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 "autoreflow.h" 8 #define approachto(a,b,c) (FXSYS_fabs((float)((a)-(b)))>(c) ? 0 : 1) 9 int FPDF_ProcessInterObj(const CPDF_PageObject* pPrevObj, const CPDF_PageObject* pObj) 10 { 11 CFX_AffineMatrix matrix; 12 FX_RECT PreRect = pPrevObj->GetBBox(&matrix); 13 FX_RECT rect = pObj->GetBBox(&matrix); 14 int flag = 0; 15 if(PreRect.top > rect.bottom) { 16 flag = 0; 17 } else if(rect.top > PreRect.bottom) { 18 flag = 1; 19 } else if(PreRect.right < rect.left) { 20 flag = 0; 21 } else if(PreRect.left > rect.right) { 22 flag = 1; 23 } else if(pObj->m_Type != PDFPAGE_TEXT) { 24 flag = 1; 25 } else if(pPrevObj->m_Type != PDFPAGE_TEXT) { 26 flag = 0; 27 } else { 28 if((PreRect.top < rect.top && PreRect.bottom > rect.bottom) || 29 (PreRect.top > rect.top && PreRect.bottom < rect.bottom)) { 30 if(PreRect.left > rect.left) { 31 flag = 1; 32 } else { 33 flag = 0; 34 } 35 } else { 36 CPDF_TextObject* pPrevTextObj = (CPDF_TextObject* )pPrevObj; 37 CPDF_TextObject* pTextObj = (CPDF_TextObject* )pObj; 38 CPDF_TextObjectItem item, prevItem; 39 pPrevTextObj->GetItemInfo(0, &prevItem); 40 pTextObj->GetItemInfo(0, &item); 41 CFX_AffineMatrix TextMatrix; 42 pTextObj->GetTextMatrix(&TextMatrix); 43 FX_FLOAT originX, originY, prevOriginX, preOriginY; 44 TextMatrix.Transform(item.m_OriginX, item.m_OriginY, originX, originY); 45 pPrevTextObj->GetTextMatrix(&TextMatrix); 46 TextMatrix.Transform(prevItem.m_OriginX, prevItem.m_OriginY, prevOriginX, preOriginY); 47 if(preOriginY > originY) { 48 flag = 0; 49 } else { 50 flag = 1; 51 } 52 } 53 } 54 return flag; 55 } 56 void CPDF_AutoReflowLayoutProvider::Conver2AppreceOrder(const CPDF_PageObjects* pStreamOrderObjs, CPDF_PageObjects* pAppraceOrderObjs) 57 { 58 FX_POSITION pos = pStreamOrderObjs->GetFirstObjectPosition(); 59 CFX_AffineMatrix matrix; 60 while(pos) { 61 CPDF_PageObject* pObj = pStreamOrderObjs->GetNextObject(pos); 62 CFX_AffineMatrix matrix; 63 if(pObj->m_Type != PDFPAGE_TEXT) { 64 continue; 65 } 66 FX_POSITION pos1 = pAppraceOrderObjs->GetLastObjectPosition(); 67 while(pos1) { 68 CPDF_PageObject* pTempObj = pAppraceOrderObjs->GetPrevObject(pos1); 69 if(FPDF_ProcessInterObj(pObj, pTempObj) == 1) { 70 if(!pos1) { 71 pos1 = pAppraceOrderObjs->GetFirstObjectPosition(); 72 } else { 73 pAppraceOrderObjs->GetNextObject(pos1); 74 } 75 break; 76 } 77 } 78 pAppraceOrderObjs->InsertObject(pos1, pObj); 79 } 80 pos = pStreamOrderObjs->GetFirstObjectPosition(); 81 while(pos) { 82 CPDF_PageObject* pObj = pStreamOrderObjs->GetNextObject(pos); 83 if(pObj->m_Type != PDFPAGE_IMAGE) { 84 continue; 85 } 86 FX_POSITION pos1 = pAppraceOrderObjs->GetLastObjectPosition(); 87 while(pos1) { 88 CPDF_PageObject* pTempObj = pAppraceOrderObjs->GetPrevObject(pos1); 89 if(FPDF_ProcessInterObj(pObj, pTempObj) == 1) { 90 if(!pos1) { 91 pos1 = pAppraceOrderObjs->GetFirstObjectPosition(); 92 } else { 93 pAppraceOrderObjs->GetNextObject(pos1); 94 } 95 break; 96 } 97 } 98 pAppraceOrderObjs->InsertObject(pos1, pObj); 99 } 100 } 101 IPDF_LayoutProvider* IPDF_LayoutProvider::Create_LayoutProvider_AutoReflow(CPDF_PageObjects* pPage, FX_BOOL bReadOrder) 102 { 103 return new CPDF_AutoReflowLayoutProvider(pPage, bReadOrder); 104 } 105 CPDF_AutoReflowElement::CPDF_AutoReflowElement(LayoutType layoutType , CPDF_AutoReflowElement* pParent) 106 { 107 m_ElmType = layoutType; 108 m_pParentElm = pParent; 109 if(pParent) { 110 pParent->m_ChildArray.Add(this); 111 } 112 m_SpaceBefore = 0; 113 } 114 CPDF_AutoReflowElement::~CPDF_AutoReflowElement() 115 { 116 m_ChildArray.RemoveAll(); 117 m_ObjArray.RemoveAll(); 118 } 119 int CPDF_AutoReflowElement::CountAttrValues(LayoutAttr attr_type) 120 { 121 return 1; 122 } 123 LayoutEnum CPDF_AutoReflowElement::GetEnumAttr(LayoutAttr attr_type, int index ) 124 { 125 return LayoutInvalid; 126 } 127 FX_FLOAT CPDF_AutoReflowElement::GetNumberAttr(LayoutAttr attr_type, int index ) 128 { 129 switch (attr_type) { 130 case LayoutSpaceBefore: 131 return m_SpaceBefore; 132 default: 133 return 0; 134 } 135 } 136 FX_COLORREF CPDF_AutoReflowElement::GetColorAttr(LayoutAttr attr_type, int index ) 137 { 138 return 0; 139 } 140 #define WritingMode_UNKNOW 0 141 #define WritingMode_LRTB 1 142 #define WritingMode_RLTB 2 143 #define WritingMode_TBRL 3 144 CPDF_AutoReflowLayoutProvider::CPDF_AutoReflowLayoutProvider(CPDF_PageObjects* pPage, FX_BOOL bReadOrder) 145 { 146 m_pPDFPage = (CPDF_Page*)pPage; 147 FX_FLOAT width = m_pPDFPage->GetPageWidth(); 148 FX_FLOAT height = m_pPDFPage->GetPageHeight(); 149 m_pPDFPage->GetDisplayMatrix(m_PDFDisplayMatrix, 0, 0, (int)(m_pPDFPage->GetPageWidth()), (int)(m_pPDFPage->GetPageHeight()), 0); 150 m_bReadOrder = bReadOrder; 151 m_Status = LayoutReady; 152 m_pRoot = NULL; 153 m_pCurrElm = NULL; 154 m_pPreObj = NULL; 155 m_Step = 0; 156 m_WritingMode = WritingMode_UNKNOW; 157 } 158 CPDF_AutoReflowLayoutProvider::~CPDF_AutoReflowLayoutProvider() 159 { 160 m_pPDFPage = NULL; 161 ReleaseElm(m_pRoot); 162 } 163 void CPDF_AutoReflowLayoutProvider::ReleaseElm(CPDF_AutoReflowElement*& pElm, FX_BOOL bReleaseChildren) 164 { 165 if(bReleaseChildren) { 166 int count = pElm->CountChildren(); 167 for(int i = 0; i < count; i++) { 168 CPDF_AutoReflowElement* pChild = (CPDF_AutoReflowElement*)pElm->GetChild(i); 169 ReleaseElm(pChild); 170 } 171 } 172 delete pElm; 173 pElm = NULL; 174 } 175 void CPDF_AutoReflowLayoutProvider::AddObjectArray(CPDF_AutoReflowElement* pElm, CFX_PtrList& ObjList) 176 { 177 if(!pElm) { 178 return; 179 } 180 FX_POSITION pos = ObjList.GetHeadPosition(); 181 while (pos) { 182 pElm->m_ObjArray.Add((CPDF_PageObject*)ObjList.GetNext(pos)); 183 } 184 } 185 void CPDF_AutoReflowLayoutProvider::GenerateStructTree() 186 { 187 if (m_Step < AUTOREFLOW_STEP_GENERATELINE) { 188 GenerateLine(m_cellArray); 189 if(m_cellArray.GetSize() == 0) { 190 m_Status = LayoutError; 191 return; 192 } 193 if(m_pPause && m_pPause->NeedToPauseNow()) { 194 m_Step = AUTOREFLOW_STEP_GENERATELINE; 195 m_Status = LayoutToBeContinued; 196 return; 197 } 198 } 199 if (m_Step < AUTOREFLOW_STEP_GENERATEParagraph) { 200 GenerateParagraph(m_cellArray); 201 if(m_pPause && m_pPause->NeedToPauseNow()) { 202 m_Step = AUTOREFLOW_STEP_GENERATEParagraph; 203 m_Status = LayoutToBeContinued; 204 return; 205 } 206 } 207 if (m_Step < AUTOREFLOW_STEP_CREATEELEMENT) { 208 CreateElement(); 209 if(m_pPause && m_pPause->NeedToPauseNow()) { 210 m_Step = AUTOREFLOW_STEP_CREATEELEMENT; 211 m_Status = LayoutToBeContinued; 212 return; 213 } 214 } 215 if (m_Step < AUTOREFLOW_STEP_REMOVEDATA) { 216 int count = m_cellArray.GetSize(); 217 for(int i = 0; i < count; i++) { 218 CRF_CELL* pCell = (CRF_CELL*)m_cellArray.GetAt(i); 219 if(pCell) { 220 pCell->m_ObjList.RemoveAll(); 221 delete pCell; 222 } 223 } 224 m_cellArray.RemoveAll(); 225 if(m_pPause && m_pPause->NeedToPauseNow()) { 226 m_Step = AUTOREFLOW_STEP_REMOVEDATA; 227 m_Status = LayoutToBeContinued; 228 return; 229 } 230 } 231 m_Step = AUTOREFLOW_STEP_REMOVEDATA; 232 m_Status = LayoutFinished; 233 return; 234 } 235 void CPDF_AutoReflowLayoutProvider::CreateElement() 236 { 237 int count = m_cellArray.GetSize(); 238 CRF_CELL* plastCell = NULL; 239 CRF_CELL* pCell = NULL; 240 CRF_CELL* pNextCell = NULL; 241 CPDF_AutoReflowElement* pParent = m_pRoot; 242 CPDF_AutoReflowElement* pCurrElm = NULL; 243 int i; 244 for(i = 0; i < count; i++) { 245 pCell = (CRF_CELL*)m_cellArray.GetAt(i); 246 if(!pCell) { 247 continue; 248 } 249 if(i < count - 1) { 250 pNextCell = (CRF_CELL*)m_cellArray.GetAt(i + 1); 251 } else { 252 pNextCell = NULL; 253 } 254 pCurrElm = new CPDF_AutoReflowElement(LayoutParagraph, pParent); 255 if(pCurrElm->GetType() == LayoutParagraph && plastCell) { 256 int SpaceBefore = 0; 257 if(pCell->m_CellWritingMode != plastCell->m_CellWritingMode ) { 258 SpaceBefore = 20; 259 } else if(pCell->m_CellWritingMode == WritingMode_LRTB) { 260 SpaceBefore = plastCell->m_BBox.bottom - pCell->m_BBox.top; 261 } else if(pCell->m_CellWritingMode == WritingMode_TBRL) { 262 SpaceBefore = plastCell->m_BBox.left - pCell->m_BBox.right; 263 } 264 if(SpaceBefore > 0) { 265 pCurrElm->m_SpaceBefore = SpaceBefore > 50 ? 50.0f : SpaceBefore; 266 } 267 } 268 AddObjectArray(pCurrElm, pCell->m_ObjList); 269 plastCell = pCell; 270 } 271 } 272 void CPDF_AutoReflowLayoutProvider::GenerateParagraph(CFX_PtrArray& cellArray) 273 { 274 int count = cellArray.GetSize(); 275 if(count <= 1) { 276 return; 277 } 278 CRF_CELL* plastCell = (CRF_CELL*)cellArray.GetAt(0); 279 if(plastCell->m_BBox.Height() > plastCell->m_BBox.Width()) { 280 m_WritingMode = WritingMode_TBRL; 281 } else { 282 m_WritingMode = WritingMode_LRTB; 283 } 284 FX_BOOL bEnforce = FALSE; 285 int i = 0; 286 for(i = 1; i < count; i++) { 287 CRF_CELL* pCell = (CRF_CELL*)cellArray.GetAt(i); 288 if(!pCell) { 289 continue; 290 } 291 int c = pCell->m_ObjList.GetCount(); 292 FX_BOOL bMerge = FALSE; 293 FX_POSITION pos1 = plastCell->m_ObjList.GetTailPosition(); 294 CPDF_PageObject* pLastObj = (CPDF_PageObject*)plastCell->m_ObjList.GetPrev(pos1); 295 pos1 = pCell->m_ObjList.GetHeadPosition(); 296 CPDF_PageObject* pCurObj = (CPDF_PageObject*)pCell->m_ObjList.GetNext(pos1); 297 int WritingMode = GetRectEnd(pCell->m_BBox); 298 if(pCell->m_CellWritingMode == WritingMode_UNKNOW) { 299 if(pCell->m_BBox.Height() > pCell->m_BBox.Width()) { 300 pCell->m_CellWritingMode = WritingMode_TBRL; 301 } else { 302 pCell->m_CellWritingMode = WritingMode_LRTB; 303 } 304 } 305 WritingMode = pCell->m_CellWritingMode; 306 if(WritingMode == WritingMode_LRTB && (m_Style.m_Language & LP_Lang_ChinesePRC || m_Style.m_Language & LP_Lang_ChineseTaiwan 307 || m_Style.m_Language & LP_Lang_Japanese || m_Style.m_Language & LP_Lang_Korean)) { 308 if(pCurObj->m_Type == PDFPAGE_TEXT) { 309 CPDF_TextObject* pText; 310 pText = (CPDF_TextObject*)pCurObj; 311 if(pText->CountItems()) { 312 CPDF_TextObjectItem item; 313 pText->GetItemInfo(0, &item); 314 CFX_WideString str = pText->GetFont()->UnicodeFromCharCode(item.m_CharCode); 315 FX_WCHAR unicode = str.GetAt(0); 316 if(unicode == 32) { 317 plastCell = pCell; 318 bMerge = FALSE; 319 bEnforce = FALSE; 320 continue; 321 } 322 } 323 } 324 } 325 if(m_WritingMode == WritingMode) { 326 if(bEnforce) { 327 bMerge = FALSE; 328 bEnforce = FALSE; 329 if(pCurObj->m_Type == PDFPAGE_TEXT) { 330 CPDF_TextObject* pText; 331 pText = (CPDF_TextObject*)pCurObj; 332 if(pText->CountItems()) { 333 CPDF_TextObjectItem item; 334 pText->GetItemInfo(0, &item); 335 CFX_WideString str = pText->GetFont()->UnicodeFromCharCode(item.m_CharCode); 336 FX_WCHAR unicode = str.GetAt(0); 337 if(unicode > 96 && unicode < 123) { 338 bMerge = TRUE; 339 } 340 } 341 } else { 342 CPDF_ImageObject* pImage = (CPDF_ImageObject*)pCurObj; 343 FX_RECT imageBBox = pImage->GetBBox(&m_PDFDisplayMatrix); 344 if(GetRectEnd(plastCell->m_BBox) - GetRectEnd(pCell->m_BBox) < GetRectWidth(imageBBox)) { 345 bMerge = TRUE; 346 } 347 } 348 } else { 349 if(!approachto(GetRectStart(pCell->m_BBox), GetRectStart(plastCell->m_BBox), GetRectHeight(pCell->m_BBox) / 4)) { 350 if(approachto(GetRectStart(plastCell->m_BBox), GetRectStart(pCell->m_BBox), GetRectHeight(pCell->m_BBox) * 2.3) && 351 GetRectStart(plastCell->m_BBox) - GetRectStart(pCell->m_BBox) > 0) { 352 if(pCurObj->m_Type == PDFPAGE_TEXT || pLastObj->m_Type == PDFPAGE_TEXT) { 353 CPDF_TextObject* pText; 354 if(pCurObj->m_Type == PDFPAGE_TEXT) { 355 pText = (CPDF_TextObject*)pCurObj; 356 } else { 357 pText = (CPDF_TextObject*)pLastObj; 358 } 359 CPDF_TextObjectItem item; 360 pText->GetItemInfo(0, &item); 361 CFX_WideString str = pText->GetFont()->UnicodeFromCharCode(item.m_CharCode); 362 FX_WCHAR unicode = str.GetAt(0); 363 if(unicode > 255) { 364 bMerge = TRUE; 365 } 366 } 367 } 368 } else if(!approachto(GetRectEnd(pCell->m_BBox), GetRectEnd(plastCell->m_BBox), GetRectHeight(pCell->m_BBox) * 3)) { 369 FX_RECT rect = pLastObj->GetBBox(&m_PDFDisplayMatrix); 370 if(approachto(GetRectStart(pCell->m_BBox), GetRectStart(plastCell->m_BBox), GetRectHeight(pCell->m_BBox) / 4)) { 371 if(GetRectEnd(rect) - GetRectEnd(pCell->m_BBox) > 0) { 372 bMerge = TRUE; 373 bEnforce = TRUE; 374 } else if(GetRectEnd(rect) - GetRectEnd(pCell->m_BBox) <= 0 && 375 GetRectEnd(rect) - GetRectEnd(pCell->m_BBox) > GetRectHeight(pCell->m_BBox) * -3) { 376 if(pCurObj->m_Type == PDFPAGE_TEXT) { 377 CPDF_TextObject* pText = (CPDF_TextObject*)pCurObj; 378 CPDF_TextObjectItem item; 379 pText->GetItemInfo(0, &item); 380 CFX_WideString str = pText->GetFont()->UnicodeFromCharCode(item.m_CharCode); 381 FX_WCHAR unicode = str.GetAt(0); 382 if(unicode > 96 && unicode < 123) { 383 bMerge = TRUE; 384 } 385 } 386 } 387 } 388 } else { 389 bMerge = TRUE; 390 } 391 } 392 } else { 393 m_WritingMode = WritingMode; 394 bEnforce = FALSE; 395 } 396 if(bMerge) { 397 if(GetRectEnd(plastCell->m_BBox) - GetRectEnd(pCell->m_BBox) > 30) { 398 bEnforce = TRUE; 399 } 400 FX_POSITION pos = pCell->m_ObjList.GetHeadPosition(); 401 while(pos) { 402 plastCell->m_ObjList.AddTail(pCell->m_ObjList.GetNext(pos)); 403 } 404 plastCell->m_BBox.Union(pCell->m_BBox); 405 pCell->m_ObjList.RemoveAll(); 406 delete pCell; 407 cellArray.RemoveAt(i); 408 i--; 409 count--; 410 } else { 411 plastCell = pCell; 412 } 413 } 414 } 415 void CPDF_AutoReflowLayoutProvider::ProcessObj(CFX_PtrArray& cellArray, CPDF_PageObject* pObj, CFX_AffineMatrix matrix) 416 { 417 } 418 FX_INT32 CPDF_AutoReflowLayoutProvider::LogicPreObj(CPDF_PageObject* pObj) 419 { 420 CPDF_PageObject* pPreObj = m_pPreObj; 421 m_pPreObj = pObj; 422 if(!pPreObj) { 423 return 0; 424 } 425 if(pPreObj->m_Type != pObj->m_Type) { 426 return 0; 427 } 428 CFX_FloatRect rcCurObj(pObj->m_Left, pObj->m_Bottom, pObj->m_Right, pObj->m_Top); 429 CFX_FloatRect rcPreObj(pPreObj->m_Left, pPreObj->m_Bottom, pPreObj->m_Right, pPreObj->m_Top); 430 if(pObj->m_Type == PDFPAGE_IMAGE) { 431 if(rcPreObj.Contains(rcCurObj)) { 432 return 2; 433 } 434 if(rcCurObj.Contains(rcPreObj)) { 435 return 2; 436 } 437 return 0; 438 } 439 if(pObj->m_Type == PDFPAGE_TEXT) { 440 if(!((rcPreObj.bottom > rcCurObj.top) || (rcPreObj.top < rcCurObj.bottom))) { 441 FX_FLOAT height = FX_MIN(rcPreObj.Height(), rcCurObj.Height()); 442 if((rcCurObj.left - rcPreObj.right) > height / 3) { 443 return 3; 444 } 445 } 446 if(FXSYS_fabs(rcPreObj.Width() - rcCurObj.Width()) >= 2 || FXSYS_fabs(rcPreObj.Height() - rcCurObj.Height()) >= 2 ) { 447 return 0; 448 } 449 CPDF_TextObject* pPreTextObj = (CPDF_TextObject*)pPreObj; 450 CPDF_TextObject* pCurTextObj = (CPDF_TextObject*)pObj; 451 int nPreCount = pPreTextObj->CountItems(); 452 int nCurCount = pCurTextObj->CountItems(); 453 if (nPreCount != nCurCount) { 454 return 0; 455 } 456 FX_BOOL bSame = TRUE; 457 for (int i = 0; i < nPreCount; i++) { 458 CPDF_TextObjectItem itemPer, itemCur; 459 pPreTextObj->GetItemInfo(i, &itemPer); 460 pCurTextObj->GetItemInfo(i, &itemCur); 461 if (itemCur.m_CharCode != itemPer.m_CharCode) { 462 return 0; 463 } 464 if (itemCur.m_OriginX != itemPer.m_OriginX) { 465 bSame = FALSE; 466 } 467 if (itemCur.m_OriginY != itemPer.m_OriginY) { 468 bSame = FALSE; 469 } 470 } 471 if(rcPreObj.left == rcCurObj.left && rcPreObj.top == rcCurObj.top) { 472 return 1; 473 } 474 if(FXSYS_fabs(rcPreObj.left - rcCurObj.left) < rcPreObj.Width() / 3 475 && FXSYS_fabs(rcPreObj.top - rcCurObj.top) < rcPreObj.Height() / 3) { 476 return 2; 477 } 478 } 479 return 0; 480 } 481 void CPDF_AutoReflowLayoutProvider::GenerateLine(CFX_PtrArray& cellArray) 482 { 483 CRF_CELL* pCell = NULL; 484 CFX_AffineMatrix matrix; 485 FX_POSITION pos = m_pPDFPage->GetFirstObjectPosition(); 486 if(!pos) { 487 return; 488 } 489 FX_FLOAT PDFWidth = m_pPDFPage->GetPageWidth(); 490 FX_FLOAT PDFHeight = m_pPDFPage->GetPageHeight(); 491 m_pPDFPage->GetDisplayMatrix(m_PDFDisplayMatrix, 0, 0, (int)PDFWidth, (int)PDFHeight, 0); 492 CPDF_PageObject* pPerObj = NULL; 493 int a = 0; 494 CFX_FloatRect pageBBox = m_pPDFPage->m_BBox; 495 FX_FLOAT PrevX = 0 , PrevY = 0, PosX, PosY; 496 while(pos) { 497 CPDF_PageObject* pObj = m_pPDFPage->GetNextObject(pos); 498 if(!pObj || pObj->m_Type == PDFPAGE_PATH) { 499 continue; 500 } 501 int logic = LogicPreObj(pObj); 502 if(logic == 2) { 503 if(pCell) { 504 pCell->m_ObjList.SetAt(pCell->m_ObjList.GetTailPosition(), pObj); 505 } 506 continue; 507 } 508 if (pObj->m_Type == PDFPAGE_TEXT) { 509 CPDF_TextObject* pTextObj = (CPDF_TextObject*)pObj; 510 int textmode = pTextObj->m_TextState.GetObject()->m_TextMode; 511 if(m_Style.m_bIgnoreInvisibleText && pTextObj->m_TextState.GetObject()->m_TextMode == 3) { 512 continue; 513 } 514 PosX = pTextObj->GetPosX(); 515 PosY = pTextObj->GetPosY(); 516 m_PDFDisplayMatrix.Transform(PosX, PosY); 517 } else { 518 PosX = 0; 519 PosY = 0; 520 } 521 FX_BOOL bNewLine = TRUE; 522 FX_RECT ObjBBox = pObj->GetBBox(&m_PDFDisplayMatrix); 523 if(ObjBBox.left > PDFWidth || ObjBBox.right < 0 || 524 ObjBBox.bottom < 0 || ObjBBox.top > PDFHeight) { 525 continue; 526 } 527 if(ObjBBox.IsEmpty()) { 528 continue; 529 } 530 a++; 531 if(!pCell) { 532 bNewLine = TRUE; 533 m_WritingMode = GetWritingMode(NULL, pObj); 534 } else { 535 int WritingMode = GetWritingMode(pPerObj, pObj); 536 if(m_WritingMode == WritingMode || m_WritingMode == WritingMode_UNKNOW || WritingMode == WritingMode_UNKNOW) { 537 if(WritingMode != WritingMode_UNKNOW) { 538 m_WritingMode = WritingMode; 539 } 540 if(m_WritingMode == WritingMode_TBRL) { 541 if(!(GetRectBottom(ObjBBox) > GetRectTop(pCell->m_BBox) || 542 GetRectTop(ObjBBox) < GetRectBottom(pCell->m_BBox))) { 543 bNewLine = FALSE; 544 } 545 } else { 546 if(!(GetRectBottom(ObjBBox) < GetRectTop(pCell->m_BBox) || 547 GetRectTop(ObjBBox) > GetRectBottom(pCell->m_BBox))) { 548 bNewLine = FALSE; 549 } 550 if (pObj->m_Type == PDFPAGE_TEXT) { 551 if(FXSYS_fabs(PrevY - PosY) < 1 ) { 552 bNewLine = FALSE; 553 } 554 } 555 } 556 } else { 557 m_WritingMode = WritingMode; 558 } 559 } 560 pPerObj = pObj; 561 if(bNewLine) { 562 int c = pCell ? pCell->m_ObjList.GetCount() : 0; 563 pCell = new CRF_CELL; 564 pCell->m_CellWritingMode = m_WritingMode; 565 pCell->m_BBox = ObjBBox; 566 if(pObj->m_Type == PDFPAGE_TEXT) { 567 FX_FLOAT x = ((CPDF_TextObject*)pObj)->GetPosX(), y = ((CPDF_TextObject*)pObj)->GetPosY(); 568 m_PDFDisplayMatrix.Transform(x, y); 569 if(x < ObjBBox.left) { 570 pCell->m_BBox.left = (int)x; 571 } 572 } 573 pCell->m_ObjList.AddTail(pObj); 574 cellArray.Add(pCell); 575 } else { 576 pCell->m_ObjList.AddTail(pObj); 577 pCell->m_BBox.Union(ObjBBox); 578 } 579 PrevX = PosX; 580 PrevY = PosY; 581 } 582 } 583 FX_FLOAT CPDF_AutoReflowLayoutProvider::GetLayoutOrderHeight(CPDF_PageObject* pCurObj) 584 { 585 CFX_FloatRect rcCurObj(pCurObj->m_Left, pCurObj->m_Bottom, pCurObj->m_Right, pCurObj->m_Top); 586 if (m_WritingMode == WritingMode_TBRL) { 587 return rcCurObj.Width(); 588 } 589 return rcCurObj.Height(); 590 } 591 FX_FLOAT CPDF_AutoReflowLayoutProvider::GetLayoutOrderWidth(CPDF_PageObject* pCurObj) 592 { 593 CFX_FloatRect rcCurObj(pCurObj->m_Left, pCurObj->m_Bottom, pCurObj->m_Right, pCurObj->m_Top); 594 if (m_WritingMode == WritingMode_TBRL) { 595 return rcCurObj.Height(); 596 } 597 return rcCurObj.Width(); 598 } 599 int CPDF_AutoReflowLayoutProvider:: GetRectWidth(FX_RECT rect) 600 { 601 if(m_WritingMode == WritingMode_TBRL) { 602 return rect.Height(); 603 } 604 return rect.Width(); 605 } 606 int CPDF_AutoReflowLayoutProvider:: GetRectHeight(FX_RECT rect) 607 { 608 if(m_WritingMode == WritingMode_TBRL) { 609 return rect.Width(); 610 } 611 return rect.Height(); 612 } 613 int CPDF_AutoReflowLayoutProvider:: GetRectStart(FX_RECT rect) 614 { 615 if(m_WritingMode == WritingMode_TBRL) { 616 return rect.top; 617 } 618 return rect.left; 619 } 620 int CPDF_AutoReflowLayoutProvider:: GetRectEnd(FX_RECT rect) 621 { 622 if(m_WritingMode == WritingMode_TBRL) { 623 return rect.bottom; 624 } 625 return rect.right; 626 } 627 int CPDF_AutoReflowLayoutProvider:: GetRectTop(FX_RECT rect) 628 { 629 if(m_WritingMode == WritingMode_TBRL) { 630 return rect.right; 631 } 632 return rect.top; 633 } 634 int CPDF_AutoReflowLayoutProvider:: GetRectBottom(FX_RECT rect) 635 { 636 if(m_WritingMode == WritingMode_TBRL) { 637 return rect.left; 638 } 639 return rect.bottom; 640 } 641 int CPDF_AutoReflowLayoutProvider::GetWritingMode(CPDF_PageObject* pPreObj, CPDF_PageObject* pCurObj) 642 { 643 CFX_FloatRect rcCurObj(pCurObj->m_Left, pCurObj->m_Bottom, pCurObj->m_Right, pCurObj->m_Top); 644 if(pCurObj->m_Type == PDFPAGE_TEXT) { 645 CPDF_TextObject* ptextObj = (CPDF_TextObject* )pCurObj; 646 int count = ptextObj->CountItems(); 647 if(count > 1) { 648 CPDF_TextObjectItem Item1, Item2; 649 ptextObj->GetItemInfo(0, &Item1); 650 ptextObj->GetItemInfo(count - 1, &Item2); 651 if(Item2.m_CharCode == -1 && count > 2) { 652 ptextObj->GetItemInfo(2, &Item2); 653 } 654 CFX_AffineMatrix textMatrix; 655 ptextObj->GetTextMatrix(&textMatrix); 656 textMatrix.Transform(Item1.m_OriginX, Item1.m_OriginY); 657 textMatrix.Transform(Item2.m_OriginX, Item2.m_OriginY); 658 FX_FLOAT dx = FXSYS_fabs(Item1.m_OriginX - Item2.m_OriginX); 659 FX_FLOAT dy = FXSYS_fabs(Item1.m_OriginY - Item2.m_OriginY); 660 return dx >= dy ? WritingMode_LRTB : WritingMode_TBRL; 661 } else { 662 if(m_WritingMode != WritingMode_UNKNOW) { 663 return m_WritingMode; 664 } 665 } 666 } 667 if(pPreObj) { 668 FX_FLOAT threshold = rcCurObj.Width() / 4; 669 if(m_WritingMode == WritingMode_LRTB) { 670 if(FXSYS_fabs(pPreObj->m_Bottom - pCurObj->m_Bottom) < threshold * 2 671 && FXSYS_fabs(pPreObj->m_Top - pCurObj->m_Top) < threshold * 2) { 672 return m_WritingMode; 673 } 674 FX_FLOAT mid = (pCurObj->m_Bottom + pCurObj->m_Top) / 2; 675 if(mid > pPreObj->m_Bottom && mid < pPreObj->m_Top && pCurObj->m_Right > pPreObj->m_Right) { 676 return m_WritingMode; 677 } 678 } else if(m_WritingMode == WritingMode_TBRL) { 679 if(FXSYS_fabs(pPreObj->m_Left - pCurObj->m_Left) < threshold * 2 680 && FXSYS_fabs(pPreObj->m_Right - pCurObj->m_Right) < threshold * 2) { 681 return m_WritingMode; 682 } 683 FX_FLOAT mid = (pCurObj->m_Right + pCurObj->m_Left) / 2; 684 if(mid > pPreObj->m_Left && mid < pPreObj->m_Right && pCurObj->m_Bottom < pPreObj->m_Bottom) { 685 return m_WritingMode; 686 } 687 } 688 if(FXSYS_fabs(pPreObj->m_Left - pCurObj->m_Left) < threshold && 689 FXSYS_fabs(pPreObj->m_Bottom - pCurObj->m_Bottom) > threshold * 2) { 690 return WritingMode_TBRL; 691 } 692 if(FXSYS_fabs(pPreObj->m_Left - pCurObj->m_Left) > threshold && 693 FXSYS_fabs(pPreObj->m_Bottom - pCurObj->m_Bottom) < threshold * 2) { 694 return WritingMode_LRTB; 695 } 696 int count = 0; 697 if(pPreObj->m_Type == PDFPAGE_TEXT) { 698 CPDF_TextObject* ptextObj = (CPDF_TextObject* )pCurObj; 699 count = ptextObj->CountItems(); 700 } 701 if(pPreObj->m_Type != PDFPAGE_TEXT || count == 1) { 702 if(pCurObj->m_Left > pPreObj->m_Right) { 703 FX_FLOAT mid = (pCurObj->m_Top + pCurObj->m_Bottom) / 2; 704 if(mid < pPreObj->m_Top && mid > pPreObj->m_Bottom) { 705 return WritingMode_LRTB; 706 } 707 } 708 if(pCurObj->m_Top < pPreObj->m_Bottom) { 709 FX_FLOAT mid = (pCurObj->m_Left + pCurObj->m_Right) / 2; 710 if(mid < pPreObj->m_Right && mid > pPreObj->m_Left) { 711 return WritingMode_TBRL; 712 } 713 } 714 } 715 } 716 return WritingMode_UNKNOW; 717 } 718 LayoutStatus CPDF_AutoReflowLayoutProvider::StartLoad(IFX_Pause* pPause) 719 { 720 m_pPause = pPause; 721 m_pRoot = new CPDF_AutoReflowElement(LayoutDocument); 722 m_Step = 0; 723 return Continue(); 724 } 725 LayoutStatus CPDF_AutoReflowLayoutProvider::Continue() 726 { 727 GenerateStructTree(); 728 return m_Status; 729 } 730 int CPDF_AutoReflowLayoutProvider::GetPosition() 731 { 732 if(m_Step == 0) { 733 return 0; 734 } else { 735 return m_Step * 100 / AUTOREFLOW_STEP_REMOVEDATA; 736 } 737 } 738 FX_FLOAT CPDF_AutoReflowLayoutProvider::GetObjMinCell(CPDF_PageObject* pObj) 739 { 740 if(!pObj) { 741 return 0; 742 } 743 if(pObj->m_Type != PDFPAGE_TEXT) { 744 CFX_AffineMatrix matrix; 745 FX_RECT rect = pObj->GetBBox(&matrix); 746 return (FX_FLOAT)(rect.Width()); 747 } 748 CPDF_TextObject* pTextObj = (CPDF_TextObject* )pObj; 749 int count = pTextObj->CountItems(); 750 for(int i = 0; i < count; i++) { 751 CPDF_TextObjectItem Item; 752 pTextObj->GetItemInfo(i, &Item); 753 if(Item.m_CharCode == -1) { 754 continue; 755 } 756 if((Item.m_CharCode > 47 && Item.m_CharCode < 58) || (Item.m_CharCode > 64 && Item.m_CharCode < 91) 757 || (Item.m_CharCode > 96 && Item.m_CharCode < 123)) { 758 continue; 759 } 760 if(Item.m_CharCode > 127 || (Item.m_CharCode > 32 && Item.m_CharCode < 35) || Item.m_CharCode == 37 || 761 (Item.m_CharCode > 38 && Item.m_CharCode < 42) || Item.m_CharCode == 44 || Item.m_CharCode == 46 || 762 Item.m_CharCode == 58 || Item.m_CharCode == 59 || Item.m_CharCode == 63 || Item.m_CharCode == 93) { 763 if(i == count - 1) { 764 CFX_AffineMatrix matrix; 765 FX_RECT rect = pObj->GetBBox(&matrix); 766 return (FX_FLOAT)(rect.Width()); 767 } else { 768 pTextObj->GetItemInfo(i + 1, &Item); 769 return Item.m_OriginX; 770 } 771 } 772 return Item.m_OriginX; 773 } 774 CFX_AffineMatrix matrix; 775 FX_RECT rect = pObj->GetBBox(&matrix); 776 return (FX_FLOAT)(rect.Width()); 777 } 778
