URL https://opencores.org/ocsvn/openrisc/openrisc/trunk

Subversion Repositories openrisc

[/] [openrisc/] [trunk/] [gnu-dev/] [or1k-gcc/] [libgo/] [go/] [text/] [scanner/] [scanner.go] - Blame information for rev 858

Go to most recent revision | Details | Compare with Previous | View Log


// Copyright 2009 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 scanner provides a scanner and tokenizer for UTF-8-encoded text.
// It takes an io.Reader providing the source, which then can be tokenized
// through repeated calls to the Scan function.  For compatibility with
// existing tools, the NUL character is not allowed (implementation
// restriction).
//
// By default, a Scanner skips white space and Go comments and recognizes all
// literals as defined by the Go language specification.  It may be
// customized to recognize only a subset of those literals and to recognize
// different white space characters.
//
// Basic usage pattern:
//
//      var s scanner.Scanner
//      s.Init(src)
//      tok := s.Scan()
//      for tok != scanner.EOF {
//              // do something with tok
//              tok = s.Scan()
//      }
//
package scanner
 
import (
        "bytes"
        "fmt"
        "io"
        "os"
        "unicode"
        "unicode/utf8"
)
 
// TODO(gri): Consider changing this to use the new (token) Position package.
 
// A source position is represented by a Position value.
// A position is valid if Line > 0.
type Position struct {
        Filename string // filename, if any
        Offset   int    // byte offset, starting at 0
        Line     int    // line number, starting at 1
        Column   int    // column number, starting at 1 (character count per line)
}
 
// IsValid returns true if the position is valid.
func (pos *Position) IsValid() bool { return pos.Line > 0 }
 
func (pos Position) String() string {
        s := pos.Filename
        if pos.IsValid() {
                if s != "" {
                        s += ":"
                }
                s += fmt.Sprintf("%d:%d", pos.Line, pos.Column)
        }
        if s == "" {
                s = "???"
        }
        return s
}
 
// Predefined mode bits to control recognition of tokens. For instance,
// to configure a Scanner such that it only recognizes (Go) identifiers,
// integers, and skips comments, set the Scanner's Mode field to:
//
//      ScanIdents | ScanInts | SkipComments
//
const (
        ScanIdents     = 1 << -Ident
        ScanInts       = 1 << -Int
        ScanFloats     = 1 << -Float // includes Ints
        ScanChars      = 1 << -Char
        ScanStrings    = 1 << -String
        ScanRawStrings = 1 << -RawString
        ScanComments   = 1 << -Comment
        SkipComments   = 1 << -skipComment // if set with ScanComments, comments become white space
        GoTokens       = ScanIdents | ScanFloats | ScanChars | ScanStrings | ScanRawStrings | ScanComments | SkipComments
)
 
// The result of Scan is one of the following tokens or a Unicode character.
const (
        EOF = -(iota + 1)
        Ident
        Int
        Float
        Char
        String
        RawString
        Comment
        skipComment
)
 
var tokenString = map[rune]string{
        EOF:       "EOF",
        Ident:     "Ident",
        Int:       "Int",
        Float:     "Float",
        Char:      "Char",
        String:    "String",
        RawString: "RawString",
        Comment:   "Comment",
}
 
// TokenString returns a (visible) string for a token or Unicode character.
func TokenString(tok rune) string {
        if s, found := tokenString[tok]; found {
                return s
        }
        return fmt.Sprintf("%q", string(tok))
}
 
// GoWhitespace is the default value for the Scanner's Whitespace field.
// Its value selects Go's white space characters.
const GoWhitespace = 1<<'\t' | 1<<'\n' | 1<<'\r' | 1<<' '
 
const bufLen = 1024 // at least utf8.UTFMax
 
// A Scanner implements reading of Unicode characters and tokens from an io.Reader.
type Scanner struct {
        // Input
        src io.Reader
 
        // Source buffer
        srcBuf [bufLen + 1]byte // +1 for sentinel for common case of s.next()
        srcPos int              // reading position (srcBuf index)
        srcEnd int              // source end (srcBuf index)
 
        // Source position
        srcBufOffset int // byte offset of srcBuf[0] in source
        line         int // line count
        column       int // character count
        lastLineLen  int // length of last line in characters (for correct column reporting)
        lastCharLen  int // length of last character in bytes
 
        // Token text buffer
        // Typically, token text is stored completely in srcBuf, but in general
        // the token text's head may be buffered in tokBuf while the token text's
        // tail is stored in srcBuf.
        tokBuf bytes.Buffer // token text head that is not in srcBuf anymore
        tokPos int          // token text tail position (srcBuf index); valid if >= 0
        tokEnd int          // token text tail end (srcBuf index)
 
        // One character look-ahead
        ch rune // character before current srcPos
 
        // Error is called for each error encountered. If no Error
        // function is set, the error is reported to os.Stderr.
        Error func(s *Scanner, msg string)
 
        // ErrorCount is incremented by one for each error encountered.
        ErrorCount int
 
        // The Mode field controls which tokens are recognized. For instance,
        // to recognize Ints, set the ScanInts bit in Mode. The field may be
        // changed at any time.
        Mode uint
 
        // The Whitespace field controls which characters are recognized
        // as white space. To recognize a character ch <= ' ' as white space,
        // set the ch'th bit in Whitespace (the Scanner's behavior is undefined
        // for values ch > ' '). The field may be changed at any time.
        Whitespace uint64
 
        // Start position of most recently scanned token; set by Scan.
        // Calling Init or Next invalidates the position (Line == 0).
        // 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. Call Pos to obtain an error
        // position in that case.
        Position
}
 
// Init initializes a Scanner with a new source and returns s.
// Error is set to nil, ErrorCount is set to 0, Mode is set to GoTokens,
// and Whitespace is set to GoWhitespace.
func (s *Scanner) Init(src io.Reader) *Scanner {
        s.src = src
 
        // initialize source buffer
        // (the first call to next() will fill it by calling src.Read)
        s.srcBuf[0] = utf8.RuneSelf // sentinel
        s.srcPos = 0
        s.srcEnd = 0
 
        // initialize source position
        s.srcBufOffset = 0
        s.line = 1
        s.column = 0
        s.lastLineLen = 0
        s.lastCharLen = 0
 
        // initialize token text buffer
        // (required for first call to next()).
        s.tokPos = -1
 
        // initialize one character look-ahead
        s.ch = -1 // no char read yet
 
        // initialize public fields
        s.Error = nil
        s.ErrorCount = 0
        s.Mode = GoTokens
        s.Whitespace = GoWhitespace
        s.Line = 0 // invalidate token position
 
        return s
}
 
// TODO(gri): The code for next() and the internal scanner state could benefit
//            from a rethink. While next() is optimized for the common ASCII
//            case, the "corrections" needed for proper position tracking undo
//            some of the attempts for fast-path optimization.
 
// next reads and returns the next Unicode character. It is designed such
// that only a minimal amount of work needs to be done in the common ASCII
// case (one test to check for both ASCII and end-of-buffer, and one test
// to check for newlines).
func (s *Scanner) next() rune {
        ch, width := rune(s.srcBuf[s.srcPos]), 1
 
        if ch >= utf8.RuneSelf {
                // uncommon case: not ASCII or not enough bytes
                for s.srcPos+utf8.UTFMax > s.srcEnd && !utf8.FullRune(s.srcBuf[s.srcPos:s.srcEnd]) {
                        // not enough bytes: read some more, but first
                        // save away token text if any
                        if s.tokPos >= 0 {
                                s.tokBuf.Write(s.srcBuf[s.tokPos:s.srcPos])
                                s.tokPos = 0
                                // s.tokEnd is set by Scan()
                        }
                        // move unread bytes to beginning of buffer
                        copy(s.srcBuf[0:], s.srcBuf[s.srcPos:s.srcEnd])
                        s.srcBufOffset += s.srcPos
                        // read more bytes
                        // (an io.Reader must return io.EOF when it reaches
                        // the end of what it is reading - simply returning
                        // n == 0 will make this loop retry forever; but the
                        // error is in the reader implementation in that case)
                        i := s.srcEnd - s.srcPos
                        n, err := s.src.Read(s.srcBuf[i:bufLen])
                        s.srcPos = 0
                        s.srcEnd = i + n
                        s.srcBuf[s.srcEnd] = utf8.RuneSelf // sentinel
                        if err != nil {
                                if s.srcEnd == 0 {
                                        if s.lastCharLen > 0 {
                                                // previous character was not EOF
                                                s.column++
                                        }
                                        s.lastCharLen = 0
                                        return EOF
                                }
                                if err != io.EOF {
                                        s.error(err.Error())
                                }
                                // If err == EOF, we won't be getting more
                                // bytes; break to avoid infinite loop. If
                                // err is something else, we don't know if
                                // we can get more bytes; thus also break.
                                break
                        }
                }
                // at least one byte
                ch = rune(s.srcBuf[s.srcPos])
                if ch >= utf8.RuneSelf {
                        // uncommon case: not ASCII
                        ch, width = utf8.DecodeRune(s.srcBuf[s.srcPos:s.srcEnd])
                        if ch == utf8.RuneError && width == 1 {
                                // advance for correct error position
                                s.srcPos += width
                                s.lastCharLen = width
                                s.column++
                                s.error("illegal UTF-8 encoding")
                                return ch
                        }
                }
        }
 
        // advance
        s.srcPos += width
        s.lastCharLen = width
        s.column++
 
        // special situations
        switch ch {
        case 0:
                // implementation restriction for compatibility with other tools
                s.error("illegal character NUL")
        case '\n':
                s.line++
                s.lastLineLen = s.column
                s.column = 0
        }
 
        return ch
}
 
// Next reads and returns the next Unicode character.
// It returns EOF at the end of the source. It reports
// a read error by calling s.Error, if not nil; otherwise
// it prints an error message to os.Stderr. Next does not
// update the Scanner's Position field; use Pos() to
// get the current position.
func (s *Scanner) Next() rune {
        s.tokPos = -1 // don't collect token text
        s.Line = 0    // invalidate token position
        ch := s.Peek()
        s.ch = s.next()
        return ch
}
 
// Peek returns the next Unicode character in the source without advancing
// the scanner. It returns EOF if the scanner's position is at the last
// character of the source.
func (s *Scanner) Peek() rune {
        if s.ch < 0 {
                s.ch = s.next()
        }
        return s.ch
}
 
func (s *Scanner) error(msg string) {
        s.ErrorCount++
        if s.Error != nil {
                s.Error(s, msg)
                return
        }
        pos := s.Position
        if !pos.IsValid() {
                pos = s.Pos()
        }
        fmt.Fprintf(os.Stderr, "%s: %s\n", pos, msg)
}
 
func (s *Scanner) scanIdentifier() rune {
        ch := s.next() // read character after first '_' or letter
        for ch == '_' || unicode.IsLetter(ch) || unicode.IsDigit(ch) {
                ch = s.next()
        }
        return ch
}
 
func digitVal(ch rune) int {
        switch {
        case '0' <= ch && ch <= '9':
                return int(ch - '0')
        case 'a' <= ch && ch <= 'f':
                return int(ch - 'a' + 10)
        case 'A' <= ch && ch <= 'F':
                return int(ch - 'A' + 10)
        }
        return 16 // larger than any legal digit val
}
 
func isDecimal(ch rune) bool { return '0' <= ch && ch <= '9' }
 
func (s *Scanner) scanMantissa(ch rune) rune {
        for isDecimal(ch) {
                ch = s.next()
        }
        return ch
}
 
func (s *Scanner) scanFraction(ch rune) rune {
        if ch == '.' {
                ch = s.scanMantissa(s.next())
        }
        return ch
}
 
func (s *Scanner) scanExponent(ch rune) rune {
        if ch == 'e' || ch == 'E' {
                ch = s.next()
                if ch == '-' || ch == '+' {
                        ch = s.next()
                }
                ch = s.scanMantissa(ch)
        }
        return ch
}
 
func (s *Scanner) scanNumber(ch rune) (rune, rune) {
        // isDecimal(ch)
        if ch == '0' {
                // int or float
                ch = s.next()
                if ch == 'x' || ch == 'X' {
                        // hexadecimal int
                        ch = s.next()
                        for digitVal(ch) < 16 {
                                ch = s.next()
                        }
                } else {
                        // octal int or float
                        seenDecimalDigit := false
                        for isDecimal(ch) {
                                if ch > '7' {
                                        seenDecimalDigit = true
                                }
                                ch = s.next()
                        }
                        if s.Mode&ScanFloats != 0 && (ch == '.' || ch == 'e' || ch == 'E') {
                                // float
                                ch = s.scanFraction(ch)
                                ch = s.scanExponent(ch)
                                return Float, ch
                        }
                        // octal int
                        if seenDecimalDigit {
                                s.error("illegal octal number")
                        }
                }
                return Int, ch
        }
        // decimal int or float
        ch = s.scanMantissa(ch)
        if s.Mode&ScanFloats != 0 && (ch == '.' || ch == 'e' || ch == 'E') {
                // float
                ch = s.scanFraction(ch)
                ch = s.scanExponent(ch)
                return Float, ch
        }
        return Int, ch
}
 
func (s *Scanner) scanDigits(ch rune, base, n int) rune {
        for n > 0 && digitVal(ch) < base {
                ch = s.next()
                n--
        }
        if n > 0 {
                s.error("illegal char escape")
        }
        return ch
}
 
func (s *Scanner) scanEscape(quote rune) rune {
        ch := s.next() // read character after '/'
        switch ch {
        case 'a', 'b', 'f', 'n', 'r', 't', 'v', '\\', quote:
                // nothing to do
                ch = s.next()
        case '0', '1', '2', '3', '4', '5', '6', '7':
                ch = s.scanDigits(ch, 8, 3)
        case 'x':
                ch = s.scanDigits(s.next(), 16, 2)
        case 'u':
                ch = s.scanDigits(s.next(), 16, 4)
        case 'U':
                ch = s.scanDigits(s.next(), 16, 8)
        default:
                s.error("illegal char escape")
        }
        return ch
}
 
func (s *Scanner) scanString(quote rune) (n int) {
        ch := s.next() // read character after quote
        for ch != quote {
                if ch == '\n' || ch < 0 {
                        s.error("literal not terminated")
                        return
                }
                if ch == '\\' {
                        ch = s.scanEscape(quote)
                } else {
                        ch = s.next()
                }
                n++
        }
        return
}
 
func (s *Scanner) scanRawString() {
        ch := s.next() // read character after '`'
        for ch != '`' {
                if ch < 0 {
                        s.error("literal not terminated")
                        return
                }
                ch = s.next()
        }
}
 
func (s *Scanner) scanChar() {
        if s.scanString('\'') != 1 {
                s.error("illegal char literal")
        }
}
 
func (s *Scanner) scanComment(ch rune) rune {
        // ch == '/' || ch == '*'
        if ch == '/' {
                // line comment
                ch = s.next() // read character after "//"
                for ch != '\n' && ch >= 0 {
                        ch = s.next()
                }
                return ch
        }
 
        // general comment
        ch = s.next() // read character after "/*"
        for {
                if ch < 0 {
                        s.error("comment not terminated")
                        break
                }
                ch0 := ch
                ch = s.next()
                if ch0 == '*' && ch == '/' {
                        ch = s.next()
                        break
                }
        }
        return ch
}
 
// Scan reads the next token or Unicode character from source and returns it.
// It only recognizes tokens t for which the respective Mode bit (1<<-t) is set.
// It returns EOF at the end of the source. It reports scanner errors (read and
// token errors) by calling s.Error, if not nil; otherwise it prints an error
// message to os.Stderr.
func (s *Scanner) Scan() rune {
        ch := s.Peek()
 
        // reset token text position
        s.tokPos = -1
        s.Line = 0
 
redo:
        // skip white space
        for s.Whitespace&(1<
                ch = s.next()
        }
 
        // start collecting token text
        s.tokBuf.Reset()
        s.tokPos = s.srcPos - s.lastCharLen
 
        // set token position
        // (this is a slightly optimized version of the code in Pos())
        s.Offset = s.srcBufOffset + s.tokPos
        if s.column > 0 {
                // common case: last character was not a '\n'
                s.Line = s.line
                s.Column = s.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.Line = s.line - 1
                s.Column = s.lastLineLen
        }
 
        // determine token value
        tok := ch
        switch {
        case unicode.IsLetter(ch) || ch == '_':
                if s.Mode&ScanIdents != 0 {
                        tok = Ident
                        ch = s.scanIdentifier()
                } else {
                        ch = s.next()
                }
        case isDecimal(ch):
                if s.Mode&(ScanInts|ScanFloats) != 0 {
                        tok, ch = s.scanNumber(ch)
                } else {
                        ch = s.next()
                }
        default:
                switch ch {
                case '"':
                        if s.Mode&ScanStrings != 0 {
                                s.scanString('"')
                                tok = String
                        }
                        ch = s.next()
                case '\'':
                        if s.Mode&ScanChars != 0 {
                                s.scanChar()
                                tok = Char
                        }
                        ch = s.next()
                case '.':
                        ch = s.next()
                        if isDecimal(ch) && s.Mode&ScanFloats != 0 {
                                tok = Float
                                ch = s.scanMantissa(ch)
                                ch = s.scanExponent(ch)
                        }
                case '/':
                        ch = s.next()
                        if (ch == '/' || ch == '*') && s.Mode&ScanComments != 0 {
                                if s.Mode&SkipComments != 0 {
                                        s.tokPos = -1 // don't collect token text
                                        ch = s.scanComment(ch)
                                        goto redo
                                }
                                ch = s.scanComment(ch)
                                tok = Comment
                        }
                case '`':
                        if s.Mode&ScanRawStrings != 0 {
                                s.scanRawString()
                                tok = String
                        }
                        ch = s.next()
                default:
                        ch = s.next()
                }
        }
 
        // end of token text
        s.tokEnd = s.srcPos - s.lastCharLen
 
        s.ch = ch
        return tok
}
 
// Pos returns the position of the character immediately after
// the character or token returned by the last call to Next or Scan.
func (s *Scanner) Pos() (pos Position) {
        pos.Filename = s.Filename
        pos.Offset = s.srcBufOffset + s.srcPos - s.lastCharLen
        switch {
        case s.column > 0:
                // common case: last character was not a '\n'
                pos.Line = s.line
                pos.Column = s.column
        case s.lastLineLen > 0:
                // last character was a '\n'
                pos.Line = s.line - 1
                pos.Column = s.lastLineLen
        default:
                // at the beginning of the source
                pos.Line = 1
                pos.Column = 1
        }
        return
}
 
// TokenText returns the string corresponding to the most recently scanned token.
// Valid after calling Scan().
func (s *Scanner) TokenText() string {
        if s.tokPos < 0 {
                // no token text
                return ""
        }
 
        if s.tokEnd < 0 {
                // if EOF was reached, s.tokEnd is set to -1 (s.srcPos == 0)
                s.tokEnd = s.tokPos
        }
 
        if s.tokBuf.Len() == 0 {
                // common case: the entire token text is still in srcBuf
                return string(s.srcBuf[s.tokPos:s.tokEnd])
        }
 
        // part of the token text was saved in tokBuf: save the rest in
        // tokBuf as well and return its content
        s.tokBuf.Write(s.srcBuf[s.tokPos:s.tokEnd])
        s.tokPos = s.tokEnd // ensure idempotency of TokenText() call
        return s.tokBuf.String()
}

Browse

Tools

Subversion Repositories openrisc

[/] [openrisc/] [trunk/] [gnu-dev/] [or1k-gcc/] [libgo/] [go/] [text/] [scanner/] [scanner.go] - Blame information for rev 858

Line No.	Rev	Author	Line
1	747	jeremybenn	`// Copyright 2009 The Go Authors. All rights reserved.`
2			`// Use of this source code is governed by a BSD-style`
3			`// license that can be found in the LICENSE file.`
4
5			`// Package scanner provides a scanner and tokenizer for UTF-8-encoded text.`
6			`// It takes an io.Reader providing the source, which then can be tokenized`
7			`// through repeated calls to the Scan function. For compatibility with`
8			`// existing tools, the NUL character is not allowed (implementation`
9			`// restriction).`
10			`//`
11			`// By default, a Scanner skips white space and Go comments and recognizes all`
12			`// literals as defined by the Go language specification. It may be`
13			`// customized to recognize only a subset of those literals and to recognize`
14			`// different white space characters.`
15			`//`
16			`// Basic usage pattern:`
17			`//`
18			`// var s scanner.Scanner`
19			`// s.Init(src)`
20			`// tok := s.Scan()`
21			`// for tok != scanner.EOF {`
22			`// // do something with tok`
23			`// tok = s.Scan()`
24			`// }`
25			`//`
26			`package scanner`
27
28			`import (`
29			`"bytes"`
30			`"fmt"`
31			`"io"`
32			`"os"`
33			`"unicode"`
34			`"unicode/utf8"`
35			`)`
36
37			`// TODO(gri): Consider changing this to use the new (token) Position package.`
38
39			`// A source position is represented by a Position value.`
40			`// A position is valid if Line > 0.`
41			`type Position struct {`
42			`Filename string // filename, if any`
43			`Offset int // byte offset, starting at 0`
44			`Line int // line number, starting at 1`
45			`Column int // column number, starting at 1 (character count per line)`
46			`}`
47
48			`// IsValid returns true if the position is valid.`
49			`func (pos *Position) IsValid() bool { return pos.Line > 0 }`
50
51			`func (pos Position) String() string {`
52			`s := pos.Filename`
53			`if pos.IsValid() {`
54			`if s != "" {`
55			`s += ":"`
56			`}`
57			`s += fmt.Sprintf("%d:%d", pos.Line, pos.Column)`
58			`}`
59			`if s == "" {`
60			`s = "???"`
61			`}`
62			`return s`
63			`}`
64
65			`// Predefined mode bits to control recognition of tokens. For instance,`
66			`// to configure a Scanner such that it only recognizes (Go) identifiers,`
67			`// integers, and skips comments, set the Scanner's Mode field to:`
68			`//`
69			`// ScanIdents \| ScanInts \| SkipComments`
70			`//`
71			`const (`
72			`ScanIdents = 1 << -Ident`
73			`ScanInts = 1 << -Int`
74			`ScanFloats = 1 << -Float // includes Ints`
75			`ScanChars = 1 << -Char`
76			`ScanStrings = 1 << -String`
77			`ScanRawStrings = 1 << -RawString`
78			`ScanComments = 1 << -Comment`
79			`SkipComments = 1 << -skipComment // if set with ScanComments, comments become white space`
80			`GoTokens = ScanIdents \| ScanFloats \| ScanChars \| ScanStrings \| ScanRawStrings \| ScanComments \| SkipComments`
81			`)`
82
83			`// The result of Scan is one of the following tokens or a Unicode character.`
84			`const (`
85			`EOF = -(iota + 1)`
86			`Ident`
87			`Int`
88			`Float`
89			`Char`
90			`String`
91			`RawString`
92			`Comment`
93			`skipComment`
94			`)`
95
96			`var tokenString = map[rune]string{`
97			`EOF: "EOF",`
98			`Ident: "Ident",`
99			`Int: "Int",`
100			`Float: "Float",`
101			`Char: "Char",`
102			`String: "String",`
103			`RawString: "RawString",`
104			`Comment: "Comment",`
105			`}`
106
107			`// TokenString returns a (visible) string for a token or Unicode character.`
108			`func TokenString(tok rune) string {`
109			`if s, found := tokenString[tok]; found {`
110			`return s`
111			`}`
112			`return fmt.Sprintf("%q", string(tok))`
113			`}`
114
115			`// GoWhitespace is the default value for the Scanner's Whitespace field.`
116			`// Its value selects Go's white space characters.`
117			`const GoWhitespace = 1<<'\t' \| 1<<'\n' \| 1<<'\r' \| 1<<' '`
118
119			`const bufLen = 1024 // at least utf8.UTFMax`
120
121			`// A Scanner implements reading of Unicode characters and tokens from an io.Reader.`
122			`type Scanner struct {`
123			`// Input`
124			`src io.Reader`
125
126			`// Source buffer`
127			`srcBuf [bufLen + 1]byte // +1 for sentinel for common case of s.next()`
128			`srcPos int // reading position (srcBuf index)`
129			`srcEnd int // source end (srcBuf index)`
130
131			`// Source position`
132			`srcBufOffset int // byte offset of srcBuf[0] in source`
133			`line int // line count`
134			`column int // character count`
135			`lastLineLen int // length of last line in characters (for correct column reporting)`
136			`lastCharLen int // length of last character in bytes`
137
138			`// Token text buffer`
139			`// Typically, token text is stored completely in srcBuf, but in general`
140			`// the token text's head may be buffered in tokBuf while the token text's`
141			`// tail is stored in srcBuf.`
142			`tokBuf bytes.Buffer // token text head that is not in srcBuf anymore`
143			`tokPos int // token text tail position (srcBuf index); valid if >= 0`
144			`tokEnd int // token text tail end (srcBuf index)`
145
146			`// One character look-ahead`
147			`ch rune // character before current srcPos`
148
149			`// Error is called for each error encountered. If no Error`
150			`// function is set, the error is reported to os.Stderr.`
151			`Error func(s *Scanner, msg string)`
152
153			`// ErrorCount is incremented by one for each error encountered.`
154			`ErrorCount int`
155
156			`// The Mode field controls which tokens are recognized. For instance,`
157			`// to recognize Ints, set the ScanInts bit in Mode. The field may be`
158			`// changed at any time.`
159			`Mode uint`
160
161			`// The Whitespace field controls which characters are recognized`
162			`// as white space. To recognize a character ch <= ' ' as white space,`
163			`// set the ch'th bit in Whitespace (the Scanner's behavior is undefined`
164			`// for values ch > ' '). The field may be changed at any time.`
165			`Whitespace uint64`
166
167			`// Start position of most recently scanned token; set by Scan.`
168			`// Calling Init or Next invalidates the position (Line == 0).`
169			`// The Filename field is always left untouched by the Scanner.`
170			`// If an error is reported (via Error) and Position is invalid,`
171			`// the scanner is not inside a token. Call Pos to obtain an error`
172			`// position in that case.`
173			`Position`
174			`}`
175
176			`// Init initializes a Scanner with a new source and returns s.`
177			`// Error is set to nil, ErrorCount is set to 0, Mode is set to GoTokens,`
178			`// and Whitespace is set to GoWhitespace.`
179			`func (s Scanner) Init(src io.Reader) Scanner {`
180			`s.src = src`
181
182			`// initialize source buffer`
183			`// (the first call to next() will fill it by calling src.Read)`
184			`s.srcBuf[0] = utf8.RuneSelf // sentinel`
185			`s.srcPos = 0`
186			`s.srcEnd = 0`
187
188			`// initialize source position`
189			`s.srcBufOffset = 0`
190			`s.line = 1`
191			`s.column = 0`
192			`s.lastLineLen = 0`
193			`s.lastCharLen = 0`
194
195			`// initialize token text buffer`
196			`// (required for first call to next()).`
197			`s.tokPos = -1`
198
199			`// initialize one character look-ahead`
200			`s.ch = -1 // no char read yet`
201
202			`// initialize public fields`
203			`s.Error = nil`
204			`s.ErrorCount = 0`
205			`s.Mode = GoTokens`
206			`s.Whitespace = GoWhitespace`
207			`s.Line = 0 // invalidate token position`
208
209			`return s`
210			`}`
211
212			`// TODO(gri): The code for next() and the internal scanner state could benefit`
213			`// from a rethink. While next() is optimized for the common ASCII`
214			`// case, the "corrections" needed for proper position tracking undo`
215			`// some of the attempts for fast-path optimization.`
216
217			`// next reads and returns the next Unicode character. It is designed such`
218			`// that only a minimal amount of work needs to be done in the common ASCII`
219			`// case (one test to check for both ASCII and end-of-buffer, and one test`
220			`// to check for newlines).`
221			`func (s *Scanner) next() rune {`
222			`ch, width := rune(s.srcBuf[s.srcPos]), 1`
223
224			`if ch >= utf8.RuneSelf {`
225			`// uncommon case: not ASCII or not enough bytes`
226			`for s.srcPos+utf8.UTFMax > s.srcEnd && !utf8.FullRune(s.srcBuf[s.srcPos:s.srcEnd]) {`
227			`// not enough bytes: read some more, but first`
228			`// save away token text if any`
229			`if s.tokPos >= 0 {`
230			`s.tokBuf.Write(s.srcBuf[s.tokPos:s.srcPos])`
231			`s.tokPos = 0`
232			`// s.tokEnd is set by Scan()`
233			`}`
234			`// move unread bytes to beginning of buffer`
235			`copy(s.srcBuf[0:], s.srcBuf[s.srcPos:s.srcEnd])`
236			`s.srcBufOffset += s.srcPos`
237			`// read more bytes`
238			`// (an io.Reader must return io.EOF when it reaches`
239			`// the end of what it is reading - simply returning`
240			`// n == 0 will make this loop retry forever; but the`
241			`// error is in the reader implementation in that case)`
242			`i := s.srcEnd - s.srcPos`
243			`n, err := s.src.Read(s.srcBuf[i:bufLen])`
244			`s.srcPos = 0`
245			`s.srcEnd = i + n`
246			`s.srcBuf[s.srcEnd] = utf8.RuneSelf // sentinel`
247			`if err != nil {`
248			`if s.srcEnd == 0 {`
249			`if s.lastCharLen > 0 {`
250			`// previous character was not EOF`
251			`s.column++`
252			`}`
253			`s.lastCharLen = 0`
254			`return EOF`
255			`}`
256			`if err != io.EOF {`
257			`s.error(err.Error())`
258			`}`
259			`// If err == EOF, we won't be getting more`
260			`// bytes; break to avoid infinite loop. If`
261			`// err is something else, we don't know if`
262			`// we can get more bytes; thus also break.`
263			`break`
264			`}`
265			`}`
266			`// at least one byte`
267			`ch = rune(s.srcBuf[s.srcPos])`
268			`if ch >= utf8.RuneSelf {`
269			`// uncommon case: not ASCII`
270			`ch, width = utf8.DecodeRune(s.srcBuf[s.srcPos:s.srcEnd])`
271			`if ch == utf8.RuneError && width == 1 {`
272			`// advance for correct error position`
273			`s.srcPos += width`
274			`s.lastCharLen = width`
275			`s.column++`
276			`s.error("illegal UTF-8 encoding")`
277			`return ch`
278			`}`
279			`}`
280			`}`
281
282			`// advance`
283			`s.srcPos += width`
284			`s.lastCharLen = width`
285			`s.column++`
286
287			`// special situations`
288			`switch ch {`
289			`case 0:`
290			`// implementation restriction for compatibility with other tools`
291			`s.error("illegal character NUL")`
292			`case '\n':`
293			`s.line++`
294			`s.lastLineLen = s.column`
295			`s.column = 0`
296			`}`
297
298			`return ch`
299			`}`
300
301			`// Next reads and returns the next Unicode character.`
302			`// It returns EOF at the end of the source. It reports`
303			`// a read error by calling s.Error, if not nil; otherwise`
304			`// it prints an error message to os.Stderr. Next does not`
305			`// update the Scanner's Position field; use Pos() to`
306			`// get the current position.`
307			`func (s *Scanner) Next() rune {`
308			`s.tokPos = -1 // don't collect token text`
309			`s.Line = 0 // invalidate token position`
310			`ch := s.Peek()`
311			`s.ch = s.next()`
312			`return ch`
313			`}`
314
315			`// Peek returns the next Unicode character in the source without advancing`
316			`// the scanner. It returns EOF if the scanner's position is at the last`
317			`// character of the source.`
318			`func (s *Scanner) Peek() rune {`
319			`if s.ch < 0 {`
320			`s.ch = s.next()`
321			`}`
322			`return s.ch`
323			`}`
324
325			`func (s *Scanner) error(msg string) {`
326			`s.ErrorCount++`
327			`if s.Error != nil {`
328			`s.Error(s, msg)`
329			`return`
330			`}`
331			`pos := s.Position`
332			`if !pos.IsValid() {`
333			`pos = s.Pos()`
334			`}`
335			`fmt.Fprintf(os.Stderr, "%s: %s\n", pos, msg)`
336			`}`
337
338			`func (s *Scanner) scanIdentifier() rune {`
339			`ch := s.next() // read character after first '_' or letter`
340			`for ch == '_' \|\| unicode.IsLetter(ch) \|\| unicode.IsDigit(ch) {`
341			`ch = s.next()`
342			`}`
343			`return ch`
344			`}`
345
346			`func digitVal(ch rune) int {`
347			`switch {`
348			`case '0' <= ch && ch <= '9':`
349			`return int(ch - '0')`
350			`case 'a' <= ch && ch <= 'f':`
351			`return int(ch - 'a' + 10)`
352			`case 'A' <= ch && ch <= 'F':`
353			`return int(ch - 'A' + 10)`
354			`}`
355			`return 16 // larger than any legal digit val`
356			`}`
357
358			`func isDecimal(ch rune) bool { return '0' <= ch && ch <= '9' }`
359
360			`func (s *Scanner) scanMantissa(ch rune) rune {`
361			`for isDecimal(ch) {`
362			`ch = s.next()`
363			`}`
364			`return ch`
365			`}`
366
367			`func (s *Scanner) scanFraction(ch rune) rune {`
368			`if ch == '.' {`
369			`ch = s.scanMantissa(s.next())`
370			`}`
371			`return ch`
372			`}`
373
374			`func (s *Scanner) scanExponent(ch rune) rune {`
375			`if ch == 'e' \|\| ch == 'E' {`
376			`ch = s.next()`
377			`if ch == '-' \|\| ch == '+' {`
378			`ch = s.next()`
379			`}`
380			`ch = s.scanMantissa(ch)`
381			`}`
382			`return ch`
383			`}`
384
385			`func (s *Scanner) scanNumber(ch rune) (rune, rune) {`
386			`// isDecimal(ch)`
387			`if ch == '0' {`
388			`// int or float`
389			`ch = s.next()`
390			`if ch == 'x' \|\| ch == 'X' {`
391			`// hexadecimal int`
392			`ch = s.next()`
393			`for digitVal(ch) < 16 {`
394			`ch = s.next()`
395			`}`
396			`} else {`
397			`// octal int or float`
398			`seenDecimalDigit := false`
399			`for isDecimal(ch) {`
400			`if ch > '7' {`
401			`seenDecimalDigit = true`
402			`}`
403			`ch = s.next()`
404			`}`
405			`if s.Mode&ScanFloats != 0 && (ch == '.' \|\| ch == 'e' \|\| ch == 'E') {`
406			`// float`
407			`ch = s.scanFraction(ch)`
408			`ch = s.scanExponent(ch)`
409			`return Float, ch`
410			`}`
411			`// octal int`
412			`if seenDecimalDigit {`
413			`s.error("illegal octal number")`
414			`}`
415			`}`
416			`return Int, ch`
417			`}`
418			`// decimal int or float`
419			`ch = s.scanMantissa(ch)`
420			`if s.Mode&ScanFloats != 0 && (ch == '.' \|\| ch == 'e' \|\| ch == 'E') {`
421			`// float`
422			`ch = s.scanFraction(ch)`
423			`ch = s.scanExponent(ch)`
424			`return Float, ch`
425			`}`
426			`return Int, ch`
427			`}`
428
429			`func (s *Scanner) scanDigits(ch rune, base, n int) rune {`
430			`for n > 0 && digitVal(ch) < base {`
431			`ch = s.next()`
432			`n--`
433			`}`
434			`if n > 0 {`
435			`s.error("illegal char escape")`
436			`}`
437			`return ch`
438			`}`
439
440			`func (s *Scanner) scanEscape(quote rune) rune {`
441			`ch := s.next() // read character after '/'`
442			`switch ch {`
443			`case 'a', 'b', 'f', 'n', 'r', 't', 'v', '\\', quote:`
444			`// nothing to do`
445			`ch = s.next()`
446			`case '0', '1', '2', '3', '4', '5', '6', '7':`
447			`ch = s.scanDigits(ch, 8, 3)`
448			`case 'x':`
449			`ch = s.scanDigits(s.next(), 16, 2)`
450			`case 'u':`
451			`ch = s.scanDigits(s.next(), 16, 4)`
452			`case 'U':`
453			`ch = s.scanDigits(s.next(), 16, 8)`
454			`default:`
455			`s.error("illegal char escape")`
456			`}`
457			`return ch`
458			`}`
459
460			`func (s *Scanner) scanString(quote rune) (n int) {`
461			`ch := s.next() // read character after quote`
462			`for ch != quote {`
463			`if ch == '\n' \|\| ch < 0 {`
464			`s.error("literal not terminated")`
465			`return`
466			`}`
467			`if ch == '\\' {`
468			`ch = s.scanEscape(quote)`
469			`} else {`
470			`ch = s.next()`
471			`}`
472			`n++`
473			`}`
474			`return`
475			`}`
476
477			`func (s *Scanner) scanRawString() {`
478			ch := s.next() // read character after '`'
479			for ch != '`' {
480			`if ch < 0 {`
481			`s.error("literal not terminated")`
482			`return`
483			`}`
484			`ch = s.next()`
485			`}`
486			`}`
487
488			`func (s *Scanner) scanChar() {`
489			`if s.scanString('\'') != 1 {`
490			`s.error("illegal char literal")`
491			`}`
492			`}`
493
494			`func (s *Scanner) scanComment(ch rune) rune {`
495			`// ch == '/' \|\| ch == '*'`
496			`if ch == '/' {`
497			`// line comment`
498			`ch = s.next() // read character after "//"`
499			`for ch != '\n' && ch >= 0 {`
500			`ch = s.next()`
501			`}`
502			`return ch`
503			`}`
504
505			`// general comment`
506			`ch = s.next() // read character after "/*"`
507			`for {`
508			`if ch < 0 {`
509			`s.error("comment not terminated")`
510			`break`
511			`}`
512			`ch0 := ch`
513			`ch = s.next()`
514			`if ch0 == '*' && ch == '/' {`
515			`ch = s.next()`
516			`break`
517			`}`
518			`}`
519			`return ch`
520			`}`
521
522			`// Scan reads the next token or Unicode character from source and returns it.`
523			`// It only recognizes tokens t for which the respective Mode bit (1<<-t) is set.`
524			`// It returns EOF at the end of the source. It reports scanner errors (read and`
525			`// token errors) by calling s.Error, if not nil; otherwise it prints an error`
526			`// message to os.Stderr.`
527			`func (s *Scanner) Scan() rune {`
528			`ch := s.Peek()`
529
530			`// reset token text position`
531			`s.tokPos = -1`
532			`s.Line = 0`
533
534			`redo:`
535			`// skip white space`
536			`for s.Whitespace&(1<`
537			`ch = s.next()`
538			`}`
539
540			`// start collecting token text`
541			`s.tokBuf.Reset()`
542			`s.tokPos = s.srcPos - s.lastCharLen`
543
544			`// set token position`
545			`// (this is a slightly optimized version of the code in Pos())`
546			`s.Offset = s.srcBufOffset + s.tokPos`
547			`if s.column > 0 {`
548			`// common case: last character was not a '\n'`
549			`s.Line = s.line`
550			`s.Column = s.column`
551			`} else {`
552			`// last character was a '\n'`
553			`// (we cannot be at the beginning of the source`
554			`// since we have called next() at least once)`
555			`s.Line = s.line - 1`
556			`s.Column = s.lastLineLen`
557			`}`
558
559			`// determine token value`
560			`tok := ch`
561			`switch {`
562			`case unicode.IsLetter(ch) \|\| ch == '_':`
563			`if s.Mode&ScanIdents != 0 {`
564			`tok = Ident`
565			`ch = s.scanIdentifier()`
566			`} else {`
567			`ch = s.next()`
568			`}`
569			`case isDecimal(ch):`
570			`if s.Mode&(ScanInts\|ScanFloats) != 0 {`
571			`tok, ch = s.scanNumber(ch)`
572			`} else {`
573			`ch = s.next()`
574			`}`
575			`default:`
576			`switch ch {`
577			`case '"':`
578			`if s.Mode&ScanStrings != 0 {`
579			`s.scanString('"')`
580			`tok = String`
581			`}`
582			`ch = s.next()`
583			`case '\'':`
584			`if s.Mode&ScanChars != 0 {`
585			`s.scanChar()`
586			`tok = Char`
587			`}`
588			`ch = s.next()`
589			`case '.':`
590			`ch = s.next()`
591			`if isDecimal(ch) && s.Mode&ScanFloats != 0 {`
592			`tok = Float`
593			`ch = s.scanMantissa(ch)`
594			`ch = s.scanExponent(ch)`
595			`}`
596			`case '/':`
597			`ch = s.next()`
598			`if (ch == '/' \|\| ch == '*') && s.Mode&ScanComments != 0 {`
599			`if s.Mode&SkipComments != 0 {`
600			`s.tokPos = -1 // don't collect token text`
601			`ch = s.scanComment(ch)`
602			`goto redo`
603			`}`
604			`ch = s.scanComment(ch)`
605			`tok = Comment`
606			`}`
607			case '`':
608			`if s.Mode&ScanRawStrings != 0 {`
609			`s.scanRawString()`
610			`tok = String`
611			`}`
612			`ch = s.next()`
613			`default:`
614			`ch = s.next()`
615			`}`
616			`}`
617
618			`// end of token text`
619			`s.tokEnd = s.srcPos - s.lastCharLen`
620
621			`s.ch = ch`
622			`return tok`
623			`}`
624
625			`// Pos returns the position of the character immediately after`
626			`// the character or token returned by the last call to Next or Scan.`
627			`func (s *Scanner) Pos() (pos Position) {`
628			`pos.Filename = s.Filename`
629			`pos.Offset = s.srcBufOffset + s.srcPos - s.lastCharLen`
630			`switch {`
631			`case s.column > 0:`
632			`// common case: last character was not a '\n'`
633			`pos.Line = s.line`
634			`pos.Column = s.column`
635			`case s.lastLineLen > 0:`
636			`// last character was a '\n'`
637			`pos.Line = s.line - 1`
638			`pos.Column = s.lastLineLen`
639			`default:`
640			`// at the beginning of the source`
641			`pos.Line = 1`
642			`pos.Column = 1`
643			`}`
644			`return`
645			`}`
646
647			`// TokenText returns the string corresponding to the most recently scanned token.`
648			`// Valid after calling Scan().`
649			`func (s *Scanner) TokenText() string {`
650			`if s.tokPos < 0 {`
651			`// no token text`
652			`return ""`
653			`}`
654
655			`if s.tokEnd < 0 {`
656			`// if EOF was reached, s.tokEnd is set to -1 (s.srcPos == 0)`
657			`s.tokEnd = s.tokPos`
658			`}`
659
660			`if s.tokBuf.Len() == 0 {`
661			`// common case: the entire token text is still in srcBuf`
662			`return string(s.srcBuf[s.tokPos:s.tokEnd])`
663			`}`
664
665			`// part of the token text was saved in tokBuf: save the rest in`
666			`// tokBuf as well and return its content`
667			`s.tokBuf.Write(s.srcBuf[s.tokPos:s.tokEnd])`
668			`s.tokPos = s.tokEnd // ensure idempotency of TokenText() call`
669			`return s.tokBuf.String()`
670			`}`