FPDFText_MarkedContent CPDF_TextPage::PreMarkedContent(PDFTEXT_Obj Obj) {
CPDF_TextObject* pTextObj = Obj.m_pTextObj.Get();
size_t nContentMarks = pTextObj->m_ContentMarks.CountItems();
if (nContentMarks == 0)
return FPDFText_MarkedContent::Pass;
WideString actText;
bool bExist = false;
const CPDF_Dictionary* pDict = nullptr;
for (size_t i = 0; i < nContentMarks; ++i) {
const CPDF_ContentMarkItem* item = pTextObj->m_ContentMarks.GetItem(i);
pDict = item->GetParam();
if (!pDict)
continue;
const CPDF_String* temp = ToString(pDict->GetObjectFor("ActualText"));
if (temp) {
bExist = true;
actText = temp->GetUnicodeText();
}
}
if (!bExist)
return FPDFText_MarkedContent::Pass;
if (m_pPreTextObj) {
const CPDF_ContentMarks& marks = m_pPreTextObj->m_ContentMarks;
if (marks.CountItems() == nContentMarks &&
marks.GetItem(nContentMarks - 1)->GetParam() == pDict) {
return FPDFText_MarkedContent::Done;
}
}
if (actText.IsEmpty())
return FPDFText_MarkedContent::Pass;
CPDF_Font* pFont = pTextObj->GetFont();
bExist = false;
for (size_t i = 0; i < actText.GetLength(); ++i) {
if (pFont->CharCodeFromUnicode(actText[i]) != CPDF_Font::kInvalidCharCode) {
bExist = true;
break;
}
}
if (!bExist)
return FPDFText_MarkedContent::Pass;
bExist = false;
for (size_t i = 0; i < actText.GetLength(); ++i) {
wchar_t wChar = actText[i];
if ((wChar > 0x80 && wChar < 0xFFFD) || (wChar <= 0x80 && isprint(wChar))) {
bExist = true;
break;
}
}
if (!bExist)
return FPDFText_MarkedContent::Done;
return FPDFText_MarkedContent::Delay;
}
WideString CPDF_FileSpec::DecodeFileName(const WideString& filepath) {
if (filepath.GetLength() <= 1)
return WideString();
#if _FX_PLATFORM_ == _FX_PLATFORM_APPLE_
if (filepath.Left(sizeof("/Mac") - 1) == WideStringView(L"/Mac"))
return ChangeSlashToPlatform(filepath.c_str() + 1);
return ChangeSlashToPlatform(filepath.c_str());
#elif _FX_PLATFORM_ == _FX_PLATFORM_WINDOWS_
if (filepath[0] != L'/')
return ChangeSlashToPlatform(filepath.c_str());
if (filepath[1] == L'/')
return ChangeSlashToPlatform(filepath.c_str() + 1);
if (filepath[2] == L'/') {
WideString result;
result += filepath[1];
result += L':';
result += ChangeSlashToPlatform(filepath.c_str() + 2);
return result;
}
WideString result;
result += L'\\';
result += ChangeSlashToPlatform(filepath.c_str());
return result;
#else
return WideString(filepath);
#endif
}
WideString CPDF_FileSpec::EncodeFileName(const WideString& filepath) {
if (filepath.GetLength() <= 1)
return WideString();
#if _FX_PLATFORM_ == _FX_PLATFORM_WINDOWS_
if (filepath[1] == L':') {
WideString result(L'/');
result += filepath[0];
if (filepath[2] != L'\\')
result += L'/';
result += ChangeSlashToPDF(filepath.c_str() + 2);
return result;
}
if (filepath[0] == L'\\' && filepath[1] == L'\\')
return ChangeSlashToPDF(filepath.c_str() + 1);
if (filepath[0] == L'\\')
return L'/' + ChangeSlashToPDF(filepath.c_str());
return ChangeSlashToPDF(filepath.c_str());
#elif _FX_PLATFORM_ == _FX_PLATFORM_APPLE_
if (filepath.Left(sizeof("Mac") - 1).EqualsASCII("Mac"))
return L'/' + ChangeSlashToPDF(filepath.c_str());
return ChangeSlashToPDF(filepath.c_str());
#else
return WideString(filepath);
#endif
}
bool CPWL_ListBox::OnNotifySelectionChanged(bool bKeyDown, uint32_t nFlag) {
if (!m_pFillerNotify)
return false;
CPWL_Wnd::ObservedPtr thisObserved(this);
WideString swChange = GetText();
WideString strChangeEx;
int nSelStart = 0;
int nSelEnd = swChange.GetLength();
bool bRC;
bool bExit;
std::tie(bRC, bExit) = m_pFillerNotify->OnBeforeKeyStroke(
GetAttachedData(), swChange, strChangeEx, nSelStart, nSelEnd, bKeyDown,
nFlag);
if (!thisObserved)
return false;
return bExit;
}
void CPDF_TextPage::ProcessMarkedContent(PDFTEXT_Obj Obj) {
CPDF_TextObject* pTextObj = Obj.m_pTextObj.Get();
size_t nContentMarks = pTextObj->m_ContentMarks.CountItems();
if (nContentMarks == 0)
return;
WideString actText;
for (size_t n = 0; n < nContentMarks; ++n) {
const CPDF_ContentMarkItem* item = pTextObj->m_ContentMarks.GetItem(n);
const CPDF_Dictionary* pDict = item->GetParam();
if (pDict)
actText = pDict->GetUnicodeTextFor("ActualText");
}
if (actText.IsEmpty())
return;
CPDF_Font* pFont = pTextObj->GetFont();
CFX_Matrix matrix = pTextObj->GetTextMatrix() * Obj.m_formMatrix;
for (size_t k = 0; k < actText.GetLength(); ++k) {
wchar_t wChar = actText[k];
if (wChar <= 0x80 && !isprint(wChar))
wChar = 0x20;
if (wChar >= 0xFFFD)
continue;
PAGECHAR_INFO charinfo;
charinfo.m_Origin = pTextObj->GetPos();
charinfo.m_Index = m_TextBuf.GetLength();
charinfo.m_Unicode = wChar;
charinfo.m_CharCode = pFont->CharCodeFromUnicode(wChar);
charinfo.m_Flag = FPDFTEXT_CHAR_PIECE;
charinfo.m_pTextObj = pTextObj;
charinfo.m_CharBox = pTextObj->GetRect();
charinfo.m_Matrix = matrix;
m_TempTextBuf.AppendChar(wChar);
m_TempCharList.push_back(charinfo);
}
}
void CPDF_TextPage::CloseTempLine() {
if (m_TempCharList.empty())
return;
WideString str = m_TempTextBuf.MakeString();
bool bPrevSpace = false;
for (size_t i = 0; i < str.GetLength(); ++i) {
if (str[i] != ' ') {
bPrevSpace = false;
continue;
}
if (bPrevSpace) {
m_TempTextBuf.Delete(i, 1);
m_TempCharList.erase(m_TempCharList.begin() + i);
str.Delete(i);
--i;
}
bPrevSpace = true;
}
CFX_BidiString bidi(str);
if (m_parserflag == FPDFText_Direction::Right)
bidi.SetOverallDirectionRight();
CFX_BidiChar::Direction eCurrentDirection = bidi.OverallDirection();
for (const auto& segment : bidi) {
if (segment.direction == CFX_BidiChar::RIGHT ||
(segment.direction == CFX_BidiChar::NEUTRAL &&
eCurrentDirection == CFX_BidiChar::RIGHT)) {
eCurrentDirection = CFX_BidiChar::RIGHT;
for (int m = segment.start + segment.count; m > segment.start; --m)
AddCharInfoByRLDirection(str[m - 1], m_TempCharList[m - 1]);
} else {
eCurrentDirection = CFX_BidiChar::LEFT;
for (int m = segment.start; m < segment.start + segment.count; ++m)
AddCharInfoByLRDirection(str[m], m_TempCharList[m]);
}
}
m_TempCharList.clear();
m_TempTextBuf.Delete(0, m_TempTextBuf.GetLength());
}
void CPDF_TextPage::ProcessTextObject(PDFTEXT_Obj Obj) {
CPDF_TextObject* pTextObj = Obj.m_pTextObj.Get();
if (fabs(pTextObj->GetRect().Width()) < kSizeEpsilon)
return;
CFX_Matrix formMatrix = Obj.m_formMatrix;
CPDF_Font* pFont = pTextObj->GetFont();
CFX_Matrix matrix = pTextObj->GetTextMatrix() * formMatrix;
FPDFText_MarkedContent ePreMKC = PreMarkedContent(Obj);
if (ePreMKC == FPDFText_MarkedContent::Done) {
m_pPreTextObj = pTextObj;
m_perMatrix = formMatrix;
return;
}
GenerateCharacter result = GenerateCharacter::None;
if (m_pPreTextObj) {
result = ProcessInsertObject(pTextObj, formMatrix);
if (result == GenerateCharacter::LineBreak)
m_CurlineRect = Obj.m_pTextObj->GetRect();
else
m_CurlineRect.Union(Obj.m_pTextObj->GetRect());
switch (result) {
case GenerateCharacter::None:
break;
case GenerateCharacter::Space: {
Optional<PAGECHAR_INFO> pGenerateChar =
GenerateCharInfo(TEXT_SPACE_CHAR);
if (pGenerateChar) {
if (!formMatrix.IsIdentity())
pGenerateChar->m_Matrix = formMatrix;
m_TempTextBuf.AppendChar(TEXT_SPACE_CHAR);
m_TempCharList.push_back(*pGenerateChar);
}
break;
}
case GenerateCharacter::LineBreak:
CloseTempLine();
if (m_TextBuf.GetSize()) {
AppendGeneratedCharacter(TEXT_RETURN_CHAR, formMatrix);
AppendGeneratedCharacter(TEXT_LINEFEED_CHAR, formMatrix);
}
break;
case GenerateCharacter::Hyphen:
if (pTextObj->CountChars() == 1) {
CPDF_TextObjectItem item;
pTextObj->GetCharInfo(0, &item);
WideString wstrItem =
pTextObj->GetFont()->UnicodeFromCharCode(item.m_CharCode);
if (wstrItem.IsEmpty())
wstrItem += (wchar_t)item.m_CharCode;
wchar_t curChar = wstrItem[0];
if (IsHyphenCode(curChar))
return;
}
while (m_TempTextBuf.GetSize() > 0 &&
m_TempTextBuf.AsStringView()[m_TempTextBuf.GetLength() - 1] ==
0x20) {
m_TempTextBuf.Delete(m_TempTextBuf.GetLength() - 1, 1);
m_TempCharList.pop_back();
}
PAGECHAR_INFO* charinfo = &m_TempCharList.back();
m_TempTextBuf.Delete(m_TempTextBuf.GetLength() - 1, 1);
charinfo->m_Unicode = 0x2;
charinfo->m_Flag = FPDFTEXT_CHAR_HYPHEN;
m_TempTextBuf.AppendChar(0xfffe);
break;
}
} else {
m_CurlineRect = Obj.m_pTextObj->GetRect();
}
if (ePreMKC == FPDFText_MarkedContent::Delay) {
ProcessMarkedContent(Obj);
m_pPreTextObj = pTextObj;
m_perMatrix = formMatrix;
return;
}
m_pPreTextObj = pTextObj;
m_perMatrix = formMatrix;
float baseSpace = CalculateBaseSpace(pTextObj, matrix);
const bool bR2L = IsRightToLeft(*pTextObj, *pFont);
const bool bIsBidiAndMirrorInverse =
bR2L && (matrix.a * matrix.d - matrix.b * matrix.c) < 0;
int32_t iBufStartAppend = m_TempTextBuf.GetLength();
int32_t iCharListStartAppend =
pdfium::CollectionSize<int32_t>(m_TempCharList);
float spacing = 0;
const size_t nItems = pTextObj->CountItems();
for (size_t i = 0; i < nItems; ++i) {
CPDF_TextObjectItem item;
PAGECHAR_INFO charinfo;
pTextObj->GetItemInfo(i, &item);
if (item.m_CharCode == static_cast<uint32_t>(-1)) {
WideString str = m_TempTextBuf.MakeString();
if (str.IsEmpty())
str = m_TextBuf.AsStringView();
if (str.IsEmpty() || str[str.GetLength() - 1] == TEXT_SPACE_CHAR)
continue;
float fontsize_h = pTextObj->m_TextState.GetFontSizeH();
spacing = -fontsize_h * item.m_Origin.x / 1000;
continue;
}
float charSpace = pTextObj->m_TextState.GetCharSpace();
if (charSpace > 0.001)
spacing += matrix.TransformDistance(charSpace);
else if (charSpace < -0.001)
spacing -= matrix.TransformDistance(fabs(charSpace));
spacing -= baseSpace;
if (spacing && i > 0) {
float fontsize_h = pTextObj->m_TextState.GetFontSizeH();
uint32_t space_charcode = pFont->CharCodeFromUnicode(' ');
float threshold = 0;
if (space_charcode != CPDF_Font::kInvalidCharCode)
threshold = fontsize_h * pFont->GetCharWidthF(space_charcode) / 1000;
if (threshold > fontsize_h / 3)
threshold = 0;
else
threshold /= 2;
if (threshold == 0) {
threshold = static_cast<float>(GetCharWidth(item.m_CharCode, pFont));
threshold = NormalizeThreshold(threshold, 300, 500, 700);
threshold = fontsize_h * threshold / 1000;
}
if (threshold && (spacing && spacing >= threshold)) {
charinfo.m_Unicode = TEXT_SPACE_CHAR;
charinfo.m_Flag = FPDFTEXT_CHAR_GENERATED;
charinfo.m_pTextObj = pTextObj;
charinfo.m_Index = m_TextBuf.GetLength();
m_TempTextBuf.AppendChar(TEXT_SPACE_CHAR);
charinfo.m_CharCode = CPDF_Font::kInvalidCharCode;
charinfo.m_Matrix = formMatrix;
charinfo.m_Origin = matrix.Transform(item.m_Origin);
charinfo.m_CharBox =
CFX_FloatRect(charinfo.m_Origin.x, charinfo.m_Origin.y,
charinfo.m_Origin.x, charinfo.m_Origin.y);
m_TempCharList.push_back(charinfo);
}
if (item.m_CharCode == CPDF_Font::kInvalidCharCode)
continue;
}
spacing = 0;
WideString wstrItem = pFont->UnicodeFromCharCode(item.m_CharCode);
bool bNoUnicode = false;
if (wstrItem.IsEmpty() && item.m_CharCode) {
wstrItem += static_cast<wchar_t>(item.m_CharCode);
bNoUnicode = true;
}
charinfo.m_Index = -1;
charinfo.m_CharCode = item.m_CharCode;
charinfo.m_Flag =
bNoUnicode ? FPDFTEXT_CHAR_UNUNICODE : FPDFTEXT_CHAR_NORMAL;
charinfo.m_pTextObj = pTextObj;
charinfo.m_Origin = matrix.Transform(item.m_Origin);
const FX_RECT rect =
charinfo.m_pTextObj->GetFont()->GetCharBBox(charinfo.m_CharCode);
const float fFontSize = pTextObj->GetFontSize() / 1000;
charinfo.m_CharBox.top = rect.top * fFontSize + item.m_Origin.y;
charinfo.m_CharBox.left = rect.left * fFontSize + item.m_Origin.x;
charinfo.m_CharBox.right = rect.right * fFontSize + item.m_Origin.x;
charinfo.m_CharBox.bottom = rect.bottom * fFontSize + item.m_Origin.y;
if (fabsf(charinfo.m_CharBox.top - charinfo.m_CharBox.bottom) <
kSizeEpsilon) {
charinfo.m_CharBox.top =
charinfo.m_CharBox.bottom + pTextObj->GetFontSize();
}
if (fabsf(charinfo.m_CharBox.right - charinfo.m_CharBox.left) <
kSizeEpsilon) {
charinfo.m_CharBox.right =
charinfo.m_CharBox.left + pTextObj->GetCharWidth(charinfo.m_CharCode);
}
charinfo.m_CharBox = matrix.TransformRect(charinfo.m_CharBox);
charinfo.m_Matrix = matrix;
if (wstrItem.IsEmpty()) {
charinfo.m_Unicode = 0;
m_TempCharList.push_back(charinfo);
m_TempTextBuf.AppendChar(0xfffe);
continue;
}
int nTotal = wstrItem.GetLength();
bool bDel = false;
const int count = std::min(pdfium::CollectionSize<int>(m_TempCharList), 7);
float threshold = charinfo.m_Matrix.TransformXDistance(
static_cast<float>(TEXT_CHARRATIO_GAPDELTA) * pTextObj->GetFontSize());
for (int n = pdfium::CollectionSize<int>(m_TempCharList);
n > pdfium::CollectionSize<int>(m_TempCharList) - count; --n) {
const PAGECHAR_INFO& charinfo1 = m_TempCharList[n - 1];
CFX_PointF diff = charinfo1.m_Origin - charinfo.m_Origin;
if (charinfo1.m_CharCode == charinfo.m_CharCode &&
charinfo1.m_pTextObj->GetFont() == charinfo.m_pTextObj->GetFont() &&
fabs(diff.x) < threshold && fabs(diff.y) < threshold) {
bDel = true;
break;
}
}
if (!bDel) {
for (int nIndex = 0; nIndex < nTotal; ++nIndex) {
charinfo.m_Unicode = wstrItem[nIndex];
if (charinfo.m_Unicode) {
charinfo.m_Index = m_TextBuf.GetLength();
m_TempTextBuf.AppendChar(charinfo.m_Unicode);
} else {
m_TempTextBuf.AppendChar(0xfffe);
}
m_TempCharList.push_back(charinfo);
}
} else if (i == 0) {
WideString str = m_TempTextBuf.MakeString();
if (!str.IsEmpty() && str[str.GetLength() - 1] == TEXT_SPACE_CHAR) {
m_TempTextBuf.Delete(m_TempTextBuf.GetLength() - 1, 1);
m_TempCharList.pop_back();
}
}
}
if (bIsBidiAndMirrorInverse)
SwapTempTextBuf(iCharListStartAppend, iBufStartAppend);
}
