Chromium Code Reviews
chromiumcodereview-hr@appspot.gserviceaccount.com (chromiumcodereview-hr) | Please choose your nickname with Settings | Help | Chromium Project | Gerrit Changes | Sign out
(92)

Side by Side Diff: core/fpdfapi/fpdf_parser/cpdf_hint_tables.cpp

Issue 2392603004: Move core/fpdfapi/fpdf_parser to core/fpdfapi/parser (Closed)
Patch Set: Rebase to master Created 4 years, 2 months ago
Use n/p to move between diff chunks; N/P to move between comments. Draft comments are only viewable by you.
Jump to:
View unified diff | Download patch
OLDNEW
(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_hint_tables.h"
8
9 #include <limits>
10
11 #include "core/fpdfapi/fpdf_parser/cpdf_array.h"
12 #include "core/fpdfapi/fpdf_parser/cpdf_data_avail.h"
13 #include "core/fpdfapi/fpdf_parser/cpdf_dictionary.h"
14 #include "core/fpdfapi/fpdf_parser/cpdf_document.h"
15 #include "core/fpdfapi/fpdf_parser/cpdf_stream.h"
16 #include "core/fpdfapi/fpdf_parser/cpdf_stream_acc.h"
17 #include "core/fxcrt/fx_safe_types.h"
18 #include "third_party/base/numerics/safe_conversions.h"
19
20 namespace {
21
22 bool CanReadFromBitStream(const CFX_BitStream* hStream,
23 const FX_SAFE_UINT32& bits) {
24 return bits.IsValid() && hStream->BitsRemaining() >= bits.ValueOrDie();
25 }
26
27 // Sanity check values from the page table header. The note in the PDF 1.7
28 // reference for Table F.3 says the valid range is only 0 through 32. Though 0
29 // is not useful either.
30 bool IsValidPageOffsetHintTableBitCount(uint32_t bits) {
31 return bits > 0 && bits <= 32;
32 }
33
34 } // namespace
35
36 CPDF_HintTables::CPDF_HintTables(CPDF_DataAvail* pDataAvail,
37 CPDF_Dictionary* pLinearized)
38 : m_pDataAvail(pDataAvail),
39 m_pLinearizedDict(pLinearized),
40 m_nFirstPageSharedObjs(0),
41 m_szFirstPageObjOffset(0) {
42 ASSERT(m_pLinearizedDict);
43 }
44
45 CPDF_HintTables::~CPDF_HintTables() {}
46
47 uint32_t CPDF_HintTables::GetItemLength(
48 uint32_t index,
49 const std::vector<FX_FILESIZE>& szArray) {
50 if (szArray.size() < 2 || index > szArray.size() - 2 ||
51 szArray[index] > szArray[index + 1]) {
52 return 0;
53 }
54 return szArray[index + 1] - szArray[index];
55 }
56
57 bool CPDF_HintTables::ReadPageHintTable(CFX_BitStream* hStream) {
58 if (!hStream || hStream->IsEOF())
59 return false;
60
61 int nStreamOffset = ReadPrimaryHintStreamOffset();
62 if (nStreamOffset < 0)
63 return false;
64
65 int nStreamLen = ReadPrimaryHintStreamLength();
66 if (nStreamLen < 1 ||
67 !pdfium::base::IsValueInRangeForNumericType<FX_FILESIZE>(nStreamLen)) {
68 return false;
69 }
70
71 const uint32_t kHeaderSize = 288;
72 if (hStream->BitsRemaining() < kHeaderSize)
73 return false;
74
75 // Item 1: The least number of objects in a page.
76 const uint32_t dwObjLeastNum = hStream->GetBits(32);
77 if (!dwObjLeastNum)
78 return false;
79
80 // Item 2: The location of the first page's page object.
81 const uint32_t dwFirstObjLoc = hStream->GetBits(32);
82 if (dwFirstObjLoc > static_cast<uint32_t>(nStreamOffset)) {
83 FX_SAFE_FILESIZE safeLoc = nStreamLen;
84 safeLoc += dwFirstObjLoc;
85 if (!safeLoc.IsValid())
86 return false;
87 m_szFirstPageObjOffset = safeLoc.ValueOrDie();
88 } else {
89 if (!pdfium::base::IsValueInRangeForNumericType<FX_FILESIZE>(dwFirstObjLoc))
90 return false;
91 m_szFirstPageObjOffset = dwFirstObjLoc;
92 }
93
94 // Item 3: The number of bits needed to represent the difference
95 // between the greatest and least number of objects in a page.
96 const uint32_t dwDeltaObjectsBits = hStream->GetBits(16);
97 if (!IsValidPageOffsetHintTableBitCount(dwDeltaObjectsBits))
98 return false;
99
100 // Item 4: The least length of a page in bytes.
101 const uint32_t dwPageLeastLen = hStream->GetBits(32);
102 if (!dwPageLeastLen)
103 return false;
104
105 // Item 5: The number of bits needed to represent the difference
106 // between the greatest and least length of a page, in bytes.
107 const uint32_t dwDeltaPageLenBits = hStream->GetBits(16);
108 if (!IsValidPageOffsetHintTableBitCount(dwDeltaPageLenBits))
109 return false;
110
111 // Skip Item 6, 7, 8, 9 total 96 bits.
112 hStream->SkipBits(96);
113
114 // Item 10: The number of bits needed to represent the greatest
115 // number of shared object references.
116 const uint32_t dwSharedObjBits = hStream->GetBits(16);
117 if (!IsValidPageOffsetHintTableBitCount(dwSharedObjBits))
118 return false;
119
120 // Item 11: The number of bits needed to represent the numerically
121 // greatest shared object identifier used by the pages.
122 const uint32_t dwSharedIdBits = hStream->GetBits(16);
123 if (!IsValidPageOffsetHintTableBitCount(dwSharedIdBits))
124 return false;
125
126 // Item 12: The number of bits needed to represent the numerator of
127 // the fractional position for each shared object reference. For each
128 // shared object referenced from a page, there is an indication of
129 // where in the page's content stream the object is first referenced.
130 const uint32_t dwSharedNumeratorBits = hStream->GetBits(16);
131 if (!IsValidPageOffsetHintTableBitCount(dwSharedNumeratorBits))
132 return false;
133
134 // Item 13: Skip Item 13 which has 16 bits.
135 hStream->SkipBits(16);
136
137 const int nPages = GetNumberOfPages();
138 if (nPages < 1 || nPages >= FPDF_PAGE_MAX_NUM)
139 return false;
140
141 const uint32_t dwPages = pdfium::base::checked_cast<uint32_t>(nPages);
142 FX_SAFE_UINT32 required_bits = dwDeltaObjectsBits;
143 required_bits *= dwPages;
144 if (!CanReadFromBitStream(hStream, required_bits))
145 return false;
146
147 for (int i = 0; i < nPages; ++i) {
148 FX_SAFE_UINT32 safeDeltaObj = hStream->GetBits(dwDeltaObjectsBits);
149 safeDeltaObj += dwObjLeastNum;
150 if (!safeDeltaObj.IsValid())
151 return false;
152 m_dwDeltaNObjsArray.push_back(safeDeltaObj.ValueOrDie());
153 }
154 hStream->ByteAlign();
155
156 required_bits = dwDeltaPageLenBits;
157 required_bits *= dwPages;
158 if (!CanReadFromBitStream(hStream, required_bits))
159 return false;
160
161 std::vector<uint32_t> dwPageLenArray;
162 for (int i = 0; i < nPages; ++i) {
163 FX_SAFE_UINT32 safePageLen = hStream->GetBits(dwDeltaPageLenBits);
164 safePageLen += dwPageLeastLen;
165 if (!safePageLen.IsValid())
166 return false;
167
168 dwPageLenArray.push_back(safePageLen.ValueOrDie());
169 }
170
171 int nOffsetE = GetEndOfFirstPageOffset();
172 if (nOffsetE < 0)
173 return false;
174
175 int nFirstPageNum = GetFirstPageNumber();
176 if (nFirstPageNum < 0 || nFirstPageNum > std::numeric_limits<int>::max() - 1)
177 return false;
178
179 for (int i = 0; i < nPages; ++i) {
180 if (i == nFirstPageNum) {
181 m_szPageOffsetArray.push_back(m_szFirstPageObjOffset);
182 } else if (i == nFirstPageNum + 1) {
183 if (i == 1) {
184 m_szPageOffsetArray.push_back(nOffsetE);
185 } else {
186 m_szPageOffsetArray.push_back(m_szPageOffsetArray[i - 2] +
187 dwPageLenArray[i - 2]);
188 }
189 } else {
190 if (i == 0) {
191 m_szPageOffsetArray.push_back(nOffsetE);
192 } else {
193 m_szPageOffsetArray.push_back(m_szPageOffsetArray[i - 1] +
194 dwPageLenArray[i - 1]);
195 }
196 }
197 }
198
199 m_szPageOffsetArray.push_back(m_szPageOffsetArray[nPages - 1] +
200 dwPageLenArray[nPages - 1]);
201 hStream->ByteAlign();
202
203 // Number of shared objects.
204 required_bits = dwSharedObjBits;
205 required_bits *= dwPages;
206 if (!CanReadFromBitStream(hStream, required_bits))
207 return false;
208
209 for (int i = 0; i < nPages; i++)
210 m_dwNSharedObjsArray.push_back(hStream->GetBits(dwSharedObjBits));
211 hStream->ByteAlign();
212
213 // Array of identifiers, size = nshared_objects.
214 for (int i = 0; i < nPages; i++) {
215 required_bits = dwSharedIdBits;
216 required_bits *= m_dwNSharedObjsArray[i];
217 if (!CanReadFromBitStream(hStream, required_bits))
218 return false;
219
220 for (uint32_t j = 0; j < m_dwNSharedObjsArray[i]; j++)
221 m_dwIdentifierArray.push_back(hStream->GetBits(dwSharedIdBits));
222 }
223 hStream->ByteAlign();
224
225 for (int i = 0; i < nPages; i++) {
226 FX_SAFE_UINT32 safeSize = m_dwNSharedObjsArray[i];
227 safeSize *= dwSharedNumeratorBits;
228 if (!CanReadFromBitStream(hStream, safeSize))
229 return false;
230
231 hStream->SkipBits(safeSize.ValueOrDie());
232 }
233 hStream->ByteAlign();
234
235 FX_SAFE_UINT32 safeTotalPageLen = dwPages;
236 safeTotalPageLen *= dwDeltaPageLenBits;
237 if (!CanReadFromBitStream(hStream, safeTotalPageLen))
238 return false;
239
240 hStream->SkipBits(safeTotalPageLen.ValueOrDie());
241 hStream->ByteAlign();
242 return true;
243 }
244
245 bool CPDF_HintTables::ReadSharedObjHintTable(CFX_BitStream* hStream,
246 uint32_t offset) {
247 if (!hStream || hStream->IsEOF())
248 return false;
249
250 int nStreamOffset = ReadPrimaryHintStreamOffset();
251 int nStreamLen = ReadPrimaryHintStreamLength();
252 if (nStreamOffset < 0 || nStreamLen < 1)
253 return false;
254
255 FX_SAFE_UINT32 bit_offset = offset;
256 bit_offset *= 8;
257 if (!bit_offset.IsValid() || hStream->GetPos() > bit_offset.ValueOrDie())
258 return false;
259 hStream->SkipBits(bit_offset.ValueOrDie() - hStream->GetPos());
260
261 const uint32_t kHeaderSize = 192;
262 if (hStream->BitsRemaining() < kHeaderSize)
263 return false;
264
265 // Item 1: The object number of the first object in the shared objects
266 // section.
267 uint32_t dwFirstSharedObjNum = hStream->GetBits(32);
268
269 // Item 2: The location of the first object in the shared objects section.
270 uint32_t dwFirstSharedObjLoc = hStream->GetBits(32);
271 if (dwFirstSharedObjLoc > static_cast<uint32_t>(nStreamOffset))
272 dwFirstSharedObjLoc += nStreamLen;
273
274 // Item 3: The number of shared object entries for the first page.
275 m_nFirstPageSharedObjs = hStream->GetBits(32);
276
277 // Item 4: The number of shared object entries for the shared objects
278 // section, including the number of shared object entries for the first page.
279 uint32_t dwSharedObjTotal = hStream->GetBits(32);
280
281 // Item 5: The number of bits needed to represent the greatest number of
282 // objects in a shared object group. Skipped.
283 hStream->SkipBits(16);
284
285 // Item 6: The least length of a shared object group in bytes.
286 uint32_t dwGroupLeastLen = hStream->GetBits(32);
287
288 // Item 7: The number of bits needed to represent the difference between the
289 // greatest and least length of a shared object group, in bytes.
290 uint32_t dwDeltaGroupLen = hStream->GetBits(16);
291
292 if (dwFirstSharedObjNum >= CPDF_Parser::kMaxObjectNumber ||
293 m_nFirstPageSharedObjs >= CPDF_Parser::kMaxObjectNumber ||
294 dwSharedObjTotal >= CPDF_Parser::kMaxObjectNumber) {
295 return false;
296 }
297
298 int nFirstPageObjNum = GetFirstPageObjectNumber();
299 if (nFirstPageObjNum < 0)
300 return false;
301
302 uint32_t dwPrevObjLen = 0;
303 uint32_t dwCurObjLen = 0;
304 FX_SAFE_UINT32 required_bits = dwSharedObjTotal;
305 required_bits *= dwDeltaGroupLen;
306 if (!CanReadFromBitStream(hStream, required_bits))
307 return false;
308
309 for (uint32_t i = 0; i < dwSharedObjTotal; ++i) {
310 dwPrevObjLen = dwCurObjLen;
311 FX_SAFE_UINT32 safeObjLen = hStream->GetBits(dwDeltaGroupLen);
312 safeObjLen += dwGroupLeastLen;
313 if (!safeObjLen.IsValid())
314 return false;
315
316 dwCurObjLen = safeObjLen.ValueOrDie();
317 if (i < m_nFirstPageSharedObjs) {
318 m_dwSharedObjNumArray.push_back(nFirstPageObjNum + i);
319 if (i == 0)
320 m_szSharedObjOffsetArray.push_back(m_szFirstPageObjOffset);
321 } else {
322 FX_SAFE_UINT32 safeObjNum = dwFirstSharedObjNum;
323 safeObjNum += i - m_nFirstPageSharedObjs;
324 if (!safeObjNum.IsValid())
325 return false;
326
327 m_dwSharedObjNumArray.push_back(safeObjNum.ValueOrDie());
328 if (i == m_nFirstPageSharedObjs) {
329 FX_SAFE_FILESIZE safeLoc = dwFirstSharedObjLoc;
330 if (!safeLoc.IsValid())
331 return false;
332
333 m_szSharedObjOffsetArray.push_back(safeLoc.ValueOrDie());
334 }
335 }
336
337 if (i != 0 && i != m_nFirstPageSharedObjs) {
338 FX_SAFE_FILESIZE safeLoc = dwPrevObjLen;
339 safeLoc += m_szSharedObjOffsetArray[i - 1];
340 if (!safeLoc.IsValid())
341 return false;
342
343 m_szSharedObjOffsetArray.push_back(safeLoc.ValueOrDie());
344 }
345 }
346
347 if (dwSharedObjTotal > 0) {
348 FX_SAFE_FILESIZE safeLoc = dwCurObjLen;
349 safeLoc += m_szSharedObjOffsetArray[dwSharedObjTotal - 1];
350 if (!safeLoc.IsValid())
351 return false;
352
353 m_szSharedObjOffsetArray.push_back(safeLoc.ValueOrDie());
354 }
355
356 hStream->ByteAlign();
357 if (hStream->BitsRemaining() < dwSharedObjTotal)
358 return false;
359
360 hStream->SkipBits(dwSharedObjTotal);
361 hStream->ByteAlign();
362 return true;
363 }
364
365 bool CPDF_HintTables::GetPagePos(uint32_t index,
366 FX_FILESIZE* szPageStartPos,
367 FX_FILESIZE* szPageLength,
368 uint32_t* dwObjNum) {
369 *szPageStartPos = m_szPageOffsetArray[index];
370 *szPageLength = GetItemLength(index, m_szPageOffsetArray);
371
372 int nFirstPageObjNum = GetFirstPageObjectNumber();
373 if (nFirstPageObjNum < 0)
374 return false;
375
376 int nFirstPageNum = GetFirstPageNumber();
377 if (!pdfium::base::IsValueInRangeForNumericType<uint32_t>(nFirstPageNum))
378 return false;
379
380 uint32_t dwFirstPageNum = static_cast<uint32_t>(nFirstPageNum);
381 if (index == dwFirstPageNum) {
382 *dwObjNum = nFirstPageObjNum;
383 return true;
384 }
385
386 // The object number of remaining pages starts from 1.
387 *dwObjNum = 1;
388 for (uint32_t i = 0; i < index; ++i) {
389 if (i == dwFirstPageNum)
390 continue;
391 *dwObjNum += m_dwDeltaNObjsArray[i];
392 }
393 return true;
394 }
395
396 CPDF_DataAvail::DocAvailStatus CPDF_HintTables::CheckPage(
397 uint32_t index,
398 CPDF_DataAvail::DownloadHints* pHints) {
399 if (!pHints)
400 return CPDF_DataAvail::DataError;
401
402 int nFirstPageNum = GetFirstPageNumber();
403 if (!pdfium::base::IsValueInRangeForNumericType<uint32_t>(nFirstPageNum))
404 return CPDF_DataAvail::DataError;
405
406 if (index == static_cast<uint32_t>(nFirstPageNum))
407 return CPDF_DataAvail::DataAvailable;
408
409 uint32_t dwLength = GetItemLength(index, m_szPageOffsetArray);
410 // If two pages have the same offset, it should be treated as an error.
411 if (!dwLength)
412 return CPDF_DataAvail::DataError;
413
414 if (!m_pDataAvail->IsDataAvail(m_szPageOffsetArray[index], dwLength, pHints))
415 return CPDF_DataAvail::DataNotAvailable;
416
417 // Download data of shared objects in the page.
418 uint32_t offset = 0;
419 for (uint32_t i = 0; i < index; ++i)
420 offset += m_dwNSharedObjsArray[i];
421
422 int nFirstPageObjNum = GetFirstPageObjectNumber();
423 if (nFirstPageObjNum < 0)
424 return CPDF_DataAvail::DataError;
425
426 uint32_t dwIndex = 0;
427 uint32_t dwObjNum = 0;
428 for (uint32_t j = 0; j < m_dwNSharedObjsArray[index]; ++j) {
429 dwIndex = m_dwIdentifierArray[offset + j];
430 if (dwIndex >= m_dwSharedObjNumArray.size())
431 return CPDF_DataAvail::DataNotAvailable;
432
433 dwObjNum = m_dwSharedObjNumArray[dwIndex];
434 if (dwObjNum >= static_cast<uint32_t>(nFirstPageObjNum) &&
435 dwObjNum <
436 static_cast<uint32_t>(nFirstPageObjNum) + m_nFirstPageSharedObjs) {
437 continue;
438 }
439
440 dwLength = GetItemLength(dwIndex, m_szSharedObjOffsetArray);
441 // If two objects have the same offset, it should be treated as an error.
442 if (!dwLength)
443 return CPDF_DataAvail::DataError;
444
445 if (!m_pDataAvail->IsDataAvail(m_szSharedObjOffsetArray[dwIndex], dwLength,
446 pHints)) {
447 return CPDF_DataAvail::DataNotAvailable;
448 }
449 }
450 return CPDF_DataAvail::DataAvailable;
451 }
452
453 bool CPDF_HintTables::LoadHintStream(CPDF_Stream* pHintStream) {
454 if (!pHintStream)
455 return false;
456
457 CPDF_Dictionary* pDict = pHintStream->GetDict();
458 CPDF_Object* pOffset = pDict ? pDict->GetObjectFor("S") : nullptr;
459 if (!pOffset || !pOffset->IsNumber())
460 return false;
461
462 int shared_hint_table_offset = pOffset->GetInteger();
463 if (shared_hint_table_offset <= 0)
464 return false;
465
466 CPDF_StreamAcc acc;
467 acc.LoadAllData(pHintStream);
468
469 uint32_t size = acc.GetSize();
470 // The header section of page offset hint table is 36 bytes.
471 // The header section of shared object hint table is 24 bytes.
472 // Hint table has at least 60 bytes.
473 const uint32_t kMinStreamLength = 60;
474 if (size < kMinStreamLength)
475 return false;
476
477 FX_SAFE_UINT32 safe_shared_hint_table_offset = shared_hint_table_offset;
478 if (!safe_shared_hint_table_offset.IsValid() ||
479 size < safe_shared_hint_table_offset.ValueOrDie()) {
480 return false;
481 }
482
483 CFX_BitStream bs;
484 bs.Init(acc.GetData(), size);
485 return ReadPageHintTable(&bs) &&
486 ReadSharedObjHintTable(&bs, shared_hint_table_offset);
487 }
488
489 int CPDF_HintTables::GetEndOfFirstPageOffset() const {
490 CPDF_Object* pOffsetE = m_pLinearizedDict->GetDirectObjectFor("E");
491 return pOffsetE ? pOffsetE->GetInteger() : -1;
492 }
493
494 int CPDF_HintTables::GetNumberOfPages() const {
495 CPDF_Object* pPageNum = m_pLinearizedDict->GetDirectObjectFor("N");
496 return pPageNum ? pPageNum->GetInteger() : 0;
497 }
498
499 int CPDF_HintTables::GetFirstPageObjectNumber() const {
500 CPDF_Object* pFirstPageObj = m_pLinearizedDict->GetDirectObjectFor("O");
501 return pFirstPageObj ? pFirstPageObj->GetInteger() : -1;
502 }
503
504 int CPDF_HintTables::GetFirstPageNumber() const {
505 CPDF_Object* pFirstPageNum = m_pLinearizedDict->GetDirectObjectFor("P");
506 return pFirstPageNum ? pFirstPageNum->GetInteger() : 0;
507 }
508
509 int CPDF_HintTables::ReadPrimaryHintStreamOffset() const {
510 return ReadPrimaryHintStream(0);
511 }
512
513 int CPDF_HintTables::ReadPrimaryHintStreamLength() const {
514 return ReadPrimaryHintStream(1);
515 }
516
517 int CPDF_HintTables::ReadPrimaryHintStream(int index) const {
518 CPDF_Array* pRange = m_pLinearizedDict->GetArrayFor("H");
519 if (!pRange)
520 return -1;
521
522 CPDF_Object* pStreamLen = pRange->GetDirectObjectAt(index);
523 return pStreamLen ? pStreamLen->GetInteger() : -1;
524 }
OLDNEW
« no previous file with comments | « core/fpdfapi/fpdf_parser/cpdf_hint_tables.h ('k') | core/fpdfapi/fpdf_parser/cpdf_indirect_object_holder.h » ('j') | no next file with comments »

Powered by Google App Engine
This is Rietveld 408576698