1 // Copyright 2016 PDFium Authors. All rights reserved.
2 // Use of this source code is governed by a BSD-style license that can be
3 // found in the LICENSE file.
4
5 // Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com
6
7 #include "core/fpdfdoc/ctypeset.h"
8
9 #include <algorithm>
10
11 #include "core/fpdfdoc/cline.h"
12 #include "core/fpdfdoc/cpdf_variabletext.h"
13 #include "core/fpdfdoc/cpvt_wordinfo.h"
14 #include "core/fpdfdoc/csection.h"
15 #include "third_party/base/stl_util.h"
16
17 namespace {
18
19 const uint8_t special_chars[128] = {
20 0x00, 0x0C, 0x08, 0x0C, 0x08, 0x00, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
21 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
22 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x08, 0x08, 0x00,
23 0x10, 0x00, 0x00, 0x28, 0x0C, 0x08, 0x00, 0x00, 0x28, 0x28, 0x28, 0x28,
24 0x02, 0x02, 0x02, 0x02, 0x02, 0x02, 0x02, 0x02, 0x02, 0x02, 0x08, 0x08,
25 0x00, 0x00, 0x00, 0x08, 0x00, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
26 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
27 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x0C, 0x00, 0x08, 0x00, 0x00,
28 0x00, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
29 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
30 0x01, 0x01, 0x01, 0x0C, 0x00, 0x08, 0x00, 0x00,
31 };
32
IsLatin(uint16_t word)33 bool IsLatin(uint16_t word) {
34 if (word <= 0x007F)
35 return !!(special_chars[word] & 0x01);
36
37 return ((word >= 0x00C0 && word <= 0x00FF) ||
38 (word >= 0x0100 && word <= 0x024F) ||
39 (word >= 0x1E00 && word <= 0x1EFF) ||
40 (word >= 0x2C60 && word <= 0x2C7F) ||
41 (word >= 0xA720 && word <= 0xA7FF) ||
42 (word >= 0xFF21 && word <= 0xFF3A) ||
43 (word >= 0xFF41 && word <= 0xFF5A));
44 }
45
IsDigit(uint32_t word)46 bool IsDigit(uint32_t word) {
47 return word >= 0x0030 && word <= 0x0039;
48 }
49
IsCJK(uint32_t word)50 bool IsCJK(uint32_t word) {
51 if ((word >= 0x1100 && word <= 0x11FF) ||
52 (word >= 0x2E80 && word <= 0x2FFF) ||
53 (word >= 0x3040 && word <= 0x9FBF) ||
54 (word >= 0xAC00 && word <= 0xD7AF) ||
55 (word >= 0xF900 && word <= 0xFAFF) ||
56 (word >= 0xFE30 && word <= 0xFE4F) ||
57 (word >= 0x20000 && word <= 0x2A6DF) ||
58 (word >= 0x2F800 && word <= 0x2FA1F)) {
59 return true;
60 }
61 if (word >= 0x3000 && word <= 0x303F) {
62 return (
63 word == 0x3005 || word == 0x3006 || word == 0x3021 || word == 0x3022 ||
64 word == 0x3023 || word == 0x3024 || word == 0x3025 || word == 0x3026 ||
65 word == 0x3027 || word == 0x3028 || word == 0x3029 || word == 0x3031 ||
66 word == 0x3032 || word == 0x3033 || word == 0x3034 || word == 0x3035);
67 }
68 return word >= 0xFF66 && word <= 0xFF9D;
69 }
70
IsPunctuation(uint32_t word)71 bool IsPunctuation(uint32_t word) {
72 if (word <= 0x007F)
73 return !!(special_chars[word] & 0x08);
74
75 if (word >= 0x0080 && word <= 0x00FF) {
76 return (word == 0x0082 || word == 0x0084 || word == 0x0085 ||
77 word == 0x0091 || word == 0x0092 || word == 0x0093 ||
78 word <= 0x0094 || word == 0x0096 || word == 0x00B4 ||
79 word == 0x00B8);
80 }
81
82 if (word >= 0x2000 && word <= 0x206F) {
83 return (
84 word == 0x2010 || word == 0x2011 || word == 0x2012 || word == 0x2013 ||
85 word == 0x2018 || word == 0x2019 || word == 0x201A || word == 0x201B ||
86 word == 0x201C || word == 0x201D || word == 0x201E || word == 0x201F ||
87 word == 0x2032 || word == 0x2033 || word == 0x2034 || word == 0x2035 ||
88 word == 0x2036 || word == 0x2037 || word == 0x203C || word == 0x203D ||
89 word == 0x203E || word == 0x2044);
90 }
91
92 if (word >= 0x3000 && word <= 0x303F) {
93 return (
94 word == 0x3001 || word == 0x3002 || word == 0x3003 || word == 0x3005 ||
95 word == 0x3009 || word == 0x300A || word == 0x300B || word == 0x300C ||
96 word == 0x300D || word == 0x300F || word == 0x300E || word == 0x3010 ||
97 word == 0x3011 || word == 0x3014 || word == 0x3015 || word == 0x3016 ||
98 word == 0x3017 || word == 0x3018 || word == 0x3019 || word == 0x301A ||
99 word == 0x301B || word == 0x301D || word == 0x301E || word == 0x301F);
100 }
101
102 if (word >= 0xFE50 && word <= 0xFE6F)
103 return (word >= 0xFE50 && word <= 0xFE5E) || word == 0xFE63;
104
105 if (word >= 0xFF00 && word <= 0xFFEF) {
106 return (
107 word == 0xFF01 || word == 0xFF02 || word == 0xFF07 || word == 0xFF08 ||
108 word == 0xFF09 || word == 0xFF0C || word == 0xFF0E || word == 0xFF0F ||
109 word == 0xFF1A || word == 0xFF1B || word == 0xFF1F || word == 0xFF3B ||
110 word == 0xFF3D || word == 0xFF40 || word == 0xFF5B || word == 0xFF5C ||
111 word == 0xFF5D || word == 0xFF61 || word == 0xFF62 || word == 0xFF63 ||
112 word == 0xFF64 || word == 0xFF65 || word == 0xFF9E || word == 0xFF9F);
113 }
114
115 return false;
116 }
117
IsConnectiveSymbol(uint32_t word)118 bool IsConnectiveSymbol(uint32_t word) {
119 return word <= 0x007F && (special_chars[word] & 0x20);
120 }
121
IsOpenStylePunctuation(uint32_t word)122 bool IsOpenStylePunctuation(uint32_t word) {
123 if (word <= 0x007F)
124 return !!(special_chars[word] & 0x04);
125
126 return (word == 0x300A || word == 0x300C || word == 0x300E ||
127 word == 0x3010 || word == 0x3014 || word == 0x3016 ||
128 word == 0x3018 || word == 0x301A || word == 0xFF08 ||
129 word == 0xFF3B || word == 0xFF5B || word == 0xFF62);
130 }
131
IsCurrencySymbol(uint16_t word)132 bool IsCurrencySymbol(uint16_t word) {
133 return (word == 0x0024 || word == 0x0080 || word == 0x00A2 ||
134 word == 0x00A3 || word == 0x00A4 || word == 0x00A5 ||
135 (word >= 0x20A0 && word <= 0x20CF) || word == 0xFE69 ||
136 word == 0xFF04 || word == 0xFFE0 || word == 0xFFE1 ||
137 word == 0xFFE5 || word == 0xFFE6);
138 }
139
IsPrefixSymbol(uint16_t word)140 bool IsPrefixSymbol(uint16_t word) {
141 return IsCurrencySymbol(word) || word == 0x2116;
142 }
143
IsSpace(uint16_t word)144 bool IsSpace(uint16_t word) {
145 return word == 0x0020 || word == 0x3000;
146 }
147
NeedDivision(uint16_t prevWord,uint16_t curWord)148 bool NeedDivision(uint16_t prevWord, uint16_t curWord) {
149 if ((IsLatin(prevWord) || IsDigit(prevWord)) &&
150 (IsLatin(curWord) || IsDigit(curWord))) {
151 return false;
152 }
153 if (IsSpace(curWord) || IsPunctuation(curWord)) {
154 return false;
155 }
156 if (IsConnectiveSymbol(prevWord) || IsConnectiveSymbol(curWord)) {
157 return false;
158 }
159 if (IsSpace(prevWord) || IsPunctuation(prevWord)) {
160 return true;
161 }
162 if (IsPrefixSymbol(prevWord)) {
163 return false;
164 }
165 if (IsPrefixSymbol(curWord) || IsCJK(curWord)) {
166 return true;
167 }
168 if (IsCJK(prevWord)) {
169 return true;
170 }
171 return false;
172 }
173
174 } // namespace
175
CTypeset(CSection * pSection)176 CTypeset::CTypeset(CSection* pSection)
177 : m_pVT(pSection->m_pVT), m_pSection(pSection) {}
178
179 CTypeset::~CTypeset() = default;
180
CharArray()181 CPVT_FloatRect CTypeset::CharArray() {
182 m_rcRet = CPVT_FloatRect();
183 if (m_pSection->m_LineArray.empty())
184 return m_rcRet;
185
186 float fNodeWidth = m_pVT->GetPlateWidth() /
187 (m_pVT->GetCharArray() <= 0 ? 1 : m_pVT->GetCharArray());
188 float fLineAscent =
189 m_pVT->GetFontAscent(m_pVT->GetDefaultFontIndex(), m_pVT->GetFontSize());
190 float fLineDescent =
191 m_pVT->GetFontDescent(m_pVT->GetDefaultFontIndex(), m_pVT->GetFontSize());
192 float x = 0.0f;
193 float y = m_pVT->GetLineLeading() + fLineAscent;
194 int32_t nStart = 0;
195 CLine* pLine = m_pSection->m_LineArray.front().get();
196 switch (m_pVT->GetAlignment()) {
197 case 0:
198 pLine->m_LineInfo.fLineX = fNodeWidth * VARIABLETEXT_HALF;
199 break;
200 case 1:
201 nStart = (m_pVT->GetCharArray() -
202 pdfium::CollectionSize<int32_t>(m_pSection->m_WordArray)) /
203 2;
204 pLine->m_LineInfo.fLineX =
205 fNodeWidth * nStart - fNodeWidth * VARIABLETEXT_HALF;
206 break;
207 case 2:
208 nStart = m_pVT->GetCharArray() -
209 pdfium::CollectionSize<int32_t>(m_pSection->m_WordArray);
210 pLine->m_LineInfo.fLineX =
211 fNodeWidth * nStart - fNodeWidth * VARIABLETEXT_HALF;
212 break;
213 }
214 for (int32_t w = 0,
215 sz = pdfium::CollectionSize<int32_t>(m_pSection->m_WordArray);
216 w < sz; w++) {
217 if (w >= m_pVT->GetCharArray())
218 break;
219
220 float fNextWidth = 0;
221 if (pdfium::IndexInBounds(m_pSection->m_WordArray, w + 1)) {
222 CPVT_WordInfo* pNextWord = m_pSection->m_WordArray[w + 1].get();
223 pNextWord->fWordTail = 0;
224 fNextWidth = m_pVT->GetWordWidth(*pNextWord);
225 }
226 CPVT_WordInfo* pWord = m_pSection->m_WordArray[w].get();
227 pWord->fWordTail = 0;
228 float fWordWidth = m_pVT->GetWordWidth(*pWord);
229 float fWordAscent = m_pVT->GetWordAscent(*pWord);
230 float fWordDescent = m_pVT->GetWordDescent(*pWord);
231 x = (float)(fNodeWidth * (w + nStart + 0.5) -
232 fWordWidth * VARIABLETEXT_HALF);
233 pWord->fWordX = x;
234 pWord->fWordY = y;
235 if (w == 0) {
236 pLine->m_LineInfo.fLineX = x;
237 }
238 if (w != pdfium::CollectionSize<int32_t>(m_pSection->m_WordArray) - 1) {
239 pWord->fWordTail =
240 (fNodeWidth - (fWordWidth + fNextWidth) * VARIABLETEXT_HALF > 0
241 ? fNodeWidth - (fWordWidth + fNextWidth) * VARIABLETEXT_HALF
242 : 0);
243 } else {
244 pWord->fWordTail = 0;
245 }
246 x += fWordWidth;
247 fLineAscent = std::max(fLineAscent, fWordAscent);
248 fLineDescent = std::min(fLineDescent, fWordDescent);
249 }
250 pLine->m_LineInfo.nBeginWordIndex = 0;
251 pLine->m_LineInfo.nEndWordIndex =
252 pdfium::CollectionSize<int32_t>(m_pSection->m_WordArray) - 1;
253 pLine->m_LineInfo.fLineY = y;
254 pLine->m_LineInfo.fLineWidth = x - pLine->m_LineInfo.fLineX;
255 pLine->m_LineInfo.fLineAscent = fLineAscent;
256 pLine->m_LineInfo.fLineDescent = fLineDescent;
257 m_rcRet = CPVT_FloatRect(0, 0, x, y - fLineDescent);
258 return m_rcRet;
259 }
260
GetEditSize(float fFontSize)261 CFX_SizeF CTypeset::GetEditSize(float fFontSize) {
262 ASSERT(m_pSection);
263 ASSERT(m_pVT);
264 SplitLines(false, fFontSize);
265 return CFX_SizeF(m_rcRet.Width(), m_rcRet.Height());
266 }
267
Typeset()268 CPVT_FloatRect CTypeset::Typeset() {
269 ASSERT(m_pVT);
270 m_pSection->m_LineArray.clear();
271 SplitLines(true, 0.0f);
272 OutputLines();
273 return m_rcRet;
274 }
275
SplitLines(bool bTypeset,float fFontSize)276 void CTypeset::SplitLines(bool bTypeset, float fFontSize) {
277 ASSERT(m_pVT);
278 ASSERT(m_pSection);
279
280 CPVT_LineInfo line;
281 if (m_pSection->m_WordArray.empty()) {
282 float fLineAscent;
283 float fLineDescent;
284 if (bTypeset) {
285 fLineAscent = m_pVT->GetLineAscent();
286 fLineDescent = m_pVT->GetLineDescent();
287 } else {
288 fLineAscent =
289 m_pVT->GetFontAscent(m_pVT->GetDefaultFontIndex(), fFontSize);
290 fLineDescent =
291 m_pVT->GetFontDescent(m_pVT->GetDefaultFontIndex(), fFontSize);
292 }
293 if (bTypeset) {
294 line.nBeginWordIndex = -1;
295 line.nEndWordIndex = -1;
296 line.nTotalWord = 0;
297 line.fLineWidth = 0;
298 line.fLineAscent = fLineAscent;
299 line.fLineDescent = fLineDescent;
300 m_pSection->AddLine(line);
301 }
302 float fMaxY = m_pVT->GetLineLeading() + fLineAscent - fLineDescent;
303 m_rcRet = CPVT_FloatRect(0, 0, 0, fMaxY);
304 return;
305 }
306
307 int32_t nLineHead = 0;
308 int32_t nLineTail = 0;
309 float fMaxX = 0.0f;
310 float fMaxY = 0.0f;
311 float fLineWidth = 0.0f;
312 float fBackupLineWidth = 0.0f;
313 float fLineAscent = 0.0f;
314 float fBackupLineAscent = 0.0f;
315 float fLineDescent = 0.0f;
316 float fBackupLineDescent = 0.0f;
317 int32_t nWordStartPos = 0;
318 bool bFullWord = false;
319 int32_t nLineFullWordIndex = 0;
320 int32_t nCharIndex = 0;
321 float fWordWidth = 0;
322 float fTypesetWidth =
323 std::max(m_pVT->GetPlateWidth() - m_pVT->GetLineIndent(), 0.0f);
324 int32_t nTotalWords =
325 pdfium::CollectionSize<int32_t>(m_pSection->m_WordArray);
326 bool bOpened = false;
327 int32_t i = 0;
328 while (i < nTotalWords) {
329 CPVT_WordInfo* pWord = m_pSection->m_WordArray[i].get();
330 CPVT_WordInfo* pOldWord = pWord;
331 if (i > 0) {
332 pOldWord = m_pSection->m_WordArray[i - 1].get();
333 }
334 if (pWord) {
335 if (bTypeset) {
336 fLineAscent = std::max(fLineAscent, m_pVT->GetWordAscent(*pWord));
337 fLineDescent = std::min(fLineDescent, m_pVT->GetWordDescent(*pWord));
338 fWordWidth = m_pVT->GetWordWidth(*pWord);
339 } else {
340 fLineAscent =
341 std::max(fLineAscent, m_pVT->GetWordAscent(*pWord, fFontSize));
342 fLineDescent =
343 std::min(fLineDescent, m_pVT->GetWordDescent(*pWord, fFontSize));
344 fWordWidth = m_pVT->GetWordWidth(
345 pWord->nFontIndex, pWord->Word, m_pVT->GetSubWord(),
346 m_pVT->GetCharSpace(), fFontSize, pWord->fWordTail);
347 }
348 if (!bOpened) {
349 if (IsOpenStylePunctuation(pWord->Word)) {
350 bOpened = true;
351 bFullWord = true;
352 } else if (pOldWord) {
353 if (NeedDivision(pOldWord->Word, pWord->Word)) {
354 bFullWord = true;
355 }
356 }
357 } else {
358 if (!IsSpace(pWord->Word) && !IsOpenStylePunctuation(pWord->Word)) {
359 bOpened = false;
360 }
361 }
362 if (bFullWord) {
363 bFullWord = false;
364 if (nCharIndex > 0) {
365 nLineFullWordIndex++;
366 }
367 nWordStartPos = i;
368 fBackupLineWidth = fLineWidth;
369 fBackupLineAscent = fLineAscent;
370 fBackupLineDescent = fLineDescent;
371 }
372 nCharIndex++;
373 }
374 if (m_pVT->IsAutoReturn() && fTypesetWidth > 0 &&
375 fLineWidth + fWordWidth > fTypesetWidth) {
376 if (nLineFullWordIndex > 0) {
377 i = nWordStartPos;
378 fLineWidth = fBackupLineWidth;
379 fLineAscent = fBackupLineAscent;
380 fLineDescent = fBackupLineDescent;
381 }
382 if (nCharIndex == 1) {
383 fLineWidth = fWordWidth;
384 i++;
385 }
386 nLineTail = i - 1;
387 if (bTypeset) {
388 line.nBeginWordIndex = nLineHead;
389 line.nEndWordIndex = nLineTail;
390 line.nTotalWord = nLineTail - nLineHead + 1;
391 line.fLineWidth = fLineWidth;
392 line.fLineAscent = fLineAscent;
393 line.fLineDescent = fLineDescent;
394 m_pSection->AddLine(line);
395 }
396 fMaxY += (fLineAscent + m_pVT->GetLineLeading());
397 fMaxY -= fLineDescent;
398 fMaxX = std::max(fLineWidth, fMaxX);
399 nLineHead = i;
400 fLineWidth = 0.0f;
401 fLineAscent = 0.0f;
402 fLineDescent = 0.0f;
403 nCharIndex = 0;
404 nLineFullWordIndex = 0;
405 bFullWord = false;
406 } else {
407 fLineWidth += fWordWidth;
408 i++;
409 }
410 }
411 if (nLineHead <= nTotalWords - 1) {
412 nLineTail = nTotalWords - 1;
413 if (bTypeset) {
414 line.nBeginWordIndex = nLineHead;
415 line.nEndWordIndex = nLineTail;
416 line.nTotalWord = nLineTail - nLineHead + 1;
417 line.fLineWidth = fLineWidth;
418 line.fLineAscent = fLineAscent;
419 line.fLineDescent = fLineDescent;
420 m_pSection->AddLine(line);
421 }
422 fMaxY += (fLineAscent + m_pVT->GetLineLeading());
423 fMaxY -= fLineDescent;
424 fMaxX = std::max(fLineWidth, fMaxX);
425 }
426 m_rcRet = CPVT_FloatRect(0, 0, fMaxX, fMaxY);
427 }
428
OutputLines()429 void CTypeset::OutputLines() {
430 ASSERT(m_pVT);
431 ASSERT(m_pSection);
432 float fMinX;
433 float fLineIndent = m_pVT->GetLineIndent();
434 float fTypesetWidth = std::max(m_pVT->GetPlateWidth() - fLineIndent, 0.0f);
435 switch (m_pVT->GetAlignment()) {
436 default:
437 case 0:
438 fMinX = 0.0f;
439 break;
440 case 1:
441 fMinX = (fTypesetWidth - m_rcRet.Width()) * VARIABLETEXT_HALF;
442 break;
443 case 2:
444 fMinX = fTypesetWidth - m_rcRet.Width();
445 break;
446 }
447 float fMaxX = fMinX + m_rcRet.Width();
448 float fMinY = 0.0f;
449 float fMaxY = m_rcRet.Height();
450 int32_t nTotalLines =
451 pdfium::CollectionSize<int32_t>(m_pSection->m_LineArray);
452 if (nTotalLines > 0) {
453 float fPosX = 0.0f;
454 float fPosY = 0.0f;
455 for (int32_t l = 0; l < nTotalLines; l++) {
456 CLine* pLine = m_pSection->m_LineArray[l].get();
457 switch (m_pVT->GetAlignment()) {
458 default:
459 case 0:
460 fPosX = 0;
461 break;
462 case 1:
463 fPosX = (fTypesetWidth - pLine->m_LineInfo.fLineWidth) *
464 VARIABLETEXT_HALF;
465 break;
466 case 2:
467 fPosX = fTypesetWidth - pLine->m_LineInfo.fLineWidth;
468 break;
469 }
470 fPosX += fLineIndent;
471 fPosY += m_pVT->GetLineLeading();
472 fPosY += pLine->m_LineInfo.fLineAscent;
473 pLine->m_LineInfo.fLineX = fPosX - fMinX;
474 pLine->m_LineInfo.fLineY = fPosY - fMinY;
475 for (int32_t w = pLine->m_LineInfo.nBeginWordIndex;
476 w <= pLine->m_LineInfo.nEndWordIndex; w++) {
477 if (pdfium::IndexInBounds(m_pSection->m_WordArray, w)) {
478 CPVT_WordInfo* pWord = m_pSection->m_WordArray[w].get();
479 pWord->fWordX = fPosX - fMinX;
480 pWord->fWordY = fPosY - fMinY;
481
482 fPosX += m_pVT->GetWordWidth(*pWord);
483 }
484 }
485 fPosY -= pLine->m_LineInfo.fLineDescent;
486 }
487 }
488 m_rcRet = CPVT_FloatRect(fMinX, fMinY, fMaxX, fMaxY);
489 }
490