pdfium: Import PDF with unloaded images.

Reviewed-on: https://gerrit.libreoffice.org/56268
Reviewed-by: Ashod Nakashian <ashnakash@gmail.com>
Tested-by: Ashod Nakashian <ashnakash@gmail.com>
(cherry picked from commit e07561d2ae743b208a0807ef32d7f011614b73e5)

Reviewed-on: https://gerrit.libreoffice.org/77688
Tested-by: Jenkins
Reviewed-by: Ashod Nakashian <ashnakash@gmail.com>
(cherry picked from commit 9194f019afb0599d5e72476786fabfa996e07f20)

Change-Id: I5e4a16ff38b9643127ce16879b35f456c13bcff8
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/90834
Tested-by: Tomaž Vajngerl <quikee@gmail.com>
Reviewed-by: Tomaž Vajngerl <quikee@gmail.com>

3 files changed, 95 insertions, 23 deletions

diff --git a/include/vcl/pdfread.hxx b/include/vcl/pdfread.hxx
index 1de9cfc9b259..1e555d9841c7 100644
--- a/include/vcl/pdfread.hxx
+++ b/include/vcl/pdfread.hxx
@@ -11,6 +11,7 @@
 #define INCLUDED_VCL_SOURCE_FILTER_IPDF_PDFREAD_HXX
 
 #include <vector>
+#include <tools/gen.hxx>
 #include <tools/stream.hxx>
 
 namespace com
@@ -45,7 +46,16 @@ VCL_DLLPUBLIC bool ImportPDF(SvStream& rStream, Bitmap& rBitmap, size_t nPageInd
 VCL_DLLPUBLIC bool ImportPDF(SvStream& rStream, Graphic& rGraphic, double fResolutionDPI = 96.);
 
 VCL_DLLPUBLIC size_t ImportPDF(const OUString& rURL, std::vector<Bitmap>& rBitmaps,
-                               css::uno::Sequence<sal_Int8>& rPdfData, double fResolutionDPI = 96.);
+                               css::uno::Sequence<sal_Int8>& rPdfData,
+                               const double fResolutionDPI = 96.);
+
+/// Import PDF as Graphic images (1 per page), all unloaded.
+/// Since Graphic is unloaded, we need to return the page size (in pixels) separately.
+/// Does not set rPdfData if no conversion is done.
+/// Returns the number of pages read.
+VCL_DLLPUBLIC size_t ImportPDFUnloaded(const OUString& rURL,
+                                       std::vector<std::pair<Graphic, Size>>& rGraphics,
+                                       const double fResolutionDPI = 96.);
 }
 
 #endif // INCLUDED_VCL_SOURCE_FILTER_IPDF_PDFREAD_HXX
diff --git a/sd/source/filter/pdf/sdpdffilter.cxx b/sd/source/filter/pdf/sdpdffilter.cxx
index db7af07b58ce..acb811ae2063 100644
--- a/sd/source/filter/pdf/sdpdffilter.cxx
+++ b/sd/source/filter/pdf/sdpdffilter.cxx
@@ -103,37 +103,29 @@ bool SdPdfFilter::Import()
     // Rendering resolution.
     const double dResolutionDPI = 96.;
 
-    uno::Sequence<sal_Int8> aPdfData;
-    std::vector<Bitmap> aBitmaps;
-    if (vcl::ImportPDF(aFileName, aBitmaps, aPdfData, dResolutionDPI) == 0)
+    std::vector<std::pair<Graphic, Size>> aGraphics;
+    if (vcl::ImportPDFUnloaded(aFileName, aGraphics, dResolutionDPI) == 0)
         return false;
 
-    // Prepare the link with the PDF stream.
-    const size_t nGraphicContentSize = aPdfData.getLength();
-    std::unique_ptr<sal_uInt8[]> pGraphicContent(new sal_uInt8[nGraphicContentSize]);
-    memcpy(pGraphicContent.get(), aPdfData.get(), nGraphicContentSize);
-    std::shared_ptr<GfxLink> pGfxLink(std::make_shared<GfxLink>(
-        std::move(pGraphicContent), nGraphicContentSize, GfxLinkType::NativePdf));
-    auto pPdfData = std::make_shared<uno::Sequence<sal_Int8>>(aPdfData);
-
+    // Add as many pages as we need up-front.
     mrDocument.CreateFirstPages();
-    for (size_t i = 0; i < aBitmaps.size() - 1; ++i)
+    for (size_t i = 0; i < aGraphics.size() - 1; ++i)
     {
         mrDocument.DuplicatePage(0);
     }
 
-    size_t nPageNumber = 0;
-    for (Bitmap& aBitmap : aBitmaps)
+    for (const std::pair<Graphic, Size>& aPair : aGraphics)
     {
-        // Create the Graphic and link the original PDF stream.
-        Graphic aGraphic(aBitmap);
-        aGraphic.setPdfData(pPdfData);
-        aGraphic.setPageNumber(nPageNumber);
-        aGraphic.SetGfxLink(pGfxLink);
+        const Graphic& rGraphic = aPair.first;
+        const Size& aSize = aPair.second;
+
+        const sal_Int32 nPageNumber = rGraphic.getPageNumber();
+        if (nPageNumber < 0 || static_cast<size_t>(nPageNumber) >= aGraphics.size())
+            continue; // Page is out of range
 
         // Create the page and insert the Graphic.
-        SdPage* pPage = mrDocument.GetSdPage(nPageNumber++, PageKind::Standard);
-        Size aGrfSize(OutputDevice::LogicToLogic(aGraphic.GetPrefSize(), aGraphic.GetPrefMapMode(),
+        SdPage* pPage = mrDocument.GetSdPage(nPageNumber, PageKind::Standard);
+        Size aGrfSize(OutputDevice::LogicToLogic(aSize, rGraphic.GetPrefMapMode(),
                                                  MapMode(MapUnit::Map100thMM)));
 
         // Resize to original size based on 72 dpi to preserve page size.
@@ -144,7 +136,7 @@ bool SdPdfFilter::Import()
         pPage->SetSize(aGrfSize);
         Point aPos(0, 0);
 
-        pPage->InsertObject(new SdrGrafObj(pPage->getSdrModelFromSdrPage(), aGraphic,
+        pPage->InsertObject(new SdrGrafObj(pPage->getSdrModelFromSdrPage(), rGraphic,
                                            tools::Rectangle(aPos, aGrfSize)));
     }
 
diff --git a/vcl/source/filter/ipdf/pdfread.cxx b/vcl/source/filter/ipdf/pdfread.cxx
index 63162c6e2ad2..d9c197bffd77 100644
--- a/vcl/source/filter/ipdf/pdfread.cxx
+++ b/vcl/source/filter/ipdf/pdfread.cxx
@@ -17,6 +17,7 @@
 #include <fpdf_save.h>
 #endif
 
+#include <impgraph.hxx>
 #include <vcl/graph.hxx>
 #include <bitmapwriteaccess.hxx>
 #include <unotools/ucbstreamhelper.hxx>
@@ -269,6 +270,75 @@ size_t ImportPDF(const OUString& rURL, std::vector<Bitmap>& rBitmaps,
 
     return rBitmaps.size();
 }
+
+size_t ImportPDFUnloaded(const OUString& rURL, std::vector<std::pair<Graphic, Size>>& rGraphics,
+                         const double fResolutionDPI)
+{
+    std::unique_ptr<SvStream> xStream(
+        ::utl::UcbStreamHelper::CreateStream(rURL, StreamMode::READ | StreamMode::SHARE_DENYNONE));
+
+    // Save the original PDF stream for later use.
+    SvMemoryStream aMemoryStream;
+    if (!getCompatibleStream(*xStream, aMemoryStream, STREAM_SEEK_TO_BEGIN, STREAM_SEEK_TO_END))
+        return 0;
+
+    // Copy into PdfData
+    uno::Sequence<sal_Int8> aPdfData;
+    aMemoryStream.Seek(STREAM_SEEK_TO_END);
+    aPdfData = css::uno::Sequence<sal_Int8>(aMemoryStream.Tell());
+    aMemoryStream.Seek(STREAM_SEEK_TO_BEGIN);
+    aMemoryStream.ReadBytes(aPdfData.getArray(), aPdfData.getLength());
+
+    // Prepare the link with the PDF stream.
+    const size_t nGraphicContentSize = aPdfData.getLength();
+    std::unique_ptr<sal_uInt8[]> pGraphicContent(new sal_uInt8[nGraphicContentSize]);
+    memcpy(pGraphicContent.get(), aPdfData.get(), nGraphicContentSize);
+    std::shared_ptr<GfxLink> pGfxLink(std::make_shared<GfxLink>(
+        std::move(pGraphicContent), nGraphicContentSize, GfxLinkType::NativePdf));
+    auto pPdfData = std::make_shared<uno::Sequence<sal_Int8>>(aPdfData);
+
+    FPDF_LIBRARY_CONFIG aConfig;
+    aConfig.version = 2;
+    aConfig.m_pUserFontPaths = nullptr;
+    aConfig.m_pIsolate = nullptr;
+    aConfig.m_v8EmbedderSlot = 0;
+    FPDF_InitLibraryWithConfig(&aConfig);
+
+    // Load the buffer using pdfium.
+    FPDF_DOCUMENT pPdfDocument
+        = FPDF_LoadMemDocument(aPdfData.getArray(), aPdfData.getLength(), /*password=*/nullptr);
+    if (!pPdfDocument)
+        return 0;
+
+    const int nPageCount = FPDF_GetPageCount(pPdfDocument);
+    if (nPageCount <= 0)
+        return 0;
+
+    for (size_t nPageIndex = 0; nPageIndex < static_cast<size_t>(nPageCount); ++nPageIndex)
+    {
+        double fPageWidth = 0;
+        double fPageHeight = 0;
+        if (FPDF_GetPageSizeByIndex(pPdfDocument, nPageIndex, &fPageWidth, &fPageHeight) == 0)
+            continue;
+
+        // Returned unit is points, convert that to pixel.
+        const size_t nPageWidth = pointToPixel(fPageWidth, fResolutionDPI);
+        const size_t nPageHeight = pointToPixel(fPageHeight, fResolutionDPI);
+
+        // Create the Graphic and link the original PDF stream.
+        Graphic aGraphic;
+        aGraphic.setPdfData(pPdfData); // TODO: Skip if unchanged.
+        aGraphic.setPageNumber(nPageIndex);
+        aGraphic.SetGfxLink(pGfxLink);
+
+        rGraphics.emplace_back(std::move(aGraphic), Size(nPageWidth, nPageHeight));
+    }
+
+    FPDF_CloseDocument(pPdfDocument);
+    FPDF_DestroyLibrary();
+
+    return rGraphics.size();
+}
 }
 
 /* vim:set shiftwidth=4 softtabstop=4 expandtab: */