7#include "config-kitinerary.h"
8#include "pdfdocument.h"
9#include "pdfdocument_p.h"
10#include "pdfextractoroutputdevice_p.h"
11#include "pdfimage_p.h"
12#include "popplerglobalparams_p.h"
13#include "popplerutils_p.h"
18#include <QScopedValueRollback>
23#include <PDFDocEncoding.h>
30void PdfPagePrivate::load()
36 PopplerGlobalParams gp;
37 PdfExtractorOutputDevice device;
38 m_doc->m_popplerDoc->displayPageSlice(&device, m_pageNum + 1, 72, 72, 0,
false,
true,
false, -1, -1, -1, -1);
39 m_doc->m_popplerDoc->processLinks(&device, m_pageNum + 1);
41 const auto pageRect = m_doc->m_popplerDoc->getPage(m_pageNum + 1)->getCropBox();
42 std::unique_ptr<GooString> s(device.getText(pageRect->x1, pageRect->y1, pageRect->x2, pageRect->y2));
44#if KPOPPLER_VERSION >= QT_VERSION_CHECK(0, 72, 0)
49 m_images = std::move(device.m_images);
50 for (
auto it = m_images.begin(); it != m_images.end(); ++it) {
51 (*it).d->m_page =
this;
54 m_links = std::move(device.m_links);
55 for (
auto &link : m_links) {
56 link.convertToPageRect(pageRect);
63 : d(new PdfPagePrivate)
67PdfPage::PdfPage(
const PdfPage&) =
default;
68PdfPage::~PdfPage() =
default;
77static double ratio(
double begin,
double end,
double ratio)
79 return begin + (end - begin) * ratio;
84 PopplerGlobalParams gp;
86 const auto page = d->m_doc->m_popplerDoc->getPage(d->m_pageNum + 1);
87 const auto pageRect = page->getCropBox();
93 switch (page->getRotate()) {
95 l = ratio(pageRect->x1, pageRect->x2, left);
96 t = ratio(pageRect->y1, pageRect->y2, top);
97 r = ratio(pageRect->x1, pageRect->x2, right);
98 b = ratio(pageRect->y1, pageRect->y2, bottom);
101 l = ratio(pageRect->y1, pageRect->y2, left);
102 t = ratio(pageRect->x1, pageRect->x2, top);
103 r = ratio(pageRect->y1, pageRect->y2, right);
104 b = ratio(pageRect->x1, pageRect->x2, bottom);
107 qCWarning(
Log) <<
"Unsupported page rotation!" << page->getRotate();
111 TextOutputDev device(
nullptr,
false, 0,
false,
false);
112 d->m_doc->m_popplerDoc->displayPageSlice(&device, d->m_pageNum + 1, 72, 72, 0,
false,
true,
false, -1, -1, -1, -1);
113 std::unique_ptr<GooString> s(device.getText(l, t, r, b));
114#if KPOPPLER_VERSION >= QT_VERSION_CHECK(0, 72, 0)
124 return d->m_images.size();
130 return d->m_images[index];
133QVariantList PdfPage::imagesVariant()
const
138 std::for_each(d->m_images.begin(), d->m_images.end(), [&l](
const PdfImage& img) { l.push_back(QVariant::fromValue(img)); });
146 PopplerGlobalParams gp;
147 const auto pageRect = d->m_doc->m_popplerDoc->getPage(d->m_pageNum + 1)->getCropBox();
149 for (
const auto &img : d->m_images) {
150 if ((img.d->m_transform.dx() >= ratio(pageRect->x1, pageRect->x2, left) && img.d->m_transform.dx() <= ratio(pageRect->x1, pageRect->x2, right)) &&
151 (img.d->m_transform.dy() >= ratio(pageRect->y1, pageRect->y2, top) && img.d->m_transform.dy() <= ratio(pageRect->y1, pageRect->y2, bottom)))
162 return d->m_links.size();
168 return d->m_links[index];
171QVariantList PdfPage::linksVariant()
const
175 l.reserve(d->m_links.size());
176 std::transform(d->m_links.begin(), d->m_links.end(), std::back_inserter(l), [](
const PdfLink &
link) { return QVariant::fromValue(link); });
186 for (
const auto &
link : d->m_links) {
193 std::sort(l.begin(), l.end(), [](
const auto &lhs,
const auto &rhs) {
194 const auto lhsLink = lhs.template value<PdfLink>();
195 const auto rhsLink = rhs.template value<PdfLink>();
196 if (lhsLink.area().top() == rhsLink.area().top()) {
197 return lhsLink.area().left() < rhsLink.area().left();
199 return lhsLink.area().top() < rhsLink.area().top();
205static constexpr inline double pdfToMM(
double points)
207 return points * 25.4 / 72.0;
210int PdfPage::width()
const
212 const auto page = d->m_doc->m_popplerDoc->getPage(d->m_pageNum + 1);
213 const auto rot = page->getRotate();
214 if (rot == 90 || rot == 270) {
215 return pdfToMM(page->getCropHeight());
217 return pdfToMM(page->getCropWidth());
220int PdfPage::height()
const
222 const auto page = d->m_doc->m_popplerDoc->getPage(d->m_pageNum + 1);
223 const auto rot = page->getRotate();
224 if (rot == 90 || rot == 270) {
225 return pdfToMM(page->getCropWidth());
227 return pdfToMM(page->getCropHeight());
231PdfDocument::PdfDocument(
QObject *parent)
233 , d(new PdfDocumentPrivate)
237PdfDocument::~PdfDocument() =
default;
242 std::for_each(d->m_pages.begin(), d->m_pages.end(), [&text](
const PdfPage &p) { text += p.text(); });
246int PdfDocument::pageCount()
const
248 return d->m_popplerDoc->getNumPages();
253 return d->m_pages[index];
258 return d->m_pdfData.size();
261#if KPOPPLER_VERSION >= QT_VERSION_CHECK(21, 8, 0)
262static QDateTime parsePdfDateTime(
const GooString *str)
264static QDateTime parsePdfDateTime(
const char *str)
277 if (!parseDateString(str, &year, &month, &day, &hour, &min, &sec, &tz, &tzHours, &tzMins)) {
281 QDate date(year, month, day);
282 QTime time(hour, min, sec);
283 if (!date.isValid() || !time.isValid()) {
287 int offset = tzHours * 3600 + tzMins * 60;
289 return QDateTime(date, time, QTimeZone::fromSecondsAheadOfUtc(offset));
290 }
else if (tz ==
'-') {
291 return QDateTime(date, time, QTimeZone::fromSecondsAheadOfUtc(-offset));
293 return QDateTime(date, time, QTimeZone::UTC);
298 std::unique_ptr<GooString> dt(d->m_popplerDoc->getDocInfoCreatDate());
302#if KPOPPLER_VERSION >= QT_VERSION_CHECK(21, 8, 0)
303 return parsePdfDateTime(dt.get());
304#elif KPOPPLER_VERSION >= QT_VERSION_CHECK(0, 72, 0)
305 return parsePdfDateTime(dt->c_str());
307 return parsePdfDateTime(dt->getCString());
313 std::unique_ptr<GooString> dt(d->m_popplerDoc->getDocInfoModDate());
317#if KPOPPLER_VERSION >= QT_VERSION_CHECK(21, 8, 0)
318 return parsePdfDateTime(dt.get());
319#elif KPOPPLER_VERSION >= QT_VERSION_CHECK(0, 72, 0)
320 return parsePdfDateTime(dt->c_str());
322 return parsePdfDateTime(dt->getCString());
327QString gooStringToUnicode(
const std::unique_ptr<GooString> &s)
333 if (s->hasUnicodeMarker() || s->hasUnicodeMarkerLE()) {
334 return QString::fromUtf16(
reinterpret_cast<const char16_t*
>(s->toStr().c_str()), s->toStr().size() / 2);
337 std::unique_ptr<const char[]> utf16Data(pdfDocEncodingToUTF16(s->toStr(), &len));
338 return QString::fromUtf16(
reinterpret_cast<const char16_t*
>(utf16Data.get()), len / 2);
346 return gooStringToUnicode(d->m_popplerDoc->getDocInfoTitle());
351 return gooStringToUnicode(d->m_popplerDoc->getDocInfoProducer());
356 return gooStringToUnicode(d->m_popplerDoc->getDocInfoCreator());
361 return gooStringToUnicode(d->m_popplerDoc->getDocInfoAuthor());
364QVariantList PdfDocument::pagesVariant()
const
367 l.reserve(pageCount());
368 std::for_each(d->m_pages.begin(), d->m_pages.end(), [&l](
const PdfPage& p) { l.push_back(QVariant::fromValue(p)); });
374 PopplerGlobalParams gp;
377 doc->d->m_pdfData = data;
379#if KPOPPLER_VERSION >= QT_VERSION_CHECK(0, 58, 0)
380 auto stream =
new MemStream(
const_cast<char*
>(doc->d->m_pdfData.constData()), 0, doc->d->m_pdfData.size(), Object());
384 auto stream =
new MemStream(
const_cast<char*
>(doc->d->m_pdfData.constData()), 0, doc->d->m_pdfData.size(), &obj);
386 std::unique_ptr<PDFDoc> popplerDoc(
new PDFDoc(stream));
387 if (!popplerDoc->isOk()) {
388 qCWarning(
Log) <<
"Got invalid PDF document!" << popplerDoc->getErrorCode();
392 doc->d->m_pages.reserve(popplerDoc->getNumPages());
393 for (
int i = 0; i < popplerDoc->getNumPages(); ++i) {
395 page.d->m_pageNum = i;
396 page.d->m_doc = doc->d.get();
397 doc->d->m_pages.push_back(
page);
400 doc->d->m_popplerDoc = std::move(popplerDoc);
401 return doc.release();
409#include "moc_pdfdocument.cpp"
PDF document for extraction.
PdfPage page(int index) const
The n-thj page in this document.
int fileSize() const
File size of the entire document in bytes.
static PdfDocument * fromData(const QByteArray &data, QObject *parent=nullptr)
Creates a PdfDocument from the given raw data.
static bool maybePdf(const QByteArray &data)
Fast check whether data might be a PDF document.
An image in a PDF document.
An external link in a PDF file.
A page in a PDF document.
Q_INVOKABLE QVariantList linksInRect(double left, double top, double right, double bottom) const
Returns all links in the specified sub-rect of this page.
PdfImage image(int index) const
The n-th image found in this document.
int linkCount() const
The number of links found in this document.
Q_INVOKABLE QString textInRect(double left, double top, double right, double bottom) const
Returns the text in the specified sub-rect of this page.
PdfLink link(int index) const
The n-th link found in this document.
int imageCount() const
The number of images found in this document.
Q_INVOKABLE QVariantList imagesInRect(double left, double top, double right, double bottom) const
Returns the images in the specified sub-rect of this page.
KIOCORE_EXPORT CopyJob * link(const QList< QUrl > &src, const QUrl &destDir, JobFlags flags=DefaultFlags)
Classes for reservation/travel data models, data extraction and data augmentation.
bool startsWith(const QByteArray &ba) const const
QObject * parent() const const
bool intersects(const QRectF &rectangle) const const
QString fromUtf16(const ushort *unicode, int size)
QString fromUtf8(const char *str, int size)
QVariant fromValue(const T &value)