| 1 |
747 |
jeremybenn |
// Copyright 2010 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 html
|
| 6 |
|
|
|
| 7 |
|
|
import (
|
| 8 |
|
|
"bytes"
|
| 9 |
|
|
"io"
|
| 10 |
|
|
"strconv"
|
| 11 |
|
|
"strings"
|
| 12 |
|
|
)
|
| 13 |
|
|
|
| 14 |
|
|
// A TokenType is the type of a Token.
|
| 15 |
|
|
type TokenType int
|
| 16 |
|
|
|
| 17 |
|
|
const (
|
| 18 |
|
|
// ErrorToken means that an error occurred during tokenization.
|
| 19 |
|
|
ErrorToken TokenType = iota
|
| 20 |
|
|
// TextToken means a text node.
|
| 21 |
|
|
TextToken
|
| 22 |
|
|
// A StartTagToken looks like .
|
| 23 |
|
|
StartTagToken
|
| 24 |
|
|
// An EndTagToken looks like .
|
| 25 |
|
|
EndTagToken
|
| 26 |
|
|
// A SelfClosingTagToken tag looks like
.
|
| 27 |
|
|
SelfClosingTagToken
|
| 28 |
|
|
// A CommentToken looks like .
|
| 29 |
|
|
CommentToken
|
| 30 |
|
|
// A DoctypeToken looks like
|
| 31 |
|
|
DoctypeToken
|
| 32 |
|
|
)
|
| 33 |
|
|
|
| 34 |
|
|
// String returns a string representation of the TokenType.
|
| 35 |
|
|
func (t TokenType) String() string {
|
| 36 |
|
|
switch t {
|
| 37 |
|
|
case ErrorToken:
|
| 38 |
|
|
return "Error"
|
| 39 |
|
|
case TextToken:
|
| 40 |
|
|
return "Text"
|
| 41 |
|
|
case StartTagToken:
|
| 42 |
|
|
return "StartTag"
|
| 43 |
|
|
case EndTagToken:
|
| 44 |
|
|
return "EndTag"
|
| 45 |
|
|
case SelfClosingTagToken:
|
| 46 |
|
|
return "SelfClosingTag"
|
| 47 |
|
|
case CommentToken:
|
| 48 |
|
|
return "Comment"
|
| 49 |
|
|
case DoctypeToken:
|
| 50 |
|
|
return "Doctype"
|
| 51 |
|
|
}
|
| 52 |
|
|
return "Invalid(" + strconv.Itoa(int(t)) + ")"
|
| 53 |
|
|
}
|
| 54 |
|
|
|
| 55 |
|
|
// An Attribute is an attribute namespace-key-value triple. Namespace is
|
| 56 |
|
|
// non-empty for foreign attributes like xlink, Key is alphabetic (and hence
|
| 57 |
|
|
// does not contain escapable characters like '&', '<' or '>'), and Val is
|
| 58 |
|
|
// unescaped (it looks like "a
|
| 59 |
|
|
//
|
| 60 |
|
|
// Namespace is only used by the parser, not the tokenizer.
|
| 61 |
|
|
type Attribute struct {
|
| 62 |
|
|
Namespace, Key, Val string
|
| 63 |
|
|
}
|
| 64 |
|
|
|
| 65 |
|
|
// A Token consists of a TokenType and some Data (tag name for start and end
|
| 66 |
|
|
// tags, content for text, comments and doctypes). A tag Token may also contain
|
| 67 |
|
|
// a slice of Attributes. Data is unescaped for all Tokens (it looks like "a
|
| 68 |
|
|
// rather than "a<b").
|
| 69 |
|
|
type Token struct {
|
| 70 |
|
|
Type TokenType
|
| 71 |
|
|
Data string
|
| 72 |
|
|
Attr []Attribute
|
| 73 |
|
|
}
|
| 74 |
|
|
|
| 75 |
|
|
// tagString returns a string representation of a tag Token's Data and Attr.
|
| 76 |
|
|
func (t Token) tagString() string {
|
| 77 |
|
|
if len(t.Attr) == 0 {
|
| 78 |
|
|
return t.Data
|
| 79 |
|
|
}
|
| 80 |
|
|
buf := bytes.NewBufferString(t.Data)
|
| 81 |
|
|
for _, a := range t.Attr {
|
| 82 |
|
|
buf.WriteByte(' ')
|
| 83 |
|
|
buf.WriteString(a.Key)
|
| 84 |
|
|
buf.WriteString(`="`)
|
| 85 |
|
|
escape(buf, a.Val)
|
| 86 |
|
|
buf.WriteByte('"')
|
| 87 |
|
|
}
|
| 88 |
|
|
return buf.String()
|
| 89 |
|
|
}
|
| 90 |
|
|
|
| 91 |
|
|
// String returns a string representation of the Token.
|
| 92 |
|
|
func (t Token) String() string {
|
| 93 |
|
|
switch t.Type {
|
| 94 |
|
|
case ErrorToken:
|
| 95 |
|
|
return ""
|
| 96 |
|
|
case TextToken:
|
| 97 |
|
|
return EscapeString(t.Data)
|
| 98 |
|
|
case StartTagToken:
|
| 99 |
|
|
return "<" + t.tagString() + ">"
|
| 100 |
|
|
case EndTagToken:
|
| 101 |
|
|
return ""
|
| 102 |
|
|
case SelfClosingTagToken:
|
| 103 |
|
|
return "<" + t.tagString() + "/>"
|
| 104 |
|
|
case CommentToken:
|
| 105 |
|
|
return ""
|
| 106 |
|
|
case DoctypeToken:
|
| 107 |
|
|
return ""
|
| 108 |
|
|
}
|
| 109 |
|
|
return "Invalid(" + strconv.Itoa(int(t.Type)) + ")"
|
| 110 |
|
|
}
|
| 111 |
|
|
|
| 112 |
|
|
// span is a range of bytes in a Tokenizer's buffer. The start is inclusive,
|
| 113 |
|
|
// the end is exclusive.
|
| 114 |
|
|
type span struct {
|
| 115 |
|
|
start, end int
|
| 116 |
|
|
}
|
| 117 |
|
|
|
| 118 |
|
|
// A Tokenizer returns a stream of HTML Tokens.
|
| 119 |
|
|
type Tokenizer struct {
|
| 120 |
|
|
// r is the source of the HTML text.
|
| 121 |
|
|
r io.Reader
|
| 122 |
|
|
// tt is the TokenType of the current token.
|
| 123 |
|
|
tt TokenType
|
| 124 |
|
|
// err is the first error encountered during tokenization. It is possible
|
| 125 |
|
|
// for tt != Error && err != nil to hold: this means that Next returned a
|
| 126 |
|
|
// valid token but the subsequent Next call will return an error token.
|
| 127 |
|
|
// For example, if the HTML text input was just "plain", then the first
|
| 128 |
|
|
// Next call would set z.err to io.EOF but return a TextToken, and all
|
| 129 |
|
|
// subsequent Next calls would return an ErrorToken.
|
| 130 |
|
|
// err is never reset. Once it becomes non-nil, it stays non-nil.
|
| 131 |
|
|
err error
|
| 132 |
|
|
// buf[raw.start:raw.end] holds the raw bytes of the current token.
|
| 133 |
|
|
// buf[raw.end:] is buffered input that will yield future tokens.
|
| 134 |
|
|
raw span
|
| 135 |
|
|
buf []byte
|
| 136 |
|
|
// buf[data.start:data.end] holds the raw bytes of the current token's data:
|
| 137 |
|
|
// a text token's text, a tag token's tag name, etc.
|
| 138 |
|
|
data span
|
| 139 |
|
|
// pendingAttr is the attribute key and value currently being tokenized.
|
| 140 |
|
|
// When complete, pendingAttr is pushed onto attr. nAttrReturned is
|
| 141 |
|
|
// incremented on each call to TagAttr.
|
| 142 |
|
|
pendingAttr [2]span
|
| 143 |
|
|
attr [][2]span
|
| 144 |
|
|
nAttrReturned int
|
| 145 |
|
|
// rawTag is the "script" in "" that closes the next token. If
|
| 146 |
|
|
// non-empty, the subsequent call to Next will return a raw or RCDATA text
|
| 147 |
|
|
// token: one that treats "" as text instead of an element.
|
| 148 |
|
|
// rawTag's contents are lower-cased.
|
| 149 |
|
|
rawTag string
|
| 150 |
|
|
// textIsRaw is whether the current text token's data is not escaped.
|
| 151 |
|
|
textIsRaw bool
|
| 152 |
|
|
}
|
| 153 |
|
|
|
| 154 |
|
|
// Err returns the error associated with the most recent ErrorToken token.
|
| 155 |
|
|
// This is typically io.EOF, meaning the end of tokenization.
|
| 156 |
|
|
func (z *Tokenizer) Err() error {
|
| 157 |
|
|
if z.tt != ErrorToken {
|
| 158 |
|
|
return nil
|
| 159 |
|
|
}
|
| 160 |
|
|
return z.err
|
| 161 |
|
|
}
|
| 162 |
|
|
|
| 163 |
|
|
// readByte returns the next byte from the input stream, doing a buffered read
|
| 164 |
|
|
// from z.r into z.buf if necessary. z.buf[z.raw.start:z.raw.end] remains a contiguous byte
|
| 165 |
|
|
// slice that holds all the bytes read so far for the current token.
|
| 166 |
|
|
// It sets z.err if the underlying reader returns an error.
|
| 167 |
|
|
// Pre-condition: z.err == nil.
|
| 168 |
|
|
func (z *Tokenizer) readByte() byte {
|
| 169 |
|
|
if z.raw.end >= len(z.buf) {
|
| 170 |
|
|
// Our buffer is exhausted and we have to read from z.r.
|
| 171 |
|
|
// We copy z.buf[z.raw.start:z.raw.end] to the beginning of z.buf. If the length
|
| 172 |
|
|
// z.raw.end - z.raw.start is more than half the capacity of z.buf, then we
|
| 173 |
|
|
// allocate a new buffer before the copy.
|
| 174 |
|
|
c := cap(z.buf)
|
| 175 |
|
|
d := z.raw.end - z.raw.start
|
| 176 |
|
|
var buf1 []byte
|
| 177 |
|
|
if 2*d > c {
|
| 178 |
|
|
buf1 = make([]byte, d, 2*c)
|
| 179 |
|
|
} else {
|
| 180 |
|
|
buf1 = z.buf[:d]
|
| 181 |
|
|
}
|
| 182 |
|
|
copy(buf1, z.buf[z.raw.start:z.raw.end])
|
| 183 |
|
|
if x := z.raw.start; x != 0 {
|
| 184 |
|
|
// Adjust the data/attr spans to refer to the same contents after the copy.
|
| 185 |
|
|
z.data.start -= x
|
| 186 |
|
|
z.data.end -= x
|
| 187 |
|
|
z.pendingAttr[0].start -= x
|
| 188 |
|
|
z.pendingAttr[0].end -= x
|
| 189 |
|
|
z.pendingAttr[1].start -= x
|
| 190 |
|
|
z.pendingAttr[1].end -= x
|
| 191 |
|
|
for i := range z.attr {
|
| 192 |
|
|
z.attr[i][0].start -= x
|
| 193 |
|
|
z.attr[i][0].end -= x
|
| 194 |
|
|
z.attr[i][1].start -= x
|
| 195 |
|
|
z.attr[i][1].end -= x
|
| 196 |
|
|
}
|
| 197 |
|
|
}
|
| 198 |
|
|
z.raw.start, z.raw.end, z.buf = 0, d, buf1[:d]
|
| 199 |
|
|
// Now that we have copied the live bytes to the start of the buffer,
|
| 200 |
|
|
// we read from z.r into the remainder.
|
| 201 |
|
|
n, err := z.r.Read(buf1[d:cap(buf1)])
|
| 202 |
|
|
if err != nil {
|
| 203 |
|
|
z.err = err
|
| 204 |
|
|
return 0
|
| 205 |
|
|
}
|
| 206 |
|
|
z.buf = buf1[:d+n]
|
| 207 |
|
|
}
|
| 208 |
|
|
x := z.buf[z.raw.end]
|
| 209 |
|
|
z.raw.end++
|
| 210 |
|
|
return x
|
| 211 |
|
|
}
|
| 212 |
|
|
|
| 213 |
|
|
// skipWhiteSpace skips past any white space.
|
| 214 |
|
|
func (z *Tokenizer) skipWhiteSpace() {
|
| 215 |
|
|
if z.err != nil {
|
| 216 |
|
|
return
|
| 217 |
|
|
}
|
| 218 |
|
|
for {
|
| 219 |
|
|
c := z.readByte()
|
| 220 |
|
|
if z.err != nil {
|
| 221 |
|
|
return
|
| 222 |
|
|
}
|
| 223 |
|
|
switch c {
|
| 224 |
|
|
case ' ', '\n', '\r', '\t', '\f':
|
| 225 |
|
|
// No-op.
|
| 226 |
|
|
default:
|
| 227 |
|
|
z.raw.end--
|
| 228 |
|
|
return
|
| 229 |
|
|
}
|
| 230 |
|
|
}
|
| 231 |
|
|
}
|
| 232 |
|
|
|
| 233 |
|
|
// readRawOrRCDATA reads until the next "", where "foo" is z.rawTag and
|
| 234 |
|
|
// is typically something like "script" or "textarea".
|
| 235 |
|
|
func (z *Tokenizer) readRawOrRCDATA() {
|
| 236 |
|
|
loop:
|
| 237 |
|
|
for {
|
| 238 |
|
|
c := z.readByte()
|
| 239 |
|
|
if z.err != nil {
|
| 240 |
|
|
break loop
|
| 241 |
|
|
}
|
| 242 |
|
|
if c != '<' {
|
| 243 |
|
|
continue loop
|
| 244 |
|
|
}
|
| 245 |
|
|
c = z.readByte()
|
| 246 |
|
|
if z.err != nil {
|
| 247 |
|
|
break loop
|
| 248 |
|
|
}
|
| 249 |
|
|
if c != '/' {
|
| 250 |
|
|
continue loop
|
| 251 |
|
|
}
|
| 252 |
|
|
for i := 0; i < len(z.rawTag); i++ {
|
| 253 |
|
|
c = z.readByte()
|
| 254 |
|
|
if z.err != nil {
|
| 255 |
|
|
break loop
|
| 256 |
|
|
}
|
| 257 |
|
|
if c != z.rawTag[i] && c != z.rawTag[i]-('a'-'A') {
|
| 258 |
|
|
continue loop
|
| 259 |
|
|
}
|
| 260 |
|
|
}
|
| 261 |
|
|
c = z.readByte()
|
| 262 |
|
|
if z.err != nil {
|
| 263 |
|
|
break loop
|
| 264 |
|
|
}
|
| 265 |
|
|
switch c {
|
| 266 |
|
|
case ' ', '\n', '\r', '\t', '\f', '/', '>':
|
| 267 |
|
|
// The 3 is 2 for the leading "
|
| 268 |
|
|
z.raw.end -= 3 + len(z.rawTag)
|
| 269 |
|
|
break loop
|
| 270 |
|
|
case '<':
|
| 271 |
|
|
// Step back one, to catch "".
|
| 272 |
|
|
z.raw.end--
|
| 273 |
|
|
}
|
| 274 |
|
|
}
|
| 275 |
|
|
z.data.end = z.raw.end
|
| 276 |
|
|
// A textarea's or title's RCDATA can contain escaped entities.
|
| 277 |
|
|
z.textIsRaw = z.rawTag != "textarea" && z.rawTag != "title"
|
| 278 |
|
|
z.rawTag = ""
|
| 279 |
|
|
}
|
| 280 |
|
|
|
| 281 |
|
|
// readComment reads the next comment token starting with ".
|
| 288 |
|
|
z.data.end = z.data.start
|
| 289 |
|
|
}
|
| 290 |
|
|
}()
|
| 291 |
|
|
for dashCount := 2; ; {
|
| 292 |
|
|
c := z.readByte()
|
| 293 |
|
|
if z.err != nil {
|
| 294 |
|
|
// Ignore up to two dashes at EOF.
|
| 295 |
|
|
if dashCount > 2 {
|
| 296 |
|
|
dashCount = 2
|
| 297 |
|
|
}
|
| 298 |
|
|
z.data.end = z.raw.end - dashCount
|
| 299 |
|
|
return
|
| 300 |
|
|
}
|
| 301 |
|
|
switch c {
|
| 302 |
|
|
case '-':
|
| 303 |
|
|
dashCount++
|
| 304 |
|
|
continue
|
| 305 |
|
|
case '>':
|
| 306 |
|
|
if dashCount >= 2 {
|
| 307 |
|
|
z.data.end = z.raw.end - len("-->")
|
| 308 |
|
|
return
|
| 309 |
|
|
}
|
| 310 |
|
|
case '!':
|
| 311 |
|
|
if dashCount >= 2 {
|
| 312 |
|
|
c = z.readByte()
|
| 313 |
|
|
if z.err != nil {
|
| 314 |
|
|
z.data.end = z.raw.end
|
| 315 |
|
|
return
|
| 316 |
|
|
}
|
| 317 |
|
|
if c == '>' {
|
| 318 |
|
|
z.data.end = z.raw.end - len("--!>")
|
| 319 |
|
|
return
|
| 320 |
|
|
}
|
| 321 |
|
|
}
|
| 322 |
|
|
}
|
| 323 |
|
|
dashCount = 0
|
| 324 |
|
|
}
|
| 325 |
|
|
}
|
| 326 |
|
|
|
| 327 |
|
|
// readUntilCloseAngle reads until the next ">".
|
| 328 |
|
|
func (z *Tokenizer) readUntilCloseAngle() {
|
| 329 |
|
|
z.data.start = z.raw.end
|
| 330 |
|
|
for {
|
| 331 |
|
|
c := z.readByte()
|
| 332 |
|
|
if z.err != nil {
|
| 333 |
|
|
z.data.end = z.raw.end
|
| 334 |
|
|
return
|
| 335 |
|
|
}
|
| 336 |
|
|
if c == '>' {
|
| 337 |
|
|
z.data.end = z.raw.end - len(">")
|
| 338 |
|
|
return
|
| 339 |
|
|
}
|
| 340 |
|
|
}
|
| 341 |
|
|
}
|
| 342 |
|
|
|
| 343 |
|
|
// readMarkupDeclaration reads the next token starting with "
|
| 344 |
|
|
// a "", a "", or "
|
| 345 |
|
|
// "
|
| 346 |
|
|
func (z *Tokenizer) readMarkupDeclaration() TokenType {
|
| 347 |
|
|
z.data.start = z.raw.end
|
| 348 |
|
|
var c [2]byte
|
| 349 |
|
|
for i := 0; i < 2; i++ {
|
| 350 |
|
|
c[i] = z.readByte()
|
| 351 |
|
|
if z.err != nil {
|
| 352 |
|
|
z.data.end = z.raw.end
|
| 353 |
|
|
return CommentToken
|
| 354 |
|
|
}
|
| 355 |
|
|
}
|
| 356 |
|
|
if c[0] == '-' && c[1] == '-' {
|
| 357 |
|
|
z.readComment()
|
| 358 |
|
|
return CommentToken
|
| 359 |
|
|
}
|
| 360 |
|
|
z.raw.end -= 2
|
| 361 |
|
|
const s = "DOCTYPE"
|
| 362 |
|
|
for i := 0; i < len(s); i++ {
|
| 363 |
|
|
c := z.readByte()
|
| 364 |
|
|
if z.err != nil {
|
| 365 |
|
|
z.data.end = z.raw.end
|
| 366 |
|
|
return CommentToken
|
| 367 |
|
|
}
|
| 368 |
|
|
if c != s[i] && c != s[i]+('a'-'A') {
|
| 369 |
|
|
// Back up to read the fragment of "DOCTYPE" again.
|
| 370 |
|
|
z.raw.end = z.data.start
|
| 371 |
|
|
z.readUntilCloseAngle()
|
| 372 |
|
|
return CommentToken
|
| 373 |
|
|
}
|
| 374 |
|
|
}
|
| 375 |
|
|
if z.skipWhiteSpace(); z.err != nil {
|
| 376 |
|
|
z.data.start = z.raw.end
|
| 377 |
|
|
z.data.end = z.raw.end
|
| 378 |
|
|
return DoctypeToken
|
| 379 |
|
|
}
|
| 380 |
|
|
z.readUntilCloseAngle()
|
| 381 |
|
|
return DoctypeToken
|
| 382 |
|
|
}
|
| 383 |
|
|
|
| 384 |
|
|
// startTagIn returns whether the start tag in z.buf[z.data.start:z.data.end]
|
| 385 |
|
|
// case-insensitively matches any element of ss.
|
| 386 |
|
|
func (z *Tokenizer) startTagIn(ss ...string) bool {
|
| 387 |
|
|
loop:
|
| 388 |
|
|
for _, s := range ss {
|
| 389 |
|
|
if z.data.end-z.data.start != len(s) {
|
| 390 |
|
|
continue loop
|
| 391 |
|
|
}
|
| 392 |
|
|
for i := 0; i < len(s); i++ {
|
| 393 |
|
|
c := z.buf[z.data.start+i]
|
| 394 |
|
|
if 'A' <= c && c <= 'Z' {
|
| 395 |
|
|
c += 'a' - 'A'
|
| 396 |
|
|
}
|
| 397 |
|
|
if c != s[i] {
|
| 398 |
|
|
continue loop
|
| 399 |
|
|
}
|
| 400 |
|
|
}
|
| 401 |
|
|
return true
|
| 402 |
|
|
}
|
| 403 |
|
|
return false
|
| 404 |
|
|
}
|
| 405 |
|
|
|
| 406 |
|
|
// readStartTag reads the next start tag token. The opening "
|
| 407 |
|
|
// been consumed, where 'a' means anything in [A-Za-z].
|
| 408 |
|
|
func (z *Tokenizer) readStartTag() TokenType {
|
| 409 |
|
|
z.attr = z.attr[:0]
|
| 410 |
|
|
z.nAttrReturned = 0
|
| 411 |
|
|
// Read the tag name and attribute key/value pairs.
|
| 412 |
|
|
z.readTagName()
|
| 413 |
|
|
if z.skipWhiteSpace(); z.err != nil {
|
| 414 |
|
|
return ErrorToken
|
| 415 |
|
|
}
|
| 416 |
|
|
for {
|
| 417 |
|
|
c := z.readByte()
|
| 418 |
|
|
if z.err != nil || c == '>' {
|
| 419 |
|
|
break
|
| 420 |
|
|
}
|
| 421 |
|
|
z.raw.end--
|
| 422 |
|
|
z.readTagAttrKey()
|
| 423 |
|
|
z.readTagAttrVal()
|
| 424 |
|
|
// Save pendingAttr if it has a non-empty key.
|
| 425 |
|
|
if z.pendingAttr[0].start != z.pendingAttr[0].end {
|
| 426 |
|
|
z.attr = append(z.attr, z.pendingAttr)
|
| 427 |
|
|
}
|
| 428 |
|
|
if z.skipWhiteSpace(); z.err != nil {
|
| 429 |
|
|
break
|
| 430 |
|
|
}
|
| 431 |
|
|
}
|
| 432 |
|
|
// Several tags flag the tokenizer's next token as raw.
|
| 433 |
|
|
c, raw := z.buf[z.data.start], false
|
| 434 |
|
|
if 'A' <= c && c <= 'Z' {
|
| 435 |
|
|
c += 'a' - 'A'
|
| 436 |
|
|
}
|
| 437 |
|
|
switch c {
|
| 438 |
|
|
case 'i':
|
| 439 |
|
|
raw = z.startTagIn("iframe")
|
| 440 |
|
|
case 'n':
|
| 441 |
|
|
raw = z.startTagIn("noembed", "noframes", "noscript")
|
| 442 |
|
|
case 'p':
|
| 443 |
|
|
raw = z.startTagIn("plaintext")
|
| 444 |
|
|
case 's':
|
| 445 |
|
|
raw = z.startTagIn("script", "style")
|
| 446 |
|
|
case 't':
|
| 447 |
|
|
raw = z.startTagIn("textarea", "title")
|
| 448 |
|
|
case 'x':
|
| 449 |
|
|
raw = z.startTagIn("xmp")
|
| 450 |
|
|
}
|
| 451 |
|
|
if raw {
|
| 452 |
|
|
z.rawTag = strings.ToLower(string(z.buf[z.data.start:z.data.end]))
|
| 453 |
|
|
}
|
| 454 |
|
|
// Look for a self-closing token like "
".
|
| 455 |
|
|
if z.err == nil && z.buf[z.raw.end-2] == '/' {
|
| 456 |
|
|
return SelfClosingTagToken
|
| 457 |
|
|
}
|
| 458 |
|
|
return StartTagToken
|
| 459 |
|
|
}
|
| 460 |
|
|
|
| 461 |
|
|
// readEndTag reads the next end tag token. The opening "
|
| 462 |
|
|
// been consumed, where 'a' means anything in [A-Za-z].
|
| 463 |
|
|
func (z *Tokenizer) readEndTag() {
|
| 464 |
|
|
z.attr = z.attr[:0]
|
| 465 |
|
|
z.nAttrReturned = 0
|
| 466 |
|
|
z.readTagName()
|
| 467 |
|
|
for {
|
| 468 |
|
|
c := z.readByte()
|
| 469 |
|
|
if z.err != nil || c == '>' {
|
| 470 |
|
|
return
|
| 471 |
|
|
}
|
| 472 |
|
|
}
|
| 473 |
|
|
}
|
| 474 |
|
|
|
| 475 |
|
|
// readTagName sets z.data to the "div" in "". The reader (z.raw.end)
|
| 476 |
|
|
// is positioned such that the first byte of the tag name (the "d" in "
|
| 477 |
|
|
// has already been consumed.
|
| 478 |
|
|
func (z *Tokenizer) readTagName() {
|
| 479 |
|
|
z.data.start = z.raw.end - 1
|
| 480 |
|
|
for {
|
| 481 |
|
|
c := z.readByte()
|
| 482 |
|
|
if z.err != nil {
|
| 483 |
|
|
z.data.end = z.raw.end
|
| 484 |
|
|
return
|
| 485 |
|
|
}
|
| 486 |
|
|
switch c {
|
| 487 |
|
|
case ' ', '\n', '\r', '\t', '\f':
|
| 488 |
|
|
z.data.end = z.raw.end - 1
|
| 489 |
|
|
return
|
| 490 |
|
|
case '/', '>':
|
| 491 |
|
|
z.raw.end--
|
| 492 |
|
|
z.data.end = z.raw.end
|
| 493 |
|
|
return
|
| 494 |
|
|
}
|
| 495 |
|
|
}
|
| 496 |
|
|
}
|
| 497 |
|
|
|
| 498 |
|
|
// readTagAttrKey sets z.pendingAttr[0] to the "k" in "".
|
| 499 |
|
|
// Precondition: z.err == nil.
|
| 500 |
|
|
func (z *Tokenizer) readTagAttrKey() {
|
| 501 |
|
|
z.pendingAttr[0].start = z.raw.end
|
| 502 |
|
|
for {
|
| 503 |
|
|
c := z.readByte()
|
| 504 |
|
|
if z.err != nil {
|
| 505 |
|
|
z.pendingAttr[0].end = z.raw.end
|
| 506 |
|
|
return
|
| 507 |
|
|
}
|
| 508 |
|
|
switch c {
|
| 509 |
|
|
case ' ', '\n', '\r', '\t', '\f', '/':
|
| 510 |
|
|
z.pendingAttr[0].end = z.raw.end - 1
|
| 511 |
|
|
return
|
| 512 |
|
|
case '=', '>':
|
| 513 |
|
|
z.raw.end--
|
| 514 |
|
|
z.pendingAttr[0].end = z.raw.end
|
| 515 |
|
|
return
|
| 516 |
|
|
}
|
| 517 |
|
|
}
|
| 518 |
|
|
}
|
| 519 |
|
|
|
| 520 |
|
|
// readTagAttrVal sets z.pendingAttr[1] to the "v" in "".
|
| 521 |
|
|
func (z *Tokenizer) readTagAttrVal() {
|
| 522 |
|
|
z.pendingAttr[1].start = z.raw.end
|
| 523 |
|
|
z.pendingAttr[1].end = z.raw.end
|
| 524 |
|
|
if z.skipWhiteSpace(); z.err != nil {
|
| 525 |
|
|
return
|
| 526 |
|
|
}
|
| 527 |
|
|
c := z.readByte()
|
| 528 |
|
|
if z.err != nil {
|
| 529 |
|
|
return
|
| 530 |
|
|
}
|
| 531 |
|
|
if c != '=' {
|
| 532 |
|
|
z.raw.end--
|
| 533 |
|
|
return
|
| 534 |
|
|
}
|
| 535 |
|
|
if z.skipWhiteSpace(); z.err != nil {
|
| 536 |
|
|
return
|
| 537 |
|
|
}
|
| 538 |
|
|
quote := z.readByte()
|
| 539 |
|
|
if z.err != nil {
|
| 540 |
|
|
return
|
| 541 |
|
|
}
|
| 542 |
|
|
switch quote {
|
| 543 |
|
|
case '>':
|
| 544 |
|
|
z.raw.end--
|
| 545 |
|
|
return
|
| 546 |
|
|
|
| 547 |
|
|
case '\'', '"':
|
| 548 |
|
|
z.pendingAttr[1].start = z.raw.end
|
| 549 |
|
|
for {
|
| 550 |
|
|
c := z.readByte()
|
| 551 |
|
|
if z.err != nil {
|
| 552 |
|
|
z.pendingAttr[1].end = z.raw.end
|
| 553 |
|
|
return
|
| 554 |
|
|
}
|
| 555 |
|
|
if c == quote {
|
| 556 |
|
|
z.pendingAttr[1].end = z.raw.end - 1
|
| 557 |
|
|
return
|
| 558 |
|
|
}
|
| 559 |
|
|
}
|
| 560 |
|
|
|
| 561 |
|
|
default:
|
| 562 |
|
|
z.pendingAttr[1].start = z.raw.end - 1
|
| 563 |
|
|
for {
|
| 564 |
|
|
c := z.readByte()
|
| 565 |
|
|
if z.err != nil {
|
| 566 |
|
|
z.pendingAttr[1].end = z.raw.end
|
| 567 |
|
|
return
|
| 568 |
|
|
}
|
| 569 |
|
|
switch c {
|
| 570 |
|
|
case ' ', '\n', '\r', '\t', '\f':
|
| 571 |
|
|
z.pendingAttr[1].end = z.raw.end - 1
|
| 572 |
|
|
return
|
| 573 |
|
|
case '>':
|
| 574 |
|
|
z.raw.end--
|
| 575 |
|
|
z.pendingAttr[1].end = z.raw.end
|
| 576 |
|
|
return
|
| 577 |
|
|
}
|
| 578 |
|
|
}
|
| 579 |
|
|
}
|
| 580 |
|
|
}
|
| 581 |
|
|
|
| 582 |
|
|
// Next scans the next token and returns its type.
|
| 583 |
|
|
func (z *Tokenizer) Next() TokenType {
|
| 584 |
|
|
if z.err != nil {
|
| 585 |
|
|
z.tt = ErrorToken
|
| 586 |
|
|
return z.tt
|
| 587 |
|
|
}
|
| 588 |
|
|
z.raw.start = z.raw.end
|
| 589 |
|
|
z.data.start = z.raw.end
|
| 590 |
|
|
z.data.end = z.raw.end
|
| 591 |
|
|
if z.rawTag != "" {
|
| 592 |
|
|
if z.rawTag == "plaintext" {
|
| 593 |
|
|
// Read everything up to EOF.
|
| 594 |
|
|
for z.err == nil {
|
| 595 |
|
|
z.readByte()
|
| 596 |
|
|
}
|
| 597 |
|
|
z.textIsRaw = true
|
| 598 |
|
|
} else {
|
| 599 |
|
|
z.readRawOrRCDATA()
|
| 600 |
|
|
}
|
| 601 |
|
|
if z.data.end > z.data.start {
|
| 602 |
|
|
z.tt = TextToken
|
| 603 |
|
|
return z.tt
|
| 604 |
|
|
}
|
| 605 |
|
|
}
|
| 606 |
|
|
z.textIsRaw = false
|
| 607 |
|
|
|
| 608 |
|
|
loop:
|
| 609 |
|
|
for {
|
| 610 |
|
|
c := z.readByte()
|
| 611 |
|
|
if z.err != nil {
|
| 612 |
|
|
break loop
|
| 613 |
|
|
}
|
| 614 |
|
|
if c != '<' {
|
| 615 |
|
|
continue loop
|
| 616 |
|
|
}
|
| 617 |
|
|
|
| 618 |
|
|
// Check if the '<' we have just read is part of a tag, comment
|
| 619 |
|
|
// or doctype. If not, it's part of the accumulated text token.
|
| 620 |
|
|
c = z.readByte()
|
| 621 |
|
|
if z.err != nil {
|
| 622 |
|
|
break loop
|
| 623 |
|
|
}
|
| 624 |
|
|
var tokenType TokenType
|
| 625 |
|
|
switch {
|
| 626 |
|
|
case 'a' <= c && c <= 'z' || 'A' <= c && c <= 'Z':
|
| 627 |
|
|
tokenType = StartTagToken
|
| 628 |
|
|
case c == '/':
|
| 629 |
|
|
tokenType = EndTagToken
|
| 630 |
|
|
case c == '!' || c == '?':
|
| 631 |
|
|
// We use CommentToken to mean any of "",
|
| 632 |
|
|
// "" and "".
|
| 633 |
|
|
tokenType = CommentToken
|
| 634 |
|
|
default:
|
| 635 |
|
|
continue
|
| 636 |
|
|
}
|
| 637 |
|
|
|
| 638 |
|
|
// We have a non-text token, but we might have accumulated some text
|
| 639 |
|
|
// before that. If so, we return the text first, and return the non-
|
| 640 |
|
|
// text token on the subsequent call to Next.
|
| 641 |
|
|
if x := z.raw.end - len("
|
| 642 |
|
|
z.raw.end = x
|
| 643 |
|
|
z.data.end = x
|
| 644 |
|
|
z.tt = TextToken
|
| 645 |
|
|
return z.tt
|
| 646 |
|
|
}
|
| 647 |
|
|
switch tokenType {
|
| 648 |
|
|
case StartTagToken:
|
| 649 |
|
|
z.tt = z.readStartTag()
|
| 650 |
|
|
return z.tt
|
| 651 |
|
|
case EndTagToken:
|
| 652 |
|
|
c = z.readByte()
|
| 653 |
|
|
if z.err != nil {
|
| 654 |
|
|
break loop
|
| 655 |
|
|
}
|
| 656 |
|
|
if c == '>' {
|
| 657 |
|
|
// "" does not generate a token at all.
|
| 658 |
|
|
// Reset the tokenizer state and start again.
|
| 659 |
|
|
z.raw.start = z.raw.end
|
| 660 |
|
|
z.data.start = z.raw.end
|
| 661 |
|
|
z.data.end = z.raw.end
|
| 662 |
|
|
continue loop
|
| 663 |
|
|
}
|
| 664 |
|
|
if 'a' <= c && c <= 'z' || 'A' <= c && c <= 'Z' {
|
| 665 |
|
|
z.readEndTag()
|
| 666 |
|
|
z.tt = EndTagToken
|
| 667 |
|
|
return z.tt
|
| 668 |
|
|
}
|
| 669 |
|
|
z.raw.end--
|
| 670 |
|
|
z.readUntilCloseAngle()
|
| 671 |
|
|
z.tt = CommentToken
|
| 672 |
|
|
return z.tt
|
| 673 |
|
|
case CommentToken:
|
| 674 |
|
|
if c == '!' {
|
| 675 |
|
|
z.tt = z.readMarkupDeclaration()
|
| 676 |
|
|
return z.tt
|
| 677 |
|
|
}
|
| 678 |
|
|
z.raw.end--
|
| 679 |
|
|
z.readUntilCloseAngle()
|
| 680 |
|
|
z.tt = CommentToken
|
| 681 |
|
|
return z.tt
|
| 682 |
|
|
}
|
| 683 |
|
|
}
|
| 684 |
|
|
if z.raw.start < z.raw.end {
|
| 685 |
|
|
z.data.end = z.raw.end
|
| 686 |
|
|
z.tt = TextToken
|
| 687 |
|
|
return z.tt
|
| 688 |
|
|
}
|
| 689 |
|
|
z.tt = ErrorToken
|
| 690 |
|
|
return z.tt
|
| 691 |
|
|
}
|
| 692 |
|
|
|
| 693 |
|
|
// Raw returns the unmodified text of the current token. Calling Next, Token,
|
| 694 |
|
|
// Text, TagName or TagAttr may change the contents of the returned slice.
|
| 695 |
|
|
func (z *Tokenizer) Raw() []byte {
|
| 696 |
|
|
return z.buf[z.raw.start:z.raw.end]
|
| 697 |
|
|
}
|
| 698 |
|
|
|
| 699 |
|
|
// Text returns the unescaped text of a text, comment or doctype token. The
|
| 700 |
|
|
// contents of the returned slice may change on the next call to Next.
|
| 701 |
|
|
func (z *Tokenizer) Text() []byte {
|
| 702 |
|
|
switch z.tt {
|
| 703 |
|
|
case TextToken, CommentToken, DoctypeToken:
|
| 704 |
|
|
s := z.buf[z.data.start:z.data.end]
|
| 705 |
|
|
z.data.start = z.raw.end
|
| 706 |
|
|
z.data.end = z.raw.end
|
| 707 |
|
|
if !z.textIsRaw {
|
| 708 |
|
|
s = unescape(s)
|
| 709 |
|
|
}
|
| 710 |
|
|
return s
|
| 711 |
|
|
}
|
| 712 |
|
|
return nil
|
| 713 |
|
|
}
|
| 714 |
|
|
|
| 715 |
|
|
// TagName returns the lower-cased name of a tag token (the `img` out of
|
| 716 |
|
|
// ` `) and whether the tag has attributes.
|
| 717 |
|
|
// The contents of the returned slice may change on the next call to Next.
|
| 718 |
|
|
func (z *Tokenizer) TagName() (name []byte, hasAttr bool) {
|
| 719 |
|
|
if z.data.start < z.data.end {
|
| 720 |
|
|
switch z.tt {
|
| 721 |
|
|
case StartTagToken, EndTagToken, SelfClosingTagToken:
|
| 722 |
|
|
s := z.buf[z.data.start:z.data.end]
|
| 723 |
|
|
z.data.start = z.raw.end
|
| 724 |
|
|
z.data.end = z.raw.end
|
| 725 |
|
|
return lower(s), z.nAttrReturned < len(z.attr)
|
| 726 |
|
|
}
|
| 727 |
|
|
}
|
| 728 |
|
|
return nil, false
|
| 729 |
|
|
}
|
| 730 |
|
|
|
| 731 |
|
|
// TagAttr returns the lower-cased key and unescaped value of the next unparsed
|
| 732 |
|
|
// attribute for the current tag token and whether there are more attributes.
|
| 733 |
|
|
// The contents of the returned slices may change on the next call to Next.
|
| 734 |
|
|
func (z *Tokenizer) TagAttr() (key, val []byte, moreAttr bool) {
|
| 735 |
|
|
if z.nAttrReturned < len(z.attr) {
|
| 736 |
|
|
switch z.tt {
|
| 737 |
|
|
case StartTagToken, SelfClosingTagToken:
|
| 738 |
|
|
x := z.attr[z.nAttrReturned]
|
| 739 |
|
|
z.nAttrReturned++
|
| 740 |
|
|
key = z.buf[x[0].start:x[0].end]
|
| 741 |
|
|
val = z.buf[x[1].start:x[1].end]
|
| 742 |
|
|
return lower(key), unescape(val), z.nAttrReturned < len(z.attr)
|
| 743 |
|
|
}
|
| 744 |
|
|
}
|
| 745 |
|
|
return nil, nil, false
|
| 746 |
|
|
}
|
| 747 |
|
|
|
| 748 |
|
|
// Token returns the next Token. The result's Data and Attr values remain valid
|
| 749 |
|
|
// after subsequent Next calls.
|
| 750 |
|
|
func (z *Tokenizer) Token() Token {
|
| 751 |
|
|
t := Token{Type: z.tt}
|
| 752 |
|
|
switch z.tt {
|
| 753 |
|
|
case TextToken, CommentToken, DoctypeToken:
|
| 754 |
|
|
t.Data = string(z.Text())
|
| 755 |
|
|
case StartTagToken, SelfClosingTagToken:
|
| 756 |
|
|
var attr []Attribute
|
| 757 |
|
|
name, moreAttr := z.TagName()
|
| 758 |
|
|
for moreAttr {
|
| 759 |
|
|
var key, val []byte
|
| 760 |
|
|
key, val, moreAttr = z.TagAttr()
|
| 761 |
|
|
attr = append(attr, Attribute{"", string(key), string(val)})
|
| 762 |
|
|
}
|
| 763 |
|
|
t.Data = string(name)
|
| 764 |
|
|
t.Attr = attr
|
| 765 |
|
|
case EndTagToken:
|
| 766 |
|
|
name, _ := z.TagName()
|
| 767 |
|
|
t.Data = string(name)
|
| 768 |
|
|
}
|
| 769 |
|
|
return t
|
| 770 |
|
|
}
|
| 771 |
|
|
|
| 772 |
|
|
// NewTokenizer returns a new HTML Tokenizer for the given Reader.
|
| 773 |
|
|
// The input is assumed to be UTF-8 encoded.
|
| 774 |
|
|
func NewTokenizer(r io.Reader) *Tokenizer {
|
| 775 |
|
|
return &Tokenizer{
|
| 776 |
|
|
r: r,
|
| 777 |
|
|
buf: make([]byte, 0, 4096),
|
| 778 |
|
|
}
|
| 779 |
|
|
}
|
