int Document::CountWords(CPDF_TextObject* pTextObj)
{
if (!pTextObj) return 0;
int nWords = 0;
CPDF_Font* pFont = pTextObj->GetFont();
if (!pFont) return 0;
FX_BOOL bIsLatin = FALSE;
for (int i=0, sz=pTextObj->CountChars(); i<sz; i++)
{
FX_DWORD charcode = -1;
FX_FLOAT kerning;
pTextObj->GetCharInfo(i, charcode, kerning);
CFX_WideString swUnicode = pFont->UnicodeFromCharCode(charcode);
FX_WORD unicode = 0;
if (swUnicode.GetLength() > 0)
unicode = swUnicode[0];
if (ISLATINWORD(unicode) && bIsLatin)
continue;
bIsLatin = ISLATINWORD(unicode);
if (unicode != 0x20)
nWords++;
}
return nWords;
}
CFX_WideString Document::GetObjWordStr(CPDF_TextObject* pTextObj, int nWordIndex)
{
ASSERT(pTextObj != NULL);
CFX_WideString swRet;
CPDF_Font* pFont = pTextObj->GetFont();
if (!pFont) return L"";
int nWords = 0;
FX_BOOL bIsLatin = FALSE;
for (int i=0, sz=pTextObj->CountChars(); i<sz; i++)
{
FX_DWORD charcode = -1;
FX_FLOAT kerning;
pTextObj->GetCharInfo(i, charcode, kerning);
CFX_WideString swUnicode = pFont->UnicodeFromCharCode(charcode);
FX_WORD unicode = 0;
if (swUnicode.GetLength() > 0)
unicode = swUnicode[0];
if (ISLATINWORD(unicode) && bIsLatin)
{
}
else
{
bIsLatin = ISLATINWORD(unicode);
if (unicode != 0x20)
nWords++;
}
if (nWords-1 == nWordIndex)
swRet += unicode;
}
return swRet;
}
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);
}
FX_BOOL CPDF_TextStream::ProcessObject(const CPDF_TextObject* pObj, FX_BOOL bFirstLine)
{
CPDF_Font* pFont = pObj->GetFont();
CFX_AffineMatrix matrix;
pObj->GetTextMatrix(&matrix);
int item_index = 0;
if (m_pLastObj) {
int result = FPDFText_ProcessInterObj(m_pLastObj, pObj);
if (result == 2) {
int len = m_Buffer.GetLength();
if (len && m_bUseLF && m_Buffer.GetBuffer()[len - 1] == L'-') {
m_Buffer.Delete(len - 1, 1);
if (m_pObjArray) {
m_pObjArray->RemoveAt((len - 1) * 2, 2);
}
} else {
if (bFirstLine) {
return TRUE;
}
if (m_bUseLF) {
m_Buffer.AppendChar(L'\r');
m_Buffer.AppendChar(L'\n');
if (m_pObjArray) {
for (int i = 0; i < 4; i ++) {
m_pObjArray->Add(NULL);
}
}
} else {
m_Buffer.AppendChar(' ');
if (m_pObjArray) {
m_pObjArray->Add(NULL);
m_pObjArray->Add(NULL);
}
}
}
} else if (result == 1) {
m_Buffer.AppendChar(L' ');
if (m_pObjArray) {
m_pObjArray->Add(NULL);
m_pObjArray->Add(NULL);
}
} else if (result == -1) {
m_pLastObj = pObj;
return FALSE;
} else if (result == 3) {
item_index = 1;
}
}
m_pLastObj = pObj;
int nItems = pObj->CountItems();
FX_FLOAT Ignorekerning = 0;
for(int i = 1; i < nItems - 1; i += 2) {
CPDF_TextObjectItem item;
pObj->GetItemInfo(i, &item);
if (item.m_CharCode == (FX_DWORD) - 1) {
if(i == 1) {
Ignorekerning = item.m_OriginX;
} else if(Ignorekerning > item.m_OriginX) {
Ignorekerning = item.m_OriginX;
}
} else {
Ignorekerning = 0;
break;
}
}
FX_FLOAT spacing = 0;
for (; item_index < nItems; item_index ++) {
CPDF_TextObjectItem item;
pObj->GetItemInfo(item_index, &item);
if (item.m_CharCode == (FX_DWORD) - 1) {
CFX_WideString wstr = m_Buffer.GetWideString();
if (wstr.IsEmpty() || wstr.GetAt(wstr.GetLength() - 1) == L' ') {
continue;
}
FX_FLOAT fontsize_h = pObj->m_TextState.GetFontSizeH();
spacing = -fontsize_h * (item.m_OriginX - Ignorekerning) / 1000;
continue;
}
FX_FLOAT charSpace = pObj->m_TextState.GetObject()->m_CharSpace;
if(nItems > 3 && !spacing) {
charSpace = 0;
}
if((spacing || charSpace) && item_index > 0) {
int last_width = 0;
FX_FLOAT fontsize_h = pObj->m_TextState.GetFontSizeH();
FX_DWORD space_charcode = pFont->CharCodeFromUnicode(' ');
FX_FLOAT threshold = 0;
if (space_charcode != -1) {
threshold = fontsize_h * pFont->GetCharWidthF(space_charcode) / 1000 ;
}
if(threshold > fontsize_h / 3) {
threshold = 0;
} else {
threshold /= 2;
}
if (threshold == 0) {
threshold = fontsize_h;
int this_width = FXSYS_abs(GetCharWidth(item.m_CharCode, pFont));
threshold = this_width > last_width ? (FX_FLOAT)this_width : (FX_FLOAT)last_width;
int nDivide = 6;
if (threshold < 300) {
nDivide = 2;
} else if (threshold < 500) {
nDivide = 4;
} else if (threshold < 700) {
nDivide = 5;
}
threshold = threshold / nDivide;
threshold = fontsize_h * threshold / 1000;
}
if(charSpace > 0.001) {
spacing += matrix.TransformDistance(charSpace);
} else if(charSpace < -0.001) {
spacing -= matrix.TransformDistance(FXSYS_fabs(charSpace));
}
if (threshold && (spacing && spacing >= threshold) ) {
m_Buffer.AppendChar(L' ');
if (m_pObjArray) {
m_pObjArray->Add(NULL);
m_pObjArray->Add(NULL);
}
}
if (item.m_CharCode == (FX_DWORD) - 1) {
continue;
}
spacing = 0;
}
CFX_WideString unicode_str = pFont->UnicodeFromCharCode(item.m_CharCode);
if (unicode_str.IsEmpty()) {
m_Buffer.AppendChar((FX_WCHAR)item.m_CharCode);
if (m_pObjArray) {
m_pObjArray->Add((void*)pObj);
m_pObjArray->Add((void*)(FX_INTPTR)item_index);
}
} else {
m_Buffer << unicode_str;
if (m_pObjArray) {
for (int i = 0; i < unicode_str.GetLength(); i ++) {
m_pObjArray->Add((void*)pObj);
m_pObjArray->Add((void*)(FX_INTPTR)item_index);
}
}
}
}
return FALSE;
}