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 "public/fpdf_flatten.h"
8
9#include <algorithm>
10
11#include "fpdfsdk/include/fsdk_define.h"
12
13typedef CFX_ArrayTemplate<CPDF_Dictionary*> CPDF_ObjectArray;
14typedef CFX_ArrayTemplate<CPDF_Rect> CPDF_RectArray;
15
16enum FPDF_TYPE { MAX, MIN };
17enum FPDF_VALUE { TOP, LEFT, RIGHT, BOTTOM };
18
19FX_BOOL IsValiableRect(CPDF_Rect rect, CPDF_Rect rcPage) {
20  if (rect.left - rect.right > 0.000001f || rect.bottom - rect.top > 0.000001f)
21    return FALSE;
22
23  if (rect.left == 0.0f && rect.top == 0.0f && rect.right == 0.0f &&
24      rect.bottom == 0.0f)
25    return FALSE;
26
27  if (!rcPage.IsEmpty()) {
28    if (rect.left - rcPage.left < -10.000001f ||
29        rect.right - rcPage.right > 10.000001f ||
30        rect.top - rcPage.top > 10.000001f ||
31        rect.bottom - rcPage.bottom < -10.000001f)
32      return FALSE;
33  }
34
35  return TRUE;
36}
37
38FX_BOOL GetContentsRect(CPDF_Document* pDoc,
39                        CPDF_Dictionary* pDict,
40                        CPDF_RectArray* pRectArray) {
41  CPDF_Page* pPDFPage = new CPDF_Page;
42  pPDFPage->Load(pDoc, pDict, FALSE);
43  pPDFPage->ParseContent();
44
45  FX_POSITION pos = pPDFPage->GetFirstObjectPosition();
46
47  while (pos) {
48    CPDF_PageObject* pPageObject = pPDFPage->GetNextObject(pos);
49    if (!pPageObject)
50      continue;
51
52    CPDF_Rect rc;
53    rc.left = pPageObject->m_Left;
54    rc.right = pPageObject->m_Right;
55    rc.bottom = pPageObject->m_Bottom;
56    rc.top = pPageObject->m_Top;
57
58    if (IsValiableRect(rc, pDict->GetRect("MediaBox"))) {
59      pRectArray->Add(rc);
60    }
61  }
62
63  delete pPDFPage;
64  return TRUE;
65}
66
67void ParserStream(CPDF_Dictionary* pPageDic,
68                  CPDF_Dictionary* pStream,
69                  CPDF_RectArray* pRectArray,
70                  CPDF_ObjectArray* pObjectArray) {
71  if (!pStream)
72    return;
73  CPDF_Rect rect;
74  if (pStream->KeyExist("Rect"))
75    rect = pStream->GetRect("Rect");
76  else if (pStream->KeyExist("BBox"))
77    rect = pStream->GetRect("BBox");
78
79  if (IsValiableRect(rect, pPageDic->GetRect("MediaBox")))
80    pRectArray->Add(rect);
81
82  pObjectArray->Add(pStream);
83}
84
85int ParserAnnots(CPDF_Document* pSourceDoc,
86                 CPDF_Dictionary* pPageDic,
87                 CPDF_RectArray* pRectArray,
88                 CPDF_ObjectArray* pObjectArray,
89                 int nUsage) {
90  if (!pSourceDoc || !pPageDic)
91    return FLATTEN_FAIL;
92
93  GetContentsRect(pSourceDoc, pPageDic, pRectArray);
94  CPDF_Array* pAnnots = pPageDic->GetArray("Annots");
95  if (!pAnnots)
96    return FLATTEN_NOTHINGTODO;
97
98  FX_DWORD dwSize = pAnnots->GetCount();
99  for (int i = 0; i < (int)dwSize; i++) {
100    CPDF_Dictionary* pAnnotDic = ToDictionary(pAnnots->GetElementValue(i));
101    if (!pAnnotDic)
102      continue;
103
104    CFX_ByteString sSubtype = pAnnotDic->GetString("Subtype");
105    if (sSubtype == "Popup")
106      continue;
107
108    int nAnnotFlag = pAnnotDic->GetInteger("F");
109    if (nAnnotFlag & ANNOTFLAG_HIDDEN)
110      continue;
111
112    if (nUsage == FLAT_NORMALDISPLAY) {
113      if (nAnnotFlag & ANNOTFLAG_INVISIBLE)
114        continue;
115
116      ParserStream(pPageDic, pAnnotDic, pRectArray, pObjectArray);
117    } else {
118      if (nAnnotFlag & ANNOTFLAG_PRINT)
119        ParserStream(pPageDic, pAnnotDic, pRectArray, pObjectArray);
120    }
121  }
122  return FLATTEN_SUCCESS;
123}
124
125FX_FLOAT GetMinMaxValue(CPDF_RectArray& array,
126                        FPDF_TYPE type,
127                        FPDF_VALUE value) {
128  int nRects = array.GetSize();
129  FX_FLOAT fRet = 0.0f;
130
131  if (nRects <= 0)
132    return 0.0f;
133
134  FX_FLOAT* pArray = new FX_FLOAT[nRects];
135  switch (value) {
136    case LEFT: {
137      for (int i = 0; i < nRects; i++)
138        pArray[i] = CPDF_Rect(array.GetAt(i)).left;
139
140      break;
141    }
142    case TOP: {
143      for (int i = 0; i < nRects; i++)
144        pArray[i] = CPDF_Rect(array.GetAt(i)).top;
145
146      break;
147    }
148    case RIGHT: {
149      for (int i = 0; i < nRects; i++)
150        pArray[i] = CPDF_Rect(array.GetAt(i)).right;
151
152      break;
153    }
154    case BOTTOM: {
155      for (int i = 0; i < nRects; i++)
156        pArray[i] = CPDF_Rect(array.GetAt(i)).bottom;
157
158      break;
159    }
160    default:
161      break;
162  }
163  fRet = pArray[0];
164  if (type == MAX) {
165    for (int i = 1; i < nRects; i++)
166      if (fRet <= pArray[i])
167        fRet = pArray[i];
168  } else {
169    for (int i = 1; i < nRects; i++)
170      if (fRet >= pArray[i])
171        fRet = pArray[i];
172  }
173  delete[] pArray;
174  return fRet;
175}
176
177CPDF_Rect CalculateRect(CPDF_RectArray* pRectArray) {
178  CPDF_Rect rcRet;
179
180  rcRet.left = GetMinMaxValue(*pRectArray, MIN, LEFT);
181  rcRet.top = GetMinMaxValue(*pRectArray, MAX, TOP);
182  rcRet.right = GetMinMaxValue(*pRectArray, MAX, RIGHT);
183  rcRet.bottom = GetMinMaxValue(*pRectArray, MIN, BOTTOM);
184
185  return rcRet;
186}
187
188void SetPageContents(CFX_ByteString key,
189                     CPDF_Dictionary* pPage,
190                     CPDF_Document* pDocument) {
191  CPDF_Object* pContentsObj = pPage->GetStream("Contents");
192  if (!pContentsObj) {
193    pContentsObj = pPage->GetArray("Contents");
194  }
195
196  if (!pContentsObj) {
197    // Create a new contents dictionary
198    if (!key.IsEmpty()) {
199      CPDF_Stream* pNewContents = new CPDF_Stream(NULL, 0, new CPDF_Dictionary);
200      pPage->SetAtReference("Contents", pDocument,
201                            pDocument->AddIndirectObject(pNewContents));
202
203      CFX_ByteString sStream;
204      sStream.Format("q 1 0 0 1 0 0 cm /%s Do Q", key.c_str());
205      pNewContents->SetData((const uint8_t*)sStream, sStream.GetLength(), FALSE,
206                            FALSE);
207    }
208    return;
209  }
210
211  int iType = pContentsObj->GetType();
212  CPDF_Array* pContentsArray = NULL;
213
214  switch (iType) {
215    case PDFOBJ_STREAM: {
216      pContentsArray = new CPDF_Array;
217      CPDF_Stream* pContents = pContentsObj->AsStream();
218      FX_DWORD dwObjNum = pDocument->AddIndirectObject(pContents);
219      CPDF_StreamAcc acc;
220      acc.LoadAllData(pContents);
221      CFX_ByteString sStream = "q\n";
222      CFX_ByteString sBody =
223          CFX_ByteString((const FX_CHAR*)acc.GetData(), acc.GetSize());
224      sStream = sStream + sBody + "\nQ";
225      pContents->SetData((const uint8_t*)sStream, sStream.GetLength(), FALSE,
226                         FALSE);
227      pContentsArray->AddReference(pDocument, dwObjNum);
228      break;
229    }
230
231    case PDFOBJ_ARRAY: {
232      pContentsArray = pContentsObj->AsArray();
233      break;
234    }
235    default:
236      break;
237  }
238
239  if (!pContentsArray)
240    return;
241
242  FX_DWORD dwObjNum = pDocument->AddIndirectObject(pContentsArray);
243  pPage->SetAtReference("Contents", pDocument, dwObjNum);
244
245  if (!key.IsEmpty()) {
246    CPDF_Stream* pNewContents = new CPDF_Stream(NULL, 0, new CPDF_Dictionary);
247    dwObjNum = pDocument->AddIndirectObject(pNewContents);
248    pContentsArray->AddReference(pDocument, dwObjNum);
249
250    CFX_ByteString sStream;
251    sStream.Format("q 1 0 0 1 0 0 cm /%s Do Q", key.c_str());
252    pNewContents->SetData((const uint8_t*)sStream, sStream.GetLength(), FALSE,
253                          FALSE);
254  }
255}
256
257CFX_Matrix GetMatrix(CPDF_Rect rcAnnot,
258                     CPDF_Rect rcStream,
259                     const CFX_Matrix& matrix) {
260  if (rcStream.IsEmpty())
261    return CFX_Matrix();
262
263  matrix.TransformRect(rcStream);
264  rcStream.Normalize();
265
266  FX_FLOAT a = rcAnnot.Width() / rcStream.Width();
267  FX_FLOAT d = rcAnnot.Height() / rcStream.Height();
268
269  FX_FLOAT e = rcAnnot.left - rcStream.left * a;
270  FX_FLOAT f = rcAnnot.bottom - rcStream.bottom * d;
271  return CFX_Matrix(a, 0, 0, d, e, f);
272}
273
274void GetOffset(FX_FLOAT& fa,
275               FX_FLOAT& fd,
276               FX_FLOAT& fe,
277               FX_FLOAT& ff,
278               CPDF_Rect rcAnnot,
279               CPDF_Rect rcStream,
280               const CFX_Matrix& matrix) {
281  FX_FLOAT fStreamWidth = 0.0f;
282  FX_FLOAT fStreamHeight = 0.0f;
283
284  if (matrix.a != 0 && matrix.d != 0) {
285    fStreamWidth = rcStream.right - rcStream.left;
286    fStreamHeight = rcStream.top - rcStream.bottom;
287  } else {
288    fStreamWidth = rcStream.top - rcStream.bottom;
289    fStreamHeight = rcStream.right - rcStream.left;
290  }
291
292  FX_FLOAT x1 =
293      matrix.a * rcStream.left + matrix.c * rcStream.bottom + matrix.e;
294  FX_FLOAT y1 =
295      matrix.b * rcStream.left + matrix.d * rcStream.bottom + matrix.f;
296  FX_FLOAT x2 = matrix.a * rcStream.left + matrix.c * rcStream.top + matrix.e;
297  FX_FLOAT y2 = matrix.b * rcStream.left + matrix.d * rcStream.top + matrix.f;
298  FX_FLOAT x3 =
299      matrix.a * rcStream.right + matrix.c * rcStream.bottom + matrix.e;
300  FX_FLOAT y3 =
301      matrix.b * rcStream.right + matrix.d * rcStream.bottom + matrix.f;
302  FX_FLOAT x4 = matrix.a * rcStream.right + matrix.c * rcStream.top + matrix.e;
303  FX_FLOAT y4 = matrix.b * rcStream.right + matrix.d * rcStream.top + matrix.f;
304
305  FX_FLOAT left = std::min(std::min(x1, x2), std::min(x3, x4));
306  FX_FLOAT bottom = std::min(std::min(y1, y2), std::min(y3, y4));
307
308  fa = (rcAnnot.right - rcAnnot.left) / fStreamWidth;
309  fd = (rcAnnot.top - rcAnnot.bottom) / fStreamHeight;
310  fe = rcAnnot.left - left * fa;
311  ff = rcAnnot.bottom - bottom * fd;
312}
313
314DLLEXPORT int STDCALL FPDFPage_Flatten(FPDF_PAGE page, int nFlag) {
315  CPDF_Page* pPage = CPDFPageFromFPDFPage(page);
316  if (!page) {
317    return FLATTEN_FAIL;
318  }
319
320  CPDF_Document* pDocument = pPage->m_pDocument;
321  CPDF_Dictionary* pPageDict = pPage->m_pFormDict;
322
323  if (!pDocument || !pPageDict) {
324    return FLATTEN_FAIL;
325  }
326
327  CPDF_ObjectArray ObjectArray;
328  CPDF_RectArray RectArray;
329
330  int iRet = FLATTEN_FAIL;
331  iRet = ParserAnnots(pDocument, pPageDict, &RectArray, &ObjectArray, nFlag);
332  if (iRet == FLATTEN_NOTHINGTODO || iRet == FLATTEN_FAIL)
333    return iRet;
334
335  CPDF_Rect rcOriginalCB;
336  CPDF_Rect rcMerger = CalculateRect(&RectArray);
337  CPDF_Rect rcOriginalMB = pPageDict->GetRect("MediaBox");
338
339  if (pPageDict->KeyExist("CropBox"))
340    rcOriginalMB = pPageDict->GetRect("CropBox");
341
342  if (rcOriginalMB.IsEmpty()) {
343    rcOriginalMB = CPDF_Rect(0.0f, 0.0f, 612.0f, 792.0f);
344  }
345
346  rcMerger.left =
347      rcMerger.left < rcOriginalMB.left ? rcOriginalMB.left : rcMerger.left;
348  rcMerger.right =
349      rcMerger.right > rcOriginalMB.right ? rcOriginalMB.right : rcMerger.right;
350  rcMerger.top =
351      rcMerger.top > rcOriginalMB.top ? rcOriginalMB.top : rcMerger.top;
352  rcMerger.bottom = rcMerger.bottom < rcOriginalMB.bottom ? rcOriginalMB.bottom
353                                                          : rcMerger.bottom;
354
355  if (pPageDict->KeyExist("ArtBox"))
356    rcOriginalCB = pPageDict->GetRect("ArtBox");
357  else
358    rcOriginalCB = rcOriginalMB;
359
360  if (!rcOriginalMB.IsEmpty()) {
361    CPDF_Array* pMediaBox = new CPDF_Array();
362    pMediaBox->Add(new CPDF_Number(rcOriginalMB.left));
363    pMediaBox->Add(new CPDF_Number(rcOriginalMB.bottom));
364    pMediaBox->Add(new CPDF_Number(rcOriginalMB.right));
365    pMediaBox->Add(new CPDF_Number(rcOriginalMB.top));
366    pPageDict->SetAt("MediaBox", pMediaBox);
367  }
368
369  if (!rcOriginalCB.IsEmpty()) {
370    CPDF_Array* pCropBox = new CPDF_Array();
371    pCropBox->Add(new CPDF_Number(rcOriginalCB.left));
372    pCropBox->Add(new CPDF_Number(rcOriginalCB.bottom));
373    pCropBox->Add(new CPDF_Number(rcOriginalCB.right));
374    pCropBox->Add(new CPDF_Number(rcOriginalCB.top));
375    pPageDict->SetAt("ArtBox", pCropBox);
376  }
377
378  CPDF_Dictionary* pRes = pPageDict->GetDict("Resources");
379  if (!pRes) {
380    pRes = new CPDF_Dictionary;
381    pPageDict->SetAt("Resources", pRes);
382  }
383
384  CPDF_Stream* pNewXObject = new CPDF_Stream(NULL, 0, new CPDF_Dictionary);
385  FX_DWORD dwObjNum = pDocument->AddIndirectObject(pNewXObject);
386  CPDF_Dictionary* pPageXObject = pRes->GetDict("XObject");
387  if (!pPageXObject) {
388    pPageXObject = new CPDF_Dictionary;
389    pRes->SetAt("XObject", pPageXObject);
390  }
391
392  CFX_ByteString key = "";
393  int nStreams = ObjectArray.GetSize();
394
395  if (nStreams > 0) {
396    for (int iKey = 0; /*iKey < 100*/; iKey++) {
397      char sExtend[5] = {};
398      FXSYS_itoa(iKey, sExtend, 10);
399      key = CFX_ByteString("FFT") + CFX_ByteString(sExtend);
400
401      if (!pPageXObject->KeyExist(key))
402        break;
403    }
404  }
405
406  SetPageContents(key, pPageDict, pDocument);
407
408  CPDF_Dictionary* pNewXORes = NULL;
409
410  if (!key.IsEmpty()) {
411    pPageXObject->SetAtReference(key, pDocument, dwObjNum);
412    CPDF_Dictionary* pNewOXbjectDic = pNewXObject->GetDict();
413    pNewXORes = new CPDF_Dictionary;
414    pNewOXbjectDic->SetAt("Resources", pNewXORes);
415    pNewOXbjectDic->SetAtName("Type", "XObject");
416    pNewOXbjectDic->SetAtName("Subtype", "Form");
417    pNewOXbjectDic->SetAtInteger("FormType", 1);
418    pNewOXbjectDic->SetAtName("Name", "FRM");
419    CPDF_Rect rcBBox = pPageDict->GetRect("ArtBox");
420    pNewOXbjectDic->SetAtRect("BBox", rcBBox);
421  }
422
423  for (int i = 0; i < nStreams; i++) {
424    CPDF_Dictionary* pAnnotDic = ObjectArray.GetAt(i);
425    if (!pAnnotDic)
426      continue;
427
428    CPDF_Rect rcAnnot = pAnnotDic->GetRect("Rect");
429    rcAnnot.Normalize();
430
431    CFX_ByteString sAnnotState = pAnnotDic->GetString("AS");
432    CPDF_Dictionary* pAnnotAP = pAnnotDic->GetDict("AP");
433    if (!pAnnotAP)
434      continue;
435
436    CPDF_Stream* pAPStream = pAnnotAP->GetStream("N");
437    if (!pAPStream) {
438      CPDF_Dictionary* pAPDic = pAnnotAP->GetDict("N");
439      if (!pAPDic)
440        continue;
441
442      if (!sAnnotState.IsEmpty()) {
443        pAPStream = pAPDic->GetStream(sAnnotState);
444      } else {
445        auto it = pAPDic->begin();
446        if (it != pAPDic->end()) {
447          CPDF_Object* pFirstObj = it->second;
448          if (pFirstObj) {
449            if (pFirstObj->IsReference())
450              pFirstObj = pFirstObj->GetDirect();
451            if (!pFirstObj->IsStream())
452              continue;
453            pAPStream = pFirstObj->AsStream();
454          }
455        }
456      }
457    }
458    if (!pAPStream)
459      continue;
460
461    CPDF_Dictionary* pAPDic = pAPStream->GetDict();
462    CFX_Matrix matrix = pAPDic->GetMatrix("Matrix");
463
464    CPDF_Rect rcStream;
465    if (pAPDic->KeyExist("Rect"))
466      rcStream = pAPDic->GetRect("Rect");
467    else if (pAPDic->KeyExist("BBox"))
468      rcStream = pAPDic->GetRect("BBox");
469
470    if (rcStream.IsEmpty())
471      continue;
472
473    CPDF_Object* pObj = pAPStream;
474
475    if (pObj) {
476      CPDF_Dictionary* pObjDic = pObj->GetDict();
477      if (pObjDic) {
478        pObjDic->SetAtName("Type", "XObject");
479        pObjDic->SetAtName("Subtype", "Form");
480      }
481    }
482
483    CPDF_Dictionary* pXObject = pNewXORes->GetDict("XObject");
484    if (!pXObject) {
485      pXObject = new CPDF_Dictionary;
486      pNewXORes->SetAt("XObject", pXObject);
487    }
488
489    CFX_ByteString sFormName;
490    sFormName.Format("F%d", i);
491    FX_DWORD dwObjNum = pDocument->AddIndirectObject(pObj);
492    pXObject->SetAtReference(sFormName, pDocument, dwObjNum);
493
494    CPDF_StreamAcc acc;
495    acc.LoadAllData(pNewXObject);
496
497    const uint8_t* pData = acc.GetData();
498    CFX_ByteString sStream(pData, acc.GetSize());
499    CFX_ByteString sTemp;
500
501    if (matrix.IsIdentity()) {
502      matrix.a = 1.0f;
503      matrix.b = 0.0f;
504      matrix.c = 0.0f;
505      matrix.d = 1.0f;
506      matrix.e = 0.0f;
507      matrix.f = 0.0f;
508    }
509
510    CFX_Matrix m = GetMatrix(rcAnnot, rcStream, matrix);
511    sTemp.Format("q %f 0 0 %f %f %f cm /%s Do Q\n", m.a, m.d, m.e, m.f,
512                 sFormName.c_str());
513    sStream += sTemp;
514
515    pNewXObject->SetData((const uint8_t*)sStream, sStream.GetLength(), FALSE,
516                         FALSE);
517  }
518  pPageDict->RemoveAt("Annots");
519
520  ObjectArray.RemoveAll();
521  RectArray.RemoveAll();
522
523  return FLATTEN_SUCCESS;
524}
525