svx: improved text importing from PDF

Change-Id: I9a2fc2c8511655c1aa362c1a03a5e82ae3ba697e

2 files changed, 59 insertions, 44 deletions

diff --git a/external/pdfium/edit.patch.1 b/external/pdfium/edit.patch.1
index 86cda347f262..982560a3fab7 100644
--- a/external/pdfium/edit.patch.1
+++ b/external/pdfium/edit.patch.1
@@ -66,7 +66,7 @@ index 0d7ba56..37bdf99 100644
  FPDFImageObj_GetImageDataDecoded(FPDF_PAGEOBJECT image_object,
                                   void* buffer,
 diff --git a/fpdfsdk/fpdfeditpage.cpp b/fpdfsdk/fpdfeditpage.cpp
-index ca2cf3f..8073a18 100644
+index ca2cf3f..e7d633f 100644
 --- a/fpdfsdk/fpdfeditpage.cpp
 +++ b/fpdfsdk/fpdfeditpage.cpp
 @@ -11,12 +11,14 @@
@@ -84,7 +84,7 @@ index ca2cf3f..8073a18 100644
  #include "core/fpdfapi/page/cpdf_shadingobject.h"
  #include "core/fpdfapi/parser/cpdf_array.h"
  #include "core/fpdfapi/parser/cpdf_document.h"
-@@ -363,3 +365,123 @@ FPDFPageObj_GetBounds(FPDF_PAGEOBJECT pageObject,
+@@ -363,3 +365,128 @@ FPDFPageObj_GetBounds(FPDF_PAGEOBJECT pageObject,
    *top = bbox.top;
    return true;
  }
@@ -114,8 +114,11 @@ index ca2cf3f..8073a18 100644
 +                      double* a,
 +                      double* b,
 +                      double* c,
-+                      double* d) {
-+  if (!text_object)
++                      double* d,
++                      double* e,
++                      double* f)
++{
++  if (!text_object || !a || !b || !c || !d || !e || !f)
 +    return;
 +
 +  CPDF_TextObject* pTxtObj = CPDFTextObjectFromFPDFPageObject(text_object);
@@ -124,6 +127,8 @@ index ca2cf3f..8073a18 100644
 +  *b = matrix.b;
 +  *c = matrix.c;
 +  *d = matrix.d;
++  *e = matrix.e;
++  *f = matrix.f;
 +}
 +
 +FPDF_EXPORT int FPDF_CALLCONV
@@ -357,7 +362,7 @@ index 77c2315..db3e734 100644
  
  CPDF_PageObject* CPDFPageObjectFromFPDFPageObject(FPDF_PAGEOBJECT page_object);
 diff --git a/public/fpdf_edit.h b/public/fpdf_edit.h
-index 54735a3..a415c98 100644
+index 54735a3..c86b638 100644
 --- a/public/fpdf_edit.h
 +++ b/public/fpdf_edit.h
 @@ -520,6 +520,15 @@ FPDFPath_GetStrokeColor(FPDF_PAGEOBJECT path,
@@ -413,7 +418,7 @@ index 54735a3..a415c98 100644
  // Create a new text object using one of the standard PDF fonts.
  //
  // document   - handle to the document.
-@@ -761,6 +800,73 @@ FPDFPageObj_CreateTextObj(FPDF_DOCUMENT document,
+@@ -761,6 +800,77 @@ FPDFPageObj_CreateTextObj(FPDF_DOCUMENT document,
                            FPDF_FONT font,
                            float font_size);
  
@@ -445,12 +450,16 @@ index 54735a3..a415c98 100644
 +//   b            - Pointer to a double value receiving coefficient "b" of the matrix.
 +//   c            - Pointer to a double value receiving coefficient "c" of the matrix.
 +//   d            - Pointer to a double value receiving coefficient "d" of the matrix.
++//   e            - Pointer to a double value receiving coefficient "e" of the matrix.
++//   f            - Pointer to a double value receiving coefficient "f" of the matrix.
 +FPDF_EXPORT void FPDF_CALLCONV
 +FPDFTextObj_GetMatrix(FPDF_PAGEOBJECT text_object,
 +                      double* a,
 +                      double* b,
 +                      double* c,
-+                      double* d);
++                      double* d,
++                      double* e,
++                      double* f);
 +
 +// Get the unicode of a special character in a text object.
 +//
diff --git a/svx/source/svdraw/svdpdf.cxx b/svx/source/svdraw/svdpdf.cxx
index 2dfba0f986dc..2735ae4c94a1 100644
--- a/svx/source/svdraw/svdpdf.cxx
+++ b/svx/source/svdraw/svdpdf.cxx
@@ -103,6 +103,8 @@ static inline long lcl_ToLogic(double value)
     const long out = OutputDevice::LogicToLogic(in, MapUnit::MapPixel, MapUnit::Map100thMM);
     return out / 100;
 }
+
+static inline double sqrt2(double a, double b) { return sqrt(a * a + b * b); }
 }
 
 struct FPDFBitmapDeleter
@@ -1020,53 +1022,56 @@ void ImpSdrPdfImport::ImportText(FPDF_PAGEOBJECT pPageObject, int nPageObjectInd
         SAL_WARN("sd.filter", "FAILED to get TEXT bounds");
     }
 
-    SAL_WARN("sd.filter", "Got TEXT bounds left: " << left << ", right: " << right
-                                                   << ", top: " << top << ", bottom: " << bottom);
-    Rectangle aRect = PointsToLogic(left, right, top, bottom);
+    if (left == right || top == bottom)
+    {
+        SAL_WARN("sd.filter", "Skipping empty TEXT #" << nPageObjectIndex << " left: " << left
+                                                      << ", right: " << right << ", top: " << top
+                                                      << ", bottom: " << bottom);
+        return;
+    }
 
-    double dFontScale = 1.0;
-    geometry::Matrix2D aMatrix;
-    FPDFTextObj_GetMatrix(pPageObject, &aMatrix.m00, &aMatrix.m01, &aMatrix.m10, &aMatrix.m11);
-    if (aMatrix.m00 != aMatrix.m11 || aMatrix.m00 <= 0)
+    const int nChars = FPDFTextObj_CountChars(pPageObject);
+    std::unique_ptr<sal_Unicode[]> pText(new sal_Unicode[nChars + 1]); // + terminating null
+
+    unsigned short* pShortText = reinterpret_cast<unsigned short*>(pText.get());
+    const int nActualChars = FPDFTextObj_GetText(pPageObject, 0, nChars, pShortText);
+    if (nActualChars <= 0)
     {
-        SAL_WARN("sd.filter", "Bogus font scale matrix (" << aMatrix.m00 << ',' << aMatrix.m11
-                                                          << "), will use heuristic height of "
-                                                          << aRect.GetHeight() << ".");
-        dFontScale = aRect.GetHeight();
+        SAL_WARN("sd.filter", "Got not TEXT");
+        return;
     }
-    else
-        dFontScale = aMatrix.m00;
 
-    double dFontSize = FPDFTextObj_GetFontSize(pPageObject);
-    SAL_WARN("sd.filter", "Got Font Size: " << dFontSize);
-    dFontSize *= dFontScale;
-    SAL_WARN("sd.filter", "Got Font Size Scaled: " << dFontSize);
-    dFontSize = lcl_PointToPixel(dFontSize);
-    SAL_WARN("sd.filter", "Got Font Pixel Size: " << dFontSize);
-    dFontSize = lcl_ToLogic(dFontSize);
-    SAL_WARN("sd.filter", "Got Font Logic Size: " << dFontSize);
+    OUString sText(pText.get(), nActualChars);
+    SAL_WARN("sd.filter", "Got Text (" << nChars << "): [" << sText << "].");
+
+    double a, b, c, d, e, f;
+    FPDFTextObj_GetMatrix(pPageObject, &a, &b, &c, &d, &e, &f);
+    SAL_WARN("sd.filter", "Got font scale matrix (" << a << ", " << b << ", " << c << ", " << d
+                                                    << ", " << e << ", " << f << ')');
+    Point aPos = PointsToLogic(e, f);
+    SAL_WARN("sd.filter", "Got TEXT origin: " << aPos);
+
+    const double dFontSize = FPDFTextObj_GetFontSize(pPageObject);
+    double dFontSizeH = fabs(sqrt2(a, c) * dFontSize);
+    double dFontSizeV = fabs(sqrt2(b, d) * dFontSize);
+    SAL_WARN("sd.filter", "Got Font Size: " << dFontSize << ", Scaled Font Size H: " << dFontSizeH
+                                            << ", V: " << dFontSizeV);
+    dFontSizeH = lcl_PointToPixel(dFontSizeH);
+    dFontSizeV = lcl_PointToPixel(dFontSizeV);
+    SAL_WARN("sd.filter", "Got Pixel Font Size H: " << dFontSizeH << ", V: " << dFontSizeV);
+    dFontSizeH = lcl_ToLogic(dFontSizeH);
+    dFontSizeV = lcl_ToLogic(dFontSizeV);
+    SAL_WARN("sd.filter", "Got Logic Font Size H: " << dFontSizeH << ", V: " << dFontSizeV);
 
     unsigned int nR, nG, nB, nA;
     if (FPDFTextObj_GetStrokeColor(pPageObject, &nR, &nG, &nB, &nA))
         mpVD->SetTextColor(Color(nR, nG, nB));
 
     vcl::Font aFnt = mpVD->GetFont();
-    aFnt.SetFontSize(Size(dFontSize, dFontSize));
+    aFnt.SetFontSize(Size(dFontSizeH, dFontSizeV));
     mpVD->SetFont(aFnt);
 
-    const int nChars = FPDFTextObj_CountChars(pPageObject);
-    std::unique_ptr<sal_Unicode[]> pText(new sal_Unicode[nChars + 1]); // + terminating null
-
-    unsigned short* pShortText = reinterpret_cast<unsigned short*>(pText.get());
-    const int nActualChars = FPDFTextObj_GetText(pPageObject, 0, nChars, pShortText);
-    OUString sText(pText.get(), nActualChars);
-
-    // for (int nChar = 0; nChar < nChars; ++nChar)
-    //     pText[nChar] = static_cast<sal_Unicode>(FPDFTextObj_GetUnicode(pPageObject, nChar));
-    // OUString sText(pText.get(), nChars);
-    SAL_WARN("sd.filter", "Got Text (" << nChars << "): [" << sText << "].");
-
-    ImportText(aRect.TopLeft(), sText);
+    ImportText(aPos, sText);
 }
 
 void ImpSdrPdfImport::ImportText(const Point& rPos, const OUString& rStr)
@@ -1079,11 +1084,12 @@ void ImpSdrPdfImport::ImportText(const Point& rPos, const OUString& rStr)
 
     sal_Int32 nTextWidth = static_cast<sal_Int32>(mpVD->GetTextWidth(rStr) * mfScaleX);
     sal_Int32 nTextHeight = static_cast<sal_Int32>(mpVD->GetTextHeight() * mfScaleY);
-    SAL_WARN("sd.filter", "TextWidth: " << nTextWidth << ", TextHeight: " << nTextHeight);
+    SAL_WARN("sd.filter",
+             "Unscaled text size: " << mpVD->GetTextWidth(rStr) << 'x' << mpVD->GetTextHeight()
+                                    << ", Scaled: " << nTextWidth << 'x' << nTextHeight);
 
     Point aPos(FRound(rPos.X() * mfScaleX + maOfs.X()), FRound(rPos.Y() * mfScaleY + maOfs.Y()));
     Size aSize(nTextWidth, nTextHeight);
-    SAL_WARN("sd.filter", "Text Pos: " << aPos << ", Size: " << aSize);
 
     if (eAlg == ALIGN_BASELINE)
         aPos.Y() -= FRound(aFontMetric.GetAscent() * mfScaleY);