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 "reflowedtextpage.h"
CreateReflowTextPage(IPDF_ReflowedPage * pRefPage)8 IPDF_TextPage* IPDF_TextPage::CreateReflowTextPage(IPDF_ReflowedPage* pRefPage)
9 {
10 return FX_NEW CRF_TextPage(pRefPage);
11 }
CRF_TextPage(IPDF_ReflowedPage * pRefPage)12 CRF_TextPage::CRF_TextPage(IPDF_ReflowedPage* pRefPage)
13 {
14 m_pRefPage = (CPDF_ReflowedPage*)(pRefPage);
15 m_pDataList = NULL;
16 m_CountBSArray = NULL;
17 }
~CRF_TextPage()18 CRF_TextPage::~CRF_TextPage()
19 {
20 if(m_pDataList) {
21 delete m_pDataList;
22 m_pDataList = NULL;
23 }
24 if(m_CountBSArray) {
25 delete m_CountBSArray;
26 m_CountBSArray = NULL;
27 }
28 }
ParseTextPage()29 FX_BOOL CRF_TextPage::ParseTextPage()
30 {
31 if(!m_pRefPage) {
32 return FALSE;
33 }
34 int count = m_pRefPage->m_pReflowed->GetSize();
35 if(count < 500) {
36 m_pDataList = FX_NEW CRF_CharDataPtrArray(count);
37 } else {
38 m_pDataList = FX_NEW CRF_CharDataPtrArray(500);
39 }
40 if (NULL == m_pDataList) {
41 return FALSE;
42 }
43 for(int i = 0; i < count; i++) {
44 CRF_Data* pData = (*(m_pRefPage->m_pReflowed))[i];
45 if(pData->GetType() == CRF_Data::Text) {
46 m_pDataList->Add((CRF_CharData*)pData);
47 }
48 }
49 m_CountBSArray = FX_NEW CFX_CountBSINT32Array(20);
50 if(NULL == m_CountBSArray) {
51 return FALSE;
52 }
53 return TRUE;
54 }
IsParsered() const55 FX_BOOL CRF_TextPage::IsParsered() const
56 {
57 if(m_pDataList) {
58 return TRUE;
59 }
60 return FALSE;
61 }
CharIndexFromTextIndex(int TextIndex) const62 int CRF_TextPage::CharIndexFromTextIndex(int TextIndex) const
63 {
64 return TextIndex;
65 }
TextIndexFromCharIndex(int CharIndex) const66 int CRF_TextPage::TextIndexFromCharIndex(int CharIndex) const
67 {
68 return CharIndex;
69 }
70
CountChars() const71 int CRF_TextPage::CountChars() const
72 {
73 if (NULL == m_pDataList) {
74 return -1;
75 }
76 return m_pDataList->GetSize();
77 }
GetCharInfo(int index,FPDF_CHAR_INFO & info) const78 void CRF_TextPage::GetCharInfo(int index, FPDF_CHAR_INFO & info) const
79 {
80 if(index >= CountChars() || index < 0 || !m_pDataList) {
81 return;
82 }
83 CRF_CharData* pData = (*m_pDataList)[index];
84 FX_FLOAT ReltiveCorddDs = pData->m_pCharState->m_fDescent;
85 FX_FLOAT ReltiveCorddAs = pData->m_pCharState->m_fAscent;
86 info.m_Flag = CHAR_NORMAL;
87 info.m_pTextObj = pData->m_pCharState->m_pTextObj;
88 info.m_OriginX = pData->m_PosX;
89 info.m_OriginY = pData->m_PosY - ReltiveCorddDs;
90 info.m_FontSize = pData->m_pCharState->m_fFontSize;
91 CFX_FloatRect FloatRectTmp(pData->m_PosX, pData->m_PosY, pData->m_PosX + pData->m_Width, pData->m_PosY + ReltiveCorddAs - ReltiveCorddDs);
92 info.m_CharBox = FloatRectTmp;
93 CFX_WideString str = pData->m_pCharState->m_pFont->UnicodeFromCharCode(pData->m_CharCode);
94 if(!str.IsEmpty()) {
95 info.m_Unicode = str.GetAt(0);
96 } else {
97 info.m_Unicode = -1;
98 }
99 info.m_Charcode = (FX_WCHAR)pData->m_CharCode;
100 info.m_Matrix = CFX_Matrix(1, 0, 0, 1, 0, 0);
101 }
102 extern FX_BOOL GetIntersection(FX_FLOAT low1, FX_FLOAT high1, FX_FLOAT low2, FX_FLOAT high2, FX_FLOAT& interlow, FX_FLOAT& interhigh);
_IsInsameline(const CFX_FloatRect & rectA,const CFX_FloatRect & rectB)103 inline FX_BOOL _IsInsameline(const CFX_FloatRect& rectA, const CFX_FloatRect& rectB)
104 {
105 if((rectA.top >= rectB.bottom && rectB.top >= rectA.bottom)) {
106 return TRUE;
107 } else {
108 return FALSE;
109 }
110 }
_IsIntersect(const CFX_FloatRect & rectA,const CFX_FloatRect & rectB)111 inline FX_BOOL _IsIntersect(const CFX_FloatRect& rectA, const CFX_FloatRect& rectB)
112 {
113 FX_FLOAT interlow = .0f, interhigh = .0f;
114 if(GetIntersection(rectA.bottom, rectA.top, rectB.bottom, rectB.top, interlow, interhigh)) {
115 if(GetIntersection(rectA.left, rectA.right, rectB.left, rectB.right, interlow, interhigh)) {
116 return TRUE;
117 } else {
118 return FALSE;
119 }
120 }
121 return FALSE;
122 }
GetRectArray(int start,int nCount,CFX_RectArray & rectArray) const123 void CRF_TextPage::GetRectArray(int start, int nCount, CFX_RectArray& rectArray) const
124 {
125 int indexlen = start + nCount;
126 FPDF_CHAR_INFO info;
127 FX_BOOL bstart = TRUE;
128 CFX_FloatRect recttmp;
129 int i;
130 for(i = start; i < indexlen; i++) {
131 GetCharInfo(i, info);
132 if(bstart) {
133 recttmp = info.m_CharBox;
134 bstart = FALSE;
135 } else if(_IsInsameline(recttmp, info.m_CharBox)) {
136 recttmp.right = info.m_CharBox.right;
137 if(info.m_CharBox.top > recttmp.top) {
138 recttmp.top = info.m_CharBox.top;
139 }
140 if(info.m_CharBox.bottom < recttmp.bottom) {
141 recttmp.bottom = info.m_CharBox.bottom;
142 }
143 } else {
144 rectArray.Add(recttmp);
145 recttmp = info.m_CharBox;
146 }
147 }
148 rectArray.Add(recttmp);
149 }
_GetDistance(CFX_FloatRect floatRect,CPDF_Point point)150 inline FX_FLOAT _GetDistance(CFX_FloatRect floatRect, CPDF_Point point)
151 {
152 if(floatRect.right < point.x && floatRect.bottom > point.y) {
153 return FXSYS_sqrt(FXSYS_pow(point.x - floatRect.right, 2) + FXSYS_pow(floatRect.bottom - point.y, 2));
154 }
155 if (floatRect.right < point.x && floatRect.top < point.y) {
156 return FXSYS_sqrt(FXSYS_pow(point.x - floatRect.right, 2) + FXSYS_pow(point.y - floatRect.top, 2));
157 }
158 if(floatRect.left > point.x && floatRect.bottom > point.y) {
159 return FXSYS_sqrt(FXSYS_pow(floatRect.bottom - point.y, 2) + FXSYS_pow(floatRect.left - point.x, 2));
160 }
161 if((floatRect.right > point.x || FXSYS_fabs(floatRect.right - point.x) <= 0.0001f) &&
162 (floatRect.left < point.x || FXSYS_fabs(floatRect.left - point.x) <= 0.0001f) && floatRect.bottom > point.y) {
163 return FXSYS_fabs(floatRect.bottom - point.y);
164 }
165 if(floatRect.left > point.x && (floatRect.bottom < point.y || FXSYS_fabs(floatRect.bottom - point.y) <= 0.0001f) &&
166 (floatRect.top > point.y || FXSYS_fabs(floatRect.top - point.y) <= 0.0001f)) {
167 return FXSYS_fabs(floatRect.left - point.x);
168 }
169 if(floatRect.left > point.x && floatRect.top < point.y) {
170 return FXSYS_sqrt(FXSYS_pow(floatRect.left - point.x, 2) + FXSYS_pow(point.y - floatRect.top, 2));
171 }
172 if ((floatRect.left < point.x || FXSYS_fabs(floatRect.left - point.x) <= 0.0001f) &&
173 (floatRect.right > point.x || FXSYS_fabs(floatRect.right - point.x) <= 0.0001f) && floatRect.top < point.y) {
174 return FXSYS_fabs(point.y - floatRect.top);
175 }
176 if(floatRect.right < point.x && (floatRect.top > point.y || FXSYS_fabs(floatRect.top - point.y) <= 0.0001f) &&
177 (floatRect.bottom < point.y || FXSYS_fabs(floatRect.bottom - point.y) <= 0.0001f)) {
178 return point.x - floatRect.right;
179 }
180 return .0f;
181 }
GetIndexAtPos(CPDF_Point point,FX_FLOAT xTorelance,FX_FLOAT yTorelance) const182 int CRF_TextPage::GetIndexAtPos(CPDF_Point point, FX_FLOAT xTorelance, FX_FLOAT yTorelance) const
183 {
184 int index = -1, i = 0, j = 0;
185 FPDF_CHAR_INFO info;
186 CFX_FloatRect rectTmp;
187 FX_FLOAT MinDistance = 1000, DistanceTmp = 0;
188 FX_FLOAT rect_bottom = point.x - xTorelance;
189 CFX_FloatRect TorelanceRect(rect_bottom <= 0 ? 0 : rect_bottom, point.y - yTorelance, point.x + xTorelance, point.y + yTorelance);
190 int count = CountChars();
191 for(i = 0; i < count; i++) {
192 GetCharInfo(i, info);
193 rectTmp = info.m_CharBox;
194 if(rectTmp.Contains(point.x, point.y)) {
195 index = i;
196 break;
197 } else if(_IsIntersect(rectTmp, TorelanceRect)) {
198 DistanceTmp = _GetDistance(rectTmp, point);
199 if(DistanceTmp < MinDistance) {
200 MinDistance = DistanceTmp;
201 index = i;
202 }
203 }
204 }
205 return index;
206 }
GetIndexAtPos(FX_FLOAT x,FX_FLOAT y,FX_FLOAT xTorelance,FX_FLOAT yTorelance) const207 int CRF_TextPage::GetIndexAtPos(FX_FLOAT x, FX_FLOAT y, FX_FLOAT xTorelance, FX_FLOAT yTorelance) const
208 {
209 int index = 0;
210 CPDF_Point point(x, y);
211 if((index = GetIndexAtPos(point, xTorelance, yTorelance)) < 0) {
212 return -1;
213 } else {
214 return index;
215 }
216 }
GetOrderByDirection(int index,int direction) const217 int CRF_TextPage::GetOrderByDirection(int index, int direction) const
218 {
219 return -1;
220 }
GetTextByRect(CFX_FloatRect rect) const221 CFX_WideString CRF_TextPage::GetTextByRect(CFX_FloatRect rect) const
222 {
223 int count;
224 FPDF_CHAR_INFO info;
225 CFX_WideString str;
226 CFX_FloatRect Recttmp;
227 FX_BOOL bstart = TRUE;
228 count = CountChars();
229 if(rect.IsEmpty()) {
230 return L"";
231 }
232 for(int i = 0; i < count; i++) {
233 GetCharInfo(i, info);
234 if(_IsIntersect(rect, info.m_CharBox)) {
235 if(bstart) {
236 Recttmp = info.m_CharBox;
237 str += info.m_Unicode;
238 bstart = FALSE;
239 } else if(_IsInsameline(Recttmp, info.m_CharBox)) {
240 str += info.m_Unicode;
241 } else {
242 str += L"\r\n";
243 Recttmp = info.m_CharBox;
244 str += info.m_Unicode;
245 }
246 }
247 }
248 if(str.IsEmpty()) {
249 return L"";
250 } else {
251 return str;
252 }
253 }
GetRectsArrayByRect(CFX_FloatRect rect,CFX_RectArray & resRectArray) const254 void CRF_TextPage::GetRectsArrayByRect(CFX_FloatRect rect, CFX_RectArray& resRectArray) const
255 {
256 int count, i;
257 FX_BOOL bstart = TRUE;
258 FPDF_CHAR_INFO info;
259 CFX_FloatRect recttmp;
260 count = CountChars();
261 for(i = 0; i < count; i++) {
262 GetCharInfo(i, info);
263 if(_IsIntersect(rect, info.m_CharBox)) {
264 if(bstart) {
265 recttmp = info.m_CharBox;
266 bstart = FALSE;
267 } else if(_IsInsameline(recttmp, info.m_CharBox)) {
268 recttmp.right = info.m_CharBox.right;
269 if(info.m_CharBox.top > recttmp.top) {
270 recttmp.top = info.m_CharBox.top;
271 }
272 if(info.m_CharBox.bottom < recttmp.bottom) {
273 recttmp.bottom = info.m_CharBox.bottom;
274 }
275 } else {
276 resRectArray.Add(recttmp);
277 recttmp = info.m_CharBox;
278 }
279 }
280 }
281 resRectArray.Add(recttmp);
282 }
CountRects(int start,int nCount)283 int CRF_TextPage::CountRects(int start, int nCount)
284 {
285 m_rectArray.RemoveAll();
286 GetRectArray(start, nCount, m_rectArray);
287 return m_rectArray.GetSize();
288 }
GetRect(int rectIndex,FX_FLOAT & left,FX_FLOAT & top,FX_FLOAT & right,FX_FLOAT & bottom) const289 void CRF_TextPage::GetRect(int rectIndex, FX_FLOAT& left, FX_FLOAT& top, FX_FLOAT& right, FX_FLOAT &bottom) const
290 {
291 if(m_rectArray.GetSize() <= rectIndex) {
292 return;
293 }
294 left = m_rectArray[rectIndex].left;
295 top = m_rectArray[rectIndex].top;
296 right = m_rectArray[rectIndex].right;
297 bottom = m_rectArray[rectIndex].bottom;
298 }
GetBaselineRotate(int rectIndex,int & Rotate)299 FX_BOOL CRF_TextPage::GetBaselineRotate(int rectIndex, int& Rotate)
300 {
301 Rotate = 0;
302 return TRUE;
303 }
GetBaselineRotate(CFX_FloatRect rect,int & Rotate)304 FX_BOOL CRF_TextPage::GetBaselineRotate(CFX_FloatRect rect, int& Rotate)
305 {
306 Rotate = 0;
307 return TRUE;
308 }
CountBoundedSegments(FX_FLOAT left,FX_FLOAT top,FX_FLOAT right,FX_FLOAT bottom,FX_BOOL bContains)309 int CRF_TextPage::CountBoundedSegments(FX_FLOAT left, FX_FLOAT top, FX_FLOAT right, FX_FLOAT bottom, FX_BOOL bContains)
310 {
311 if (!m_CountBSArray) {
312 return -1;
313 }
314 m_CountBSArray->RemoveAll();
315 CFX_FloatRect floatrect(left, bottom, right, top);
316 int totalcount, i, j = 0, counttmp = 0;
317 FX_BOOL bstart = TRUE;
318 FPDF_CHAR_INFO info;
319 CFX_FloatRect recttmp;
320 totalcount = CountChars();
321 for(i = 0; i < totalcount; i++) {
322 GetCharInfo(i, info);
323 if(_IsIntersect(floatrect, info.m_CharBox)) {
324 if(bstart) {
325 m_CountBSArray->Add(i);
326 counttmp = 1;
327 recttmp = info.m_CharBox;
328 bstart = FALSE;
329 } else if(_IsInsameline(recttmp, info.m_CharBox)) {
330 recttmp.right = info.m_CharBox.right;
331 if(info.m_CharBox.top > recttmp.top) {
332 recttmp.top = info.m_CharBox.top;
333 }
334 if(info.m_CharBox.bottom < recttmp.bottom) {
335 recttmp.bottom = info.m_CharBox.bottom;
336 }
337 counttmp ++;
338 } else {
339 m_CountBSArray->Add(counttmp);
340 m_CountBSArray->Add(i);
341 counttmp = 1;
342 j++;
343 recttmp = info.m_CharBox;
344 }
345 }
346 }
347 m_CountBSArray->Add(counttmp);
348 j++;
349 return j;
350 }
GetBoundedSegment(int index,int & start,int & count) const351 void CRF_TextPage::GetBoundedSegment(int index, int& start, int& count) const
352 {
353 if (!m_CountBSArray) {
354 return;
355 }
356 if(m_CountBSArray->GetSize() <= index * 2) {
357 start = 0;
358 count = 0;
359 return;
360 }
361 start = *(int *)m_CountBSArray->GetAt(index * 2);
362 count = *(int *)m_CountBSArray->GetAt(index * 2 + 1);
363 }
364
GetWordBreak(int index,int direction) const365 int CRF_TextPage::GetWordBreak(int index, int direction) const
366 {
367 return -1;
368 }
GetPageText(int start,int nCount) const369 CFX_WideString CRF_TextPage::GetPageText(int start, int nCount ) const
370 {
371 if(nCount == -1) {
372 nCount = CountChars();
373 start = 0;
374 } else if(nCount < 1) {
375 return L"";
376 } else if(start >= CountChars()) {
377 return L"";
378 }
379 int i, index = start + nCount;
380 FPDF_CHAR_INFO info;
381 CFX_WideString str;
382 CFX_FloatRect recttmp;
383 FX_BOOL bstart = TRUE;
384 for(i = start; i < index; i++) {
385 GetCharInfo(i, info);
386 if(bstart) {
387 recttmp = info.m_CharBox;
388 str += info.m_Unicode;
389 bstart = FALSE;
390 } else if (_IsInsameline(recttmp, info.m_CharBox)) {
391 str += info.m_Unicode;
392 } else {
393 str += L"\r\n";
394 recttmp = info.m_CharBox;
395 str += info.m_Unicode;
396 }
397 }
398 if(str.IsEmpty()) {
399 return L"";
400 }
401 return str;
402 }
403