void PDF_GetPageText_Unicode(CFX_WideStringArray& lines, CPDF_Document* pDoc, CPDF_Dictionary* pPage,
int iMinWidth, FX_DWORD flags)
{
lines.RemoveAll();
if (pPage == NULL) {
return;
}
CPDF_Page page;
page.Load(pDoc, pPage);
CPDF_ParseOptions options;
options.m_bTextOnly = TRUE;
options.m_bSeparateForm = FALSE;
page.ParseContent(&options);
CFX_FloatRect page_bbox = page.GetPageBBox();
if (flags & PDF2TXT_AUTO_ROTATE) {
CheckRotate(page, page_bbox);
}
CTextPage texts;
texts.m_bAutoWidth = flags & PDF2TXT_AUTO_WIDTH;
texts.m_bKeepColumn = flags & PDF2TXT_KEEP_COLUMN;
texts.m_bBreakSpace = TRUE;
FX_POSITION pos = page.GetFirstObjectPosition();
while (pos) {
CPDF_PageObject* pObject = page.GetNextObject(pos);
if (!(flags & PDF2TXT_INCLUDE_INVISIBLE)) {
CFX_FloatRect rect(pObject->m_Left, pObject->m_Bottom, pObject->m_Right, pObject->m_Top);
if (!page_bbox.Contains(rect)) {
continue;
}
}
texts.ProcessObject(pObject);
}
texts.WriteOutput(lines, iMinWidth);
}
FX_BOOL GetContentsRect( CPDF_Document * pDoc, CPDF_Dictionary* pDict, CPDF_RectArray * pRectArray )
{
CPDF_Page* pPDFPage = FX_NEW CPDF_Page;
pPDFPage->Load( pDoc, pDict, FALSE );
pPDFPage->ParseContent();
FX_POSITION pos = pPDFPage->GetFirstObjectPosition();
while (pos)
{
CPDF_PageObject* pPageObject = pPDFPage->GetNextObject(pos);
if (!pPageObject)continue;
CPDF_Rect rc;
rc.left = pPageObject->m_Left;
rc.right = pPageObject->m_Right;
rc.bottom = pPageObject->m_Bottom;
rc.top = pPageObject->m_Top;
if (IsValiableRect(rc, pDict->GetRect("MediaBox")))
{
pRectArray->Add(rc);
}
}
delete pPDFPage;
return TRUE;
}
FX_BOOL Document::getPageNumWords(IFXJS_Context* cc, const CJS_Parameters& params, CJS_Value& vRet, CFX_WideString& sError)
{
ASSERT(m_pDocument != NULL);
if (!m_pDocument->GetPermissions(FPDFPERM_EXTRACT_ACCESS)) return FALSE;
int nPageNo = params.GetSize() > 0 ? params[0].ToInt() : 0;
CPDF_Document* pDocument = m_pDocument->GetDocument();
ASSERT(pDocument != NULL);
CJS_Context* pContext = static_cast<CJS_Context*>(cc);
if (nPageNo < 0 || nPageNo >= pDocument->GetPageCount())
{
sError = JSGetStringFromID(pContext, IDS_STRING_JSVALUEERROR);
return FALSE;
}
CPDF_Dictionary* pPageDict = pDocument->GetPage(nPageNo);
if (!pPageDict) return FALSE;
CPDF_Page page;
page.Load(pDocument, pPageDict);
page.StartParse();
page.ParseContent();
FX_POSITION pos = page.GetFirstObjectPosition();
int nWords = 0;
while (pos)
{
if (CPDF_PageObject* pPageObj = page.GetNextObject(pos))
{
if (pPageObj->m_Type == PDFPAGE_TEXT)
{
CPDF_TextObject* pTextObj = (CPDF_TextObject*)pPageObj;
nWords += CountWords(pTextObj);
}
}
}
vRet = nWords;
return TRUE;
}
static void CheckRotate(CPDF_Page& page, CFX_FloatRect& page_bbox)
{
int total_count = 0, rotated_count[3] = {0, 0, 0};
FX_POSITION pos = page.GetFirstObjectPosition();
while (pos) {
CPDF_PageObject* pObj = page.GetNextObject(pos);
if (pObj->m_Type != PDFPAGE_TEXT) {
continue;
}
total_count ++;
CPDF_TextObject* pText = (CPDF_TextObject*)pObj;
FX_FLOAT angle = pText->m_TextState.GetBaselineAngle();
if (angle == 0.0) {
continue;
}
int degree = (int)(angle * 180 / PI + 0.5);
if (degree % 90) {
continue;
}
if (degree < 0) {
degree += 360;
}
int index = degree / 90 % 3 - 1;
if (index < 0) {
continue;
}
rotated_count[index] ++;
}
if (total_count == 0) {
return;
}
CFX_AffineMatrix matrix;
if (rotated_count[0] > total_count * 2 / 3) {
matrix.Set(0, -1, 1, 0, 0, page.GetPageHeight());
} else if (rotated_count[1] > total_count * 2 / 3) {
matrix.Set(-1, 0, 0, -1, page.GetPageWidth(), page.GetPageHeight());
} else if (rotated_count[2] > total_count * 2 / 3) {
matrix.Set(0, 1, -1, 0, page.GetPageWidth(), 0);
} else {
return;
}
page.Transform(matrix);
page_bbox.Transform(&matrix);
}
CFX_WideString PDF_GetFirstTextLine_Unicode(CPDF_Document* pDoc, CPDF_Dictionary* pPage)
{
CFX_WideTextBuf buffer;
buffer.EstimateSize(0, 1024);
CPDF_Page page;
page.Load(pDoc, pPage);
CPDF_ParseOptions options;
options.m_bTextOnly = TRUE;
options.m_bSeparateForm = FALSE;
page.ParseContent(&options);
CPDF_TextStream textstream(buffer, FALSE, NULL);
FX_POSITION pos = page.GetFirstObjectPosition();
while (pos) {
CPDF_PageObject* pObject = page.GetNextObject(pos);
if (pObject->m_Type != PDFPAGE_TEXT) {
continue;
}
if (textstream.ProcessObject((CPDF_TextObject*)pObject, TRUE)) {
break;
}
}
return buffer.GetWideString();
}
FX_BOOL Document::getPageNthWord(IFXJS_Context* cc, const CJS_Parameters& params, CJS_Value& vRet, CFX_WideString& sError)
{
ASSERT(m_pDocument != NULL);
if (!m_pDocument->GetPermissions(FPDFPERM_EXTRACT_ACCESS)) return FALSE;
int nPageNo = params.GetSize() > 0 ? params[0].ToInt() : 0;
int nWordNo = params.GetSize() > 1 ? params[1].ToInt() : 0;
bool bStrip = params.GetSize() > 2 ? params[2].ToBool() : true;
CPDF_Document* pDocument = m_pDocument->GetDocument();
if (!pDocument) return FALSE;
CJS_Context* pContext = static_cast<CJS_Context*>(cc);
if (nPageNo < 0 || nPageNo >= pDocument->GetPageCount())
{
sError = JSGetStringFromID(pContext, IDS_STRING_JSVALUEERROR);
return FALSE;
}
CPDF_Dictionary* pPageDict = pDocument->GetPage(nPageNo);
if (!pPageDict) return FALSE;
CPDF_Page page;
page.Load(pDocument, pPageDict);
page.StartParse();
page.ParseContent();
FX_POSITION pos = page.GetFirstObjectPosition();
int nWords = 0;
CFX_WideString swRet;
while (pos)
{
if (CPDF_PageObject* pPageObj = page.GetNextObject(pos))
{
if (pPageObj->m_Type == PDFPAGE_TEXT)
{
int nObjWords = CountWords((CPDF_TextObject*)pPageObj);
if (nWords + nObjWords >= nWordNo)
{
swRet = GetObjWordStr((CPDF_TextObject*)pPageObj, nWordNo - nWords);
break;
}
nWords += nObjWords;
}
}
}
if (bStrip)
{
swRet.TrimLeft();
swRet.TrimRight();
}
vRet = swRet.c_str();
return TRUE;
}