tdf#127471 Detect&Correct EMF/WMF with wrong FontScale

Before correcting our EMF/WMF export to write the Windows-
specific data in the case of FontScaling, we wrote these
files with wrong FontScaling.
This change tries to detect and correct this at import, so
that newer versions of the office on all plattforms can
again load old, from us but not on Windows written EMF/WMF
files.
With this change we can read again all new and old EMF/WMF
files (see table in task, comment 80).

Change-Id: I1a0b0ab5f57c7cd40520401568af05cab4ecb4c8
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/111399
Tested-by: Jenkins
Reviewed-by: Armin Le Grand <Armin.Le.Grand@me.com>

4 files changed, 239 insertions, 3 deletions

diff --git a/emfio/inc/mtftools.hxx b/emfio/inc/mtftools.hxx
index 70471b66ea21..286442767d3a 100644
--- a/emfio/inc/mtftools.hxx
+++ b/emfio/inc/mtftools.hxx
@@ -27,6 +27,7 @@
 #include <vcl/bitmapex.hxx>
 #include <vcl/lineinfo.hxx>
 #include <vcl/outdevstate.hxx>
+#include <rtl/ref.hxx>
 
 #include "emfiodllapi.h"
 
@@ -259,6 +260,8 @@ namespace emfio
 #define HUNDREDTH_MILLIMETERS_PER_MILLIINCH 2.54
 #define MILLIINCH_PER_TWIPS   1.44
 
+class MetaFontAction;
+
 //============================ WmfReader ==================================
 
 namespace emfio
@@ -458,6 +461,24 @@ namespace emfio
         {}
     };
 
+    // tdf#127471 implement detection and correction of wrongly scaled
+    // fonts in own-written, old (before this fix) EMF/WMF files
+    class ScaledFontDetectCorrectHelper
+    {
+    private:
+        rtl::Reference<MetaFontAction>                                  maCurrentMetaFontAction;
+        std::vector<double>                                             maAlternativeFontScales;
+        std::vector<std::pair<rtl::Reference<MetaFontAction>, double>>  maPositiveIdentifiedCases;
+        std::vector<std::pair<rtl::Reference<MetaFontAction>, double>>  maNegativeIdentifiedCases;
+
+    public:
+        ScaledFontDetectCorrectHelper();
+        void endCurrentMetaFontAction();
+        void newCurrentMetaFontAction(rtl::Reference<MetaFontAction>& rNewMetaFontAction);
+        void evaluateAlternativeFontScale(OUString const & rText, tools::Long nImportedTextLength);
+        void applyAlternativeFontScale();
+    };
+
     class MtfTools
     {
         MtfTools(MtfTools const &) = delete;
@@ -521,6 +542,9 @@ namespace emfio
         sal_uInt32          mnEndPos;
         std::vector<std::unique_ptr<BSaveStruct>>    maBmpSaveList;
 
+        // tdf#127471 always try to detect - only used with ScaledText
+        ScaledFontDetectCorrectHelper maScaledFontHelper;
+
         bool                mbNopMode : 1;
         bool                mbFillStyleSelected : 1;
         bool                mbClipNeedsUpdate : 1;
diff --git a/emfio/source/reader/emfreader.cxx b/emfio/source/reader/emfreader.cxx
index c2240ed612e5..30730725a182 100644
--- a/emfio/source/reader/emfreader.cxx
+++ b/emfio/source/reader/emfreader.cxx
@@ -2043,6 +2043,10 @@ namespace emfio
             }
             mpInputStream->Seek( nNextPos );
         }
+
+        // tdf#127471
+        maScaledFontHelper.applyAlternativeFontScale();
+
         if( !maBmpSaveList.empty() )
             ResolveBitmapActions( maBmpSaveList );
 
diff --git a/emfio/source/reader/mtftools.cxx b/emfio/source/reader/mtftools.cxx
index 6064ca937a16..e64bb488a458 100644
--- a/emfio/source/reader/mtftools.cxx
+++ b/emfio/source/reader/mtftools.cxx
@@ -299,6 +299,184 @@ namespace emfio
 #endif
     };
 
+    // tdf#127471
+    ScaledFontDetectCorrectHelper::ScaledFontDetectCorrectHelper()
+    :   maCurrentMetaFontAction(),
+        maAlternativeFontScales(),
+        maPositiveIdentifiedCases(),
+        maNegativeIdentifiedCases()
+    {
+    }
+
+    void ScaledFontDetectCorrectHelper::endCurrentMetaFontAction()
+    {
+        if(maCurrentMetaFontAction.is() && !maAlternativeFontScales.empty())
+        {
+            // create average corrected FontScale value and count
+            // positive/negative hits
+            sal_uInt32 nPositive(0);
+            sal_uInt32 nNegative(0);
+            double fAverage(0.0);
+
+            for(double fPart : maAlternativeFontScales)
+            {
+                if(fPart < 0.0)
+                {
+                    nNegative++;
+                    fAverage += -fPart;
+                }
+                else
+                {
+                    nPositive++;
+                    fAverage += fPart;
+                }
+            }
+
+            fAverage /= static_cast<double>(maAlternativeFontScales.size());
+
+            if(nPositive >= nNegative)
+            {
+                // correction intended, it is probably an old imported file
+                maPositiveIdentifiedCases.push_back(std::pair<rtl::Reference<MetaFontAction>, double>(maCurrentMetaFontAction, fAverage));
+            }
+            else
+            {
+                // correction not favorable in the majority of cases for this Font, still
+                // remember to have a weight in the last decision for correction
+                maNegativeIdentifiedCases.push_back(std::pair<rtl::Reference<MetaFontAction>, double>(maCurrentMetaFontAction, fAverage));
+            }
+        }
+
+        maCurrentMetaFontAction.clear();
+        maAlternativeFontScales.clear();
+    }
+
+    void ScaledFontDetectCorrectHelper::newCurrentMetaFontAction(rtl::Reference<MetaFontAction>& rNewMetaFontAction)
+    {
+        maCurrentMetaFontAction.clear();
+        maAlternativeFontScales.clear();
+
+        if(rNewMetaFontAction.is())
+        {
+            // check 1st criteria for FontScale active. We usually write this,
+            // so this will already sort out most situations
+            const vcl::Font& rCandidate(rNewMetaFontAction->GetFont());
+
+            if(0 != rCandidate.GetAverageFontWidth())
+            {
+                const tools::Long nUnscaledAverageFontWidth(rCandidate.GetOrCalculateAverageFontWidth());
+
+                // check 2nd (system-dependent) criteria for FontScale
+                if(nUnscaledAverageFontWidth != rCandidate.GetFontHeight())
+                {
+                    // FontScale is active, remrmber and use as current
+                    maCurrentMetaFontAction = rNewMetaFontAction;
+                }
+            }
+        }
+    }
+
+    void ScaledFontDetectCorrectHelper::evaluateAlternativeFontScale(OUString const & rText, tools::Long nImportedTextLength)
+    {
+        if(maCurrentMetaFontAction.is())
+        {
+            SolarMutexGuard aGuard; // VirtualDevice is not thread-safe
+            ScopedVclPtrInstance< VirtualDevice > pTempVirtualDevice;
+
+            // calculate measured TextLength
+            const vcl::Font& rFontCandidate(maCurrentMetaFontAction->GetFont());
+            pTempVirtualDevice->SetFont(rFontCandidate);
+            const tools::Long nMeasuredTextLength(pTempVirtualDevice->GetTextWidth(rText));
+
+            // compare expected and imported TextLengths
+            if(nImportedTextLength != nMeasuredTextLength)
+            {
+                const double fFactorText(static_cast<double>(nImportedTextLength) / static_cast<double>(nMeasuredTextLength));
+                const double fFactorTextPercent(fabs(1.0 - fFactorText) * 100.0);
+
+                // if we assume that loaded file was written on old linux, we have to
+                // back-convert the scale value depending on which system we run
+#ifdef _WIN32
+                // When running on Windows the value was not adapted at font import (see WinMtfFontStyle
+                // constructor), so it is still NormedFontScaling and we need to convert to Windows-style
+                // scaling
+#else
+                // When running on unx (non-Windows) the value was already adapted at font import (see WinMtfFontStyle
+                // constructor). It was wrongly assumed to be Windows-style FontScaling, so we need to revert that
+                // to get back to the needed unx-style FontScale
+#endif
+                // Interestingly this leads to the *same* correction, so no need to make this
+                // system-dependent (!)
+                const tools::Long nUnscaledAverageFontWidth(rFontCandidate.GetOrCalculateAverageFontWidth());
+                const tools::Long nScaledAverageFontWidth(rFontCandidate.GetAverageFontWidth());
+                const double fScaleFactor(static_cast<double>(nUnscaledAverageFontWidth) / static_cast<double>(rFontCandidate.GetFontHeight()));
+                const double fCorrectedAverageFontWidth(static_cast<double>(nScaledAverageFontWidth) * fScaleFactor);
+                tools::Long nCorrectedTextLength(0);
+
+                { // do in own scope, only need nUnscaledAverageFontWidth
+                    vcl::Font rFontCandidate2(rFontCandidate);
+                    rFontCandidate2.SetAverageFontWidth(static_cast<tools::Long>(fCorrectedAverageFontWidth));
+                    pTempVirtualDevice->SetFont(rFontCandidate2);
+                    nCorrectedTextLength = pTempVirtualDevice->GetTextWidth(rText);
+                }
+
+                const double fFactorCorrectedText(static_cast<double>(nImportedTextLength) / static_cast<double>(nCorrectedTextLength));
+                const double fFactorCorrectedTextPercent(fabs(1.0 - fFactorCorrectedText) * 100.0);
+
+                // If FactorCorrectedText fits better than FactorText this is probably
+                // an import of an old EMF/WMF written by LibreOffice on a non-Windows (unx) system
+                // and should be corrected.
+                // Usually in tested cases this lies inside 5% of range, so detecting this just using
+                //  fFactorTextPercent indside 5% -> no old file
+                //  fFactorCorrectedTextPercent indside 5% -> is old file
+                // works not too bad, but there are some strange not so often used fonts where that
+                // values do deviate, so better just compare if old corrected would fit better than
+                // the uncorrected case, that is usually safe.
+                if(fFactorCorrectedTextPercent < fFactorTextPercent)
+                {
+                    maAlternativeFontScales.push_back(fCorrectedAverageFontWidth);
+                }
+                else
+                {
+                    // also push, but negative to remember non-fitting case
+                    maAlternativeFontScales.push_back(-fCorrectedAverageFontWidth);
+                }
+            }
+        }
+    }
+
+    void ScaledFontDetectCorrectHelper::applyAlternativeFontScale()
+    {
+        // make sure last evtl. detected current FontAction gets added to identified cases
+        endCurrentMetaFontAction();
+
+        // Take final decision to correct FontScaling for this imported Metafile or not.
+        // It is possible to weight positive against negative cases, so to only finally
+        // correct when more positive cases were detected.
+        // But that would be inconsequent and wrong. *If* the detected case is an old import
+        // the whole file was written with wrong FontScale values and all Font actions
+        // need to be corrected. Thus, for now, correct all when there are/is positive
+        // cases detected.
+        // On the other hand it *may* be that for some strange fonts there is a false-positive
+        // in the positive cases, so at least insist on positive cases being more than negative.
+        // Still, do then correct *all* cases.
+        if(!maPositiveIdentifiedCases.empty()
+            && maPositiveIdentifiedCases.size() >= maNegativeIdentifiedCases.size())
+        {
+            for(std::pair<rtl::Reference<MetaFontAction>, double>& rCandidate : maPositiveIdentifiedCases)
+            {
+                rCandidate.first->correctFontScale(static_cast<tools::Long>(rCandidate.second));
+            }
+            for(std::pair<rtl::Reference<MetaFontAction>, double>& rCandidate : maNegativeIdentifiedCases)
+            {
+                rCandidate.first->correctFontScale(static_cast<tools::Long>(rCandidate.second));
+            }
+        }
+
+        maPositiveIdentifiedCases.clear();
+        maNegativeIdentifiedCases.clear();
+    }
+
     Color MtfTools::ReadColor()
     {
         sal_uInt32 nColor;
@@ -927,6 +1105,7 @@ namespace emfio
         mnStartPos(0),
         mnEndPos(0),
         maBmpSaveList(),
+        maScaledFontHelper(),
         mbNopMode(false),
         mbFillStyleSelected(false),
         mbClipNeedsUpdate(true),
@@ -1617,14 +1796,29 @@ namespace emfio
                 maActPos = rPosition + aActPosDelta;
             }
         }
-        if ( bChangeFont || ( maLatestFont != aTmp ) )
+
+        if(bChangeFont || (maLatestFont != aTmp))
         {
             maLatestFont = aTmp;
-            mpGDIMetaFile->AddAction( new MetaFontAction( aTmp ) );
+            rtl::Reference<MetaFontAction> aNewMetaFontAction(new MetaFontAction(aTmp));
+
+            // tdf#127471 end evtl active MetaFontAction scale corrector detector/collector
+            maScaledFontHelper.endCurrentMetaFontAction();
+
+            // !bRecordPath: else no MetaTextArrayAction will be created
+            // nullptr != pDXArry: detection only possible when text size is given
+            // rText.getLength(): no useful check without text
+            if(!bRecordPath && nullptr != pDXArry && 0 != rText.getLength())
+            {
+                maScaledFontHelper.newCurrentMetaFontAction(aNewMetaFontAction);
+            }
+
+            mpGDIMetaFile->AddAction( aNewMetaFontAction );
             mpGDIMetaFile->AddAction( new MetaTextAlignAction( aTmp.GetAlignment() ) );
             mpGDIMetaFile->AddAction( new MetaTextColorAction( aTmp.GetColor() ) );
             mpGDIMetaFile->AddAction( new MetaTextFillColorAction( aTmp.GetFillColor(), !aTmp.IsTransparent() ) );
         }
+
         if ( bRecordPath )
         {
             // TODO
@@ -1645,7 +1839,18 @@ namespace emfio
                 /* because text without dx array is badly scaled, we
                    will create such an array if necessary */
                 tools::Long* pDX = pDXArry;
-                if (!pDXArry)
+                if (pDXArry)
+                {
+                    // only useful when we have an imported DXArray
+                    if(!rText.isEmpty())
+                    {
+                        maScaledFontHelper.evaluateAlternativeFontScale(
+                            rText,
+                            pDXArry[rText.getLength() - 1] // extract imported TextLength
+                        );
+                    }
+                }
+                else
                 {
                     // #i117968# VirtualDevice is not thread safe, but filter is used in multithreading
                     SolarMutexGuard aGuard;
diff --git a/emfio/source/reader/wmfreader.cxx b/emfio/source/reader/wmfreader.cxx
index 999584b02294..f285cacc1674 100644
--- a/emfio/source/reader/wmfreader.cxx
+++ b/emfio/source/reader/wmfreader.cxx
@@ -1234,6 +1234,9 @@ namespace emfio
             case W_META_ENDDOC:
             break;
         }
+
+        // tdf#127471
+        maScaledFontHelper.applyAlternativeFontScale();
     }
 
     const tools::Long   aMaxWidth = 1024;