pdf: add text page object attribs, refactor ImpSdrPdfImport, tests

This refactors ImpSdrPdfImport to push more functions into the
PDFium wrapper. The focus is on text page object attributes.

Reviewed-on: https://gerrit.libreoffice.org/c/core/+/97366
Tested-by: Jenkins
Reviewed-by: Tomaž Vajngerl <quikee@gmail.com>
(cherry picked from commit 4e9b03d04f740a0cbafa22a4f3cedfae7f37a994)

Change-Id: Ie1faf5e3743eec7c77050835651533f9e227c2a7
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/97451
Tested-by: Jenkins CollaboraOffice <jenkinscollaboraoffice@gmail.com>
Reviewed-by: Tomaž Vajngerl <quikee@gmail.com>

6 files changed, 110 insertions, 57 deletions

diff --git a/include/vcl/filter/PDFiumLibrary.hxx b/include/vcl/filter/PDFiumLibrary.hxx
index 9f34bdb92ad6..f7dcc4b2c99e 100644
--- a/include/vcl/filter/PDFiumLibrary.hxx
+++ b/include/vcl/filter/PDFiumLibrary.hxx
@@ -22,6 +22,7 @@
 #include <basegfx/range/b2drectangle.hxx>
 #include <basegfx/matrix/b2dhommatrix.hxx>
 #include <rtl/ustring.hxx>
+#include <tools/color.hxx>
 
 #include <fpdf_doc.h>
 
@@ -93,6 +94,12 @@ public:
     std::unique_ptr<PDFiumPageObject> getFormObject(int nIndex);
 
     basegfx::B2DHomMatrix getMatrix();
+    basegfx::B2DRectangle getBounds();
+    double getFontSize();
+    OUString getFontName();
+    int getTextRenderMode();
+    Color getFillColor();
+    Color getStrokeColor();
 };
 
 class VCL_DLLPUBLIC PDFiumTextPage final
diff --git a/sd/qa/unit/SdrPdfImportTest.cxx b/sd/qa/unit/SdrPdfImportTest.cxx
index ca022d585c48..2ac548bb6c85 100644
--- a/sd/qa/unit/SdrPdfImportTest.cxx
+++ b/sd/qa/unit/SdrPdfImportTest.cxx
@@ -89,6 +89,9 @@ CPPUNIT_TEST_FIXTURE(SdrPdfImportTest, testImportSimpleText)
     SdPage* pPage = pViewShell->GetActualPage();
     CPPUNIT_ASSERT(pPage);
 
+    // Check there is one object on the page only
+    CPPUNIT_ASSERT_EQUAL(size_t(1), pPage->GetObjCount());
+
     // Get the first object - there should be only one.
     SdrObject* pObject = pPage->GetObj(0);
     CPPUNIT_ASSERT(pObject);
@@ -110,11 +113,17 @@ CPPUNIT_TEST_FIXTURE(SdrPdfImportTest, testImportSimpleText)
     // Execute the break operation - to turn the PDF into shapes/objects
     pViewShell->GetDrawView()->DoImportMarkedMtf();
 
-    // Check Objects after import
+    // Check there is one object on the page only
+    CPPUNIT_ASSERT_EQUAL(size_t(1), pPage->GetObjCount());
 
+    // Get the object
     SdrObject* pImportedObject = pPage->GetObj(0);
     CPPUNIT_ASSERT(pImportedObject);
 
+    // Check the object position
+    CPPUNIT_ASSERT_EQUAL(tools::Rectangle(Point(2011, 2098), Size(2106 + 1, 302 + 1)),
+                         pImportedObject->GetLogicRect());
+
     // Object should be a text object containing one paragraph with
     // content "This is PDF!"
 
diff --git a/svx/source/inc/svdpdf.hxx b/svx/source/inc/svdpdf.hxx
index 49c8d3e1b3c2..b1ad9d69ad12 100644
--- a/svx/source/inc/svdpdf.hxx
+++ b/svx/source/inc/svdpdf.hxx
@@ -120,7 +120,9 @@ class SVXCORE_DLLPUBLIC ImpSdrPdfImport final
                     int nPageObjectIndex);
     void ImportImage(FPDF_PAGEOBJECT pPageObject, int nPageObjectIndex);
     void ImportPath(FPDF_PAGEOBJECT pPageObject, int nPageObjectIndex);
-    void ImportText(FPDF_PAGEOBJECT pPageObject, FPDF_TEXTPAGE pTextPage, int nPageObjectIndex);
+    void ImportText(std::unique_ptr<vcl::pdf::PDFiumPageObject> const& pPageObject,
+                    std::unique_ptr<vcl::pdf::PDFiumTextPage> const& pTextPage,
+                    int nPageObjectIndex);
     void InsertTextObject(const Point& rPos, const Size& rSize, const OUString& rStr);
 
     void SetupPageScale(const double dPageWidth, const double dPageHeight);
diff --git a/svx/source/svdraw/svdpdf.cxx b/svx/source/svdraw/svdpdf.cxx
index 066c77f0798f..ff4029a00b54 100644
--- a/svx/source/svdraw/svdpdf.cxx
+++ b/svx/source/svdraw/svdpdf.cxx
@@ -694,7 +694,7 @@ void ImpSdrPdfImport::ImportPdfObject(
     switch (nPageObjectType)
     {
         case FPDF_PAGEOBJ_TEXT:
-            ImportText(pPageObject->getPointer(), pTextPage->getPointer(), nPageObjectIndex);
+            ImportText(pPageObject, pTextPage, nPageObjectIndex);
             break;
         case FPDF_PAGEOBJ_PATH:
             ImportPath(pPageObject->getPointer(), nPageObjectIndex);
@@ -736,47 +736,23 @@ void ImpSdrPdfImport::ImportForm(std::unique_ptr<vcl::pdf::PDFiumPageObject> con
     maCurrentMatrix = aOldMatrix;
 }
 
-void ImpSdrPdfImport::ImportText(FPDF_PAGEOBJECT pPageObject, FPDF_TEXTPAGE pTextPage,
+void ImpSdrPdfImport::ImportText(std::unique_ptr<vcl::pdf::PDFiumPageObject> const& pPageObject,
+                                 std::unique_ptr<vcl::pdf::PDFiumTextPage> const& pTextPage,
                                  int /*nPageObjectIndex*/)
 {
-    float left;
-    float bottom;
-    float right;
-    float top;
-    if (!FPDFPageObj_GetBounds(pPageObject, &left, &bottom, &right, &top))
-    {
-        SAL_WARN("sd.filter", "FAILED to get TEXT bounds");
-    }
-
-    if (left == right || top == bottom)
-        return;
-
-    double a, b, c, d, e, f;
-    FPDFTextObj_GetMatrix(pPageObject, &a, &b, &c, &d, &e, &f);
-
+    basegfx::B2DRectangle aTextRect = pPageObject->getBounds();
+    basegfx::B2DHomMatrix aMatrix = pPageObject->getMatrix();
     basegfx::B2DHomMatrix aTextMatrix(maCurrentMatrix);
-    basegfx::B2DRange aTextRect(left, top, right, bottom);
+
     aTextRect *= aTextMatrix;
     const tools::Rectangle aRect = PointsToLogic(aTextRect.getMinX(), aTextRect.getMaxX(),
                                                  aTextRect.getMinY(), aTextRect.getMaxY());
 
-    const int nBytes = FPDFTextObj_GetText(pPageObject, pTextPage, nullptr, 0);
-    std::unique_ptr<sal_Unicode[]> pText(new sal_Unicode[nBytes]);
-
-    const int nActualBytes = FPDFTextObj_GetText(pPageObject, pTextPage, pText.get(), nBytes);
-    if (nActualBytes <= 0)
-    {
-        return;
-    }
+    OUString sText = pPageObject->getText(pTextPage);
 
-    // Let's rely on null-terminaton for the length of the string. We
-    // just know the number of bytes the string takes, but in OUString
-    // needs the number of charaters.
-    OUString sText(pText.get());
-
-    const double dFontSize = FPDFTextObj_GetFontSize(pPageObject);
-    double dFontSizeH = fabs(sqrt2(a, c) * dFontSize);
-    double dFontSizeV = fabs(sqrt2(b, d) * dFontSize);
+    const double dFontSize = pPageObject->getFontSize();
+    double dFontSizeH = fabs(sqrt2(aMatrix.a(), aMatrix.c()) * dFontSize);
+    double dFontSizeV = fabs(sqrt2(aMatrix.b(), aMatrix.d()) * dFontSize);
 
     dFontSizeH = convertPointToMm100(dFontSizeH);
     dFontSizeV = convertPointToMm100(dFontSizeV);
@@ -790,25 +766,18 @@ void ImpSdrPdfImport::ImportText(FPDF_PAGEOBJECT pPageObject, FPDF_TEXTPAGE pTex
         mbFntDirty = true;
     }
 
-    const int nFontName = 80 + 1;
-    std::unique_ptr<char[]> pFontName(new char[nFontName]); // + terminating null
-    char* pCharFontName = reinterpret_cast<char*>(pFontName.get());
-    int nFontNameChars = FPDFTextObj_GetFontName(pPageObject, pCharFontName, nFontName);
-    if (nFontName >= nFontNameChars)
+    OUString sFontName = pPageObject->getFontName();
+    if (!sFontName.isEmpty() && sFontName != aFnt.GetFamilyName())
     {
-        OUString sFontName = OUString::createFromAscii(pFontName.get());
-        if (sFontName != aFnt.GetFamilyName())
-        {
-            aFnt.SetFamilyName(sFontName);
-            mpVD->SetFont(aFnt);
-            mbFntDirty = true;
-        }
+        aFnt.SetFamilyName(sFontName);
+        mpVD->SetFont(aFnt);
+        mbFntDirty = true;
     }
 
     Color aTextColor(COL_TRANSPARENT);
     bool bFill = false;
     bool bUse = true;
-    switch (FPDFTextObj_GetTextRenderMode(pPageObject))
+    switch (pPageObject->getTextRenderMode())
     {
         case FPDF_TEXTRENDERMODE_FILL:
         case FPDF_TEXTRENDERMODE_FILL_CLIP:
@@ -826,11 +795,9 @@ void ImpSdrPdfImport::ImportText(FPDF_PAGEOBJECT pPageObject, FPDF_TEXTPAGE pTex
     }
     if (bUse)
     {
-        unsigned int nR, nG, nB, nA;
-        bool bRet = bFill ? FPDFPageObj_GetFillColor(pPageObject, &nR, &nG, &nB, &nA)
-                          : FPDFPageObj_GetStrokeColor(pPageObject, &nR, &nG, &nB, &nA);
-        if (bRet)
-            aTextColor = Color(nR, nG, nB);
+        Color aColor = bFill ? pPageObject->getFillColor() : pPageObject->getStrokeColor();
+        if (aColor != COL_TRANSPARENT)
+            aTextColor = aColor.GetRGBColor();
     }
 
     if (aTextColor != mpVD->GetTextColor())
diff --git a/vcl/qa/cppunit/PDFiumLibraryTest.cxx b/vcl/qa/cppunit/PDFiumLibraryTest.cxx
index 9c0c92607b14..f18681c0adda 100644
--- a/vcl/qa/cppunit/PDFiumLibraryTest.cxx
+++ b/vcl/qa/cppunit/PDFiumLibraryTest.cxx
@@ -134,10 +134,22 @@ void PDFiumLibraryTest::testPageObjects()
     auto pPageObject = pPage->getObject(0);
     auto pTextPage = pPage->getTextPage();
 
-    CPPUNIT_ASSERT_EQUAL(1, pPageObject->getType());
+    CPPUNIT_ASSERT_EQUAL(1, pPageObject->getType()); // FPDF_PAGEOBJ_TEXT
+
     CPPUNIT_ASSERT_EQUAL(OUString("The quick, brown fox jumps over a lazy dog. DJs flock by when "
                                   "MTV ax quiz prog. Junk MTV quiz "),
                          pPageObject->getText(pTextPage));
+
+    CPPUNIT_ASSERT_EQUAL(12.0, pPageObject->getFontSize());
+    CPPUNIT_ASSERT_EQUAL(OUString("Liberation Serif"), pPageObject->getFontName());
+    CPPUNIT_ASSERT_EQUAL(0, pPageObject->getTextRenderMode()); // FPDF_TEXTRENDERMODE_FILL
+    CPPUNIT_ASSERT_EQUAL(COL_BLACK, pPageObject->getFillColor());
+    CPPUNIT_ASSERT_EQUAL(COL_BLACK, pPageObject->getStrokeColor());
+
+    CPPUNIT_ASSERT_DOUBLES_EQUAL(057.01, pPageObject->getBounds().getMinX(), 1E-2);
+    CPPUNIT_ASSERT_DOUBLES_EQUAL(721.51, pPageObject->getBounds().getMinY(), 1E-2);
+    CPPUNIT_ASSERT_DOUBLES_EQUAL(539.48, pPageObject->getBounds().getMaxX(), 1E-2);
+    CPPUNIT_ASSERT_DOUBLES_EQUAL(732.54, pPageObject->getBounds().getMaxY(), 1E-2);
 }
 
 void PDFiumLibraryTest::testAnnotationsMadeInEvince()
diff --git a/vcl/source/pdf/PDFiumLibrary.cxx b/vcl/source/pdf/PDFiumLibrary.cxx
index 7e30ba47b0bc..e86b7565dd1b 100644
--- a/vcl/source/pdf/PDFiumLibrary.cxx
+++ b/vcl/source/pdf/PDFiumLibrary.cxx
@@ -254,9 +254,65 @@ std::unique_ptr<PDFiumPageObject> PDFiumPageObject::getFormObject(int nIndex)
 
 basegfx::B2DHomMatrix PDFiumPageObject::getMatrix()
 {
+    basegfx::B2DHomMatrix aB2DMatrix;
     double a, b, c, d, e, f;
-    FPDFTextObj_GetMatrix(mpPageObject, &a, &b, &c, &d, &e, &f);
-    return basegfx::B2DHomMatrix::abcdef(a, b, c, d, e, f);
+    if (FPDFTextObj_GetMatrix(mpPageObject, &a, &b, &c, &d, &e, &f))
+        aB2DMatrix = basegfx::B2DHomMatrix::abcdef(a, b, c, d, e, f);
+    return aB2DMatrix;
+}
+
+basegfx::B2DRectangle PDFiumPageObject::getBounds()
+{
+    basegfx::B2DRectangle aB2DRectangle;
+
+    float left = 0;
+    float bottom = 0;
+    float right = 0;
+    float top = 0;
+    if (FPDFPageObj_GetBounds(mpPageObject, &left, &bottom, &right, &top))
+    {
+        aB2DRectangle = basegfx::B2DRectangle(left, top, right, bottom);
+    }
+    return aB2DRectangle;
+}
+
+double PDFiumPageObject::getFontSize() { return FPDFTextObj_GetFontSize(mpPageObject); }
+
+OUString PDFiumPageObject::getFontName()
+{
+    OUString sFontName;
+    const int nFontName = 80 + 1;
+    std::unique_ptr<char[]> pFontName(new char[nFontName]); // + terminating null
+    int nFontNameChars = FPDFTextObj_GetFontName(mpPageObject, pFontName.get(), nFontName);
+    if (nFontName >= nFontNameChars)
+    {
+        sFontName = OUString::createFromAscii(pFontName.get());
+    }
+    return sFontName;
+}
+
+int PDFiumPageObject::getTextRenderMode() { return FPDFTextObj_GetTextRenderMode(mpPageObject); }
+
+Color PDFiumPageObject::getFillColor()
+{
+    Color aColor = COL_TRANSPARENT;
+    unsigned int nR, nG, nB, nA;
+    if (FPDFPageObj_GetFillColor(mpPageObject, &nR, &nG, &nB, &nA))
+    {
+        aColor = Color(0xFF - nA, nR, nG, nB);
+    }
+    return aColor;
+}
+
+Color PDFiumPageObject::getStrokeColor()
+{
+    Color aColor = COL_TRANSPARENT;
+    unsigned int nR, nG, nB, nA;
+    if (FPDFPageObj_GetStrokeColor(mpPageObject, &nR, &nG, &nB, &nA))
+    {
+        aColor = Color(0xFF - nA, nR, nG, nB);
+    }
+    return aColor;
 }
 
 PDFiumAnnotation::PDFiumAnnotation(FPDF_ANNOTATION pAnnotation)