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 "../../include/fpdfapi/fpdf_resource.h"
8 #include "../../include/fpdfapi/fpdf_pageobj.h"
9 #include "../../include/fpdftext/fpdf_text.h"
10 #include "../../include/fpdfapi/fpdf_page.h"
11 #include "../../include/fpdfapi/fpdf_module.h"
12 #include <ctype.h>
13 #include <algorithm>
14 #include "text_int.h"
15
16 namespace {
17
_IsIgnoreSpaceCharacter(FX_WCHAR curChar)18 FX_BOOL _IsIgnoreSpaceCharacter(FX_WCHAR curChar)
19 {
20 if(curChar < 255 ) {
21 return FALSE;
22 }
23 if ( (curChar >= 0x0600 && curChar <= 0x06FF)
24 || (curChar >= 0xFE70 && curChar <= 0xFEFF)
25 || (curChar >= 0xFB50 && curChar <= 0xFDFF)
26 || (curChar >= 0x0400 && curChar <= 0x04FF)
27 || (curChar >= 0x0500 && curChar <= 0x052F)
28 || (curChar >= 0xA640 && curChar <= 0xA69F)
29 || (curChar >= 0x2DE0 && curChar <= 0x2DFF)
30 || curChar == 8467
31 || (curChar >= 0x2000 && curChar <= 0x206F)) {
32 return FALSE;
33 }
34 return TRUE;
35 }
36
_NormalizeThreshold(FX_FLOAT threshold)37 FX_FLOAT _NormalizeThreshold(FX_FLOAT threshold)
38 {
39 if (threshold < 300) {
40 return threshold / 2.0f;
41 } else if (threshold < 500) {
42 return threshold / 4.0f;
43 } else if (threshold < 700) {
44 return threshold / 5.0f;
45 }
46 return threshold / 6.0f;
47 }
48
_CalculateBaseSpace(const CPDF_TextObject * pTextObj,const CFX_AffineMatrix & matrix)49 FX_FLOAT _CalculateBaseSpace(const CPDF_TextObject* pTextObj,
50 const CFX_AffineMatrix& matrix)
51 {
52 FX_FLOAT baseSpace = 0.0;
53 const int nItems = pTextObj->CountItems();
54 if (pTextObj->m_TextState.GetObject()->m_CharSpace && nItems >= 3) {
55 FX_BOOL bAllChar = TRUE;
56 FX_FLOAT spacing = matrix.TransformDistance(
57 pTextObj->m_TextState.GetObject()->m_CharSpace);
58 baseSpace = spacing;
59 for (int i = 0; i < nItems; i++) {
60 CPDF_TextObjectItem item;
61 pTextObj->GetItemInfo(i, &item);
62 if (item.m_CharCode == (FX_DWORD) - 1) {
63 FX_FLOAT fontsize_h = pTextObj->m_TextState.GetFontSizeH();
64 FX_FLOAT kerning = -fontsize_h * item.m_OriginX / 1000;
65 baseSpace = std::min(baseSpace, kerning + spacing);
66 bAllChar = FALSE;
67 }
68 }
69 if (baseSpace < 0.0 || (nItems == 3 && !bAllChar)) {
70 baseSpace = 0.0;
71 }
72 }
73 return baseSpace;
74 }
75
76 } // namespace
77
CPDFText_ParseOptions()78 CPDFText_ParseOptions::CPDFText_ParseOptions()
79 : m_bGetCharCodeOnly(FALSE), m_bNormalizeObjs(TRUE), m_bOutputHyphen(FALSE)
80 {
81 }
CreateTextPage(const CPDF_Page * pPage,CPDFText_ParseOptions ParserOptions)82 IPDF_TextPage* IPDF_TextPage::CreateTextPage(const CPDF_Page* pPage, CPDFText_ParseOptions ParserOptions)
83 {
84 return new CPDF_TextPage(pPage, ParserOptions);
85 }
CreateTextPage(const CPDF_Page * pPage,int flags)86 IPDF_TextPage* IPDF_TextPage::CreateTextPage(const CPDF_Page* pPage, int flags)
87 {
88 return new CPDF_TextPage(pPage, flags);
89 }
CreateTextPage(const CPDF_PageObjects * pObjs,int flags)90 IPDF_TextPage* IPDF_TextPage::CreateTextPage(const CPDF_PageObjects* pObjs, int flags)
91 {
92 return new CPDF_TextPage(pObjs, flags);
93 }
CreatePageFind(const IPDF_TextPage * pTextPage)94 IPDF_TextPageFind* IPDF_TextPageFind::CreatePageFind(const IPDF_TextPage* pTextPage)
95 {
96 if (!pTextPage) {
97 return NULL;
98 }
99 return new CPDF_TextPageFind(pTextPage);
100 }
CreateLinkExtract()101 IPDF_LinkExtract* IPDF_LinkExtract::CreateLinkExtract()
102 {
103 return new CPDF_LinkExtract();
104 }
105 #define TEXT_BLANK_CHAR L' '
106 #define TEXT_LINEFEED_CHAR L'\n'
107 #define TEXT_RETURN_CHAR L'\r'
108 #define TEXT_EMPTY L""
109 #define TEXT_BLANK L" "
110 #define TEXT_RETURN_LINEFEED L"\r\n"
111 #define TEXT_LINEFEED L"\n"
112 #define TEXT_CHARRATIO_GAPDELTA 0.070
CPDF_TextPage(const CPDF_Page * pPage,int flags)113 CPDF_TextPage::CPDF_TextPage(const CPDF_Page* pPage, int flags)
114 : m_charList(512),
115 m_TempCharList(50),
116 m_pPreTextObj(NULL),
117 m_IsParsered(FALSE),
118 m_TextlineDir(-1),
119 m_CurlineRect(0, 0, 0, 0)
120 {
121 m_pPage = pPage;
122 m_parserflag = flags;
123 m_TextBuf.EstimateSize(0, 10240);
124 pPage->GetDisplayMatrix(m_DisplayMatrix, 0, 0, (int) pPage->GetPageWidth(), (int)pPage->GetPageHeight(), 0);
125 }
CPDF_TextPage(const CPDF_Page * pPage,CPDFText_ParseOptions ParserOptions)126 CPDF_TextPage::CPDF_TextPage(const CPDF_Page* pPage, CPDFText_ParseOptions ParserOptions)
127 : m_ParseOptions(ParserOptions)
128 , m_charList(512)
129 , m_TempCharList(50)
130 , m_pPreTextObj(NULL)
131 , m_IsParsered(FALSE)
132 , m_TextlineDir(-1)
133 , m_CurlineRect(0, 0, 0, 0)
134 {
135 m_pPage = pPage;
136 m_parserflag = 0;
137 m_TextBuf.EstimateSize(0, 10240);
138 pPage->GetDisplayMatrix(m_DisplayMatrix, 0, 0, (int) pPage->GetPageWidth(), (int)pPage->GetPageHeight(), 0);
139 }
CPDF_TextPage(const CPDF_PageObjects * pPage,int flags)140 CPDF_TextPage::CPDF_TextPage(const CPDF_PageObjects* pPage, int flags)
141 : m_charList(512),
142 m_TempCharList(50),
143 m_pPreTextObj(NULL),
144 m_IsParsered(FALSE),
145 m_TextlineDir(-1),
146 m_CurlineRect(0, 0, 0, 0)
147 {
148 m_pPage = pPage;
149 m_parserflag = flags;
150 m_TextBuf.EstimateSize(0, 10240);
151 CFX_FloatRect pageRect = pPage->CalcBoundingBox();
152 m_DisplayMatrix = CFX_AffineMatrix(1, 0, 0, -1, pageRect.right, pageRect.top);
153 }
NormalizeObjects(FX_BOOL bNormalize)154 void CPDF_TextPage::NormalizeObjects(FX_BOOL bNormalize)
155 {
156 m_ParseOptions.m_bNormalizeObjs = bNormalize;
157 }
IsControlChar(PAGECHAR_INFO * pCharInfo)158 FX_BOOL CPDF_TextPage::IsControlChar(PAGECHAR_INFO* pCharInfo)
159 {
160 if(!pCharInfo) {
161 return FALSE;
162 }
163 switch(pCharInfo->m_Unicode) {
164 case 0x2:
165 case 0x3:
166 case 0x93:
167 case 0x94:
168 case 0x96:
169 case 0x97:
170 case 0x98:
171 case 0xfffe:
172 if(pCharInfo->m_Flag == FPDFTEXT_CHAR_HYPHEN) {
173 return FALSE;
174 } else {
175 return TRUE;
176 }
177 default:
178 return FALSE;
179 }
180 }
ParseTextPage()181 FX_BOOL CPDF_TextPage::ParseTextPage()
182 {
183 if (!m_pPage) {
184 m_IsParsered = FALSE;
185 return FALSE;
186 }
187 m_IsParsered = FALSE;
188 m_TextBuf.Clear();
189 m_charList.RemoveAll();
190 m_pPreTextObj = NULL;
191 ProcessObject();
192 m_IsParsered = TRUE;
193 if(!m_ParseOptions.m_bGetCharCodeOnly) {
194 m_CharIndex.RemoveAll();
195 int nCount = m_charList.GetSize();
196 if(nCount) {
197 m_CharIndex.Add(0);
198 }
199 for(int i = 0; i < nCount; i++) {
200 int indexSize = m_CharIndex.GetSize();
201 FX_BOOL bNormal = FALSE;
202 PAGECHAR_INFO charinfo = *(PAGECHAR_INFO*)m_charList.GetAt(i);
203 if(charinfo.m_Flag == FPDFTEXT_CHAR_GENERATED) {
204 bNormal = TRUE;
205 }
206 else if(charinfo.m_Unicode == 0 || IsControlChar(&charinfo))
207 bNormal = FALSE;
208 else {
209 bNormal = TRUE;
210 }
211 if(bNormal) {
212 if(indexSize % 2) {
213 m_CharIndex.Add(1);
214 } else {
215 if(indexSize <= 0) {
216 continue;
217 }
218 m_CharIndex.SetAt(indexSize - 1, m_CharIndex.GetAt(indexSize - 1) + 1);
219 }
220 } else {
221 if(indexSize % 2) {
222 if(indexSize <= 0) {
223 continue;
224 }
225 m_CharIndex.SetAt(indexSize - 1, i + 1);
226 } else {
227 m_CharIndex.Add(i + 1);
228 }
229 }
230 }
231 int indexSize = m_CharIndex.GetSize();
232 if(indexSize % 2) {
233 m_CharIndex.RemoveAt(indexSize - 1);
234 }
235 }
236 return TRUE;
237 }
CountChars() const238 int CPDF_TextPage::CountChars() const
239 {
240 if(m_ParseOptions.m_bGetCharCodeOnly) {
241 return m_TextBuf.GetSize();
242 }
243 return m_charList.GetSize();
244 }
CharIndexFromTextIndex(int TextIndex) const245 int CPDF_TextPage::CharIndexFromTextIndex(int TextIndex) const
246 {
247 int indexSize = m_CharIndex.GetSize();
248 int count = 0;
249 for(int i = 0; i < indexSize; i += 2) {
250 count += m_CharIndex.GetAt(i + 1);
251 if(count > TextIndex) {
252 return TextIndex - count + m_CharIndex.GetAt(i + 1) + m_CharIndex.GetAt(i);
253 }
254 }
255 return -1;
256 }
TextIndexFromCharIndex(int CharIndex) const257 int CPDF_TextPage::TextIndexFromCharIndex(int CharIndex) const
258 {
259 int indexSize = m_CharIndex.GetSize();
260 int count = 0;
261 for(int i = 0; i < indexSize; i += 2) {
262 count += m_CharIndex.GetAt(i + 1);
263 if(m_CharIndex.GetAt(i + 1) + m_CharIndex.GetAt(i) > CharIndex) {
264 if(CharIndex - m_CharIndex.GetAt(i) < 0) {
265 return -1;
266 }
267 return CharIndex - m_CharIndex.GetAt(i) + count - m_CharIndex.GetAt(i + 1);
268 }
269 }
270 return -1;
271 }
GetRectArray(int start,int nCount,CFX_RectArray & rectArray) const272 void CPDF_TextPage::GetRectArray(int start, int nCount, CFX_RectArray& rectArray) const
273 {
274 if(m_ParseOptions.m_bGetCharCodeOnly) {
275 return;
276 }
277 if(start < 0 || nCount == 0) {
278 return;
279 }
280 if (!m_IsParsered) {
281 return;
282 }
283 PAGECHAR_INFO info_curchar;
284 CPDF_TextObject* pCurObj = NULL;
285 CFX_FloatRect rect;
286 int curPos = start;
287 FX_BOOL flagNewRect = TRUE;
288 if (nCount + start > m_charList.GetSize() || nCount == -1) {
289 nCount = m_charList.GetSize() - start;
290 }
291 while (nCount--) {
292 info_curchar = *(PAGECHAR_INFO*)m_charList.GetAt(curPos++);
293 if (info_curchar.m_Flag == FPDFTEXT_CHAR_GENERATED) {
294 continue;
295 }
296 if(info_curchar.m_CharBox.Width() < 0.01 || info_curchar.m_CharBox.Height() < 0.01) {
297 continue;
298 }
299 if(!pCurObj) {
300 pCurObj = info_curchar.m_pTextObj;
301 }
302 if (pCurObj != info_curchar.m_pTextObj) {
303 rectArray.Add(rect);
304 pCurObj = info_curchar.m_pTextObj;
305 flagNewRect = TRUE;
306 }
307 if (flagNewRect) {
308 FX_FLOAT orgX = info_curchar.m_OriginX, orgY = info_curchar.m_OriginY;
309 CFX_AffineMatrix matrix, matrix_reverse;
310 info_curchar.m_pTextObj->GetTextMatrix(&matrix);
311 matrix.Concat(info_curchar.m_Matrix);
312 matrix_reverse.SetReverse(matrix);
313 matrix_reverse.Transform(orgX, orgY);
314 rect.left = info_curchar.m_CharBox.left;
315 rect.right = info_curchar.m_CharBox.right;
316 if (pCurObj->GetFont()->GetTypeDescent()) {
317 rect.bottom = orgY + pCurObj->GetFont()->GetTypeDescent() * pCurObj->GetFontSize() / 1000;
318 FX_FLOAT xPosTemp = orgX;
319 matrix.Transform(xPosTemp, rect.bottom);
320 } else {
321 rect.bottom = info_curchar.m_CharBox.bottom;
322 }
323 if (pCurObj->GetFont()->GetTypeAscent()) {
324 rect.top = orgY + pCurObj->GetFont()->GetTypeAscent() * pCurObj->GetFontSize() / 1000;
325 FX_FLOAT xPosTemp = orgX + GetCharWidth(info_curchar.m_CharCode, pCurObj->GetFont()) * pCurObj->GetFontSize() / 1000;
326 matrix.Transform(xPosTemp, rect.top);
327 } else {
328 rect.top = info_curchar.m_CharBox.top;
329 }
330 flagNewRect = FALSE;
331 rect = info_curchar.m_CharBox;
332 rect.Normalize();
333 } else {
334 info_curchar.m_CharBox.Normalize();
335 if (rect.left > info_curchar.m_CharBox.left) {
336 rect.left = info_curchar.m_CharBox.left;
337 }
338 if (rect.right < info_curchar.m_CharBox.right) {
339 rect.right = info_curchar.m_CharBox.right;
340 }
341 if ( rect.top < info_curchar.m_CharBox.top) {
342 rect.top = info_curchar.m_CharBox.top;
343 }
344 if (rect.bottom > info_curchar.m_CharBox.bottom) {
345 rect.bottom = info_curchar.m_CharBox.bottom;
346 }
347 }
348 }
349 rectArray.Add(rect);
350 return;
351 }
GetIndexAtPos(CPDF_Point point,FX_FLOAT xTorelance,FX_FLOAT yTorelance) const352 int CPDF_TextPage::GetIndexAtPos(CPDF_Point point , FX_FLOAT xTorelance, FX_FLOAT yTorelance) const
353 {
354 if(m_ParseOptions.m_bGetCharCodeOnly) {
355 return -3;
356 }
357 if (!m_IsParsered) {
358 return -3;
359 }
360 int pos = 0;
361 int NearPos = -1;
362 double xdif = 5000, ydif = 5000;
363 while(pos < m_charList.GetSize()) {
364 PAGECHAR_INFO charinfo = *(PAGECHAR_INFO*)(m_charList.GetAt(pos));
365 CFX_FloatRect charrect = charinfo.m_CharBox;
366 if (charrect.Contains(point.x, point.y)) {
367 break;
368 }
369 if (xTorelance > 0 || yTorelance > 0) {
370 CFX_FloatRect charRectExt;
371 charrect.Normalize();
372 charRectExt.left = charrect.left - xTorelance / 2;
373 charRectExt.right = charrect.right + xTorelance / 2;
374 charRectExt.top = charrect.top + yTorelance / 2;
375 charRectExt.bottom = charrect.bottom - yTorelance / 2;
376 if (charRectExt.Contains(point.x, point.y)) {
377 double curXdif, curYdif;
378 curXdif = FXSYS_fabs(point.x - charrect.left) < FXSYS_fabs(point.x - charrect.right) ? FXSYS_fabs(point.x - charrect.left) : FXSYS_fabs(point.x - charrect.right);
379 curYdif = FXSYS_fabs(point.y - charrect.bottom) < FXSYS_fabs(point.y - charrect.top ) ? FXSYS_fabs(point.y - charrect.bottom) : FXSYS_fabs(point.y - charrect.top);
380 if (curYdif + curXdif < xdif + ydif) {
381 ydif = curYdif;
382 xdif = curXdif;
383 NearPos = pos;
384 }
385 }
386 }
387 ++pos;
388 }
389 if (pos >= m_charList.GetSize()) {
390 pos = NearPos;
391 }
392 return pos;
393 }
GetTextByRect(const CFX_FloatRect & rect) const394 CFX_WideString CPDF_TextPage::GetTextByRect(const CFX_FloatRect& rect) const
395 {
396 CFX_WideString strText;
397 if(m_ParseOptions.m_bGetCharCodeOnly || !m_IsParsered) {
398 return strText;
399 }
400 int nCount = m_charList.GetSize();
401 int pos = 0;
402 FX_FLOAT posy = 0;
403 FX_BOOL IsContainPreChar = FALSE;
404 FX_BOOL ISAddLineFeed = FALSE;
405 while (pos < nCount) {
406 PAGECHAR_INFO charinfo = *(PAGECHAR_INFO*)m_charList.GetAt(pos++);
407 if (IsRectIntersect(rect, charinfo.m_CharBox)) {
408 if (FXSYS_fabs(posy - charinfo.m_OriginY) > 0 && !IsContainPreChar && ISAddLineFeed) {
409 posy = charinfo.m_OriginY;
410 if (strText.GetLength() > 0) {
411 strText += L"\r\n";
412 }
413 }
414 IsContainPreChar = TRUE;
415 ISAddLineFeed = FALSE;
416 if (charinfo.m_Unicode) {
417 strText += charinfo.m_Unicode;
418 }
419 } else if (charinfo.m_Unicode == 32) {
420 if (IsContainPreChar && charinfo.m_Unicode) {
421 strText += charinfo.m_Unicode;
422 IsContainPreChar = FALSE;
423 ISAddLineFeed = FALSE;
424 }
425 } else {
426 IsContainPreChar = FALSE;
427 ISAddLineFeed = TRUE;
428 }
429 }
430 return strText;
431 }
GetRectsArrayByRect(const CFX_FloatRect & rect,CFX_RectArray & resRectArray) const432 void CPDF_TextPage::GetRectsArrayByRect(const CFX_FloatRect& rect, CFX_RectArray& resRectArray) const
433 {
434 if(m_ParseOptions.m_bGetCharCodeOnly) {
435 return;
436 }
437 if (!m_IsParsered) {
438 return;
439 }
440 CFX_FloatRect curRect;
441 FX_BOOL flagNewRect = TRUE;
442 CPDF_TextObject* pCurObj = NULL;
443 int nCount = m_charList.GetSize();
444 int pos = 0;
445 while (pos < nCount) {
446 PAGECHAR_INFO info_curchar = *(PAGECHAR_INFO*)m_charList.GetAt(pos++);
447 if (info_curchar.m_Flag == FPDFTEXT_CHAR_GENERATED) {
448 continue;
449 }
450 if (IsRectIntersect(rect, info_curchar.m_CharBox)) {
451 if(!pCurObj) {
452 pCurObj = info_curchar.m_pTextObj;
453 }
454 if (pCurObj != info_curchar.m_pTextObj) {
455 resRectArray.Add(curRect);
456 pCurObj = info_curchar.m_pTextObj;
457 flagNewRect = TRUE;
458 }
459 if (flagNewRect) {
460 curRect = info_curchar.m_CharBox;
461 flagNewRect = FALSE;
462 curRect.Normalize();
463 } else {
464 info_curchar.m_CharBox.Normalize();
465 if (curRect.left > info_curchar.m_CharBox.left) {
466 curRect.left = info_curchar.m_CharBox.left;
467 }
468 if (curRect.right < info_curchar.m_CharBox.right) {
469 curRect.right = info_curchar.m_CharBox.right;
470 }
471 if ( curRect.top < info_curchar.m_CharBox.top) {
472 curRect.top = info_curchar.m_CharBox.top;
473 }
474 if (curRect.bottom > info_curchar.m_CharBox.bottom) {
475 curRect.bottom = info_curchar.m_CharBox.bottom;
476 }
477 }
478 }
479 }
480 resRectArray.Add(curRect);
481 return;
482 }
GetIndexAtPos(FX_FLOAT x,FX_FLOAT y,FX_FLOAT xTorelance,FX_FLOAT yTorelance) const483 int CPDF_TextPage::GetIndexAtPos(FX_FLOAT x, FX_FLOAT y, FX_FLOAT xTorelance, FX_FLOAT yTorelance) const
484 {
485 if(m_ParseOptions.m_bGetCharCodeOnly) {
486 return -3;
487 }
488 CPDF_Point point(x, y);
489 return GetIndexAtPos(point, xTorelance, yTorelance);
490 }
GetOrderByDirection(int order,int direction) const491 int CPDF_TextPage::GetOrderByDirection(int order, int direction) const
492 {
493 if(m_ParseOptions.m_bGetCharCodeOnly) {
494 return -3;
495 }
496 if (!m_IsParsered) {
497 return -3;
498 }
499 if (direction == FPDFTEXT_RIGHT || direction == FPDFTEXT_LEFT) {
500 order += direction;
501 while(order >= 0 && order < m_charList.GetSize()) {
502 PAGECHAR_INFO cinfo = *(PAGECHAR_INFO*)m_charList.GetAt(order);
503 if (cinfo.m_Flag != FPDFTEXT_CHAR_GENERATED) {
504 break;
505 } else {
506 if (cinfo.m_Unicode == TEXT_LINEFEED_CHAR || cinfo.m_Unicode == TEXT_RETURN_CHAR) {
507 order += direction;
508 } else {
509 break;
510 }
511 }
512 }
513 if (order >= m_charList.GetSize()) {
514 order = -2;
515 }
516 return order;
517 }
518 PAGECHAR_INFO charinfo;
519 charinfo = *(PAGECHAR_INFO*)m_charList.GetAt(order);
520 CPDF_Point curPos(charinfo.m_OriginX, charinfo.m_OriginY);
521 FX_FLOAT difPosY = 0.0, minXdif = 1000;
522 int minIndex = -2;
523 int index = order;
524 FX_FLOAT height = charinfo.m_CharBox.Height();
525 if (direction == FPDFTEXT_UP) {
526 minIndex = -1;
527 while (1) {
528 if (--index < 0) {
529 return -1;
530 }
531 charinfo = *(PAGECHAR_INFO*)m_charList.GetAt(index);
532 if (FXSYS_fabs(charinfo.m_OriginY - curPos.y) > FX_MAX(height, charinfo.m_CharBox.Height()) / 2) {
533 difPosY = charinfo.m_OriginY;
534 minIndex = index;
535 break;
536 }
537 }
538 FX_FLOAT PreXdif = charinfo.m_OriginX - curPos.x;
539 minXdif = PreXdif;
540 if (PreXdif == 0) {
541 return index;
542 }
543 FX_FLOAT curXdif = 0;
544 while (--index >= 0) {
545 charinfo = *(PAGECHAR_INFO*)m_charList.GetAt(index);
546 if (difPosY != charinfo.m_OriginY) {
547 break;
548 }
549 curXdif = charinfo.m_OriginX - curPos.x;
550 if (curXdif == 0) {
551 return index;
552 }
553 int signflag = 0;
554 if (curXdif > 0) {
555 signflag = 1;
556 } else {
557 signflag = -1;
558 }
559 if (signflag * PreXdif < 0) {
560 if (FXSYS_fabs(PreXdif) < FXSYS_fabs(curXdif)) {
561 return index + 1;
562 } else {
563 return index;
564 }
565 }
566 if (FXSYS_fabs(curXdif) < FXSYS_fabs(minXdif)) {
567 minIndex = index;
568 minXdif = curXdif;
569 }
570 PreXdif = curXdif;
571 if (difPosY != charinfo.m_OriginY) {
572 break;
573 }
574 }
575 return minIndex;
576 } else if(FPDFTEXT_DOWN) {
577 minIndex = -2;
578 while (1) {
579 if (++index > m_charList.GetSize() - 1) {
580 return minIndex;
581 }
582 charinfo = *(PAGECHAR_INFO*)m_charList.GetAt(index);
583 if (FXSYS_fabs(charinfo.m_OriginY - curPos.y) > FX_MAX(height, charinfo.m_CharBox.Height()) / 2) {
584 difPosY = charinfo.m_OriginY;
585 minIndex = index;
586 break;
587 }
588 }
589 FX_FLOAT PreXdif = charinfo.m_OriginX - curPos.x;
590 minXdif = PreXdif;
591 if (PreXdif == 0) {
592 return index;
593 }
594 FX_FLOAT curXdif = 0;
595 while (++index < m_charList.GetSize()) {
596 charinfo = *(PAGECHAR_INFO*)m_charList.GetAt(index);
597 if (difPosY != charinfo.m_OriginY) {
598 break;
599 }
600 curXdif = charinfo.m_OriginX - curPos.x;
601 if (curXdif == 0) {
602 return index;
603 }
604 int signflag = 0;
605 if (curXdif > 0) {
606 signflag = 1;
607 } else {
608 signflag = -1;
609 }
610 if (signflag * PreXdif < 0) {
611 if (FXSYS_fabs(PreXdif) < FXSYS_fabs(curXdif)) {
612 return index - 1;
613 } else {
614 return index;
615 }
616 }
617 if (FXSYS_fabs(curXdif) < FXSYS_fabs(minXdif)) {
618 minXdif = curXdif;
619 minIndex = index;
620 }
621 PreXdif = curXdif;
622 }
623 return minIndex;
624 }
625 }
GetCharInfo(int index,FPDF_CHAR_INFO & info) const626 void CPDF_TextPage::GetCharInfo(int index, FPDF_CHAR_INFO & info) const
627 {
628 if(m_ParseOptions.m_bGetCharCodeOnly) {
629 return;
630 }
631 if (!m_IsParsered) {
632 return;
633 }
634 if (index < 0 || index >= m_charList.GetSize()) {
635 return;
636 }
637 PAGECHAR_INFO charinfo;
638 charinfo = *(PAGECHAR_INFO*)m_charList.GetAt(index);
639 info.m_Charcode = charinfo.m_CharCode;
640 info.m_OriginX = charinfo.m_OriginX;
641 info.m_OriginY = charinfo.m_OriginY;
642 info.m_Unicode = charinfo.m_Unicode;
643 info.m_Flag = charinfo.m_Flag;
644 info.m_CharBox = charinfo.m_CharBox;
645 info.m_pTextObj = charinfo.m_pTextObj;
646 if (charinfo.m_pTextObj && charinfo.m_pTextObj->GetFont()) {
647 info.m_FontSize = charinfo.m_pTextObj->GetFontSize();
648 }
649 info.m_Matrix.Copy(charinfo.m_Matrix);
650 return;
651 }
CheckMarkedContentObject(FX_INT32 & start,FX_INT32 & nCount) const652 void CPDF_TextPage::CheckMarkedContentObject(FX_INT32& start, FX_INT32& nCount) const
653 {
654 PAGECHAR_INFO charinfo = *(PAGECHAR_INFO*)m_charList.GetAt(start);
655 PAGECHAR_INFO charinfo2 = *(PAGECHAR_INFO*)m_charList.GetAt(start + nCount - 1);
656 if (FPDFTEXT_CHAR_PIECE != charinfo.m_Flag && FPDFTEXT_CHAR_PIECE != charinfo2.m_Flag) {
657 return;
658 }
659 if (FPDFTEXT_CHAR_PIECE == charinfo.m_Flag) {
660 PAGECHAR_INFO charinfo1 = charinfo;
661 int startIndex = start;
662 while(FPDFTEXT_CHAR_PIECE == charinfo1.m_Flag && charinfo1.m_Index == charinfo.m_Index) {
663 startIndex--;
664 if (startIndex < 0) {
665 break;
666 }
667 charinfo1 = *(PAGECHAR_INFO*)m_charList.GetAt(startIndex);
668 }
669 startIndex++;
670 start = startIndex;
671 }
672 if (FPDFTEXT_CHAR_PIECE == charinfo2.m_Flag) {
673 PAGECHAR_INFO charinfo3 = charinfo2;
674 int endIndex = start + nCount - 1;
675 while(FPDFTEXT_CHAR_PIECE == charinfo3.m_Flag && charinfo3.m_Index == charinfo2.m_Index) {
676 endIndex++;
677 if (endIndex >= m_charList.GetSize()) {
678 break;
679 }
680 charinfo3 = *(PAGECHAR_INFO*)m_charList.GetAt(endIndex);
681 }
682 endIndex--;
683 nCount = endIndex - start + 1;
684 }
685 }
GetPageText(int start,int nCount) const686 CFX_WideString CPDF_TextPage::GetPageText(int start , int nCount) const
687 {
688 if (!m_IsParsered || nCount == 0) {
689 return L"";
690 }
691 if (start < 0) {
692 start = 0;
693 }
694 if (nCount == -1) {
695 nCount = m_charList.GetSize() - start;
696 return m_TextBuf.GetWideString().Mid(start, m_TextBuf.GetWideString().GetLength());
697 }
698 if(nCount <= 0 || m_charList.GetSize() <= 0) {
699 return L"";
700 }
701 if(nCount + start > m_charList.GetSize() - 1) {
702 nCount = m_charList.GetSize() - start;
703 }
704 if (nCount <= 0) {
705 return L"";
706 }
707 CheckMarkedContentObject(start, nCount);
708 int startindex = 0;
709 PAGECHAR_INFO charinfo = *(PAGECHAR_INFO*)m_charList.GetAt(start);
710 int startOffset = 0;
711 while(charinfo.m_Index == -1) {
712 startOffset++;
713 if (startOffset > nCount || start + startOffset >= m_charList.GetSize()) {
714 return L"";
715 }
716 charinfo = *(PAGECHAR_INFO*)m_charList.GetAt(start + startOffset);
717 }
718 startindex = charinfo.m_Index;
719 charinfo = *(PAGECHAR_INFO*)m_charList.GetAt(start + nCount - 1);
720 int nCountOffset = 0;
721 while (charinfo.m_Index == -1) {
722 nCountOffset++;
723 if (nCountOffset >= nCount) {
724 return L"";
725 }
726 charinfo = *(PAGECHAR_INFO*)m_charList.GetAt(start + nCount - nCountOffset - 1);
727 }
728 nCount = start + nCount - nCountOffset - startindex;
729 if(nCount <= 0) {
730 return L"";
731 }
732 return m_TextBuf.GetWideString().Mid(startindex, nCount);
733 }
CountRects(int start,int nCount)734 int CPDF_TextPage::CountRects(int start, int nCount)
735 {
736 if(m_ParseOptions.m_bGetCharCodeOnly) {
737 return -1;
738 }
739 if (!m_IsParsered) {
740 return -1;
741 }
742 if (start < 0) {
743 return -1;
744 }
745 if (nCount == -1 || nCount + start > m_charList.GetSize() ) {
746 nCount = m_charList.GetSize() - start;
747 }
748 m_SelRects.RemoveAll();
749 GetRectArray(start, nCount, m_SelRects);
750 return m_SelRects.GetSize();
751 }
GetRect(int rectIndex,FX_FLOAT & left,FX_FLOAT & top,FX_FLOAT & right,FX_FLOAT & bottom) const752 void CPDF_TextPage::GetRect(int rectIndex, FX_FLOAT& left, FX_FLOAT& top, FX_FLOAT& right, FX_FLOAT &bottom) const
753 {
754 if(m_ParseOptions.m_bGetCharCodeOnly) {
755 return ;
756 }
757 if (!m_IsParsered || rectIndex < 0 || rectIndex >= m_SelRects.GetSize()) {
758 return;
759 }
760 left = m_SelRects.GetAt(rectIndex).left;
761 top = m_SelRects.GetAt(rectIndex).top;
762 right = m_SelRects.GetAt(rectIndex).right;
763 bottom = m_SelRects.GetAt(rectIndex).bottom;
764 }
GetBaselineRotate(int start,int end,int & Rotate)765 FX_BOOL CPDF_TextPage::GetBaselineRotate(int start, int end, int& Rotate)
766 {
767 if(m_ParseOptions.m_bGetCharCodeOnly) {
768 return FALSE;
769 }
770 if(end == start) {
771 return FALSE;
772 }
773 FX_FLOAT dx, dy;
774 FPDF_CHAR_INFO info1, info2;
775 GetCharInfo(start, info1);
776 GetCharInfo(end, info2);
777 while(info2.m_CharBox.Width() == 0 || info2.m_CharBox.Height() == 0) {
778 end--;
779 if(end <= start) {
780 return FALSE;
781 }
782 GetCharInfo(end, info2);
783 }
784 dx = (info2.m_OriginX - info1.m_OriginX);
785 dy = (info2.m_OriginY - info1.m_OriginY);
786 if(dx == 0) {
787 if(dy > 0) {
788 Rotate = 90;
789 } else if (dy < 0) {
790 Rotate = 270;
791 } else {
792 Rotate = 0;
793 }
794 } else {
795 float a = FXSYS_atan2(dy, dx);
796 Rotate = (int)(a * 180 / FX_PI + 0.5);
797 }
798 if(Rotate < 0) {
799 Rotate = -Rotate;
800 } else if(Rotate > 0) {
801 Rotate = 360 - Rotate;
802 }
803 return TRUE;
804 }
GetBaselineRotate(const CFX_FloatRect & rect,int & Rotate)805 FX_BOOL CPDF_TextPage::GetBaselineRotate(const CFX_FloatRect& rect , int& Rotate)
806 {
807 if(m_ParseOptions.m_bGetCharCodeOnly) {
808 return FALSE;
809 }
810 int start, end, count, n = CountBoundedSegments(rect.left, rect.top, rect.right, rect.bottom, TRUE);
811 if(n < 1) {
812 return FALSE;
813 }
814 if(n > 1) {
815 GetBoundedSegment(n - 1, start, count);
816 end = start + count - 1;
817 GetBoundedSegment(0, start, count);
818 } else {
819 GetBoundedSegment(0, start, count);
820 end = start + count - 1;
821 }
822 return GetBaselineRotate(start, end, Rotate);
823 }
GetBaselineRotate(int rectIndex,int & Rotate)824 FX_BOOL CPDF_TextPage::GetBaselineRotate(int rectIndex, int& Rotate)
825 {
826 if(m_ParseOptions.m_bGetCharCodeOnly) {
827 return FALSE;
828 }
829 if (!m_IsParsered || rectIndex < 0 || rectIndex > m_SelRects.GetSize()) {
830 return FALSE;
831 }
832 CFX_FloatRect rect = m_SelRects.GetAt(rectIndex);
833 return GetBaselineRotate(rect , Rotate);
834 }
CountBoundedSegments(FX_FLOAT left,FX_FLOAT top,FX_FLOAT right,FX_FLOAT bottom,FX_BOOL bContains)835 int CPDF_TextPage::CountBoundedSegments(FX_FLOAT left, FX_FLOAT top, FX_FLOAT right, FX_FLOAT bottom, FX_BOOL bContains )
836 {
837 if(m_ParseOptions.m_bGetCharCodeOnly) {
838 return -1;
839 }
840 m_Segment.RemoveAll();
841 if (!m_IsParsered) {
842 return -1;
843 }
844 CFX_FloatRect rect(left, bottom, right, top);
845 rect.Normalize();
846 int nCount = m_charList.GetSize();
847 int pos = 0;
848 FPDF_SEGMENT segment;
849 segment.m_Start = 0;
850 segment.m_nCount = 0;
851 FX_BOOL segmentStatus = 0;
852 FX_BOOL IsContainPreChar = FALSE;
853 while (pos < nCount) {
854 PAGECHAR_INFO charinfo = *(PAGECHAR_INFO*)m_charList.GetAt(pos);
855 if(bContains && rect.Contains(charinfo.m_CharBox)) {
856 if (segmentStatus == 0 || segmentStatus == 2) {
857 segment.m_Start = pos;
858 segment.m_nCount = 1;
859 segmentStatus = 1;
860 } else if (segmentStatus == 1) {
861 segment.m_nCount++;
862 }
863 IsContainPreChar = TRUE;
864 } else if (!bContains && (IsRectIntersect(rect, charinfo.m_CharBox) || rect.Contains(charinfo.m_OriginX, charinfo.m_OriginY))) {
865 if (segmentStatus == 0 || segmentStatus == 2) {
866 segment.m_Start = pos;
867 segment.m_nCount = 1;
868 segmentStatus = 1;
869 } else if (segmentStatus == 1) {
870 segment.m_nCount++;
871 }
872 IsContainPreChar = TRUE;
873 } else if (charinfo.m_Unicode == 32) {
874 if (IsContainPreChar == TRUE) {
875 if (segmentStatus == 0 || segmentStatus == 2) {
876 segment.m_Start = pos;
877 segment.m_nCount = 1;
878 segmentStatus = 1;
879 } else if (segmentStatus == 1) {
880 segment.m_nCount++;
881 }
882 IsContainPreChar = FALSE;
883 } else {
884 if (segmentStatus == 1) {
885 segmentStatus = 2;
886 m_Segment.Add(segment);
887 segment.m_Start = 0;
888 segment.m_nCount = 0;
889 }
890 }
891 } else {
892 if (segmentStatus == 1) {
893 segmentStatus = 2;
894 m_Segment.Add(segment);
895 segment.m_Start = 0;
896 segment.m_nCount = 0;
897 }
898 IsContainPreChar = FALSE;
899 }
900 pos++;
901 }
902 if (segmentStatus == 1) {
903 segmentStatus = 2;
904 m_Segment.Add(segment);
905 segment.m_Start = 0;
906 segment.m_nCount = 0;
907 }
908 return m_Segment.GetSize();
909 }
GetBoundedSegment(int index,int & start,int & count) const910 void CPDF_TextPage::GetBoundedSegment(int index, int& start, int& count) const
911 {
912 if(m_ParseOptions.m_bGetCharCodeOnly) {
913 return ;
914 }
915 if (index < 0 || index >= m_Segment.GetSize()) {
916 return;
917 }
918 start = m_Segment.GetAt(index).m_Start;
919 count = m_Segment.GetAt(index).m_nCount;
920 }
GetWordBreak(int index,int direction) const921 int CPDF_TextPage::GetWordBreak(int index, int direction) const
922 {
923 if(m_ParseOptions.m_bGetCharCodeOnly) {
924 return -1;
925 }
926 if (!m_IsParsered) {
927 return -1;
928 }
929 if (direction != FPDFTEXT_LEFT && direction != FPDFTEXT_RIGHT) {
930 return -1;
931 }
932 if (index < 0 || index >= m_charList.GetSize()) {
933 return -1;
934 }
935 PAGECHAR_INFO charinfo;
936 charinfo = *(PAGECHAR_INFO*)m_charList.GetAt(index);
937 if (charinfo.m_Index == -1 || charinfo.m_Flag == FPDFTEXT_CHAR_GENERATED) {
938 return index;
939 }
940 if (!IsLetter(charinfo.m_Unicode)) {
941 return index;
942 }
943 int breakPos = index;
944 if (direction == FPDFTEXT_LEFT) {
945 while (--breakPos > 0) {
946 charinfo = *(PAGECHAR_INFO*)m_charList.GetAt(breakPos);
947 if (!IsLetter(charinfo.m_Unicode)) {
948 return breakPos;
949 }
950 }
951 return breakPos;
952 } else if (direction == FPDFTEXT_RIGHT) {
953 while (++breakPos < m_charList.GetSize()) {
954 charinfo = *(PAGECHAR_INFO*)m_charList.GetAt(breakPos);
955 if (!IsLetter(charinfo.m_Unicode)) {
956 return breakPos;
957 }
958 }
959 return breakPos;
960 }
961 return breakPos;
962 }
FindTextlineFlowDirection()963 FX_INT32 CPDF_TextPage::FindTextlineFlowDirection()
964 {
965 if (!m_pPage) {
966 return -1;
967 }
968 const FX_INT32 nPageWidth = (FX_INT32)((CPDF_Page*)m_pPage)->GetPageWidth();
969 const FX_INT32 nPageHeight = (FX_INT32)((CPDF_Page*)m_pPage)->GetPageHeight();
970 CFX_ByteArray nHorizontalMask;
971 if (!nHorizontalMask.SetSize(nPageWidth)) {
972 return -1;
973 }
974 FX_BYTE* pDataH = nHorizontalMask.GetData();
975 CFX_ByteArray nVerticalMask;
976 if (!nVerticalMask.SetSize(nPageHeight)) {
977 return -1;
978 }
979 FX_BYTE* pDataV = nVerticalMask.GetData();
980 FX_INT32 index = 0;
981 FX_FLOAT fLineHeight = 0.0f;
982 CPDF_PageObject* pPageObj = NULL;
983 FX_POSITION pos = NULL;
984 pos = m_pPage->GetFirstObjectPosition();
985 if(!pos) {
986 return -1;
987 }
988 while(pos) {
989 pPageObj = m_pPage->GetNextObject(pos);
990 if(NULL == pPageObj) {
991 continue;
992 }
993 if(PDFPAGE_TEXT != pPageObj->m_Type) {
994 continue;
995 }
996 FX_INT32 minH = (FX_INT32)pPageObj->m_Left < 0 ? 0 : (FX_INT32)pPageObj->m_Left;
997 FX_INT32 maxH = (FX_INT32)pPageObj->m_Right > nPageWidth ? nPageWidth : (FX_INT32)pPageObj->m_Right;
998 FX_INT32 minV = (FX_INT32)pPageObj->m_Bottom < 0 ? 0 : (FX_INT32)pPageObj->m_Bottom;
999 FX_INT32 maxV = (FX_INT32)pPageObj->m_Top > nPageHeight ? nPageHeight : (FX_INT32)pPageObj->m_Top;
1000 if (minH >= maxH || minV >= maxV) {
1001 continue;
1002 }
1003 FXSYS_memset8(pDataH + minH, 1, maxH - minH);
1004 FXSYS_memset8(pDataV + minV, 1, maxV - minV);
1005 if (fLineHeight <= 0.0f) {
1006 fLineHeight = pPageObj->m_Top - pPageObj->m_Bottom;
1007 }
1008 pPageObj = NULL;
1009 }
1010 FX_INT32 nStartH = 0;
1011 FX_INT32 nEndH = 0;
1012 FX_FLOAT nSumH = 0.0f;
1013 for (index = 0; index < nPageWidth; index++)
1014 if(1 == nHorizontalMask[index]) {
1015 break;
1016 }
1017 nStartH = index;
1018 for (index = nPageWidth; index > 0; index--)
1019 if(1 == nHorizontalMask[index - 1]) {
1020 break;
1021 }
1022 nEndH = index;
1023 for (index = nStartH; index < nEndH; index++) {
1024 nSumH += nHorizontalMask[index];
1025 }
1026 nSumH /= nEndH - nStartH;
1027 FX_INT32 nStartV = 0;
1028 FX_INT32 nEndV = 0;
1029 FX_FLOAT nSumV = 0.0f;
1030 for (index = 0; index < nPageHeight; index++)
1031 if(1 == nVerticalMask[index]) {
1032 break;
1033 }
1034 nStartV = index;
1035 for (index = nPageHeight; index > 0; index--)
1036 if(1 == nVerticalMask[index - 1]) {
1037 break;
1038 }
1039 nEndV = index;
1040 for (index = nStartV; index < nEndV; index++) {
1041 nSumV += nVerticalMask[index];
1042 }
1043 nSumV /= nEndV - nStartV;
1044 if ((nEndV - nStartV) < (FX_INT32)(2 * fLineHeight)) {
1045 return 0;
1046 }
1047 if ((nEndH - nStartH) < (FX_INT32)(2 * fLineHeight)) {
1048 return 1;
1049 }
1050 if (nSumH > 0.8f) {
1051 return 0;
1052 }
1053 if (nSumH - nSumV > 0.0f) {
1054 return 0;
1055 }
1056 if (nSumV - nSumH > 0.0f) {
1057 return 1;
1058 }
1059 return -1;
1060 }
ProcessObject()1061 void CPDF_TextPage::ProcessObject()
1062 {
1063 CPDF_PageObject* pPageObj = NULL;
1064 if (!m_pPage) {
1065 return;
1066 }
1067 FX_POSITION pos;
1068 pos = m_pPage->GetFirstObjectPosition();
1069 if (!pos) {
1070 return;
1071 }
1072 m_TextlineDir = FindTextlineFlowDirection();
1073 int nCount = 0;
1074 while (pos) {
1075 pPageObj = m_pPage->GetNextObject(pos);
1076 if(pPageObj) {
1077 if(pPageObj->m_Type == PDFPAGE_TEXT) {
1078 CFX_AffineMatrix matrix;
1079 ProcessTextObject((CPDF_TextObject*)pPageObj, matrix, pos);
1080 nCount++;
1081 } else if (pPageObj->m_Type == PDFPAGE_FORM) {
1082 CFX_AffineMatrix formMatrix(1, 0, 0, 1, 0, 0);
1083 ProcessFormObject((CPDF_FormObject*)pPageObj, formMatrix);
1084 }
1085 }
1086 pPageObj = NULL;
1087 }
1088 int count = m_LineObj.GetSize();
1089 for(int i = 0; i < count; i++) {
1090 ProcessTextObject(m_LineObj.GetAt(i));
1091 }
1092 m_LineObj.RemoveAll();
1093 CloseTempLine();
1094 }
ProcessFormObject(CPDF_FormObject * pFormObj,const CFX_AffineMatrix & formMatrix)1095 void CPDF_TextPage::ProcessFormObject(CPDF_FormObject* pFormObj, const CFX_AffineMatrix& formMatrix)
1096 {
1097 CPDF_PageObject* pPageObj = NULL;
1098 FX_POSITION pos;
1099 if (!pFormObj) {
1100 return;
1101 }
1102 pos = pFormObj->m_pForm->GetFirstObjectPosition();
1103 if (!pos) {
1104 return;
1105 }
1106 CFX_AffineMatrix curFormMatrix;
1107 curFormMatrix.Copy(pFormObj->m_FormMatrix);
1108 curFormMatrix.Concat(formMatrix);
1109 while (pos) {
1110 pPageObj = pFormObj->m_pForm->GetNextObject(pos);
1111 if(pPageObj) {
1112 if(pPageObj->m_Type == PDFPAGE_TEXT) {
1113 ProcessTextObject((CPDF_TextObject*)pPageObj, curFormMatrix, pos);
1114 } else if (pPageObj->m_Type == PDFPAGE_FORM) {
1115 ProcessFormObject((CPDF_FormObject*)pPageObj, curFormMatrix);
1116 }
1117 }
1118 pPageObj = NULL;
1119 }
1120 }
GetCharWidth(FX_DWORD charCode,CPDF_Font * pFont) const1121 int CPDF_TextPage::GetCharWidth(FX_DWORD charCode, CPDF_Font* pFont) const
1122 {
1123 if(charCode == -1) {
1124 return 0;
1125 }
1126 int w = pFont->GetCharWidthF(charCode);
1127 if(w == 0) {
1128 CFX_ByteString str;
1129 pFont->AppendChar(str, charCode);
1130 w = pFont->GetStringWidth(str, 1);
1131 if(w == 0) {
1132 FX_RECT BBox;
1133 pFont->GetCharBBox(charCode, BBox);
1134 w = BBox.right - BBox.left;
1135 }
1136 }
1137 return w;
1138 }
OnPiece(IFX_BidiChar * pBidi,CFX_WideString & str)1139 void CPDF_TextPage::OnPiece(IFX_BidiChar* pBidi, CFX_WideString& str)
1140 {
1141 FX_INT32 start, count;
1142 FX_INT32 ret = pBidi->GetBidiInfo(start, count);
1143 if(ret == 2) {
1144 for(int i = start + count - 1; i >= start; i--) {
1145 m_TextBuf.AppendChar(str.GetAt(i));
1146 m_charList.Add(*(PAGECHAR_INFO*)m_TempCharList.GetAt(i));
1147 }
1148 } else {
1149 int end = start + count ;
1150 for(int i = start; i < end; i++) {
1151 m_TextBuf.AppendChar(str.GetAt(i));
1152 m_charList.Add(*(PAGECHAR_INFO*)m_TempCharList.GetAt(i));
1153 }
1154 }
1155 }
AddCharInfoByLRDirection(CFX_WideString & str,int i)1156 void CPDF_TextPage::AddCharInfoByLRDirection(CFX_WideString& str, int i)
1157 {
1158 PAGECHAR_INFO Info = *(PAGECHAR_INFO*)m_TempCharList.GetAt(i);
1159 FX_WCHAR wChar = str.GetAt(i);
1160 if(!IsControlChar(&Info)) {
1161 Info.m_Index = m_TextBuf.GetLength();
1162 if (wChar >= 0xFB00 && wChar <= 0xFB06) {
1163 FX_LPWSTR pDst = NULL;
1164 FX_STRSIZE nCount = FX_Unicode_GetNormalization(wChar, pDst);
1165 if (nCount >= 1) {
1166 pDst = FX_Alloc(FX_WCHAR, nCount);
1167 FX_Unicode_GetNormalization(wChar, pDst);
1168 for (int nIndex = 0; nIndex < nCount; nIndex++) {
1169 PAGECHAR_INFO Info2 = Info;
1170 Info2.m_Unicode = pDst[nIndex];
1171 Info2.m_Flag = FPDFTEXT_CHAR_PIECE;
1172 m_TextBuf.AppendChar(Info2.m_Unicode);
1173 if( !m_ParseOptions.m_bGetCharCodeOnly) {
1174 m_charList.Add(Info2);
1175 }
1176 }
1177 FX_Free(pDst);
1178 return;
1179 }
1180 }
1181 m_TextBuf.AppendChar(wChar);
1182 } else {
1183 Info.m_Index = -1;
1184 }
1185 if( !m_ParseOptions.m_bGetCharCodeOnly) {
1186 m_charList.Add(Info);
1187 }
1188 }
AddCharInfoByRLDirection(CFX_WideString & str,int i)1189 void CPDF_TextPage::AddCharInfoByRLDirection(CFX_WideString& str, int i)
1190 {
1191 PAGECHAR_INFO Info = *(PAGECHAR_INFO*)m_TempCharList.GetAt(i);
1192 if(!IsControlChar(&Info)) {
1193 Info.m_Index = m_TextBuf.GetLength();
1194 FX_WCHAR wChar = FX_GetMirrorChar(str.GetAt(i), TRUE, FALSE);
1195 FX_LPWSTR pDst = NULL;
1196 FX_STRSIZE nCount = FX_Unicode_GetNormalization(wChar, pDst);
1197 if (nCount >= 1) {
1198 pDst = FX_Alloc(FX_WCHAR, nCount);
1199 FX_Unicode_GetNormalization(wChar, pDst);
1200 for (int nIndex = 0; nIndex < nCount; nIndex++) {
1201 PAGECHAR_INFO Info2 = Info;
1202 Info2.m_Unicode = pDst[nIndex];
1203 Info2.m_Flag = FPDFTEXT_CHAR_PIECE;
1204 m_TextBuf.AppendChar(Info2.m_Unicode);
1205 if( !m_ParseOptions.m_bGetCharCodeOnly) {
1206 m_charList.Add(Info2);
1207 }
1208 }
1209 FX_Free(pDst);
1210 return;
1211 } else {
1212 Info.m_Unicode = wChar;
1213 }
1214 m_TextBuf.AppendChar(Info.m_Unicode);
1215 } else {
1216 Info.m_Index = -1;
1217 }
1218 if( !m_ParseOptions.m_bGetCharCodeOnly) {
1219 m_charList.Add(Info);
1220 }
1221 }
CloseTempLine()1222 void CPDF_TextPage::CloseTempLine()
1223 {
1224 int count1 = m_TempCharList.GetSize();
1225 if (count1 <= 0) {
1226 return;
1227 }
1228 IFX_BidiChar* BidiChar = IFX_BidiChar::Create();
1229 CFX_WideString str = m_TempTextBuf.GetWideString();
1230 CFX_WordArray order;
1231 FX_BOOL bR2L = FALSE;
1232 FX_INT32 start = 0, count = 0;
1233 int nR2L = 0, nL2R = 0;
1234 FX_BOOL bPrevSpace = FALSE;
1235 for (int i = 0; i < str.GetLength(); i++) {
1236 if(str.GetAt(i) == 32) {
1237 if(bPrevSpace) {
1238 m_TempTextBuf.Delete(i, 1);
1239 m_TempCharList.Delete(i);
1240 str.Delete(i);
1241 count1--;
1242 i--;
1243 continue;
1244 }
1245 bPrevSpace = TRUE;
1246 } else {
1247 bPrevSpace = FALSE;
1248 }
1249 if(BidiChar && BidiChar->AppendChar(str.GetAt(i))) {
1250 FX_INT32 ret = BidiChar->GetBidiInfo(start, count);
1251 order.Add(start);
1252 order.Add(count);
1253 order.Add(ret);
1254 if(!bR2L) {
1255 if(ret == 2) {
1256 nR2L++;
1257 } else if (ret == 1) {
1258 nL2R++;
1259 }
1260 }
1261 }
1262 }
1263 if(BidiChar && BidiChar->EndChar()) {
1264 FX_INT32 ret = BidiChar->GetBidiInfo(start, count);
1265 order.Add(start);
1266 order.Add(count);
1267 order.Add(ret);
1268 if(!bR2L) {
1269 if(ret == 2) {
1270 nR2L++;
1271 } else if(ret == 1) {
1272 nL2R++;
1273 }
1274 }
1275 }
1276 if(nR2L > 0 && nR2L >= nL2R) {
1277 bR2L = TRUE;
1278 }
1279 if(this->m_parserflag == FPDFTEXT_RLTB || bR2L) {
1280 int count = order.GetSize();
1281 for(int i = count - 1; i > 0; i -= 3) {
1282 int ret = order.GetAt(i);
1283 int start = order.GetAt(i - 2);
1284 int count1 = order.GetAt(i - 1);
1285 if(ret == 2 || ret == 0) {
1286 for(int j = start + count1 - 1; j >= start; j--) {
1287 AddCharInfoByRLDirection(str, j);
1288 }
1289 } else {
1290 int j = i;
1291 FX_BOOL bSymbol = FALSE;
1292 while(j > 0 && order.GetAt(j) != 2) {
1293 bSymbol = !order.GetAt(j);
1294 j -= 3;
1295 }
1296 int end = start + count1 ;
1297 int n = 0;
1298 if(bSymbol) {
1299 n = j + 6;
1300 } else {
1301 n = j + 3;
1302 }
1303 if(n >= i) {
1304 for(int m = start; m < end; m++) {
1305 AddCharInfoByLRDirection(str, m);
1306 }
1307 } else {
1308 j = i;
1309 i = n;
1310 for(; n <= j; n += 3) {
1311 int start = order.GetAt(n - 2);
1312 int count1 = order.GetAt(n - 1);
1313 int end = start + count1 ;
1314 for(int m = start; m < end; m++) {
1315 AddCharInfoByLRDirection(str, m);
1316 }
1317 }
1318 }
1319 }
1320 }
1321 } else {
1322 int count = order.GetSize();
1323 FX_BOOL bL2R = FALSE;
1324 for(int i = 0; i < count; i += 3) {
1325 int ret = order.GetAt(i + 2);
1326 int start = order.GetAt(i);
1327 int count1 = order.GetAt(i + 1);
1328 if(ret == 2 || (i == 0 && ret == 0 && !bL2R)) {
1329 int j = i + 3;
1330 while(bR2L && j < count) {
1331 if(order.GetAt(j + 2) == 1) {
1332 break;
1333 } else {
1334 j += 3;
1335 }
1336 }
1337 if(j == 3) {
1338 i = -3;
1339 bL2R = TRUE;
1340 continue;
1341 }
1342 int end = m_TempCharList.GetSize() - 1;
1343 if(j < count) {
1344 end = order.GetAt(j) - 1;
1345 }
1346 i = j - 3;
1347 for(int n = end; n >= start; n--) {
1348 AddCharInfoByRLDirection(str, n);
1349 }
1350 } else {
1351 int end = start + count1 ;
1352 for(int n = start; n < end; n++) {
1353 AddCharInfoByLRDirection(str, n);
1354 }
1355 }
1356 }
1357 }
1358 order.RemoveAll();
1359 m_TempCharList.RemoveAll();
1360 m_TempTextBuf.Delete(0, m_TempTextBuf.GetLength());
1361 BidiChar->Release();
1362 }
ProcessTextObject(CPDF_TextObject * pTextObj,const CFX_AffineMatrix & formMatrix,FX_POSITION ObjPos)1363 void CPDF_TextPage::ProcessTextObject(CPDF_TextObject* pTextObj, const CFX_AffineMatrix& formMatrix, FX_POSITION ObjPos)
1364 {
1365 CFX_FloatRect re(pTextObj->m_Left, pTextObj->m_Bottom, pTextObj->m_Right, pTextObj->m_Top);
1366 if(FXSYS_fabs(pTextObj->m_Right - pTextObj->m_Left) < 0.01f ) {
1367 return;
1368 }
1369 int count = m_LineObj.GetSize();
1370 PDFTEXT_Obj Obj;
1371 Obj.m_pTextObj = pTextObj;
1372 Obj.m_formMatrix = formMatrix;
1373 if(count == 0) {
1374 m_LineObj.Add(Obj);
1375 return;
1376 }
1377 if (IsSameAsPreTextObject(pTextObj, ObjPos)) {
1378 return;
1379 }
1380 PDFTEXT_Obj prev_Obj = m_LineObj.GetAt(count - 1);
1381 CPDF_TextObjectItem item;
1382 int nItem = prev_Obj.m_pTextObj->CountItems();
1383 prev_Obj.m_pTextObj->GetItemInfo(nItem - 1, &item);
1384 FX_FLOAT prev_width = GetCharWidth(item.m_CharCode, prev_Obj.m_pTextObj->GetFont()) * prev_Obj.m_pTextObj->GetFontSize() / 1000;
1385 CFX_AffineMatrix prev_matrix;
1386 prev_Obj.m_pTextObj->GetTextMatrix(&prev_matrix);
1387 prev_width = FXSYS_fabs(prev_width);
1388 prev_matrix.Concat(prev_Obj.m_formMatrix);
1389 prev_width = prev_matrix.TransformDistance(prev_width);
1390 pTextObj->GetItemInfo(0, &item);
1391 FX_FLOAT this_width = GetCharWidth(item.m_CharCode, pTextObj->GetFont()) * pTextObj->GetFontSize() / 1000;
1392 this_width = FXSYS_fabs(this_width);
1393 CFX_AffineMatrix this_matrix;
1394 pTextObj->GetTextMatrix(&this_matrix);
1395 this_width = FXSYS_fabs(this_width);
1396 this_matrix.Concat(formMatrix);
1397 this_width = this_matrix.TransformDistance(this_width);
1398 FX_FLOAT threshold = prev_width > this_width ? prev_width / 4 : this_width / 4;
1399 FX_FLOAT prev_x = prev_Obj.m_pTextObj->GetPosX(), prev_y = prev_Obj.m_pTextObj->GetPosY();
1400 prev_Obj.m_formMatrix.Transform(prev_x, prev_y);
1401 m_DisplayMatrix.Transform(prev_x, prev_y);
1402 FX_FLOAT this_x = pTextObj->GetPosX(), this_y = pTextObj->GetPosY();
1403 formMatrix.Transform(this_x, this_y);
1404 m_DisplayMatrix.Transform(this_x, this_y);
1405 if (FXSYS_fabs(this_y - prev_y) > threshold * 2) {
1406 for(int i = 0; i < count; i++) {
1407 ProcessTextObject(m_LineObj.GetAt(i));
1408 }
1409 m_LineObj.RemoveAll();
1410 m_LineObj.Add(Obj);
1411 return;
1412 }
1413 int i = 0;
1414 if(m_ParseOptions.m_bNormalizeObjs) {
1415 for(i = count - 1; i >= 0; i--) {
1416 PDFTEXT_Obj prev_Obj = m_LineObj.GetAt(i);
1417 CFX_AffineMatrix prev_matrix;
1418 prev_Obj.m_pTextObj->GetTextMatrix(&prev_matrix);
1419 FX_FLOAT Prev_x = prev_Obj.m_pTextObj->GetPosX(), Prev_y = prev_Obj.m_pTextObj->GetPosY();
1420 prev_Obj.m_formMatrix.Transform(Prev_x, Prev_y);
1421 m_DisplayMatrix.Transform(Prev_x, Prev_y);
1422 if(this_x >= Prev_x) {
1423 if(i == count - 1) {
1424 m_LineObj.Add(Obj);
1425 } else {
1426 m_LineObj.InsertAt(i + 1, Obj);
1427 }
1428 break;
1429 }
1430 }
1431 if(i < 0) {
1432 m_LineObj.InsertAt(0, Obj);
1433 }
1434 } else {
1435 m_LineObj.Add(Obj);
1436 }
1437 }
PreMarkedContent(PDFTEXT_Obj Obj)1438 FX_INT32 CPDF_TextPage::PreMarkedContent(PDFTEXT_Obj Obj)
1439 {
1440 CPDF_TextObject* pTextObj = Obj.m_pTextObj;
1441 CPDF_ContentMarkData* pMarkData = (CPDF_ContentMarkData*)pTextObj->m_ContentMark.GetObject();
1442 if(!pMarkData) {
1443 return FPDFTEXT_MC_PASS;
1444 }
1445 int nContentMark = pMarkData->CountItems();
1446 if (nContentMark < 1) {
1447 return FPDFTEXT_MC_PASS;
1448 }
1449 CFX_WideString actText;
1450 FX_BOOL bExist = FALSE;
1451 CPDF_Dictionary* pDict = NULL;
1452 int n = 0;
1453 for (n = 0; n < nContentMark; n++) {
1454 CPDF_ContentMarkItem& item = pMarkData->GetItem(n);
1455 CFX_ByteString tagStr = (CFX_ByteString)item.GetName();
1456 pDict = (CPDF_Dictionary*)item.GetParam();
1457 CPDF_String* temp = (CPDF_String*)(pDict ? pDict->GetElement(FX_BSTRC("ActualText")) : NULL);
1458 if (temp) {
1459 bExist = TRUE;
1460 actText = temp->GetUnicodeText();
1461 }
1462 }
1463 if (!bExist) {
1464 return FPDFTEXT_MC_PASS;
1465 }
1466 if (m_pPreTextObj) {
1467 if (CPDF_ContentMarkData* pPreMarkData = (CPDF_ContentMarkData*)m_pPreTextObj->m_ContentMark.GetObject()) {
1468 if (pPreMarkData->CountItems() == n) {
1469 CPDF_ContentMarkItem& item = pPreMarkData->GetItem(n - 1);
1470 if (pDict == item.GetParam()) {
1471 return FPDFTEXT_MC_DONE;
1472 }
1473 }
1474 }
1475 }
1476 CPDF_Font* pFont = pTextObj->GetFont();
1477 FX_STRSIZE nItems = actText.GetLength();
1478 if (nItems < 1) {
1479 return FPDFTEXT_MC_PASS;
1480 }
1481 bExist = FALSE;
1482 for (FX_STRSIZE i = 0; i < nItems; i++) {
1483 FX_WCHAR wChar = actText.GetAt(i);
1484 if (-1 == pFont->CharCodeFromUnicode(wChar)) {
1485 continue;
1486 } else {
1487 bExist = TRUE;
1488 break;
1489 }
1490 }
1491 if (!bExist) {
1492 return FPDFTEXT_MC_PASS;
1493 }
1494 bExist = FALSE;
1495 for (FX_STRSIZE i = 0; i < nItems; i++) {
1496 FX_WCHAR wChar = actText.GetAt(i);
1497 if ((wChar > 0x80 && wChar < 0xFFFD) || (wChar <= 0x80 && isprint(wChar))) {
1498 bExist = TRUE;
1499 break;
1500 }
1501 }
1502 if (!bExist) {
1503 return FPDFTEXT_MC_DONE;
1504 }
1505 return FPDFTEXT_MC_DELAY;
1506 }
ProcessMarkedContent(PDFTEXT_Obj Obj)1507 void CPDF_TextPage::ProcessMarkedContent(PDFTEXT_Obj Obj)
1508 {
1509 CPDF_TextObject* pTextObj = Obj.m_pTextObj;
1510 CPDF_ContentMarkData* pMarkData = (CPDF_ContentMarkData*)pTextObj->m_ContentMark.GetObject();
1511 if(!pMarkData) {
1512 return;
1513 }
1514 int nContentMark = pMarkData->CountItems();
1515 if (nContentMark < 1) {
1516 return;
1517 }
1518 CFX_WideString actText;
1519 CPDF_Dictionary* pDict = NULL;
1520 int n = 0;
1521 for (n = 0; n < nContentMark; n++) {
1522 CPDF_ContentMarkItem& item = pMarkData->GetItem(n);
1523 CFX_ByteString tagStr = (CFX_ByteString)item.GetName();
1524 pDict = (CPDF_Dictionary*)item.GetParam();
1525 CPDF_String* temp = (CPDF_String*)(pDict ? pDict->GetElement(FX_BSTRC("ActualText")) : NULL);
1526 if (temp) {
1527 actText = temp->GetUnicodeText();
1528 }
1529 }
1530 FX_STRSIZE nItems = actText.GetLength();
1531 if (nItems < 1) {
1532 return;
1533 }
1534 CPDF_Font* pFont = pTextObj->GetFont();
1535 CFX_AffineMatrix formMatrix = Obj.m_formMatrix;
1536 CFX_AffineMatrix matrix;
1537 pTextObj->GetTextMatrix(&matrix);
1538 matrix.Concat(formMatrix);
1539 FX_FLOAT fPosX = pTextObj->GetPosX();
1540 FX_FLOAT fPosY = pTextObj->GetPosY();
1541 int nCharInfoIndex = m_TextBuf.GetLength();
1542 CFX_FloatRect charBox;
1543 charBox.top = pTextObj->m_Top;
1544 charBox.left = pTextObj->m_Left;
1545 charBox.right = pTextObj->m_Right;
1546 charBox.bottom = pTextObj->m_Bottom;
1547 for (FX_STRSIZE k = 0; k < nItems; k++) {
1548 FX_WCHAR wChar = actText.GetAt(k);
1549 if (wChar <= 0x80 && !isprint(wChar)) {
1550 wChar = 0x20;
1551 }
1552 if (wChar >= 0xFFFD) {
1553 continue;
1554 }
1555 PAGECHAR_INFO charinfo;
1556 charinfo.m_OriginX = fPosX;
1557 charinfo.m_OriginY = fPosY;
1558 charinfo.m_Index = nCharInfoIndex;
1559 charinfo.m_Unicode = wChar;
1560 charinfo.m_CharCode = pFont->CharCodeFromUnicode(wChar);
1561 charinfo.m_Flag = FPDFTEXT_CHAR_PIECE;
1562 charinfo.m_pTextObj = pTextObj;
1563 charinfo.m_CharBox.top = charBox.top;
1564 charinfo.m_CharBox.left = charBox.left;
1565 charinfo.m_CharBox.right = charBox.right;
1566 charinfo.m_CharBox.bottom = charBox.bottom;
1567 charinfo.m_Matrix.Copy(matrix);
1568 m_TempTextBuf.AppendChar(wChar);
1569 m_TempCharList.Add(charinfo);
1570 }
1571 }
FindPreviousTextObject(void)1572 void CPDF_TextPage::FindPreviousTextObject(void)
1573 {
1574 if (m_TempCharList.GetSize() < 1 && m_charList.GetSize() < 1) {
1575 return;
1576 }
1577 PAGECHAR_INFO preChar;
1578 if (m_TempCharList.GetSize() >= 1) {
1579 preChar = *(PAGECHAR_INFO*)m_TempCharList.GetAt(m_TempCharList.GetSize() - 1);
1580 } else {
1581 preChar = *(PAGECHAR_INFO*)m_charList.GetAt(m_charList.GetSize() - 1);
1582 }
1583 if (preChar.m_pTextObj) {
1584 m_pPreTextObj = preChar.m_pTextObj;
1585 }
1586 }
ProcessTextObject(PDFTEXT_Obj Obj)1587 void CPDF_TextPage::ProcessTextObject(PDFTEXT_Obj Obj)
1588 {
1589 CPDF_TextObject* pTextObj = Obj.m_pTextObj;
1590 if(FXSYS_fabs(pTextObj->m_Right - pTextObj->m_Left) < 0.01f ) {
1591 return;
1592 }
1593 CFX_AffineMatrix formMatrix = Obj.m_formMatrix;
1594 CPDF_Font* pFont = pTextObj->GetFont();
1595 CFX_AffineMatrix matrix;
1596 pTextObj->GetTextMatrix(&matrix);
1597 matrix.Concat(formMatrix);
1598 FX_INT32 bPreMKC = PreMarkedContent(Obj);
1599 if (FPDFTEXT_MC_DONE == bPreMKC) {
1600 m_pPreTextObj = pTextObj;
1601 m_perMatrix.Copy(formMatrix);
1602 return;
1603 }
1604 int result = 0;
1605 if (m_pPreTextObj) {
1606 result = ProcessInsertObject(pTextObj, formMatrix);
1607 if (2 == result) {
1608 m_CurlineRect = CFX_FloatRect(Obj.m_pTextObj->m_Left, Obj.m_pTextObj->m_Bottom, Obj.m_pTextObj->m_Right, Obj.m_pTextObj->m_Top);
1609 } else {
1610 m_CurlineRect.Union(CFX_FloatRect(Obj.m_pTextObj->m_Left, Obj.m_pTextObj->m_Bottom, Obj.m_pTextObj->m_Right, Obj.m_pTextObj->m_Top));
1611 }
1612 PAGECHAR_INFO generateChar;
1613 if (result == 1) {
1614 if (GenerateCharInfo(TEXT_BLANK_CHAR, generateChar)) {
1615 if (!formMatrix.IsIdentity()) {
1616 generateChar.m_Matrix.Copy(formMatrix);
1617 }
1618 m_TempTextBuf.AppendChar(TEXT_BLANK_CHAR);
1619 m_TempCharList.Add(generateChar);
1620 }
1621 } else if(result == 2) {
1622 CloseTempLine();
1623 if(m_TextBuf.GetSize()) {
1624 if(m_ParseOptions.m_bGetCharCodeOnly) {
1625 m_TextBuf.AppendChar(TEXT_RETURN_CHAR);
1626 m_TextBuf.AppendChar(TEXT_LINEFEED_CHAR);
1627 } else {
1628 if(GenerateCharInfo(TEXT_RETURN_CHAR, generateChar)) {
1629 m_TextBuf.AppendChar(TEXT_RETURN_CHAR);
1630 if (!formMatrix.IsIdentity()) {
1631 generateChar.m_Matrix.Copy(formMatrix);
1632 }
1633 m_charList.Add(generateChar);
1634 }
1635 if(GenerateCharInfo(TEXT_LINEFEED_CHAR, generateChar)) {
1636 m_TextBuf.AppendChar(TEXT_LINEFEED_CHAR);
1637 if (!formMatrix.IsIdentity()) {
1638 generateChar.m_Matrix.Copy(formMatrix);
1639 }
1640 m_charList.Add(generateChar);
1641 }
1642 }
1643 }
1644 } else if (result == 3 && !m_ParseOptions.m_bOutputHyphen) {
1645 FX_INT32 nChars = pTextObj->CountChars();
1646 if (nChars == 1) {
1647 CPDF_TextObjectItem item;
1648 pTextObj->GetCharInfo(0, &item);
1649 CFX_WideString wstrItem = pTextObj->GetFont()->UnicodeFromCharCode(item.m_CharCode);
1650 if(wstrItem.IsEmpty()) {
1651 wstrItem += (FX_WCHAR)item.m_CharCode;
1652 }
1653 FX_WCHAR curChar = wstrItem.GetAt(0);
1654 if (0x2D == curChar || 0xAD == curChar) {
1655 return;
1656 }
1657 }
1658 while (m_TempTextBuf.GetSize() > 0 && m_TempTextBuf.GetWideString().GetAt(m_TempTextBuf.GetLength() - 1) == 0x20) {
1659 m_TempTextBuf.Delete(m_TempTextBuf.GetLength() - 1, 1);
1660 m_TempCharList.Delete(m_TempCharList.GetSize() - 1);
1661 }
1662 PAGECHAR_INFO* cha = (PAGECHAR_INFO*)m_TempCharList.GetAt(m_TempCharList.GetSize() - 1);
1663 m_TempTextBuf.Delete(m_TempTextBuf.GetLength() - 1, 1);
1664 cha->m_Unicode = 0x2;
1665 cha->m_Flag = FPDFTEXT_CHAR_HYPHEN;
1666 m_TempTextBuf.AppendChar(0xfffe);
1667 }
1668 } else {
1669 m_CurlineRect = CFX_FloatRect(Obj.m_pTextObj->m_Left, Obj.m_pTextObj->m_Bottom, Obj.m_pTextObj->m_Right, Obj.m_pTextObj->m_Top);
1670 }
1671 if (FPDFTEXT_MC_DELAY == bPreMKC) {
1672 ProcessMarkedContent(Obj);
1673 m_pPreTextObj = pTextObj;
1674 m_perMatrix.Copy(formMatrix);
1675 return;
1676 }
1677 m_pPreTextObj = pTextObj;
1678 m_perMatrix.Copy(formMatrix);
1679 int nItems = pTextObj->CountItems();
1680 FX_FLOAT baseSpace = _CalculateBaseSpace(pTextObj, matrix);
1681
1682 const FX_BOOL bR2L = IsRightToLeft(pTextObj, pFont, nItems);
1683 const FX_BOOL bIsBidiAndMirrorInverse =
1684 bR2L && (matrix.a * matrix.d - matrix.b * matrix.c) < 0;
1685 FX_INT32 iBufStartAppend = m_TempTextBuf.GetLength();
1686 FX_INT32 iCharListStartAppend = m_TempCharList.GetSize();
1687
1688 FX_FLOAT spacing = 0;
1689 for (int i = 0; i < nItems; i++) {
1690 CPDF_TextObjectItem item;
1691 PAGECHAR_INFO charinfo;
1692 charinfo.m_OriginX = 0;
1693 charinfo.m_OriginY = 0;
1694 pTextObj->GetItemInfo(i, &item);
1695 if (item.m_CharCode == (FX_DWORD) - 1) {
1696 CFX_WideString str = m_TempTextBuf.GetWideString();
1697 if(str.IsEmpty()) {
1698 str = m_TextBuf.GetWideString();
1699 }
1700 if (str.IsEmpty() || str.GetAt(str.GetLength() - 1) == TEXT_BLANK_CHAR) {
1701 continue;
1702 }
1703 FX_FLOAT fontsize_h = pTextObj->m_TextState.GetFontSizeH();
1704 spacing = -fontsize_h * item.m_OriginX / 1000;
1705 continue;
1706 }
1707 FX_FLOAT charSpace = pTextObj->m_TextState.GetObject()->m_CharSpace;
1708 if (charSpace > 0.001) {
1709 spacing += matrix.TransformDistance(charSpace);
1710 } else if(charSpace < -0.001) {
1711 spacing -= matrix.TransformDistance(FXSYS_fabs(charSpace));
1712 }
1713 spacing -= baseSpace;
1714 if (spacing && i > 0) {
1715 int last_width = 0;
1716 FX_FLOAT fontsize_h = pTextObj->m_TextState.GetFontSizeH();
1717 FX_DWORD space_charcode = pFont->CharCodeFromUnicode(' ');
1718 FX_FLOAT threshold = 0;
1719 if (space_charcode != -1) {
1720 threshold = fontsize_h * pFont->GetCharWidthF(space_charcode) / 1000 ;
1721 }
1722 if (threshold > fontsize_h / 3) {
1723 threshold = 0;
1724 } else {
1725 threshold /= 2;
1726 }
1727 if (threshold == 0) {
1728 threshold = fontsize_h;
1729 int this_width = FXSYS_abs(GetCharWidth(item.m_CharCode, pFont));
1730 threshold = this_width > last_width ? (FX_FLOAT)this_width : (FX_FLOAT)last_width;
1731 threshold = _NormalizeThreshold(threshold);
1732 threshold = fontsize_h * threshold / 1000;
1733 }
1734 if (threshold && (spacing && spacing >= threshold) ) {
1735 charinfo.m_Unicode = TEXT_BLANK_CHAR;
1736 charinfo.m_Flag = FPDFTEXT_CHAR_GENERATED;
1737 charinfo.m_pTextObj = pTextObj;
1738 charinfo.m_Index = m_TextBuf.GetLength();
1739 m_TempTextBuf.AppendChar(TEXT_BLANK_CHAR);
1740 charinfo.m_CharCode = -1;
1741 charinfo.m_Matrix.Copy(formMatrix);
1742 matrix.Transform(item.m_OriginX, item.m_OriginY, charinfo.m_OriginX, charinfo.m_OriginY);
1743 charinfo.m_CharBox = CFX_FloatRect(charinfo.m_OriginX, charinfo.m_OriginY, charinfo.m_OriginX, charinfo.m_OriginY);
1744 m_TempCharList.Add(charinfo);
1745 }
1746 if (item.m_CharCode == (FX_DWORD) - 1) {
1747 continue;
1748 }
1749 }
1750 spacing = 0;
1751 CFX_WideString wstrItem = pFont->UnicodeFromCharCode(item.m_CharCode);
1752 FX_BOOL bNoUnicode = FALSE;
1753 FX_WCHAR wChar = wstrItem.GetAt(0);
1754 if ((wstrItem.IsEmpty() || wChar == 0) && item.m_CharCode) {
1755 if(wstrItem.IsEmpty()) {
1756 wstrItem += (FX_WCHAR)item.m_CharCode;
1757 } else {
1758 wstrItem.SetAt(0, (FX_WCHAR)item.m_CharCode);
1759 }
1760 bNoUnicode = TRUE;
1761 }
1762 charinfo.m_Index = -1;
1763 charinfo.m_CharCode = item.m_CharCode;
1764 if(bNoUnicode) {
1765 charinfo.m_Flag = FPDFTEXT_CHAR_UNUNICODE;
1766 } else {
1767 charinfo.m_Flag = FPDFTEXT_CHAR_NORMAL;
1768 }
1769 charinfo.m_pTextObj = pTextObj;
1770 charinfo.m_OriginX = 0, charinfo.m_OriginY = 0;
1771 matrix.Transform(item.m_OriginX, item.m_OriginY, charinfo.m_OriginX, charinfo.m_OriginY);
1772 FX_RECT rect(0, 0, 0, 0);
1773 rect.Intersect(0, 0, 0, 0);
1774 charinfo.m_pTextObj->GetFont()->GetCharBBox(charinfo.m_CharCode, rect);
1775 charinfo.m_CharBox.top = rect.top * pTextObj->GetFontSize() / 1000 + item.m_OriginY;
1776 charinfo.m_CharBox.left = rect.left * pTextObj->GetFontSize() / 1000 + item.m_OriginX;
1777 charinfo.m_CharBox.right = rect.right * pTextObj->GetFontSize() / 1000 + item.m_OriginX;
1778 charinfo.m_CharBox.bottom = rect.bottom * pTextObj->GetFontSize() / 1000 + item.m_OriginY;
1779 if (fabsf(charinfo.m_CharBox.top - charinfo.m_CharBox.bottom) < 0.01f) {
1780 charinfo.m_CharBox.top = charinfo.m_CharBox.bottom + pTextObj->GetFontSize();
1781 }
1782 if (fabsf(charinfo.m_CharBox.right - charinfo.m_CharBox.left) < 0.01f) {
1783 charinfo.m_CharBox.right = charinfo.m_CharBox.left + pTextObj->GetCharWidth(charinfo.m_CharCode);
1784 }
1785 matrix.TransformRect(charinfo.m_CharBox);
1786 charinfo.m_Matrix.Copy(matrix);
1787 if (wstrItem.IsEmpty()) {
1788 charinfo.m_Unicode = 0;
1789 m_TempCharList.Add(charinfo);
1790 m_TempTextBuf.AppendChar(0xfffe);
1791 continue;
1792 } else {
1793 int nTotal = wstrItem.GetLength();
1794 FX_BOOL bDel = FALSE;
1795 const int count = std::min(m_TempCharList.GetSize(), 7);
1796 FX_FLOAT threshold = charinfo.m_Matrix.TransformXDistance((FX_FLOAT)TEXT_CHARRATIO_GAPDELTA * pTextObj->GetFontSize());
1797 for (int n = m_TempCharList.GetSize();
1798 n > m_TempCharList.GetSize() - count;
1799 n--) {
1800 PAGECHAR_INFO* charinfo1 = (PAGECHAR_INFO*)m_TempCharList.GetAt(n - 1);
1801 if(charinfo1->m_CharCode == charinfo.m_CharCode &&
1802 charinfo1->m_pTextObj->GetFont() == charinfo.m_pTextObj->GetFont() &&
1803 FXSYS_fabs(charinfo1->m_OriginX - charinfo.m_OriginX) < threshold &&
1804 FXSYS_fabs(charinfo1->m_OriginY - charinfo.m_OriginY) < threshold) {
1805 bDel = TRUE;
1806 break;
1807 }
1808 }
1809 if(!bDel) {
1810 for (int nIndex = 0; nIndex < nTotal; nIndex++) {
1811 charinfo.m_Unicode = wstrItem.GetAt(nIndex);
1812 if (charinfo.m_Unicode) {
1813 charinfo.m_Index = m_TextBuf.GetLength();
1814 m_TempTextBuf.AppendChar(charinfo.m_Unicode);
1815 } else {
1816 m_TempTextBuf.AppendChar(0xfffe);
1817 }
1818 m_TempCharList.Add(charinfo);
1819 }
1820 } else if(i == 0) {
1821 CFX_WideString str = m_TempTextBuf.GetWideString();
1822 if (!str.IsEmpty() && str.GetAt(str.GetLength() - 1) == TEXT_BLANK_CHAR) {
1823 m_TempTextBuf.Delete(m_TempTextBuf.GetLength() - 1, 1);
1824 m_TempCharList.Delete(m_TempCharList.GetSize() - 1);
1825 }
1826 }
1827 }
1828 }
1829 if (bIsBidiAndMirrorInverse) {
1830 SwapTempTextBuf(iCharListStartAppend, iBufStartAppend);
1831 }
1832 }
SwapTempTextBuf(FX_INT32 iCharListStartAppend,FX_INT32 iBufStartAppend)1833 void CPDF_TextPage::SwapTempTextBuf(FX_INT32 iCharListStartAppend,
1834 FX_INT32 iBufStartAppend)
1835 {
1836 FX_INT32 i, j;
1837 i = iCharListStartAppend;
1838 j = m_TempCharList.GetSize() - 1;
1839 for (; i < j; i++, j--) {
1840 std::swap(m_TempCharList[i], m_TempCharList[j]);
1841 std::swap(m_TempCharList[i].m_Index, m_TempCharList[j].m_Index);
1842 }
1843 FX_WCHAR * pTempBuffer = m_TempTextBuf.GetBuffer();
1844 i = iBufStartAppend;
1845 j = m_TempTextBuf.GetLength() - 1;
1846 for (; i < j; i++, j--) {
1847 std::swap(pTempBuffer[i], pTempBuffer[j]);
1848 }
1849 }
IsRightToLeft(const CPDF_TextObject * pTextObj,const CPDF_Font * pFont,int nItems) const1850 FX_BOOL CPDF_TextPage::IsRightToLeft(const CPDF_TextObject* pTextObj,
1851 const CPDF_Font* pFont,
1852 int nItems) const
1853 {
1854 IFX_BidiChar* BidiChar = IFX_BidiChar::Create();
1855 FX_INT32 nR2L = 0;
1856 FX_INT32 nL2R = 0;
1857 FX_INT32 start = 0, count = 0;
1858 CPDF_TextObjectItem item;
1859 for (FX_INT32 i = 0; i < nItems; i++) {
1860 pTextObj->GetItemInfo(i, &item);
1861 if (item.m_CharCode == (FX_DWORD)-1) {
1862 continue;
1863 }
1864 CFX_WideString wstrItem = pFont->UnicodeFromCharCode(item.m_CharCode);
1865 FX_WCHAR wChar = wstrItem.GetAt(0);
1866 if ((wstrItem.IsEmpty() || wChar == 0) && item.m_CharCode) {
1867 wChar = (FX_WCHAR)item.m_CharCode;
1868 }
1869 if (!wChar) {
1870 continue;
1871 }
1872 if (BidiChar && BidiChar->AppendChar(wChar)) {
1873 FX_INT32 ret = BidiChar->GetBidiInfo(start, count);
1874 if (ret == 2) {
1875 nR2L++;
1876 }
1877 else if (ret == 1) {
1878 nL2R++;
1879 }
1880 }
1881 }
1882 if (BidiChar && BidiChar->EndChar()) {
1883 FX_INT32 ret = BidiChar->GetBidiInfo(start, count);
1884 if (ret == 2) {
1885 nR2L++;
1886 }
1887 else if (ret == 1) {
1888 nL2R++;
1889 }
1890 }
1891 if (BidiChar)
1892 BidiChar->Release();
1893 return (nR2L > 0 && nR2L >= nL2R);
1894 }
GetTextObjectWritingMode(const CPDF_TextObject * pTextObj)1895 FX_INT32 CPDF_TextPage::GetTextObjectWritingMode(const CPDF_TextObject* pTextObj)
1896 {
1897 FX_INT32 nChars = pTextObj->CountChars();
1898 if (nChars == 1) {
1899 return m_TextlineDir;
1900 }
1901 CPDF_TextObjectItem first, last;
1902 pTextObj->GetCharInfo(0, &first);
1903 pTextObj->GetCharInfo(nChars - 1, &last);
1904 CFX_Matrix textMatrix;
1905 pTextObj->GetTextMatrix(&textMatrix);
1906 textMatrix.TransformPoint(first.m_OriginX, first.m_OriginY);
1907 textMatrix.TransformPoint(last.m_OriginX, last.m_OriginY);
1908 FX_FLOAT dX = FXSYS_fabs(last.m_OriginX - first.m_OriginX);
1909 FX_FLOAT dY = FXSYS_fabs(last.m_OriginY - first.m_OriginY);
1910 if (dX <= 0.0001f && dY <= 0.0001f) {
1911 return -1;
1912 }
1913 CFX_VectorF v;
1914 v.Set(dX, dY);
1915 v.Normalize();
1916 if (v.y <= 0.0872f) {
1917 if (v.x <= 0.0872f) {
1918 return m_TextlineDir;
1919 }
1920 return 0;
1921 } else if (v.x <= 0.0872f) {
1922 return 1;
1923 }
1924 return m_TextlineDir;
1925 }
IsHyphen(FX_WCHAR curChar)1926 FX_BOOL CPDF_TextPage::IsHyphen(FX_WCHAR curChar)
1927 {
1928 CFX_WideString strCurText = m_TempTextBuf.GetWideString();
1929 if(strCurText.GetLength() == 0) {
1930 strCurText = m_TextBuf.GetWideString();
1931 }
1932 FX_STRSIZE nCount = strCurText.GetLength();
1933 int nIndex = nCount - 1;
1934 FX_WCHAR wcTmp = strCurText.GetAt(nIndex);
1935 while(wcTmp == 0x20 && nIndex <= nCount - 1 && nIndex >= 0) {
1936 wcTmp = strCurText.GetAt(--nIndex);
1937 }
1938 if (0x2D == wcTmp || 0xAD == wcTmp) {
1939 if (--nIndex > 0) {
1940 FX_WCHAR preChar = strCurText.GetAt((nIndex));
1941 if (((preChar >= L'A' && preChar <= L'Z') || (preChar >= L'a' && preChar <= L'z'))
1942 && ((curChar >= L'A' && curChar <= L'Z') || (curChar >= L'a' && curChar <= L'z'))) {
1943 return TRUE;
1944 }
1945 }
1946 int size = m_TempCharList.GetSize();
1947 PAGECHAR_INFO preChar;
1948 if (size) {
1949 preChar = (PAGECHAR_INFO)m_TempCharList[size - 1];
1950 } else {
1951 size = m_charList.GetSize();
1952 if(size == 0) {
1953 return FALSE;
1954 }
1955 preChar = (PAGECHAR_INFO)m_charList[size - 1];
1956 }
1957 if (FPDFTEXT_CHAR_PIECE == preChar.m_Flag)
1958 if (0xAD == preChar.m_Unicode || 0x2D == preChar.m_Unicode) {
1959 return TRUE;
1960 }
1961 }
1962 return FALSE;
1963 }
ProcessInsertObject(const CPDF_TextObject * pObj,const CFX_AffineMatrix & formMatrix)1964 int CPDF_TextPage::ProcessInsertObject(const CPDF_TextObject* pObj, const CFX_AffineMatrix& formMatrix)
1965 {
1966 FindPreviousTextObject();
1967 FX_BOOL bNewline = FALSE;
1968 int WritingMode = GetTextObjectWritingMode(pObj);
1969 if(WritingMode == -1) {
1970 WritingMode = GetTextObjectWritingMode(m_pPreTextObj);
1971 }
1972 CFX_FloatRect this_rect(pObj->m_Left, pObj->m_Bottom, pObj->m_Right, pObj->m_Top);
1973 CFX_FloatRect prev_rect(m_pPreTextObj->m_Left, m_pPreTextObj->m_Bottom, m_pPreTextObj->m_Right, m_pPreTextObj->m_Top);
1974 CPDF_TextObjectItem PrevItem, item;
1975 int nItem = m_pPreTextObj->CountItems();
1976 m_pPreTextObj->GetItemInfo(nItem - 1, &PrevItem);
1977 pObj->GetItemInfo(0, &item);
1978 CFX_WideString wstrItem = pObj->GetFont()->UnicodeFromCharCode(item.m_CharCode);
1979 if(wstrItem.IsEmpty()) {
1980 wstrItem += (FX_WCHAR)item.m_CharCode;
1981 }
1982 FX_WCHAR curChar = wstrItem.GetAt(0);
1983 if(WritingMode == 0) {
1984 if(this_rect.Height() > 4.5 && prev_rect.Height() > 4.5) {
1985 FX_FLOAT top = this_rect.top < prev_rect.top ? this_rect.top : prev_rect.top;
1986 FX_FLOAT bottom = this_rect.bottom > prev_rect.bottom ? this_rect.bottom : prev_rect.bottom;
1987 if(bottom >= top) {
1988 if(IsHyphen(curChar)) {
1989 return 3;
1990 }
1991 return 2;
1992 }
1993 }
1994 } else if (WritingMode == 1) {
1995 if(this_rect.Width() > pObj->GetFontSize() * 0.1f && prev_rect.Width() > m_pPreTextObj->GetFontSize() * 0.1f) {
1996 FX_FLOAT left = this_rect.left > m_CurlineRect.left ? this_rect.left : m_CurlineRect.left;
1997 FX_FLOAT right = this_rect.right < m_CurlineRect.right ? this_rect.right : m_CurlineRect.right;
1998 if(right <= left) {
1999 if(IsHyphen(curChar)) {
2000 return 3;
2001 }
2002 return 2;
2003 }
2004 }
2005 }
2006 FX_FLOAT last_pos = PrevItem.m_OriginX;
2007 int nLastWidth = GetCharWidth(PrevItem.m_CharCode, m_pPreTextObj->GetFont());
2008 FX_FLOAT last_width = nLastWidth * m_pPreTextObj->GetFontSize() / 1000;
2009 last_width = FXSYS_fabs(last_width);
2010 int nThisWidth = GetCharWidth(item.m_CharCode, pObj->GetFont());
2011 FX_FLOAT this_width = nThisWidth * pObj->GetFontSize() / 1000;
2012 this_width = FXSYS_fabs(this_width);
2013 FX_FLOAT threshold = last_width > this_width ? last_width / 4 : this_width / 4;
2014 CFX_AffineMatrix prev_matrix, prev_reverse;
2015 m_pPreTextObj->GetTextMatrix(&prev_matrix);
2016 prev_matrix.Concat(m_perMatrix);
2017 prev_reverse.SetReverse(prev_matrix);
2018 FX_FLOAT x = pObj->GetPosX();
2019 FX_FLOAT y = pObj->GetPosY();
2020 formMatrix.Transform(x, y);
2021 prev_reverse.Transform(x, y);
2022 if(last_width < this_width) {
2023 threshold = prev_reverse.TransformDistance(threshold);
2024 }
2025 CFX_FloatRect rect1(m_pPreTextObj->m_Left, pObj->m_Bottom, m_pPreTextObj->m_Right, pObj->m_Top);
2026 CFX_FloatRect rect2(m_pPreTextObj->m_Left, m_pPreTextObj->m_Bottom, m_pPreTextObj->m_Right, m_pPreTextObj->m_Top);
2027 CFX_FloatRect rect3 = rect1;
2028 rect1.Intersect(rect2);
2029 if (WritingMode == 0) {
2030 if ((rect1.IsEmpty() && rect2.Height() > 5 && rect3.Height() > 5)
2031 || ((y > threshold * 2 || y < threshold * -3) && (FXSYS_fabs(y) < 1 ? FXSYS_fabs(x) < FXSYS_fabs(y) : TRUE))) {
2032 bNewline = TRUE;
2033 if(nItem > 1 ) {
2034 CPDF_TextObjectItem tempItem;
2035 m_pPreTextObj->GetItemInfo(0, &tempItem);
2036 CFX_AffineMatrix m;
2037 m_pPreTextObj->GetTextMatrix(&m);
2038 if(PrevItem.m_OriginX > tempItem.m_OriginX &&
2039 m_DisplayMatrix.a > 0.9 && m_DisplayMatrix.b < 0.1 &&
2040 m_DisplayMatrix.c < 0.1 && m_DisplayMatrix.d < -0.9
2041 && m.b < 0.1 && m.c < 0.1 ) {
2042 CFX_FloatRect re(0, m_pPreTextObj->m_Bottom, 1000, m_pPreTextObj->m_Top);
2043 if(re.Contains(pObj->GetPosX(), pObj->GetPosY())) {
2044 bNewline = FALSE;
2045 } else {
2046 CFX_FloatRect re(0, pObj->m_Bottom, 1000, pObj->m_Top);
2047 if(re.Contains(m_pPreTextObj->GetPosX(), m_pPreTextObj->GetPosY())) {
2048 bNewline = FALSE;
2049 }
2050 }
2051 }
2052 }
2053 }
2054 }
2055 if(bNewline) {
2056 if(IsHyphen(curChar)) {
2057 return 3;
2058 }
2059 return 2;
2060 }
2061 FX_INT32 nChars = pObj->CountChars();
2062 if (nChars == 1 && ( 0x2D == curChar || 0xAD == curChar))
2063 if (IsHyphen(curChar)) {
2064 return 3;
2065 }
2066 CFX_WideString PrevStr = m_pPreTextObj->GetFont()->UnicodeFromCharCode(PrevItem.m_CharCode);
2067 FX_WCHAR preChar = PrevStr.GetAt(PrevStr.GetLength() - 1);
2068 CFX_AffineMatrix matrix;
2069 pObj->GetTextMatrix(&matrix);
2070 matrix.Concat(formMatrix);
2071 threshold = (FX_FLOAT)(nLastWidth > nThisWidth ? nLastWidth : nThisWidth);
2072 threshold = threshold > 400 ? (threshold < 700 ? threshold / 4 : (threshold > 800 ? threshold / 6 : threshold / 5)) : (threshold / 2);
2073 if(nLastWidth >= nThisWidth) {
2074 threshold *= FXSYS_fabs(m_pPreTextObj->GetFontSize());
2075 } else {
2076 threshold *= FXSYS_fabs(pObj->GetFontSize());
2077 threshold = matrix.TransformDistance(threshold);
2078 threshold = prev_reverse.TransformDistance(threshold);
2079 }
2080 threshold /= 1000;
2081 if((threshold < 1.4881 && threshold > 1.4879)
2082 || (threshold < 1.39001 && threshold > 1.38999)) {
2083 threshold *= 1.5;
2084 }
2085 if (FXSYS_fabs(last_pos + last_width - x) > threshold && curChar != L' ' && preChar != L' ')
2086 if (curChar != L' ' && preChar != L' ') {
2087 if((x - last_pos - last_width) > threshold || (last_pos - x - last_width) > threshold) {
2088 return 1;
2089 }
2090 if(x < 0 && (last_pos - x - last_width) > threshold) {
2091 return 1;
2092 }
2093 if((x - last_pos - last_width) > this_width || (x - last_pos - this_width) > last_width ) {
2094 return 1;
2095 }
2096 }
2097 return 0;
2098 }
IsSameTextObject(CPDF_TextObject * pTextObj1,CPDF_TextObject * pTextObj2)2099 FX_BOOL CPDF_TextPage::IsSameTextObject(CPDF_TextObject* pTextObj1, CPDF_TextObject* pTextObj2)
2100 {
2101 if (!pTextObj1 || !pTextObj2) {
2102 return FALSE;
2103 }
2104 CFX_FloatRect rcPreObj(pTextObj2->m_Left, pTextObj2->m_Bottom, pTextObj2->m_Right, pTextObj2->m_Top);
2105 CFX_FloatRect rcCurObj(pTextObj1->m_Left, pTextObj1->m_Bottom, pTextObj1->m_Right, pTextObj1->m_Top);
2106 if (rcPreObj.IsEmpty() && rcCurObj.IsEmpty() && !m_ParseOptions.m_bGetCharCodeOnly) {
2107 FX_FLOAT dbXdif = FXSYS_fabs(rcPreObj.left - rcCurObj.left);
2108 int nCount = m_charList.GetSize();
2109 if (nCount >= 2) {
2110 PAGECHAR_INFO perCharTemp = (PAGECHAR_INFO)m_charList[nCount - 2];
2111 FX_FLOAT dbSpace = perCharTemp.m_CharBox.Width();
2112 if (dbXdif > dbSpace) {
2113 return FALSE;
2114 }
2115 }
2116 }
2117 if (!rcPreObj.IsEmpty() || !rcCurObj.IsEmpty()) {
2118 rcPreObj.Intersect(rcCurObj);
2119 if (rcPreObj.IsEmpty()) {
2120 return FALSE;
2121 }
2122 if (FXSYS_fabs(rcPreObj.Width() - rcCurObj.Width()) > rcCurObj.Width() / 2) {
2123 return FALSE;
2124 }
2125 if (pTextObj2->GetFontSize() != pTextObj1->GetFontSize()) {
2126 return FALSE;
2127 }
2128 }
2129 int nPreCount = pTextObj2->CountItems();
2130 int nCurCount = pTextObj1->CountItems();
2131 if (nPreCount != nCurCount) {
2132 return FALSE;
2133 }
2134 CPDF_TextObjectItem itemPer, itemCur;
2135 for (int i = 0; i < nPreCount; i++) {
2136 pTextObj2->GetItemInfo(i, &itemPer);
2137 pTextObj1->GetItemInfo(i, &itemCur);
2138 if (itemCur.m_CharCode != itemPer.m_CharCode) {
2139 return FALSE;
2140 }
2141 }
2142 if(FXSYS_fabs(pTextObj1->GetPosX() - pTextObj2->GetPosX()) > GetCharWidth(itemPer.m_CharCode, pTextObj2->GetFont())*pTextObj2->GetFontSize() / 1000 * 0.9 ||
2143 FXSYS_fabs(pTextObj1->GetPosY() - pTextObj2->GetPosY()) >
2144 FX_MAX(FX_MAX(rcPreObj.Height() , rcPreObj.Width()), pTextObj2->GetFontSize()) / 8) {
2145 return FALSE;
2146 }
2147 return TRUE;
2148 }
IsSameAsPreTextObject(CPDF_TextObject * pTextObj,FX_POSITION ObjPos)2149 FX_BOOL CPDF_TextPage::IsSameAsPreTextObject(CPDF_TextObject* pTextObj, FX_POSITION ObjPos)
2150 {
2151 if (!pTextObj) {
2152 return FALSE;
2153 }
2154 int i = 0;
2155 if (!ObjPos) {
2156 ObjPos = m_pPage->GetLastObjectPosition();
2157 }
2158 CPDF_PageObject* pObj = m_pPage->GetPrevObject(ObjPos);
2159 while (i < 5 && ObjPos) {
2160 pObj = m_pPage->GetPrevObject(ObjPos);
2161 if(pObj == pTextObj) {
2162 continue;
2163 }
2164 if(pObj->m_Type != PDFPAGE_TEXT) {
2165 continue;
2166 }
2167 if(IsSameTextObject((CPDF_TextObject*)pObj, pTextObj)) {
2168 return TRUE;
2169 }
2170 i++;
2171 }
2172 return FALSE;
2173 }
GenerateCharInfo(FX_WCHAR unicode,PAGECHAR_INFO & info)2174 FX_BOOL CPDF_TextPage::GenerateCharInfo(FX_WCHAR unicode, PAGECHAR_INFO& info)
2175 {
2176 int size = m_TempCharList.GetSize();
2177 PAGECHAR_INFO preChar;
2178 if (size) {
2179 preChar = (PAGECHAR_INFO)m_TempCharList[size - 1];
2180 } else {
2181 size = m_charList.GetSize();
2182 if(size == 0) {
2183 return FALSE;
2184 }
2185 preChar = (PAGECHAR_INFO)m_charList[size - 1];
2186 }
2187 info.m_Index = m_TextBuf.GetLength();
2188 info.m_Unicode = unicode;
2189 info.m_pTextObj = NULL;
2190 info.m_CharCode = -1;
2191 info.m_Flag = FPDFTEXT_CHAR_GENERATED;
2192 int preWidth = 0;
2193 if (preChar.m_pTextObj && preChar.m_CharCode != (FX_DWORD) - 1) {
2194 preWidth = GetCharWidth(preChar.m_CharCode, preChar.m_pTextObj->GetFont());
2195 }
2196 FX_FLOAT fs = 0;
2197 if(preChar.m_pTextObj) {
2198 fs = preChar.m_pTextObj->GetFontSize();
2199 } else {
2200 fs = preChar.m_CharBox.Height();
2201 }
2202 if(!fs) {
2203 fs = 1;
2204 }
2205 info.m_OriginX = preChar.m_OriginX + preWidth * (fs) / 1000;
2206 info.m_OriginY = preChar.m_OriginY;
2207 info.m_CharBox = CFX_FloatRect(info.m_OriginX, info.m_OriginY, info.m_OriginX, info.m_OriginY);
2208 return TRUE;
2209 }
IsRectIntersect(const CFX_FloatRect & rect1,const CFX_FloatRect & rect2)2210 FX_BOOL CPDF_TextPage::IsRectIntersect(const CFX_FloatRect& rect1, const CFX_FloatRect& rect2)
2211 {
2212 CFX_FloatRect rect = rect1;
2213 rect.Intersect(rect2);
2214 return !rect.IsEmpty();
2215 }
IsLetter(FX_WCHAR unicode)2216 FX_BOOL CPDF_TextPage::IsLetter(FX_WCHAR unicode)
2217 {
2218 if (unicode < L'A') {
2219 return FALSE;
2220 }
2221 if (unicode > L'Z' && unicode < L'a') {
2222 return FALSE;
2223 }
2224 if (unicode > L'z') {
2225 return FALSE;
2226 }
2227 return TRUE;
2228 }
CPDF_TextPageFind(const IPDF_TextPage * pTextPage)2229 CPDF_TextPageFind::CPDF_TextPageFind(const IPDF_TextPage* pTextPage)
2230 : m_pTextPage(pTextPage),
2231 m_flags(0),
2232 m_findNextStart(-1),
2233 m_findPreStart(-1),
2234 m_bMatchCase(FALSE),
2235 m_bMatchWholeWord(FALSE),
2236 m_resStart(0),
2237 m_resEnd(-1),
2238 m_IsFind(FALSE)
2239 {
2240 m_strText = m_pTextPage->GetPageText();
2241 int nCount = pTextPage->CountChars();
2242 if(nCount) {
2243 m_CharIndex.Add(0);
2244 }
2245 for(int i = 0; i < nCount; i++) {
2246 FPDF_CHAR_INFO info;
2247 pTextPage->GetCharInfo(i, info);
2248 int indexSize = m_CharIndex.GetSize();
2249 if(info.m_Flag == CHAR_NORMAL || info.m_Flag == CHAR_GENERATED) {
2250 if(indexSize % 2) {
2251 m_CharIndex.Add(1);
2252 } else {
2253 if(indexSize <= 0) {
2254 continue;
2255 }
2256 m_CharIndex.SetAt(indexSize - 1, m_CharIndex.GetAt(indexSize - 1) + 1);
2257 }
2258 } else {
2259 if(indexSize % 2) {
2260 if(indexSize <= 0) {
2261 continue;
2262 }
2263 m_CharIndex.SetAt(indexSize - 1, i + 1);
2264 } else {
2265 m_CharIndex.Add(i + 1);
2266 }
2267 }
2268 }
2269 int indexSize = m_CharIndex.GetSize();
2270 if(indexSize % 2) {
2271 m_CharIndex.RemoveAt(indexSize - 1);
2272 }
2273 }
GetCharIndex(int index) const2274 int CPDF_TextPageFind::GetCharIndex(int index) const
2275 {
2276 return m_pTextPage->CharIndexFromTextIndex(index);
2277 int indexSize = m_CharIndex.GetSize();
2278 int count = 0;
2279 for(int i = 0; i < indexSize; i += 2) {
2280 count += m_CharIndex.GetAt(i + 1);
2281 if(count > index) {
2282 return index - count + m_CharIndex.GetAt(i + 1) + m_CharIndex.GetAt(i);
2283 }
2284 }
2285 return -1;
2286 }
FindFirst(const CFX_WideString & findwhat,int flags,int startPos)2287 FX_BOOL CPDF_TextPageFind::FindFirst(const CFX_WideString& findwhat, int flags, int startPos)
2288 {
2289 if (!m_pTextPage) {
2290 return FALSE;
2291 }
2292 if (m_strText.IsEmpty() || m_bMatchCase != (flags & FPDFTEXT_MATCHCASE)) {
2293 m_strText = m_pTextPage->GetPageText();
2294 }
2295 CFX_WideString findwhatStr = findwhat;
2296 m_findWhat = findwhatStr;
2297 m_flags = flags;
2298 m_bMatchCase = flags & FPDFTEXT_MATCHCASE;
2299 if (m_strText.IsEmpty()) {
2300 m_IsFind = FALSE;
2301 return TRUE;
2302 }
2303 FX_STRSIZE len = findwhatStr.GetLength();
2304 if (!m_bMatchCase) {
2305 findwhatStr.MakeLower();
2306 m_strText.MakeLower();
2307 }
2308 m_bMatchWholeWord = flags & FPDFTEXT_MATCHWHOLEWORD;
2309 m_findNextStart = startPos;
2310 if (startPos == -1) {
2311 m_findPreStart = m_strText.GetLength() - 1;
2312 } else {
2313 m_findPreStart = startPos;
2314 }
2315 m_csFindWhatArray.RemoveAll();
2316 int i = 0;
2317 while(i < len) {
2318 if(findwhatStr.GetAt(i) != ' ') {
2319 break;
2320 }
2321 i++;
2322 }
2323 if(i < len) {
2324 ExtractFindWhat(findwhatStr);
2325 } else {
2326 m_csFindWhatArray.Add(findwhatStr);
2327 }
2328 if(m_csFindWhatArray.GetSize() <= 0) {
2329 return FALSE;
2330 }
2331 m_IsFind = TRUE;
2332 m_resStart = 0;
2333 m_resEnd = -1;
2334 return TRUE;
2335 }
FindNext()2336 FX_BOOL CPDF_TextPageFind::FindNext()
2337 {
2338 if (!m_pTextPage) {
2339 return FALSE;
2340 }
2341 m_resArray.RemoveAll();
2342 if(m_findNextStart == -1) {
2343 return FALSE;
2344 }
2345 if(m_strText.IsEmpty()) {
2346 m_IsFind = FALSE;
2347 return m_IsFind;
2348 }
2349 int strLen = m_strText.GetLength();
2350 if (m_findNextStart > strLen - 1) {
2351 m_IsFind = FALSE;
2352 return m_IsFind;
2353 }
2354 int nCount = m_csFindWhatArray.GetSize();
2355 int nResultPos = 0;
2356 int nStartPos = 0;
2357 nStartPos = m_findNextStart;
2358 FX_BOOL bSpaceStart = FALSE;
2359 for(int iWord = 0; iWord < nCount; iWord++) {
2360 CFX_WideString csWord = m_csFindWhatArray[iWord];
2361 if(csWord.IsEmpty()) {
2362 if(iWord == nCount - 1) {
2363 FX_WCHAR strInsert = m_strText.GetAt(nStartPos);
2364 if(strInsert == TEXT_LINEFEED_CHAR || strInsert == TEXT_BLANK_CHAR || strInsert == TEXT_RETURN_CHAR || strInsert == 160) {
2365 nResultPos = nStartPos + 1;
2366 break;
2367 }
2368 iWord = -1;
2369 } else if(iWord == 0) {
2370 bSpaceStart = TRUE;
2371 }
2372 continue;
2373 }
2374 int endIndex;
2375 nResultPos = m_strText.Find(csWord.c_str(), nStartPos);
2376 if (nResultPos == -1) {
2377 m_IsFind = FALSE;
2378 return m_IsFind;
2379 }
2380 endIndex = nResultPos + csWord.GetLength() - 1;
2381 if(iWord == 0) {
2382 m_resStart = nResultPos;
2383 }
2384 FX_BOOL bMatch = TRUE;
2385 if(iWord != 0 && !bSpaceStart) {
2386 int PreResEndPos = nStartPos;
2387 int curChar = csWord.GetAt(0);
2388 CFX_WideString lastWord = m_csFindWhatArray[iWord - 1];
2389 int lastChar = lastWord.GetAt(lastWord.GetLength() - 1);
2390 if(nStartPos == nResultPos && !(_IsIgnoreSpaceCharacter(lastChar) || _IsIgnoreSpaceCharacter(curChar))) {
2391 bMatch = FALSE;
2392 }
2393 for(int d = PreResEndPos; d < nResultPos; d++) {
2394 FX_WCHAR strInsert = m_strText.GetAt(d);
2395 if(strInsert != TEXT_LINEFEED_CHAR && strInsert != TEXT_BLANK_CHAR && strInsert != TEXT_RETURN_CHAR && strInsert != 160) {
2396 bMatch = FALSE;
2397 break;
2398 }
2399 }
2400 } else if(bSpaceStart) {
2401 if(nResultPos > 0) {
2402 FX_WCHAR strInsert = m_strText.GetAt(nResultPos - 1);
2403 if(strInsert != TEXT_LINEFEED_CHAR && strInsert != TEXT_BLANK_CHAR && strInsert != TEXT_RETURN_CHAR && strInsert != 160) {
2404 bMatch = FALSE;
2405 m_resStart = nResultPos;
2406 } else {
2407 m_resStart = nResultPos - 1;
2408 }
2409 }
2410 }
2411 if(m_bMatchWholeWord && bMatch) {
2412 bMatch = IsMatchWholeWord(m_strText, nResultPos, endIndex);
2413 }
2414 nStartPos = endIndex + 1;
2415 if(!bMatch) {
2416 iWord = -1;
2417 if(bSpaceStart) {
2418 nStartPos = m_resStart + m_csFindWhatArray[1].GetLength();
2419 } else {
2420 nStartPos = m_resStart + m_csFindWhatArray[0].GetLength();
2421 }
2422 }
2423 }
2424 m_resEnd = nResultPos + m_csFindWhatArray[m_csFindWhatArray.GetSize() - 1].GetLength() - 1;
2425 m_IsFind = TRUE;
2426 int resStart = GetCharIndex(m_resStart);
2427 int resEnd = GetCharIndex(m_resEnd);
2428 m_pTextPage->GetRectArray(resStart, resEnd - resStart + 1, m_resArray);
2429 if(m_flags & FPDFTEXT_CONSECUTIVE) {
2430 m_findNextStart = m_resStart + 1;
2431 m_findPreStart = m_resEnd - 1;
2432 } else {
2433 m_findNextStart = m_resEnd + 1;
2434 m_findPreStart = m_resStart - 1;
2435 }
2436 return m_IsFind;
2437 }
FindPrev()2438 FX_BOOL CPDF_TextPageFind::FindPrev()
2439 {
2440 if (!m_pTextPage) {
2441 return FALSE;
2442 }
2443 m_resArray.RemoveAll();
2444 if(m_strText.IsEmpty() || m_findPreStart < 0) {
2445 m_IsFind = FALSE;
2446 return m_IsFind;
2447 }
2448 CPDF_TextPageFind findEngine(m_pTextPage);
2449 FX_BOOL ret = findEngine.FindFirst(m_findWhat, m_flags);
2450 if(!ret) {
2451 m_IsFind = FALSE;
2452 return m_IsFind;
2453 }
2454 int order = -1, MatchedCount = 0;
2455 while(ret) {
2456 ret = findEngine.FindNext();
2457 if(ret) {
2458 int order1 = findEngine.GetCurOrder() ;
2459 int MatchedCount1 = findEngine.GetMatchedCount();
2460 if(((order1 + MatchedCount1) - 1) > m_findPreStart) {
2461 break;
2462 }
2463 order = order1;
2464 MatchedCount = MatchedCount1;
2465 }
2466 }
2467 if(order == -1) {
2468 m_IsFind = FALSE;
2469 return m_IsFind;
2470 }
2471 m_resStart = m_pTextPage->TextIndexFromCharIndex(order);
2472 m_resEnd = m_pTextPage->TextIndexFromCharIndex(order + MatchedCount - 1);
2473 m_IsFind = TRUE;
2474 m_pTextPage->GetRectArray(order, MatchedCount, m_resArray);
2475 if(m_flags & FPDFTEXT_CONSECUTIVE) {
2476 m_findNextStart = m_resStart + 1;
2477 m_findPreStart = m_resEnd - 1;
2478 } else {
2479 m_findNextStart = m_resEnd + 1;
2480 m_findPreStart = m_resStart - 1;
2481 }
2482 return m_IsFind;
2483 }
ExtractFindWhat(const CFX_WideString & findwhat)2484 void CPDF_TextPageFind::ExtractFindWhat(const CFX_WideString& findwhat)
2485 {
2486 if(findwhat.IsEmpty()) {
2487 return ;
2488 }
2489 int index = 0;
2490 while(1) {
2491 CFX_WideString csWord = TEXT_EMPTY;
2492 int ret = ExtractSubString(csWord, findwhat.c_str(), index, TEXT_BLANK_CHAR);
2493 if(csWord.IsEmpty()) {
2494 if(ret) {
2495 m_csFindWhatArray.Add(CFX_WideString(L""));
2496 index++;
2497 continue;
2498 } else {
2499 break;
2500 }
2501 }
2502 int pos = 0;
2503 while(pos < csWord.GetLength()) {
2504 CFX_WideString curStr = csWord.Mid(pos, 1);
2505 FX_WCHAR curChar = csWord.GetAt(pos);
2506 if (_IsIgnoreSpaceCharacter(curChar)) {
2507 if (pos > 0 && curChar == 0x2019) {
2508 pos++;
2509 continue;
2510 }
2511 if (pos > 0 ) {
2512 CFX_WideString preStr = csWord.Mid(0, pos);
2513 m_csFindWhatArray.Add(preStr);
2514 }
2515 m_csFindWhatArray.Add(curStr);
2516 if (pos == csWord.GetLength() - 1) {
2517 csWord.Empty();
2518 break;
2519 }
2520 csWord = csWord.Right(csWord.GetLength() - pos - 1);
2521 pos = 0;
2522 continue;
2523 }
2524 pos++;
2525 }
2526 if (!csWord.IsEmpty()) {
2527 m_csFindWhatArray.Add(csWord);
2528 }
2529 index++;
2530 }
2531 }
IsMatchWholeWord(const CFX_WideString & csPageText,int startPos,int endPos)2532 FX_BOOL CPDF_TextPageFind::IsMatchWholeWord(const CFX_WideString& csPageText, int startPos, int endPos)
2533 {
2534 int char_left = 0;
2535 int char_right = 0;
2536 int char_count = endPos - startPos + 1;
2537 if(char_count < 1) {
2538 return FALSE;
2539 }
2540 if (char_count == 1 && csPageText.GetAt(startPos) > 255) {
2541 return TRUE;
2542 }
2543 if(startPos - 1 >= 0 ) {
2544 char_left = csPageText.GetAt(startPos - 1);
2545 }
2546 if(startPos + char_count < csPageText.GetLength()) {
2547 char_right = csPageText.GetAt(startPos + char_count);
2548 }
2549 if ((char_left > 'A' && char_left < 'a') || (char_left > 'a' && char_left < 'z') || (char_left > 0xfb00 && char_left < 0xfb06) || (char_left >= '0' && char_left <= '9') ||
2550 (char_right > 'A' && char_right < 'a') || (char_right > 'a' && char_right < 'z') || (char_right > 0xfb00 && char_right < 0xfb06) || (char_right >= '0' && char_right <= '9')) {
2551 return FALSE;
2552 }
2553 if(!(('A' > char_left || char_left > 'Z') && ('a' > char_left || char_left > 'z')
2554 && ('A' > char_right || char_right > 'Z') && ('a' > char_right || char_right > 'z'))) {
2555 return FALSE;
2556 }
2557 if (char_count > 0) {
2558 if (csPageText.GetAt(startPos) >= L'0' && csPageText.GetAt(startPos) <= L'9' && char_left >= L'0' && char_left <= L'9') {
2559 return FALSE;
2560 }
2561 if (csPageText.GetAt(endPos) >= L'0' && csPageText.GetAt(endPos) <= L'9' && char_right >= L'0' && char_right <= L'9') {
2562 return FALSE;
2563 }
2564 }
2565 return TRUE;
2566 }
ExtractSubString(CFX_WideString & rString,FX_LPCWSTR lpszFullString,int iSubString,FX_WCHAR chSep)2567 FX_BOOL CPDF_TextPageFind::ExtractSubString(CFX_WideString& rString, FX_LPCWSTR lpszFullString,
2568 int iSubString, FX_WCHAR chSep)
2569 {
2570 if (lpszFullString == NULL) {
2571 return FALSE;
2572 }
2573 while (iSubString--) {
2574 lpszFullString = FXSYS_wcschr(lpszFullString, chSep);
2575 if (lpszFullString == NULL) {
2576 rString.Empty();
2577 return FALSE;
2578 }
2579 lpszFullString++;
2580 while(*lpszFullString == chSep) {
2581 lpszFullString++;
2582 }
2583 }
2584 FX_LPCWSTR lpchEnd = FXSYS_wcschr(lpszFullString, chSep);
2585 int nLen = (lpchEnd == NULL) ?
2586 (int)FXSYS_wcslen(lpszFullString) : (int)(lpchEnd - lpszFullString);
2587 ASSERT(nLen >= 0);
2588 FXSYS_memcpy32(rString.GetBuffer(nLen), lpszFullString, nLen * sizeof(FX_WCHAR));
2589 rString.ReleaseBuffer();
2590 return TRUE;
2591 }
MakeReverse(const CFX_WideString & str)2592 CFX_WideString CPDF_TextPageFind::MakeReverse(const CFX_WideString& str)
2593 {
2594 CFX_WideString str2;
2595 str2.Empty();
2596 int nlen = str.GetLength();
2597 for(int i = nlen - 1; i >= 0; i--) {
2598 str2 += str.GetAt(i);
2599 }
2600 return str2;
2601 }
GetRectArray(CFX_RectArray & rects) const2602 void CPDF_TextPageFind::GetRectArray(CFX_RectArray& rects) const
2603 {
2604 rects.Copy(m_resArray);
2605 }
GetCurOrder() const2606 int CPDF_TextPageFind::GetCurOrder() const
2607 {
2608 return GetCharIndex(m_resStart);
2609 }
GetMatchedCount() const2610 int CPDF_TextPageFind::GetMatchedCount()const
2611 {
2612 int resStart = GetCharIndex(m_resStart);
2613 int resEnd = GetCharIndex(m_resEnd);
2614 return resEnd - resStart + 1;
2615 }
CPDF_LinkExtract()2616 CPDF_LinkExtract::CPDF_LinkExtract()
2617 : m_pTextPage(NULL),
2618 m_IsParserd(FALSE)
2619 {
2620 }
~CPDF_LinkExtract()2621 CPDF_LinkExtract::~CPDF_LinkExtract()
2622 {
2623 DeleteLinkList();
2624 }
ExtractLinks(const IPDF_TextPage * pTextPage)2625 FX_BOOL CPDF_LinkExtract::ExtractLinks(const IPDF_TextPage* pTextPage)
2626 {
2627 if (!pTextPage || !pTextPage->IsParsered()) {
2628 return FALSE;
2629 }
2630 m_pTextPage = (const CPDF_TextPage*)pTextPage;
2631 m_strPageText = m_pTextPage->GetPageText(0, -1);
2632 DeleteLinkList();
2633 if (m_strPageText.IsEmpty()) {
2634 return FALSE;
2635 }
2636 parserLink();
2637 m_IsParserd = TRUE;
2638 return TRUE;
2639 }
DeleteLinkList()2640 void CPDF_LinkExtract::DeleteLinkList()
2641 {
2642 while (m_LinkList.GetSize()) {
2643 CPDF_LinkExt* linkinfo = NULL;
2644 linkinfo = m_LinkList.GetAt(0);
2645 m_LinkList.RemoveAt(0);
2646 delete linkinfo;
2647 }
2648 m_LinkList.RemoveAll();
2649 }
CountLinks() const2650 int CPDF_LinkExtract::CountLinks() const
2651 {
2652 if (!m_IsParserd) {
2653 return -1;
2654 }
2655 return m_LinkList.GetSize();
2656 }
parserLink()2657 void CPDF_LinkExtract::parserLink()
2658 {
2659 int start = 0, pos = 0;
2660 int TotalChar = m_pTextPage->CountChars();
2661 while (pos < TotalChar) {
2662 FPDF_CHAR_INFO pageChar;
2663 m_pTextPage->GetCharInfo(pos, pageChar);
2664 if (pageChar.m_Flag == CHAR_GENERATED || pageChar.m_Unicode == 0x20 || pos == TotalChar - 1) {
2665 int nCount = pos - start;
2666 if(pos == TotalChar - 1) {
2667 nCount++;
2668 }
2669 CFX_WideString strBeCheck;
2670 strBeCheck = m_pTextPage->GetPageText(start, nCount);
2671 if (strBeCheck.GetLength() > 5) {
2672 while(strBeCheck.GetLength() > 0) {
2673 FX_WCHAR ch = strBeCheck.GetAt(strBeCheck.GetLength() - 1);
2674 if (ch == L')' || ch == L',' || ch == L'>' || ch == L'.') {
2675 strBeCheck = strBeCheck.Mid(0, strBeCheck.GetLength() - 1);
2676 nCount--;
2677 } else {
2678 break;
2679 }
2680 }
2681 if (nCount > 5 && (CheckWebLink(strBeCheck) || CheckMailLink(strBeCheck))) {
2682 if (!AppendToLinkList(start, nCount, strBeCheck)) {
2683 break;
2684 }
2685 }
2686 }
2687 start = ++pos;
2688 } else {
2689 pos++;
2690 }
2691 }
2692 }
CheckWebLink(CFX_WideString & strBeCheck)2693 FX_BOOL CPDF_LinkExtract::CheckWebLink(CFX_WideString& strBeCheck)
2694 {
2695 CFX_WideString str = strBeCheck;
2696 str.MakeLower();
2697 if (str.Find(L"http://www.") != -1) {
2698 strBeCheck = strBeCheck.Right(str.GetLength() - str.Find(L"http://www."));
2699 return TRUE;
2700 } else if (str.Find(L"http://") != -1) {
2701 strBeCheck = strBeCheck.Right(str.GetLength() - str.Find(L"http://"));
2702 return TRUE;
2703 } else if (str.Find(L"https://www.") != -1) {
2704 strBeCheck = strBeCheck.Right(str.GetLength() - str.Find(L"https://www."));
2705 return TRUE;
2706 } else if (str.Find(L"https://") != -1) {
2707 strBeCheck = strBeCheck.Right(str.GetLength() - str.Find(L"https://"));
2708 return TRUE;
2709 } else if (str.Find(L"www.") != -1) {
2710 strBeCheck = strBeCheck.Right(str.GetLength() - str.Find(L"www."));
2711 strBeCheck = L"http://" + strBeCheck;
2712 return TRUE;
2713 } else {
2714 return FALSE;
2715 }
2716 }
CheckMailLink(CFX_WideString & str)2717 FX_BOOL CPDF_LinkExtract::CheckMailLink(CFX_WideString& str)
2718 {
2719 str.MakeLower();
2720 int aPos = str.Find(L'@');
2721 if (aPos < 1) {
2722 return FALSE;
2723 }
2724 if (str.GetAt(aPos - 1) == L'.' || str.GetAt(aPos - 1) == L'_') {
2725 return FALSE;
2726 }
2727 int i;
2728 for (i = aPos - 1; i >= 0; i--) {
2729 FX_WCHAR ch = str.GetAt(i);
2730 if (ch == L'_' || ch == L'.' || (ch >= L'a' && ch <= L'z') || (ch >= L'0' && ch <= L'9')) {
2731 continue;
2732 } else {
2733 if (i == aPos - 1) {
2734 return FALSE;
2735 }
2736 str = str.Right(str.GetLength() - i - 1);
2737 break;
2738 }
2739 }
2740 aPos = str.Find(L'@');
2741 if (aPos < 1) {
2742 return FALSE;
2743 }
2744 CFX_WideString strtemp = L"";
2745 for (i = 0; i < aPos; i++) {
2746 FX_WCHAR wch = str.GetAt(i);
2747 if (wch >= L'a' && wch <= L'z') {
2748 break;
2749 } else {
2750 strtemp = str.Right(str.GetLength() - i + 1);
2751 }
2752 }
2753 if (strtemp != L"") {
2754 str = strtemp;
2755 }
2756 aPos = str.Find(L'@');
2757 if (aPos < 1) {
2758 return FALSE;
2759 }
2760 str.TrimRight(L'.');
2761 strtemp = str;
2762 int ePos = str.Find(L'.');
2763 if (ePos == -1) {
2764 return FALSE;
2765 }
2766 while (ePos != -1) {
2767 strtemp = strtemp.Right(strtemp.GetLength() - ePos - 1);
2768 ePos = strtemp.Find('.');
2769 }
2770 ePos = strtemp.GetLength();
2771 for (i = 0; i < ePos; i++) {
2772 FX_WCHAR wch = str.GetAt(i);
2773 if ((wch >= L'a' && wch <= L'z') || (wch >= L'0' && wch <= L'9')) {
2774 continue;
2775 } else {
2776 str = str.Left(str.GetLength() - ePos + i + 1);
2777 ePos = ePos - i - 1;
2778 break;
2779 }
2780 }
2781 int nLen = str.GetLength();
2782 for (i = aPos + 1; i < nLen - ePos; i++) {
2783 FX_WCHAR wch = str.GetAt(i);
2784 if (wch == L'-' || wch == L'.' || (wch >= L'a' && wch <= L'z') || (wch >= L'0' && wch <= L'9')) {
2785 continue;
2786 } else {
2787 return FALSE;
2788 }
2789 }
2790 if (str.Find(L"mailto:") == -1) {
2791 str = L"mailto:" + str;
2792 }
2793 return TRUE;
2794 }
AppendToLinkList(int start,int count,const CFX_WideString & strUrl)2795 FX_BOOL CPDF_LinkExtract::AppendToLinkList(int start, int count, const CFX_WideString& strUrl)
2796 {
2797 CPDF_LinkExt* linkInfo = new CPDF_LinkExt;
2798 linkInfo->m_strUrl = strUrl;
2799 linkInfo->m_Start = start;
2800 linkInfo->m_Count = count;
2801 m_LinkList.Add(linkInfo);
2802 return TRUE;
2803 }
GetURL(int index) const2804 CFX_WideString CPDF_LinkExtract::GetURL(int index) const
2805 {
2806 if (!m_IsParserd || index < 0 || index >= m_LinkList.GetSize()) {
2807 return L"";
2808 }
2809 CPDF_LinkExt* link = NULL;
2810 link = m_LinkList.GetAt(index);
2811 if (!link) {
2812 return L"";
2813 }
2814 return link->m_strUrl;
2815 }
GetBoundedSegment(int index,int & start,int & count) const2816 void CPDF_LinkExtract::GetBoundedSegment(int index, int& start, int& count) const
2817 {
2818 if (!m_IsParserd || index < 0 || index >= m_LinkList.GetSize()) {
2819 return ;
2820 }
2821 CPDF_LinkExt* link = NULL;
2822 link = m_LinkList.GetAt(index);
2823 if (!link) {
2824 return ;
2825 }
2826 start = link->m_Start;
2827 count = link->m_Count;
2828 }
GetRects(int index,CFX_RectArray & rects) const2829 void CPDF_LinkExtract::GetRects(int index, CFX_RectArray& rects) const
2830 {
2831 if (!m_IsParserd || index < 0 || index >= m_LinkList.GetSize()) {
2832 return;
2833 }
2834 CPDF_LinkExt* link = NULL;
2835 link = m_LinkList.GetAt(index);
2836 if (!link) {
2837 return ;
2838 }
2839 m_pTextPage->GetRectArray(link->m_Start, link->m_Count, rects);
2840 }
2841