view.cpp

Go to the documentation of this file.

/*
 * This program source code file is part of KiCad, a free EDA CAD application.
 *
 * Copyright (C) 2013-2017 CERN
 * Copyright (C) 2021 KiCad Developers, see AUTHORS.txt for contributors.
 *
 * @author Tomasz Wlostowski <[email protected]>
 * @author Maciej Suminski <[email protected]>
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License
 * as published by the Free Software Foundation; either version 2
 * of the License, or (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, you may find one here:
 * http://www.gnu.org/licenses/old-licenses/gpl-2.0.html
 * or you may search the http://www.gnu.org website for the version 2 license,
 * or you may write to the Free Software Foundation, Inc.,
 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
 */


#include <eda_item.h>
#include <layer_ids.h>
#include <trace_helpers.h>

#include <view/view.h>
#include <view/view_group.h>
#include <view/view_item.h>
#include <view/view_rtree.h>
#include <view/view_overlay.h>

#include <gal/definitions.h>
#include <gal/graphics_abstraction_layer.h>
#include <painter.h>

#include <profile.h>

#ifdef KICAD_GAL_PROFILE
#include <wx/log.h>
#endif

namespace KIGFX {

class VIEW;

class VIEW_ITEM_DATA
{
public:
 VIEW_ITEM_DATA() :
 m_view( nullptr ),
 m_flags( KIGFX::VISIBLE ),
 m_requiredUpdate( KIGFX::NONE ),
 m_drawPriority( 0 ),
 m_groups( nullptr ),
 m_groupsSize( 0 ) {}

 ~VIEW_ITEM_DATA()
 {
 deleteGroups();
 }

 int GetFlags() const
 {
 return m_flags;
 }

private:
 friend class VIEW;

 void getLayers( int* aLayers, int& aCount ) const
 {
 int* layersPtr = aLayers;

 for( int layer : m_layers )
 *layersPtr++ = layer;

 aCount = m_layers.size();
 }

 int getGroup( int aLayer ) const
 {
  for( int i = 0; i < m_groupsSize; ++i )
 {
 if( m_groups[i].first == aLayer )
 return m_groups[i].second;
 }

 return -1;
 }

 void setGroup( int aLayer, int aGroup )
 {
 // Look if there is already an entry for the layer
 for( int i = 0; i < m_groupsSize; ++i )
 {
 if( m_groups[i].first == aLayer )
 {
 m_groups[i].second = aGroup;
 return;
 }
 }

 // If there was no entry for the given layer - create one
 std::pair<int, int>* newGroups = new std::pair<int, int>[m_groupsSize + 1];

 if( m_groupsSize > 0 )
 {
 std::copy( m_groups, m_groups + m_groupsSize, newGroups );
 delete[] m_groups;
 }

 m_groups = newGroups;
 newGroups[m_groupsSize++] = { aLayer, aGroup };
 }


 void deleteGroups()
 {
 delete[] m_groups;
 m_groups = nullptr;
 m_groupsSize = 0;
 }


 inline bool storesGroups() const
 {
 return m_groupsSize > 0;
 }


 void reorderGroups( std::unordered_map<int, int> aReorderMap )
 {
 for( int i = 0; i < m_groupsSize; ++i )
 {
 int orig_layer = m_groups[i].first;
 int new_layer = orig_layer;

 try
 {
 new_layer = aReorderMap.at( orig_layer );
 }
 catch( const std::out_of_range& )
 {}

 m_groups[i].first = new_layer;
 }
 }

 void saveLayers( int* aLayers, int aCount )
 {
 m_layers.clear();

 for( int i = 0; i < aCount; ++i )
 {
 // this fires on some eagle board after EAGLE_PLUGIN::Load()
  wxASSERT( unsigned( aLayers[i] ) <= unsigned( VIEW::VIEW_MAX_LAYERS ) );

 m_layers.push_back( aLayers[i] );
 }
 }

 int requiredUpdate() const
 {
 return m_requiredUpdate;
 }

 void clearUpdateFlags()
 {
 m_requiredUpdate = NONE;
 }

 bool isRenderable() const
 {
 return m_flags == VISIBLE;
 }

 VIEW* m_view; 
 int m_flags; 
 int m_requiredUpdate; 
 int m_drawPriority; 

 std::pair<int, int>* m_groups; 
 int m_groupsSize;

 std::vector<int> m_layers; 
};


void VIEW::OnDestroy( VIEW_ITEM* aItem )
{
 VIEW_ITEM_DATA* data = aItem->viewPrivData();

 if( !data )
 return;

 if( data->m_view )
 data->m_view->VIEW::Remove( aItem );

 delete data;
 aItem->ClearViewPrivData();
}


VIEW::VIEW( bool aIsDynamic ) :
 m_enableOrderModifier( true ),
 m_scale( 4.0 ),
 m_minScale( 0.2 ), m_maxScale( 50000.0 ),
 m_mirrorX( false ), m_mirrorY( false ),
 m_painter( nullptr ),
 m_gal( nullptr ),
 m_dynamic( aIsDynamic ),
 m_useDrawPriority( false ),
 m_nextDrawPriority( 0 ),
 m_reverseDrawOrder( false )
{
 // Set m_boundary to define the max area size. The default area size
 // is defined here as the max value of a int.
 // this is a default value acceptable for Pcbnew and Gerbview, but too large for Eeschema.
 // So in eeschema a call to SetBoundary() with a smaller value will be needed.
 typedef std::numeric_limits<int> coord_limits;
 double pos = coord_limits::lowest() / 2 + coord_limits::epsilon();
 double size = coord_limits::max() - coord_limits::epsilon();
 m_boundary.SetOrigin( pos, pos );
 m_boundary.SetSize( size, size );

 m_allItems.reset( new std::vector<VIEW_ITEM*> );
 m_allItems->reserve( 32768 );

 // Redraw everything at the beginning
 MarkDirty();

 m_layers.reserve( VIEW_MAX_LAYERS );

 // View uses layers to display EDA_ITEMs (item may be displayed on several layers, for example
 // pad may be shown on pad, pad hole and solder paste layers). There are usual copper layers
 // (eg. F.Cu, B.Cu, internal and so on) and layers for displaying objects such as texts,
 // silkscreen, pads, vias, etc.
 for( int ii = 0; ii < VIEW_MAX_LAYERS; ++ii )
 {
 m_layers.emplace_back();
 m_layers[ii].items = std::make_shared<VIEW_RTREE>();
 m_layers[ii].id = ii;
 m_layers[ii].renderingOrder = ii;
 m_layers[ii].visible = true;
 m_layers[ii].displayOnly = false;
 m_layers[ii].diffLayer = false;
 m_layers[ii].hasNegatives = false;
 m_layers[ii].target = TARGET_CACHED;
 }

 sortLayers();

 m_preview.reset( new KIGFX::VIEW_GROUP() );
 Add( m_preview.get() );
}


VIEW::~VIEW()
{
 Remove( m_preview.get() );
}


void VIEW::Add( VIEW_ITEM* aItem, int aDrawPriority )
{
 int layers[VIEW_MAX_LAYERS], layers_count;

 if( aDrawPriority < 0 )
 aDrawPriority = m_nextDrawPriority++;

 if( !aItem->m_viewPrivData )
 aItem->m_viewPrivData = new VIEW_ITEM_DATA;

 aItem->m_viewPrivData->m_view = this;
 aItem->m_viewPrivData->m_drawPriority = aDrawPriority;

 aItem->ViewGetLayers( layers, layers_count );
 aItem->viewPrivData()->saveLayers( layers, layers_count );

 m_allItems->push_back( aItem );

 for( int i = 0; i < layers_count; ++i )
 {
 VIEW_LAYER& l = m_layers[layers[i]];
 l.items->Insert( aItem );
 MarkTargetDirty( l.target );
 }

 SetVisible( aItem, true );
 Update( aItem, KIGFX::INITIAL_ADD );
}


void VIEW::Remove( VIEW_ITEM* aItem )
{
 if( !aItem )
 return;

 auto viewData = aItem->viewPrivData();

 if( !viewData )
 return;

 wxCHECK( viewData->m_view == this, /*void*/ );
 auto item = std::find( m_allItems->begin(), m_allItems->end(), aItem );

 if( item != m_allItems->end() )
 {
 m_allItems->erase( item );
 viewData->clearUpdateFlags();
 }

 int layers[VIEW::VIEW_MAX_LAYERS], layers_count;
 viewData->getLayers( layers, layers_count );

 for( int i = 0; i < layers_count; ++i )
 {
 VIEW_LAYER& l = m_layers[layers[i]];
 l.items->Remove( aItem );
 MarkTargetDirty( l.target );

 // Clear the GAL cache
 int prevGroup = viewData->getGroup( layers[i] );

 if( prevGroup >= 0 )
 m_gal->DeleteGroup( prevGroup );
 }

 viewData->deleteGroups();
 viewData->m_view = nullptr;
}


void VIEW::SetRequired( int aLayerId, int aRequiredId, bool aRequired )
{
 wxCHECK( (unsigned) aLayerId < m_layers.size(), /*void*/ );
 wxCHECK( (unsigned) aRequiredId < m_layers.size(), /*void*/ );

 if( aRequired )
 m_layers[aLayerId].requiredLayers.insert( aRequiredId );
 else
 m_layers[aLayerId].requiredLayers.erase( aRequired );
}


// stupid C++... python lambda would do this in one line
template <class CONTAINER>
struct QUERY_VISITOR
{
 typedef typename CONTAINER::value_type item_type;

 QUERY_VISITOR( CONTAINER& aCont, int aLayer ) :
 m_cont( aCont ), m_layer( aLayer )
 {
 }

 bool operator()( VIEW_ITEM* aItem )
 {
 if( aItem->viewPrivData()->GetFlags() & VISIBLE )
 m_cont.push_back( VIEW::LAYER_ITEM_PAIR( aItem, m_layer ) );

 return true;
 }

 CONTAINER& m_cont;
 int m_layer;
};


int VIEW::Query( const BOX2I& aRect, std::vector<LAYER_ITEM_PAIR>& aResult ) const
{
 if( m_orderedLayers.empty() )
 return 0;

 std::vector<VIEW_LAYER*>::const_reverse_iterator i;

 // execute queries in reverse direction, so that items that are on the top of
 // the rendering stack are returned first.
 for( i = m_orderedLayers.rbegin(); i != m_orderedLayers.rend(); ++i )
 {
 // ignore layers that do not contain actual items (i.e. the selection box, menus, floats)
 if( ( *i )->displayOnly || !( *i )->visible )
 continue;

 QUERY_VISITOR<std::vector<LAYER_ITEM_PAIR> > visitor( aResult, ( *i )->id );
 ( *i )->items->Query( aRect, visitor );
 }

 return aResult.size();
}


VECTOR2D VIEW::ToWorld( const VECTOR2D& aCoord, bool aAbsolute ) const
{
 const MATRIX3x3D& matrix = m_gal->GetScreenWorldMatrix();

 if( aAbsolute )
 return VECTOR2D( matrix * aCoord );
 else
 return VECTOR2D( matrix.GetScale().x * aCoord.x, matrix.GetScale().y * aCoord.y );
}


double VIEW::ToWorld( double aSize ) const
{
 const MATRIX3x3D& matrix = m_gal->GetScreenWorldMatrix();

 return fabs( matrix.GetScale().x * aSize );
}


VECTOR2D VIEW::ToScreen( const VECTOR2D& aCoord, bool aAbsolute ) const
{
 const MATRIX3x3D& matrix = m_gal->GetWorldScreenMatrix();

 if( aAbsolute )
 return VECTOR2D( matrix * aCoord );
 else
 return VECTOR2D( matrix.GetScale().x * aCoord.x, matrix.GetScale().y * aCoord.y );
}


double VIEW::ToScreen( double aSize ) const
{
 const MATRIX3x3D& matrix = m_gal->GetWorldScreenMatrix();

 return matrix.GetScale().x * aSize;
}


void VIEW::CopySettings( const VIEW* aOtherView )
{
 wxASSERT_MSG( false, wxT( "This is not implemented" ) );
}


void VIEW::SetGAL( GAL* aGal )
{
 bool recacheGroups = ( m_gal != nullptr ); // recache groups only if GAL is reassigned
 m_gal = aGal;

 // clear group numbers, so everything is going to be recached
 if( recacheGroups )
 clearGroupCache();

 // every target has to be refreshed
 MarkDirty();

 // force the new GAL to display the current viewport.
 SetCenter( m_center );
 SetScale( m_scale );
 SetMirror( m_mirrorX, m_mirrorY );
}


BOX2D VIEW::GetViewport() const
{
 BOX2D rect;
 VECTOR2D screenSize = m_gal->GetScreenPixelSize();

 rect.SetOrigin( ToWorld( VECTOR2D( 0, 0 ) ) );
 rect.SetEnd( ToWorld( screenSize ) );

 return rect.Normalize();
}


void VIEW::SetViewport( const BOX2D& aViewport )
{
 VECTOR2D ssize = ToWorld( m_gal->GetScreenPixelSize(), false );

 wxCHECK( ssize.x > 0 && ssize.y > 0, /*void*/ );

 VECTOR2D centre = aViewport.Centre();
 VECTOR2D vsize = aViewport.GetSize();
 double zoom = 1.0 / std::max( fabs( vsize.x / ssize.x ), fabs( vsize.y / ssize.y ) );

 SetCenter( centre );
 SetScale( GetScale() * zoom );
}


void VIEW::SetMirror( bool aMirrorX, bool aMirrorY )
{
 wxASSERT_MSG( !aMirrorY, _( "Mirroring for Y axis is not supported yet" ) );

 m_mirrorX = aMirrorX;
 m_mirrorY = aMirrorY;
 m_gal->SetFlip( aMirrorX, aMirrorY );

 // Redraw everything
 MarkDirty();
}


void VIEW::SetScale( double aScale, VECTOR2D aAnchor )
{
 if( aAnchor == VECTOR2D( 0, 0 ) )
 aAnchor = m_center;

 VECTOR2D a = ToScreen( aAnchor );

 if( aScale < m_minScale )
 m_scale = m_minScale;
 else if( aScale > m_maxScale )
 m_scale = m_maxScale;
 else
 m_scale = aScale;

 m_gal->SetZoomFactor( m_scale );
 m_gal->ComputeWorldScreenMatrix();

 VECTOR2D delta = ToWorld( a ) - aAnchor;

 SetCenter( m_center - delta );

 // Redraw everything after the viewport has changed
 MarkDirty();
}


void VIEW::SetCenter( const VECTOR2D& aCenter )
{
 m_center = aCenter;

 if( !m_boundary.Contains( aCenter ) )
 {
 if( m_center.x < m_boundary.GetLeft() )
 m_center.x = m_boundary.GetLeft();
 else if( aCenter.x > m_boundary.GetRight() )
 m_center.x = m_boundary.GetRight();

 if( m_center.y < m_boundary.GetTop() )
 m_center.y = m_boundary.GetTop();
 else if( m_center.y > m_boundary.GetBottom() )
 m_center.y = m_boundary.GetBottom();
 }

 m_gal->SetLookAtPoint( m_center );
 m_gal->ComputeWorldScreenMatrix();

 // Redraw everything after the viewport has changed
 MarkDirty();
}


void VIEW::SetCenter( const VECTOR2D& aCenter, const std::vector<BOX2D>& obscuringScreenRects )
{
 if( obscuringScreenRects.empty() )
 return SetCenter( aCenter );

 BOX2D screenRect( { 0, 0 }, m_gal->GetScreenPixelSize() );
 SHAPE_POLY_SET unobscuredPoly( screenRect );
 VECTOR2D unobscuredCenter = screenRect.Centre();

 for( const BOX2D& obscuringScreenRect : obscuringScreenRects )
 {
 SHAPE_POLY_SET obscuringPoly( obscuringScreenRect );
 unobscuredPoly.BooleanSubtract( obscuringPoly, SHAPE_POLY_SET::PM_FAST );
 }

 /*
 * Perform a step-wise deflate to find the center of the largest unobscured area
 */

 BOX2I bbox = unobscuredPoly.BBox();
 int step = std::min( bbox.GetWidth(), bbox.GetHeight() ) / 10;

 while( !unobscuredPoly.IsEmpty() )
 {
 unobscuredCenter = (wxPoint) unobscuredPoly.BBox().Centre();
 unobscuredPoly.Deflate( step, 4 );
 }

 SetCenter( aCenter - ToWorld( unobscuredCenter - screenRect.Centre(), false ) );
}


void VIEW::SetLayerOrder( int aLayer, int aRenderingOrder )
{
 m_layers[aLayer].renderingOrder = aRenderingOrder;

 sortLayers();
}


int VIEW::GetLayerOrder( int aLayer ) const
{
 return m_layers.at( aLayer ).renderingOrder;
}


void VIEW::SortLayers( int aLayers[], int& aCount ) const
{
 int maxLay, maxOrd, maxIdx;

 for( int i = 0; i < aCount; ++i )
 {
 maxLay = aLayers[i];
 maxOrd = GetLayerOrder( maxLay );
 maxIdx = i;

 // Look for the max element in the range (j..aCount)
 for( int j = i; j < aCount; ++j )
 {
 if( maxOrd < GetLayerOrder( aLayers[j] ) )
 {
 maxLay = aLayers[j];
 maxOrd = GetLayerOrder( maxLay );
 maxIdx = j;
 }
 }

 // Swap elements
 aLayers[maxIdx] = aLayers[i];
 aLayers[i] = maxLay;
 }
}


void VIEW::ReorderLayerData( std::unordered_map<int, int> aReorderMap )
{
 std::vector<VIEW_LAYER> new_map;
 new_map.reserve( m_layers.size() );

 for( const VIEW_LAYER& layer : m_layers )
 new_map.push_back( layer );

 for( auto& pair : aReorderMap )
 {
  new_map[pair.second] = m_layers[pair.first];
 new_map[pair.second].id = pair.second;
 }

 m_layers = new_map;

 for( VIEW_ITEM* item : *m_allItems )
 {
 VIEW_ITEM_DATA* viewData = item->viewPrivData();

 if( !viewData )
 continue;

 int layers[VIEW::VIEW_MAX_LAYERS], layers_count;

 item->ViewGetLayers( layers, layers_count );
 viewData->saveLayers( layers, layers_count );

 viewData->reorderGroups( aReorderMap );

 viewData->m_requiredUpdate |= COLOR;
 }

 UpdateItems();
}


struct VIEW::UPDATE_COLOR_VISITOR
{
 UPDATE_COLOR_VISITOR( int aLayer, PAINTER* aPainter, GAL* aGal ) :
 layer( aLayer ),
 painter( aPainter ),
 gal( aGal )
 {
 }

 bool operator()( VIEW_ITEM* aItem )
 {
 // Obtain the color that should be used for coloring the item
 const COLOR4D color = painter->GetSettings()->GetColor( aItem, layer );
 int group = aItem->viewPrivData()->getGroup( layer );

 if( group >= 0 )
 gal->ChangeGroupColor( group, color );

 return true;
 }

 int layer;
 PAINTER* painter;
 GAL* gal;
};


void VIEW::UpdateLayerColor( int aLayer )
{
 // There is no point in updating non-cached layers
 if( !IsCached( aLayer ) )
 return;

 BOX2I r;

 r.SetMaximum();

 if( m_gal->IsVisible() )
 {
 GAL_UPDATE_CONTEXT ctx( m_gal );

 UPDATE_COLOR_VISITOR visitor( aLayer, m_painter, m_gal );
 m_layers[aLayer].items->Query( r, visitor );
 MarkTargetDirty( m_layers[aLayer].target );
 }
}


void VIEW::UpdateAllLayersColor()
{
 if( m_gal->IsVisible() )
 {
 GAL_UPDATE_CONTEXT ctx( m_gal );

 for( VIEW_ITEM* item : *m_allItems )
 {
 VIEW_ITEM_DATA* viewData = item->viewPrivData();

 if( !viewData )
 continue;

 int layers[VIEW::VIEW_MAX_LAYERS], layers_count;
 viewData->getLayers( layers, layers_count );

 for( int i = 0; i < layers_count; ++i )
 {
 const COLOR4D color = m_painter->GetSettings()->GetColor( item, layers[i] );
 int group = viewData->getGroup( layers[i] );

 if( group >= 0 )
 m_gal->ChangeGroupColor( group, color );
 }
 }
 }

 MarkDirty();
}


struct VIEW::UPDATE_DEPTH_VISITOR
{
 UPDATE_DEPTH_VISITOR( int aLayer, int aDepth, GAL* aGal ) :
 layer( aLayer ),
 depth( aDepth ),
 gal( aGal )
 {
 }

 bool operator()( VIEW_ITEM* aItem )
 {
 int group = aItem->viewPrivData()->getGroup( layer );

 if( group >= 0 )
 gal->ChangeGroupDepth( group, depth );

 return true;
 }

 int layer, depth;
 GAL* gal;
};


int VIEW::GetTopLayer() const
{
 if( m_topLayers.size() == 0 )
 return 0;

 return *m_topLayers.begin();
}


void VIEW::SetTopLayer( int aLayer, bool aEnabled )
{
 if( aEnabled )
 {
 if( m_topLayers.count( aLayer ) == 1 )
 return;

 m_topLayers.insert( aLayer );

 // Move the layer closer to front
 if( m_enableOrderModifier )
 m_layers[aLayer].renderingOrder += TOP_LAYER_MODIFIER;
 }
 else
 {
 if( m_topLayers.count( aLayer ) == 0 )
 return;

 m_topLayers.erase( aLayer );

 // Restore the previous rendering order
 if( m_enableOrderModifier )
 m_layers[aLayer].renderingOrder -= TOP_LAYER_MODIFIER;
 }
}


void VIEW::EnableTopLayer( bool aEnable )
{
 if( aEnable == m_enableOrderModifier )
 return;

 m_enableOrderModifier = aEnable;

 std::set<unsigned int>::iterator it;

 if( aEnable )
 {
 for( it = m_topLayers.begin(); it != m_topLayers.end(); ++it )
 m_layers[*it].renderingOrder += TOP_LAYER_MODIFIER;
 }
 else
 {
 for( it = m_topLayers.begin(); it != m_topLayers.end(); ++it )
 m_layers[*it].renderingOrder -= TOP_LAYER_MODIFIER;
 }

 UpdateAllLayersOrder();
 UpdateAllLayersColor();
}


void VIEW::ClearTopLayers()
{
 std::set<unsigned int>::iterator it;

 if( m_enableOrderModifier )
 {
 // Restore the previous rendering order for layers that were marked as top
 for( it = m_topLayers.begin(); it != m_topLayers.end(); ++it )
 m_layers[*it].renderingOrder -= TOP_LAYER_MODIFIER;
 }

 m_topLayers.clear();
}


void VIEW::UpdateAllLayersOrder()
{
 sortLayers();

 if( m_gal->IsVisible() )
 {
 GAL_UPDATE_CONTEXT ctx( m_gal );

 for( VIEW_ITEM* item : *m_allItems )
 {
 VIEW_ITEM_DATA* viewData = item->viewPrivData();

 if( !viewData )
 continue;

 int layers[VIEW::VIEW_MAX_LAYERS], layers_count;
 viewData->getLayers( layers, layers_count );

 for( int i = 0; i < layers_count; ++i )
 {
  int group = viewData->getGroup( layers[i] );

 if( group >= 0 )
 m_gal->ChangeGroupDepth( group, m_layers[layers[i]].renderingOrder );
 }
 }
 }

 MarkDirty();
}


struct VIEW::DRAW_ITEM_VISITOR
{
 DRAW_ITEM_VISITOR( VIEW* aView, int aLayer, bool aUseDrawPriority, bool aReverseDrawOrder ) :
 view( aView ),
 layer( aLayer ),
 useDrawPriority( aUseDrawPriority ),
 reverseDrawOrder( aReverseDrawOrder )
 {
 }

 bool operator()( VIEW_ITEM* aItem )
 {
 wxCHECK( aItem->viewPrivData(), false );

 // Conditions that have to be fulfilled for an item to be drawn
 bool drawCondition = aItem->viewPrivData()->isRenderable()
 && aItem->ViewGetLOD( layer, view ) < view->m_scale;
 if( !drawCondition )
 return true;

 if( useDrawPriority )
 drawItems.push_back( aItem );
 else
 view->draw( aItem, layer );

 return true;
 }

 void deferredDraw()
 {
 if( reverseDrawOrder )
 {
 std::sort( drawItems.begin(), drawItems.end(),
 []( VIEW_ITEM* a, VIEW_ITEM* b ) -> bool
 {
 return b->viewPrivData()->m_drawPriority < a->viewPrivData()->m_drawPriority;
 });
 }
 else
 {
 std::sort( drawItems.begin(), drawItems.end(),
 []( VIEW_ITEM* a, VIEW_ITEM* b ) -> bool
 {
 return a->viewPrivData()->m_drawPriority < b->viewPrivData()->m_drawPriority;
 });
 }

 for( VIEW_ITEM* item : drawItems )
 view->draw( item, layer );
 }

 VIEW* view;
 int layer, layers[VIEW_MAX_LAYERS];
 bool useDrawPriority, reverseDrawOrder;
 std::vector<VIEW_ITEM*> drawItems;
};


void VIEW::redrawRect( const BOX2I& aRect )
{
 for( VIEW_LAYER* l : m_orderedLayers )
 {
 if( l->visible && IsTargetDirty( l->target ) && areRequiredLayersEnabled( l->id ) )
 {
 DRAW_ITEM_VISITOR drawFunc( this, l->id, m_useDrawPriority, m_reverseDrawOrder );

 m_gal->SetTarget( l->target );
 m_gal->SetLayerDepth( l->renderingOrder );

 // Differential layer also work for the negatives, since both special layer types
 // will composite on separate layers (at least in Cairo)
 if( l->diffLayer )
 m_gal->StartDiffLayer();
 else if( l->hasNegatives )
 m_gal->StartNegativesLayer();


 l->items->Query( aRect, drawFunc );

 if( m_useDrawPriority )
 drawFunc.deferredDraw();

 if( l->diffLayer )
 m_gal->EndDiffLayer();
 else if( l->hasNegatives )
 m_gal->EndNegativesLayer();
 }
 }
}


void VIEW::draw( VIEW_ITEM* aItem, int aLayer, bool aImmediate )
{
 VIEW_ITEM_DATA* viewData = aItem->viewPrivData();

 if( !viewData )
 return;

 if( IsCached( aLayer ) && !aImmediate )
 {
 // Draw using cached information or create one
 int group = viewData->getGroup( aLayer );

 if( group >= 0 )
 m_gal->DrawGroup( group );
 else
 Update( aItem );
 }
 else
 {
 // Immediate mode
 if( !m_painter->Draw( aItem, aLayer ) )
 aItem->ViewDraw( aLayer, this ); // Alternative drawing method
 }
}


void VIEW::draw( VIEW_ITEM* aItem, bool aImmediate )
{
 int layers[VIEW_MAX_LAYERS], layers_count;

 aItem->ViewGetLayers( layers, layers_count );

 // Sorting is needed for drawing order dependent GALs (like Cairo)
 SortLayers( layers, layers_count );

 for( int i = 0; i < layers_count; ++i )
 {
 m_gal->SetLayerDepth( m_layers.at( layers[i] ).renderingOrder );
 draw( aItem, layers[i], aImmediate );
 }
}


void VIEW::draw( VIEW_GROUP* aGroup, bool aImmediate )
{
 for( unsigned int i = 0; i < aGroup->GetSize(); i++)
 draw( aGroup->GetItem(i), aImmediate );
}


struct VIEW::RECACHE_ITEM_VISITOR
{
 RECACHE_ITEM_VISITOR( VIEW* aView, GAL* aGal, int aLayer ) :
 view( aView ),
 gal( aGal ),
 layer( aLayer )
 {
 }

 bool operator()( VIEW_ITEM* aItem )
 {
 VIEW_ITEM_DATA* viewData = aItem->viewPrivData();

 if( !viewData )
 return false;

 // Remove previously cached group
 int group = viewData->getGroup( layer );

 if( group >= 0 )
 gal->DeleteGroup( group );

 viewData->setGroup( layer, -1 );
 view->Update( aItem );

 return true;
 }

 VIEW* view;
 GAL* gal;
 int layer;
};


void VIEW::Clear()
{
 BOX2I r;
 r.SetMaximum();
 m_allItems->clear();

 for( VIEW_LAYER& layer : m_layers )
 layer.items->RemoveAll();

 m_nextDrawPriority = 0;

 m_gal->ClearCache();
}


void VIEW::ClearTargets()
{
 if( IsTargetDirty( TARGET_CACHED ) || IsTargetDirty( TARGET_NONCACHED ) )
 {
 // TARGET_CACHED and TARGET_NONCACHED have to be redrawn together, as they contain
  // layers that rely on each other (eg. netnames are noncached, but tracks - are cached)
 m_gal->ClearTarget( TARGET_NONCACHED );
 m_gal->ClearTarget( TARGET_CACHED );

 MarkDirty();
 }

 if( IsTargetDirty( TARGET_OVERLAY ) )
 {
 m_gal->ClearTarget( TARGET_OVERLAY );
 }
}


void VIEW::Redraw()
{
#ifdef KICAD_GAL_PROFILE
 PROF_COUNTER totalRealTime;
#endif /* KICAD_GAL_PROFILE */

 VECTOR2D screenSize = m_gal->GetScreenPixelSize();
 BOX2D rect( ToWorld( VECTOR2D( 0, 0 ) ),
 ToWorld( screenSize ) - ToWorld( VECTOR2D( 0, 0 ) ) );

 rect.Normalize();
 BOX2I recti( rect.GetPosition(), rect.GetSize() );

 // The view rtree uses integer positions. Large screens can overflow this size so in
 // this case, simply set the rectangle to the full rtree.
 if( rect.GetWidth() > std::numeric_limits<int>::max()
 || rect.GetHeight() > std::numeric_limits<int>::max() )
 {
 recti.SetMaximum();
 }

 redrawRect( recti );

 // All targets were redrawn, so nothing is dirty
 MarkClean();

#ifdef KICAD_GAL_PROFILE
 totalRealTime.Stop();
 wxLogTrace( traceGalProfile, wxT( "VIEW::Redraw(): %.1f ms" ), totalRealTime.msecs() );
#endif /* KICAD_GAL_PROFILE */
}


const VECTOR2I& VIEW::GetScreenPixelSize() const
{
 return m_gal->GetScreenPixelSize();
}


struct VIEW::CLEAR_LAYER_CACHE_VISITOR
{
 CLEAR_LAYER_CACHE_VISITOR( VIEW* aView ) :
 view( aView )
 {
 }

 bool operator()( VIEW_ITEM* aItem )
 {
 aItem->viewPrivData()->deleteGroups();

 return true;
 }

 VIEW* view;
};


void VIEW::clearGroupCache()
{
 BOX2I r;

 r.SetMaximum();
 CLEAR_LAYER_CACHE_VISITOR visitor( this );

 for( VIEW_LAYER& layer : m_layers )
 layer.items->Query( r, visitor );
}


void VIEW::invalidateItem( VIEW_ITEM* aItem, int aUpdateFlags )
{
 if( aUpdateFlags & INITIAL_ADD )
 {
 // Don't update layers or bbox, since it was done in VIEW::Add()
 // Now that we have initialized, set flags to ALL for the code below
 aUpdateFlags = ALL;
 }
 else
 {
 // updateLayers updates geometry too, so we do not have to update both of them at the
 // same time
 if( aUpdateFlags & LAYERS )
 updateLayers( aItem );
 else if( aUpdateFlags & GEOMETRY )
 updateBbox( aItem );
 }

 int layers[VIEW_MAX_LAYERS], layers_count;
 aItem->ViewGetLayers( layers, layers_count );

 // Iterate through layers used by the item and recache it immediately
 for( int i = 0; i < layers_count; ++i )
 {
 int layerId = layers[i];

 if( IsCached( layerId ) )
 {
 if( aUpdateFlags & ( GEOMETRY | LAYERS | REPAINT ) )
 updateItemGeometry( aItem, layerId );
 else if( aUpdateFlags & COLOR )
 updateItemColor( aItem, layerId );
 }

 // Mark those layers as dirty, so the VIEW will be refreshed
 MarkTargetDirty( m_layers[layerId].target );
 }

 aItem->viewPrivData()->clearUpdateFlags();
}


void VIEW::sortLayers()
{
 int n = 0;

 m_orderedLayers.resize( m_layers.size() );

 for( VIEW_LAYER& layer : m_layers )
 m_orderedLayers[n++] = &layer;

 sort( m_orderedLayers.begin(), m_orderedLayers.end(), compareRenderingOrder );

 MarkDirty();
}


void VIEW::updateItemColor( VIEW_ITEM* aItem, int aLayer )
{
 VIEW_ITEM_DATA* viewData = aItem->viewPrivData();
 wxCHECK( (unsigned) aLayer < m_layers.size(), /*void*/ );
 wxCHECK( IsCached( aLayer ), /*void*/ );

 if( !viewData )
 return;

 // Obtain the color that should be used for coloring the item on the specific layerId
 const COLOR4D color = m_painter->GetSettings()->GetColor( aItem, aLayer );
 int group = viewData->getGroup( aLayer );

 // Change the color, only if it has group assigned
 if( group >= 0 )
 m_gal->ChangeGroupColor( group, color );
}


void VIEW::updateItemGeometry( VIEW_ITEM* aItem, int aLayer )
{
 VIEW_ITEM_DATA* viewData = aItem->viewPrivData();
 wxCHECK( (unsigned) aLayer < m_layers.size(), /*void*/ );
 wxCHECK( IsCached( aLayer ), /*void*/ );

 if( !viewData )
 return;

 VIEW_LAYER& l = m_layers.at( aLayer );

 m_gal->SetTarget( l.target );
 m_gal->SetLayerDepth( l.renderingOrder );

 // Redraw the item from scratch
 int group = viewData->getGroup( aLayer );

 if( group >= 0 )
 m_gal->DeleteGroup( group );

 group = m_gal->BeginGroup();
 viewData->setGroup( aLayer, group );

 if( !m_painter->Draw( static_cast<EDA_ITEM*>( aItem ), aLayer ) )
 aItem->ViewDraw( aLayer, this ); // Alternative drawing method

 m_gal->EndGroup();
}


void VIEW::updateBbox( VIEW_ITEM* aItem )
{
 int layers[VIEW_MAX_LAYERS], layers_count;

 aItem->ViewGetLayers( layers, layers_count );

 for( int i = 0; i < layers_count; ++i )
 {
 VIEW_LAYER& l = m_layers[layers[i]];
 l.items->Remove( aItem );
 l.items->Insert( aItem );
 MarkTargetDirty( l.target );
 }
}


void VIEW::updateLayers( VIEW_ITEM* aItem )
{
 VIEW_ITEM_DATA* viewData = aItem->viewPrivData();
 int layers[VIEW_MAX_LAYERS], layers_count;

 if( !viewData )
 return;

 // Remove the item from previous layer set
 viewData->getLayers( layers, layers_count );

 for( int i = 0; i < layers_count; ++i )
 {
 VIEW_LAYER& l = m_layers[layers[i]];
 l.items->Remove( aItem );
 MarkTargetDirty( l.target );

 if( IsCached( l.id ) )
 {
 // Redraw the item from scratch
 int prevGroup = viewData->getGroup( layers[i] );

 if( prevGroup >= 0 )
 {
 m_gal->DeleteGroup( prevGroup );
 viewData->setGroup( l.id, -1 );
 }
 }
 }

 // Add the item to new layer set
 aItem->ViewGetLayers( layers, layers_count );
 viewData->saveLayers( layers, layers_count );

 for( int i = 0; i < layers_count; i++ )
 {
 VIEW_LAYER& l = m_layers[layers[i]];
 l.items->Insert( aItem );
 MarkTargetDirty( l.target );
 }
}


bool VIEW::areRequiredLayersEnabled( int aLayerId ) const
{
 wxCHECK( (unsigned) aLayerId < m_layers.size(), false );

 std::set<int>::const_iterator it, it_end;

 for( int layer : m_layers.at( aLayerId ).requiredLayers )
 {
 // That is enough if just one layer is not enabled
 if( !m_layers.at( layer ).visible || !areRequiredLayersEnabled( layer ) )
 return false;
 }

 return true;
}


void VIEW::RecacheAllItems()
{
 BOX2I r;

 r.SetMaximum();

 for( const VIEW_LAYER& l : m_layers )
 {
 if( IsCached( l.id ) )
 {
 RECACHE_ITEM_VISITOR visitor( this, m_gal, l.id );
 l.items->Query( r, visitor );
 }
 }
}


void VIEW::UpdateItems()
{
 if( !m_gal->IsVisible() )
 return;

 unsigned int cntGeomUpdate = 0;
 unsigned int cntAnyUpdate = 0;

 for( VIEW_ITEM* item : *m_allItems )
 {
 auto vpd = item->viewPrivData();

 if( !vpd )
 continue;

 if( vpd->m_requiredUpdate & ( GEOMETRY | LAYERS ) )
 {
 cntGeomUpdate++;
 }
 if( vpd->m_requiredUpdate != NONE )
 {
 cntAnyUpdate++;
 }
 }

 unsigned int cntTotal = m_allItems->size();

 double ratio = (double) cntGeomUpdate / (double) cntTotal;

 // Optimization to improve view update time. If a lot of items (say, 30%) have their
 // bboxes/geometry changed it's way faster (around 10 times) to rebuild the R-Trees
 // from scratch rather than update the bbox of each changed item. Pcbnew does multiple
 // full geometry updates during file load, this can save a solid 30 seconds on load time
 // for larger designs...

 if( ratio > 0.3 )
 {
 auto allItems = *m_allItems;
 int layers[VIEW_MAX_LAYERS], layers_count;

 // kill all Rtrees
 for( VIEW_LAYER& layer : m_layers )
 layer.items->RemoveAll();

 // and re-insert items from scratch
 for( VIEW_ITEM* item : allItems )
 {
 item->ViewGetLayers( layers, layers_count );
 item->viewPrivData()->saveLayers( layers, layers_count );

 for( int i = 0; i < layers_count; ++i )
 {
 VIEW_LAYER& l = m_layers[layers[i]];
 l.items->Insert( item );
 MarkTargetDirty( l.target );
 }

 item->viewPrivData()->m_requiredUpdate &= ~( LAYERS | GEOMETRY );
 }
 }

 if( cntAnyUpdate )
 {
 GAL_UPDATE_CONTEXT ctx( m_gal );

 for( VIEW_ITEM* item : *m_allItems.get() )
 {
 if( item->viewPrivData() && item->viewPrivData()->m_requiredUpdate != NONE )
 {
 invalidateItem( item, item->viewPrivData()->m_requiredUpdate );
 item->viewPrivData()->m_requiredUpdate = NONE;
 }
 }
 }

 KI_TRACE( traceGalProfile, wxT( "View update: total items %u, geom %u updates %u\n" ), cntTotal,
 cntGeomUpdate, cntAnyUpdate );
}


void VIEW::UpdateAllItems( int aUpdateFlags )
{
 for( VIEW_ITEM* item : *m_allItems )
 {
 if( item->viewPrivData() )
 item->viewPrivData()->m_requiredUpdate |= aUpdateFlags;
 }
}


void VIEW::UpdateAllItemsConditionally( int aUpdateFlags,
 std::function<bool( VIEW_ITEM* )> aCondition )
{
 for( VIEW_ITEM* item : *m_allItems )
 {
 if( aCondition( item ) )
 {
 if( item->viewPrivData() )
 item->viewPrivData()->m_requiredUpdate |= aUpdateFlags;
 }
 }
}


std::unique_ptr<VIEW> VIEW::DataReference() const
{
 std::unique_ptr<VIEW> ret = std::make_unique<VIEW>();
 ret->m_allItems = m_allItems;
 ret->m_layers = m_layers;
 ret->sortLayers();
 return ret;
}


void VIEW::SetVisible( VIEW_ITEM* aItem, bool aIsVisible )
{
 VIEW_ITEM_DATA* viewData = aItem->viewPrivData();

 if( !viewData )
 return;

 bool cur_visible = viewData->m_flags & VISIBLE;

 if( cur_visible != aIsVisible )
 {
 if( aIsVisible )
 viewData->m_flags |= VISIBLE;
 else
 viewData->m_flags &= ~VISIBLE;

 Update( aItem, APPEARANCE | COLOR );
 }
}


void VIEW::Hide( VIEW_ITEM* aItem, bool aHide )
{
 VIEW_ITEM_DATA* viewData = aItem->viewPrivData();

 if( !viewData )
 return;

 if( !( viewData->m_flags & VISIBLE ) )
 return;

 if( aHide )
 viewData->m_flags |= HIDDEN;
 else
 viewData->m_flags &= ~HIDDEN;

 Update( aItem, APPEARANCE );
}


bool VIEW::IsVisible( const VIEW_ITEM* aItem ) const
{
 const VIEW_ITEM_DATA* viewData = aItem->viewPrivData();

 return viewData && ( viewData->m_flags & VISIBLE );
}


void VIEW::Update( const VIEW_ITEM* aItem ) const
{
 Update( aItem, ALL );
}


void VIEW::Update( const VIEW_ITEM* aItem, int aUpdateFlags ) const
{
 VIEW_ITEM_DATA* viewData = aItem->viewPrivData();

 if( !viewData )
 return;

 assert( aUpdateFlags != NONE );

 viewData->m_requiredUpdate |= aUpdateFlags;
}


std::shared_ptr<VIEW_OVERLAY> VIEW::MakeOverlay()
{
 std::shared_ptr<VIEW_OVERLAY> overlay( new VIEW_OVERLAY );

 Add( overlay.get() );
 return overlay;
}


void VIEW::ClearPreview()
{
 if( !m_preview )
 return;

 m_preview->Clear();

 for( EDA_ITEM *item : m_ownedItems )
 delete item;

 m_ownedItems.clear();
 Update( m_preview.get() );
}


void VIEW::InitPreview()
{
 m_preview.reset( new KIGFX::VIEW_GROUP() );
 Add( m_preview.get() );
}


void VIEW::AddToPreview( EDA_ITEM* aItem, bool aTakeOwnership )
{
 Hide( aItem, false );
 m_preview->Add( aItem );

 if( aTakeOwnership )
 m_ownedItems.push_back( aItem );

 SetVisible( m_preview.get(), true );
 Hide( m_preview.get(), false );
 Update( m_preview.get() );
}


void VIEW::ShowPreview( bool aShow )
{
 SetVisible( m_preview.get(), aShow );
}


const int VIEW::TOP_LAYER_MODIFIER = -VIEW_MAX_LAYERS;

} // namespace KIGFX