svx: improved text importing from PDF

Change-Id: I9a2fc2c8511655c1aa362c1a03a5e82ae3ba697e
(cherry picked from commit d057cf3d9184cc5d96af9c957411911f5e788f4d)

3 files changed, 108 insertions, 39 deletions

diff --git a/external/pdfium/0007-svx-improved-text-importing-from-PDF.patch.2 b/external/pdfium/0007-svx-improved-text-importing-from-PDF.patch.2
new file mode 100644
index 000000000000..e7afda576d15
--- /dev/null
+++ b/external/pdfium/0007-svx-improved-text-importing-from-PDF.patch.2
@@ -0,0 +1,62 @@
+From 87f3da183a87f3ff5df854971a0c3bc2134ecd61 Mon Sep 17 00:00:00 2001
+From: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
+Date: Tue, 5 Jun 2018 11:31:35 +0200
+Subject: [PATCH 07/14] svx: improved text importing from PDF
+
+---
+ pdfium/fpdfsdk/fpdf_editpage.cpp | 9 +++++++--
+ pdfium/public/fpdf_edit.h        | 6 +++++-
+ 2 files changed, 12 insertions(+), 3 deletions(-)
+
+diff --git a/pdfium/fpdfsdk/fpdf_editpage.cpp b/pdfium/fpdfsdk/fpdf_editpage.cpp
+index f8e2418..2249e8e 100644
+--- a/pdfium/fpdfsdk/fpdf_editpage.cpp
++++ b/pdfium/fpdfsdk/fpdf_editpage.cpp
+@@ -652,8 +652,11 @@ FPDFTextObj_GetMatrix(FPDF_PAGEOBJECT text_object,
+                       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);
+@@ -662,6 +665,8 @@ FPDFTextObj_GetMatrix(FPDF_PAGEOBJECT text_object,
+   *b = matrix.b;
+   *c = matrix.c;
+   *d = matrix.d;
++  *e = matrix.e;
++  *f = matrix.f;
+ }
+ 
+ FPDF_EXPORT int FPDF_CALLCONV
+diff --git a/pdfium/public/fpdf_edit.h b/pdfium/public/fpdf_edit.h
+index 89ec8cf..fc906f4 100644
+--- a/pdfium/public/fpdf_edit.h
++++ b/pdfium/public/fpdf_edit.h
+@@ -1038,12 +1038,16 @@ FPDFTextObj_GetFontSize(FPDF_PAGEOBJECT text_object);
+ //   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.
+ //
+-- 
+2.16.3
+
diff --git a/external/pdfium/UnpackedTarball_pdfium.mk b/external/pdfium/UnpackedTarball_pdfium.mk
index af7841dcd120..f0ee740ddf94 100644
--- a/external/pdfium/UnpackedTarball_pdfium.mk
+++ b/external/pdfium/UnpackedTarball_pdfium.mk
@@ -20,6 +20,7 @@ pdfium_patches += 0003-svx-import-PDF-images-as-BGRA.patch.2
 pdfium_patches += 0004-svx-support-PDF-text-color.patch.2
 pdfium_patches += 0005-svx-support-Paths-in-PDFs-while-importing.patch.2
 pdfium_patches += 0006-svx-improve-path-importing-from-PDF.patch.2
+pdfium_patches += 0007-svx-improved-text-importing-from-PDF.patch.2
 
 $(eval $(call gb_UnpackedTarball_UnpackedTarball,pdfium))
 
diff --git a/svx/source/svdraw/svdpdf.cxx b/svx/source/svdraw/svdpdf.cxx
index 04e5f3f7f48e..7ef3d8614e7a 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
@@ -1024,53 +1026,56 @@ 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);
-    tools::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)
@@ -1083,11 +1088,12 @@ 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.AdjustY(-(FRound(aFontMetric.GetAscent() * mfScaleY)));
@@ -1325,7 +1331,7 @@ void ImpSdrPdfImport::ImportPath(FPDF_PAGEOBJECT pPageObject, int nPageObjectInd
     }
 
     const basegfx::B2DHomMatrix aTransform(
-        basegfx::tools::createScaleTranslateB2DHomMatrix(mfScaleX, mfScaleY, maOfs.X(), maOfs.Y()));
+        basegfx::utils::createScaleTranslateB2DHomMatrix(mfScaleX, mfScaleY, maOfs.X(), maOfs.Y()));
     aPoly.transform(aTransform);
 
     float fWidth = 1;
@@ -1353,7 +1359,7 @@ void ImpSdrPdfImport::ImportPath(FPDF_PAGEOBJECT pPageObject, int nPageObjectInd
     // if(!mbLastObjWasPolyWithoutLine || !CheckLastPolyLineAndFillMerge(basegfx::B2DPolyPolygon(aSource)))
 
     aPoly.setClosed(true); // TODO: Review
-    SdrPathObj* pPath = new SdrPathObj(OBJ_POLY, basegfx::B2DPolyPolygon(aPoly));
+    SdrPathObj* pPath = new SdrPathObj(*mpModel, OBJ_POLY, basegfx::B2DPolyPolygon(aPoly));
     SetAttributes(pPath);
     InsertObj(pPath, false);
 }