Fri, 16 Jan 2015 18:13:44 +0100
Integrate suggestion from review to improve consistency with existing code.
michael@0 | 1 | /* |
michael@0 | 2 | * Copyright (c) 2007 Henri Sivonen |
michael@0 | 3 | * Copyright (c) 2008-2010 Mozilla Foundation |
michael@0 | 4 | * |
michael@0 | 5 | * Permission is hereby granted, free of charge, to any person obtaining a |
michael@0 | 6 | * copy of this software and associated documentation files (the "Software"), |
michael@0 | 7 | * to deal in the Software without restriction, including without limitation |
michael@0 | 8 | * the rights to use, copy, modify, merge, publish, distribute, sublicense, |
michael@0 | 9 | * and/or sell copies of the Software, and to permit persons to whom the |
michael@0 | 10 | * Software is furnished to do so, subject to the following conditions: |
michael@0 | 11 | * |
michael@0 | 12 | * The above copyright notice and this permission notice shall be included in |
michael@0 | 13 | * all copies or substantial portions of the Software. |
michael@0 | 14 | * |
michael@0 | 15 | * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR |
michael@0 | 16 | * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, |
michael@0 | 17 | * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL |
michael@0 | 18 | * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER |
michael@0 | 19 | * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING |
michael@0 | 20 | * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER |
michael@0 | 21 | * DEALINGS IN THE SOFTWARE. |
michael@0 | 22 | */ |
michael@0 | 23 | |
michael@0 | 24 | /* |
michael@0 | 25 | * THIS IS A GENERATED FILE. PLEASE DO NOT EDIT. |
michael@0 | 26 | * Please edit MetaScanner.java instead and regenerate. |
michael@0 | 27 | */ |
michael@0 | 28 | |
michael@0 | 29 | #define nsHtml5MetaScanner_cpp__ |
michael@0 | 30 | |
michael@0 | 31 | #include "nsIAtom.h" |
michael@0 | 32 | #include "nsHtml5AtomTable.h" |
michael@0 | 33 | #include "nsString.h" |
michael@0 | 34 | #include "nsNameSpaceManager.h" |
michael@0 | 35 | #include "nsIContent.h" |
michael@0 | 36 | #include "nsTraceRefcnt.h" |
michael@0 | 37 | #include "jArray.h" |
michael@0 | 38 | #include "nsHtml5ArrayCopy.h" |
michael@0 | 39 | #include "nsAHtml5TreeBuilderState.h" |
michael@0 | 40 | #include "nsHtml5Atoms.h" |
michael@0 | 41 | #include "nsHtml5ByteReadable.h" |
michael@0 | 42 | #include "nsIUnicodeDecoder.h" |
michael@0 | 43 | #include "nsHtml5Macros.h" |
michael@0 | 44 | #include "nsIContentHandle.h" |
michael@0 | 45 | |
michael@0 | 46 | #include "nsHtml5Tokenizer.h" |
michael@0 | 47 | #include "nsHtml5TreeBuilder.h" |
michael@0 | 48 | #include "nsHtml5AttributeName.h" |
michael@0 | 49 | #include "nsHtml5ElementName.h" |
michael@0 | 50 | #include "nsHtml5HtmlAttributes.h" |
michael@0 | 51 | #include "nsHtml5StackNode.h" |
michael@0 | 52 | #include "nsHtml5UTF16Buffer.h" |
michael@0 | 53 | #include "nsHtml5StateSnapshot.h" |
michael@0 | 54 | #include "nsHtml5Portability.h" |
michael@0 | 55 | |
michael@0 | 56 | #include "nsHtml5MetaScanner.h" |
michael@0 | 57 | |
michael@0 | 58 | static char16_t const CHARSET_DATA[] = { 'h', 'a', 'r', 's', 'e', 't' }; |
michael@0 | 59 | staticJArray<char16_t,int32_t> nsHtml5MetaScanner::CHARSET = { CHARSET_DATA, MOZ_ARRAY_LENGTH(CHARSET_DATA) }; |
michael@0 | 60 | static char16_t const CONTENT_DATA[] = { 'o', 'n', 't', 'e', 'n', 't' }; |
michael@0 | 61 | staticJArray<char16_t,int32_t> nsHtml5MetaScanner::CONTENT = { CONTENT_DATA, MOZ_ARRAY_LENGTH(CONTENT_DATA) }; |
michael@0 | 62 | static char16_t const HTTP_EQUIV_DATA[] = { 't', 't', 'p', '-', 'e', 'q', 'u', 'i', 'v' }; |
michael@0 | 63 | staticJArray<char16_t,int32_t> nsHtml5MetaScanner::HTTP_EQUIV = { HTTP_EQUIV_DATA, MOZ_ARRAY_LENGTH(HTTP_EQUIV_DATA) }; |
michael@0 | 64 | static char16_t const CONTENT_TYPE_DATA[] = { 'c', 'o', 'n', 't', 'e', 'n', 't', '-', 't', 'y', 'p', 'e' }; |
michael@0 | 65 | staticJArray<char16_t,int32_t> nsHtml5MetaScanner::CONTENT_TYPE = { CONTENT_TYPE_DATA, MOZ_ARRAY_LENGTH(CONTENT_TYPE_DATA) }; |
michael@0 | 66 | |
michael@0 | 67 | nsHtml5MetaScanner::nsHtml5MetaScanner() |
michael@0 | 68 | : readable(nullptr), |
michael@0 | 69 | metaState(NS_HTML5META_SCANNER_NO), |
michael@0 | 70 | contentIndex(INT32_MAX), |
michael@0 | 71 | charsetIndex(INT32_MAX), |
michael@0 | 72 | httpEquivIndex(INT32_MAX), |
michael@0 | 73 | contentTypeIndex(INT32_MAX), |
michael@0 | 74 | stateSave(NS_HTML5META_SCANNER_DATA), |
michael@0 | 75 | strBufLen(0), |
michael@0 | 76 | strBuf(jArray<char16_t,int32_t>::newJArray(36)), |
michael@0 | 77 | content(nullptr), |
michael@0 | 78 | charset(nullptr), |
michael@0 | 79 | httpEquivState(NS_HTML5META_SCANNER_HTTP_EQUIV_NOT_SEEN) |
michael@0 | 80 | { |
michael@0 | 81 | MOZ_COUNT_CTOR(nsHtml5MetaScanner); |
michael@0 | 82 | } |
michael@0 | 83 | |
michael@0 | 84 | |
michael@0 | 85 | nsHtml5MetaScanner::~nsHtml5MetaScanner() |
michael@0 | 86 | { |
michael@0 | 87 | MOZ_COUNT_DTOR(nsHtml5MetaScanner); |
michael@0 | 88 | nsHtml5Portability::releaseString(content); |
michael@0 | 89 | nsHtml5Portability::releaseString(charset); |
michael@0 | 90 | } |
michael@0 | 91 | |
michael@0 | 92 | void |
michael@0 | 93 | nsHtml5MetaScanner::stateLoop(int32_t state) |
michael@0 | 94 | { |
michael@0 | 95 | int32_t c = -1; |
michael@0 | 96 | bool reconsume = false; |
michael@0 | 97 | stateloop: for (; ; ) { |
michael@0 | 98 | switch(state) { |
michael@0 | 99 | case NS_HTML5META_SCANNER_DATA: { |
michael@0 | 100 | for (; ; ) { |
michael@0 | 101 | if (reconsume) { |
michael@0 | 102 | reconsume = false; |
michael@0 | 103 | } else { |
michael@0 | 104 | c = read(); |
michael@0 | 105 | } |
michael@0 | 106 | switch(c) { |
michael@0 | 107 | case -1: { |
michael@0 | 108 | NS_HTML5_BREAK(stateloop); |
michael@0 | 109 | } |
michael@0 | 110 | case '<': { |
michael@0 | 111 | state = NS_HTML5META_SCANNER_TAG_OPEN; |
michael@0 | 112 | NS_HTML5_BREAK(dataloop); |
michael@0 | 113 | } |
michael@0 | 114 | default: { |
michael@0 | 115 | continue; |
michael@0 | 116 | } |
michael@0 | 117 | } |
michael@0 | 118 | } |
michael@0 | 119 | dataloop_end: ; |
michael@0 | 120 | } |
michael@0 | 121 | case NS_HTML5META_SCANNER_TAG_OPEN: { |
michael@0 | 122 | for (; ; ) { |
michael@0 | 123 | c = read(); |
michael@0 | 124 | switch(c) { |
michael@0 | 125 | case -1: { |
michael@0 | 126 | NS_HTML5_BREAK(stateloop); |
michael@0 | 127 | } |
michael@0 | 128 | case 'm': |
michael@0 | 129 | case 'M': { |
michael@0 | 130 | metaState = NS_HTML5META_SCANNER_M; |
michael@0 | 131 | state = NS_HTML5META_SCANNER_TAG_NAME; |
michael@0 | 132 | NS_HTML5_BREAK(tagopenloop); |
michael@0 | 133 | } |
michael@0 | 134 | case '!': { |
michael@0 | 135 | state = NS_HTML5META_SCANNER_MARKUP_DECLARATION_OPEN; |
michael@0 | 136 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 137 | } |
michael@0 | 138 | case '\?': |
michael@0 | 139 | case '/': { |
michael@0 | 140 | state = NS_HTML5META_SCANNER_SCAN_UNTIL_GT; |
michael@0 | 141 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 142 | } |
michael@0 | 143 | case '>': { |
michael@0 | 144 | state = NS_HTML5META_SCANNER_DATA; |
michael@0 | 145 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 146 | } |
michael@0 | 147 | default: { |
michael@0 | 148 | if ((c >= 'A' && c <= 'Z') || (c >= 'a' && c <= 'z')) { |
michael@0 | 149 | metaState = NS_HTML5META_SCANNER_NO; |
michael@0 | 150 | state = NS_HTML5META_SCANNER_TAG_NAME; |
michael@0 | 151 | NS_HTML5_BREAK(tagopenloop); |
michael@0 | 152 | } |
michael@0 | 153 | state = NS_HTML5META_SCANNER_DATA; |
michael@0 | 154 | reconsume = true; |
michael@0 | 155 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 156 | } |
michael@0 | 157 | } |
michael@0 | 158 | } |
michael@0 | 159 | tagopenloop_end: ; |
michael@0 | 160 | } |
michael@0 | 161 | case NS_HTML5META_SCANNER_TAG_NAME: { |
michael@0 | 162 | for (; ; ) { |
michael@0 | 163 | c = read(); |
michael@0 | 164 | switch(c) { |
michael@0 | 165 | case -1: { |
michael@0 | 166 | NS_HTML5_BREAK(stateloop); |
michael@0 | 167 | } |
michael@0 | 168 | case ' ': |
michael@0 | 169 | case '\t': |
michael@0 | 170 | case '\n': |
michael@0 | 171 | case '\f': { |
michael@0 | 172 | state = NS_HTML5META_SCANNER_BEFORE_ATTRIBUTE_NAME; |
michael@0 | 173 | NS_HTML5_BREAK(tagnameloop); |
michael@0 | 174 | } |
michael@0 | 175 | case '/': { |
michael@0 | 176 | state = NS_HTML5META_SCANNER_SELF_CLOSING_START_TAG; |
michael@0 | 177 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 178 | } |
michael@0 | 179 | case '>': { |
michael@0 | 180 | state = NS_HTML5META_SCANNER_DATA; |
michael@0 | 181 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 182 | } |
michael@0 | 183 | case 'e': |
michael@0 | 184 | case 'E': { |
michael@0 | 185 | if (metaState == NS_HTML5META_SCANNER_M) { |
michael@0 | 186 | metaState = NS_HTML5META_SCANNER_E; |
michael@0 | 187 | } else { |
michael@0 | 188 | metaState = NS_HTML5META_SCANNER_NO; |
michael@0 | 189 | } |
michael@0 | 190 | continue; |
michael@0 | 191 | } |
michael@0 | 192 | case 't': |
michael@0 | 193 | case 'T': { |
michael@0 | 194 | if (metaState == NS_HTML5META_SCANNER_E) { |
michael@0 | 195 | metaState = NS_HTML5META_SCANNER_T; |
michael@0 | 196 | } else { |
michael@0 | 197 | metaState = NS_HTML5META_SCANNER_NO; |
michael@0 | 198 | } |
michael@0 | 199 | continue; |
michael@0 | 200 | } |
michael@0 | 201 | case 'a': |
michael@0 | 202 | case 'A': { |
michael@0 | 203 | if (metaState == NS_HTML5META_SCANNER_T) { |
michael@0 | 204 | metaState = NS_HTML5META_SCANNER_A; |
michael@0 | 205 | } else { |
michael@0 | 206 | metaState = NS_HTML5META_SCANNER_NO; |
michael@0 | 207 | } |
michael@0 | 208 | continue; |
michael@0 | 209 | } |
michael@0 | 210 | default: { |
michael@0 | 211 | metaState = NS_HTML5META_SCANNER_NO; |
michael@0 | 212 | continue; |
michael@0 | 213 | } |
michael@0 | 214 | } |
michael@0 | 215 | } |
michael@0 | 216 | tagnameloop_end: ; |
michael@0 | 217 | } |
michael@0 | 218 | case NS_HTML5META_SCANNER_BEFORE_ATTRIBUTE_NAME: { |
michael@0 | 219 | for (; ; ) { |
michael@0 | 220 | if (reconsume) { |
michael@0 | 221 | reconsume = false; |
michael@0 | 222 | } else { |
michael@0 | 223 | c = read(); |
michael@0 | 224 | } |
michael@0 | 225 | switch(c) { |
michael@0 | 226 | case -1: { |
michael@0 | 227 | NS_HTML5_BREAK(stateloop); |
michael@0 | 228 | } |
michael@0 | 229 | case ' ': |
michael@0 | 230 | case '\t': |
michael@0 | 231 | case '\n': |
michael@0 | 232 | case '\f': { |
michael@0 | 233 | continue; |
michael@0 | 234 | } |
michael@0 | 235 | case '/': { |
michael@0 | 236 | state = NS_HTML5META_SCANNER_SELF_CLOSING_START_TAG; |
michael@0 | 237 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 238 | } |
michael@0 | 239 | case '>': { |
michael@0 | 240 | if (handleTag()) { |
michael@0 | 241 | NS_HTML5_BREAK(stateloop); |
michael@0 | 242 | } |
michael@0 | 243 | state = NS_HTML5META_SCANNER_DATA; |
michael@0 | 244 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 245 | } |
michael@0 | 246 | case 'c': |
michael@0 | 247 | case 'C': { |
michael@0 | 248 | contentIndex = 0; |
michael@0 | 249 | charsetIndex = 0; |
michael@0 | 250 | httpEquivIndex = INT32_MAX; |
michael@0 | 251 | contentTypeIndex = INT32_MAX; |
michael@0 | 252 | state = NS_HTML5META_SCANNER_ATTRIBUTE_NAME; |
michael@0 | 253 | NS_HTML5_BREAK(beforeattributenameloop); |
michael@0 | 254 | } |
michael@0 | 255 | case 'h': |
michael@0 | 256 | case 'H': { |
michael@0 | 257 | contentIndex = INT32_MAX; |
michael@0 | 258 | charsetIndex = INT32_MAX; |
michael@0 | 259 | httpEquivIndex = 0; |
michael@0 | 260 | contentTypeIndex = INT32_MAX; |
michael@0 | 261 | state = NS_HTML5META_SCANNER_ATTRIBUTE_NAME; |
michael@0 | 262 | NS_HTML5_BREAK(beforeattributenameloop); |
michael@0 | 263 | } |
michael@0 | 264 | default: { |
michael@0 | 265 | contentIndex = INT32_MAX; |
michael@0 | 266 | charsetIndex = INT32_MAX; |
michael@0 | 267 | httpEquivIndex = INT32_MAX; |
michael@0 | 268 | contentTypeIndex = INT32_MAX; |
michael@0 | 269 | state = NS_HTML5META_SCANNER_ATTRIBUTE_NAME; |
michael@0 | 270 | NS_HTML5_BREAK(beforeattributenameloop); |
michael@0 | 271 | } |
michael@0 | 272 | } |
michael@0 | 273 | } |
michael@0 | 274 | beforeattributenameloop_end: ; |
michael@0 | 275 | } |
michael@0 | 276 | case NS_HTML5META_SCANNER_ATTRIBUTE_NAME: { |
michael@0 | 277 | for (; ; ) { |
michael@0 | 278 | c = read(); |
michael@0 | 279 | switch(c) { |
michael@0 | 280 | case -1: { |
michael@0 | 281 | NS_HTML5_BREAK(stateloop); |
michael@0 | 282 | } |
michael@0 | 283 | case ' ': |
michael@0 | 284 | case '\t': |
michael@0 | 285 | case '\n': |
michael@0 | 286 | case '\f': { |
michael@0 | 287 | state = NS_HTML5META_SCANNER_AFTER_ATTRIBUTE_NAME; |
michael@0 | 288 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 289 | } |
michael@0 | 290 | case '/': { |
michael@0 | 291 | state = NS_HTML5META_SCANNER_SELF_CLOSING_START_TAG; |
michael@0 | 292 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 293 | } |
michael@0 | 294 | case '=': { |
michael@0 | 295 | strBufLen = 0; |
michael@0 | 296 | contentTypeIndex = 0; |
michael@0 | 297 | state = NS_HTML5META_SCANNER_BEFORE_ATTRIBUTE_VALUE; |
michael@0 | 298 | NS_HTML5_BREAK(attributenameloop); |
michael@0 | 299 | } |
michael@0 | 300 | case '>': { |
michael@0 | 301 | if (handleTag()) { |
michael@0 | 302 | NS_HTML5_BREAK(stateloop); |
michael@0 | 303 | } |
michael@0 | 304 | state = NS_HTML5META_SCANNER_DATA; |
michael@0 | 305 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 306 | } |
michael@0 | 307 | default: { |
michael@0 | 308 | if (metaState == NS_HTML5META_SCANNER_A) { |
michael@0 | 309 | if (c >= 'A' && c <= 'Z') { |
michael@0 | 310 | c += 0x20; |
michael@0 | 311 | } |
michael@0 | 312 | if (contentIndex < CONTENT.length && c == CONTENT[contentIndex]) { |
michael@0 | 313 | ++contentIndex; |
michael@0 | 314 | } else { |
michael@0 | 315 | contentIndex = INT32_MAX; |
michael@0 | 316 | } |
michael@0 | 317 | if (charsetIndex < CHARSET.length && c == CHARSET[charsetIndex]) { |
michael@0 | 318 | ++charsetIndex; |
michael@0 | 319 | } else { |
michael@0 | 320 | charsetIndex = INT32_MAX; |
michael@0 | 321 | } |
michael@0 | 322 | if (httpEquivIndex < HTTP_EQUIV.length && c == HTTP_EQUIV[httpEquivIndex]) { |
michael@0 | 323 | ++httpEquivIndex; |
michael@0 | 324 | } else { |
michael@0 | 325 | httpEquivIndex = INT32_MAX; |
michael@0 | 326 | } |
michael@0 | 327 | } |
michael@0 | 328 | continue; |
michael@0 | 329 | } |
michael@0 | 330 | } |
michael@0 | 331 | } |
michael@0 | 332 | attributenameloop_end: ; |
michael@0 | 333 | } |
michael@0 | 334 | case NS_HTML5META_SCANNER_BEFORE_ATTRIBUTE_VALUE: { |
michael@0 | 335 | for (; ; ) { |
michael@0 | 336 | c = read(); |
michael@0 | 337 | switch(c) { |
michael@0 | 338 | case -1: { |
michael@0 | 339 | NS_HTML5_BREAK(stateloop); |
michael@0 | 340 | } |
michael@0 | 341 | case ' ': |
michael@0 | 342 | case '\t': |
michael@0 | 343 | case '\n': |
michael@0 | 344 | case '\f': { |
michael@0 | 345 | continue; |
michael@0 | 346 | } |
michael@0 | 347 | case '\"': { |
michael@0 | 348 | state = NS_HTML5META_SCANNER_ATTRIBUTE_VALUE_DOUBLE_QUOTED; |
michael@0 | 349 | NS_HTML5_BREAK(beforeattributevalueloop); |
michael@0 | 350 | } |
michael@0 | 351 | case '\'': { |
michael@0 | 352 | state = NS_HTML5META_SCANNER_ATTRIBUTE_VALUE_SINGLE_QUOTED; |
michael@0 | 353 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 354 | } |
michael@0 | 355 | case '>': { |
michael@0 | 356 | if (handleTag()) { |
michael@0 | 357 | NS_HTML5_BREAK(stateloop); |
michael@0 | 358 | } |
michael@0 | 359 | state = NS_HTML5META_SCANNER_DATA; |
michael@0 | 360 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 361 | } |
michael@0 | 362 | default: { |
michael@0 | 363 | handleCharInAttributeValue(c); |
michael@0 | 364 | state = NS_HTML5META_SCANNER_ATTRIBUTE_VALUE_UNQUOTED; |
michael@0 | 365 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 366 | } |
michael@0 | 367 | } |
michael@0 | 368 | } |
michael@0 | 369 | beforeattributevalueloop_end: ; |
michael@0 | 370 | } |
michael@0 | 371 | case NS_HTML5META_SCANNER_ATTRIBUTE_VALUE_DOUBLE_QUOTED: { |
michael@0 | 372 | for (; ; ) { |
michael@0 | 373 | if (reconsume) { |
michael@0 | 374 | reconsume = false; |
michael@0 | 375 | } else { |
michael@0 | 376 | c = read(); |
michael@0 | 377 | } |
michael@0 | 378 | switch(c) { |
michael@0 | 379 | case -1: { |
michael@0 | 380 | NS_HTML5_BREAK(stateloop); |
michael@0 | 381 | } |
michael@0 | 382 | case '\"': { |
michael@0 | 383 | handleAttributeValue(); |
michael@0 | 384 | state = NS_HTML5META_SCANNER_AFTER_ATTRIBUTE_VALUE_QUOTED; |
michael@0 | 385 | NS_HTML5_BREAK(attributevaluedoublequotedloop); |
michael@0 | 386 | } |
michael@0 | 387 | default: { |
michael@0 | 388 | handleCharInAttributeValue(c); |
michael@0 | 389 | continue; |
michael@0 | 390 | } |
michael@0 | 391 | } |
michael@0 | 392 | } |
michael@0 | 393 | attributevaluedoublequotedloop_end: ; |
michael@0 | 394 | } |
michael@0 | 395 | case NS_HTML5META_SCANNER_AFTER_ATTRIBUTE_VALUE_QUOTED: { |
michael@0 | 396 | for (; ; ) { |
michael@0 | 397 | c = read(); |
michael@0 | 398 | switch(c) { |
michael@0 | 399 | case -1: { |
michael@0 | 400 | NS_HTML5_BREAK(stateloop); |
michael@0 | 401 | } |
michael@0 | 402 | case ' ': |
michael@0 | 403 | case '\t': |
michael@0 | 404 | case '\n': |
michael@0 | 405 | case '\f': { |
michael@0 | 406 | state = NS_HTML5META_SCANNER_BEFORE_ATTRIBUTE_NAME; |
michael@0 | 407 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 408 | } |
michael@0 | 409 | case '/': { |
michael@0 | 410 | state = NS_HTML5META_SCANNER_SELF_CLOSING_START_TAG; |
michael@0 | 411 | NS_HTML5_BREAK(afterattributevaluequotedloop); |
michael@0 | 412 | } |
michael@0 | 413 | case '>': { |
michael@0 | 414 | if (handleTag()) { |
michael@0 | 415 | NS_HTML5_BREAK(stateloop); |
michael@0 | 416 | } |
michael@0 | 417 | state = NS_HTML5META_SCANNER_DATA; |
michael@0 | 418 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 419 | } |
michael@0 | 420 | default: { |
michael@0 | 421 | state = NS_HTML5META_SCANNER_BEFORE_ATTRIBUTE_NAME; |
michael@0 | 422 | reconsume = true; |
michael@0 | 423 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 424 | } |
michael@0 | 425 | } |
michael@0 | 426 | } |
michael@0 | 427 | afterattributevaluequotedloop_end: ; |
michael@0 | 428 | } |
michael@0 | 429 | case NS_HTML5META_SCANNER_SELF_CLOSING_START_TAG: { |
michael@0 | 430 | c = read(); |
michael@0 | 431 | switch(c) { |
michael@0 | 432 | case -1: { |
michael@0 | 433 | NS_HTML5_BREAK(stateloop); |
michael@0 | 434 | } |
michael@0 | 435 | case '>': { |
michael@0 | 436 | if (handleTag()) { |
michael@0 | 437 | NS_HTML5_BREAK(stateloop); |
michael@0 | 438 | } |
michael@0 | 439 | state = NS_HTML5META_SCANNER_DATA; |
michael@0 | 440 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 441 | } |
michael@0 | 442 | default: { |
michael@0 | 443 | state = NS_HTML5META_SCANNER_BEFORE_ATTRIBUTE_NAME; |
michael@0 | 444 | reconsume = true; |
michael@0 | 445 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 446 | } |
michael@0 | 447 | } |
michael@0 | 448 | } |
michael@0 | 449 | case NS_HTML5META_SCANNER_ATTRIBUTE_VALUE_UNQUOTED: { |
michael@0 | 450 | for (; ; ) { |
michael@0 | 451 | if (reconsume) { |
michael@0 | 452 | reconsume = false; |
michael@0 | 453 | } else { |
michael@0 | 454 | c = read(); |
michael@0 | 455 | } |
michael@0 | 456 | switch(c) { |
michael@0 | 457 | case -1: { |
michael@0 | 458 | NS_HTML5_BREAK(stateloop); |
michael@0 | 459 | } |
michael@0 | 460 | case ' ': |
michael@0 | 461 | case '\t': |
michael@0 | 462 | case '\n': |
michael@0 | 463 | case '\f': { |
michael@0 | 464 | handleAttributeValue(); |
michael@0 | 465 | state = NS_HTML5META_SCANNER_BEFORE_ATTRIBUTE_NAME; |
michael@0 | 466 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 467 | } |
michael@0 | 468 | case '>': { |
michael@0 | 469 | handleAttributeValue(); |
michael@0 | 470 | if (handleTag()) { |
michael@0 | 471 | NS_HTML5_BREAK(stateloop); |
michael@0 | 472 | } |
michael@0 | 473 | state = NS_HTML5META_SCANNER_DATA; |
michael@0 | 474 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 475 | } |
michael@0 | 476 | default: { |
michael@0 | 477 | handleCharInAttributeValue(c); |
michael@0 | 478 | continue; |
michael@0 | 479 | } |
michael@0 | 480 | } |
michael@0 | 481 | } |
michael@0 | 482 | } |
michael@0 | 483 | case NS_HTML5META_SCANNER_AFTER_ATTRIBUTE_NAME: { |
michael@0 | 484 | for (; ; ) { |
michael@0 | 485 | c = read(); |
michael@0 | 486 | switch(c) { |
michael@0 | 487 | case -1: { |
michael@0 | 488 | NS_HTML5_BREAK(stateloop); |
michael@0 | 489 | } |
michael@0 | 490 | case ' ': |
michael@0 | 491 | case '\t': |
michael@0 | 492 | case '\n': |
michael@0 | 493 | case '\f': { |
michael@0 | 494 | continue; |
michael@0 | 495 | } |
michael@0 | 496 | case '/': { |
michael@0 | 497 | handleAttributeValue(); |
michael@0 | 498 | state = NS_HTML5META_SCANNER_SELF_CLOSING_START_TAG; |
michael@0 | 499 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 500 | } |
michael@0 | 501 | case '=': { |
michael@0 | 502 | strBufLen = 0; |
michael@0 | 503 | contentTypeIndex = 0; |
michael@0 | 504 | state = NS_HTML5META_SCANNER_BEFORE_ATTRIBUTE_VALUE; |
michael@0 | 505 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 506 | } |
michael@0 | 507 | case '>': { |
michael@0 | 508 | handleAttributeValue(); |
michael@0 | 509 | if (handleTag()) { |
michael@0 | 510 | NS_HTML5_BREAK(stateloop); |
michael@0 | 511 | } |
michael@0 | 512 | state = NS_HTML5META_SCANNER_DATA; |
michael@0 | 513 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 514 | } |
michael@0 | 515 | case 'c': |
michael@0 | 516 | case 'C': { |
michael@0 | 517 | contentIndex = 0; |
michael@0 | 518 | charsetIndex = 0; |
michael@0 | 519 | state = NS_HTML5META_SCANNER_ATTRIBUTE_NAME; |
michael@0 | 520 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 521 | } |
michael@0 | 522 | default: { |
michael@0 | 523 | contentIndex = INT32_MAX; |
michael@0 | 524 | charsetIndex = INT32_MAX; |
michael@0 | 525 | state = NS_HTML5META_SCANNER_ATTRIBUTE_NAME; |
michael@0 | 526 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 527 | } |
michael@0 | 528 | } |
michael@0 | 529 | } |
michael@0 | 530 | } |
michael@0 | 531 | case NS_HTML5META_SCANNER_MARKUP_DECLARATION_OPEN: { |
michael@0 | 532 | for (; ; ) { |
michael@0 | 533 | c = read(); |
michael@0 | 534 | switch(c) { |
michael@0 | 535 | case -1: { |
michael@0 | 536 | NS_HTML5_BREAK(stateloop); |
michael@0 | 537 | } |
michael@0 | 538 | case '-': { |
michael@0 | 539 | state = NS_HTML5META_SCANNER_MARKUP_DECLARATION_HYPHEN; |
michael@0 | 540 | NS_HTML5_BREAK(markupdeclarationopenloop); |
michael@0 | 541 | } |
michael@0 | 542 | default: { |
michael@0 | 543 | state = NS_HTML5META_SCANNER_SCAN_UNTIL_GT; |
michael@0 | 544 | reconsume = true; |
michael@0 | 545 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 546 | } |
michael@0 | 547 | } |
michael@0 | 548 | } |
michael@0 | 549 | markupdeclarationopenloop_end: ; |
michael@0 | 550 | } |
michael@0 | 551 | case NS_HTML5META_SCANNER_MARKUP_DECLARATION_HYPHEN: { |
michael@0 | 552 | for (; ; ) { |
michael@0 | 553 | c = read(); |
michael@0 | 554 | switch(c) { |
michael@0 | 555 | case -1: { |
michael@0 | 556 | NS_HTML5_BREAK(stateloop); |
michael@0 | 557 | } |
michael@0 | 558 | case '-': { |
michael@0 | 559 | state = NS_HTML5META_SCANNER_COMMENT_START; |
michael@0 | 560 | NS_HTML5_BREAK(markupdeclarationhyphenloop); |
michael@0 | 561 | } |
michael@0 | 562 | default: { |
michael@0 | 563 | state = NS_HTML5META_SCANNER_SCAN_UNTIL_GT; |
michael@0 | 564 | reconsume = true; |
michael@0 | 565 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 566 | } |
michael@0 | 567 | } |
michael@0 | 568 | } |
michael@0 | 569 | markupdeclarationhyphenloop_end: ; |
michael@0 | 570 | } |
michael@0 | 571 | case NS_HTML5META_SCANNER_COMMENT_START: { |
michael@0 | 572 | for (; ; ) { |
michael@0 | 573 | c = read(); |
michael@0 | 574 | switch(c) { |
michael@0 | 575 | case -1: { |
michael@0 | 576 | NS_HTML5_BREAK(stateloop); |
michael@0 | 577 | } |
michael@0 | 578 | case '-': { |
michael@0 | 579 | state = NS_HTML5META_SCANNER_COMMENT_START_DASH; |
michael@0 | 580 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 581 | } |
michael@0 | 582 | case '>': { |
michael@0 | 583 | state = NS_HTML5META_SCANNER_DATA; |
michael@0 | 584 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 585 | } |
michael@0 | 586 | default: { |
michael@0 | 587 | state = NS_HTML5META_SCANNER_COMMENT; |
michael@0 | 588 | NS_HTML5_BREAK(commentstartloop); |
michael@0 | 589 | } |
michael@0 | 590 | } |
michael@0 | 591 | } |
michael@0 | 592 | commentstartloop_end: ; |
michael@0 | 593 | } |
michael@0 | 594 | case NS_HTML5META_SCANNER_COMMENT: { |
michael@0 | 595 | for (; ; ) { |
michael@0 | 596 | c = read(); |
michael@0 | 597 | switch(c) { |
michael@0 | 598 | case -1: { |
michael@0 | 599 | NS_HTML5_BREAK(stateloop); |
michael@0 | 600 | } |
michael@0 | 601 | case '-': { |
michael@0 | 602 | state = NS_HTML5META_SCANNER_COMMENT_END_DASH; |
michael@0 | 603 | NS_HTML5_BREAK(commentloop); |
michael@0 | 604 | } |
michael@0 | 605 | default: { |
michael@0 | 606 | continue; |
michael@0 | 607 | } |
michael@0 | 608 | } |
michael@0 | 609 | } |
michael@0 | 610 | commentloop_end: ; |
michael@0 | 611 | } |
michael@0 | 612 | case NS_HTML5META_SCANNER_COMMENT_END_DASH: { |
michael@0 | 613 | for (; ; ) { |
michael@0 | 614 | c = read(); |
michael@0 | 615 | switch(c) { |
michael@0 | 616 | case -1: { |
michael@0 | 617 | NS_HTML5_BREAK(stateloop); |
michael@0 | 618 | } |
michael@0 | 619 | case '-': { |
michael@0 | 620 | state = NS_HTML5META_SCANNER_COMMENT_END; |
michael@0 | 621 | NS_HTML5_BREAK(commentenddashloop); |
michael@0 | 622 | } |
michael@0 | 623 | default: { |
michael@0 | 624 | state = NS_HTML5META_SCANNER_COMMENT; |
michael@0 | 625 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 626 | } |
michael@0 | 627 | } |
michael@0 | 628 | } |
michael@0 | 629 | commentenddashloop_end: ; |
michael@0 | 630 | } |
michael@0 | 631 | case NS_HTML5META_SCANNER_COMMENT_END: { |
michael@0 | 632 | for (; ; ) { |
michael@0 | 633 | c = read(); |
michael@0 | 634 | switch(c) { |
michael@0 | 635 | case -1: { |
michael@0 | 636 | NS_HTML5_BREAK(stateloop); |
michael@0 | 637 | } |
michael@0 | 638 | case '>': { |
michael@0 | 639 | state = NS_HTML5META_SCANNER_DATA; |
michael@0 | 640 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 641 | } |
michael@0 | 642 | case '-': { |
michael@0 | 643 | continue; |
michael@0 | 644 | } |
michael@0 | 645 | default: { |
michael@0 | 646 | state = NS_HTML5META_SCANNER_COMMENT; |
michael@0 | 647 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 648 | } |
michael@0 | 649 | } |
michael@0 | 650 | } |
michael@0 | 651 | } |
michael@0 | 652 | case NS_HTML5META_SCANNER_COMMENT_START_DASH: { |
michael@0 | 653 | c = read(); |
michael@0 | 654 | switch(c) { |
michael@0 | 655 | case -1: { |
michael@0 | 656 | NS_HTML5_BREAK(stateloop); |
michael@0 | 657 | } |
michael@0 | 658 | case '-': { |
michael@0 | 659 | state = NS_HTML5META_SCANNER_COMMENT_END; |
michael@0 | 660 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 661 | } |
michael@0 | 662 | case '>': { |
michael@0 | 663 | state = NS_HTML5META_SCANNER_DATA; |
michael@0 | 664 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 665 | } |
michael@0 | 666 | default: { |
michael@0 | 667 | state = NS_HTML5META_SCANNER_COMMENT; |
michael@0 | 668 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 669 | } |
michael@0 | 670 | } |
michael@0 | 671 | } |
michael@0 | 672 | case NS_HTML5META_SCANNER_ATTRIBUTE_VALUE_SINGLE_QUOTED: { |
michael@0 | 673 | for (; ; ) { |
michael@0 | 674 | if (reconsume) { |
michael@0 | 675 | reconsume = false; |
michael@0 | 676 | } else { |
michael@0 | 677 | c = read(); |
michael@0 | 678 | } |
michael@0 | 679 | switch(c) { |
michael@0 | 680 | case -1: { |
michael@0 | 681 | NS_HTML5_BREAK(stateloop); |
michael@0 | 682 | } |
michael@0 | 683 | case '\'': { |
michael@0 | 684 | handleAttributeValue(); |
michael@0 | 685 | state = NS_HTML5META_SCANNER_AFTER_ATTRIBUTE_VALUE_QUOTED; |
michael@0 | 686 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 687 | } |
michael@0 | 688 | default: { |
michael@0 | 689 | handleCharInAttributeValue(c); |
michael@0 | 690 | continue; |
michael@0 | 691 | } |
michael@0 | 692 | } |
michael@0 | 693 | } |
michael@0 | 694 | } |
michael@0 | 695 | case NS_HTML5META_SCANNER_SCAN_UNTIL_GT: { |
michael@0 | 696 | for (; ; ) { |
michael@0 | 697 | if (reconsume) { |
michael@0 | 698 | reconsume = false; |
michael@0 | 699 | } else { |
michael@0 | 700 | c = read(); |
michael@0 | 701 | } |
michael@0 | 702 | switch(c) { |
michael@0 | 703 | case -1: { |
michael@0 | 704 | NS_HTML5_BREAK(stateloop); |
michael@0 | 705 | } |
michael@0 | 706 | case '>': { |
michael@0 | 707 | state = NS_HTML5META_SCANNER_DATA; |
michael@0 | 708 | NS_HTML5_CONTINUE(stateloop); |
michael@0 | 709 | } |
michael@0 | 710 | default: { |
michael@0 | 711 | continue; |
michael@0 | 712 | } |
michael@0 | 713 | } |
michael@0 | 714 | } |
michael@0 | 715 | } |
michael@0 | 716 | } |
michael@0 | 717 | } |
michael@0 | 718 | stateloop_end: ; |
michael@0 | 719 | stateSave = state; |
michael@0 | 720 | } |
michael@0 | 721 | |
michael@0 | 722 | void |
michael@0 | 723 | nsHtml5MetaScanner::handleCharInAttributeValue(int32_t c) |
michael@0 | 724 | { |
michael@0 | 725 | if (metaState == NS_HTML5META_SCANNER_A) { |
michael@0 | 726 | if (contentIndex == CONTENT.length || charsetIndex == CHARSET.length) { |
michael@0 | 727 | addToBuffer(c); |
michael@0 | 728 | } else if (httpEquivIndex == HTTP_EQUIV.length) { |
michael@0 | 729 | if (contentTypeIndex < CONTENT_TYPE.length && toAsciiLowerCase(c) == CONTENT_TYPE[contentTypeIndex]) { |
michael@0 | 730 | ++contentTypeIndex; |
michael@0 | 731 | } else { |
michael@0 | 732 | contentTypeIndex = INT32_MAX; |
michael@0 | 733 | } |
michael@0 | 734 | } |
michael@0 | 735 | } |
michael@0 | 736 | } |
michael@0 | 737 | |
michael@0 | 738 | void |
michael@0 | 739 | nsHtml5MetaScanner::addToBuffer(int32_t c) |
michael@0 | 740 | { |
michael@0 | 741 | if (strBufLen == strBuf.length) { |
michael@0 | 742 | jArray<char16_t,int32_t> newBuf = jArray<char16_t,int32_t>::newJArray(strBuf.length + (strBuf.length << 1)); |
michael@0 | 743 | nsHtml5ArrayCopy::arraycopy(strBuf, newBuf, strBuf.length); |
michael@0 | 744 | strBuf = newBuf; |
michael@0 | 745 | } |
michael@0 | 746 | strBuf[strBufLen++] = (char16_t) c; |
michael@0 | 747 | } |
michael@0 | 748 | |
michael@0 | 749 | void |
michael@0 | 750 | nsHtml5MetaScanner::handleAttributeValue() |
michael@0 | 751 | { |
michael@0 | 752 | if (metaState != NS_HTML5META_SCANNER_A) { |
michael@0 | 753 | return; |
michael@0 | 754 | } |
michael@0 | 755 | if (contentIndex == CONTENT.length && !content) { |
michael@0 | 756 | content = nsHtml5Portability::newStringFromBuffer(strBuf, 0, strBufLen); |
michael@0 | 757 | return; |
michael@0 | 758 | } |
michael@0 | 759 | if (charsetIndex == CHARSET.length && !charset) { |
michael@0 | 760 | charset = nsHtml5Portability::newStringFromBuffer(strBuf, 0, strBufLen); |
michael@0 | 761 | return; |
michael@0 | 762 | } |
michael@0 | 763 | if (httpEquivIndex == HTTP_EQUIV.length && httpEquivState == NS_HTML5META_SCANNER_HTTP_EQUIV_NOT_SEEN) { |
michael@0 | 764 | httpEquivState = (contentTypeIndex == CONTENT_TYPE.length) ? NS_HTML5META_SCANNER_HTTP_EQUIV_CONTENT_TYPE : NS_HTML5META_SCANNER_HTTP_EQUIV_OTHER; |
michael@0 | 765 | return; |
michael@0 | 766 | } |
michael@0 | 767 | } |
michael@0 | 768 | |
michael@0 | 769 | bool |
michael@0 | 770 | nsHtml5MetaScanner::handleTag() |
michael@0 | 771 | { |
michael@0 | 772 | bool stop = handleTagInner(); |
michael@0 | 773 | nsHtml5Portability::releaseString(content); |
michael@0 | 774 | content = nullptr; |
michael@0 | 775 | nsHtml5Portability::releaseString(charset); |
michael@0 | 776 | charset = nullptr; |
michael@0 | 777 | httpEquivState = NS_HTML5META_SCANNER_HTTP_EQUIV_NOT_SEEN; |
michael@0 | 778 | return stop; |
michael@0 | 779 | } |
michael@0 | 780 | |
michael@0 | 781 | bool |
michael@0 | 782 | nsHtml5MetaScanner::handleTagInner() |
michael@0 | 783 | { |
michael@0 | 784 | if (!!charset && tryCharset(charset)) { |
michael@0 | 785 | return true; |
michael@0 | 786 | } |
michael@0 | 787 | if (!!content && httpEquivState == NS_HTML5META_SCANNER_HTTP_EQUIV_CONTENT_TYPE) { |
michael@0 | 788 | nsString* extract = nsHtml5TreeBuilder::extractCharsetFromContent(content); |
michael@0 | 789 | if (!extract) { |
michael@0 | 790 | return false; |
michael@0 | 791 | } |
michael@0 | 792 | bool success = tryCharset(extract); |
michael@0 | 793 | nsHtml5Portability::releaseString(extract); |
michael@0 | 794 | return success; |
michael@0 | 795 | } |
michael@0 | 796 | return false; |
michael@0 | 797 | } |
michael@0 | 798 | |
michael@0 | 799 | void |
michael@0 | 800 | nsHtml5MetaScanner::initializeStatics() |
michael@0 | 801 | { |
michael@0 | 802 | } |
michael@0 | 803 | |
michael@0 | 804 | void |
michael@0 | 805 | nsHtml5MetaScanner::releaseStatics() |
michael@0 | 806 | { |
michael@0 | 807 | } |
michael@0 | 808 | |
michael@0 | 809 | |
michael@0 | 810 | #include "nsHtml5MetaScannerCppSupplement.h" |
michael@0 | 811 |