == '.' {
		<