| OLD | NEW |
|---|
| (Empty) |
| 1 // Copyright 2016 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 "core/fpdfapi/fpdf_parser/cpdf_data_avail.h" | |
| 8 | |
| 9 #include <algorithm> | |
| 10 #include <memory> | |
| 11 #include <utility> | |
| 12 | |
| 13 #include "core/fpdfapi/cpdf_modulemgr.h" | |
| 14 #include "core/fpdfapi/fpdf_parser/cpdf_array.h" | |
| 15 #include "core/fpdfapi/fpdf_parser/cpdf_dictionary.h" | |
| 16 #include "core/fpdfapi/fpdf_parser/cpdf_document.h" | |
| 17 #include "core/fpdfapi/fpdf_parser/cpdf_hint_tables.h" | |
| 18 #include "core/fpdfapi/fpdf_parser/cpdf_name.h" | |
| 19 #include "core/fpdfapi/fpdf_parser/cpdf_number.h" | |
| 20 #include "core/fpdfapi/fpdf_parser/cpdf_reference.h" | |
| 21 #include "core/fpdfapi/fpdf_parser/cpdf_stream.h" | |
| 22 #include "core/fpdfapi/fpdf_parser/fpdf_parser_utility.h" | |
| 23 #include "core/fxcrt/fx_ext.h" | |
| 24 #include "core/fxcrt/fx_safe_types.h" | |
| 25 #include "third_party/base/stl_util.h" | |
| 26 | |
| 27 CPDF_DataAvail::FileAvail::~FileAvail() {} | |
| 28 | |
| 29 CPDF_DataAvail::DownloadHints::~DownloadHints() {} | |
| 30 | |
| 31 // static | |
| 32 int CPDF_DataAvail::s_CurrentDataAvailRecursionDepth = 0; | |
| 33 | |
| 34 CPDF_DataAvail::CPDF_DataAvail(FileAvail* pFileAvail, | |
| 35 IFX_FileRead* pFileRead, | |
| 36 FX_BOOL bSupportHintTable) | |
| 37 : m_pFileAvail(pFileAvail), m_pFileRead(pFileRead) { | |
| 38 m_Pos = 0; | |
| 39 m_dwFileLen = 0; | |
| 40 if (m_pFileRead) { | |
| 41 m_dwFileLen = (uint32_t)m_pFileRead->GetSize(); | |
| 42 } | |
| 43 m_dwCurrentOffset = 0; | |
| 44 m_dwXRefOffset = 0; | |
| 45 m_bufferOffset = 0; | |
| 46 m_dwFirstPageNo = 0; | |
| 47 m_bufferSize = 0; | |
| 48 m_PagesObjNum = 0; | |
| 49 m_dwCurrentXRefSteam = 0; | |
| 50 m_dwAcroFormObjNum = 0; | |
| 51 m_dwInfoObjNum = 0; | |
| 52 m_pDocument = 0; | |
| 53 m_dwEncryptObjNum = 0; | |
| 54 m_dwPrevXRefOffset = 0; | |
| 55 m_dwLastXRefOffset = 0; | |
| 56 m_bDocAvail = FALSE; | |
| 57 m_bMainXRefLoadTried = FALSE; | |
| 58 m_bDocAvail = FALSE; | |
| 59 m_bLinearized = FALSE; | |
| 60 m_bPagesLoad = FALSE; | |
| 61 m_bPagesTreeLoad = FALSE; | |
| 62 m_bMainXRefLoadedOK = FALSE; | |
| 63 m_bAnnotsLoad = FALSE; | |
| 64 m_bHaveAcroForm = FALSE; | |
| 65 m_bAcroFormLoad = FALSE; | |
| 66 m_bPageLoadedOK = FALSE; | |
| 67 m_bNeedDownLoadResource = FALSE; | |
| 68 m_bLinearizedFormParamLoad = FALSE; | |
| 69 m_pLinearized = nullptr; | |
| 70 m_pRoot = nullptr; | |
| 71 m_pTrailer = nullptr; | |
| 72 m_pCurrentParser = nullptr; | |
| 73 m_pAcroForm = nullptr; | |
| 74 m_pPageDict = nullptr; | |
| 75 m_pPageResource = nullptr; | |
| 76 m_docStatus = PDF_DATAAVAIL_HEADER; | |
| 77 m_parser.m_bOwnFileRead = false; | |
| 78 m_bTotalLoadPageTree = FALSE; | |
| 79 m_bCurPageDictLoadOK = FALSE; | |
| 80 m_bLinearedDataOK = FALSE; | |
| 81 m_bSupportHintTable = bSupportHintTable; | |
| 82 } | |
| 83 CPDF_DataAvail::~CPDF_DataAvail() { | |
| 84 m_pHintTables.reset(); | |
| 85 if (m_pLinearized) | |
| 86 m_pLinearized->Release(); | |
| 87 | |
| 88 if (m_pRoot) | |
| 89 m_pRoot->Release(); | |
| 90 | |
| 91 if (m_pTrailer) | |
| 92 m_pTrailer->Release(); | |
| 93 | |
| 94 int iSize = m_arrayAcroforms.GetSize(); | |
| 95 for (int i = 0; i < iSize; ++i) | |
| 96 m_arrayAcroforms.GetAt(i)->Release(); | |
| 97 } | |
| 98 | |
| 99 void CPDF_DataAvail::SetDocument(CPDF_Document* pDoc) { | |
| 100 m_pDocument = pDoc; | |
| 101 } | |
| 102 | |
| 103 uint32_t CPDF_DataAvail::GetObjectSize(uint32_t objnum, FX_FILESIZE& offset) { | |
| 104 CPDF_Parser* pParser = m_pDocument->GetParser(); | |
| 105 if (!pParser || !pParser->IsValidObjectNumber(objnum)) | |
| 106 return 0; | |
| 107 | |
| 108 if (pParser->GetObjectType(objnum) == 2) | |
| 109 objnum = pParser->GetObjectPositionOrZero(objnum); | |
| 110 | |
| 111 if (pParser->GetObjectType(objnum) != 1 && | |
| 112 pParser->GetObjectType(objnum) != 255) { | |
| 113 return 0; | |
| 114 } | |
| 115 | |
| 116 offset = pParser->GetObjectPositionOrZero(objnum); | |
| 117 if (offset == 0) | |
| 118 return 0; | |
| 119 | |
| 120 auto it = pParser->m_SortedOffset.find(offset); | |
| 121 if (it == pParser->m_SortedOffset.end() || | |
| 122 ++it == pParser->m_SortedOffset.end()) { | |
| 123 return 0; | |
| 124 } | |
| 125 return *it - offset; | |
| 126 } | |
| 127 | |
| 128 FX_BOOL CPDF_DataAvail::IsObjectsAvail( | |
| 129 CFX_ArrayTemplate<CPDF_Object*>& obj_array, | |
| 130 FX_BOOL bParsePage, | |
| 131 DownloadHints* pHints, | |
| 132 CFX_ArrayTemplate<CPDF_Object*>& ret_array) { | |
| 133 if (!obj_array.GetSize()) | |
| 134 return TRUE; | |
| 135 | |
| 136 uint32_t count = 0; | |
| 137 CFX_ArrayTemplate<CPDF_Object*> new_obj_array; | |
| 138 for (int i = 0; i < obj_array.GetSize(); i++) { | |
| 139 CPDF_Object* pObj = obj_array[i]; | |
| 140 if (!pObj) | |
| 141 continue; | |
| 142 | |
| 143 int32_t type = pObj->GetType(); | |
| 144 switch (type) { | |
| 145 case CPDF_Object::ARRAY: { | |
| 146 CPDF_Array* pArray = pObj->AsArray(); | |
| 147 for (size_t k = 0; k < pArray->GetCount(); ++k) | |
| 148 new_obj_array.Add(pArray->GetObjectAt(k)); | |
| 149 } break; | |
| 150 case CPDF_Object::STREAM: | |
| 151 pObj = pObj->GetDict(); | |
| 152 case CPDF_Object::DICTIONARY: { | |
| 153 CPDF_Dictionary* pDict = pObj->GetDict(); | |
| 154 if (pDict && pDict->GetStringFor("Type") == "Page" && !bParsePage) | |
| 155 continue; | |
| 156 | |
| 157 for (const auto& it : *pDict) { | |
| 158 const CFX_ByteString& key = it.first; | |
| 159 CPDF_Object* value = it.second; | |
| 160 if (key != "Parent") | |
| 161 new_obj_array.Add(value); | |
| 162 } | |
| 163 } break; | |
| 164 case CPDF_Object::REFERENCE: { | |
| 165 CPDF_Reference* pRef = pObj->AsReference(); | |
| 166 uint32_t dwNum = pRef->GetRefObjNum(); | |
| 167 | |
| 168 FX_FILESIZE offset; | |
| 169 uint32_t size = GetObjectSize(dwNum, offset); | |
| 170 if (size == 0 || offset < 0 || offset >= m_dwFileLen) | |
| 171 break; | |
| 172 | |
| 173 if (!IsDataAvail(offset, size, pHints)) { | |
| 174 ret_array.Add(pObj); | |
| 175 count++; | |
| 176 } else if (!pdfium::ContainsKey(m_ObjectSet, dwNum)) { | |
| 177 m_ObjectSet.insert(dwNum); | |
| 178 CPDF_Object* pReferred = | |
| 179 m_pDocument->GetOrParseIndirectObject(pRef->GetRefObjNum()); | |
| 180 if (pReferred) | |
| 181 new_obj_array.Add(pReferred); | |
| 182 } | |
| 183 } break; | |
| 184 } | |
| 185 } | |
| 186 | |
| 187 if (count > 0) { | |
| 188 for (int i = 0; i < new_obj_array.GetSize(); ++i) { | |
| 189 CPDF_Object* pObj = new_obj_array[i]; | |
| 190 if (CPDF_Reference* pRef = pObj->AsReference()) { | |
| 191 uint32_t dwNum = pRef->GetRefObjNum(); | |
| 192 if (!pdfium::ContainsKey(m_ObjectSet, dwNum)) | |
| 193 ret_array.Add(pObj); | |
| 194 } else { | |
| 195 ret_array.Add(pObj); | |
| 196 } | |
| 197 } | |
| 198 return FALSE; | |
| 199 } | |
| 200 | |
| 201 obj_array.RemoveAll(); | |
| 202 obj_array.Append(new_obj_array); | |
| 203 return IsObjectsAvail(obj_array, FALSE, pHints, ret_array); | |
| 204 } | |
| 205 | |
| 206 CPDF_DataAvail::DocAvailStatus CPDF_DataAvail::IsDocAvail( | |
| 207 DownloadHints* pHints) { | |
| 208 if (!m_dwFileLen && m_pFileRead) { | |
| 209 m_dwFileLen = (uint32_t)m_pFileRead->GetSize(); | |
| 210 if (!m_dwFileLen) | |
| 211 return DataError; | |
| 212 } | |
| 213 | |
| 214 while (!m_bDocAvail) { | |
| 215 if (!CheckDocStatus(pHints)) | |
| 216 return DataNotAvailable; | |
| 217 } | |
| 218 | |
| 219 return DataAvailable; | |
| 220 } | |
| 221 | |
| 222 FX_BOOL CPDF_DataAvail::CheckAcroFormSubObject(DownloadHints* pHints) { | |
| 223 if (!m_objs_array.GetSize()) { | |
| 224 m_objs_array.RemoveAll(); | |
| 225 m_ObjectSet.clear(); | |
| 226 CFX_ArrayTemplate<CPDF_Object*> obj_array; | |
| 227 obj_array.Append(m_arrayAcroforms); | |
| 228 FX_BOOL bRet = IsObjectsAvail(obj_array, FALSE, pHints, m_objs_array); | |
| 229 if (bRet) | |
| 230 m_objs_array.RemoveAll(); | |
| 231 return bRet; | |
| 232 } | |
| 233 | |
| 234 CFX_ArrayTemplate<CPDF_Object*> new_objs_array; | |
| 235 FX_BOOL bRet = IsObjectsAvail(m_objs_array, FALSE, pHints, new_objs_array); | |
| 236 if (bRet) { | |
| 237 int32_t iSize = m_arrayAcroforms.GetSize(); | |
| 238 for (int32_t i = 0; i < iSize; ++i) { | |
| 239 m_arrayAcroforms.GetAt(i)->Release(); | |
| 240 } | |
| 241 m_arrayAcroforms.RemoveAll(); | |
| 242 } else { | |
| 243 m_objs_array.RemoveAll(); | |
| 244 m_objs_array.Append(new_objs_array); | |
| 245 } | |
| 246 return bRet; | |
| 247 } | |
| 248 | |
| 249 FX_BOOL CPDF_DataAvail::CheckAcroForm(DownloadHints* pHints) { | |
| 250 FX_BOOL bExist = FALSE; | |
| 251 m_pAcroForm = GetObject(m_dwAcroFormObjNum, pHints, &bExist); | |
| 252 if (!bExist) { | |
| 253 m_docStatus = PDF_DATAAVAIL_PAGETREE; | |
| 254 return TRUE; | |
| 255 } | |
| 256 | |
| 257 if (!m_pAcroForm) { | |
| 258 if (m_docStatus == PDF_DATAAVAIL_ERROR) { | |
| 259 m_docStatus = PDF_DATAAVAIL_LOADALLFILE; | |
| 260 return TRUE; | |
| 261 } | |
| 262 return FALSE; | |
| 263 } | |
| 264 | |
| 265 m_arrayAcroforms.Add(m_pAcroForm); | |
| 266 m_docStatus = PDF_DATAAVAIL_PAGETREE; | |
| 267 return TRUE; | |
| 268 } | |
| 269 | |
| 270 FX_BOOL CPDF_DataAvail::CheckDocStatus(DownloadHints* pHints) { | |
| 271 switch (m_docStatus) { | |
| 272 case PDF_DATAAVAIL_HEADER: | |
| 273 return CheckHeader(pHints); | |
| 274 case PDF_DATAAVAIL_FIRSTPAGE: | |
| 275 case PDF_DATAAVAIL_FIRSTPAGE_PREPARE: | |
| 276 return CheckFirstPage(pHints); | |
| 277 case PDF_DATAAVAIL_HINTTABLE: | |
| 278 return CheckHintTables(pHints); | |
| 279 case PDF_DATAAVAIL_END: | |
| 280 return CheckEnd(pHints); | |
| 281 case PDF_DATAAVAIL_CROSSREF: | |
| 282 return CheckCrossRef(pHints); | |
| 283 case PDF_DATAAVAIL_CROSSREF_ITEM: | |
| 284 return CheckCrossRefItem(pHints); | |
| 285 case PDF_DATAAVAIL_CROSSREF_STREAM: | |
| 286 return CheckAllCrossRefStream(pHints); | |
| 287 case PDF_DATAAVAIL_TRAILER: | |
| 288 return CheckTrailer(pHints); | |
| 289 case PDF_DATAAVAIL_TRAILER_APPEND: | |
| 290 return CheckTrailerAppend(pHints); | |
| 291 case PDF_DATAAVAIL_LOADALLCROSSREF: | |
| 292 return LoadAllXref(pHints); | |
| 293 case PDF_DATAAVAIL_LOADALLFILE: | |
| 294 return LoadAllFile(pHints); | |
| 295 case PDF_DATAAVAIL_ROOT: | |
| 296 return CheckRoot(pHints); | |
| 297 case PDF_DATAAVAIL_INFO: | |
| 298 return CheckInfo(pHints); | |
| 299 case PDF_DATAAVAIL_ACROFORM: | |
| 300 return CheckAcroForm(pHints); | |
| 301 case PDF_DATAAVAIL_PAGETREE: | |
| 302 if (m_bTotalLoadPageTree) | |
| 303 return CheckPages(pHints); | |
| 304 return LoadDocPages(pHints); | |
| 305 case PDF_DATAAVAIL_PAGE: | |
| 306 if (m_bTotalLoadPageTree) | |
| 307 return CheckPage(pHints); | |
| 308 m_docStatus = PDF_DATAAVAIL_PAGE_LATERLOAD; | |
| 309 return TRUE; | |
| 310 case PDF_DATAAVAIL_ERROR: | |
| 311 return LoadAllFile(pHints); | |
| 312 case PDF_DATAAVAIL_PAGE_LATERLOAD: | |
| 313 m_docStatus = PDF_DATAAVAIL_PAGE; | |
| 314 default: | |
| 315 m_bDocAvail = TRUE; | |
| 316 return TRUE; | |
| 317 } | |
| 318 } | |
| 319 | |
| 320 FX_BOOL CPDF_DataAvail::CheckPageStatus(DownloadHints* pHints) { | |
| 321 switch (m_docStatus) { | |
| 322 case PDF_DATAAVAIL_PAGETREE: | |
| 323 return CheckPages(pHints); | |
| 324 case PDF_DATAAVAIL_PAGE: | |
| 325 return CheckPage(pHints); | |
| 326 case PDF_DATAAVAIL_ERROR: | |
| 327 return LoadAllFile(pHints); | |
| 328 default: | |
| 329 m_bPagesTreeLoad = TRUE; | |
| 330 m_bPagesLoad = TRUE; | |
| 331 return TRUE; | |
| 332 } | |
| 333 } | |
| 334 | |
| 335 FX_BOOL CPDF_DataAvail::LoadAllFile(DownloadHints* pHints) { | |
| 336 if (m_pFileAvail->IsDataAvail(0, (uint32_t)m_dwFileLen)) { | |
| 337 m_docStatus = PDF_DATAAVAIL_DONE; | |
| 338 return TRUE; | |
| 339 } | |
| 340 | |
| 341 pHints->AddSegment(0, (uint32_t)m_dwFileLen); | |
| 342 return FALSE; | |
| 343 } | |
| 344 | |
| 345 FX_BOOL CPDF_DataAvail::LoadAllXref(DownloadHints* pHints) { | |
| 346 m_parser.m_pSyntax->InitParser(m_pFileRead, (uint32_t)m_dwHeaderOffset); | |
| 347 m_parser.m_bOwnFileRead = false; | |
| 348 if (!m_parser.LoadAllCrossRefV4(m_dwLastXRefOffset) && | |
| 349 !m_parser.LoadAllCrossRefV5(m_dwLastXRefOffset)) { | |
| 350 m_docStatus = PDF_DATAAVAIL_LOADALLFILE; | |
| 351 return FALSE; | |
| 352 } | |
| 353 | |
| 354 m_dwRootObjNum = m_parser.GetRootObjNum(); | |
| 355 m_dwInfoObjNum = m_parser.GetInfoObjNum(); | |
| 356 m_pCurrentParser = &m_parser; | |
| 357 m_docStatus = PDF_DATAAVAIL_ROOT; | |
| 358 return TRUE; | |
| 359 } | |
| 360 | |
| 361 CPDF_Object* CPDF_DataAvail::GetObject(uint32_t objnum, | |
| 362 DownloadHints* pHints, | |
| 363 FX_BOOL* pExistInFile) { | |
| 364 CPDF_Object* pRet = nullptr; | |
| 365 uint32_t size = 0; | |
| 366 FX_FILESIZE offset = 0; | |
| 367 CPDF_Parser* pParser = nullptr; | |
| 368 | |
| 369 if (pExistInFile) | |
| 370 *pExistInFile = TRUE; | |
| 371 | |
| 372 if (m_pDocument) { | |
| 373 size = GetObjectSize(objnum, offset); | |
| 374 pParser = m_pDocument->GetParser(); | |
| 375 } else { | |
| 376 size = (uint32_t)m_parser.GetObjectSize(objnum); | |
| 377 offset = m_parser.GetObjectOffset(objnum); | |
| 378 pParser = &m_parser; | |
| 379 } | |
| 380 | |
| 381 if (!IsDataAvail(offset, size, pHints)) | |
| 382 return nullptr; | |
| 383 | |
| 384 if (pParser) | |
| 385 pRet = pParser->ParseIndirectObject(nullptr, objnum); | |
| 386 | |
| 387 if (!pRet && pExistInFile) | |
| 388 *pExistInFile = FALSE; | |
| 389 | |
| 390 return pRet; | |
| 391 } | |
| 392 | |
| 393 FX_BOOL CPDF_DataAvail::CheckInfo(DownloadHints* pHints) { | |
| 394 FX_BOOL bExist = FALSE; | |
| 395 CPDF_Object* pInfo = GetObject(m_dwInfoObjNum, pHints, &bExist); | |
| 396 if (!bExist) { | |
| 397 m_docStatus = | |
| 398 (m_bHaveAcroForm ? PDF_DATAAVAIL_ACROFORM : PDF_DATAAVAIL_PAGETREE); | |
| 399 return TRUE; | |
| 400 } | |
| 401 | |
| 402 if (!pInfo) { | |
| 403 if (m_docStatus == PDF_DATAAVAIL_ERROR) { | |
| 404 m_docStatus = PDF_DATAAVAIL_LOADALLFILE; | |
| 405 return TRUE; | |
| 406 } | |
| 407 | |
| 408 if (m_Pos == m_dwFileLen) | |
| 409 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 410 return FALSE; | |
| 411 } | |
| 412 | |
| 413 if (pInfo) | |
| 414 pInfo->Release(); | |
| 415 | |
| 416 m_docStatus = | |
| 417 (m_bHaveAcroForm ? PDF_DATAAVAIL_ACROFORM : PDF_DATAAVAIL_PAGETREE); | |
| 418 | |
| 419 return TRUE; | |
| 420 } | |
| 421 | |
| 422 FX_BOOL CPDF_DataAvail::CheckRoot(DownloadHints* pHints) { | |
| 423 FX_BOOL bExist = FALSE; | |
| 424 m_pRoot = GetObject(m_dwRootObjNum, pHints, &bExist); | |
| 425 if (!bExist) { | |
| 426 m_docStatus = PDF_DATAAVAIL_LOADALLFILE; | |
| 427 return TRUE; | |
| 428 } | |
| 429 | |
| 430 if (!m_pRoot) { | |
| 431 if (m_docStatus == PDF_DATAAVAIL_ERROR) { | |
| 432 m_docStatus = PDF_DATAAVAIL_LOADALLFILE; | |
| 433 return TRUE; | |
| 434 } | |
| 435 return FALSE; | |
| 436 } | |
| 437 | |
| 438 CPDF_Dictionary* pDict = m_pRoot->GetDict(); | |
| 439 if (!pDict) { | |
| 440 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 441 return FALSE; | |
| 442 } | |
| 443 | |
| 444 CPDF_Reference* pRef = ToReference(pDict->GetObjectFor("Pages")); | |
| 445 if (!pRef) { | |
| 446 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 447 return FALSE; | |
| 448 } | |
| 449 | |
| 450 m_PagesObjNum = pRef->GetRefObjNum(); | |
| 451 CPDF_Reference* pAcroFormRef = | |
| 452 ToReference(m_pRoot->GetDict()->GetObjectFor("AcroForm")); | |
| 453 if (pAcroFormRef) { | |
| 454 m_bHaveAcroForm = TRUE; | |
| 455 m_dwAcroFormObjNum = pAcroFormRef->GetRefObjNum(); | |
| 456 } | |
| 457 | |
| 458 if (m_dwInfoObjNum) { | |
| 459 m_docStatus = PDF_DATAAVAIL_INFO; | |
| 460 } else { | |
| 461 m_docStatus = | |
| 462 m_bHaveAcroForm ? PDF_DATAAVAIL_ACROFORM : PDF_DATAAVAIL_PAGETREE; | |
| 463 } | |
| 464 return TRUE; | |
| 465 } | |
| 466 | |
| 467 FX_BOOL CPDF_DataAvail::PreparePageItem() { | |
| 468 CPDF_Dictionary* pRoot = m_pDocument->GetRoot(); | |
| 469 CPDF_Reference* pRef = | |
| 470 ToReference(pRoot ? pRoot->GetObjectFor("Pages") : nullptr); | |
| 471 if (!pRef) { | |
| 472 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 473 return FALSE; | |
| 474 } | |
| 475 | |
| 476 m_PagesObjNum = pRef->GetRefObjNum(); | |
| 477 m_pCurrentParser = m_pDocument->GetParser(); | |
| 478 m_docStatus = PDF_DATAAVAIL_PAGETREE; | |
| 479 return TRUE; | |
| 480 } | |
| 481 | |
| 482 bool CPDF_DataAvail::IsFirstCheck(uint32_t dwPage) { | |
| 483 return m_pageMapCheckState.insert(dwPage).second; | |
| 484 } | |
| 485 | |
| 486 void CPDF_DataAvail::ResetFirstCheck(uint32_t dwPage) { | |
| 487 m_pageMapCheckState.erase(dwPage); | |
| 488 } | |
| 489 | |
| 490 FX_BOOL CPDF_DataAvail::CheckPage(DownloadHints* pHints) { | |
| 491 uint32_t iPageObjs = m_PageObjList.GetSize(); | |
| 492 CFX_ArrayTemplate<uint32_t> UnavailObjList; | |
| 493 for (uint32_t i = 0; i < iPageObjs; ++i) { | |
| 494 uint32_t dwPageObjNum = m_PageObjList.GetAt(i); | |
| 495 FX_BOOL bExist = FALSE; | |
| 496 CPDF_Object* pObj = GetObject(dwPageObjNum, pHints, &bExist); | |
| 497 if (!pObj) { | |
| 498 if (bExist) | |
| 499 UnavailObjList.Add(dwPageObjNum); | |
| 500 continue; | |
| 501 } | |
| 502 | |
| 503 CPDF_Array* pArray = ToArray(pObj); | |
| 504 if (pArray) { | |
| 505 for (CPDF_Object* pArrayObj : *pArray) { | |
| 506 if (CPDF_Reference* pRef = ToReference(pArrayObj)) | |
| 507 UnavailObjList.Add(pRef->GetRefObjNum()); | |
| 508 } | |
| 509 } | |
| 510 | |
| 511 if (!pObj->IsDictionary()) { | |
| 512 pObj->Release(); | |
| 513 continue; | |
| 514 } | |
| 515 | |
| 516 CFX_ByteString type = pObj->GetDict()->GetStringFor("Type"); | |
| 517 if (type == "Pages") { | |
| 518 m_PagesArray.Add(pObj); | |
| 519 continue; | |
| 520 } | |
| 521 pObj->Release(); | |
| 522 } | |
| 523 | |
| 524 m_PageObjList.RemoveAll(); | |
| 525 if (UnavailObjList.GetSize()) { | |
| 526 m_PageObjList.Append(UnavailObjList); | |
| 527 return FALSE; | |
| 528 } | |
| 529 | |
| 530 uint32_t iPages = m_PagesArray.GetSize(); | |
| 531 for (uint32_t i = 0; i < iPages; i++) { | |
| 532 CPDF_Object* pPages = m_PagesArray.GetAt(i); | |
| 533 if (!pPages) | |
| 534 continue; | |
| 535 | |
| 536 if (!GetPageKids(m_pCurrentParser, pPages)) { | |
| 537 pPages->Release(); | |
| 538 while (++i < iPages) { | |
| 539 pPages = m_PagesArray.GetAt(i); | |
| 540 pPages->Release(); | |
| 541 } | |
| 542 m_PagesArray.RemoveAll(); | |
| 543 | |
| 544 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 545 return FALSE; | |
| 546 } | |
| 547 pPages->Release(); | |
| 548 } | |
| 549 | |
| 550 m_PagesArray.RemoveAll(); | |
| 551 if (!m_PageObjList.GetSize()) | |
| 552 m_docStatus = PDF_DATAAVAIL_DONE; | |
| 553 return TRUE; | |
| 554 } | |
| 555 | |
| 556 FX_BOOL CPDF_DataAvail::GetPageKids(CPDF_Parser* pParser, CPDF_Object* pPages) { | |
| 557 if (!pParser) { | |
| 558 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 559 return FALSE; | |
| 560 } | |
| 561 | |
| 562 CPDF_Dictionary* pDict = pPages->GetDict(); | |
| 563 CPDF_Object* pKids = pDict ? pDict->GetObjectFor("Kids") : nullptr; | |
| 564 if (!pKids) | |
| 565 return TRUE; | |
| 566 | |
| 567 switch (pKids->GetType()) { | |
| 568 case CPDF_Object::REFERENCE: | |
| 569 m_PageObjList.Add(pKids->AsReference()->GetRefObjNum()); | |
| 570 break; | |
| 571 case CPDF_Object::ARRAY: { | |
| 572 CPDF_Array* pKidsArray = pKids->AsArray(); | |
| 573 for (size_t i = 0; i < pKidsArray->GetCount(); ++i) { | |
| 574 if (CPDF_Reference* pRef = ToReference(pKidsArray->GetObjectAt(i))) | |
| 575 m_PageObjList.Add(pRef->GetRefObjNum()); | |
| 576 } | |
| 577 } break; | |
| 578 default: | |
| 579 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 580 return FALSE; | |
| 581 } | |
| 582 return TRUE; | |
| 583 } | |
| 584 | |
| 585 FX_BOOL CPDF_DataAvail::CheckPages(DownloadHints* pHints) { | |
| 586 FX_BOOL bExist = FALSE; | |
| 587 CPDF_Object* pPages = GetObject(m_PagesObjNum, pHints, &bExist); | |
| 588 if (!bExist) { | |
| 589 m_docStatus = PDF_DATAAVAIL_LOADALLFILE; | |
| 590 return TRUE; | |
| 591 } | |
| 592 | |
| 593 if (!pPages) { | |
| 594 if (m_docStatus == PDF_DATAAVAIL_ERROR) { | |
| 595 m_docStatus = PDF_DATAAVAIL_LOADALLFILE; | |
| 596 return TRUE; | |
| 597 } | |
| 598 return FALSE; | |
| 599 } | |
| 600 | |
| 601 if (!GetPageKids(m_pCurrentParser, pPages)) { | |
| 602 pPages->Release(); | |
| 603 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 604 return FALSE; | |
| 605 } | |
| 606 | |
| 607 pPages->Release(); | |
| 608 m_docStatus = PDF_DATAAVAIL_PAGE; | |
| 609 return TRUE; | |
| 610 } | |
| 611 | |
| 612 FX_BOOL CPDF_DataAvail::CheckHeader(DownloadHints* pHints) { | |
| 613 ASSERT(m_dwFileLen >= 0); | |
| 614 const uint32_t kReqSize = std::min(static_cast<uint32_t>(m_dwFileLen), 1024U); | |
| 615 | |
| 616 if (m_pFileAvail->IsDataAvail(0, kReqSize)) { | |
| 617 uint8_t buffer[1024]; | |
| 618 m_pFileRead->ReadBlock(buffer, 0, kReqSize); | |
| 619 | |
| 620 if (IsLinearizedFile(buffer, kReqSize)) { | |
| 621 m_docStatus = PDF_DATAAVAIL_FIRSTPAGE; | |
| 622 } else { | |
| 623 if (m_docStatus == PDF_DATAAVAIL_ERROR) | |
| 624 return FALSE; | |
| 625 m_docStatus = PDF_DATAAVAIL_END; | |
| 626 } | |
| 627 return TRUE; | |
| 628 } | |
| 629 | |
| 630 pHints->AddSegment(0, kReqSize); | |
| 631 return FALSE; | |
| 632 } | |
| 633 | |
| 634 FX_BOOL CPDF_DataAvail::CheckFirstPage(DownloadHints* pHints) { | |
| 635 CPDF_Dictionary* pDict = m_pLinearized->GetDict(); | |
| 636 CPDF_Object* pEndOffSet = pDict ? pDict->GetObjectFor("E") : nullptr; | |
| 637 if (!pEndOffSet) { | |
| 638 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 639 return FALSE; | |
| 640 } | |
| 641 | |
| 642 CPDF_Object* pXRefOffset = pDict ? pDict->GetObjectFor("T") : nullptr; | |
| 643 if (!pXRefOffset) { | |
| 644 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 645 return FALSE; | |
| 646 } | |
| 647 | |
| 648 CPDF_Object* pFileLen = pDict ? pDict->GetObjectFor("L") : nullptr; | |
| 649 if (!pFileLen) { | |
| 650 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 651 return FALSE; | |
| 652 } | |
| 653 | |
| 654 FX_BOOL bNeedDownLoad = FALSE; | |
| 655 if (pEndOffSet->IsNumber()) { | |
| 656 uint32_t dwEnd = pEndOffSet->GetInteger(); | |
| 657 dwEnd += 512; | |
| 658 if ((FX_FILESIZE)dwEnd > m_dwFileLen) | |
| 659 dwEnd = (uint32_t)m_dwFileLen; | |
| 660 | |
| 661 int32_t iStartPos = (int32_t)(m_dwFileLen > 1024 ? 1024 : m_dwFileLen); | |
| 662 int32_t iSize = dwEnd > 1024 ? dwEnd - 1024 : 0; | |
| 663 if (!m_pFileAvail->IsDataAvail(iStartPos, iSize)) { | |
| 664 pHints->AddSegment(iStartPos, iSize); | |
| 665 bNeedDownLoad = TRUE; | |
| 666 } | |
| 667 } | |
| 668 | |
| 669 m_dwLastXRefOffset = 0; | |
| 670 FX_FILESIZE dwFileLen = 0; | |
| 671 if (pXRefOffset->IsNumber()) | |
| 672 m_dwLastXRefOffset = pXRefOffset->GetInteger(); | |
| 673 | |
| 674 if (pFileLen->IsNumber()) | |
| 675 dwFileLen = pFileLen->GetInteger(); | |
| 676 | |
| 677 if (!m_pFileAvail->IsDataAvail(m_dwLastXRefOffset, | |
| 678 (uint32_t)(dwFileLen - m_dwLastXRefOffset))) { | |
| 679 if (m_docStatus == PDF_DATAAVAIL_FIRSTPAGE) { | |
| 680 uint32_t dwSize = (uint32_t)(dwFileLen - m_dwLastXRefOffset); | |
| 681 FX_FILESIZE offset = m_dwLastXRefOffset; | |
| 682 if (dwSize < 512 && dwFileLen > 512) { | |
| 683 dwSize = 512; | |
| 684 offset = dwFileLen - 512; | |
| 685 } | |
| 686 pHints->AddSegment(offset, dwSize); | |
| 687 } | |
| 688 } else { | |
| 689 m_docStatus = PDF_DATAAVAIL_FIRSTPAGE_PREPARE; | |
| 690 } | |
| 691 | |
| 692 if (bNeedDownLoad || m_docStatus != PDF_DATAAVAIL_FIRSTPAGE_PREPARE) { | |
| 693 m_docStatus = PDF_DATAAVAIL_FIRSTPAGE_PREPARE; | |
| 694 return FALSE; | |
| 695 } | |
| 696 | |
| 697 m_docStatus = | |
| 698 m_bSupportHintTable ? PDF_DATAAVAIL_HINTTABLE : PDF_DATAAVAIL_DONE; | |
| 699 return TRUE; | |
| 700 } | |
| 701 | |
| 702 FX_BOOL CPDF_DataAvail::IsDataAvail(FX_FILESIZE offset, | |
| 703 uint32_t size, | |
| 704 DownloadHints* pHints) { | |
| 705 if (offset < 0 || offset > m_dwFileLen) | |
| 706 return TRUE; | |
| 707 | |
| 708 FX_SAFE_FILESIZE safeSize = offset; | |
| 709 safeSize += size; | |
| 710 safeSize += 512; | |
| 711 if (!safeSize.IsValid() || safeSize.ValueOrDie() > m_dwFileLen) | |
| 712 size = m_dwFileLen - offset; | |
| 713 else | |
| 714 size += 512; | |
| 715 | |
| 716 if (!m_pFileAvail->IsDataAvail(offset, size)) { | |
| 717 pHints->AddSegment(offset, size); | |
| 718 return FALSE; | |
| 719 } | |
| 720 return TRUE; | |
| 721 } | |
| 722 | |
| 723 FX_BOOL CPDF_DataAvail::CheckHintTables(DownloadHints* pHints) { | |
| 724 CPDF_Dictionary* pDict = m_pLinearized->GetDict(); | |
| 725 if (!pDict) { | |
| 726 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 727 return FALSE; | |
| 728 } | |
| 729 | |
| 730 // The actual value is not required here, but validate its existence and type. | |
| 731 CPDF_Number* pFirstPage = ToNumber(pDict->GetDirectObjectFor("O")); | |
| 732 if (!pFirstPage || !pFirstPage->IsInteger()) { | |
| 733 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 734 return FALSE; | |
| 735 } | |
| 736 | |
| 737 CPDF_Number* pPageCount = ToNumber(pDict->GetDirectObjectFor("N")); | |
| 738 if (!pPageCount || !pPageCount->IsInteger()) { | |
| 739 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 740 return FALSE; | |
| 741 } | |
| 742 | |
| 743 int nPageCount = pPageCount->GetInteger(); | |
| 744 if (nPageCount <= 1) { | |
| 745 m_docStatus = PDF_DATAAVAIL_DONE; | |
| 746 return TRUE; | |
| 747 } | |
| 748 | |
| 749 CPDF_Array* pHintStreamRange = pDict->GetArrayFor("H"); | |
| 750 size_t nHintStreamSize = pHintStreamRange ? pHintStreamRange->GetCount() : 0; | |
| 751 if (nHintStreamSize != 2 && nHintStreamSize != 4) { | |
| 752 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 753 return FALSE; | |
| 754 } | |
| 755 | |
| 756 for (const CPDF_Object* pArrayObject : *pHintStreamRange) { | |
| 757 const CPDF_Number* pNumber = ToNumber(pArrayObject->GetDirect()); | |
| 758 if (!pNumber || !pNumber->IsInteger()) { | |
| 759 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 760 return FALSE; | |
| 761 } | |
| 762 } | |
| 763 | |
| 764 FX_FILESIZE szHintStart = pHintStreamRange->GetIntegerAt(0); | |
| 765 FX_FILESIZE szHintLength = pHintStreamRange->GetIntegerAt(1); | |
| 766 if (szHintStart < 0 || szHintLength <= 0) { | |
| 767 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 768 return FALSE; | |
| 769 } | |
| 770 | |
| 771 if (!IsDataAvail(szHintStart, szHintLength, pHints)) | |
| 772 return FALSE; | |
| 773 | |
| 774 m_syntaxParser.InitParser(m_pFileRead, m_dwHeaderOffset); | |
| 775 | |
| 776 std::unique_ptr<CPDF_HintTables> pHintTables( | |
| 777 new CPDF_HintTables(this, pDict)); | |
| 778 std::unique_ptr<CPDF_Object, ReleaseDeleter<CPDF_Object>> pHintStream( | |
| 779 ParseIndirectObjectAt(szHintStart, 0)); | |
| 780 CPDF_Stream* pStream = ToStream(pHintStream.get()); | |
| 781 if (pStream && pHintTables->LoadHintStream(pStream)) | |
| 782 m_pHintTables = std::move(pHintTables); | |
| 783 | |
| 784 m_docStatus = PDF_DATAAVAIL_DONE; | |
| 785 return TRUE; | |
| 786 } | |
| 787 | |
| 788 CPDF_Object* CPDF_DataAvail::ParseIndirectObjectAt( | |
| 789 FX_FILESIZE pos, | |
| 790 uint32_t objnum, | |
| 791 CPDF_IndirectObjectHolder* pObjList) { | |
| 792 FX_FILESIZE SavedPos = m_syntaxParser.SavePos(); | |
| 793 m_syntaxParser.RestorePos(pos); | |
| 794 | |
| 795 bool bIsNumber; | |
| 796 CFX_ByteString word = m_syntaxParser.GetNextWord(&bIsNumber); | |
| 797 if (!bIsNumber) | |
| 798 return nullptr; | |
| 799 | |
| 800 uint32_t parser_objnum = FXSYS_atoui(word.c_str()); | |
| 801 if (objnum && parser_objnum != objnum) | |
| 802 return nullptr; | |
| 803 | |
| 804 word = m_syntaxParser.GetNextWord(&bIsNumber); | |
| 805 if (!bIsNumber) | |
| 806 return nullptr; | |
| 807 | |
| 808 uint32_t gennum = FXSYS_atoui(word.c_str()); | |
| 809 if (m_syntaxParser.GetKeyword() != "obj") { | |
| 810 m_syntaxParser.RestorePos(SavedPos); | |
| 811 return nullptr; | |
| 812 } | |
| 813 | |
| 814 CPDF_Object* pObj = | |
| 815 m_syntaxParser.GetObject(pObjList, parser_objnum, gennum, true); | |
| 816 m_syntaxParser.RestorePos(SavedPos); | |
| 817 return pObj; | |
| 818 } | |
| 819 | |
| 820 CPDF_DataAvail::DocLinearizationStatus CPDF_DataAvail::IsLinearizedPDF() { | |
| 821 const uint32_t kReqSize = 1024; | |
| 822 if (!m_pFileAvail->IsDataAvail(0, kReqSize)) | |
| 823 return LinearizationUnknown; | |
| 824 | |
| 825 if (!m_pFileRead) | |
| 826 return NotLinearized; | |
| 827 | |
| 828 FX_FILESIZE dwSize = m_pFileRead->GetSize(); | |
| 829 if (dwSize < (FX_FILESIZE)kReqSize) | |
| 830 return LinearizationUnknown; | |
| 831 | |
| 832 uint8_t buffer[1024]; | |
| 833 m_pFileRead->ReadBlock(buffer, 0, kReqSize); | |
| 834 if (IsLinearizedFile(buffer, kReqSize)) | |
| 835 return Linearized; | |
| 836 | |
| 837 return NotLinearized; | |
| 838 } | |
| 839 | |
| 840 FX_BOOL CPDF_DataAvail::IsLinearized() { | |
| 841 return m_bLinearized; | |
| 842 } | |
| 843 | |
| 844 FX_BOOL CPDF_DataAvail::IsLinearizedFile(uint8_t* pData, uint32_t dwLen) { | |
| 845 if (m_pLinearized) | |
| 846 return m_bLinearized; | |
| 847 | |
| 848 ScopedFileStream file(FX_CreateMemoryStream(pData, (size_t)dwLen, FALSE)); | |
| 849 | |
| 850 int32_t offset = GetHeaderOffset(file.get()); | |
| 851 if (offset == -1) { | |
| 852 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 853 return FALSE; | |
| 854 } | |
| 855 | |
| 856 m_dwHeaderOffset = offset; | |
| 857 m_syntaxParser.InitParser(file.get(), offset); | |
| 858 m_syntaxParser.RestorePos(m_syntaxParser.m_HeaderOffset + 9); | |
| 859 | |
| 860 bool bNumber; | |
| 861 CFX_ByteString wordObjNum = m_syntaxParser.GetNextWord(&bNumber); | |
| 862 if (!bNumber) | |
| 863 return FALSE; | |
| 864 | |
| 865 uint32_t objnum = FXSYS_atoui(wordObjNum.c_str()); | |
| 866 m_pLinearized = | |
| 867 ParseIndirectObjectAt(m_syntaxParser.m_HeaderOffset + 9, objnum); | |
| 868 if (!m_pLinearized) | |
| 869 return FALSE; | |
| 870 | |
| 871 CPDF_Dictionary* pDict = m_pLinearized->GetDict(); | |
| 872 if (!pDict || !pDict->GetObjectFor("Linearized")) | |
| 873 return FALSE; | |
| 874 | |
| 875 CPDF_Object* pLen = pDict->GetObjectFor("L"); | |
| 876 if (!pLen) | |
| 877 return FALSE; | |
| 878 | |
| 879 if ((FX_FILESIZE)pLen->GetInteger() != m_pFileRead->GetSize()) | |
| 880 return FALSE; | |
| 881 | |
| 882 m_bLinearized = TRUE; | |
| 883 | |
| 884 if (CPDF_Number* pNo = ToNumber(pDict->GetObjectFor("P"))) | |
| 885 m_dwFirstPageNo = pNo->GetInteger(); | |
| 886 | |
| 887 return TRUE; | |
| 888 } | |
| 889 | |
| 890 FX_BOOL CPDF_DataAvail::CheckEnd(DownloadHints* pHints) { | |
| 891 uint32_t req_pos = (uint32_t)(m_dwFileLen > 1024 ? m_dwFileLen - 1024 : 0); | |
| 892 uint32_t dwSize = (uint32_t)(m_dwFileLen - req_pos); | |
| 893 | |
| 894 if (m_pFileAvail->IsDataAvail(req_pos, dwSize)) { | |
| 895 uint8_t buffer[1024]; | |
| 896 m_pFileRead->ReadBlock(buffer, req_pos, dwSize); | |
| 897 | |
| 898 ScopedFileStream file(FX_CreateMemoryStream(buffer, (size_t)dwSize, FALSE)); | |
| 899 m_syntaxParser.InitParser(file.get(), 0); | |
| 900 m_syntaxParser.RestorePos(dwSize - 1); | |
| 901 | |
| 902 if (m_syntaxParser.SearchWord("startxref", TRUE, FALSE, dwSize)) { | |
| 903 m_syntaxParser.GetNextWord(nullptr); | |
| 904 | |
| 905 bool bNumber; | |
| 906 CFX_ByteString xrefpos_str = m_syntaxParser.GetNextWord(&bNumber); | |
| 907 if (!bNumber) { | |
| 908 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 909 return FALSE; | |
| 910 } | |
| 911 | |
| 912 m_dwXRefOffset = (FX_FILESIZE)FXSYS_atoi64(xrefpos_str.c_str()); | |
| 913 if (!m_dwXRefOffset || m_dwXRefOffset > m_dwFileLen) { | |
| 914 m_docStatus = PDF_DATAAVAIL_LOADALLFILE; | |
| 915 return TRUE; | |
| 916 } | |
| 917 | |
| 918 m_dwLastXRefOffset = m_dwXRefOffset; | |
| 919 SetStartOffset(m_dwXRefOffset); | |
| 920 m_docStatus = PDF_DATAAVAIL_CROSSREF; | |
| 921 return TRUE; | |
| 922 } | |
| 923 | |
| 924 m_docStatus = PDF_DATAAVAIL_LOADALLFILE; | |
| 925 return TRUE; | |
| 926 } | |
| 927 | |
| 928 pHints->AddSegment(req_pos, dwSize); | |
| 929 return FALSE; | |
| 930 } | |
| 931 | |
| 932 int32_t CPDF_DataAvail::CheckCrossRefStream(DownloadHints* pHints, | |
| 933 FX_FILESIZE& xref_offset) { | |
| 934 xref_offset = 0; | |
| 935 uint32_t req_size = | |
| 936 (uint32_t)(m_Pos + 512 > m_dwFileLen ? m_dwFileLen - m_Pos : 512); | |
| 937 | |
| 938 if (m_pFileAvail->IsDataAvail(m_Pos, req_size)) { | |
| 939 int32_t iSize = (int32_t)(m_Pos + req_size - m_dwCurrentXRefSteam); | |
| 940 CFX_BinaryBuf buf(iSize); | |
| 941 uint8_t* pBuf = buf.GetBuffer(); | |
| 942 | |
| 943 m_pFileRead->ReadBlock(pBuf, m_dwCurrentXRefSteam, iSize); | |
| 944 | |
| 945 ScopedFileStream file(FX_CreateMemoryStream(pBuf, (size_t)iSize, FALSE)); | |
| 946 m_parser.m_pSyntax->InitParser(file.get(), 0); | |
| 947 | |
| 948 bool bNumber; | |
| 949 CFX_ByteString objnum = m_parser.m_pSyntax->GetNextWord(&bNumber); | |
| 950 if (!bNumber) | |
| 951 return -1; | |
| 952 | |
| 953 uint32_t objNum = FXSYS_atoui(objnum.c_str()); | |
| 954 CPDF_Object* pObj = m_parser.ParseIndirectObjectAt(nullptr, 0, objNum); | |
| 955 if (!pObj) { | |
| 956 m_Pos += m_parser.m_pSyntax->SavePos(); | |
| 957 return 0; | |
| 958 } | |
| 959 | |
| 960 CPDF_Dictionary* pDict = pObj->GetDict(); | |
| 961 CPDF_Name* pName = ToName(pDict ? pDict->GetObjectFor("Type") : nullptr); | |
| 962 if (pName) { | |
| 963 if (pName->GetString() == "XRef") { | |
| 964 m_Pos += m_parser.m_pSyntax->SavePos(); | |
| 965 xref_offset = pObj->GetDict()->GetIntegerFor("Prev"); | |
| 966 pObj->Release(); | |
| 967 return 1; | |
| 968 } | |
| 969 } | |
| 970 pObj->Release(); | |
| 971 return -1; | |
| 972 } | |
| 973 pHints->AddSegment(m_Pos, req_size); | |
| 974 return 0; | |
| 975 } | |
| 976 | |
| 977 void CPDF_DataAvail::SetStartOffset(FX_FILESIZE dwOffset) { | |
| 978 m_Pos = dwOffset; | |
| 979 } | |
| 980 | |
| 981 FX_BOOL CPDF_DataAvail::GetNextToken(CFX_ByteString& token) { | |
| 982 uint8_t ch; | |
| 983 if (!GetNextChar(ch)) | |
| 984 return FALSE; | |
| 985 | |
| 986 while (1) { | |
| 987 while (PDFCharIsWhitespace(ch)) { | |
| 988 if (!GetNextChar(ch)) | |
| 989 return FALSE; | |
| 990 } | |
| 991 | |
| 992 if (ch != '%') | |
| 993 break; | |
| 994 | |
| 995 while (1) { | |
| 996 if (!GetNextChar(ch)) | |
| 997 return FALSE; | |
| 998 if (PDFCharIsLineEnding(ch)) | |
| 999 break; | |
| 1000 } | |
| 1001 } | |
| 1002 | |
| 1003 uint8_t buffer[256]; | |
| 1004 uint32_t index = 0; | |
| 1005 if (PDFCharIsDelimiter(ch)) { | |
| 1006 buffer[index++] = ch; | |
| 1007 if (ch == '/') { | |
| 1008 while (1) { | |
| 1009 if (!GetNextChar(ch)) | |
| 1010 return FALSE; | |
| 1011 | |
| 1012 if (!PDFCharIsOther(ch) && !PDFCharIsNumeric(ch)) { | |
| 1013 m_Pos--; | |
| 1014 CFX_ByteString ret(buffer, index); | |
| 1015 token = ret; | |
| 1016 return TRUE; | |
| 1017 } | |
| 1018 | |
| 1019 if (index < sizeof(buffer)) | |
| 1020 buffer[index++] = ch; | |
| 1021 } | |
| 1022 } else if (ch == '<') { | |
| 1023 if (!GetNextChar(ch)) | |
| 1024 return FALSE; | |
| 1025 | |
| 1026 if (ch == '<') | |
| 1027 buffer[index++] = ch; | |
| 1028 else | |
| 1029 m_Pos--; | |
| 1030 } else if (ch == '>') { | |
| 1031 if (!GetNextChar(ch)) | |
| 1032 return FALSE; | |
| 1033 | |
| 1034 if (ch == '>') | |
| 1035 buffer[index++] = ch; | |
| 1036 else | |
| 1037 m_Pos--; | |
| 1038 } | |
| 1039 | |
| 1040 CFX_ByteString ret(buffer, index); | |
| 1041 token = ret; | |
| 1042 return TRUE; | |
| 1043 } | |
| 1044 | |
| 1045 while (1) { | |
| 1046 if (index < sizeof(buffer)) | |
| 1047 buffer[index++] = ch; | |
| 1048 | |
| 1049 if (!GetNextChar(ch)) | |
| 1050 return FALSE; | |
| 1051 | |
| 1052 if (PDFCharIsDelimiter(ch) || PDFCharIsWhitespace(ch)) { | |
| 1053 m_Pos--; | |
| 1054 break; | |
| 1055 } | |
| 1056 } | |
| 1057 | |
| 1058 token = CFX_ByteString(buffer, index); | |
| 1059 return TRUE; | |
| 1060 } | |
| 1061 | |
| 1062 FX_BOOL CPDF_DataAvail::GetNextChar(uint8_t& ch) { | |
| 1063 FX_FILESIZE pos = m_Pos; | |
| 1064 if (pos >= m_dwFileLen) | |
| 1065 return FALSE; | |
| 1066 | |
| 1067 if (m_bufferOffset >= pos || | |
| 1068 (FX_FILESIZE)(m_bufferOffset + m_bufferSize) <= pos) { | |
| 1069 FX_FILESIZE read_pos = pos; | |
| 1070 uint32_t read_size = 512; | |
| 1071 if ((FX_FILESIZE)read_size > m_dwFileLen) | |
| 1072 read_size = (uint32_t)m_dwFileLen; | |
| 1073 | |
| 1074 if ((FX_FILESIZE)(read_pos + read_size) > m_dwFileLen) | |
| 1075 read_pos = m_dwFileLen - read_size; | |
| 1076 | |
| 1077 if (!m_pFileRead->ReadBlock(m_bufferData, read_pos, read_size)) | |
| 1078 return FALSE; | |
| 1079 | |
| 1080 m_bufferOffset = read_pos; | |
| 1081 m_bufferSize = read_size; | |
| 1082 } | |
| 1083 ch = m_bufferData[pos - m_bufferOffset]; | |
| 1084 m_Pos++; | |
| 1085 return TRUE; | |
| 1086 } | |
| 1087 | |
| 1088 FX_BOOL CPDF_DataAvail::CheckCrossRefItem(DownloadHints* pHints) { | |
| 1089 int32_t iSize = 0; | |
| 1090 CFX_ByteString token; | |
| 1091 while (1) { | |
| 1092 if (!GetNextToken(token)) { | |
| 1093 iSize = (int32_t)(m_Pos + 512 > m_dwFileLen ? m_dwFileLen - m_Pos : 512); | |
| 1094 pHints->AddSegment(m_Pos, iSize); | |
| 1095 return FALSE; | |
| 1096 } | |
| 1097 | |
| 1098 if (token == "trailer") { | |
| 1099 m_dwTrailerOffset = m_Pos; | |
| 1100 m_docStatus = PDF_DATAAVAIL_TRAILER; | |
| 1101 return TRUE; | |
| 1102 } | |
| 1103 } | |
| 1104 } | |
| 1105 | |
| 1106 FX_BOOL CPDF_DataAvail::CheckAllCrossRefStream(DownloadHints* pHints) { | |
| 1107 FX_FILESIZE xref_offset = 0; | |
| 1108 | |
| 1109 int32_t nRet = CheckCrossRefStream(pHints, xref_offset); | |
| 1110 if (nRet == 1) { | |
| 1111 if (!xref_offset) { | |
| 1112 m_docStatus = PDF_DATAAVAIL_LOADALLCROSSREF; | |
| 1113 } else { | |
| 1114 m_dwCurrentXRefSteam = xref_offset; | |
| 1115 m_Pos = xref_offset; | |
| 1116 } | |
| 1117 return TRUE; | |
| 1118 } | |
| 1119 | |
| 1120 if (nRet == -1) | |
| 1121 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 1122 return FALSE; | |
| 1123 } | |
| 1124 | |
| 1125 FX_BOOL CPDF_DataAvail::CheckCrossRef(DownloadHints* pHints) { | |
| 1126 int32_t iSize = 0; | |
| 1127 CFX_ByteString token; | |
| 1128 if (!GetNextToken(token)) { | |
| 1129 iSize = (int32_t)(m_Pos + 512 > m_dwFileLen ? m_dwFileLen - m_Pos : 512); | |
| 1130 pHints->AddSegment(m_Pos, iSize); | |
| 1131 return FALSE; | |
| 1132 } | |
| 1133 | |
| 1134 if (token == "xref") { | |
| 1135 while (1) { | |
| 1136 if (!GetNextToken(token)) { | |
| 1137 iSize = | |
| 1138 (int32_t)(m_Pos + 512 > m_dwFileLen ? m_dwFileLen - m_Pos : 512); | |
| 1139 pHints->AddSegment(m_Pos, iSize); | |
| 1140 m_docStatus = PDF_DATAAVAIL_CROSSREF_ITEM; | |
| 1141 return FALSE; | |
| 1142 } | |
| 1143 | |
| 1144 if (token == "trailer") { | |
| 1145 m_dwTrailerOffset = m_Pos; | |
| 1146 m_docStatus = PDF_DATAAVAIL_TRAILER; | |
| 1147 return TRUE; | |
| 1148 } | |
| 1149 } | |
| 1150 } else { | |
| 1151 m_docStatus = PDF_DATAAVAIL_LOADALLFILE; | |
| 1152 return TRUE; | |
| 1153 } | |
| 1154 return FALSE; | |
| 1155 } | |
| 1156 | |
| 1157 FX_BOOL CPDF_DataAvail::CheckTrailerAppend(DownloadHints* pHints) { | |
| 1158 if (m_Pos < m_dwFileLen) { | |
| 1159 FX_FILESIZE dwAppendPos = m_Pos + m_syntaxParser.SavePos(); | |
| 1160 int32_t iSize = (int32_t)( | |
| 1161 dwAppendPos + 512 > m_dwFileLen ? m_dwFileLen - dwAppendPos : 512); | |
| 1162 | |
| 1163 if (!m_pFileAvail->IsDataAvail(dwAppendPos, iSize)) { | |
| 1164 pHints->AddSegment(dwAppendPos, iSize); | |
| 1165 return FALSE; | |
| 1166 } | |
| 1167 } | |
| 1168 | |
| 1169 if (m_dwPrevXRefOffset) { | |
| 1170 SetStartOffset(m_dwPrevXRefOffset); | |
| 1171 m_docStatus = PDF_DATAAVAIL_CROSSREF; | |
| 1172 } else { | |
| 1173 m_docStatus = PDF_DATAAVAIL_LOADALLCROSSREF; | |
| 1174 } | |
| 1175 return TRUE; | |
| 1176 } | |
| 1177 | |
| 1178 FX_BOOL CPDF_DataAvail::CheckTrailer(DownloadHints* pHints) { | |
| 1179 int32_t iTrailerSize = | |
| 1180 (int32_t)(m_Pos + 512 > m_dwFileLen ? m_dwFileLen - m_Pos : 512); | |
| 1181 if (m_pFileAvail->IsDataAvail(m_Pos, iTrailerSize)) { | |
| 1182 int32_t iSize = (int32_t)(m_Pos + iTrailerSize - m_dwTrailerOffset); | |
| 1183 CFX_BinaryBuf buf(iSize); | |
| 1184 uint8_t* pBuf = buf.GetBuffer(); | |
| 1185 if (!pBuf) { | |
| 1186 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 1187 return FALSE; | |
| 1188 } | |
| 1189 | |
| 1190 if (!m_pFileRead->ReadBlock(pBuf, m_dwTrailerOffset, iSize)) | |
| 1191 return FALSE; | |
| 1192 | |
| 1193 ScopedFileStream file(FX_CreateMemoryStream(pBuf, (size_t)iSize, FALSE)); | |
| 1194 m_syntaxParser.InitParser(file.get(), 0); | |
| 1195 | |
| 1196 std::unique_ptr<CPDF_Object, ReleaseDeleter<CPDF_Object>> pTrailer( | |
| 1197 m_syntaxParser.GetObject(nullptr, 0, 0, true)); | |
| 1198 if (!pTrailer) { | |
| 1199 m_Pos += m_syntaxParser.SavePos(); | |
| 1200 pHints->AddSegment(m_Pos, iTrailerSize); | |
| 1201 return FALSE; | |
| 1202 } | |
| 1203 | |
| 1204 if (!pTrailer->IsDictionary()) | |
| 1205 return FALSE; | |
| 1206 | |
| 1207 CPDF_Dictionary* pTrailerDict = pTrailer->GetDict(); | |
| 1208 CPDF_Object* pEncrypt = pTrailerDict->GetObjectFor("Encrypt"); | |
| 1209 if (ToReference(pEncrypt)) { | |
| 1210 m_docStatus = PDF_DATAAVAIL_LOADALLFILE; | |
| 1211 return TRUE; | |
| 1212 } | |
| 1213 | |
| 1214 uint32_t xrefpos = GetDirectInteger(pTrailerDict, "Prev"); | |
| 1215 if (xrefpos) { | |
| 1216 m_dwPrevXRefOffset = GetDirectInteger(pTrailerDict, "XRefStm"); | |
| 1217 if (m_dwPrevXRefOffset) { | |
| 1218 m_docStatus = PDF_DATAAVAIL_LOADALLFILE; | |
| 1219 } else { | |
| 1220 m_dwPrevXRefOffset = xrefpos; | |
| 1221 if (m_dwPrevXRefOffset >= m_dwFileLen) { | |
| 1222 m_docStatus = PDF_DATAAVAIL_LOADALLFILE; | |
| 1223 } else { | |
| 1224 SetStartOffset(m_dwPrevXRefOffset); | |
| 1225 m_docStatus = PDF_DATAAVAIL_TRAILER_APPEND; | |
| 1226 } | |
| 1227 } | |
| 1228 return TRUE; | |
| 1229 } | |
| 1230 m_dwPrevXRefOffset = 0; | |
| 1231 m_docStatus = PDF_DATAAVAIL_TRAILER_APPEND; | |
| 1232 return TRUE; | |
| 1233 } | |
| 1234 pHints->AddSegment(m_Pos, iTrailerSize); | |
| 1235 return FALSE; | |
| 1236 } | |
| 1237 | |
| 1238 FX_BOOL CPDF_DataAvail::CheckPage(uint32_t dwPage, DownloadHints* pHints) { | |
| 1239 while (TRUE) { | |
| 1240 switch (m_docStatus) { | |
| 1241 case PDF_DATAAVAIL_PAGETREE: | |
| 1242 if (!LoadDocPages(pHints)) | |
| 1243 return FALSE; | |
| 1244 break; | |
| 1245 case PDF_DATAAVAIL_PAGE: | |
| 1246 if (!LoadDocPage(dwPage, pHints)) | |
| 1247 return FALSE; | |
| 1248 break; | |
| 1249 case PDF_DATAAVAIL_ERROR: | |
| 1250 return LoadAllFile(pHints); | |
| 1251 default: | |
| 1252 m_bPagesTreeLoad = TRUE; | |
| 1253 m_bPagesLoad = TRUE; | |
| 1254 m_bCurPageDictLoadOK = TRUE; | |
| 1255 m_docStatus = PDF_DATAAVAIL_PAGE; | |
| 1256 return TRUE; | |
| 1257 } | |
| 1258 } | |
| 1259 } | |
| 1260 | |
| 1261 FX_BOOL CPDF_DataAvail::CheckArrayPageNode(uint32_t dwPageNo, | |
| 1262 PageNode* pPageNode, | |
| 1263 DownloadHints* pHints) { | |
| 1264 FX_BOOL bExist = FALSE; | |
| 1265 CPDF_Object* pPages = GetObject(dwPageNo, pHints, &bExist); | |
| 1266 if (!bExist) { | |
| 1267 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 1268 return FALSE; | |
| 1269 } | |
| 1270 | |
| 1271 if (!pPages) { | |
| 1272 if (m_docStatus == PDF_DATAAVAIL_ERROR) { | |
| 1273 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 1274 return FALSE; | |
| 1275 } | |
| 1276 return FALSE; | |
| 1277 } | |
| 1278 | |
| 1279 CPDF_Array* pArray = pPages->AsArray(); | |
| 1280 if (!pArray) { | |
| 1281 pPages->Release(); | |
| 1282 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 1283 return FALSE; | |
| 1284 } | |
| 1285 | |
| 1286 pPageNode->m_type = PDF_PAGENODE_PAGES; | |
| 1287 for (size_t i = 0; i < pArray->GetCount(); ++i) { | |
| 1288 CPDF_Reference* pKid = ToReference(pArray->GetObjectAt(i)); | |
| 1289 if (!pKid) | |
| 1290 continue; | |
| 1291 | |
| 1292 PageNode* pNode = new PageNode(); | |
| 1293 pPageNode->m_childNode.Add(pNode); | |
| 1294 pNode->m_dwPageNo = pKid->GetRefObjNum(); | |
| 1295 } | |
| 1296 pPages->Release(); | |
| 1297 return TRUE; | |
| 1298 } | |
| 1299 | |
| 1300 FX_BOOL CPDF_DataAvail::CheckUnkownPageNode(uint32_t dwPageNo, | |
| 1301 PageNode* pPageNode, | |
| 1302 DownloadHints* pHints) { | |
| 1303 FX_BOOL bExist = FALSE; | |
| 1304 CPDF_Object* pPage = GetObject(dwPageNo, pHints, &bExist); | |
| 1305 if (!bExist) { | |
| 1306 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 1307 return FALSE; | |
| 1308 } | |
| 1309 | |
| 1310 if (!pPage) { | |
| 1311 if (m_docStatus == PDF_DATAAVAIL_ERROR) | |
| 1312 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 1313 return FALSE; | |
| 1314 } | |
| 1315 | |
| 1316 if (pPage->IsArray()) { | |
| 1317 pPageNode->m_dwPageNo = dwPageNo; | |
| 1318 pPageNode->m_type = PDF_PAGENODE_ARRAY; | |
| 1319 pPage->Release(); | |
| 1320 return TRUE; | |
| 1321 } | |
| 1322 | |
| 1323 if (!pPage->IsDictionary()) { | |
| 1324 pPage->Release(); | |
| 1325 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 1326 return FALSE; | |
| 1327 } | |
| 1328 | |
| 1329 pPageNode->m_dwPageNo = dwPageNo; | |
| 1330 CPDF_Dictionary* pDict = pPage->GetDict(); | |
| 1331 CFX_ByteString type = pDict->GetStringFor("Type"); | |
| 1332 if (type == "Pages") { | |
| 1333 pPageNode->m_type = PDF_PAGENODE_PAGES; | |
| 1334 CPDF_Object* pKids = pDict->GetObjectFor("Kids"); | |
| 1335 if (!pKids) { | |
| 1336 m_docStatus = PDF_DATAAVAIL_PAGE; | |
| 1337 return TRUE; | |
| 1338 } | |
| 1339 | |
| 1340 switch (pKids->GetType()) { | |
| 1341 case CPDF_Object::REFERENCE: { | |
| 1342 CPDF_Reference* pKid = pKids->AsReference(); | |
| 1343 PageNode* pNode = new PageNode(); | |
| 1344 pPageNode->m_childNode.Add(pNode); | |
| 1345 pNode->m_dwPageNo = pKid->GetRefObjNum(); | |
| 1346 } break; | |
| 1347 case CPDF_Object::ARRAY: { | |
| 1348 CPDF_Array* pKidsArray = pKids->AsArray(); | |
| 1349 for (size_t i = 0; i < pKidsArray->GetCount(); ++i) { | |
| 1350 CPDF_Reference* pKid = ToReference(pKidsArray->GetObjectAt(i)); | |
| 1351 if (!pKid) | |
| 1352 continue; | |
| 1353 | |
| 1354 PageNode* pNode = new PageNode(); | |
| 1355 pPageNode->m_childNode.Add(pNode); | |
| 1356 pNode->m_dwPageNo = pKid->GetRefObjNum(); | |
| 1357 } | |
| 1358 } break; | |
| 1359 default: | |
| 1360 break; | |
| 1361 } | |
| 1362 } else if (type == "Page") { | |
| 1363 pPageNode->m_type = PDF_PAGENODE_PAGE; | |
| 1364 } else { | |
| 1365 pPage->Release(); | |
| 1366 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 1367 return FALSE; | |
| 1368 } | |
| 1369 pPage->Release(); | |
| 1370 return TRUE; | |
| 1371 } | |
| 1372 | |
| 1373 FX_BOOL CPDF_DataAvail::CheckPageNode(CPDF_DataAvail::PageNode& pageNodes, | |
| 1374 int32_t iPage, | |
| 1375 int32_t& iCount, | |
| 1376 DownloadHints* pHints, | |
| 1377 int level) { | |
| 1378 if (level >= kMaxPageRecursionDepth) | |
| 1379 return FALSE; | |
| 1380 | |
| 1381 int32_t iSize = pageNodes.m_childNode.GetSize(); | |
| 1382 if (iSize <= 0 || iPage >= iSize) { | |
| 1383 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 1384 return FALSE; | |
| 1385 } | |
| 1386 | |
| 1387 for (int32_t i = 0; i < iSize; ++i) { | |
| 1388 PageNode* pNode = pageNodes.m_childNode.GetAt(i); | |
| 1389 if (!pNode) | |
| 1390 continue; | |
| 1391 | |
| 1392 switch (pNode->m_type) { | |
| 1393 case PDF_PAGENODE_UNKNOWN: | |
| 1394 if (!CheckUnkownPageNode(pNode->m_dwPageNo, pNode, pHints)) { | |
| 1395 return FALSE; | |
| 1396 } | |
| 1397 --i; | |
| 1398 break; | |
| 1399 case PDF_PAGENODE_PAGE: | |
| 1400 iCount++; | |
| 1401 if (iPage == iCount && m_pDocument) | |
| 1402 m_pDocument->SetPageObjNum(iPage, pNode->m_dwPageNo); | |
| 1403 break; | |
| 1404 case PDF_PAGENODE_PAGES: | |
| 1405 if (!CheckPageNode(*pNode, iPage, iCount, pHints, level + 1)) | |
| 1406 return FALSE; | |
| 1407 break; | |
| 1408 case PDF_PAGENODE_ARRAY: | |
| 1409 if (!CheckArrayPageNode(pNode->m_dwPageNo, pNode, pHints)) | |
| 1410 return FALSE; | |
| 1411 --i; | |
| 1412 break; | |
| 1413 } | |
| 1414 | |
| 1415 if (iPage == iCount) { | |
| 1416 m_docStatus = PDF_DATAAVAIL_DONE; | |
| 1417 return TRUE; | |
| 1418 } | |
| 1419 } | |
| 1420 return TRUE; | |
| 1421 } | |
| 1422 | |
| 1423 FX_BOOL CPDF_DataAvail::LoadDocPage(uint32_t dwPage, DownloadHints* pHints) { | |
| 1424 FX_SAFE_INT32 safePage = pdfium::base::checked_cast<int32_t>(dwPage); | |
| 1425 int32_t iPage = safePage.ValueOrDie(); | |
| 1426 if (m_pDocument->GetPageCount() <= iPage || | |
| 1427 m_pDocument->IsPageLoaded(iPage)) { | |
| 1428 m_docStatus = PDF_DATAAVAIL_DONE; | |
| 1429 return TRUE; | |
| 1430 } | |
| 1431 | |
| 1432 if (m_pageNodes.m_type == PDF_PAGENODE_PAGE) { | |
| 1433 if (iPage == 0) { | |
| 1434 m_docStatus = PDF_DATAAVAIL_DONE; | |
| 1435 return TRUE; | |
| 1436 } | |
| 1437 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 1438 return TRUE; | |
| 1439 } | |
| 1440 int32_t iCount = -1; | |
| 1441 return CheckPageNode(m_pageNodes, iPage, iCount, pHints, 0); | |
| 1442 } | |
| 1443 | |
| 1444 FX_BOOL CPDF_DataAvail::CheckPageCount(DownloadHints* pHints) { | |
| 1445 FX_BOOL bExist = FALSE; | |
| 1446 CPDF_Object* pPages = GetObject(m_PagesObjNum, pHints, &bExist); | |
| 1447 if (!bExist) { | |
| 1448 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 1449 return FALSE; | |
| 1450 } | |
| 1451 | |
| 1452 if (!pPages) | |
| 1453 return FALSE; | |
| 1454 | |
| 1455 CPDF_Dictionary* pPagesDict = pPages->GetDict(); | |
| 1456 if (!pPagesDict) { | |
| 1457 pPages->Release(); | |
| 1458 m_docStatus = PDF_DATAAVAIL_ERROR; | |
| 1459 return FALSE; | |
| 1460 } | |
| 1461 | |
| 1462 if (!pPagesDict->KeyExist("Kids")) { | |
| 1463 pPages->Release(); | |
| 1464 return TRUE; | |
| 1465 } | |
| 1466 | |
| 1467 int count = pPagesDict->GetIntegerFor("Count"); | |
| 1468 if (count > 0) { | |
| 1469 pPages->Release(); | |
| 1470 return TRUE; | |
| 1471 } | |
| 1472 | |
| 1473 pPages->Release(); | |
| 1474 return FALSE; | |
| 1475 } | |
| 1476 | |
| 1477 FX_BOOL CPDF_DataAvail::LoadDocPages(DownloadHints* pHints) { | |
| 1478 if (!CheckUnkownPageNode(m_PagesObjNum, &m_pageNodes, pHints)) | |
| 1479 return FALSE; | |
| 1480 | |
| 1481 if (CheckPageCount(pHints)) { | |
| 1482 m_docStatus = PDF_DATAAVAIL_PAGE; | |
| 1483 return TRUE; | |
| 1484 } | |
| 1485 | |
| 1486 m_bTotalLoadPageTree = TRUE; | |
| 1487 return FALSE; | |
| 1488 } | |
| 1489 | |
| 1490 FX_BOOL CPDF_DataAvail::LoadPages(DownloadHints* pHints) { | |
| 1491 while (!m_bPagesTreeLoad) { | |
| 1492 if (!CheckPageStatus(pHints)) | |
| 1493 return FALSE; | |
| 1494 } | |
| 1495 | |
| 1496 if (m_bPagesLoad) | |
| 1497 return TRUE; | |
| 1498 | |
| 1499 m_pDocument->LoadPages(); | |
| 1500 return FALSE; | |
| 1501 } | |
| 1502 | |
| 1503 CPDF_DataAvail::DocAvailStatus CPDF_DataAvail::CheckLinearizedData( | |
| 1504 DownloadHints* pHints) { | |
| 1505 if (m_bLinearedDataOK) | |
| 1506 return DataAvailable; | |
| 1507 | |
| 1508 if (!m_bMainXRefLoadTried) { | |
| 1509 FX_SAFE_UINT32 data_size = m_dwFileLen; | |
| 1510 data_size -= m_dwLastXRefOffset; | |
| 1511 if (!data_size.IsValid()) | |
| 1512 return DataError; | |
| 1513 | |
| 1514 if (!m_pFileAvail->IsDataAvail(m_dwLastXRefOffset, | |
| 1515 data_size.ValueOrDie())) { | |
| 1516 pHints->AddSegment(m_dwLastXRefOffset, data_size.ValueOrDie()); | |
| 1517 return DataNotAvailable; | |
| 1518 } | |
| 1519 | |
| 1520 CPDF_Parser::Error eRet = | |
| 1521 m_pDocument->GetParser()->LoadLinearizedMainXRefTable(); | |
| 1522 m_bMainXRefLoadTried = TRUE; | |
| 1523 if (eRet != CPDF_Parser::SUCCESS) | |
| 1524 return DataError; | |
| 1525 | |
| 1526 if (!PreparePageItem()) | |
| 1527 return DataNotAvailable; | |
| 1528 | |
| 1529 m_bMainXRefLoadedOK = TRUE; | |
| 1530 m_bLinearedDataOK = TRUE; | |
| 1531 } | |
| 1532 | |
| 1533 return m_bLinearedDataOK ? DataAvailable : DataNotAvailable; | |
| 1534 } | |
| 1535 | |
| 1536 FX_BOOL CPDF_DataAvail::CheckPageAnnots(uint32_t dwPage, | |
| 1537 DownloadHints* pHints) { | |
| 1538 if (!m_objs_array.GetSize()) { | |
| 1539 m_objs_array.RemoveAll(); | |
| 1540 m_ObjectSet.clear(); | |
| 1541 | |
| 1542 FX_SAFE_INT32 safePage = pdfium::base::checked_cast<int32_t>(dwPage); | |
| 1543 CPDF_Dictionary* pPageDict = m_pDocument->GetPage(safePage.ValueOrDie()); | |
| 1544 if (!pPageDict) | |
| 1545 return TRUE; | |
| 1546 | |
| 1547 CPDF_Object* pAnnots = pPageDict->GetObjectFor("Annots"); | |
| 1548 if (!pAnnots) | |
| 1549 return TRUE; | |
| 1550 | |
| 1551 CFX_ArrayTemplate<CPDF_Object*> obj_array; | |
| 1552 obj_array.Add(pAnnots); | |
| 1553 | |
| 1554 FX_BOOL bRet = IsObjectsAvail(obj_array, FALSE, pHints, m_objs_array); | |
| 1555 if (bRet) | |
| 1556 m_objs_array.RemoveAll(); | |
| 1557 | |
| 1558 return bRet; | |
| 1559 } | |
| 1560 | |
| 1561 CFX_ArrayTemplate<CPDF_Object*> new_objs_array; | |
| 1562 FX_BOOL bRet = IsObjectsAvail(m_objs_array, FALSE, pHints, new_objs_array); | |
| 1563 m_objs_array.RemoveAll(); | |
| 1564 if (!bRet) | |
| 1565 m_objs_array.Append(new_objs_array); | |
| 1566 | |
| 1567 return bRet; | |
| 1568 } | |
| 1569 | |
| 1570 CPDF_DataAvail::DocAvailStatus CPDF_DataAvail::CheckLinearizedFirstPage( | |
| 1571 uint32_t dwPage, | |
| 1572 DownloadHints* pHints) { | |
| 1573 if (!m_bAnnotsLoad) { | |
| 1574 if (!CheckPageAnnots(dwPage, pHints)) | |
| 1575 return DataNotAvailable; | |
| 1576 m_bAnnotsLoad = TRUE; | |
| 1577 } | |
| 1578 | |
| 1579 DocAvailStatus nRet = CheckLinearizedData(pHints); | |
| 1580 if (nRet == DataAvailable) | |
| 1581 m_bPageLoadedOK = FALSE; | |
| 1582 return nRet; | |
| 1583 } | |
| 1584 | |
| 1585 FX_BOOL CPDF_DataAvail::HaveResourceAncestor(CPDF_Dictionary* pDict) { | |
| 1586 CFX_AutoRestorer<int> restorer(&s_CurrentDataAvailRecursionDepth); | |
| 1587 if (++s_CurrentDataAvailRecursionDepth > kMaxDataAvailRecursionDepth) | |
| 1588 return FALSE; | |
| 1589 | |
| 1590 CPDF_Object* pParent = pDict->GetObjectFor("Parent"); | |
| 1591 if (!pParent) | |
| 1592 return FALSE; | |
| 1593 | |
| 1594 CPDF_Dictionary* pParentDict = pParent->GetDict(); | |
| 1595 if (!pParentDict) | |
| 1596 return FALSE; | |
| 1597 | |
| 1598 CPDF_Object* pRet = pParentDict->GetObjectFor("Resources"); | |
| 1599 if (pRet) { | |
| 1600 m_pPageResource = pRet; | |
| 1601 return TRUE; | |
| 1602 } | |
| 1603 | |
| 1604 return HaveResourceAncestor(pParentDict); | |
| 1605 } | |
| 1606 | |
| 1607 CPDF_DataAvail::DocAvailStatus CPDF_DataAvail::IsPageAvail( | |
| 1608 uint32_t dwPage, | |
| 1609 DownloadHints* pHints) { | |
| 1610 if (!m_pDocument) | |
| 1611 return DataError; | |
| 1612 | |
| 1613 if (IsFirstCheck(dwPage)) { | |
| 1614 m_bCurPageDictLoadOK = FALSE; | |
| 1615 m_bPageLoadedOK = FALSE; | |
| 1616 m_bAnnotsLoad = FALSE; | |
| 1617 m_bNeedDownLoadResource = FALSE; | |
| 1618 m_objs_array.RemoveAll(); | |
| 1619 m_ObjectSet.clear(); | |
| 1620 } | |
| 1621 | |
| 1622 if (pdfium::ContainsKey(m_pagesLoadState, dwPage)) | |
| 1623 return DataAvailable; | |
| 1624 | |
| 1625 if (m_bLinearized) { | |
| 1626 if (dwPage == m_dwFirstPageNo) { | |
| 1627 DocAvailStatus nRet = CheckLinearizedFirstPage(dwPage, pHints); | |
| 1628 if (nRet == DataAvailable) | |
| 1629 m_pagesLoadState.insert(dwPage); | |
| 1630 return nRet; | |
| 1631 } | |
| 1632 | |
| 1633 DocAvailStatus nResult = CheckLinearizedData(pHints); | |
| 1634 if (nResult != DataAvailable) | |
| 1635 return nResult; | |
| 1636 | |
| 1637 if (m_pHintTables) { | |
| 1638 nResult = m_pHintTables->CheckPage(dwPage, pHints); | |
| 1639 if (nResult != DataAvailable) | |
| 1640 return nResult; | |
| 1641 m_pagesLoadState.insert(dwPage); | |
| 1642 return DataAvailable; | |
| 1643 } | |
| 1644 | |
| 1645 if (m_bMainXRefLoadedOK) { | |
| 1646 if (m_bTotalLoadPageTree) { | |
| 1647 if (!LoadPages(pHints)) | |
| 1648 return DataNotAvailable; | |
| 1649 } else { | |
| 1650 if (!m_bCurPageDictLoadOK && !CheckPage(dwPage, pHints)) | |
| 1651 return DataNotAvailable; | |
| 1652 } | |
| 1653 } else { | |
| 1654 if (!LoadAllFile(pHints)) | |
| 1655 return DataNotAvailable; | |
| 1656 m_pDocument->GetParser()->RebuildCrossRef(); | |
| 1657 ResetFirstCheck(dwPage); | |
| 1658 return DataAvailable; | |
| 1659 } | |
| 1660 } else { | |
| 1661 if (!m_bTotalLoadPageTree && !m_bCurPageDictLoadOK && | |
| 1662 !CheckPage(dwPage, pHints)) { | |
| 1663 return DataNotAvailable; | |
| 1664 } | |
| 1665 } | |
| 1666 | |
| 1667 if (m_bHaveAcroForm && !m_bAcroFormLoad) { | |
| 1668 if (!CheckAcroFormSubObject(pHints)) | |
| 1669 return DataNotAvailable; | |
| 1670 m_bAcroFormLoad = TRUE; | |
| 1671 } | |
| 1672 | |
| 1673 if (!m_bPageLoadedOK) { | |
| 1674 if (!m_objs_array.GetSize()) { | |
| 1675 m_objs_array.RemoveAll(); | |
| 1676 m_ObjectSet.clear(); | |
| 1677 | |
| 1678 FX_SAFE_INT32 safePage = pdfium::base::checked_cast<int32_t>(dwPage); | |
| 1679 m_pPageDict = m_pDocument->GetPage(safePage.ValueOrDie()); | |
| 1680 if (!m_pPageDict) { | |
| 1681 ResetFirstCheck(dwPage); | |
| 1682 return DataAvailable; | |
| 1683 } | |
| 1684 | |
| 1685 CFX_ArrayTemplate<CPDF_Object*> obj_array; | |
| 1686 obj_array.Add(m_pPageDict); | |
| 1687 FX_BOOL bRet = IsObjectsAvail(obj_array, TRUE, pHints, m_objs_array); | |
| 1688 if (!bRet) | |
| 1689 return DataNotAvailable; | |
| 1690 | |
| 1691 m_objs_array.RemoveAll(); | |
| 1692 } else { | |
| 1693 CFX_ArrayTemplate<CPDF_Object*> new_objs_array; | |
| 1694 FX_BOOL bRet = | |
| 1695 IsObjectsAvail(m_objs_array, FALSE, pHints, new_objs_array); | |
| 1696 | |
| 1697 m_objs_array.RemoveAll(); | |
| 1698 if (!bRet) { | |
| 1699 m_objs_array.Append(new_objs_array); | |
| 1700 return DataNotAvailable; | |
| 1701 } | |
| 1702 } | |
| 1703 m_bPageLoadedOK = TRUE; | |
| 1704 } | |
| 1705 | |
| 1706 if (!m_bAnnotsLoad) { | |
| 1707 if (!CheckPageAnnots(dwPage, pHints)) | |
| 1708 return DataNotAvailable; | |
| 1709 m_bAnnotsLoad = TRUE; | |
| 1710 } | |
| 1711 | |
| 1712 if (m_pPageDict && !m_bNeedDownLoadResource) { | |
| 1713 m_pPageResource = m_pPageDict->GetObjectFor("Resources"); | |
| 1714 m_bNeedDownLoadResource = | |
| 1715 m_pPageResource || HaveResourceAncestor(m_pPageDict); | |
| 1716 } | |
| 1717 | |
| 1718 if (m_bNeedDownLoadResource) { | |
| 1719 if (!CheckResources(pHints)) | |
| 1720 return DataNotAvailable; | |
| 1721 m_bNeedDownLoadResource = FALSE; | |
| 1722 } | |
| 1723 | |
| 1724 m_bPageLoadedOK = FALSE; | |
| 1725 m_bAnnotsLoad = FALSE; | |
| 1726 m_bCurPageDictLoadOK = FALSE; | |
| 1727 | |
| 1728 ResetFirstCheck(dwPage); | |
| 1729 m_pagesLoadState.insert(dwPage); | |
| 1730 return DataAvailable; | |
| 1731 } | |
| 1732 | |
| 1733 FX_BOOL CPDF_DataAvail::CheckResources(DownloadHints* pHints) { | |
| 1734 if (!m_objs_array.GetSize()) { | |
| 1735 m_objs_array.RemoveAll(); | |
| 1736 CFX_ArrayTemplate<CPDF_Object*> obj_array; | |
| 1737 obj_array.Add(m_pPageResource); | |
| 1738 | |
| 1739 FX_BOOL bRet = IsObjectsAvail(obj_array, TRUE, pHints, m_objs_array); | |
| 1740 if (bRet) | |
| 1741 m_objs_array.RemoveAll(); | |
| 1742 return bRet; | |
| 1743 } | |
| 1744 | |
| 1745 CFX_ArrayTemplate<CPDF_Object*> new_objs_array; | |
| 1746 FX_BOOL bRet = IsObjectsAvail(m_objs_array, FALSE, pHints, new_objs_array); | |
| 1747 m_objs_array.RemoveAll(); | |
| 1748 if (!bRet) | |
| 1749 m_objs_array.Append(new_objs_array); | |
| 1750 return bRet; | |
| 1751 } | |
| 1752 | |
| 1753 void CPDF_DataAvail::GetLinearizedMainXRefInfo(FX_FILESIZE* pPos, | |
| 1754 uint32_t* pSize) { | |
| 1755 if (pPos) | |
| 1756 *pPos = m_dwLastXRefOffset; | |
| 1757 if (pSize) | |
| 1758 *pSize = (uint32_t)(m_dwFileLen - m_dwLastXRefOffset); | |
| 1759 } | |
| 1760 | |
| 1761 int CPDF_DataAvail::GetPageCount() const { | |
| 1762 if (m_pLinearized) { | |
| 1763 CPDF_Dictionary* pDict = m_pLinearized->GetDict(); | |
| 1764 CPDF_Object* pObj = pDict ? pDict->GetDirectObjectFor("N") : nullptr; | |
| 1765 return pObj ? pObj->GetInteger() : 0; | |
| 1766 } | |
| 1767 return m_pDocument ? m_pDocument->GetPageCount() : 0; | |
| 1768 } | |
| 1769 | |
| 1770 CPDF_Dictionary* CPDF_DataAvail::GetPage(int index) { | |
| 1771 if (!m_pDocument || index < 0 || index >= GetPageCount()) | |
| 1772 return nullptr; | |
| 1773 | |
| 1774 if (m_pLinearized) { | |
| 1775 CPDF_Dictionary* pDict = m_pLinearized->GetDict(); | |
| 1776 CPDF_Object* pObj = pDict ? pDict->GetDirectObjectFor("P") : nullptr; | |
| 1777 | |
| 1778 int pageNum = pObj ? pObj->GetInteger() : 0; | |
| 1779 if (m_pHintTables && index != pageNum) { | |
| 1780 FX_FILESIZE szPageStartPos = 0; | |
| 1781 FX_FILESIZE szPageLength = 0; | |
| 1782 uint32_t dwObjNum = 0; | |
| 1783 bool bPagePosGot = m_pHintTables->GetPagePos(index, &szPageStartPos, | |
| 1784 &szPageLength, &dwObjNum); | |
| 1785 if (!bPagePosGot) | |
| 1786 return nullptr; | |
| 1787 | |
| 1788 m_syntaxParser.InitParser(m_pFileRead, (uint32_t)szPageStartPos); | |
| 1789 CPDF_Object* pPageDict = ParseIndirectObjectAt(0, dwObjNum, m_pDocument); | |
| 1790 if (!pPageDict) | |
| 1791 return nullptr; | |
| 1792 | |
| 1793 if (!m_pDocument->ReplaceIndirectObjectIfHigherGeneration(dwObjNum, | |
| 1794 pPageDict)) { | |
| 1795 return nullptr; | |
| 1796 } | |
| 1797 return pPageDict->GetDict(); | |
| 1798 } | |
| 1799 } | |
| 1800 return m_pDocument->GetPage(index); | |
| 1801 } | |
| 1802 | |
| 1803 CPDF_DataAvail::DocFormStatus CPDF_DataAvail::IsFormAvail( | |
| 1804 DownloadHints* pHints) { | |
| 1805 if (!m_pDocument) | |
| 1806 return FormAvailable; | |
| 1807 | |
| 1808 if (!m_bLinearizedFormParamLoad) { | |
| 1809 CPDF_Dictionary* pRoot = m_pDocument->GetRoot(); | |
| 1810 if (!pRoot) | |
| 1811 return FormAvailable; | |
| 1812 | |
| 1813 CPDF_Object* pAcroForm = pRoot->GetObjectFor("AcroForm"); | |
| 1814 if (!pAcroForm) | |
| 1815 return FormNotExist; | |
| 1816 | |
| 1817 DocAvailStatus nDocStatus = CheckLinearizedData(pHints); | |
| 1818 if (nDocStatus == DataError) | |
| 1819 return FormError; | |
| 1820 if (nDocStatus == DataNotAvailable) | |
| 1821 return FormNotAvailable; | |
| 1822 | |
| 1823 if (!m_objs_array.GetSize()) | |
| 1824 m_objs_array.Add(pAcroForm->GetDict()); | |
| 1825 m_bLinearizedFormParamLoad = TRUE; | |
| 1826 } | |
| 1827 | |
| 1828 CFX_ArrayTemplate<CPDF_Object*> new_objs_array; | |
| 1829 FX_BOOL bRet = IsObjectsAvail(m_objs_array, FALSE, pHints, new_objs_array); | |
| 1830 m_objs_array.RemoveAll(); | |
| 1831 if (!bRet) { | |
| 1832 m_objs_array.Append(new_objs_array); | |
| 1833 return FormNotAvailable; | |
| 1834 } | |
| 1835 return FormAvailable; | |
| 1836 } | |
| 1837 | |
| 1838 CPDF_DataAvail::PageNode::PageNode() : m_type(PDF_PAGENODE_UNKNOWN) {} | |
| 1839 | |
| 1840 CPDF_DataAvail::PageNode::~PageNode() { | |
| 1841 for (int32_t i = 0; i < m_childNode.GetSize(); ++i) | |
| 1842 delete m_childNode[i]; | |
| 1843 m_childNode.RemoveAll(); | |
| 1844 } | |
| OLD | NEW |
|---|
Chromium Code Reviews
Issue 2392603004:
Move core/fpdfapi/fpdf_parser to core/fpdfapi/parser (Closed)
