...
Run Format

Source file src/fmt/scan.go

Documentation: fmt

  // Copyright 2010 The Go Authors. All rights reserved.
  // Use of this source code is governed by a BSD-style
  // license that can be found in the LICENSE file.
  
  package fmt
  
  import (
  	"errors"
  	"io"
  	"math"
  	"os"
  	"reflect"
  	"strconv"
  	"sync"
  	"unicode/utf8"
  )
  
  // ScanState represents the scanner state passed to custom scanners.
  // Scanners may do rune-at-a-time scanning or ask the ScanState
  // to discover the next space-delimited token.
  type ScanState interface {
  	// ReadRune reads the next rune (Unicode code point) from the input.
  	// If invoked during Scanln, Fscanln, or Sscanln, ReadRune() will
  	// return EOF after returning the first '\n' or when reading beyond
  	// the specified width.
  	ReadRune() (r rune, size int, err error)
  	// UnreadRune causes the next call to ReadRune to return the same rune.
  	UnreadRune() error
  	// SkipSpace skips space in the input. Newlines are treated appropriately
  	// for the operation being performed; see the package documentation
  	// for more information.
  	SkipSpace()
  	// Token skips space in the input if skipSpace is true, then returns the
  	// run of Unicode code points c satisfying f(c).  If f is nil,
  	// !unicode.IsSpace(c) is used; that is, the token will hold non-space
  	// characters. Newlines are treated appropriately for the operation being
  	// performed; see the package documentation for more information.
  	// The returned slice points to shared data that may be overwritten
  	// by the next call to Token, a call to a Scan function using the ScanState
  	// as input, or when the calling Scan method returns.
  	Token(skipSpace bool, f func(rune) bool) (token []byte, err error)
  	// Width returns the value of the width option and whether it has been set.
  	// The unit is Unicode code points.
  	Width() (wid int, ok bool)
  	// Because ReadRune is implemented by the interface, Read should never be
  	// called by the scanning routines and a valid implementation of
  	// ScanState may choose always to return an error from Read.
  	Read(buf []byte) (n int, err error)
  }
  
  // Scanner is implemented by any value that has a Scan method, which scans
  // the input for the representation of a value and stores the result in the
  // receiver, which must be a pointer to be useful. The Scan method is called
  // for any argument to Scan, Scanf, or Scanln that implements it.
  type Scanner interface {
  	Scan(state ScanState, verb rune) error
  }
  
  // Scan scans text read from standard input, storing successive
  // space-separated values into successive arguments. Newlines count
  // as space. It returns the number of items successfully scanned.
  // If that is less than the number of arguments, err will report why.
  func Scan(a ...interface{}) (n int, err error) {
  	return Fscan(os.Stdin, a...)
  }
  
  // Scanln is similar to Scan, but stops scanning at a newline and
  // after the final item there must be a newline or EOF.
  func Scanln(a ...interface{}) (n int, err error) {
  	return Fscanln(os.Stdin, a...)
  }
  
  // Scanf scans text read from standard input, storing successive
  // space-separated values into successive arguments as determined by
  // the format. It returns the number of items successfully scanned.
  // If that is less than the number of arguments, err will report why.
  // Newlines in the input must match newlines in the format.
  // The one exception: the verb %c always scans the next rune in the
  // input, even if it is a space (or tab etc.) or newline.
  func Scanf(format string, a ...interface{}) (n int, err error) {
  	return Fscanf(os.Stdin, format, a...)
  }
  
  type stringReader string
  
  func (r *stringReader) Read(b []byte) (n int, err error) {
  	n = copy(b, *r)
  	*r = (*r)[n:]
  	if n == 0 {
  		err = io.EOF
  	}
  	return
  }
  
  // Sscan scans the argument string, storing successive space-separated
  // values into successive arguments. Newlines count as space. It
  // returns the number of items successfully scanned. If that is less
  // than the number of arguments, err will report why.
  func Sscan(str string, a ...interface{}) (n int, err error) {
  	return Fscan((*stringReader)(&str), a...)
  }
  
  // Sscanln is similar to Sscan, but stops scanning at a newline and
  // after the final item there must be a newline or EOF.
  func Sscanln(str string, a ...interface{}) (n int, err error) {
  	return Fscanln((*stringReader)(&str), a...)
  }
  
  // Sscanf scans the argument string, storing successive space-separated
  // values into successive arguments as determined by the format. It
  // returns the number of items successfully parsed.
  // Newlines in the input must match newlines in the format.
  func Sscanf(str string, format string, a ...interface{}) (n int, err error) {
  	return Fscanf((*stringReader)(&str), format, a...)
  }
  
  // Fscan scans text read from r, storing successive space-separated
  // values into successive arguments. Newlines count as space. It
  // returns the number of items successfully scanned. If that is less
  // than the number of arguments, err will report why.
  func Fscan(r io.Reader, a ...interface{}) (n int, err error) {
  	s, old := newScanState(r, true, false)
  	n, err = s.doScan(a)
  	s.free(old)
  	return
  }
  
  // Fscanln is similar to Fscan, but stops scanning at a newline and
  // after the final item there must be a newline or EOF.
  func Fscanln(r io.Reader, a ...interface{}) (n int, err error) {
  	s, old := newScanState(r, false, true)
  	n, err = s.doScan(a)
  	s.free(old)
  	return
  }
  
  // Fscanf scans text read from r, storing successive space-separated
  // values into successive arguments as determined by the format. It
  // returns the number of items successfully parsed.
  // Newlines in the input must match newlines in the format.
  func Fscanf(r io.Reader, format string, a ...interface{}) (n int, err error) {
  	s, old := newScanState(r, false, false)
  	n, err = s.doScanf(format, a)
  	s.free(old)
  	return
  }
  
  // scanError represents an error generated by the scanning software.
  // It's used as a unique signature to identify such errors when recovering.
  type scanError struct {
  	err error
  }
  
  const eof = -1
  
  // ss is the internal implementation of ScanState.
  type ss struct {
  	rs    io.RuneScanner // where to read input
  	buf   buffer         // token accumulator
  	count int            // runes consumed so far.
  	atEOF bool           // already read EOF
  	ssave
  }
  
  // ssave holds the parts of ss that need to be
  // saved and restored on recursive scans.
  type ssave struct {
  	validSave bool // is or was a part of an actual ss.
  	nlIsEnd   bool // whether newline terminates scan
  	nlIsSpace bool // whether newline counts as white space
  	argLimit  int  // max value of ss.count for this arg; argLimit <= limit
  	limit     int  // max value of ss.count.
  	maxWid    int  // width of this arg.
  }
  
  // The Read method is only in ScanState so that ScanState
  // satisfies io.Reader. It will never be called when used as
  // intended, so there is no need to make it actually work.
  func (s *ss) Read(buf []byte) (n int, err error) {
  	return 0, errors.New("ScanState's Read should not be called. Use ReadRune")
  }
  
  func (s *ss) ReadRune() (r rune, size int, err error) {
  	if s.atEOF || s.count >= s.argLimit {
  		err = io.EOF
  		return
  	}
  
  	r, size, err = s.rs.ReadRune()
  	if err == nil {
  		s.count++
  		if s.nlIsEnd && r == '\n' {
  			s.atEOF = true
  		}
  	} else if err == io.EOF {
  		s.atEOF = true
  	}
  	return
  }
  
  func (s *ss) Width() (wid int, ok bool) {
  	if s.maxWid == hugeWid {
  		return 0, false
  	}
  	return s.maxWid, true
  }
  
  // The public method returns an error; this private one panics.
  // If getRune reaches EOF, the return value is EOF (-1).
  func (s *ss) getRune() (r rune) {
  	r, _, err := s.ReadRune()
  	if err != nil {
  		if err == io.EOF {
  			return eof
  		}
  		s.error(err)
  	}
  	return
  }
  
  // mustReadRune turns io.EOF into a panic(io.ErrUnexpectedEOF).
  // It is called in cases such as string scanning where an EOF is a
  // syntax error.
  func (s *ss) mustReadRune() (r rune) {
  	r = s.getRune()
  	if r == eof {
  		s.error(io.ErrUnexpectedEOF)
  	}
  	return
  }
  
  func (s *ss) UnreadRune() error {
  	s.rs.UnreadRune()
  	s.atEOF = false
  	s.count--
  	return nil
  }
  
  func (s *ss) error(err error) {
  	panic(scanError{err})
  }
  
  func (s *ss) errorString(err string) {
  	panic(scanError{errors.New(err)})
  }
  
  func (s *ss) Token(skipSpace bool, f func(rune) bool) (tok []byte, err error) {
  	defer func() {
  		if e := recover(); e != nil {
  			if se, ok := e.(scanError); ok {
  				err = se.err
  			} else {
  				panic(e)
  			}
  		}
  	}()
  	if f == nil {
  		f = notSpace
  	}
  	s.buf = s.buf[:0]
  	tok = s.token(skipSpace, f)
  	return
  }
  
  // space is a copy of the unicode.White_Space ranges,
  // to avoid depending on package unicode.
  var space = [][2]uint16{
  	{0x0009, 0x000d},
  	{0x0020, 0x0020},
  	{0x0085, 0x0085},
  	{0x00a0, 0x00a0},
  	{0x1680, 0x1680},
  	{0x2000, 0x200a},
  	{0x2028, 0x2029},
  	{0x202f, 0x202f},
  	{0x205f, 0x205f},
  	{0x3000, 0x3000},
  }
  
  func isSpace(r rune) bool {
  	if r >= 1<<16 {
  		return false
  	}
  	rx := uint16(r)
  	for _, rng := range space {
  		if rx < rng[0] {
  			return false
  		}
  		if rx <= rng[1] {
  			return true
  		}
  	}
  	return false
  }
  
  // notSpace is the default scanning function used in Token.
  func notSpace(r rune) bool {
  	return !isSpace(r)
  }
  
  // SkipSpace provides Scan methods the ability to skip space and newline
  // characters in keeping with the current scanning mode set by format strings
  // and Scan/Scanln.
  func (s *ss) SkipSpace() {
  	s.skipSpace(false)
  }
  
  // readRune is a structure to enable reading UTF-8 encoded code points
  // from an io.Reader. It is used if the Reader given to the scanner does
  // not already implement io.RuneScanner.
  type readRune struct {
  	reader   io.Reader
  	buf      [utf8.UTFMax]byte // used only inside ReadRune
  	pending  int               // number of bytes in pendBuf; only >0 for bad UTF-8
  	pendBuf  [utf8.UTFMax]byte // bytes left over
  	peekRune rune              // if >=0 next rune; when <0 is ^(previous Rune)
  }
  
  // readByte returns the next byte from the input, which may be
  // left over from a previous read if the UTF-8 was ill-formed.
  func (r *readRune) readByte() (b byte, err error) {
  	if r.pending > 0 {
  		b = r.pendBuf[0]
  		copy(r.pendBuf[0:], r.pendBuf[1:])
  		r.pending--
  		return
  	}
  	n, err := io.ReadFull(r.reader, r.pendBuf[:1])
  	if n != 1 {
  		return 0, err
  	}
  	return r.pendBuf[0], err
  }
  
  // ReadRune returns the next UTF-8 encoded code point from the
  // io.Reader inside r.
  func (r *readRune) ReadRune() (rr rune, size int, err error) {
  	if r.peekRune >= 0 {
  		rr = r.peekRune
  		r.peekRune = ^r.peekRune
  		size = utf8.RuneLen(rr)
  		return
  	}
  	r.buf[0], err = r.readByte()
  	if err != nil {
  		return
  	}
  	if r.buf[0] < utf8.RuneSelf { // fast check for common ASCII case
  		rr = rune(r.buf[0])
  		size = 1 // Known to be 1.
  		// Flip the bits of the rune so it's available to UnreadRune.
  		r.peekRune = ^rr
  		return
  	}
  	var n int
  	for n = 1; !utf8.FullRune(r.buf[:n]); n++ {
  		r.buf[n], err = r.readByte()
  		if err != nil {
  			if err == io.EOF {
  				err = nil
  				break
  			}
  			return
  		}
  	}
  	rr, size = utf8.DecodeRune(r.buf[:n])
  	if size < n { // an error, save the bytes for the next read
  		copy(r.pendBuf[r.pending:], r.buf[size:n])
  		r.pending += n - size
  	}
  	// Flip the bits of the rune so it's available to UnreadRune.
  	r.peekRune = ^rr
  	return
  }
  
  func (r *readRune) UnreadRune() error {
  	if r.peekRune >= 0 {
  		return errors.New("fmt: scanning called UnreadRune with no rune available")
  	}
  	// Reverse bit flip of previously read rune to obtain valid >=0 state.
  	r.peekRune = ^r.peekRune
  	return nil
  }
  
  var ssFree = sync.Pool{
  	New: func() interface{} { return new(ss) },
  }
  
  // newScanState allocates a new ss struct or grab a cached one.
  func newScanState(r io.Reader, nlIsSpace, nlIsEnd bool) (s *ss, old ssave) {
  	s = ssFree.Get().(*ss)
  	if rs, ok := r.(io.RuneScanner); ok {
  		s.rs = rs
  	} else {
  		s.rs = &readRune{reader: r, peekRune: -1}
  	}
  	s.nlIsSpace = nlIsSpace
  	s.nlIsEnd = nlIsEnd
  	s.atEOF = false
  	s.limit = hugeWid
  	s.argLimit = hugeWid
  	s.maxWid = hugeWid
  	s.validSave = true
  	s.count = 0
  	return
  }
  
  // free saves used ss structs in ssFree; avoid an allocation per invocation.
  func (s *ss) free(old ssave) {
  	// If it was used recursively, just restore the old state.
  	if old.validSave {
  		s.ssave = old
  		return
  	}
  	// Don't hold on to ss structs with large buffers.
  	if cap(s.buf) > 1024 {
  		return
  	}
  	s.buf = s.buf[:0]
  	s.rs = nil
  	ssFree.Put(s)
  }
  
  // skipSpace skips spaces and maybe newlines.
  func (s *ss) skipSpace(stopAtNewline bool) {
  	for {
  		r := s.getRune()
  		if r == eof {
  			return
  		}
  		if r == '\r' && s.peek("\n") {
  			continue
  		}
  		if r == '\n' {
  			if stopAtNewline {
  				break
  			}
  			if s.nlIsSpace {
  				continue
  			}
  			s.errorString("unexpected newline")
  			return
  		}
  		if !isSpace(r) {
  			s.UnreadRune()
  			break
  		}
  	}
  }
  
  // token returns the next space-delimited string from the input. It
  // skips white space. For Scanln, it stops at newlines. For Scan,
  // newlines are treated as spaces.
  func (s *ss) token(skipSpace bool, f func(rune) bool) []byte {
  	if skipSpace {
  		s.skipSpace(false)
  	}
  	// read until white space or newline
  	for {
  		r := s.getRune()
  		if r == eof {
  			break
  		}
  		if !f(r) {
  			s.UnreadRune()
  			break
  		}
  		s.buf.WriteRune(r)
  	}
  	return s.buf
  }
  
  var complexError = errors.New("syntax error scanning complex number")
  var boolError = errors.New("syntax error scanning boolean")
  
  func indexRune(s string, r rune) int {
  	for i, c := range s {
  		if c == r {
  			return i
  		}
  	}
  	return -1
  }
  
  // consume reads the next rune in the input and reports whether it is in the ok string.
  // If accept is true, it puts the character into the input token.
  func (s *ss) consume(ok string, accept bool) bool {
  	r := s.getRune()
  	if r == eof {
  		return false
  	}
  	if indexRune(ok, r) >= 0 {
  		if accept {
  			s.buf.WriteRune(r)
  		}
  		return true
  	}
  	if r != eof && accept {
  		s.UnreadRune()
  	}
  	return false
  }
  
  // peek reports whether the next character is in the ok string, without consuming it.
  func (s *ss) peek(ok string) bool {
  	r := s.getRune()
  	if r != eof {
  		s.UnreadRune()
  	}
  	return indexRune(ok, r) >= 0
  }
  
  func (s *ss) notEOF() {
  	// Guarantee there is data to be read.
  	if r := s.getRune(); r == eof {
  		panic(io.EOF)
  	}
  	s.UnreadRune()
  }
  
  // accept checks the next rune in the input. If it's a byte (sic) in the string, it puts it in the
  // buffer and returns true. Otherwise it return false.
  func (s *ss) accept(ok string) bool {
  	return s.consume(ok, true)
  }
  
  // okVerb verifies that the verb is present in the list, setting s.err appropriately if not.
  func (s *ss) okVerb(verb rune, okVerbs, typ string) bool {
  	for _, v := range okVerbs {
  		if v == verb {
  			return true
  		}
  	}
  	s.errorString("bad verb '%" + string(verb) + "' for " + typ)
  	return false
  }
  
  // scanBool returns the value of the boolean represented by the next token.
  func (s *ss) scanBool(verb rune) bool {
  	s.skipSpace(false)
  	s.notEOF()
  	if !s.okVerb(verb, "tv", "boolean") {
  		return false
  	}
  	// Syntax-checking a boolean is annoying. We're not fastidious about case.
  	switch s.getRune() {
  	case '0':
  		return false
  	case '1':
  		return true
  	case 't', 'T':
  		if s.accept("rR") && (!s.accept("uU") || !s.accept("eE")) {
  			s.error(boolError)
  		}
  		return true
  	case 'f', 'F':
  		if s.accept("aA") && (!s.accept("lL") || !s.accept("sS") || !s.accept("eE")) {
  			s.error(boolError)
  		}
  		return false
  	}
  	return false
  }
  
  // Numerical elements
  const (
  	binaryDigits      = "01"
  	octalDigits       = "01234567"
  	decimalDigits     = "0123456789"
  	hexadecimalDigits = "0123456789aAbBcCdDeEfF"
  	sign              = "+-"
  	period            = "."
  	exponent          = "eEp"
  )
  
  // getBase returns the numeric base represented by the verb and its digit string.
  func (s *ss) getBase(verb rune) (base int, digits string) {
  	s.okVerb(verb, "bdoUxXv", "integer") // sets s.err
  	base = 10
  	digits = decimalDigits
  	switch verb {
  	case 'b':
  		base = 2
  		digits = binaryDigits
  	case 'o':
  		base = 8
  		digits = octalDigits
  	case 'x', 'X', 'U':
  		base = 16
  		digits = hexadecimalDigits
  	}
  	return
  }
  
  // scanNumber returns the numerical string with specified digits starting here.
  func (s *ss) scanNumber(digits string, haveDigits bool) string {
  	if !haveDigits {
  		s.notEOF()
  		if !s.accept(digits) {
  			s.errorString("expected integer")
  		}
  	}
  	for s.accept(digits) {
  	}
  	return string(s.buf)
  }
  
  // scanRune returns the next rune value in the input.
  func (s *ss) scanRune(bitSize int) int64 {
  	s.notEOF()
  	r := int64(s.getRune())
  	n := uint(bitSize)
  	x := (r << (64 - n)) >> (64 - n)
  	if x != r {
  		s.errorString("overflow on character value " + string(r))
  	}
  	return r
  }
  
  // scanBasePrefix reports whether the integer begins with a 0 or 0x,
  // and returns the base, digit string, and whether a zero was found.
  // It is called only if the verb is %v.
  func (s *ss) scanBasePrefix() (base int, digits string, found bool) {
  	if !s.peek("0") {
  		return 10, decimalDigits, false
  	}
  	s.accept("0")
  	found = true // We've put a digit into the token buffer.
  	// Special cases for '0' && '0x'
  	base, digits = 8, octalDigits
  	if s.peek("xX") {
  		s.consume("xX", false)
  		base, digits = 16, hexadecimalDigits
  	}
  	return
  }
  
  // scanInt returns the value of the integer represented by the next
  // token, checking for overflow. Any error is stored in s.err.
  func (s *ss) scanInt(verb rune, bitSize int) int64 {
  	if verb == 'c' {
  		return s.scanRune(bitSize)
  	}
  	s.skipSpace(false)
  	s.notEOF()
  	base, digits := s.getBase(verb)
  	haveDigits := false
  	if verb == 'U' {
  		if !s.consume("U", false) || !s.consume("+", false) {
  			s.errorString("bad unicode format ")
  		}
  	} else {
  		s.accept(sign) // If there's a sign, it will be left in the token buffer.
  		if verb == 'v' {
  			base, digits, haveDigits = s.scanBasePrefix()
  		}
  	}
  	tok := s.scanNumber(digits, haveDigits)
  	i, err := strconv.ParseInt(tok, base, 64)
  	if err != nil {
  		s.error(err)
  	}
  	n := uint(bitSize)
  	x := (i << (64 - n)) >> (64 - n)
  	if x != i {
  		s.errorString("integer overflow on token " + tok)
  	}
  	return i
  }
  
  // scanUint returns the value of the unsigned integer represented
  // by the next token, checking for overflow. Any error is stored in s.err.
  func (s *ss) scanUint(verb rune, bitSize int) uint64 {
  	if verb == 'c' {
  		return uint64(s.scanRune(bitSize))
  	}
  	s.skipSpace(false)
  	s.notEOF()
  	base, digits := s.getBase(verb)
  	haveDigits := false
  	if verb == 'U' {
  		if !s.consume("U", false) || !s.consume("+", false) {
  			s.errorString("bad unicode format ")
  		}
  	} else if verb == 'v' {
  		base, digits, haveDigits = s.scanBasePrefix()
  	}
  	tok := s.scanNumber(digits, haveDigits)
  	i, err := strconv.ParseUint(tok, base, 64)
  	if err != nil {
  		s.error(err)
  	}
  	n := uint(bitSize)
  	x := (i << (64 - n)) >> (64 - n)
  	if x != i {
  		s.errorString("unsigned integer overflow on token " + tok)
  	}
  	return i
  }
  
  // floatToken returns the floating-point number starting here, no longer than swid
  // if the width is specified. It's not rigorous about syntax because it doesn't check that
  // we have at least some digits, but Atof will do that.
  func (s *ss) floatToken() string {
  	s.buf = s.buf[:0]
  	// NaN?
  	if s.accept("nN") && s.accept("aA") && s.accept("nN") {
  		return string(s.buf)
  	}
  	// leading sign?
  	s.accept(sign)
  	// Inf?
  	if s.accept("iI") && s.accept("nN") && s.accept("fF") {
  		return string(s.buf)
  	}
  	// digits?
  	for s.accept(decimalDigits) {
  	}
  	// decimal point?
  	if s.accept(period) {
  		// fraction?
  		for s.accept(decimalDigits) {
  		}
  	}
  	// exponent?
  	if s.accept(exponent) {
  		// leading sign?
  		s.accept(sign)
  		// digits?
  		for s.accept(decimalDigits) {
  		}
  	}
  	return string(s.buf)
  }
  
  // complexTokens returns the real and imaginary parts of the complex number starting here.
  // The number might be parenthesized and has the format (N+Ni) where N is a floating-point
  // number and there are no spaces within.
  func (s *ss) complexTokens() (real, imag string) {
  	// TODO: accept N and Ni independently?
  	parens := s.accept("(")
  	real = s.floatToken()
  	s.buf = s.buf[:0]
  	// Must now have a sign.
  	if !s.accept("+-") {
  		s.error(complexError)
  	}
  	// Sign is now in buffer
  	imagSign := string(s.buf)
  	imag = s.floatToken()
  	if !s.accept("i") {
  		s.error(complexError)
  	}
  	if parens && !s.accept(")") {
  		s.error(complexError)
  	}
  	return real, imagSign + imag
  }
  
  // convertFloat converts the string to a float64value.
  func (s *ss) convertFloat(str string, n int) float64 {
  	if p := indexRune(str, 'p'); p >= 0 {
  		// Atof doesn't handle power-of-2 exponents,
  		// but they're easy to evaluate.
  		f, err := strconv.ParseFloat(str[:p], n)
  		if err != nil {
  			// Put full string into error.
  			if e, ok := err.(*strconv.NumError); ok {
  				e.Num = str
  			}
  			s.error(err)
  		}
  		m, err := strconv.Atoi(str[p+1:])
  		if err != nil {
  			// Put full string into error.
  			if e, ok := err.(*strconv.NumError); ok {
  				e.Num = str
  			}
  			s.error(err)
  		}
  		return math.Ldexp(f, m)
  	}
  	f, err := strconv.ParseFloat(str, n)
  	if err != nil {
  		s.error(err)
  	}
  	return f
  }
  
  // convertComplex converts the next token to a complex128 value.
  // The atof argument is a type-specific reader for the underlying type.
  // If we're reading complex64, atof will parse float32s and convert them
  // to float64's to avoid reproducing this code for each complex type.
  func (s *ss) scanComplex(verb rune, n int) complex128 {
  	if !s.okVerb(verb, floatVerbs, "complex") {
  		return 0
  	}
  	s.skipSpace(false)
  	s.notEOF()
  	sreal, simag := s.complexTokens()
  	real := s.convertFloat(sreal, n/2)
  	imag := s.convertFloat(simag, n/2)
  	return complex(real, imag)
  }
  
  // convertString returns the string represented by the next input characters.
  // The format of the input is determined by the verb.
  func (s *ss) convertString(verb rune) (str string) {
  	if !s.okVerb(verb, "svqxX", "string") {
  		return ""
  	}
  	s.skipSpace(false)
  	s.notEOF()
  	switch verb {
  	case 'q':
  		str = s.quotedString()
  	case 'x', 'X':
  		str = s.hexString()
  	default:
  		str = string(s.token(true, notSpace)) // %s and %v just return the next word
  	}
  	return
  }
  
  // quotedString returns the double- or back-quoted string represented by the next input characters.
  func (s *ss) quotedString() string {
  	s.notEOF()
  	quote := s.getRune()
  	switch quote {
  	case '`':
  		// Back-quoted: Anything goes until EOF or back quote.
  		for {
  			r := s.mustReadRune()
  			if r == quote {
  				break
  			}
  			s.buf.WriteRune(r)
  		}
  		return string(s.buf)
  	case '"':
  		// Double-quoted: Include the quotes and let strconv.Unquote do the backslash escapes.
  		s.buf.WriteByte('"')
  		for {
  			r := s.mustReadRune()
  			s.buf.WriteRune(r)
  			if r == '\\' {
  				// In a legal backslash escape, no matter how long, only the character
  				// immediately after the escape can itself be a backslash or quote.
  				// Thus we only need to protect the first character after the backslash.
  				s.buf.WriteRune(s.mustReadRune())
  			} else if r == '"' {
  				break
  			}
  		}
  		result, err := strconv.Unquote(string(s.buf))
  		if err != nil {
  			s.error(err)
  		}
  		return result
  	default:
  		s.errorString("expected quoted string")
  	}
  	return ""
  }
  
  // hexDigit returns the value of the hexadecimal digit.
  func hexDigit(d rune) (int, bool) {
  	digit := int(d)
  	switch digit {
  	case '0', '1', '2', '3', '4', '5', '6', '7', '8', '9':
  		return digit - '0', true
  	case 'a', 'b', 'c', 'd', 'e', 'f':
  		return 10 + digit - 'a', true
  	case 'A', 'B', 'C', 'D', 'E', 'F':
  		return 10 + digit - 'A', true
  	}
  	return -1, false
  }
  
  // hexByte returns the next hex-encoded (two-character) byte from the input.
  // It returns ok==false if the next bytes in the input do not encode a hex byte.
  // If the first byte is hex and the second is not, processing stops.
  func (s *ss) hexByte() (b byte, ok bool) {
  	rune1 := s.getRune()
  	if rune1 == eof {
  		return
  	}
  	value1, ok := hexDigit(rune1)
  	if !ok {
  		s.UnreadRune()
  		return
  	}
  	value2, ok := hexDigit(s.mustReadRune())
  	if !ok {
  		s.errorString("illegal hex digit")
  		return
  	}
  	return byte(value1<<4 | value2), true
  }
  
  // hexString returns the space-delimited hexpair-encoded string.
  func (s *ss) hexString() string {
  	s.notEOF()
  	for {
  		b, ok := s.hexByte()
  		if !ok {
  			break
  		}
  		s.buf.WriteByte(b)
  	}
  	if len(s.buf) == 0 {
  		s.errorString("no hex data for %x string")
  		return ""
  	}
  	return string(s.buf)
  }
  
  const (
  	floatVerbs = "beEfFgGv"
  
  	hugeWid = 1 << 30
  
  	intBits     = 32 << (^uint(0) >> 63)
  	uintptrBits = 32 << (^uintptr(0) >> 63)
  )
  
  // scanOne scans a single value, deriving the scanner from the type of the argument.
  func (s *ss) scanOne(verb rune, arg interface{}) {
  	s.buf = s.buf[:0]
  	var err error
  	// If the parameter has its own Scan method, use that.
  	if v, ok := arg.(Scanner); ok {
  		err = v.Scan(s, verb)
  		if err != nil {
  			if err == io.EOF {
  				err = io.ErrUnexpectedEOF
  			}
  			s.error(err)
  		}
  		return
  	}
  
  	switch v := arg.(type) {
  	case *bool:
  		*v = s.scanBool(verb)
  	case *complex64:
  		*v = complex64(s.scanComplex(verb, 64))
  	case *complex128:
  		*v = s.scanComplex(verb, 128)
  	case *int:
  		*v = int(s.scanInt(verb, intBits))
  	case *int8:
  		*v = int8(s.scanInt(verb, 8))
  	case *int16:
  		*v = int16(s.scanInt(verb, 16))
  	case *int32:
  		*v = int32(s.scanInt(verb, 32))
  	case *int64:
  		*v = s.scanInt(verb, 64)
  	case *uint:
  		*v = uint(s.scanUint(verb, intBits))
  	case *uint8:
  		*v = uint8(s.scanUint(verb, 8))
  	case *uint16:
  		*v = uint16(s.scanUint(verb, 16))
  	case *uint32:
  		*v = uint32(s.scanUint(verb, 32))
  	case *uint64:
  		*v = s.scanUint(verb, 64)
  	case *uintptr:
  		*v = uintptr(s.scanUint(verb, uintptrBits))
  	// Floats are tricky because you want to scan in the precision of the result, not
  	// scan in high precision and convert, in order to preserve the correct error condition.
  	case *float32:
  		if s.okVerb(verb, floatVerbs, "float32") {
  			s.skipSpace(false)
  			s.notEOF()
  			*v = float32(s.convertFloat(s.floatToken(), 32))
  		}
  	case *float64:
  		if s.okVerb(verb, floatVerbs, "float64") {
  			s.skipSpace(false)
  			s.notEOF()
  			*v = s.convertFloat(s.floatToken(), 64)
  		}
  	case *string:
  		*v = s.convertString(verb)
  	case *[]byte:
  		// We scan to string and convert so we get a copy of the data.
  		// If we scanned to bytes, the slice would point at the buffer.
  		*v = []byte(s.convertString(verb))
  	default:
  		val := reflect.ValueOf(v)
  		ptr := val
  		if ptr.Kind() != reflect.Ptr {
  			s.errorString("type not a pointer: " + val.Type().String())
  			return
  		}
  		switch v := ptr.Elem(); v.Kind() {
  		case reflect.Bool:
  			v.SetBool(s.scanBool(verb))
  		case reflect.Int, reflect.Int8, reflect.Int16, reflect.Int32, reflect.Int64:
  			v.SetInt(s.scanInt(verb, v.Type().Bits()))
  		case reflect.Uint, reflect.Uint8, reflect.Uint16, reflect.Uint32, reflect.Uint64, reflect.Uintptr:
  			v.SetUint(s.scanUint(verb, v.Type().Bits()))
  		case reflect.String:
  			v.SetString(s.convertString(verb))
  		case reflect.Slice:
  			// For now, can only handle (renamed) []byte.
  			typ := v.Type()
  			if typ.Elem().Kind() != reflect.Uint8 {
  				s.errorString("can't scan type: " + val.Type().String())
  			}
  			str := s.convertString(verb)
  			v.Set(reflect.MakeSlice(typ, len(str), len(str)))
  			for i := 0; i < len(str); i++ {
  				v.Index(i).SetUint(uint64(str[i]))
  			}
  		case reflect.Float32, reflect.Float64:
  			s.skipSpace(false)
  			s.notEOF()
  			v.SetFloat(s.convertFloat(s.floatToken(), v.Type().Bits()))
  		case reflect.Complex64, reflect.Complex128:
  			v.SetComplex(s.scanComplex(verb, v.Type().Bits()))
  		default:
  			s.errorString("can't scan type: " + val.Type().String())
  		}
  	}
  }
  
  // errorHandler turns local panics into error returns.
  func errorHandler(errp *error) {
  	if e := recover(); e != nil {
  		if se, ok := e.(scanError); ok { // catch local error
  			*errp = se.err
  		} else if eof, ok := e.(error); ok && eof == io.EOF { // out of input
  			*errp = eof
  		} else {
  			panic(e)
  		}
  	}
  }
  
  // doScan does the real work for scanning without a format string.
  func (s *ss) doScan(a []interface{}) (numProcessed int, err error) {
  	defer errorHandler(&err)
  	for _, arg := range a {
  		s.scanOne('v', arg)
  		numProcessed++
  	}
  	// Check for newline (or EOF) if required (Scanln etc.).
  	if s.nlIsEnd {
  		for {
  			r := s.getRune()
  			if r == '\n' || r == eof {
  				break
  			}
  			if !isSpace(r) {
  				s.errorString("expected newline")
  				break
  			}
  		}
  	}
  	return
  }
  
  // advance determines whether the next characters in the input match
  // those of the format. It returns the number of bytes (sic) consumed
  // in the format. All runs of space characters in either input or
  // format behave as a single space. Newlines are special, though:
  // newlines in the format must match those in the input and vice versa.
  // This routine also handles the %% case. If the return value is zero,
  // either format starts with a % (with no following %) or the input
  // is empty. If it is negative, the input did not match the string.
  func (s *ss) advance(format string) (i int) {
  	for i < len(format) {
  		fmtc, w := utf8.DecodeRuneInString(format[i:])
  
  		// Space processing.
  		// In the rest of this comment "space" means spaces other than newline.
  		// Newline in the format matches input of zero or more spaces and then newline or end-of-input.
  		// Spaces in the format before the newline are collapsed into the newline.
  		// Spaces in the format after the newline match zero or more spaces after the corresponding input newline.
  		// Other spaces in the format match input of one or more spaces or end-of-input.
  		if isSpace(fmtc) {
  			newlines := 0
  			trailingSpace := false
  			for isSpace(fmtc) && i < len(format) {
  				if fmtc == '\n' {
  					newlines++
  					trailingSpace = false
  				} else {
  					trailingSpace = true
  				}
  				i += w
  				fmtc, w = utf8.DecodeRuneInString(format[i:])
  			}
  			for j := 0; j < newlines; j++ {
  				inputc := s.getRune()
  				for isSpace(inputc) && inputc != '\n' {
  					inputc = s.getRune()
  				}
  				if inputc != '\n' && inputc != eof {
  					s.errorString("newline in format does not match input")
  				}
  			}
  			if trailingSpace {
  				inputc := s.getRune()
  				if newlines == 0 {
  					// If the trailing space stood alone (did not follow a newline),
  					// it must find at least one space to consume.
  					if !isSpace(inputc) && inputc != eof {
  						s.errorString("expected space in input to match format")
  					}
  					if inputc == '\n' {
  						s.errorString("newline in input does not match format")
  					}
  				}
  				for isSpace(inputc) && inputc != '\n' {
  					inputc = s.getRune()
  				}
  				if inputc != eof {
  					s.UnreadRune()
  				}
  			}
  			continue
  		}
  
  		// Verbs.
  		if fmtc == '%' {
  			// % at end of string is an error.
  			if i+w == len(format) {
  				s.errorString("missing verb: % at end of format string")
  			}
  			// %% acts like a real percent
  			nextc, _ := utf8.DecodeRuneInString(format[i+w:]) // will not match % if string is empty
  			if nextc != '%' {
  				return
  			}
  			i += w // skip the first %
  		}
  
  		// Literals.
  		inputc := s.mustReadRune()
  		if fmtc != inputc {
  			s.UnreadRune()
  			return -1
  		}
  		i += w
  	}
  	return
  }
  
  // doScanf does the real work when scanning with a format string.
  // At the moment, it handles only pointers to basic types.
  func (s *ss) doScanf(format string, a []interface{}) (numProcessed int, err error) {
  	defer errorHandler(&err)
  	end := len(format) - 1
  	// We process one item per non-trivial format
  	for i := 0; i <= end; {
  		w := s.advance(format[i:])
  		if w > 0 {
  			i += w
  			continue
  		}
  		// Either we failed to advance, we have a percent character, or we ran out of input.
  		if format[i] != '%' {
  			// Can't advance format. Why not?
  			if w < 0 {
  				s.errorString("input does not match format")
  			}
  			// Otherwise at EOF; "too many operands" error handled below
  			break
  		}
  		i++ // % is one byte
  
  		// do we have 20 (width)?
  		var widPresent bool
  		s.maxWid, widPresent, i = parsenum(format, i, end)
  		if !widPresent {
  			s.maxWid = hugeWid
  		}
  
  		c, w := utf8.DecodeRuneInString(format[i:])
  		i += w
  
  		if c != 'c' {
  			s.SkipSpace()
  		}
  		s.argLimit = s.limit
  		if f := s.count + s.maxWid; f < s.argLimit {
  			s.argLimit = f
  		}
  
  		if numProcessed >= len(a) { // out of operands
  			s.errorString("too few operands for format '%" + format[i-w:] + "'")
  			break
  		}
  		arg := a[numProcessed]
  
  		s.scanOne(c, arg)
  		numProcessed++
  		s.argLimit = s.limit
  	}
  	if numProcessed < len(a) {
  		s.errorString("too many operands")
  	}
  	return
  }
  

View as plain text