matterbridge/vendor/github.com/d5/tengo/objects/formatter.go

1213 lines
28 KiB
Go
Raw Normal View History

2019-06-16 23:33:25 +02:00
package objects
import (
"strconv"
"sync"
"unicode/utf8"
"github.com/d5/tengo"
)
// Strings for use with buffer.WriteString.
// This is less overhead than using buffer.Write with byte arrays.
const (
commaSpaceString = ", "
nilParenString = "(nil)"
percentBangString = "%!"
missingString = "(MISSING)"
badIndexString = "(BADINDEX)"
extraString = "%!(EXTRA "
badWidthString = "%!(BADWIDTH)"
badPrecString = "%!(BADPREC)"
noVerbString = "%!(NOVERB)"
)
const (
ldigits = "0123456789abcdefx"
udigits = "0123456789ABCDEFX"
)
const (
signed = true
unsigned = false
)
// flags placed in a separate struct for easy clearing.
type fmtFlags struct {
widPresent bool
precPresent bool
minus bool
plus bool
sharp bool
space bool
zero bool
// For the formats %+v %#v, we set the plusV/sharpV flags
// and clear the plus/sharp flags since %+v and %#v are in effect
// different, flagless formats set at the top level.
plusV bool
sharpV bool
// error-related flags.
inDetail bool
needNewline bool
needColon bool
}
// A formatter is the raw formatter used by Printf etc.
// It prints into a buffer that must be set up separately.
type formatter struct {
buf *buffer
fmtFlags
wid int // width
prec int // precision
// intbuf is large enough to store %b of an int64 with a sign and
// avoids padding at the end of the struct on 32 bit architectures.
intbuf [68]byte
}
func (f *formatter) clearflags() {
f.fmtFlags = fmtFlags{}
}
func (f *formatter) init(buf *buffer) {
f.buf = buf
f.clearflags()
}
// writePadding generates n bytes of padding.
func (f *formatter) writePadding(n int) {
if n <= 0 { // No padding bytes needed.
return
}
buf := *f.buf
oldLen := len(buf)
newLen := oldLen + n
if newLen > tengo.MaxStringLen {
panic(ErrStringLimit)
}
// Make enough room for padding.
if newLen > cap(buf) {
buf = make(buffer, cap(buf)*2+n)
copy(buf, *f.buf)
}
// Decide which byte the padding should be filled with.
padByte := byte(' ')
if f.zero {
padByte = byte('0')
}
// Fill padding with padByte.
padding := buf[oldLen:newLen]
for i := range padding {
padding[i] = padByte
}
*f.buf = buf[:newLen]
}
// pad appends b to f.buf, padded on left (!f.minus) or right (f.minus).
func (f *formatter) pad(b []byte) {
if !f.widPresent || f.wid == 0 {
f.buf.Write(b)
return
}
width := f.wid - utf8.RuneCount(b)
if !f.minus {
// left padding
f.writePadding(width)
f.buf.Write(b)
} else {
// right padding
f.buf.Write(b)
f.writePadding(width)
}
}
// padString appends s to f.buf, padded on left (!f.minus) or right (f.minus).
func (f *formatter) padString(s string) {
if !f.widPresent || f.wid == 0 {
f.buf.WriteString(s)
return
}
width := f.wid - utf8.RuneCountInString(s)
if !f.minus {
// left padding
f.writePadding(width)
f.buf.WriteString(s)
} else {
// right padding
f.buf.WriteString(s)
f.writePadding(width)
}
}
// fmtBoolean formats a boolean.
func (f *formatter) fmtBoolean(v bool) {
if v {
f.padString("true")
} else {
f.padString("false")
}
}
// fmtUnicode formats a uint64 as "U+0078" or with f.sharp set as "U+0078 'x'".
func (f *formatter) fmtUnicode(u uint64) {
buf := f.intbuf[0:]
// With default precision set the maximum needed buf length is 18
// for formatting -1 with %#U ("U+FFFFFFFFFFFFFFFF") which fits
// into the already allocated intbuf with a capacity of 68 bytes.
prec := 4
if f.precPresent && f.prec > 4 {
prec = f.prec
// Compute space needed for "U+" , number, " '", character, "'".
width := 2 + prec + 2 + utf8.UTFMax + 1
if width > len(buf) {
buf = make([]byte, width)
}
}
// Format into buf, ending at buf[i]. Formatting numbers is easier right-to-left.
i := len(buf)
// For %#U we want to add a space and a quoted character at the end of the buffer.
if f.sharp && u <= utf8.MaxRune && strconv.IsPrint(rune(u)) {
i--
buf[i] = '\''
i -= utf8.RuneLen(rune(u))
utf8.EncodeRune(buf[i:], rune(u))
i--
buf[i] = '\''
i--
buf[i] = ' '
}
// Format the Unicode code point u as a hexadecimal number.
for u >= 16 {
i--
buf[i] = udigits[u&0xF]
prec--
u >>= 4
}
i--
buf[i] = udigits[u]
prec--
// Add zeros in front of the number until requested precision is reached.
for prec > 0 {
i--
buf[i] = '0'
prec--
}
// Add a leading "U+".
i--
buf[i] = '+'
i--
buf[i] = 'U'
oldZero := f.zero
f.zero = false
f.pad(buf[i:])
f.zero = oldZero
}
// fmtInteger formats signed and unsigned integers.
func (f *formatter) fmtInteger(u uint64, base int, isSigned bool, verb rune, digits string) {
negative := isSigned && int64(u) < 0
if negative {
u = -u
}
buf := f.intbuf[0:]
// The already allocated f.intbuf with a capacity of 68 bytes
// is large enough for integer formatting when no precision or width is set.
if f.widPresent || f.precPresent {
// Account 3 extra bytes for possible addition of a sign and "0x".
width := 3 + f.wid + f.prec // wid and prec are always positive.
if width > len(buf) {
// We're going to need a bigger boat.
buf = make([]byte, width)
}
}
// Two ways to ask for extra leading zero digits: %.3d or %03d.
// If both are specified the f.zero flag is ignored and
// padding with spaces is used instead.
prec := 0
if f.precPresent {
prec = f.prec
// Precision of 0 and value of 0 means "print nothing" but padding.
if prec == 0 && u == 0 {
oldZero := f.zero
f.zero = false
f.writePadding(f.wid)
f.zero = oldZero
return
}
} else if f.zero && f.widPresent {
prec = f.wid
if negative || f.plus || f.space {
prec-- // leave room for sign
}
}
// Because printing is easier right-to-left: format u into buf, ending at buf[i].
// We could make things marginally faster by splitting the 32-bit case out
// into a separate block but it's not worth the duplication, so u has 64 bits.
i := len(buf)
// Use constants for the division and modulo for more efficient code.
// Switch cases ordered by popularity.
switch base {
case 10:
for u >= 10 {
i--
next := u / 10
buf[i] = byte('0' + u - next*10)
u = next
}
case 16:
for u >= 16 {
i--
buf[i] = digits[u&0xF]
u >>= 4
}
case 8:
for u >= 8 {
i--
buf[i] = byte('0' + u&7)
u >>= 3
}
case 2:
for u >= 2 {
i--
buf[i] = byte('0' + u&1)
u >>= 1
}
default:
panic("fmt: unknown base; can't happen")
}
i--
buf[i] = digits[u]
for i > 0 && prec > len(buf)-i {
i--
buf[i] = '0'
}
// Various prefixes: 0x, -, etc.
if f.sharp {
switch base {
case 2:
// Add a leading 0b.
i--
buf[i] = 'b'
i--
buf[i] = '0'
case 8:
if buf[i] != '0' {
i--
buf[i] = '0'
}
case 16:
// Add a leading 0x or 0X.
i--
buf[i] = digits[16]
i--
buf[i] = '0'
}
}
if verb == 'O' {
i--
buf[i] = 'o'
i--
buf[i] = '0'
}
if negative {
i--
buf[i] = '-'
} else if f.plus {
i--
buf[i] = '+'
} else if f.space {
i--
buf[i] = ' '
}
// Left padding with zeros has already been handled like precision earlier
// or the f.zero flag is ignored due to an explicitly set precision.
oldZero := f.zero
f.zero = false
f.pad(buf[i:])
f.zero = oldZero
}
// truncate truncates the string s to the specified precision, if present.
func (f *formatter) truncateString(s string) string {
if f.precPresent {
n := f.prec
for i := range s {
n--
if n < 0 {
return s[:i]
}
}
}
return s
}
// truncate truncates the byte slice b as a string of the specified precision, if present.
func (f *formatter) truncate(b []byte) []byte {
if f.precPresent {
n := f.prec
for i := 0; i < len(b); {
n--
if n < 0 {
return b[:i]
}
wid := 1
if b[i] >= utf8.RuneSelf {
_, wid = utf8.DecodeRune(b[i:])
}
i += wid
}
}
return b
}
// fmtS formats a string.
func (f *formatter) fmtS(s string) {
s = f.truncateString(s)
f.padString(s)
}
// fmtBs formats the byte slice b as if it was formatted as string with fmtS.
func (f *formatter) fmtBs(b []byte) {
b = f.truncate(b)
f.pad(b)
}
// fmtSbx formats a string or byte slice as a hexadecimal encoding of its bytes.
func (f *formatter) fmtSbx(s string, b []byte, digits string) {
length := len(b)
if b == nil {
// No byte slice present. Assume string s should be encoded.
length = len(s)
}
// Set length to not process more bytes than the precision demands.
if f.precPresent && f.prec < length {
length = f.prec
}
// Compute width of the encoding taking into account the f.sharp and f.space flag.
width := 2 * length
if width > 0 {
if f.space {
// Each element encoded by two hexadecimals will get a leading 0x or 0X.
if f.sharp {
width *= 2
}
// Elements will be separated by a space.
width += length - 1
} else if f.sharp {
// Only a leading 0x or 0X will be added for the whole string.
width += 2
}
} else { // The byte slice or string that should be encoded is empty.
if f.widPresent {
f.writePadding(f.wid)
}
return
}
// Handle padding to the left.
if f.widPresent && f.wid > width && !f.minus {
f.writePadding(f.wid - width)
}
// Write the encoding directly into the output buffer.
buf := *f.buf
if f.sharp {
// Add leading 0x or 0X.
buf = append(buf, '0', digits[16])
}
var c byte
for i := 0; i < length; i++ {
if f.space && i > 0 {
// Separate elements with a space.
buf = append(buf, ' ')
if f.sharp {
// Add leading 0x or 0X for each element.
buf = append(buf, '0', digits[16])
}
}
if b != nil {
c = b[i] // Take a byte from the input byte slice.
} else {
c = s[i] // Take a byte from the input string.
}
// Encode each byte as two hexadecimal digits.
buf = append(buf, digits[c>>4], digits[c&0xF])
}
*f.buf = buf
// Handle padding to the right.
if f.widPresent && f.wid > width && f.minus {
f.writePadding(f.wid - width)
}
}
// fmtSx formats a string as a hexadecimal encoding of its bytes.
func (f *formatter) fmtSx(s, digits string) {
f.fmtSbx(s, nil, digits)
}
// fmtBx formats a byte slice as a hexadecimal encoding of its bytes.
func (f *formatter) fmtBx(b []byte, digits string) {
f.fmtSbx("", b, digits)
}
// fmtQ formats a string as a double-quoted, escaped Go string constant.
// If f.sharp is set a raw (backquoted) string may be returned instead
// if the string does not contain any control characters other than tab.
func (f *formatter) fmtQ(s string) {
s = f.truncateString(s)
if f.sharp && strconv.CanBackquote(s) {
f.padString("`" + s + "`")
return
}
buf := f.intbuf[:0]
if f.plus {
f.pad(strconv.AppendQuoteToASCII(buf, s))
} else {
f.pad(strconv.AppendQuote(buf, s))
}
}
// fmtC formats an integer as a Unicode character.
// If the character is not valid Unicode, it will print '\ufffd'.
func (f *formatter) fmtC(c uint64) {
r := rune(c)
if c > utf8.MaxRune {
r = utf8.RuneError
}
buf := f.intbuf[:0]
w := utf8.EncodeRune(buf[:utf8.UTFMax], r)
f.pad(buf[:w])
}
// fmtQc formats an integer as a single-quoted, escaped Go character constant.
// If the character is not valid Unicode, it will print '\ufffd'.
func (f *formatter) fmtQc(c uint64) {
r := rune(c)
if c > utf8.MaxRune {
r = utf8.RuneError
}
buf := f.intbuf[:0]
if f.plus {
f.pad(strconv.AppendQuoteRuneToASCII(buf, r))
} else {
f.pad(strconv.AppendQuoteRune(buf, r))
}
}
// fmtFloat formats a float64. It assumes that verb is a valid format specifier
// for strconv.AppendFloat and therefore fits into a byte.
func (f *formatter) fmtFloat(v float64, size int, verb rune, prec int) {
// Explicit precision in format specifier overrules default precision.
if f.precPresent {
prec = f.prec
}
// Format number, reserving space for leading + sign if needed.
num := strconv.AppendFloat(f.intbuf[:1], v, byte(verb), prec, size)
if num[1] == '-' || num[1] == '+' {
num = num[1:]
} else {
num[0] = '+'
}
// f.space means to add a leading space instead of a "+" sign unless
// the sign is explicitly asked for by f.plus.
if f.space && num[0] == '+' && !f.plus {
num[0] = ' '
}
// Special handling for infinities and NaN,
// which don't look like a number so shouldn't be padded with zeros.
if num[1] == 'I' || num[1] == 'N' {
oldZero := f.zero
f.zero = false
// Remove sign before NaN if not asked for.
if num[1] == 'N' && !f.space && !f.plus {
num = num[1:]
}
f.pad(num)
f.zero = oldZero
return
}
// The sharp flag forces printing a decimal point for non-binary formats
// and retains trailing zeros, which we may need to restore.
if f.sharp && verb != 'b' {
digits := 0
switch verb {
case 'v', 'g', 'G', 'x':
digits = prec
// If no precision is set explicitly use a precision of 6.
if digits == -1 {
digits = 6
}
}
// Buffer pre-allocated with enough room for
// exponent notations of the form "e+123" or "p-1023".
var tailBuf [6]byte
tail := tailBuf[:0]
hasDecimalPoint := false
// Starting from i = 1 to skip sign at num[0].
for i := 1; i < len(num); i++ {
switch num[i] {
case '.':
hasDecimalPoint = true
case 'p', 'P':
tail = append(tail, num[i:]...)
num = num[:i]
case 'e', 'E':
if verb != 'x' && verb != 'X' {
tail = append(tail, num[i:]...)
num = num[:i]
break
}
fallthrough
default:
digits--
}
}
if !hasDecimalPoint {
num = append(num, '.')
}
for digits > 0 {
num = append(num, '0')
digits--
}
num = append(num, tail...)
}
// We want a sign if asked for and if the sign is not positive.
if f.plus || num[0] != '+' {
// If we're zero padding to the left we want the sign before the leading zeros.
// Achieve this by writing the sign out and then padding the unsigned number.
if f.zero && f.widPresent && f.wid > len(num) {
f.buf.WriteSingleByte(num[0])
f.writePadding(f.wid - len(num))
f.buf.Write(num[1:])
return
}
f.pad(num)
return
}
// No sign to show and the number is positive; just print the unsigned number.
f.pad(num[1:])
}
// Use simple []byte instead of bytes.Buffer to avoid large dependency.
type buffer []byte
func (b *buffer) Write(p []byte) {
if len(*b)+len(p) > tengo.MaxStringLen {
panic(ErrStringLimit)
}
*b = append(*b, p...)
}
func (b *buffer) WriteString(s string) {
if len(*b)+len(s) > tengo.MaxStringLen {
panic(ErrStringLimit)
}
*b = append(*b, s...)
}
func (b *buffer) WriteSingleByte(c byte) {
if len(*b) >= tengo.MaxStringLen {
panic(ErrStringLimit)
}
*b = append(*b, c)
}
func (b *buffer) WriteRune(r rune) {
if len(*b)+utf8.RuneLen(r) > tengo.MaxStringLen {
panic(ErrStringLimit)
}
if r < utf8.RuneSelf {
*b = append(*b, byte(r))
return
}
b2 := *b
n := len(b2)
for n+utf8.UTFMax > cap(b2) {
b2 = append(b2, 0)
}
w := utf8.EncodeRune(b2[n:n+utf8.UTFMax], r)
*b = b2[:n+w]
}
// pp is used to store a printer's state and is reused with sync.Pool to avoid allocations.
type pp struct {
buf buffer
// arg holds the current item.
arg Object
// fmt is used to format basic items such as integers or strings.
fmt formatter
// reordered records whether the format string used argument reordering.
reordered bool
// goodArgNum records whether the most recent reordering directive was valid.
goodArgNum bool
// erroring is set when printing an error string to guard against calling handleMethods.
erroring bool
}
var ppFree = sync.Pool{
New: func() interface{} { return new(pp) },
}
// newPrinter allocates a new pp struct or grabs a cached one.
func newPrinter() *pp {
p := ppFree.Get().(*pp)
p.erroring = false
p.fmt.init(&p.buf)
return p
}
// free saves used pp structs in ppFree; avoids an allocation per invocation.
func (p *pp) free() {
// Proper usage of a sync.Pool requires each entry to have approximately
// the same memory cost. To obtain this property when the stored type
// contains a variably-sized buffer, we add a hard limit on the maximum buffer
// to place back in the pool.
//
// See https://golang.org/issue/23199
if cap(p.buf) > 64<<10 {
return
}
p.buf = p.buf[:0]
p.arg = nil
ppFree.Put(p)
}
func (p *pp) Width() (wid int, ok bool) { return p.fmt.wid, p.fmt.widPresent }
func (p *pp) Precision() (prec int, ok bool) { return p.fmt.prec, p.fmt.precPresent }
func (p *pp) Flag(b int) bool {
switch b {
case '-':
return p.fmt.minus
case '+':
return p.fmt.plus || p.fmt.plusV
case '#':
return p.fmt.sharp || p.fmt.sharpV
case ' ':
return p.fmt.space
case '0':
return p.fmt.zero
}
return false
}
// Implement Write so we can call Fprintf on a pp (through State), for
// recursive use in custom verbs.
func (p *pp) Write(b []byte) (ret int, err error) {
p.buf.Write(b)
return len(b), nil
}
// Implement WriteString so that we can call io.WriteString
// on a pp (through state), for efficiency.
func (p *pp) WriteString(s string) (ret int, err error) {
p.buf.WriteString(s)
return len(s), nil
}
func (p *pp) WriteRune(r rune) (ret int, err error) {
p.buf.WriteRune(r)
return utf8.RuneLen(r), nil
}
func (p *pp) WriteSingleByte(c byte) (ret int, err error) {
p.buf.WriteSingleByte(c)
return 1, nil
}
// tooLarge reports whether the magnitude of the integer is
// too large to be used as a formatting width or precision.
func tooLarge(x int) bool {
const max int = 1e6
return x > max || x < -max
}
// parsenum converts ASCII to integer. num is 0 (and isnum is false) if no number present.
func parsenum(s string, start, end int) (num int, isnum bool, newi int) {
if start >= end {
return 0, false, end
}
for newi = start; newi < end && '0' <= s[newi] && s[newi] <= '9'; newi++ {
if tooLarge(num) {
return 0, false, end // Overflow; crazy long number most likely.
}
num = num*10 + int(s[newi]-'0')
isnum = true
}
return
}
func (p *pp) badVerb(verb rune) {
p.erroring = true
_, _ = p.WriteString(percentBangString)
_, _ = p.WriteRune(verb)
_, _ = p.WriteSingleByte('(')
switch {
case p.arg != nil:
_, _ = p.WriteString(p.arg.String())
_, _ = p.WriteSingleByte('=')
p.printArg(p.arg, 'v')
default:
_, _ = p.WriteString(UndefinedValue.String())
}
_, _ = p.WriteSingleByte(')')
p.erroring = false
}
func (p *pp) fmtBool(v bool, verb rune) {
switch verb {
case 't', 'v':
p.fmt.fmtBoolean(v)
default:
p.badVerb(verb)
}
}
// fmt0x64 formats a uint64 in hexadecimal and prefixes it with 0x or
// not, as requested, by temporarily setting the sharp flag.
func (p *pp) fmt0x64(v uint64, leading0x bool) {
sharp := p.fmt.sharp
p.fmt.sharp = leading0x
p.fmt.fmtInteger(v, 16, unsigned, 'v', ldigits)
p.fmt.sharp = sharp
}
// fmtInteger formats a signed or unsigned integer.
func (p *pp) fmtInteger(v uint64, isSigned bool, verb rune) {
switch verb {
case 'v':
if p.fmt.sharpV && !isSigned {
p.fmt0x64(v, true)
} else {
p.fmt.fmtInteger(v, 10, isSigned, verb, ldigits)
}
case 'd':
p.fmt.fmtInteger(v, 10, isSigned, verb, ldigits)
case 'b':
p.fmt.fmtInteger(v, 2, isSigned, verb, ldigits)
case 'o', 'O':
p.fmt.fmtInteger(v, 8, isSigned, verb, ldigits)
case 'x':
p.fmt.fmtInteger(v, 16, isSigned, verb, ldigits)
case 'X':
p.fmt.fmtInteger(v, 16, isSigned, verb, udigits)
case 'c':
p.fmt.fmtC(v)
case 'q':
if v <= utf8.MaxRune {
p.fmt.fmtQc(v)
} else {
p.badVerb(verb)
}
case 'U':
p.fmt.fmtUnicode(v)
default:
p.badVerb(verb)
}
}
// fmtFloat formats a float. The default precision for each verb
// is specified as last argument in the call to fmt_float.
func (p *pp) fmtFloat(v float64, size int, verb rune) {
switch verb {
case 'v':
p.fmt.fmtFloat(v, size, 'g', -1)
case 'b', 'g', 'G', 'x', 'X':
p.fmt.fmtFloat(v, size, verb, -1)
case 'f', 'e', 'E':
p.fmt.fmtFloat(v, size, verb, 6)
case 'F':
p.fmt.fmtFloat(v, size, 'f', 6)
default:
p.badVerb(verb)
}
}
func (p *pp) fmtString(v string, verb rune) {
switch verb {
case 'v':
if p.fmt.sharpV {
p.fmt.fmtQ(v)
} else {
p.fmt.fmtS(v)
}
case 's':
p.fmt.fmtS(v)
case 'x':
p.fmt.fmtSx(v, ldigits)
case 'X':
p.fmt.fmtSx(v, udigits)
case 'q':
p.fmt.fmtQ(v)
default:
p.badVerb(verb)
}
}
func (p *pp) fmtBytes(v []byte, verb rune, typeString string) {
switch verb {
case 'v', 'd':
if p.fmt.sharpV {
_, _ = p.WriteString(typeString)
if v == nil {
_, _ = p.WriteString(nilParenString)
return
}
_, _ = p.WriteSingleByte('{')
for i, c := range v {
if i > 0 {
_, _ = p.WriteString(commaSpaceString)
}
p.fmt0x64(uint64(c), true)
}
_, _ = p.WriteSingleByte('}')
} else {
_, _ = p.WriteSingleByte('[')
for i, c := range v {
if i > 0 {
_, _ = p.WriteSingleByte(' ')
}
p.fmt.fmtInteger(uint64(c), 10, unsigned, verb, ldigits)
}
_, _ = p.WriteSingleByte(']')
}
case 's':
p.fmt.fmtBs(v)
case 'x':
p.fmt.fmtBx(v, ldigits)
case 'X':
p.fmt.fmtBx(v, udigits)
case 'q':
p.fmt.fmtQ(string(v))
}
}
func (p *pp) printArg(arg Object, verb rune) {
p.arg = arg
if arg == nil {
arg = UndefinedValue
}
// Special processing considerations.
// %T (the value's type) and %p (its address) are special; we always do them first.
switch verb {
case 'T':
p.fmt.fmtS(arg.TypeName())
return
case 'v':
p.fmt.fmtS(arg.String())
return
}
// Some types can be done without reflection.
switch f := arg.(type) {
case *Bool:
p.fmtBool(!f.IsFalsy(), verb)
case *Float:
p.fmtFloat(f.Value, 64, verb)
case *Int:
p.fmtInteger(uint64(f.Value), signed, verb)
case *String:
p.fmtString(f.Value, verb)
case *Bytes:
p.fmtBytes(f.Value, verb, "[]byte")
default:
p.fmtString(f.String(), verb)
}
}
// intFromArg gets the argNumth element of a. On return, isInt reports whether the argument has integer type.
func intFromArg(a []Object, argNum int) (num int, isInt bool, newArgNum int) {
newArgNum = argNum
if argNum < len(a) {
var num64 int64
num64, isInt = ToInt64(a[argNum])
num = int(num64)
newArgNum = argNum + 1
if tooLarge(num) {
num = 0
isInt = false
}
}
return
}
// parseArgNumber returns the value of the bracketed number, minus 1
// (explicit argument numbers are one-indexed but we want zero-indexed).
// The opening bracket is known to be present at format[0].
// The returned values are the index, the number of bytes to consume
// up to the closing paren, if present, and whether the number parsed
// ok. The bytes to consume will be 1 if no closing paren is present.
func parseArgNumber(format string) (index int, wid int, ok bool) {
// There must be at least 3 bytes: [n].
if len(format) < 3 {
return 0, 1, false
}
// Find closing bracket.
for i := 1; i < len(format); i++ {
if format[i] == ']' {
width, ok, newi := parsenum(format, 1, i)
if !ok || newi != i {
return 0, i + 1, false
}
return width - 1, i + 1, true // arg numbers are one-indexed and skip paren.
}
}
return 0, 1, false
}
// argNumber returns the next argument to evaluate, which is either the value of the passed-in
// argNum or the value of the bracketed integer that begins format[i:]. It also returns
// the new value of i, that is, the index of the next byte of the format to process.
func (p *pp) argNumber(argNum int, format string, i int, numArgs int) (newArgNum, newi int, found bool) {
if len(format) <= i || format[i] != '[' {
return argNum, i, false
}
p.reordered = true
index, wid, ok := parseArgNumber(format[i:])
if ok && 0 <= index && index < numArgs {
return index, i + wid, true
}
p.goodArgNum = false
return argNum, i + wid, ok
}
func (p *pp) badArgNum(verb rune) {
_, _ = p.WriteString(percentBangString)
_, _ = p.WriteRune(verb)
_, _ = p.WriteString(badIndexString)
}
func (p *pp) missingArg(verb rune) {
_, _ = p.WriteString(percentBangString)
_, _ = p.WriteRune(verb)
_, _ = p.WriteString(missingString)
}
func (p *pp) doFormat(format string, a []Object) (err error) {
defer func() {
if r := recover(); r != nil {
if e, ok := r.(error); ok && e == ErrStringLimit {
err = e
return
}
panic(r)
}
}()
end := len(format)
argNum := 0 // we process one argument per non-trivial format
afterIndex := false // previous item in format was an index like [3].
p.reordered = false
formatLoop:
for i := 0; i < end; {
p.goodArgNum = true
lasti := i
for i < end && format[i] != '%' {
i++
}
if i > lasti {
_, _ = p.WriteString(format[lasti:i])
}
if i >= end {
// done processing format string
break
}
// Process one verb
i++
// Do we have flags?
p.fmt.clearflags()
simpleFormat:
for ; i < end; i++ {
c := format[i]
switch c {
case '#':
p.fmt.sharp = true
case '0':
p.fmt.zero = !p.fmt.minus // Only allow zero padding to the left.
case '+':
p.fmt.plus = true
case '-':
p.fmt.minus = true
p.fmt.zero = false // Do not pad with zeros to the right.
case ' ':
p.fmt.space = true
default:
// Fast path for common case of ascii lower case simple verbs
// without precision or width or argument indices.
if 'a' <= c && c <= 'z' && argNum < len(a) {
if c == 'v' {
// Go syntax
p.fmt.sharpV = p.fmt.sharp
p.fmt.sharp = false
// Struct-field syntax
p.fmt.plusV = p.fmt.plus
p.fmt.plus = false
}
p.printArg(a[argNum], rune(c))
argNum++
i++
continue formatLoop
}
// Format is more complex than simple flags and a verb or is malformed.
break simpleFormat
}
}
// Do we have an explicit argument index?
argNum, i, afterIndex = p.argNumber(argNum, format, i, len(a))
// Do we have width?
if i < end && format[i] == '*' {
i++
p.fmt.wid, p.fmt.widPresent, argNum = intFromArg(a, argNum)
if !p.fmt.widPresent {
_, _ = p.WriteString(badWidthString)
}
// We have a negative width, so take its value and ensure
// that the minus flag is set
if p.fmt.wid < 0 {
p.fmt.wid = -p.fmt.wid
p.fmt.minus = true
p.fmt.zero = false // Do not pad with zeros to the right.
}
afterIndex = false
} else {
p.fmt.wid, p.fmt.widPresent, i = parsenum(format, i, end)
if afterIndex && p.fmt.widPresent { // "%[3]2d"
p.goodArgNum = false
}
}
// Do we have precision?
if i+1 < end && format[i] == '.' {
i++
if afterIndex { // "%[3].2d"
p.goodArgNum = false
}
argNum, i, afterIndex = p.argNumber(argNum, format, i, len(a))
if i < end && format[i] == '*' {
i++
p.fmt.prec, p.fmt.precPresent, argNum = intFromArg(a, argNum)
// Negative precision arguments don't make sense
if p.fmt.prec < 0 {
p.fmt.prec = 0
p.fmt.precPresent = false
}
if !p.fmt.precPresent {
_, _ = p.WriteString(badPrecString)
}
afterIndex = false
} else {
p.fmt.prec, p.fmt.precPresent, i = parsenum(format, i, end)
if !p.fmt.precPresent {
p.fmt.prec = 0
p.fmt.precPresent = true
}
}
}
if !afterIndex {
argNum, i, afterIndex = p.argNumber(argNum, format, i, len(a))
}
if i >= end {
_, _ = p.WriteString(noVerbString)
break
}
verb, size := rune(format[i]), 1
if verb >= utf8.RuneSelf {
verb, size = utf8.DecodeRuneInString(format[i:])
}
i += size
switch {
case verb == '%': // Percent does not absorb operands and ignores f.wid and f.prec.
_, _ = p.WriteSingleByte('%')
case !p.goodArgNum:
p.badArgNum(verb)
case argNum >= len(a): // No argument left over to print for the current verb.
p.missingArg(verb)
case verb == 'v':
// Go syntax
p.fmt.sharpV = p.fmt.sharp
p.fmt.sharp = false
// Struct-field syntax
p.fmt.plusV = p.fmt.plus
p.fmt.plus = false
fallthrough
default:
p.printArg(a[argNum], verb)
argNum++
}
}
// Check for extra arguments unless the call accessed the arguments
// out of order, in which case it's too expensive to detect if they've all
// been used and arguably OK if they're not.
if !p.reordered && argNum < len(a) {
p.fmt.clearflags()
_, _ = p.WriteString(extraString)
for i, arg := range a[argNum:] {
if i > 0 {
_, _ = p.WriteString(commaSpaceString)
}
if arg == nil {
_, _ = p.WriteString(UndefinedValue.String())
} else {
_, _ = p.WriteString(arg.TypeName())
_, _ = p.WriteSingleByte('=')
p.printArg(arg, 'v')
}
}
_, _ = p.WriteSingleByte(')')
}
return nil
}
// Format formats according to a format specifier and returns the resulting string.
func Format(format string, a ...Object) (string, error) {
p := newPrinter()
err := p.doFormat(format, a)
s := string(p.buf)
p.free()
return s, err
}