| OLD | NEW |
| (Empty) |
| 1 // Copyright 2014 PDFium Authors. All rights reserved. | |
| 2 // Use of this source code is governed by a BSD-style license that can be | |
| 3 // found in the LICENSE file. | |
| 4 | |
| 5 // Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com | |
| 6 | |
| 7 #include "xfa/fgas/xml/fgas_sax.h" | |
| 8 | |
| 9 #include <algorithm> | |
| 10 | |
| 11 #include "xfa/fxfa/include/xfa_checksum.h" | |
| 12 | |
| 13 namespace { | |
| 14 | |
| 15 const uint32_t kSaxFileBufSize = 32768; | |
| 16 | |
| 17 } // namespace | |
| 18 | |
| 19 CFX_SAXFile::CFX_SAXFile() | |
| 20 : m_pFile(NULL), | |
| 21 m_dwStart(0), | |
| 22 m_dwEnd(0), | |
| 23 m_dwCur(0), | |
| 24 m_pBuf(NULL), | |
| 25 m_dwBufSize(0), | |
| 26 m_dwBufIndex(0) {} | |
| 27 FX_BOOL CFX_SAXFile::StartFile(IFX_FileRead* pFile, | |
| 28 uint32_t dwStart, | |
| 29 uint32_t dwLen) { | |
| 30 ASSERT(m_pFile == NULL && pFile != NULL); | |
| 31 uint32_t dwSize = pFile->GetSize(); | |
| 32 if (dwStart >= dwSize) { | |
| 33 return FALSE; | |
| 34 } | |
| 35 if (dwLen == static_cast<uint32_t>(-1) || dwStart + dwLen > dwSize) { | |
| 36 dwLen = dwSize - dwStart; | |
| 37 } | |
| 38 if (dwLen == 0) { | |
| 39 return FALSE; | |
| 40 } | |
| 41 m_dwBufSize = std::min(dwLen, kSaxFileBufSize); | |
| 42 m_pBuf = FX_Alloc(uint8_t, m_dwBufSize); | |
| 43 if (!pFile->ReadBlock(m_pBuf, dwStart, m_dwBufSize)) { | |
| 44 return FALSE; | |
| 45 } | |
| 46 m_dwStart = dwStart; | |
| 47 m_dwEnd = dwStart + dwLen; | |
| 48 m_dwCur = dwStart; | |
| 49 m_pFile = pFile; | |
| 50 m_dwBufIndex = 0; | |
| 51 return TRUE; | |
| 52 } | |
| 53 FX_BOOL CFX_SAXFile::ReadNextBlock() { | |
| 54 ASSERT(m_pFile != NULL); | |
| 55 uint32_t dwSize = m_dwEnd - m_dwCur; | |
| 56 if (dwSize == 0) { | |
| 57 return FALSE; | |
| 58 } | |
| 59 m_dwBufSize = std::min(dwSize, kSaxFileBufSize); | |
| 60 if (!m_pFile->ReadBlock(m_pBuf, m_dwCur, m_dwBufSize)) { | |
| 61 return FALSE; | |
| 62 } | |
| 63 m_dwBufIndex = 0; | |
| 64 return TRUE; | |
| 65 } | |
| 66 void CFX_SAXFile::Reset() { | |
| 67 if (m_pBuf) { | |
| 68 FX_Free(m_pBuf); | |
| 69 m_pBuf = NULL; | |
| 70 } | |
| 71 m_pFile = NULL; | |
| 72 } | |
| 73 CFX_SAXReader::CFX_SAXReader() | |
| 74 : m_File(), | |
| 75 m_pHandler(nullptr), | |
| 76 m_iState(-1), | |
| 77 m_pRoot(nullptr), | |
| 78 m_pCurItem(nullptr), | |
| 79 m_dwItemID(0), | |
| 80 m_iDataSize(256), | |
| 81 m_iNameSize(256), | |
| 82 m_dwParseMode(0), | |
| 83 m_pCommentContext(nullptr) { | |
| 84 m_pszData = FX_Alloc(uint8_t, m_iDataSize); | |
| 85 m_pszName = FX_Alloc(uint8_t, m_iNameSize); | |
| 86 } | |
| 87 CFX_SAXReader::~CFX_SAXReader() { | |
| 88 Reset(); | |
| 89 if (m_pszData) { | |
| 90 FX_Free(m_pszData); | |
| 91 m_pszData = NULL; | |
| 92 } | |
| 93 if (m_pszName) { | |
| 94 FX_Free(m_pszName); | |
| 95 m_pszName = NULL; | |
| 96 } | |
| 97 } | |
| 98 void CFX_SAXReader::Reset() { | |
| 99 m_File.Reset(); | |
| 100 CFX_SAXItem* pItem = m_pRoot; | |
| 101 while (pItem) { | |
| 102 CFX_SAXItem* pNext = pItem->m_pNext; | |
| 103 delete pItem; | |
| 104 pItem = pNext; | |
| 105 } | |
| 106 m_pRoot = NULL; | |
| 107 m_pCurItem = NULL; | |
| 108 m_dwItemID = 0; | |
| 109 m_SkipStack.RemoveAll(); | |
| 110 m_SkipChar = 0; | |
| 111 m_iDataLength = 0; | |
| 112 m_iEntityStart = -1; | |
| 113 m_iNameLength = 0; | |
| 114 m_iDataPos = 0; | |
| 115 if (m_pCommentContext) { | |
| 116 delete m_pCommentContext; | |
| 117 m_pCommentContext = NULL; | |
| 118 } | |
| 119 } | |
| 120 inline void CFX_SAXReader::Push() { | |
| 121 CFX_SAXItem* pNew = new CFX_SAXItem; | |
| 122 pNew->m_dwID = ++m_dwItemID; | |
| 123 pNew->m_bSkip = m_pCurItem->m_bSkip; | |
| 124 pNew->m_pPrev = m_pCurItem; | |
| 125 m_pCurItem->m_pNext = pNew; | |
| 126 m_pCurItem = pNew; | |
| 127 } | |
| 128 inline void CFX_SAXReader::Pop() { | |
| 129 if (!m_pCurItem) { | |
| 130 return; | |
| 131 } | |
| 132 CFX_SAXItem* pPrev = m_pCurItem->m_pPrev; | |
| 133 pPrev->m_pNext = NULL; | |
| 134 delete m_pCurItem; | |
| 135 m_pCurItem = pPrev; | |
| 136 } | |
| 137 inline void CFX_SAXReader::AppendData(uint8_t ch) { | |
| 138 ReallocDataBuffer(); | |
| 139 m_pszData[m_iDataPos++] = ch; | |
| 140 } | |
| 141 inline void CFX_SAXReader::AppendName(uint8_t ch) { | |
| 142 ReallocNameBuffer(); | |
| 143 m_pszName[m_iDataPos++] = ch; | |
| 144 } | |
| 145 void CFX_SAXReader::ReallocDataBuffer() { | |
| 146 if (m_iDataPos < m_iDataSize) { | |
| 147 return; | |
| 148 } | |
| 149 if (m_iDataSize <= 1024 * 1024) { | |
| 150 m_iDataSize *= 2; | |
| 151 } else { | |
| 152 m_iDataSize += 1024 * 1024; | |
| 153 } | |
| 154 m_pszData = (uint8_t*)FX_Realloc(uint8_t, m_pszData, m_iDataSize); | |
| 155 } | |
| 156 void CFX_SAXReader::ReallocNameBuffer() { | |
| 157 if (m_iDataPos < m_iNameSize) { | |
| 158 return; | |
| 159 } | |
| 160 if (m_iNameSize <= 1024 * 1024) { | |
| 161 m_iNameSize *= 2; | |
| 162 } else { | |
| 163 m_iNameSize += 1024 * 1024; | |
| 164 } | |
| 165 m_pszName = (uint8_t*)FX_Realloc(uint8_t, m_pszName, m_iNameSize); | |
| 166 } | |
| 167 inline FX_BOOL CFX_SAXReader::SkipSpace(uint8_t ch) { | |
| 168 return (m_dwParseMode & FX_SAXPARSEMODE_NotSkipSpace) == 0 && ch < 0x21; | |
| 169 } | |
| 170 int32_t CFX_SAXReader::StartParse(IFX_FileRead* pFile, | |
| 171 uint32_t dwStart, | |
| 172 uint32_t dwLen, | |
| 173 uint32_t dwParseMode) { | |
| 174 m_iState = -1; | |
| 175 Reset(); | |
| 176 if (!m_File.StartFile(pFile, dwStart, dwLen)) { | |
| 177 return -1; | |
| 178 } | |
| 179 m_iState = 0; | |
| 180 m_eMode = FX_SAXMODE_Text; | |
| 181 m_ePrevMode = FX_SAXMODE_Text; | |
| 182 m_bCharData = FALSE; | |
| 183 m_dwDataOffset = 0; | |
| 184 m_pRoot = m_pCurItem = new CFX_SAXItem; | |
| 185 m_pCurItem->m_dwID = ++m_dwItemID; | |
| 186 m_dwParseMode = dwParseMode; | |
| 187 return 0; | |
| 188 } | |
| 189 typedef void (CFX_SAXReader::*FX_SAXReader_LPFParse)(); | |
| 190 static const FX_SAXReader_LPFParse g_FX_SAXReader_LPFParse[FX_SAXMODE_MAX] = { | |
| 191 &CFX_SAXReader::ParseText, | |
| 192 &CFX_SAXReader::ParseNodeStart, | |
| 193 &CFX_SAXReader::ParseDeclOrComment, | |
| 194 &CFX_SAXReader::ParseDeclNode, | |
| 195 &CFX_SAXReader::ParseComment, | |
| 196 &CFX_SAXReader::ParseCommentContent, | |
| 197 &CFX_SAXReader::ParseTagName, | |
| 198 &CFX_SAXReader::ParseTagAttributeName, | |
| 199 &CFX_SAXReader::ParseTagAttributeEqual, | |
| 200 &CFX_SAXReader::ParseTagAttributeValue, | |
| 201 &CFX_SAXReader::ParseMaybeClose, | |
| 202 &CFX_SAXReader::ParseTagClose, | |
| 203 &CFX_SAXReader::ParseTagEnd, | |
| 204 &CFX_SAXReader::ParseTargetData, | |
| 205 }; | |
| 206 int32_t CFX_SAXReader::ContinueParse(IFX_Pause* pPause) { | |
| 207 if (m_iState < 0 || m_iState > 99) { | |
| 208 return m_iState; | |
| 209 } | |
| 210 while (m_File.m_dwCur < m_File.m_dwEnd) { | |
| 211 uint32_t& index = m_File.m_dwBufIndex; | |
| 212 uint32_t size = m_File.m_dwBufSize; | |
| 213 const uint8_t* pBuf = m_File.m_pBuf; | |
| 214 while (index < size) { | |
| 215 m_CurByte = pBuf[index]; | |
| 216 (this->*g_FX_SAXReader_LPFParse[m_eMode])(); | |
| 217 index++; | |
| 218 } | |
| 219 m_File.m_dwCur += index; | |
| 220 m_iState = (m_File.m_dwCur - m_File.m_dwStart) * 100 / | |
| 221 (m_File.m_dwEnd - m_File.m_dwStart); | |
| 222 if (m_File.m_dwCur >= m_File.m_dwEnd) { | |
| 223 break; | |
| 224 } | |
| 225 if (!m_File.ReadNextBlock()) { | |
| 226 m_iState = -2; | |
| 227 break; | |
| 228 } | |
| 229 m_dwDataOffset = 0; | |
| 230 if (pPause && pPause->NeedToPauseNow()) { | |
| 231 break; | |
| 232 } | |
| 233 } | |
| 234 return m_iState; | |
| 235 } | |
| 236 void CFX_SAXReader::ParseChar(uint8_t ch) { | |
| 237 ReallocDataBuffer(); | |
| 238 m_pszData[m_iDataPos] = ch; | |
| 239 if (m_iEntityStart > -1 && ch == ';') { | |
| 240 int32_t iSaveEntityStart = m_iEntityStart; | |
| 241 CFX_ByteString csEntity(m_pszData + m_iEntityStart + 1, | |
| 242 m_iDataPos - m_iEntityStart - 1); | |
| 243 int32_t iLen = csEntity.GetLength(); | |
| 244 if (iLen > 0) { | |
| 245 if (csEntity[0] == '#') { | |
| 246 if ((m_dwParseMode & FX_SAXPARSEMODE_NotConvert_sharp) == 0) { | |
| 247 ch = 0; | |
| 248 uint8_t w; | |
| 249 if (iLen > 1 && csEntity[1] == 'x') { | |
| 250 for (int32_t i = 2; i < iLen; i++) { | |
| 251 w = csEntity[i]; | |
| 252 if (w >= '0' && w <= '9') { | |
| 253 ch = (ch << 4) + w - '0'; | |
| 254 } else if (w >= 'A' && w <= 'F') { | |
| 255 ch = (ch << 4) + w - 55; | |
| 256 } else if (w >= 'a' && w <= 'f') { | |
| 257 ch = (ch << 4) + w - 87; | |
| 258 } else { | |
| 259 break; | |
| 260 } | |
| 261 } | |
| 262 } else { | |
| 263 for (int32_t i = 1; i < iLen; i++) { | |
| 264 w = csEntity[i]; | |
| 265 if (w < '0' || w > '9') { | |
| 266 break; | |
| 267 } | |
| 268 ch = ch * 10 + w - '0'; | |
| 269 } | |
| 270 } | |
| 271 if (ch != 0) { | |
| 272 m_pszData[m_iEntityStart++] = ch; | |
| 273 } | |
| 274 } | |
| 275 } else { | |
| 276 if (csEntity.Compare("amp") == 0) { | |
| 277 if ((m_dwParseMode & FX_SAXPARSEMODE_NotConvert_amp) == 0) { | |
| 278 m_pszData[m_iEntityStart++] = '&'; | |
| 279 } | |
| 280 } else if (csEntity.Compare("lt") == 0) { | |
| 281 if ((m_dwParseMode & FX_SAXPARSEMODE_NotConvert_lt) == 0) { | |
| 282 m_pszData[m_iEntityStart++] = '<'; | |
| 283 } | |
| 284 } else if (csEntity.Compare("gt") == 0) { | |
| 285 if ((m_dwParseMode & FX_SAXPARSEMODE_NotConvert_gt) == 0) { | |
| 286 m_pszData[m_iEntityStart++] = '>'; | |
| 287 } | |
| 288 } else if (csEntity.Compare("apos") == 0) { | |
| 289 if ((m_dwParseMode & FX_SAXPARSEMODE_NotConvert_apos) == 0) { | |
| 290 m_pszData[m_iEntityStart++] = '\''; | |
| 291 } | |
| 292 } else if (csEntity.Compare("quot") == 0) { | |
| 293 if ((m_dwParseMode & FX_SAXPARSEMODE_NotConvert_quot) == 0) { | |
| 294 m_pszData[m_iEntityStart++] = '\"'; | |
| 295 } | |
| 296 } | |
| 297 } | |
| 298 } | |
| 299 if (iSaveEntityStart != m_iEntityStart) { | |
| 300 m_iDataPos = m_iEntityStart; | |
| 301 m_iEntityStart = -1; | |
| 302 } else { | |
| 303 m_iDataPos++; | |
| 304 m_iEntityStart = -1; | |
| 305 } | |
| 306 } else { | |
| 307 if (m_iEntityStart < 0 && ch == '&') { | |
| 308 m_iEntityStart = m_iDataPos; | |
| 309 } | |
| 310 m_iDataPos++; | |
| 311 } | |
| 312 } | |
| 313 void CFX_SAXReader::ParseText() { | |
| 314 if (m_CurByte == '<') { | |
| 315 if (m_iDataPos > 0) { | |
| 316 m_iDataLength = m_iDataPos; | |
| 317 m_iDataPos = 0; | |
| 318 if (m_pHandler) { | |
| 319 NotifyData(); | |
| 320 } | |
| 321 } | |
| 322 Push(); | |
| 323 m_dwNodePos = m_File.m_dwCur + m_File.m_dwBufIndex; | |
| 324 m_eMode = FX_SAXMODE_NodeStart; | |
| 325 return; | |
| 326 } | |
| 327 if (m_iDataPos < 1 && SkipSpace(m_CurByte)) { | |
| 328 return; | |
| 329 } | |
| 330 ParseChar(m_CurByte); | |
| 331 } | |
| 332 void CFX_SAXReader::ParseNodeStart() { | |
| 333 if (m_CurByte == '?') { | |
| 334 m_pCurItem->m_eNode = FX_SAXNODE_Instruction; | |
| 335 m_eMode = FX_SAXMODE_TagName; | |
| 336 return; | |
| 337 } | |
| 338 if (m_CurByte == '!') { | |
| 339 m_eMode = FX_SAXMODE_DeclOrComment; | |
| 340 return; | |
| 341 } | |
| 342 if (m_CurByte == '/') { | |
| 343 m_eMode = FX_SAXMODE_TagEnd; | |
| 344 return; | |
| 345 } | |
| 346 if (m_CurByte == '>') { | |
| 347 Pop(); | |
| 348 m_eMode = FX_SAXMODE_Text; | |
| 349 return; | |
| 350 } | |
| 351 if (m_CurByte > 0x20) { | |
| 352 m_dwDataOffset = m_File.m_dwBufIndex; | |
| 353 m_pCurItem->m_eNode = FX_SAXNODE_Tag; | |
| 354 m_eMode = FX_SAXMODE_TagName; | |
| 355 AppendData(m_CurByte); | |
| 356 } | |
| 357 } | |
| 358 void CFX_SAXReader::ParseDeclOrComment() { | |
| 359 if (m_CurByte == '-') { | |
| 360 m_eMode = FX_SAXMODE_Comment; | |
| 361 m_pCurItem->m_eNode = FX_SAXNODE_Comment; | |
| 362 if (m_pCommentContext == NULL) { | |
| 363 m_pCommentContext = new CFX_SAXCommentContext; | |
| 364 } | |
| 365 m_pCommentContext->m_iHeaderCount = 1; | |
| 366 m_pCommentContext->m_iTailCount = 0; | |
| 367 } else { | |
| 368 m_eMode = FX_SAXMODE_DeclNode; | |
| 369 m_dwDataOffset = m_File.m_dwBufIndex; | |
| 370 m_SkipChar = '>'; | |
| 371 m_SkipStack.Add('>'); | |
| 372 SkipNode(); | |
| 373 } | |
| 374 } | |
| 375 void CFX_SAXReader::ParseComment() { | |
| 376 m_pCommentContext->m_iHeaderCount = 2; | |
| 377 m_dwNodePos = m_File.m_dwCur + m_File.m_dwBufIndex; | |
| 378 m_eMode = FX_SAXMODE_CommentContent; | |
| 379 } | |
| 380 void CFX_SAXReader::ParseCommentContent() { | |
| 381 if (m_CurByte == '-') { | |
| 382 m_pCommentContext->m_iTailCount++; | |
| 383 } else if (m_CurByte == '>' && m_pCommentContext->m_iTailCount == 2) { | |
| 384 m_iDataLength = m_iDataPos; | |
| 385 m_iDataPos = 0; | |
| 386 if (m_pHandler) { | |
| 387 NotifyTargetData(); | |
| 388 } | |
| 389 Pop(); | |
| 390 m_eMode = FX_SAXMODE_Text; | |
| 391 } else { | |
| 392 while (m_pCommentContext->m_iTailCount > 0) { | |
| 393 AppendData('-'); | |
| 394 m_pCommentContext->m_iTailCount--; | |
| 395 } | |
| 396 AppendData(m_CurByte); | |
| 397 } | |
| 398 } | |
| 399 void CFX_SAXReader::ParseDeclNode() { | |
| 400 SkipNode(); | |
| 401 } | |
| 402 void CFX_SAXReader::ParseTagName() { | |
| 403 if (m_CurByte < 0x21 || m_CurByte == '/' || m_CurByte == '>' || | |
| 404 m_CurByte == '?') { | |
| 405 m_iDataLength = m_iDataPos; | |
| 406 m_iDataPos = 0; | |
| 407 if (m_pHandler) { | |
| 408 NotifyEnter(); | |
| 409 } | |
| 410 if (m_CurByte < 0x21) { | |
| 411 m_eMode = FX_SAXMODE_TagAttributeName; | |
| 412 } else if (m_CurByte == '/' || m_CurByte == '?') { | |
| 413 m_ePrevMode = m_eMode; | |
| 414 m_eMode = FX_SAXMODE_TagMaybeClose; | |
| 415 } else { | |
| 416 if (m_pHandler) { | |
| 417 NotifyBreak(); | |
| 418 } | |
| 419 m_eMode = FX_SAXMODE_Text; | |
| 420 } | |
| 421 } else { | |
| 422 AppendData(m_CurByte); | |
| 423 } | |
| 424 } | |
| 425 void CFX_SAXReader::ParseTagAttributeName() { | |
| 426 if (m_CurByte < 0x21 || m_CurByte == '=') { | |
| 427 if (m_iDataPos < 1 && m_CurByte < 0x21) { | |
| 428 return; | |
| 429 } | |
| 430 m_iNameLength = m_iDataPos; | |
| 431 m_iDataPos = 0; | |
| 432 m_SkipChar = 0; | |
| 433 m_eMode = m_CurByte == '=' ? FX_SAXMODE_TagAttributeValue | |
| 434 : FX_SAXMODE_TagAttributeEqual; | |
| 435 return; | |
| 436 } | |
| 437 if (m_CurByte == '/' || m_CurByte == '>' || m_CurByte == '?') { | |
| 438 if (m_CurByte == '/' || m_CurByte == '?') { | |
| 439 m_ePrevMode = m_eMode; | |
| 440 m_eMode = FX_SAXMODE_TagMaybeClose; | |
| 441 } else { | |
| 442 if (m_pHandler) { | |
| 443 NotifyBreak(); | |
| 444 } | |
| 445 m_eMode = FX_SAXMODE_Text; | |
| 446 } | |
| 447 return; | |
| 448 } | |
| 449 if (m_iDataPos < 1) { | |
| 450 m_dwDataOffset = m_File.m_dwBufIndex; | |
| 451 } | |
| 452 AppendName(m_CurByte); | |
| 453 } | |
| 454 void CFX_SAXReader::ParseTagAttributeEqual() { | |
| 455 if (m_CurByte == '=') { | |
| 456 m_SkipChar = 0; | |
| 457 m_eMode = FX_SAXMODE_TagAttributeValue; | |
| 458 return; | |
| 459 } else if (m_pCurItem->m_eNode == FX_SAXNODE_Instruction) { | |
| 460 m_iDataPos = m_iNameLength; | |
| 461 AppendName(0x20); | |
| 462 m_eMode = FX_SAXMODE_TargetData; | |
| 463 ParseTargetData(); | |
| 464 } | |
| 465 } | |
| 466 void CFX_SAXReader::ParseTagAttributeValue() { | |
| 467 if (m_SkipChar) { | |
| 468 if (m_SkipChar == m_CurByte) { | |
| 469 { | |
| 470 m_iDataLength = m_iDataPos; | |
| 471 m_iDataPos = 0; | |
| 472 if (m_pHandler) { | |
| 473 NotifyAttribute(); | |
| 474 } | |
| 475 } | |
| 476 m_SkipChar = 0; | |
| 477 m_eMode = FX_SAXMODE_TagAttributeName; | |
| 478 return; | |
| 479 } | |
| 480 ParseChar(m_CurByte); | |
| 481 return; | |
| 482 } | |
| 483 if (m_CurByte < 0x21) { | |
| 484 return; | |
| 485 } | |
| 486 if (m_iDataPos < 1) { | |
| 487 if (m_CurByte == '\'' || m_CurByte == '\"') { | |
| 488 m_SkipChar = m_CurByte; | |
| 489 } | |
| 490 } | |
| 491 } | |
| 492 void CFX_SAXReader::ParseMaybeClose() { | |
| 493 if (m_CurByte == '>') { | |
| 494 if (m_pCurItem->m_eNode == FX_SAXNODE_Instruction) { | |
| 495 m_iNameLength = m_iDataPos; | |
| 496 m_iDataPos = 0; | |
| 497 if (m_pHandler) { | |
| 498 NotifyTargetData(); | |
| 499 } | |
| 500 } | |
| 501 ParseTagClose(); | |
| 502 m_eMode = FX_SAXMODE_Text; | |
| 503 } else if (m_ePrevMode == FX_SAXMODE_TagName) { | |
| 504 AppendData('/'); | |
| 505 m_eMode = FX_SAXMODE_TagName; | |
| 506 m_ePrevMode = FX_SAXMODE_Text; | |
| 507 ParseTagName(); | |
| 508 } else if (m_ePrevMode == FX_SAXMODE_TagAttributeName) { | |
| 509 AppendName('/'); | |
| 510 m_eMode = FX_SAXMODE_TagAttributeName; | |
| 511 m_ePrevMode = FX_SAXMODE_Text; | |
| 512 ParseTagAttributeName(); | |
| 513 } else if (m_ePrevMode == FX_SAXMODE_TargetData) { | |
| 514 AppendName('?'); | |
| 515 m_eMode = FX_SAXMODE_TargetData; | |
| 516 m_ePrevMode = FX_SAXMODE_Text; | |
| 517 ParseTargetData(); | |
| 518 } | |
| 519 } | |
| 520 void CFX_SAXReader::ParseTagClose() { | |
| 521 m_dwNodePos = m_File.m_dwCur + m_File.m_dwBufIndex; | |
| 522 if (m_pHandler) { | |
| 523 NotifyClose(); | |
| 524 } | |
| 525 Pop(); | |
| 526 } | |
| 527 void CFX_SAXReader::ParseTagEnd() { | |
| 528 if (m_CurByte < 0x21) { | |
| 529 return; | |
| 530 } | |
| 531 if (m_CurByte == '>') { | |
| 532 Pop(); | |
| 533 m_dwNodePos = m_File.m_dwCur + m_File.m_dwBufIndex; | |
| 534 m_iDataLength = m_iDataPos; | |
| 535 m_iDataPos = 0; | |
| 536 if (m_pHandler) { | |
| 537 NotifyEnd(); | |
| 538 } | |
| 539 Pop(); | |
| 540 m_eMode = FX_SAXMODE_Text; | |
| 541 } else { | |
| 542 ParseChar(m_CurByte); | |
| 543 } | |
| 544 } | |
| 545 void CFX_SAXReader::ParseTargetData() { | |
| 546 if (m_CurByte == '?') { | |
| 547 m_ePrevMode = m_eMode; | |
| 548 m_eMode = FX_SAXMODE_TagMaybeClose; | |
| 549 } else { | |
| 550 AppendName(m_CurByte); | |
| 551 } | |
| 552 } | |
| 553 void CFX_SAXReader::SkipNode() { | |
| 554 int32_t iLen = m_SkipStack.GetSize(); | |
| 555 if (m_SkipChar == '\'' || m_SkipChar == '\"') { | |
| 556 if (m_CurByte != m_SkipChar) { | |
| 557 return; | |
| 558 } | |
| 559 iLen--; | |
| 560 ASSERT(iLen > -1); | |
| 561 m_SkipStack.RemoveAt(iLen, 1); | |
| 562 m_SkipChar = iLen ? m_SkipStack[iLen - 1] : 0; | |
| 563 return; | |
| 564 } | |
| 565 switch (m_CurByte) { | |
| 566 case '<': | |
| 567 m_SkipChar = '>'; | |
| 568 m_SkipStack.Add('>'); | |
| 569 break; | |
| 570 case '[': | |
| 571 m_SkipChar = ']'; | |
| 572 m_SkipStack.Add(']'); | |
| 573 break; | |
| 574 case '(': | |
| 575 m_SkipChar = ')'; | |
| 576 m_SkipStack.Add(')'); | |
| 577 break; | |
| 578 case '\'': | |
| 579 m_SkipChar = '\''; | |
| 580 m_SkipStack.Add('\''); | |
| 581 break; | |
| 582 case '\"': | |
| 583 m_SkipChar = '\"'; | |
| 584 m_SkipStack.Add('\"'); | |
| 585 break; | |
| 586 default: | |
| 587 if (m_CurByte == m_SkipChar) { | |
| 588 iLen--; | |
| 589 m_SkipStack.RemoveAt(iLen, 1); | |
| 590 m_SkipChar = iLen ? m_SkipStack[iLen - 1] : 0; | |
| 591 if (iLen == 0 && m_CurByte == '>') { | |
| 592 m_iDataLength = m_iDataPos; | |
| 593 m_iDataPos = 0; | |
| 594 if (m_iDataLength >= 9 && | |
| 595 FXSYS_memcmp(m_pszData, "[CDATA[", 7 * sizeof(uint8_t)) == 0 && | |
| 596 FXSYS_memcmp(m_pszData + m_iDataLength - 2, "]]", | |
| 597 2 * sizeof(uint8_t)) == 0) { | |
| 598 Pop(); | |
| 599 m_iDataLength -= 9; | |
| 600 m_dwDataOffset += 7; | |
| 601 FXSYS_memmove(m_pszData, m_pszData + 7, | |
| 602 m_iDataLength * sizeof(uint8_t)); | |
| 603 m_bCharData = TRUE; | |
| 604 if (m_pHandler) { | |
| 605 NotifyData(); | |
| 606 } | |
| 607 m_bCharData = FALSE; | |
| 608 } else { | |
| 609 Pop(); | |
| 610 } | |
| 611 m_eMode = FX_SAXMODE_Text; | |
| 612 } | |
| 613 } | |
| 614 break; | |
| 615 } | |
| 616 if (iLen > 0) { | |
| 617 ParseChar(m_CurByte); | |
| 618 } | |
| 619 } | |
| 620 | |
| 621 void CFX_SAXReader::NotifyData() { | |
| 622 if (m_pCurItem->m_eNode == FX_SAXNODE_Tag) | |
| 623 m_pHandler->OnTagData(m_pCurItem->m_pNode, | |
| 624 m_bCharData ? FX_SAXNODE_CharData : FX_SAXNODE_Text, | |
| 625 CFX_ByteStringC(m_pszData, m_iDataLength), | |
| 626 m_File.m_dwCur + m_dwDataOffset); | |
| 627 } | |
| 628 | |
| 629 void CFX_SAXReader::NotifyEnter() { | |
| 630 if (m_pCurItem->m_eNode == FX_SAXNODE_Tag || | |
| 631 m_pCurItem->m_eNode == FX_SAXNODE_Instruction) { | |
| 632 m_pCurItem->m_pNode = | |
| 633 m_pHandler->OnTagEnter(CFX_ByteStringC(m_pszData, m_iDataLength), | |
| 634 m_pCurItem->m_eNode, m_dwNodePos); | |
| 635 } | |
| 636 } | |
| 637 | |
| 638 void CFX_SAXReader::NotifyAttribute() { | |
| 639 if (m_pCurItem->m_eNode == FX_SAXNODE_Tag || | |
| 640 m_pCurItem->m_eNode == FX_SAXNODE_Instruction) { | |
| 641 m_pHandler->OnTagAttribute(m_pCurItem->m_pNode, | |
| 642 CFX_ByteStringC(m_pszName, m_iNameLength), | |
| 643 CFX_ByteStringC(m_pszData, m_iDataLength)); | |
| 644 } | |
| 645 } | |
| 646 | |
| 647 void CFX_SAXReader::NotifyBreak() { | |
| 648 if (m_pCurItem->m_eNode == FX_SAXNODE_Tag) | |
| 649 m_pHandler->OnTagBreak(m_pCurItem->m_pNode); | |
| 650 } | |
| 651 | |
| 652 void CFX_SAXReader::NotifyClose() { | |
| 653 if (m_pCurItem->m_eNode == FX_SAXNODE_Tag || | |
| 654 m_pCurItem->m_eNode == FX_SAXNODE_Instruction) { | |
| 655 m_pHandler->OnTagClose(m_pCurItem->m_pNode, m_dwNodePos); | |
| 656 } | |
| 657 } | |
| 658 | |
| 659 void CFX_SAXReader::NotifyEnd() { | |
| 660 if (m_pCurItem->m_eNode != FX_SAXNODE_Tag) | |
| 661 return; | |
| 662 | |
| 663 m_pHandler->OnTagEnd(m_pCurItem->m_pNode, | |
| 664 CFX_ByteStringC(m_pszData, m_iDataLength), m_dwNodePos); | |
| 665 } | |
| 666 | |
| 667 void CFX_SAXReader::NotifyTargetData() { | |
| 668 if (m_pCurItem->m_eNode == FX_SAXNODE_Instruction) { | |
| 669 m_pHandler->OnTargetData(m_pCurItem->m_pNode, m_pCurItem->m_eNode, | |
| 670 CFX_ByteStringC(m_pszName, m_iNameLength), | |
| 671 m_dwNodePos); | |
| 672 } else if (m_pCurItem->m_eNode == FX_SAXNODE_Comment) { | |
| 673 m_pHandler->OnTargetData(m_pCurItem->m_pNode, m_pCurItem->m_eNode, | |
| 674 CFX_ByteStringC(m_pszData, m_iDataLength), | |
| 675 m_dwNodePos); | |
| 676 } | |
| 677 } | |
| 678 | |
| 679 void CFX_SAXReader::SkipCurrentNode() { | |
| 680 if (!m_pCurItem) | |
| 681 return; | |
| 682 | |
| 683 m_pCurItem->m_bSkip = TRUE; | |
| 684 } | |
| 685 | |
| 686 void CFX_SAXReader::SetHandler(CXFA_SAXReaderHandler* pHandler) { | |
| 687 m_pHandler = pHandler; | |
| 688 } | |
| OLD | NEW |