virtual QImage scaleAndRender(const Poppler::Page *page, const QWidget *widget) {
QList<Poppler::TextBox *> textBoxen = page->textList();
QRectF bounds;
while (!textBoxen.isEmpty()) {
Poppler::TextBox *box = textBoxen.takeFirst();
bounds = bounds.united(box->boundingBox());
delete box;
}
bounds.adjust(-4, -4, 8, 8);
qreal win = bounds.width() / 72., hin = bounds.height() / 72.;
qreal xsc = widget->width() / win, ysc = widget->height() / hin;
qreal scale = qMin(xsc, ysc);
QRectF viewRect(bounds.left() / 72. * scale, bounds.top() / 72. * scale,
bounds.width() / 72. * scale, bounds.height() / 72. * scale);
QRect iRect = viewRect.toAlignedRect();
return page->renderToImage(scale, scale,
iRect.left(), iRect.top(), iRect.width(), iRect.height());
}
void Worker::run() {
while (1) {
res->requestSemaphore.acquire(1);
if (die) {
break;
}
// get next page to render
res->requestMutex.lock();
int page, width, index;
map<int,pair<int,int> >::iterator less = res->requests.lower_bound(res->center_page);
map<int,pair<int,int> >::iterator greater = less--;
if (greater != res->requests.end()) {
if (greater != res->requests.begin()) {
// favour nearby page, go down first
if (greater->first + less->first <= res->center_page * 2) {
page = greater->first;
index = greater->second.first;
width = greater->second.second;
res->requests.erase(greater);
} else {
page = less->first;
index = less->second.first;
width = less->second.second;
res->requests.erase(less);
}
} else {
page = greater->first;
index = greater->second.first;
width = greater->second.second;
res->requests.erase(greater);
}
} else {
page = less->first;
index = less->second.first;
width = less->second.second;
res->requests.erase(less);
}
res->requestMutex.unlock();
// check for duplicate requests
res->k_page[page].mutex.lock();
if (res->k_page[page].status[index] == width &&
res->k_page[page].rotation[index] == res->rotation) {
res->k_page[page].mutex.unlock();
continue;
}
int rotation = res->rotation;
res->k_page[page].mutex.unlock();
// open page
#ifdef DEBUG
cerr << " rendering page " << page << " for index " << index << endl;
#endif
Poppler::Page *p = res->doc->page(page);
if (p == NULL) {
cerr << "failed to load page " << page << endl;
continue;
}
// render page
float dpi = 72.0 * width / res->get_page_width(page);
QImage img = p->renderToImage(dpi, dpi, -1, -1, -1, -1,
static_cast<Poppler::Page::Rotation>(rotation));
if (img.isNull()) {
cerr << "failed to render page " << page << endl;
continue;
}
// invert to current color setting
if (res->inverted_colors) {
img.invertPixels();
}
// put page
res->k_page[page].mutex.lock();
if (!res->k_page[page].img[index].isNull()) {
res->k_page[page].img[index] = QImage(); // assign null image
}
// adjust all available images to current color setting
if (res->k_page[page].inverted_colors != res->inverted_colors) {
res->k_page[page].inverted_colors = res->inverted_colors;
for (int i = 0; i < 3; i++) {
res->k_page[page].img[i].invertPixels();
}
res->k_page[page].thumbnail.invertPixels();
}
res->k_page[page].img[index] = img;
res->k_page[page].status[index] = width;
res->k_page[page].rotation[index] = rotation;
res->k_page[page].mutex.unlock();
res->garbageMutex.lock();
res->garbage.insert(page); // TODO add index information?
res->garbageMutex.unlock();
emit page_rendered(page);
// collect goto links
res->link_mutex.lock();
if (res->k_page[page].links == NULL) {
res->link_mutex.unlock();
QList<Poppler::Link *> *links = new QList<Poppler::Link *>;
QList<Poppler::Link *> l = p->links();
links->swap(l);
res->link_mutex.lock();
res->k_page[page].links = links;
}
if (res->k_page[page].text == NULL) {
res->link_mutex.unlock();
QList<Poppler::TextBox *> text = p->textList();
// assign boxes to lines
// make single parts from chained boxes
set<Poppler::TextBox *> used;
QList<SelectionPart *> selection_parts;
Q_FOREACH(Poppler::TextBox *box, text) {
if (used.find(box) != used.end()) {
continue;
}
used.insert(box);
SelectionPart *p = new SelectionPart(box);
selection_parts.push_back(p);
Poppler::TextBox *next = box->nextWord();
while (next != NULL) {
used.insert(next);
p->add_word(next);
next = next->nextWord();
}
}
// sort by y coordinate
qStableSort(selection_parts.begin(), selection_parts.end(), selection_less_y);
QRectF line_box;
QList<SelectionLine *> *lines = new QList<SelectionLine *>();
Q_FOREACH(SelectionPart *part, selection_parts) {
QRectF box = part->get_bbox();
// box fits into line_box's line
if (!lines->empty() && box.y() <= line_box.center().y() && box.bottom() > line_box.center().y()) {
float ratio_w = box.width() / line_box.width();
float ratio_h = box.height() / line_box.height();
if (ratio_w < 1.0f) {
ratio_w = 1.0f / ratio_w;
}
if (ratio_h < 1.0f) {
ratio_h = 1.0f / ratio_h;
}
if (ratio_w > 1.3f && ratio_h > 1.3f) {
lines->back()->sort();
lines->push_back(new SelectionLine(part));
line_box = part->get_bbox();
} else {
lines->back()->add_part(part);
}
// it doesn't fit, create new line
} else {
if (!lines->empty()) {
lines->back()->sort();
}
lines->push_back(new SelectionLine(part));
line_box = part->get_bbox();
}
}
if (!lines->empty()) {
lines->back()->sort();
}
res->link_mutex.lock();
res->k_page[page].text = lines;
}
QString PopplerExtractor::parseFirstPage(Poppler::Document* pdfDoc, const QString& fileUrl)
{
QScopedPointer<Poppler::Page> p(pdfDoc->page(0));
if (!p) {
qWarning() << "Could not read page content from" << fileUrl;
return QString();
}
QList<Poppler::TextBox*> tbList = p->textList();
QMap<int, QString> possibleTitleMap;
int currentLargestChar = 0;
int skipTextboxes = 0;
// Iterate over all textboxes. Each textbox can be a single character/word or textblock
// Here we combine the etxtboxes back together based on the textsize
// Important are the words with the biggest font size
foreach(Poppler::TextBox * tb, tbList) {
// if we added followup words, skip the textboxes here now
if (skipTextboxes > 0) {
skipTextboxes--;
continue;
}
int height = tb->charBoundingBox(0).height();
// if the following text is smaller than the biggest we found up to now, ignore it
if (height >= currentLargestChar) {
QString possibleTitle;
possibleTitle.append(tb->text());
currentLargestChar = height;
// if the text has follow up words add them to to create the full title
Poppler::TextBox* next = tb->nextWord();
while (next) {
possibleTitle.append(QLatin1Char(' '));
possibleTitle.append(next->text());
next = next->nextWord();
skipTextboxes++;
}
// now combine text for each font size together, very likeley it must be connected
QString existingTitlePart = possibleTitleMap.value(currentLargestChar, QString());
existingTitlePart.append(QLatin1Char(' '));
existingTitlePart.append(possibleTitle);
possibleTitleMap.insert(currentLargestChar, existingTitlePart);
}
}
qDeleteAll(tbList);
QList<int> titleSizes = possibleTitleMap.keys();
qSort(titleSizes.begin(), titleSizes.end(), qGreater<int>());
QString newPossibleTitle;
// find the text with the largest font that is not just 1 character
foreach(int i, titleSizes) {
QString title = possibleTitleMap.value(i);
// sometime the biggest part is a single letter
// as a starting paragraph letter
if (title.size() < 5) {
continue;
} else {
newPossibleTitle = title.trimmed();
break;
}
}