allegro_builder.cpp

Go to the documentation of this file.

/*
 * This program source code file is part of KiCad, a free EDA CAD application.
 *
 * Copyright Quilter
 * Copyright The KiCad Developers, see AUTHORS.txt for contributors.
 *
 * 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 3
 * 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, see <https://www.gnu.org/licenses/>.
 */
 
#include "allegro_builder.h"
#include "allegro_db_utils.h"
 
#include <cmath>
#include <limits>
#include <regex>
#include <set>
#include <tuple>
#include <unordered_set>
 
#include <convert/allegro_pcb_structs.h>
 
#include <wx/log.h>
 
#include <core/profile.h>
#include <core/throttle.h>
 
#include <base_units.h>
#include <board_design_settings.h>
#include <convert_basic_shapes_to_polygon.h>
#include <geometry/shape_utils.h>
#include <project/net_settings.h>
#include <footprint.h>
#include <netclass.h>
#include <pad.h>
#include <pcb_group.h>
#include <pcb_text.h>
#include <pcb_shape.h>
#include <pcb_track.h>
#include <priority_thread_pool_task.h>
#include <zone.h>
#include <zone_utils.h>
 
 
using namespace ALLEGRO;
 

static const wxChar* const traceAllegroBuilder = wxT( "KICAD_ALLEGRO_BUILDER" );
static const wxChar* const traceAllegroPerf = wxT( "KICAD_ALLEGRO_PERF" );
 
 
#define BLK_FIELD( BLK_T, FIELD ) BlockDataAs<BLK_T>( aBlock ).FIELD
 

static uint32_t PadGetNextInFootprint( const BLOCK_BASE& aBlock )
{
    const uint8_t type = aBlock.GetBlockType();
 
    if( type != 0x32 )
    {
        THROW_IO_ERROR(
                wxString::Format( "Unexpected next item in 0x32 pad list: block type %#04x, offset %#lx, key %#010x",
                                  type, aBlock.GetOffset(), aBlock.GetKey() ) );
    }
 
    // When iterating in a footprint use this field, not m_Next.
    return BLK_FIELD( BLK_0x32_PLACED_PAD, m_NextInFp );
}
 
 
template <>
struct std::hash<LAYER_INFO>
{
    size_t operator()( const LAYER_INFO& aLayerInfo ) const noexcept
    {
        return ( aLayerInfo.m_Class << 8 ) + aLayerInfo.m_Subclass;
    }
};
 

// clang-format off
static const std::unordered_map<LAYER_INFO, PCB_LAYER_ID> s_LayerKiMap = {
 
    { { LAYER_INFO::CLASS::BOARD_GEOMETRY,   LAYER_INFO::SUBCLASS::BGEOM_OUTLINE},              Edge_Cuts},
    { { LAYER_INFO::CLASS::BOARD_GEOMETRY,   LAYER_INFO::SUBCLASS::BGEOM_DESIGN_OUTLINE},       Edge_Cuts},
    { { LAYER_INFO::CLASS::BOARD_GEOMETRY,   LAYER_INFO::SUBCLASS::BGEOM_SILKSCREEN_TOP},       F_SilkS},
    { { LAYER_INFO::CLASS::BOARD_GEOMETRY,   LAYER_INFO::SUBCLASS::BGEOM_SILKSCREEN_BOTTOM},    B_SilkS},
    { { LAYER_INFO::CLASS::BOARD_GEOMETRY,   LAYER_INFO::SUBCLASS::BGEOM_SOLDERMASK_TOP},       F_Mask},
    { { LAYER_INFO::CLASS::BOARD_GEOMETRY,   LAYER_INFO::SUBCLASS::BGEOM_SOLDERMASK_BOTTOM},    B_Mask},
 
    { { LAYER_INFO::CLASS::COMPONENT_VALUE,  LAYER_INFO::SUBCLASS::ASSEMBLY_BOTTOM},            B_Fab},
    { { LAYER_INFO::CLASS::COMPONENT_VALUE,  LAYER_INFO::SUBCLASS::ASSEMBLY_TOP},               F_Fab},
 
    { { LAYER_INFO::CLASS::DEVICE_TYPE,      LAYER_INFO::SUBCLASS::ASSEMBLY_BOTTOM},            B_Fab},
    { { LAYER_INFO::CLASS::DEVICE_TYPE,      LAYER_INFO::SUBCLASS::ASSEMBLY_TOP},               F_Fab},
 
    { { LAYER_INFO::CLASS::PACKAGE_GEOMETRY, LAYER_INFO::SUBCLASS::PGEOM_SILKSCREEN_BOTTOM},    B_SilkS},
    { { LAYER_INFO::CLASS::PACKAGE_GEOMETRY, LAYER_INFO::SUBCLASS::PGEOM_SILKSCREEN_TOP},       F_SilkS},
    { { LAYER_INFO::CLASS::PACKAGE_GEOMETRY, LAYER_INFO::SUBCLASS::PGEOM_ASSEMBLY_BOTTOM},      B_Fab},
    { { LAYER_INFO::CLASS::PACKAGE_GEOMETRY, LAYER_INFO::SUBCLASS::PGEOM_ASSEMBLY_TOP},         F_Fab},
    { { LAYER_INFO::CLASS::PACKAGE_GEOMETRY, LAYER_INFO::SUBCLASS::PGEOM_PLACE_BOUND_BOTTOM},   B_CrtYd},
    { { LAYER_INFO::CLASS::PACKAGE_GEOMETRY, LAYER_INFO::SUBCLASS::PGEOM_PLACE_BOUND_TOP},      F_CrtYd},
    { { LAYER_INFO::CLASS::PACKAGE_GEOMETRY, LAYER_INFO::SUBCLASS::PGEOM_PASTEMASK_BOTTOM},     B_Paste},
    { { LAYER_INFO::CLASS::PACKAGE_GEOMETRY, LAYER_INFO::SUBCLASS::PGEOM_PASTEMASK_TOP},        F_Paste},
 
    { { LAYER_INFO::CLASS::REF_DES,          LAYER_INFO::SUBCLASS::SILKSCREEN_BOTTOM},          B_SilkS},
    { { LAYER_INFO::CLASS::REF_DES,          LAYER_INFO::SUBCLASS::SILKSCREEN_TOP},             F_SilkS},
    { { LAYER_INFO::CLASS::REF_DES,          LAYER_INFO::SUBCLASS::ASSEMBLY_BOTTOM},            B_Fab},
    { { LAYER_INFO::CLASS::REF_DES,          LAYER_INFO::SUBCLASS::ASSEMBLY_TOP},               F_Fab},
 
    { { LAYER_INFO::CLASS::MANUFACTURING,    LAYER_INFO::SUBCLASS::MFR_AUTOSILK_BOTTOM},        B_SilkS},
    { { LAYER_INFO::CLASS::MANUFACTURING,    LAYER_INFO::SUBCLASS::MFR_AUTOSILK_TOP},           F_SilkS},
};

static const std::unordered_map<LAYER_INFO, wxString> s_OptionalFixedMappings = {
    { { LAYER_INFO::CLASS::PACKAGE_GEOMETRY, LAYER_INFO::SUBCLASS::PGEOM_DISPLAY_TOP},          "DISPLAY_TOP" },
    { { LAYER_INFO::CLASS::PACKAGE_GEOMETRY, LAYER_INFO::SUBCLASS::PGEOM_DISPLAY_BOTTOM},       "DISPLAY_BOTTOM" },
    { { LAYER_INFO::CLASS::PACKAGE_GEOMETRY, LAYER_INFO::SUBCLASS::PGEOM_BODY_CENTER},          "BODY_CENTER" },
 
    { { LAYER_INFO::CLASS::BOARD_GEOMETRY,   LAYER_INFO::SUBCLASS::BGEOM_DIMENSION},            "DIMENSION" },
 
    { { LAYER_INFO::CLASS::DRAWING_FORMAT,   LAYER_INFO::SUBCLASS::DFMT_OUTLINE},               "PAGE_OUTLINE" },
 
    { { LAYER_INFO::CLASS::COMPONENT_VALUE,  LAYER_INFO::SUBCLASS::DISPLAY_BOTTOM},             "DISPLAY_BOTTOM" },
    { { LAYER_INFO::CLASS::COMPONENT_VALUE,  LAYER_INFO::SUBCLASS::DISPLAY_TOP},                "DISPLAY_TOP" },
    { { LAYER_INFO::CLASS::COMPONENT_VALUE,  LAYER_INFO::SUBCLASS::SILKSCREEN_BOTTOM},          "COMPVAL_TYPE_BOTTOM"},
    { { LAYER_INFO::CLASS::COMPONENT_VALUE,  LAYER_INFO::SUBCLASS::SILKSCREEN_TOP},             "COMPVAL_TYPE_TOP"},
 
    { { LAYER_INFO::CLASS::DEVICE_TYPE,      LAYER_INFO::SUBCLASS::DISPLAY_BOTTOM},             "DISPLAY_BOTTOM" },
    { { LAYER_INFO::CLASS::DEVICE_TYPE,      LAYER_INFO::SUBCLASS::DISPLAY_TOP},                "DISPLAY_TOP" },
    { { LAYER_INFO::CLASS::DEVICE_TYPE,      LAYER_INFO::SUBCLASS::SILKSCREEN_BOTTOM},          "DEVICE_TYPE_BOTTOM"},
    { { LAYER_INFO::CLASS::DEVICE_TYPE,      LAYER_INFO::SUBCLASS::SILKSCREEN_TOP},             "DEVICE_TYPE_TOP"},
 
    { { LAYER_INFO::CLASS::TOLERANCE,        LAYER_INFO::SUBCLASS::DISPLAY_BOTTOM},             "DISPLAY_BOTTOM" },
    { { LAYER_INFO::CLASS::TOLERANCE,        LAYER_INFO::SUBCLASS::DISPLAY_TOP},                "DISPLAY_TOP" },
    { { LAYER_INFO::CLASS::TOLERANCE,        LAYER_INFO::SUBCLASS::SILKSCREEN_BOTTOM},          "TOLERANCE_BOTTOM"},
    { { LAYER_INFO::CLASS::TOLERANCE,        LAYER_INFO::SUBCLASS::SILKSCREEN_TOP},             "TOLERANCE_TOP"},
 
    { { LAYER_INFO::CLASS::USER_PART_NUMBER, LAYER_INFO::SUBCLASS::DISPLAY_BOTTOM},             "DISPLAY_BOTTOM" },
    { { LAYER_INFO::CLASS::USER_PART_NUMBER, LAYER_INFO::SUBCLASS::DISPLAY_TOP},                "DISPLAY_TOP" },
    { { LAYER_INFO::CLASS::USER_PART_NUMBER, LAYER_INFO::SUBCLASS::SILKSCREEN_BOTTOM},          "USER_PART_NUM_BOTTOM"},
    { { LAYER_INFO::CLASS::USER_PART_NUMBER, LAYER_INFO::SUBCLASS::SILKSCREEN_TOP},             "USER_PART_NUM_TOP"},
 
    { { LAYER_INFO::CLASS::MANUFACTURING,    LAYER_INFO::SUBCLASS::MFR_XSECTION_CHART},         "XSECTION_CHART" },
};
 
// clang-format on
 

static wxString layerInfoDisplayName( const LAYER_INFO& aLayerInfo )
{
    // clang-format off
    static const std::unordered_map<uint8_t, wxString> s_ClassNames = {
        { LAYER_INFO::CLASS::BOARD_GEOMETRY,        wxS( "Board Geometry" ) },
        { LAYER_INFO::CLASS::COMPONENT_VALUE,       wxS( "Component Value" ) },
        { LAYER_INFO::CLASS::DEVICE_TYPE,           wxS( "Device Type" ) },
        { LAYER_INFO::CLASS::DRAWING_FORMAT,        wxS( "Drawing Format" ) },
        { LAYER_INFO::CLASS::ETCH,                  wxS( "Etch" ) },
        { LAYER_INFO::CLASS::MANUFACTURING,         wxS( "Manufacturing" ) },
        { LAYER_INFO::CLASS::PACKAGE_GEOMETRY,      wxS( "Package Geometry" ) },
        { LAYER_INFO::CLASS::PACKAGE_KEEPIN,        wxS( "Package Keepin" ) },
        { LAYER_INFO::CLASS::PACKAGE_KEEPOUT,       wxS( "Package Keepout" ) },
        { LAYER_INFO::CLASS::PIN,                   wxS( "Pin" ) },
        { LAYER_INFO::CLASS::REF_DES,               wxS( "Ref Des" ) },
        { LAYER_INFO::CLASS::ROUTE_KEEPIN,          wxS( "Route Keepin" ) },
        { LAYER_INFO::CLASS::ROUTE_KEEPOUT,         wxS( "Route Keepout" ) },
        { LAYER_INFO::CLASS::TOLERANCE,             wxS( "Tolerance" ) },
        { LAYER_INFO::CLASS::USER_PART_NUMBER,      wxS( "User Part Number" ) },
        { LAYER_INFO::CLASS::VIA_CLASS,             wxS( "Via Class" ) },
        { LAYER_INFO::CLASS::VIA_KEEPOUT,           wxS( "Via Keepout" ) },
        { LAYER_INFO::CLASS::ANTI_ETCH,             wxS( "Anti Etch" ) },
        { LAYER_INFO::CLASS::BOUNDARY,              wxS( "Boundary" ) },
        { LAYER_INFO::CLASS::CONSTRAINTS_REGION,    wxS( "Constraints Region" ) },
    };
 
    static const std::unordered_map<uint8_t, wxString> s_BoardGeomSubclassNames = {
        { LAYER_INFO::SUBCLASS::BGEOM_OUTLINE,              wxS( "Outline" ) },
        { LAYER_INFO::SUBCLASS::BGEOM_CONSTRAINT_AREA,      wxS( "Constraint Area" ) },
        { LAYER_INFO::SUBCLASS::BGEOM_OFF_GRID_AREA,        wxS( "Off Grid Area" ) },
        { LAYER_INFO::SUBCLASS::BGEOM_SOLDERMASK_BOTTOM,    wxS( "Soldermask Bottom" ) },
        { LAYER_INFO::SUBCLASS::BGEOM_SOLDERMASK_TOP,       wxS( "Soldermask Top" ) },
        { LAYER_INFO::SUBCLASS::BGEOM_ASSEMBLY_DETAIL,      wxS( "Assembly Detail" ) },
        { LAYER_INFO::SUBCLASS::BGEOM_SILKSCREEN_BOTTOM,    wxS( "Silkscreen Bottom" ) },
        { LAYER_INFO::SUBCLASS::BGEOM_SILKSCREEN_TOP,       wxS( "Silkscreen Top" ) },
        { LAYER_INFO::SUBCLASS::BGEOM_SWITCH_AREA_BOTTOM,   wxS( "Switch Area Bottom" ) },
        { LAYER_INFO::SUBCLASS::BGEOM_SWITCH_AREA_TOP,      wxS( "Switch Area Top" ) },
        { LAYER_INFO::SUBCLASS::BGEOM_BOTH_ROOMS,           wxS( "Both Rooms" ) },
        { LAYER_INFO::SUBCLASS::BGEOM_BOTTOM_ROOM,          wxS( "Bottom Room" ) },
        { LAYER_INFO::SUBCLASS::BGEOM_TOP_ROOM,             wxS( "Top Room" ) },
        { LAYER_INFO::SUBCLASS::BGEOM_PLACE_GRID_BOTTOM,    wxS( "Place Grid Bottom" ) },
        { LAYER_INFO::SUBCLASS::BGEOM_PLACE_GRID_TOP,       wxS( "Place Grid Top" ) },
        { LAYER_INFO::SUBCLASS::BGEOM_DIMENSION,            wxS( "Dimension" ) },
        { LAYER_INFO::SUBCLASS::BGEOM_TOOLING_CORNERS,      wxS( "Tooling Corners" ) },
        { LAYER_INFO::SUBCLASS::BGEOM_ASSEMBLY_NOTES,       wxS( "Assembly Notes" ) },
        { LAYER_INFO::SUBCLASS::BGEOM_PLATING_BAR,          wxS( "Plating Bar" ) },
        { LAYER_INFO::SUBCLASS::BGEOM_DESIGN_OUTLINE,       wxS( "Design Outline" ) },
 
    };
 
    static const std::unordered_map<uint8_t, wxString> s_ComponentValueSubclassNames = {
        { LAYER_INFO::SUBCLASS::DISPLAY_BOTTOM,             wxS( "Display Bottom" ) },
        { LAYER_INFO::SUBCLASS::DISPLAY_TOP,                wxS( "Display Top" ) },
        { LAYER_INFO::SUBCLASS::SILKSCREEN_BOTTOM,          wxS( "Silkscreen Bottom" ) },
        { LAYER_INFO::SUBCLASS::SILKSCREEN_TOP,             wxS( "Silkscreen Top" ) },
        { LAYER_INFO::SUBCLASS::ASSEMBLY_BOTTOM,            wxS( "Assembly Bottom" ) },
        { LAYER_INFO::SUBCLASS::ASSEMBLY_TOP,               wxS( "Assembly Top" ) },
    };
 
    static const std::unordered_map<uint8_t, wxString> s_DrawingFormatSubclassNames = {
        { LAYER_INFO::SUBCLASS::DFMT_REVISION_DATA,         wxS( "Revision Data" ) },
        { LAYER_INFO::SUBCLASS::DFMT_REVISION_BLOCK,        wxS( "Revision Block" ) },
        { LAYER_INFO::SUBCLASS::DFMT_TITLE_DATA,            wxS( "Title Data" ) },
        { LAYER_INFO::SUBCLASS::DFMT_TITLE_BLOCK,           wxS( "Title Block" ) },
        { LAYER_INFO::SUBCLASS::DFMT_OUTLINE,               wxS( "Outline" ) },
    };
 
    static const std::unordered_map<uint8_t, wxString> s_PackageGeometrySubclassNames = {
        { LAYER_INFO::SUBCLASS::PGEOM_PASTEMASK_BOTTOM,     wxS( "Pastemask Bottom" ) },
        { LAYER_INFO::SUBCLASS::PGEOM_PASTEMASK_TOP,        wxS( "Pastemask Top" ) },
        { LAYER_INFO::SUBCLASS::PGEOM_DFA_BOUND_BOTTOM,     wxS( "DFA Bound Bottom" ) },
        { LAYER_INFO::SUBCLASS::PGEOM_DFA_BOUND_TOP,        wxS( "DFA Bound Top" ) },
        { LAYER_INFO::SUBCLASS::PGEOM_DISPLAY_BOTTOM,       wxS( "Display Bottom" ) },
        { LAYER_INFO::SUBCLASS::PGEOM_DISPLAY_TOP,          wxS( "Display Top" ) },
        { LAYER_INFO::SUBCLASS::PGEOM_SOLDERMASK_BOTTOM,    wxS( "Soldermask Bottom" ) },
        { LAYER_INFO::SUBCLASS::PGEOM_SOLDERMASK_TOP,       wxS( "Soldermask Top" ) },
        { LAYER_INFO::SUBCLASS::PGEOM_BODY_CENTER,          wxS( "Body Center" ) },
        { LAYER_INFO::SUBCLASS::PGEOM_SILKSCREEN_BOTTOM,    wxS( "Silkscreen Bottom" ) },
        { LAYER_INFO::SUBCLASS::PGEOM_SILKSCREEN_TOP,       wxS( "Silkscreen Top" ) },
        { LAYER_INFO::SUBCLASS::PGEOM_PAD_STACK_NAME,       wxS( "Pad Stack Name" ) },
        { LAYER_INFO::SUBCLASS::PGEOM_PIN_NUMBER,           wxS( "Pin Number" ) },
        { LAYER_INFO::SUBCLASS::PGEOM_PLACE_BOUND_BOTTOM,   wxS( "Place Bound Bottom" ) },
        { LAYER_INFO::SUBCLASS::PGEOM_PLACE_BOUND_TOP,      wxS( "Place Bound Top" ) },
        { LAYER_INFO::SUBCLASS::PGEOM_ASSEMBLY_BOTTOM,      wxS( "Assembly Bottom" ) },
        { LAYER_INFO::SUBCLASS::PGEOM_ASSEMBLY_TOP,         wxS( "Assembly Top" ) },
    };
 
    static const std::unordered_map<uint8_t, wxString> s_ManufacturingSubclassNames = {
        { LAYER_INFO::SUBCLASS::MFR_XSECTION_CHART,           wxS( "X-Section Chart" ) },
        { LAYER_INFO::SUBCLASS::MFR_NO_PROBE_BOTTOM,          wxS( "No Probe Bottom" ) },
        { LAYER_INFO::SUBCLASS::MFR_NO_PROBE_TOP,             wxS( "No Probe Top" ) },
        { LAYER_INFO::SUBCLASS::MFR_AUTOSILK_BOTTOM,          wxS( "AutoSilk Bottom" ) },
        { LAYER_INFO::SUBCLASS::MFR_AUTOSILK_TOP,             wxS( "AutoSilk Top" ) },
        { LAYER_INFO::SUBCLASS::MFR_PROBE_BOTTOM,             wxS( "Probe Bottom" ) },
        { LAYER_INFO::SUBCLASS::MFR_PROBE_TOP,                wxS( "Probe Top" ) },
        { LAYER_INFO::SUBCLASS::MFR_NCDRILL_FIGURE,           wxS( "NC Drill Figure" ) },
        { LAYER_INFO::SUBCLASS::MFR_NCDRILL_LEGEND,           wxS( "NC Drill Legend" ) },
        { LAYER_INFO::SUBCLASS::MFR_NO_GLOSS_INTERNAL,        wxS( "No Gloss Internal" ) },
        { LAYER_INFO::SUBCLASS::MFR_NO_GLOSS_BOTTOM,          wxS( "No Gloss Bottom" ) },
        { LAYER_INFO::SUBCLASS::MFR_NO_GLOSS_TOP,             wxS( "No Gloss Top" ) },
        { LAYER_INFO::SUBCLASS::MFR_NO_GLOSS_ALL,             wxS( "No Gloss All" ) },
        { LAYER_INFO::SUBCLASS::MFR_PHOTOPLOT_OUTLINE,        wxS( "Photoplot Outline" ) },
    };
 
    static const std::unordered_map<uint8_t, wxString> s_AnalysisSubclassNames = {
        { LAYER_INFO::SUBCLASS::ANALYSIS_PCB_TEMPERATURE,     wxS( "PCB Temperature" ) },
        { LAYER_INFO::SUBCLASS::ANALYSIS_HIGH_ISOCONTOUR,     wxS( "High IsoContour" ) },
        { LAYER_INFO::SUBCLASS::ANALYSIS_MEDIUM3_ISOCONTOUR,  wxS( "Medium3 IsoContour" ) },
        { LAYER_INFO::SUBCLASS::ANALYSIS_MEDIUM2_ISOCONTOUR,  wxS( "Medium2 IsoContour" ) },
        { LAYER_INFO::SUBCLASS::ANALYSIS_MEDIUM1_ISOCONTOUR,  wxS( "Medium1 IsoContour" ) },
        { LAYER_INFO::SUBCLASS::ANALYSIS_LOW_ISOCONTOUR,      wxS( "Low IsoContour" ) },
    };
 
    static const std::unordered_map<uint8_t, wxString> s_ConstraintSubclassNames = {
        { LAYER_INFO::SUBCLASS::CREG_ALL,                     wxS( "All" ) },
    };
 
    static const std::unordered_map<uint8_t, wxString> s_KeepinSubclassNames = {
        { LAYER_INFO::SUBCLASS::KEEPIN_ALL,                   wxS( "All" ) },
    };
 
    static const std::unordered_map<uint8_t, wxString> s_KeepoutSubclassNames = {
        { LAYER_INFO::SUBCLASS::KEEPOUT_ALL,                  wxS( "All" ) },
        { LAYER_INFO::SUBCLASS::KEEPOUT_TOP,                  wxS( "Top" ) },
        { LAYER_INFO::SUBCLASS::KEEPOUT_BOTTOM,               wxS( "Bottom" ) },
    };
 
    static const std::unordered_map<uint8_t, const std::unordered_map<uint8_t, wxString>&> s_SubclassNameMaps = {
        { LAYER_INFO::CLASS::BOARD_GEOMETRY,        s_BoardGeomSubclassNames },
 
        // These classes all share the same subclass names
        { LAYER_INFO::CLASS::COMPONENT_VALUE,       s_ComponentValueSubclassNames },
        { LAYER_INFO::CLASS::DEVICE_TYPE,           s_ComponentValueSubclassNames },
        { LAYER_INFO::CLASS::REF_DES,               s_ComponentValueSubclassNames },
        { LAYER_INFO::CLASS::TOLERANCE,             s_ComponentValueSubclassNames },
        { LAYER_INFO::CLASS::USER_PART_NUMBER,      s_ComponentValueSubclassNames },
 
        { LAYER_INFO::CLASS::DRAWING_FORMAT,        s_DrawingFormatSubclassNames },
        { LAYER_INFO::CLASS::PACKAGE_GEOMETRY,      s_PackageGeometrySubclassNames },
        { LAYER_INFO::CLASS::MANUFACTURING,         s_ManufacturingSubclassNames },
        { LAYER_INFO::CLASS::ANALYSIS,              s_AnalysisSubclassNames },
        { LAYER_INFO::CLASS::CONSTRAINTS_REGION,    s_ConstraintSubclassNames },
        { LAYER_INFO::CLASS::PACKAGE_KEEPIN,        s_KeepinSubclassNames },
        { LAYER_INFO::CLASS::PACKAGE_KEEPOUT,       s_KeepoutSubclassNames },
        { LAYER_INFO::CLASS::ROUTE_KEEPIN,          s_KeepinSubclassNames },
        { LAYER_INFO::CLASS::ROUTE_KEEPOUT,         s_KeepoutSubclassNames },
        { LAYER_INFO::CLASS::VIA_KEEPOUT,           s_KeepoutSubclassNames },
    };
    // clang-format on
 
    wxString   className;
    const auto classIt = s_ClassNames.find( aLayerInfo.m_Class );
 
    if( classIt != s_ClassNames.end() )
        className = classIt->second;
    else
        className = wxString::Format( wxS( "Class_%02X" ), aLayerInfo.m_Class );
 
    wxString subclassName;
 
    // Find the right subclass name map for this class
    auto classMapIt = s_SubclassNameMaps.find( aLayerInfo.m_Class );
 
    if( classMapIt != s_SubclassNameMaps.end() )
    {
        const std::unordered_map<uint8_t, wxString>& subclassMap = classMapIt->second;
 
        const auto subIt = subclassMap.find( aLayerInfo.m_Subclass );
 
        if( subIt != subclassMap.end() )
            subclassName = subIt->second;
        else
        {
            // This subclass seems not to have a known name
            subclassName = wxString::Format( wxS( "Subclass_%02X" ), aLayerInfo.m_Subclass );
        }
    }
    else
    {
        // Don't have a specific map for this class, just do a generic one.
        subclassName = wxString::Format( wxS( "Subclass_%02X" ), aLayerInfo.m_Subclass );
    }
 
    return className + wxS( "/" ) + subclassName;
}
 

static bool layerIsZone( const LAYER_INFO& aLayerInfo )
{
    if ( aLayerInfo.m_Class == LAYER_INFO::CLASS::CONSTRAINTS_REGION ||
         aLayerInfo.m_Class == LAYER_INFO::CLASS::BOUNDARY ||
         aLayerInfo.m_Class == LAYER_INFO::CLASS::PACKAGE_KEEPIN ||
         aLayerInfo.m_Class == LAYER_INFO::CLASS::ROUTE_KEEPIN ||
         aLayerInfo.m_Class == LAYER_INFO::CLASS::PACKAGE_KEEPOUT ||
         aLayerInfo.m_Class == LAYER_INFO::CLASS::ROUTE_KEEPOUT ||
         aLayerInfo.m_Class == LAYER_INFO::CLASS::VIA_KEEPOUT )
        return true;
 
    return false;
}
 

static std::optional<LAYER_INFO> tryLayerFromBlock( const BLOCK_BASE& aBlock )
{
    switch( aBlock.GetBlockType() )
    {
    case 0x0e:
    {
        const auto& net = BlockDataAs<BLK_0x0E_RECT>( aBlock );
        return net.m_Layer;
    }
    case 0x14:
    {
        const auto& trace = BlockDataAs<BLK_0x14_GRAPHIC>( aBlock );
        return trace.m_Layer;
    }
    case 0x24:
    {
        const auto& rect = BlockDataAs<BLK_0x24_RECT>( aBlock );
        return rect.m_Layer;
    }
    case 0x28:
    {
        const auto& shape = BlockDataAs<BLK_0x28_SHAPE>( aBlock );
        return shape.m_Layer;
    }
    }
 
    return std::nullopt;
}
 

LAYER_INFO expectLayerFromBlock( const BLOCK_BASE& aBlock )
{
    std::optional<LAYER_INFO> layerInfo = tryLayerFromBlock( aBlock );
 
    // Programming error - should only call this function if we're sure the block has layer info
    wxCHECK( layerInfo.has_value(), LAYER_INFO() );
 
    return layerInfo.value();
}
 

static PCB_LAYER_ID nthCopperLayerId( int aIndex, int aTotal )
{
    if( aIndex <= 0 )
        return F_Cu;
 
    if( aIndex >= aTotal - 1 )
        return B_Cu;
 
    return ToLAYER_ID( In1_Cu + ( aIndex - 1 ) * 2 );
}
 

static int64_t packPoint( int32_t aX, int32_t aY )
{
    return ( static_cast<int64_t>( static_cast<uint32_t>( aX ) ) << 32 )
           | static_cast<uint32_t>( aY );
}
 

class ALLEGRO::LAYER_MAPPER
{
    struct CUSTOM_LAYER
    {
        wxString m_Name;
        // LAYER_ARTWORK: POSITIVE/NEGATIVE
        // LAYER_USE: empty, EMBEDDED_PLANE, ...?
        // bool m_IsConductor;
    };
 
public:
    LAYER_MAPPER( const BRD_DB& aRawBoard, BOARD& aBoard, const LAYER_MAPPING_HANDLER& aLayerMappingHandler ) :
            m_layerMappingHandler( aLayerMappingHandler ),
            m_brdDb( aRawBoard ),
            m_board( aBoard )
    {}
 
    void ProcessLayerList( uint8_t aClass, const BLK_0x2A_LAYER_LIST& aList )
    {
        // If we haven't seen this list yet, create and store the CUSTOM_LAYER list
        if( m_Lists.count( &aList ) == 0 )
        {
            std::vector<CUSTOM_LAYER>& classLayers = m_Lists[&aList];
 
            if( aList.m_RefEntries.has_value() )
            {
                for( const BLK_0x2A_LAYER_LIST::REF_ENTRY& entry : aList.m_RefEntries.value() )
                {
                    const wxString& layerName = m_brdDb.GetString( entry.mLayerNameId );
 
                    classLayers.emplace_back( CUSTOM_LAYER( layerName ) );
                }
            }
            else if( aList.m_NonRefEntries.has_value() )
            {
                for( const BLK_0x2A_LAYER_LIST::NONREF_ENTRY& entry : aList.m_NonRefEntries.value() )
                {
                    classLayers.emplace_back( CUSTOM_LAYER( entry.m_Name ) );
                }
            }
            else
            {
                // Presumably a parsing error.
                THROW_IO_ERROR( "No ETCH layer list found." );
            }
 
            wxLogTrace( traceAllegroBuilder, "Added %zu layers for class %#04x, from 0x2A key %#010x",
                        classLayers.size(), aClass, aList.m_Key );
        }
 
        // Store the class ID -> 0x2A mapping
        m_ClassCustomLayerLists[aClass] = &m_Lists[&aList];
    }

    void FinalizeLayers()
    {
        auto customLayerIt = m_ClassCustomLayerLists.find( LAYER_INFO::CLASS::ETCH );
 
        if( customLayerIt == m_ClassCustomLayerLists.end() || !customLayerIt->second )
        {
            wxLogTrace( traceAllegroBuilder, "No ETCH layer class found; cannot finalize layers" );
            return;
        }
 
        const std::vector<CUSTOM_LAYER>& etchLayers = *customLayerIt->second;
        const size_t                     numCuLayers = etchLayers.size();
 
        m_board.GetDesignSettings().SetCopperLayerCount( numCuLayers );
 
        std::vector<INPUT_LAYER_DESC> inputLayers;
 
        for( size_t li = 0; li < numCuLayers; ++li )
        {
            INPUT_LAYER_DESC desc;
            desc.Name = etchLayers[li].m_Name;
            desc.AutoMapLayer = getNthCopperLayer( li, numCuLayers );
            desc.PermittedLayers = LSET::AllCuMask();
            desc.Required = true;
            inputLayers.push_back( desc );
        }
 
        // Add non-ETCH custom layers so they appear in the layer mapping dialog
        const std::vector<CUSTOM_LAYER>* etchList = m_ClassCustomLayerLists[LAYER_INFO::CLASS::ETCH];
        int nextAutoUser = 0;
 
        for( const auto& [classId, layerList] : m_ClassCustomLayerLists )
        {
            if( classId == LAYER_INFO::CLASS::ETCH || layerList == etchList )
                continue;
 
            for( size_t si = 0; si < layerList->size(); ++si )
            {
                const LAYER_INFO li{ classId, static_cast<uint8_t>( si ) };
 
                // Skip entries already covered by s_LayerKiMap
                if( s_LayerKiMap.count( li ) )
                    continue;
 
                INPUT_LAYER_DESC desc;
                desc.Name = layerInfoDisplayName( li );
 
                if( layerList->at( si ).m_Name.length() > 0 )
                    desc.Name = layerList->at( si ).m_Name;
 
                desc.AutoMapLayer = getNthUserLayer( nextAutoUser++ );
                desc.PermittedLayers = LSET::AllLayersMask();
                desc.Required = false;
                inputLayers.push_back( desc );
 
                m_customLayerDialogNames[li] = desc.Name;
            }
        }
 
        // The layers that maybe lump together multiple Allegro class:subclasses
        // into a single, named, KiCad layer
        for( const auto& [layerName, kiLayer] : m_MappedOptionalLayers )
        {
            INPUT_LAYER_DESC desc;
            desc.Name = layerName;
            desc.AutoMapLayer = kiLayer;
            desc.PermittedLayers = LSET::AllLayersMask();
            desc.Required = false;
            inputLayers.push_back( desc );
        }
 
        for( const auto& [layerInfo, kiLayer] : s_LayerKiMap )
        {
            INPUT_LAYER_DESC desc;
            desc.Name = layerInfoDisplayName( layerInfo );
            desc.AutoMapLayer = kiLayer;
            desc.PermittedLayers = LSET::AllLayersMask();
            desc.Required = false;
            inputLayers.push_back( desc );
        }
 
        std::map<wxString, PCB_LAYER_ID> resolvedMapping = m_layerMappingHandler( inputLayers );
 
        // Apply copper layer mapping
        for( size_t li = 0; li < numCuLayers; ++li )
        {
            const LAYER_INFO layerInfo{ LAYER_INFO::CLASS::ETCH, static_cast<uint8_t>( li ) };
            const wxString&  layerName = etchLayers[li].m_Name;
 
            auto it = resolvedMapping.find( layerName );
            PCB_LAYER_ID lId = ( it != resolvedMapping.end() ) ? it->second
                                                               : getNthCopperLayer( li, numCuLayers );
 
            m_customLayerToKiMap[layerInfo] = lId;
            m_board.SetLayerName( lId, layerName );
        }
 
        // Apply non-copper static layer mapping from the handler result
        for( const auto& [layerInfo, defaultKiLayer] : s_LayerKiMap )
        {
            const wxString displayName = layerInfoDisplayName( layerInfo );
 
            auto rmIt = resolvedMapping.find( displayName );
 
            if( rmIt != resolvedMapping.end() && rmIt->second != PCB_LAYER_ID::UNDEFINED_LAYER )
            {
                m_staticLayerOverrides[layerInfo] = rmIt->second;
            }
        }
 
        // Apply custom layer mapping from the handler result
        for( const auto& [layerInfo, dialogName] : m_customLayerDialogNames )
        {
            auto rmIt = resolvedMapping.find( dialogName );
 
            if( rmIt != resolvedMapping.end() && rmIt->second != PCB_LAYER_ID::UNDEFINED_LAYER )
            {
                m_customLayerToKiMap[layerInfo] = rmIt->second;
                m_board.SetLayerName( rmIt->second, dialogName );
            }
        }
 
        // Enable all the layers we ended up mapping to
        LSET enabledLayersMask = m_board.GetEnabledLayers();
        int userLayers = 0;
        for( const auto& [name, layerId] : resolvedMapping )
        {
            if( layerId != PCB_LAYER_ID::UNDEFINED_LAYER )
                enabledLayersMask |= LSET{ layerId };
 
            if( IsUserLayer( layerId ) )
                userLayers++;
 
            wxLogTrace( traceAllegroBuilder, "Mapping Allegro layer '%s' to KiCad layer '%s' (%d)", name,
                        m_board.GetLayerName( layerId ), layerId );
 
            m_board.SetLayerName( layerId, name );
        }
        m_board.SetEnabledLayers( enabledLayersMask );
        wxLogTrace( traceAllegroBuilder, "After mapping, there are %d user layers", userLayers );
        m_board.GetDesignSettings().SetUserDefinedLayerCount( userLayers );
    }
 
    PCB_LAYER_ID GetLayer( const LAYER_INFO& aLayerInfo )
    {
        // We already mapped and created the layer
        if( m_customLayerToKiMap.count( aLayerInfo ) )
            return m_customLayerToKiMap.at( aLayerInfo );
 
        // Check for user-remapped static layers first
        if( m_staticLayerOverrides.count( aLayerInfo ) )
            return m_staticLayerOverrides.at( aLayerInfo );
 
        // Next, have a look and see if the class:subclass was recorded as a custom layer
        if( m_ClassCustomLayerLists.count( aLayerInfo.m_Class ) )
        {
            const std::vector<CUSTOM_LAYER>* cLayerList = m_ClassCustomLayerLists.at( aLayerInfo.m_Class );
 
            // If it is using the copper layer list and the subclass is within the
            // copper layer range, return the mapped copper layer. Non-ETCH classes
            // can share the same layer list pointer, but their subclass values may
            // exceed the copper layer count and must fall through to the custom
            // layer mapping below.
            const auto etchIt = m_ClassCustomLayerLists.find( LAYER_INFO::CLASS::ETCH );
            if( etchIt != m_ClassCustomLayerLists.end()
                && cLayerList == etchIt->second
                && aLayerInfo.m_Subclass < cLayerList->size() )
            {
                const PCB_LAYER_ID cuLayer = getNthCopperLayer( aLayerInfo.m_Subclass, cLayerList->size() );
                // Remember this mapping
                m_customLayerToKiMap[aLayerInfo] = cuLayer;
                return cuLayer;
            }
 
            if( aLayerInfo.m_Subclass < cLayerList->size() )
            {
                // This subclass maps to a custom layer in this class
                const CUSTOM_LAYER& cLayer = cLayerList->at( aLayerInfo.m_Subclass );
                return MapCustomLayer( aLayerInfo, cLayer.m_Name );
            }
        }
 
        // Now, there may be layers that map to custom layers in KiCad, but are fixed in Allegro
        // (perhaps, DFA_BOUND_TOP), which means we won't find them in the layer lists.
        // We add them if we encounter them, with the names defined.
        if( s_OptionalFixedMappings.count( aLayerInfo ) )
        {
            const wxString& layerName = s_OptionalFixedMappings.at( aLayerInfo );
            return MapCustomLayer( aLayerInfo, layerName );
        }
 
        // Finally, fallback to the static mapping for any layers we haven't got a custom map for
        // We do this last so that it can be overridden for example if we want to remap
        // OUTLINE and DESIGN_OUTLINE to different layers.
        if( s_LayerKiMap.count( aLayerInfo ) )
            return s_LayerKiMap.at( aLayerInfo );
 
        // Keep a record of what we failed to map
        if( m_unknownLayers.count( aLayerInfo ) == 0 )
        {
            wxLogTrace( traceAllegroBuilder, "Failed to map class:subclass to layer: %#04x:%#04x", aLayerInfo.m_Class,
                        aLayerInfo.m_Subclass );
            m_unknownLayers[aLayerInfo] = 1;
        }
        m_unknownLayers[aLayerInfo]++;
 
        // Dump everything else here
        return m_unmappedLayer;
    }

    bool IsLayerMapped( PCB_LAYER_ID aLayerId ) const
    {
        return aLayerId != m_unmappedLayer;
    }

    PCB_LAYER_ID GetPlaceBounds( bool aTop )
    {
        const wxString name = aTop ? "PLACE_BOUND_TOP" : "PLACE_BOUND_BOTTOM";
        return mapCustomLayerByName( name );
    }

    bool IsOutlineLayer( const LAYER_INFO& aLayerInfo ) const
    {
        if( aLayerInfo.m_Class != LAYER_INFO::CLASS::BOARD_GEOMETRY
            && aLayerInfo.m_Class != LAYER_INFO::CLASS::DRAWING_FORMAT )
        {
            return false;
        }
 
        return aLayerInfo.m_Subclass == LAYER_INFO::SUBCLASS::DFMT_OUTLINE
               || aLayerInfo.m_Subclass == LAYER_INFO::SUBCLASS::BGEOM_OUTLINE;
    }

    PCB_LAYER_ID MapCustomLayer( const LAYER_INFO& aLayerInfo, const wxString& aLayerName )
    {
        // See if we have mapped this layer name under a different class:subclass
        if( m_MappedOptionalLayers.count( aLayerName ) )
        {
            const PCB_LAYER_ID existingLId = m_MappedOptionalLayers.at( aLayerName );
            // Record the reuse
            m_customLayerToKiMap[aLayerInfo] = existingLId;
            return existingLId;
        }
 
        // First time we needed this name:
        // Add as a user layer and store for next time
        const PCB_LAYER_ID lId = addUserLayer( aLayerName );
        m_customLayerToKiMap[aLayerInfo] = lId;
        m_MappedOptionalLayers[aLayerName] = lId;
 
        wxLogTrace( traceAllegroBuilder, "Adding mapping for %#04x:%#04x to %s", aLayerInfo.m_Class,
                    aLayerInfo.m_Subclass, aLayerName );
        return lId;
    }
 
 
    LSET GetRuleAreaLayers( const LAYER_INFO& aLayerInfo )
    {
        LSET layerSet{};
 
        switch( aLayerInfo.m_Class )
        {
        case LAYER_INFO::CLASS::ROUTE_KEEPOUT:
        case LAYER_INFO::CLASS::VIA_KEEPOUT:
        case LAYER_INFO::CLASS::PACKAGE_KEEPOUT:
        {
            switch( aLayerInfo.m_Subclass )
            {
            case LAYER_INFO::SUBCLASS::KEEPOUT_ALL:
                layerSet = LSET::AllCuMask();
                break;
            case LAYER_INFO::SUBCLASS::KEEPOUT_TOP:
                layerSet = LSET{ F_Cu };
                break;
            case LAYER_INFO::SUBCLASS::KEEPOUT_BOTTOM:
                layerSet = LSET{ B_Cu };
                break;
            default:
                layerSet = LSET{ GetLayer( aLayerInfo ) };
                break;
            }
            break;
        }
        case LAYER_INFO::CLASS::ROUTE_KEEPIN:
        case LAYER_INFO::CLASS::PACKAGE_KEEPIN:
        {
            // This can be ALL, but can it be anything else?
            if( aLayerInfo.m_Subclass == LAYER_INFO::SUBCLASS::KEEPOUT_ALL )
                layerSet = LSET::AllCuMask();
            else
                layerSet = LSET{ GetLayer( aLayerInfo ) };
            break;
        }
        default:
            wxLogTrace( traceAllegroBuilder, "  Unhandled non-copper zone layer class %#02x, using default layers",
                        aLayerInfo.m_Class );
            layerSet = LSET{ GetLayer( aLayerInfo ) };
            break;
        }
 
        return layerSet;
    }
 
private:
    static PCB_LAYER_ID getNthCopperLayer( int aNum, int aTotal )
    {
        return nthCopperLayerId( aNum, aTotal );
    }
 
    static PCB_LAYER_ID getNthUserLayer( int aNum )
    {
        aNum = std::min( aNum, MAX_USER_DEFINED_LAYERS - 1 );
        return ToLAYER_ID( static_cast<int>( User_1 ) + 2 * aNum );
    }

    PCB_LAYER_ID mapCustomLayerByName( const wxString& aLayerName )
    {
        // If it's been added already, use it
        if( m_MappedOptionalLayers.count( aLayerName ) )
        {
            return m_MappedOptionalLayers.at( aLayerName );
        }
 
        const PCB_LAYER_ID newLId = addUserLayer( aLayerName );
        m_MappedOptionalLayers[aLayerName] = newLId;
        return newLId;
    }
 
    PCB_LAYER_ID addUserLayer( const wxString& aName )
    {
        const PCB_LAYER_ID lId = getNthUserLayer( m_numUserLayersUsed++ );
        m_board.GetDesignSettings().SetUserDefinedLayerCount( m_board.GetDesignSettings().GetUserDefinedLayerCount() + 1 );
        m_board.SetLayerName( lId, aName );
        wxLogTrace( traceAllegroBuilder, "Adding user layer %s: %s", LayerName( lId ), aName );
        return lId;
    }
 
    // Map of original layer list - we use this to store the CUSTOM_LAYERs, as well
    // as check that we only handle each one once
    std::unordered_map<const BLK_0x2A_LAYER_LIST*, std::vector<CUSTOM_LAYER>> m_Lists;
 
    // Which classes point to which layer lists (more than one class can point to one list.
    std::unordered_map<uint8_t, std::vector<CUSTOM_LAYER>*> m_ClassCustomLayerLists;

    std::unordered_map<LAYER_INFO, PCB_LAYER_ID> m_customLayerToKiMap;

    std::unordered_map<wxString, PCB_LAYER_ID> m_MappedOptionalLayers;

    std::unordered_map<LAYER_INFO, PCB_LAYER_ID> m_staticLayerOverrides;

    std::unordered_map<LAYER_INFO, wxString> m_customLayerDialogNames;

    std::unordered_map<LAYER_INFO, int> m_unknownLayers;
 
    int m_numUserLayersUsed = 0;
 
    // The layer to use for mapping failures;
    PCB_LAYER_ID m_unmappedLayer = Cmts_User;
 
    const LAYER_MAPPING_HANDLER& m_layerMappingHandler;
 
    const BRD_DB& m_brdDb;
    BOARD&        m_board;
};
 
 
BOARD_BUILDER::BOARD_BUILDER( const BRD_DB& aRawBoard, BOARD& aBoard, REPORTER& aReporter,
                              PROGRESS_REPORTER* aProgressReporter,
                              const LAYER_MAPPING_HANDLER& aLayerMappingHandler ) :
        m_brdDb( aRawBoard ), m_board( aBoard ), m_reporter( aReporter ), m_progressReporter( aProgressReporter ),
        m_layerMappingHandler( aLayerMappingHandler ),
        m_layerMapper( std::make_unique<LAYER_MAPPER>( m_brdDb, m_board, m_layerMappingHandler ) )
{
    // Internal coordinates are stored in <base> / <divisor> units.
 
    const std::map<BOARD_UNITS, int> c_baseScales = { { BOARD_UNITS::MILS, pcbIUScale.MilsToIU( 1 ) },
                                                      { BOARD_UNITS::INCHES, pcbIUScale.MilsToIU( 1000 ) },
                                                      { BOARD_UNITS::MILLIMETERS, pcbIUScale.mmToIU( 1 ) },
                                                      { BOARD_UNITS::CENTIMETERS, pcbIUScale.mmToIU( 10 ) },
                                                      { BOARD_UNITS::MICROMETERS, pcbIUScale.mmToIU( 0.001 ) } };
 
    if( m_brdDb.m_Header->m_UnitsDivisor == 0 )
        THROW_IO_ERROR( "Board units divisor is 0" );
 
    if( !c_baseScales.contains( m_brdDb.m_Header->m_BoardUnits ) )
        THROW_IO_ERROR( "Unknown board units" );
 
    double baseScale( c_baseScales.at( m_brdDb.m_Header->m_BoardUnits ) );
 
    m_scale = baseScale / m_brdDb.m_Header->m_UnitsDivisor;
}
 

class BOARD_BUILDER::ZONE_FILL_HANDLER
{
public:
    struct FILL_INFO
    {
        ZONE*        m_Zone;
        PCB_LAYER_ID m_Layer;
        SHAPE_POLY_SET m_CombinedFill;
    };

    class COMPLEX_FIRST_FILL_TASK : public PRIORITY_THREAD_POOL_TASK<std::vector<FILL_INFO>>
    {
    public:
        COMPLEX_FIRST_FILL_TASK( bool aSimplify ) : m_simplify( aSimplify ) {}
 
    private:
        int computePriorityKey( const FILL_INFO& a ) const override
        {
            return static_cast<int>( a.m_CombinedFill.TotalVertices() );
        }
 
        size_t task( FILL_INFO& fillInfo ) override
        {
            SHAPE_POLY_SET finalFillPolys = *fillInfo.m_Zone->Outline();
 
            finalFillPolys.ClearArcs();
            fillInfo.m_CombinedFill.ClearArcs();
 
            // Intersect the zone outline with the combined fill that was assembled
            // from all the related objects.
            finalFillPolys.BooleanIntersection( fillInfo.m_CombinedFill );
            finalFillPolys.Fracture( m_simplify );
 
            // This is already mutex-ed, so this is safe
            fillInfo.m_Zone->SetFilledPolysList( fillInfo.m_Layer, finalFillPolys );
            return 1;
        }
 
        bool m_simplify;
    };

    void ProcessPolygons( bool aSimplify )
    {
        PROF_TIMER timer( "Zone fill processing" );
 
        COMPLEX_FIRST_FILL_TASK fillTask( aSimplify );
        fillTask.Execute( m_FillInfos );
 
        wxLogTrace( traceAllegroPerf, wxT( "   Intersected and fractured zone fills in %.3f ms" ), // format:allow
                    timer.msecs() );
    }
 
    void QueuePolygonForZone( ZONE& aZone, SHAPE_POLY_SET aFilledArea, PCB_LAYER_ID aLayer )
    {
        // Rule areas don't need filling
        if( aZone.GetIsRuleArea() )
            return;
 
        m_FillInfos.emplace_back( &aZone, aLayer, std::move( aFilledArea ) );
    }
 
private:
    std::vector<FILL_INFO> m_FillInfos;
};
 
 
BOARD_BUILDER::~BOARD_BUILDER()
{
}
 
 
static int clampForScale( double aValue )
{
    double result = std::round( aValue );
 
    if( result > std::numeric_limits<int>::max() )
        return std::numeric_limits<int>::max();
 
    if( result < std::numeric_limits<int>::min() )
        return std::numeric_limits<int>::min();
 
    return static_cast<int>( result );
}
 
 
VECTOR2I BOARD_BUILDER::scale( const VECTOR2I& aVector ) const
{
    return VECTOR2I{
        clampForScale( aVector.x * m_scale ),
        clampForScale( -aVector.y * m_scale ),
    };
}
 
int BOARD_BUILDER::scale( int aValue ) const
{
    return clampForScale( aValue * m_scale );
}
 
 
VECTOR2I BOARD_BUILDER::scaleSize( const VECTOR2I& aSize ) const
{
    return VECTOR2I{
        clampForScale( std::abs( aSize.x ) * m_scale ),
        clampForScale( std::abs( aSize.y ) * m_scale ),
    };
}
 
 
static EDA_ANGLE fromMillidegrees( uint32_t aMilliDegrees )
{
    return EDA_ANGLE{ static_cast<double>( aMilliDegrees ) / 1000.0, DEGREES_T };
}
 
 
void BOARD_BUILDER::reportMissingBlock( uint32_t aKey, uint8_t aType ) const
{
    m_reporter.Report( wxString::Format( "Could not find expected block with key %#010x and type %#04x", aKey, aType ),
                       RPT_SEVERITY_WARNING );
}
 
 
void BOARD_BUILDER::reportUnexpectedBlockType( uint8_t aGot, uint8_t aExpected, uint32_t aKey, size_t aOffset,
                                               const wxString& aName ) const
{
    wxString name = aName.IsEmpty() ? wxString( "Object" ) : aName;
    wxString withKey = ( aKey == 0 ) ? wxString( "" ) : wxString::Format( ", with key %#010x ", aKey );
    wxString withOffset = ( aOffset == 0 ) ? wxString( "" ) : wxString::Format( ", at offset %#zx ", aOffset );
 
    wxString s = wxString::Format( "%s has unexpected type %#04x (expected %#04x)%s%s", name, aGot, aExpected, withKey,
                                   withOffset );
 
    m_reporter.Report( s, RPT_SEVERITY_WARNING );
}
 
 
wxString BOARD_BUILDER::get0x30StringValue( uint32_t a0x30Key ) const
{
    const BLK_0x30_STR_WRAPPER* blk0x30 = expectBlockByKey<BLK_0x30_STR_WRAPPER>( a0x30Key );
 
    if( blk0x30 == nullptr )
        THROW_IO_ERROR( "Failed to get 0x30 for string lookup" );
 
    const BLK_0x31_SGRAPHIC* blk0x31 = expectBlockByKey<BLK_0x31_SGRAPHIC>( blk0x30->m_StrGraphicPtr );
 
    if( blk0x31 == nullptr )
        THROW_IO_ERROR( "Failed to get 0x31 for string lookup" );
 
    return blk0x31->m_Value;
}
 
 
void BOARD_BUILDER::cacheFontDefs()
{
    TYPED_LL_WALKER<BLK_0x36_DEF_TABLE> x36Walker{ m_brdDb.m_Header->m_LL_0x36.m_Head,
                                                   m_brdDb.m_Header->m_LL_0x36.m_Tail, m_brdDb };
 
    bool encountered = false;
 
    for( const BLK_0x36_DEF_TABLE& blk0x36 : x36Walker )
    {
        if( blk0x36.m_Code != 0x08 )
            continue;
 
        if( encountered )
        {
            // This would be bad, because we won't get the indexes into the list right if there
            // it's made up of entries from more than one list of entries.
            m_reporter.Report( "Found more than one font definition lists in the 0x36 list.", RPT_SEVERITY_WARNING );
            break;
        }
 
        for( const auto& item : blk0x36.m_Items )
        {
            const auto& fontDef = std::get<BLK_0x36_DEF_TABLE::FontDef_X08>( item );
            m_fontDefList.push_back( &fontDef );
        }
 
        encountered = true;
    }
}
 
 
void BOARD_BUILDER::createNets()
{
    wxLogTrace( traceAllegroBuilder, "Creating nets from Allegro data" );
 
    // Incrementing netcode. We could also choose to, say, use the 0x1B key if we wanted
    int netCode = 1;
 
    std::vector<BOARD_ITEM*> bulkAdded;
 
    TYPED_LL_WALKER<BLK_0x1B_NET> netWalker{
        m_brdDb.m_Header->m_LL_0x1B_Nets,
        m_brdDb,
        MISMATCH_POLICY::LOG_TRACE,
    };
 
    for( const BLK_0x1B_NET& netBlk : netWalker )
    {
        wxString netName = m_brdDb.GetString( netBlk.m_NetName );
 
        // Allegro allows unnamed nets. KiCad's NETINFO_LIST matches nets by name, and all
        // empty-named nets would collapse to the unconnected net (code 0). Generate a unique
        // name so each Allegro net gets its own KiCad net code.
        if( netName.IsEmpty() )
            netName = wxString::Format( wxS( "Net_%d" ), netCode );
 
        auto kiNetInfo = std::make_unique<NETINFO_ITEM>( &m_board, netName, netCode );
        netCode++;
 
        m_netCache[netBlk.m_Key] = kiNetInfo.get();
        bulkAdded.push_back( kiNetInfo.get() );
        m_board.Add( kiNetInfo.release(), ADD_MODE::BULK_APPEND );
    }
 
    m_board.FinalizeBulkAdd( bulkAdded );
 
    wxLogTrace( traceAllegroBuilder, "Added %zu nets", m_netCache.size() );
}
 
 
wxString BOARD_BUILDER::resolveConstraintSetNameFromField( uint32_t aFieldKey ) const
{
    const BLOCK_BASE* fieldBlock = m_brdDb.GetObjectByKey( aFieldKey );
 
    if( !fieldBlock || fieldBlock->GetBlockType() != 0x03 )
        return wxEmptyString;
 
    const BLK_0x03_FIELD& field = BlockDataAs<BLK_0x03_FIELD>( *fieldBlock );
    const std::string* str = std::get_if<std::string>( &field.m_Substruct );
 
    if( !str )
        return wxEmptyString;
 
    // Extract name from schematic cross-reference format: @lib.xxx(view):\NAME\.
    // Find the last colon-backslash separator in the raw std::string and extract from there.
    size_t sep = str->find( ":\\" );
 
    if( sep == std::string::npos )
        return wxEmptyString;
 
    std::string extracted = str->substr( sep + 2 );
 
    if( !extracted.empty() && extracted.back() == '\\' )
        extracted.pop_back();
 
    return wxString( extracted );
}
 
 
void BOARD_BUILDER::applyConstraintSets()
{
    wxLogTrace( traceAllegroBuilder, "Importing physical constraint sets from 0x1D blocks" );
 
    BOARD_DESIGN_SETTINGS& bds = m_board.GetDesignSettings();
    std::shared_ptr<NET_SETTINGS> netSettings = bds.m_NetSettings;
 
    bool isV172Plus = ( m_brdDb.m_FmtVer >= FMT_VER::V_172 );
 
    struct CS_DEF
    {
        wxString name;
        int      lineWidth = 0;
        int      clearance = 0;
        int      diffPairGap = 0;
    };
 
    // Map from constraint set name to its definition
    std::map<wxString, CS_DEF> constraintSets;
 
    // Also map string table keys to set names for net lookup
    std::map<uint32_t, wxString> keyToSetName;
 
    int                                      csIndex = 0;
    TYPED_LL_WALKER<BLK_0x1D_CONSTRAINT_SET> csWalker( m_brdDb.m_Header->m_LL_0x1D_0x1E_0x1F, m_brdDb );
 
    for( const BLK_0x1D_CONSTRAINT_SET& csBlock : csWalker )
    {
        wxString        setName;
        const wxString& resolved = m_brdDb.GetString( csBlock.m_NameStrKey );
 
        if( !resolved.IsEmpty() )
        {
            setName = resolved;
        }
        else if( csBlock.m_FieldPtr != 0 )
        {
            // Some boards store the name in a 0x03 FIELD block as a schematic cross-reference
            setName = resolveConstraintSetNameFromField( csBlock.m_FieldPtr );
        }
 
        if( setName.IsEmpty() )
            setName = wxString::Format( wxS( "CS_%d" ), csIndex );
 
        csIndex++;
 
        if( csBlock.m_DataB.empty() )
        {
            wxLogTrace( traceAllegroBuilder, "Constraint set '%s' has no DataB records, skipping", setName );
            continue;
        }
 
        // Parse first DataB record (first copper layer) as 14 x int32
        const auto& record = csBlock.m_DataB[0];
        int32_t fields[14];
        static_assert( sizeof( fields ) == std::tuple_size_v<std::decay_t<decltype( record )>> );
        memcpy( fields, record.data(), sizeof( fields ) );
 
        CS_DEF def;
        def.name = setName;
 
        if( isV172Plus )
        {
            def.lineWidth = scale( fields[1] );
            def.clearance = scale( fields[4] );
        }
        else
        {
            // Pre-V172: f[0] is preferred line width, f[1] is line spacing (used as clearance).
            // f[4] is sometimes also clearance when non-zero, but f[1] is the primary source.
            def.lineWidth = scale( fields[0] );
            def.clearance = scale( fields[1] );
        }
 
        def.diffPairGap = scale( fields[7] );
 
        constraintSets[setName] = def;
        keyToSetName[csBlock.m_NameStrKey] = setName;
 
        wxLogTrace( traceAllegroBuilder,
                    "Constraint set '%s': line_width=%d nm, clearance=%d nm, dp_gap=%d nm",
                    setName, def.lineWidth, def.clearance, def.diffPairGap );
    }
 
    if( constraintSets.empty() )
    {
        wxLogTrace( traceAllegroBuilder, "No physical constraint sets found" );
        return;
    }
 
    // Create a netclass for each constraint set that has nonzero values
    for( const auto& [name, def] : constraintSets )
    {
        wxString ncName = name;
 
        if( ncName.CmpNoCase( NETCLASS::Default ) == 0 )
            ncName = wxS( "Allegro_Default" );
 
        if( netSettings->HasNetclass( ncName ) )
            continue;
 
        auto nc = std::make_shared<NETCLASS>( ncName );
 
        if( def.lineWidth > 0 )
            nc->SetTrackWidth( def.lineWidth );
 
        if( def.clearance > 0 )
            nc->SetClearance( def.clearance );
 
        if( def.diffPairGap > 0 )
        {
            nc->SetDiffPairGap( def.diffPairGap );
 
            // Diff pair width is the same as track width for the pair's netclass
            if( def.lineWidth > 0 )
                nc->SetDiffPairWidth( def.lineWidth );
        }
 
        netSettings->SetNetclass( ncName, nc );
 
        wxLogTrace( traceAllegroBuilder, "Created netclass '%s' from constraint set '%s'", ncName, name );
    }
 
    // Walk all NETs and assign them to constraint set netclasses via field 0x1a0
    wxString defaultSetName;
 
    for( const auto& [name, def] : constraintSets )
    {
        if( name.CmpNoCase( wxS( "DEFAULT" ) ) == 0 )
        {
            defaultSetName = name;
            break;
        }
    }
 
    TYPED_LL_WALKER<BLK_0x1B_NET> csNetWalker{ m_brdDb.m_Header->m_LL_0x1B_Nets, m_brdDb };
 
    for( const BLK_0x1B_NET& netBlk : csNetWalker )
    {
        // Field 0x1a0 references the constraint set. It can be an integer (string table key
        // that matches 0x1D.m_NameStrKey) or a direct string (the constraint set name).
        auto csField =
                GetFirstFieldOfType( m_brdDb, netBlk.m_FieldsPtr, netBlk.m_Key, FIELD_KEYS::PHYS_CONSTRAINT_SET );
 
        wxString assignedSetName;
 
        if( csField.has_value() )
        {
            if( auto* intVal = std::get_if<uint32_t>( &csField.value() ) )
            {
                auto it = keyToSetName.find( *intVal );
 
                if( it != keyToSetName.end() )
                    assignedSetName = it->second;
            }
            else if( auto* strVal = std::get_if<wxString>( &csField.value() ) )
            {
                if( constraintSets.count( *strVal ) )
                    assignedSetName = *strVal;
            }
        }
 
        // Nets without field 0x1a0 use the DEFAULT constraint set
        if( assignedSetName.IsEmpty() && !defaultSetName.IsEmpty() )
            assignedSetName = defaultSetName;
 
        if( assignedSetName.IsEmpty() )
            continue;
 
        wxString ncName = assignedSetName;
 
        if( ncName.CmpNoCase( NETCLASS::Default ) == 0 )
            ncName = wxS( "Allegro_Default" );
 
        if( !netSettings->HasNetclass( ncName ) )
            continue;
 
        auto netIt = m_netCache.find( netBlk.m_Key );
 
        if( netIt == m_netCache.end() )
            continue;
 
        NETINFO_ITEM* kiNet = netIt->second;
        netSettings->SetNetclassPatternAssignment( kiNet->GetNetname(), ncName );
        kiNet->SetNetClass( netSettings->GetNetClassByName( ncName ) );
    }
 
    wxLogTrace( traceAllegroBuilder, "Applied %zu physical constraint sets", constraintSets.size() );
}
 
 
void BOARD_BUILDER::applyNetConstraints()
{
    wxLogTrace( traceAllegroBuilder, "Applying per-net trace width constraints" );
 
    BOARD_DESIGN_SETTINGS& bds = m_board.GetDesignSettings();
    std::shared_ptr<NET_SETTINGS> netSettings = bds.m_NetSettings;
 
    // Group nets by their minimum trace width to create netclasses.
    // Allegro stores per-net min/max trace width in FIELD blocks attached to each NET.
    std::map<int, std::vector<uint32_t>> widthToNetKeys;
 
    TYPED_LL_WALKER<BLK_0x1B_NET> widthNetWalker{ m_brdDb.m_Header->m_LL_0x1B_Nets, m_brdDb };
 
    for( const BLK_0x1B_NET& netBlk : widthNetWalker )
    {
        std::optional<int> minWidth =
                GetFirstFieldOfTypeInt( m_brdDb, netBlk.m_FieldsPtr, netBlk.m_Key, FIELD_KEYS::MIN_LINE_WIDTH );
 
        if( !minWidth.has_value() || minWidth.value() <= 0 )
            continue;
 
        int widthNm = scale( minWidth.value() );
        widthToNetKeys[widthNm].push_back( netBlk.m_Key );
    }
 
    if( widthToNetKeys.empty() )
    {
        wxLogTrace( traceAllegroBuilder, "No per-net trace width constraints found" );
        return;
    }
 
    for( const auto& [widthNm, netKeys] : widthToNetKeys )
    {
        int widthMils = ( widthNm + 12700 ) / 25400;
        wxString ncName = wxString::Format( wxS( "W%dmil" ), widthMils );
 
        if( netSettings->HasNetclass( ncName ) )
            continue;
 
        auto nc = std::make_shared<NETCLASS>( ncName );
        nc->SetTrackWidth( widthNm );
        netSettings->SetNetclass( ncName, nc );
 
        for( uint32_t netKey : netKeys )
        {
            auto it = m_netCache.find( netKey );
 
            if( it == m_netCache.end() )
                continue;
 
            NETINFO_ITEM* kiNet = it->second;
            netSettings->SetNetclassPatternAssignment( kiNet->GetNetname(), ncName );
            kiNet->SetNetClass( nc );
        }
 
        wxLogTrace( traceAllegroBuilder, "Created netclass '%s' (track width %d nm) with %zu nets",
                    ncName, widthNm, netKeys.size() );
    }
 
    wxLogTrace( traceAllegroBuilder, "Applied trace width constraints from %zu unique width groups",
                widthToNetKeys.size() );
}
 
 
wxString BOARD_BUILDER::resolveMatchGroupName( const BLK_0x1B_NET& aNet ) const
{
    if( aNet.m_MatchGroupPtr == 0 )
        return wxEmptyString;
 
    const BLOCK_BASE* block = m_brdDb.GetObjectByKey( aNet.m_MatchGroupPtr );
 
    if( !block )
        return wxEmptyString;
 
    uint32_t tableKey = 0;
 
    if( block->GetBlockType() == 0x26 )
    {
        // V172+ path: NET -> 0x26 -> m_GroupPtr -> 0x2C TABLE
        const auto& x26 = BlockDataAs<BLK_0x26_MATCH_GROUP>( *block );
        tableKey = x26.m_GroupPtr;
 
        // Some boards have chained 0x26 blocks (m_GroupPtr -> another 0x26 -> 0x2C)
        if( tableKey != 0 )
        {
            const BLOCK_BASE* next = m_brdDb.GetObjectByKey( tableKey );
 
            if( next && next->GetBlockType() == 0x26 )
            {
                const auto& x26b = BlockDataAs<BLK_0x26_MATCH_GROUP>( *next );
                tableKey = x26b.m_GroupPtr;
            }
        }
    }
    else if( block->GetBlockType() == 0x2C )
    {
        // Pre-V172 path: NET -> 0x2C TABLE directly
        tableKey = aNet.m_MatchGroupPtr;
    }
    else
    {
        return wxEmptyString;
    }
 
    if( tableKey == 0 )
        return wxEmptyString;
 
    // Verify the target is actually a 0x2C TABLE before calling expectBlockByKey to
    // avoid noisy warnings on boards with unexpected pointer chain configurations.
    const BLOCK_BASE* tableBlock = m_brdDb.GetObjectByKey( tableKey );
 
    if( !tableBlock || tableBlock->GetBlockType() != 0x2C )
        return wxEmptyString;
 
    const BLK_0x2C_TABLE* tbl = expectBlockByKey<BLK_0x2C_TABLE>( tableKey );
 
    if( !tbl || tbl->m_StringPtr == 0 )
        return wxEmptyString;
 
    const wxString& name = m_brdDb.GetString( tbl->m_StringPtr );
 
    wxLogTrace( traceAllegroBuilder, "Resolving match group name for NET '%s': found table at key %#010x, subtype %#x, name '%s'",
                m_brdDb.GetString( aNet.m_NetName ), tableKey, tbl->m_SubType, name );
 
    return name;
}
 
 
void BOARD_BUILDER::applyMatchGroups()
{
    wxLogTrace( traceAllegroBuilder, "Applying match group / differential pair assignments" );
 
    BOARD_DESIGN_SETTINGS& bds = m_board.GetDesignSettings();
    std::shared_ptr<NET_SETTINGS> netSettings = bds.m_NetSettings;
 
    // Group NET keys by their match group name
    std::map<wxString, std::vector<uint32_t>> groupToNetKeys;
 
    TYPED_LL_WALKER<BLK_0x1B_NET> netWalker{ m_brdDb.m_Header->m_LL_0x1B_Nets, m_brdDb };
 
    for( const BLK_0x1B_NET& netBlk : netWalker )
    {
        wxString groupName = resolveMatchGroupName( netBlk );
 
        if( groupName.empty() )
            continue;
 
        groupToNetKeys[groupName].push_back( netBlk.m_Key );
    }
 
    if( groupToNetKeys.empty() )
    {
        wxLogTrace( traceAllegroBuilder, "No match groups found" );
        return;
    }
 
    int dpCount = 0;
    int mgCount = 0;
 
    for( const auto& [groupName, netKeys] : groupToNetKeys )
    {
        // A diff pair has exactly 2 nets. We don't check P/N naming because Allegro doesn't
        // require any naming convention for paired nets.
        bool isDiffPair = ( netKeys.size() == 2 );
        wxString ncPrefix = isDiffPair ? wxS( "DP_" ) : wxS( "MG_" );
        wxString ncName = ncPrefix + groupName;
 
        if( netSettings->HasNetclass( ncName ) )
            continue;
 
        auto nc = std::make_shared<NETCLASS>( ncName );
 
        // Inherit constraint set values from the first net's current netclass so that
        // clearance and track width from the underlying constraint set are not lost.
        for( uint32_t netKey : netKeys )
        {
            auto it = m_netCache.find( netKey );
 
            if( it == m_netCache.end() )
                continue;
 
            NETCLASS* existing = it->second->GetNetClass();
 
            if( existing && existing->GetName() != NETCLASS::Default )
            {
                if( existing->HasClearance() )
                    nc->SetClearance( existing->GetClearance() );
 
                if( existing->HasTrackWidth() )
                    nc->SetTrackWidth( existing->GetTrackWidth() );
 
                if( existing->HasDiffPairGap() )
                    nc->SetDiffPairGap( existing->GetDiffPairGap() );
 
                if( existing->HasDiffPairWidth() )
                    nc->SetDiffPairWidth( existing->GetDiffPairWidth() );
 
                break;
            }
        }
 
        netSettings->SetNetclass( ncName, nc );
 
        for( uint32_t netKey : netKeys )
        {
            auto it = m_netCache.find( netKey );
 
            if( it == m_netCache.end() )
                continue;
 
            NETINFO_ITEM* kiNet = it->second;
            netSettings->SetNetclassPatternAssignment( kiNet->GetNetname(), ncName );
            kiNet->SetNetClass( nc );
        }
 
        if( isDiffPair )
            dpCount++;
        else
            mgCount++;
 
        wxLogTrace( traceAllegroBuilder, "%s group '%s' -> netclass '%s' with %zu nets",
                    isDiffPair ? wxS( "Diff pair" ) : wxS( "Match" ),
                    groupName, ncName, netKeys.size() );
    }
 
    wxLogTrace( traceAllegroBuilder,
                "Applied match groups: %d diff pairs, %d match groups (%zu total groups)",
                dpCount, mgCount, groupToNetKeys.size() );
}
 

static std::unordered_set<LAYER_INFO> ScanForLayers( const BRD_DB& aDb )
{
    std::unordered_set<LAYER_INFO> layersFound;
 
    const auto& addLayer = [&]( std::optional<LAYER_INFO>& info )
    {
        if( info.has_value() )
        {
            layersFound.insert( std::move( info.value() ) );
        }
    };
 
    const auto& simpleWalker = [&]( const FILE_HEADER::LINKED_LIST& aLL )
    {
        LL_WALKER walker{ aLL, aDb };
        for( const BLOCK_BASE* block : walker )
        {
            std::optional<LAYER_INFO> info = tryLayerFromBlock( *block );
            addLayer( info );
        }
    };
 
    simpleWalker( aDb.m_Header->m_LL_Shapes );
    simpleWalker( aDb.m_Header->m_LL_0x24_0x28 );
    simpleWalker( aDb.m_Header->m_LL_0x14 );
 
    return layersFound;
}
 
 
void BOARD_BUILDER::setupLayers()
{
    wxLogTrace( traceAllegroBuilder, "Setting up layer mapping from Allegro to KiCad" );
 
    const auto& layerMap = m_brdDb.m_Header->m_LayerMap;
 
    for( size_t i = 0; i < layerMap.size(); ++i )
    {
        const uint8_t classNum = static_cast<uint8_t>( i );
 
        const uint32_t x2aKey = layerMap[i].m_LayerList0x2A;
 
        if( x2aKey == 0 )
            continue;
 
        const BLK_0x2A_LAYER_LIST* layerList = expectBlockByKey<BLK_0x2A_LAYER_LIST>( x2aKey );
 
        // Probably an error
        if( !layerList )
            continue;
 
        m_layerMapper->ProcessLayerList( classNum, *layerList );
    }
 
    std::unordered_set<LAYER_INFO> layersFound = ScanForLayers( m_brdDb );
 
    wxLogTrace( traceAllegroBuilder, "Scanned %zu layers", layersFound.size() );
    for( const LAYER_INFO& info : layersFound )
    {
        wxLogTrace( traceAllegroBuilder, " - %#02x:%#02x (%s)", info.m_Class, info.m_Subclass,
                    layerInfoDisplayName( info ) );
    }
 
    // The outline is sometimes on OUTLINE and sometimes on DESIGN_OUTLINE, and sometimes
    // on both. In the first two cases, whichever it is goes to Edge.Cuts, but in the both case,
    // we send one to a User layer
    const LAYER_INFO outlineInfo{ LAYER_INFO::CLASS::BOARD_GEOMETRY, LAYER_INFO::SUBCLASS::BGEOM_OUTLINE };
    const LAYER_INFO designOutlineInfo{ LAYER_INFO::CLASS::BOARD_GEOMETRY, LAYER_INFO::SUBCLASS::BGEOM_DESIGN_OUTLINE };
 
    if( layersFound.count( outlineInfo ) && layersFound.count( designOutlineInfo ) )
    {
        // Both layers found, remap DESIGN_OUTLINE to a user layer
        wxLogTrace( traceAllegroBuilder,
                    "Both OUTLINE and DESIGN_OUTLINE layers found, remapping DESIGN_OUTLINE to a user layer" );
        m_layerMapper->MapCustomLayer( designOutlineInfo, layerInfoDisplayName( designOutlineInfo ) );
    }
 
    m_layerMapper->FinalizeLayers();
}
 
 
const BLK_0x36_DEF_TABLE::FontDef_X08* BOARD_BUILDER::getFontDef( unsigned aIndex ) const
{
    if( aIndex == 0 || aIndex > m_fontDefList.size() )
    {
        m_reporter.Report(
                wxString::Format( "Font def index %u requested, have %zu entries", aIndex, m_fontDefList.size() ),
                RPT_SEVERITY_WARNING );
        return nullptr;
    }
 
    // The index appears to be 1-indexed (maybe 0 means something special?)
    aIndex -= 1;
 
    return m_fontDefList[aIndex];
}
 
 
std::unique_ptr<PCB_SHAPE> BOARD_BUILDER::buildLineSegment( const BLK_0x15_16_17_SEGMENT& aSegment,
                                                            const LAYER_INFO& aLayerInfo, PCB_LAYER_ID aLayer,
                                                            BOARD_ITEM_CONTAINER& aParent )
{
    VECTOR2I  start = scale( { aSegment.m_StartX, aSegment.m_StartY } );
    VECTOR2I  end = scale( { aSegment.m_EndX, aSegment.m_EndY } );
    const int width = scale( aSegment.m_Width );
 
    if( !m_layerMapper->IsLayerMapped( aLayer ) )
    {
        wxLogTrace( traceAllegroBuilder, "Unmapped Seg: %#04x %#04x %s, %s", aLayerInfo.m_Class, aLayerInfo.m_Subclass,
                    start.Format(), end.Format() );
    }
 
    std::unique_ptr<PCB_SHAPE> shape = std::make_unique<PCB_SHAPE>( &aParent, SHAPE_T::SEGMENT );
    shape->SetLayer( aLayer );
    shape->SetStart( start );
    shape->SetEnd( end );
 
    {
        int adjustedWidth = width;
 
        if( adjustedWidth <= 0 )
            adjustedWidth = m_board.GetDesignSettings().GetLineThickness( aLayer );
 
        shape->SetWidth( adjustedWidth );
    }
 
    return shape;
}
 
 
std::unique_ptr<PCB_SHAPE> BOARD_BUILDER::buildArc( const BLK_0x01_ARC& aArc, const LAYER_INFO& aLayerInfo,
                                                    PCB_LAYER_ID aLayer, BOARD_ITEM_CONTAINER& aParent )
{
    VECTOR2I start{ aArc.m_StartX, aArc.m_StartY };
    VECTOR2I end{ aArc.m_EndX, aArc.m_EndY };
 
    std::unique_ptr<PCB_SHAPE> shape = std::make_unique<PCB_SHAPE>( &aParent, SHAPE_T::ARC );
 
    shape->SetLayer( aLayer );
 
    if( !m_layerMapper->IsLayerMapped( aLayer ) )
    {
        wxLogTrace( traceAllegroBuilder, "Unmapped Arc: %#04x %#04x %s, %s", aLayerInfo.m_Class, aLayerInfo.m_Subclass,
                    start.Format(), end.Format() );
    }
 
    start = scale( start );
    end = scale( end );
 
    VECTOR2I c = scale( KiROUND( VECTOR2D{ aArc.m_CenterX, aArc.m_CenterY } ) );
 
    int radius = scale( KiROUND( aArc.m_Radius ) );
 
    bool clockwise = ( aArc.m_SubType & 0x40 ) != 0;
 
    {
        int arcWidth = scale( aArc.m_Width );
 
        if( arcWidth <= 0 )
            arcWidth = m_board.GetDesignSettings().GetLineThickness( aLayer );
 
        shape->SetWidth( arcWidth );
    }
 
    if( start == end )
    {
        shape->SetShape( SHAPE_T::CIRCLE );
        shape->SetCenter( c );
        shape->SetRadius( radius );
    }
    else
    {
        shape->SetShape( SHAPE_T::ARC );
        EDA_ANGLE startangle( start - c );
        EDA_ANGLE endangle( end - c );
 
        startangle.Normalize();
        endangle.Normalize();
 
        EDA_ANGLE angle = endangle - startangle;
 
        if( clockwise && angle < ANGLE_0 )
            angle += ANGLE_360;
        if( !clockwise && angle > ANGLE_0 )
            angle -= ANGLE_360;
 
        if( start == end )
            angle = -ANGLE_360;
 
        VECTOR2I mid = start;
        RotatePoint( mid, c, -angle / 2.0 );
 
        shape->SetArcGeometry( start, mid, end );
    }
 
    return shape;
}
 
 
std::unique_ptr<PCB_TEXT> BOARD_BUILDER::buildPcbText( const BLK_0x30_STR_WRAPPER& aStrWrapper,
                                                       BOARD_ITEM_CONTAINER&       aParent )
{
    std::unique_ptr<PCB_TEXT> text = std::make_unique<PCB_TEXT>( &aParent );
 
    PCB_LAYER_ID layer = getLayer( aStrWrapper.m_Layer );
    text->SetLayer( layer );
 
    const BLK_0x31_SGRAPHIC* strGraphic = expectBlockByKey<BLK_0x31_SGRAPHIC>( aStrWrapper.m_StrGraphicPtr );
 
    if( !strGraphic )
    {
        m_reporter.Report( wxString::Format( "Failed to find string graphic (0x31) with key %#010x "
                                             "in string wrapper (0x30) with key %#010x",
                                             aStrWrapper.m_StrGraphicPtr, aStrWrapper.m_Key ),
                           RPT_SEVERITY_WARNING );
        return nullptr;
    }
 
    const BLK_0x30_STR_WRAPPER::TEXT_PROPERTIES* props = nullptr;
 
    if( aStrWrapper.m_Font.has_value() )
        props = &aStrWrapper.m_Font.value();
 
    if( !props && aStrWrapper.m_Font16x.has_value() )
        props = &aStrWrapper.m_Font16x.value();
 
    if( !props )
    {
        m_reporter.Report(
                wxString::Format( "Expected one of the font properties fields in 0x30 object (key %#010x) to be set.",
                                  aStrWrapper.m_Key ),
                RPT_SEVERITY_WARNING );
        return nullptr;
    }
 
    const BLK_0x36_DEF_TABLE::FontDef_X08* fontDef = getFontDef( props->m_Key );
 
    if( !fontDef )
        return nullptr;
 
    text->SetText( strGraphic->m_Value );
    text->SetTextWidth( scale( fontDef->m_CharWidth ) );
    text->SetTextHeight( scale( fontDef->m_CharHeight ) );
    text->SetTextThickness( std::max( 1, scale( fontDef->m_StrokeWidth ) ) );
 
    const EDA_ANGLE textAngle = fromMillidegrees( aStrWrapper.m_Rotation );
    text->SetTextAngle( textAngle );
 
    VECTOR2I textPos = scale( VECTOR2I{ aStrWrapper.m_CoordsX, aStrWrapper.m_CoordsY } );
 
    // KiCad's stroke font has a different baseline than Allegro's, so apply a vertical offset to compensate.
    // The exact offset is a bit of guesswork based on visually matching Allegro and KiCad text, but the
    // stoke font itself isn't the same anyway, so we can't be 100% here.
    VECTOR2I textFontOffset = VECTOR2I{ 0, -( scale( fontDef->m_CharHeight ) * 45 ) / 100 };
    RotatePoint( textFontOffset, textAngle );
    text->SetPosition( textPos + textFontOffset );
 
    if( props->m_Reversal == BLK_0x30_STR_WRAPPER::TEXT_REVERSAL::REVERSED )
        text->SetMirrored( true );
 
    switch( props->m_Alignment )
    {
    case BLK_0x30_STR_WRAPPER::TEXT_ALIGNMENT::LEFT:
        text->SetHorizJustify( GR_TEXT_H_ALIGN_LEFT );
        break;
    case BLK_0x30_STR_WRAPPER::TEXT_ALIGNMENT::CENTER:
        text->SetHorizJustify( GR_TEXT_H_ALIGN_CENTER );
        break;
    case BLK_0x30_STR_WRAPPER::TEXT_ALIGNMENT::RIGHT:
        text->SetHorizJustify( GR_TEXT_H_ALIGN_RIGHT );
        break;
    default:
        break;
    }
 
    return text;
}
 
 
std::vector<std::unique_ptr<BOARD_ITEM>> BOARD_BUILDER::buildDrillMarker( const BLK_0x0C_PIN_DEF& aPinDef,
                                                                          BOARD_ITEM_CONTAINER&   aParent )
{
    using MS = BLK_0x0C_PIN_DEF::MARKER_SHAPE;
    std::vector<std::unique_ptr<PCB_SHAPE>> shapes;
 
    const uint32_t markerShape = aPinDef.GetShape();
 
    PCB_LAYER_ID   layer = getLayer( aPinDef.m_Layer );
    const VECTOR2I center = scale( VECTOR2I{ aPinDef.m_Coords[0], aPinDef.m_Coords[1] } );
    const VECTOR2I size = scaleSize( VECTOR2I{ aPinDef.m_Size[0], aPinDef.m_Size[1] } );
 
    const auto addLine = [&]( const SEG& aSeg )
    {
        std::unique_ptr<PCB_SHAPE> shape = std::make_unique<PCB_SHAPE>( &aParent, SHAPE_T::SEGMENT );
        shape->SetStart( aSeg.A );
        shape->SetEnd( aSeg.B );
        shapes.push_back( std::move( shape ) );
    };
 
    const auto addPolyPts = [&]( const std::vector<VECTOR2I>& aPts )
    {
        std::unique_ptr<PCB_SHAPE> shape = std::make_unique<PCB_SHAPE>( &aParent, SHAPE_T::POLY );
        shape->SetPolyPoints( aPts );
        shapes.push_back( std::move( shape ) );
    };
 
    switch( markerShape )
    {
    case MS::CIRCLE:
    {
        std::unique_ptr<PCB_SHAPE> shape = std::make_unique<PCB_SHAPE>( &aParent, SHAPE_T::CIRCLE );
        shape->SetCenter( center );
        shape->SetRadius( size.x / 2 );
        shapes.push_back( std::move( shape ) );
        break;
    }
    case MS::SQUARE:
    case MS::RECTANGLE:
    {
        std::unique_ptr<PCB_SHAPE> shape = std::make_unique<PCB_SHAPE>( &aParent, SHAPE_T::RECTANGLE );
        shape->SetStart( center - size / 2 );
        shape->SetEnd( center + size / 2 );
        shapes.push_back( std::move( shape ) );
        break;
    }
    case MS::CROSS:
    {
        std::unique_ptr<PCB_SHAPE> shape;
 
        std::vector<SEG> segs = KIGEOM::MakeCrossSegments( center, size, ANGLE_0 );
 
        for( const SEG& seg : segs )
        {
            addLine( seg );
        }
        break;
    }
    case MS::OBLONG_X:
    case MS::OBLONG_Y:
    {
        std::unique_ptr<PCB_SHAPE> shape = std::make_unique<PCB_SHAPE>( &aParent, SHAPE_T::RECTANGLE );
        shape->SetStart( center - size / 2 );
        shape->SetEnd( center + size / 2 );
 
        int minSize = std::min( size.x, size.y );
        shape->SetCornerRadius( minSize / 2 );
        shapes.push_back( std::move( shape ) );
        break;
    }
    case MS::TRIANGLE:
    {
        // This triangle is point-up
        // Size follows fabmaster - the circumscribed circle of the triangle
        std::vector<VECTOR2I> pts = KIGEOM::MakeRegularPolygonPoints( center, 3, size.x / 2, true, ANGLE_90 );
        addPolyPts( pts );
        break;
    }
    case MS::DIAMOND:
    {
        std::vector<VECTOR2I> pts = KIGEOM::MakeRegularPolygonPoints( center, 4, size.x / 2, true, ANGLE_90 );
        addPolyPts( pts );
        break;
    }
    case MS::PENTAGON:
    {
        // Not 100% sure which way this should point
        std::vector<VECTOR2I> pts = KIGEOM::MakeRegularPolygonPoints( center, 5, size.x / 2, true, ANGLE_90 );
        addPolyPts( pts );
        break;
    }
    case MS::HEXAGON_X:
    case MS::HEXAGON_Y:
    {
        EDA_ANGLE             startAngle = ( markerShape == MS::HEXAGON_X ) ? ANGLE_0 : ANGLE_90;
        std::vector<VECTOR2I> pts = KIGEOM::MakeRegularPolygonPoints( center, 6, size.x / 2, true, startAngle );
        addPolyPts( pts );
        break;
    }
    case MS::OCTAGON:
    {
        EDA_ANGLE startAngle = FULL_CIRCLE / 16; // Start at 22.5 degrees to align flat sides with axes
        // Octagons are measured across flats
        std::vector<VECTOR2I> pts = KIGEOM::MakeRegularPolygonPoints( center, 8, size.x / 2, false, startAngle );
        addPolyPts( pts );
        break;
    }
    default:
    {
        wxLogTrace( traceAllegroBuilder, "Unsupported drill marker shape type %#04x for pin definition with key %#010x",
                    markerShape, aPinDef.m_Key );
        break;
    }
    }
 
    std::vector<std::unique_ptr<BOARD_ITEM>> items;
    for( std::unique_ptr<PCB_SHAPE>& shape : shapes )
    {
        shape->SetLayer( layer );
        shape->SetWidth( 0 );
 
        items.push_back( std::move( shape ) );
    }
 
    return items;
}
 
 
std::vector<std::unique_ptr<BOARD_ITEM>> BOARD_BUILDER::buildGraphicItems( const BLOCK_BASE&     aBlock,
                                                                           BOARD_ITEM_CONTAINER& aParent )
{
    std::vector<std::unique_ptr<BOARD_ITEM>> newItems;
 
    switch( aBlock.GetBlockType() )
    {
    case 0x0c:
    {
        const auto& pinDef = BlockDataAs<BLK_0x0C_PIN_DEF>( aBlock );
        newItems = buildDrillMarker( pinDef, aParent );
        break;
    }
    case 0x0e:
    {
        const auto& rect = BlockDataAs<BLK_0x0E_RECT>( aBlock );
 
        std::unique_ptr<PCB_SHAPE> shape = buildRect( rect, aParent );
        if( shape )
            newItems.push_back( std::move( shape ) );
        break;
    }
    case 0x14:
    {
        const auto& graphicContainer = BlockDataAs<BLK_0x14_GRAPHIC>( aBlock );
 
        std::vector<std::unique_ptr<PCB_SHAPE>> shapes = buildShapes( graphicContainer, aParent );
        for( std::unique_ptr<PCB_SHAPE>& shape : shapes )
            newItems.push_back( std::move( shape ) );
        break;
    }
    case 0x24:
    {
        const auto& rect = BlockDataAs<BLK_0x24_RECT>( aBlock );
 
        std::unique_ptr<PCB_SHAPE> shape = buildRect( rect, aParent );
        if( shape )
            newItems.push_back( std::move( shape ) );
        break;
    }
    case 0x28:
    {
        const auto&                shapeData = BlockDataAs<BLK_0x28_SHAPE>( aBlock );
        std::unique_ptr<PCB_SHAPE> shape = buildPolygon( shapeData, aParent );
        if( shape )
            newItems.push_back( std::move( shape ) );
        break;
    }
    case 0x30:
    {
        const auto& strWrapper = BlockDataAs<BLK_0x30_STR_WRAPPER>( aBlock );
 
        std::unique_ptr<BOARD_ITEM> newItem = buildPcbText( strWrapper, aParent );
        if( newItem )
            newItems.push_back( std::move( newItem ) );
        break;
    }
    default:
    {
        wxLogTrace( traceAllegroBuilder, "    Unhandled block type for buildItems: %#04x", aBlock.GetBlockType() );
        break;
    }
    }
 
    return newItems;
};
 
 
PCB_LAYER_ID BOARD_BUILDER::getLayer( const LAYER_INFO& aLayerInfo ) const
{
    return m_layerMapper->GetLayer( aLayerInfo );
}
 
 
std::vector<std::unique_ptr<PCB_SHAPE>> BOARD_BUILDER::buildShapes( const BLK_0x14_GRAPHIC&       aGraphic,
                                                                    BOARD_ITEM_CONTAINER& aParent )
{
    std::vector<std::unique_ptr<PCB_SHAPE>> shapes;
 
    PCB_LAYER_ID layer = getLayer( aGraphic.m_Layer );
 
    // Within the graphics list, we can get various lines and arcs on PLACE_BOUND_TOP, which
    // aren't actually the courtyard, which is a polygon in the 0x28 list. So, if we see such items,
    // remap them now to a specific other layer
    if( layer == F_CrtYd )
        layer = m_layerMapper->GetPlaceBounds( true );
    else if( layer == B_CrtYd )
        layer = m_layerMapper->GetPlaceBounds( false );
 
    const LL_WALKER segWalker{ aGraphic.m_SegmentPtr, aGraphic.m_Key, m_brdDb };
 
    for( const BLOCK_BASE* segBlock : segWalker )
    {
        std::unique_ptr<PCB_SHAPE> shape;
 
        switch( segBlock->GetBlockType() )
        {
        case 0x01:
        {
            const auto& arc = BlockDataAs<BLK_0x01_ARC>( *segBlock );
            shape = buildArc( arc, aGraphic.m_Layer, layer, aParent );
            break;
        }
        case 0x15:
        case 0x16:
        case 0x17:
        {
            const auto& seg = BlockDataAs<BLK_0x15_16_17_SEGMENT>( *segBlock );
            shape = buildLineSegment( seg, aGraphic.m_Layer, layer, aParent );
            break;
        }
        default:
        {
            wxLogTrace( traceAllegroBuilder, "    Unhandled block type in BLK_0x14_GRAPHIC: %#04x",
                        segBlock->GetBlockType() );
            break;
        }
        }
 
        if( shape )
            shapes.push_back( std::move( shape ) );
    }
 
    return shapes;
}
 
 
std::unique_ptr<PCB_SHAPE> BOARD_BUILDER::buildRect( const BLK_0x24_RECT& aRect, BOARD_ITEM_CONTAINER& aParent )
{
    std::unique_ptr<PCB_SHAPE> shape = std::make_unique<PCB_SHAPE>( &aParent );
 
    PCB_LAYER_ID layer = getLayer( aRect.m_Layer );
    shape->SetLayer( layer );
 
    shape->SetShape( SHAPE_T::RECTANGLE );
 
    const VECTOR2I cornerA = scale( VECTOR2I{ aRect.m_Coords[0], aRect.m_Coords[1] } );
    const VECTOR2I cornerB = scale( VECTOR2I{ aRect.m_Coords[2], aRect.m_Coords[3] } );
 
    shape->SetStart( cornerA );
    shape->SetEnd( cornerB );
 
    const EDA_ANGLE angle = fromMillidegrees( aRect.m_Rotation );
    shape->Rotate( cornerA, angle );
 
    const int lineWidth = 0;
    shape->SetWidth( lineWidth );
 
    return shape;
}
 
 
std::unique_ptr<PCB_SHAPE> BOARD_BUILDER::buildRect( const BLK_0x0E_RECT& aRect, BOARD_ITEM_CONTAINER& aParent )
{
    std::unique_ptr<PCB_SHAPE> shape = std::make_unique<PCB_SHAPE>( &aParent );
 
    PCB_LAYER_ID layer = getLayer( aRect.m_Layer );
    shape->SetLayer( layer );
 
    shape->SetShape( SHAPE_T::RECTANGLE );
 
    const VECTOR2I cornerA = scale( VECTOR2I{ aRect.m_Coords[0], aRect.m_Coords[1] } );
    const VECTOR2I cornerB = scale( VECTOR2I{ aRect.m_Coords[2], aRect.m_Coords[3] } );
 
    shape->SetStart( cornerA );
    shape->SetEnd( cornerB );
 
    const EDA_ANGLE angle = fromMillidegrees( aRect.m_Rotation );
    shape->Rotate( cornerA, angle );
 
    const int lineWidth = 0;
    shape->SetWidth( lineWidth );
 
    return shape;
}
 
 
std::unique_ptr<PCB_SHAPE> BOARD_BUILDER::buildPolygon( const BLK_0x28_SHAPE& aPolygon, BOARD_ITEM_CONTAINER& aParent )
{
    std::unique_ptr<PCB_SHAPE> shape = std::make_unique<PCB_SHAPE>( &aParent );
 
    PCB_LAYER_ID layer = getLayer( aPolygon.m_Layer );
    shape->SetLayer( layer );
 
    shape->SetShape( SHAPE_T::POLY );
 
    SHAPE_LINE_CHAIN chain = buildOutline( aPolygon );
 
    if( chain.PointCount() < 3 )
    {
        wxLogTrace( traceAllegroBuilder, "Polygon (0x28) with key %#010x has fewer than 3 points, skipping",
                    aPolygon.m_Key );
        return nullptr;
    }
 
    chain.SetClosed( true );
    shape->SetPolyShape( chain );
 
    const int lineWidth = 0;
    shape->SetWidth( lineWidth );
 
    return shape;
}
 
 
std::vector<std::unique_ptr<PCB_SHAPE>> BOARD_BUILDER::buildPolygonShapes( const BLK_0x28_SHAPE& aShapeData,
                                                                           BOARD_ITEM_CONTAINER& aParent )
{
    std::vector<std::unique_ptr<PCB_SHAPE>> shapes;
 
    PCB_LAYER_ID layer = getLayer( aShapeData.m_Layer );
 
    // Walk the segments in this shape and create PCB_SHAPE objects on Edge_Cuts
    const LL_WALKER segWalker{ aShapeData.m_FirstSegmentPtr, aShapeData.m_Key, m_brdDb };
 
    for( const BLOCK_BASE* segBlock : segWalker )
    {
        std::unique_ptr<PCB_SHAPE> shape = std::make_unique<PCB_SHAPE>( &m_board );
        shape->SetLayer( layer );
        shape->SetWidth( m_board.GetDesignSettings().GetLineThickness( layer ) );
 
        switch( segBlock->GetBlockType() )
        {
        case 0x01:
        {
            const auto& arc = BlockDataAs<BLK_0x01_ARC>( *segBlock );
 
            VECTOR2I start = scale( { arc.m_StartX, arc.m_StartY } );
            VECTOR2I end = scale( { arc.m_EndX, arc.m_EndY } );
            VECTOR2I c = scale( KiROUND( VECTOR2D{ arc.m_CenterX, arc.m_CenterY } ) );
 
            int radius = scale( arc.m_Radius );
            if( start == end )
            {
                shape->SetShape( SHAPE_T::CIRCLE );
                shape->SetCenter( c );
                shape->SetRadius( radius );
            }
            else
            {
                shape->SetShape( SHAPE_T::ARC );
 
                bool clockwise = ( arc.m_SubType & 0x40 ) != 0;
 
                EDA_ANGLE startangle( start - c );
                EDA_ANGLE endangle( end - c );
 
                startangle.Normalize();
                endangle.Normalize();
 
                EDA_ANGLE angle = endangle - startangle;
 
                if( clockwise && angle < ANGLE_0 )
                    angle += ANGLE_360;
 
                if( !clockwise && angle > ANGLE_0 )
                    angle -= ANGLE_360;
 
                VECTOR2I mid = start;
                RotatePoint( mid, c, -angle / 2.0 );
 
                shape->SetArcGeometry( start, mid, end );
            }
            break;
        }
        case 0x15:
        case 0x16:
        case 0x17:
        {
            const auto& seg = BlockDataAs<BLK_0x15_16_17_SEGMENT>( *segBlock );
            VECTOR2I    start = scale( { seg.m_StartX, seg.m_StartY } );
            VECTOR2I    end = scale( { seg.m_EndX, seg.m_EndY } );
 
            shape->SetShape( SHAPE_T::SEGMENT );
            shape->SetStart( start );
            shape->SetEnd( end );
            shape->SetWidth( m_board.GetDesignSettings().GetLineThickness( layer ) );
            break;
        }
        default:
            wxLogTrace( traceAllegroBuilder, "  Unhandled segment type in outline: %#04x", segBlock->GetBlockType() );
            continue;
        }
 
        shapes.push_back( std::move( shape ) );
    }
 
    return shapes;
}
 
 
const BLK_0x07_COMPONENT_INST* BOARD_BUILDER::getFpInstRef( const BLK_0x2D_FOOTPRINT_INST& aFpInstance ) const
{
    uint32_t refKey = 0x00;
 
    if( aFpInstance.m_InstRef.has_value() )
        refKey = aFpInstance.m_InstRef.value();
 
    if( !refKey && aFpInstance.m_InstRef16x.has_value() )
        refKey = aFpInstance.m_InstRef16x.value();
 
    // This can happen, for example for dimension "symbols".
    if( refKey == 0 )
        return nullptr;
 
    const BLK_0x07_COMPONENT_INST* blk07 = expectBlockByKey<BLK_0x07_COMPONENT_INST>( refKey );
    return blk07;
}
 
 
std::vector<std::unique_ptr<BOARD_ITEM>> BOARD_BUILDER::buildPadItems( const BLK_0x1C_PADSTACK& aPadstack,
                                                                       FOOTPRINT& aFp, const wxString& aPadName, int aNetcode )
{
    // Not all Allegro PADSTACKS can be represented by a single KiCad pad. For example, the
    // paste and mask layers can have completely independent shapes in Allegro, but in KiCad that
    // would require a separate aperture pad.
    // Also if there are multiple drills, we will need to make a pad for each
    std::vector<std::unique_ptr<BOARD_ITEM>> padItems;
 
    std::vector<std::unique_ptr<PADSTACK::COPPER_LAYER_PROPS>> copperLayers( aPadstack.GetLayerCount() );
 
    // Thermal relief gap from antipad/pad size difference on the first layer that has both.
    std::optional<int> thermalGap;
 
    const wxString& padStackName = m_brdDb.GetString( aPadstack.m_PadStr );
 
    wxLogTrace( traceAllegroBuilder, "Building pad '%s' with %u layers", padStackName, aPadstack.GetLayerCount() );
 
    // First, gather all the copper layers into a set of shape props, which we can then use to decide on the padstack mode
    for( size_t i = 0; i < aPadstack.GetLayerCount(); ++i )
    {
        const size_t layerBaseIndex = aPadstack.m_NumFixedCompEntries + i * aPadstack.m_NumCompsPerLayer;
        const ALLEGRO::PADSTACK_COMPONENT& padComp = aPadstack.m_Components[layerBaseIndex + BLK_0x1C_PADSTACK::LAYER_COMP_SLOT::PAD];
        const ALLEGRO::PADSTACK_COMPONENT& antiPadComp = aPadstack.m_Components[layerBaseIndex + BLK_0x1C_PADSTACK::LAYER_COMP_SLOT::ANTIPAD];
        const ALLEGRO::PADSTACK_COMPONENT& thermalComp = aPadstack.m_Components[layerBaseIndex + BLK_0x1C_PADSTACK::LAYER_COMP_SLOT::THERMAL_RELIEF];
 
        // If this is zero just skip entirely - I don't think we can usefully make pads with just thermal relief
        // Flag up if that happens.
        if( padComp.m_Type == PADSTACK_COMPONENT::TYPE_NULL)
        {
            if( antiPadComp.m_Type != PADSTACK_COMPONENT::TYPE_NULL )
            {
                m_reporter.Report(
                        wxString::Format( "Padstack %s: Copper layer %zu has no pad component, but has antipad",
                                          padStackName, i ),
                        RPT_SEVERITY_WARNING );
            }
            if( thermalComp.m_Type != PADSTACK_COMPONENT::TYPE_NULL )
            {
                m_reporter.Report(
                        wxString::Format( "Copper layer %zu has no pad component, but has thermal relief", i ),
                        RPT_SEVERITY_WARNING );
            }
            continue;
        }
 
        auto& layerCuProps = copperLayers[i];
        wxLogTrace( traceAllegroBuilder, "  Adding copper layer %zu with pad type %d", i, (int) padComp.m_Type );
        layerCuProps = std::make_unique<PADSTACK::COPPER_LAYER_PROPS>();
 
        switch( padComp.m_Type )
        {
        case PADSTACK_COMPONENT::TYPE_RECTANGLE:
            layerCuProps->shape.shape = PAD_SHAPE::RECTANGLE;
            layerCuProps->shape.size = scaleSize( VECTOR2I{ padComp.m_W, padComp.m_H } );
            layerCuProps->shape.offset = scale( VECTOR2I{ padComp.m_X3, padComp.m_X4 } );
            break;
        case PADSTACK_COMPONENT::TYPE_SQUARE:
            layerCuProps->shape.shape = PAD_SHAPE::RECTANGLE;
            layerCuProps->shape.size = scaleSize( VECTOR2I{ padComp.m_W, padComp.m_W } );
            layerCuProps->shape.offset = scale( VECTOR2I{ padComp.m_X3, padComp.m_X4 } );
            break;
        case PADSTACK_COMPONENT::TYPE_CIRCLE:
            layerCuProps->shape.shape = PAD_SHAPE::CIRCLE;
            layerCuProps->shape.size = scaleSize( VECTOR2I{ padComp.m_W, padComp.m_H } );
            layerCuProps->shape.offset = scale( VECTOR2I{ padComp.m_X3, padComp.m_X4 } );
            break;
        case PADSTACK_COMPONENT::TYPE_OBLONG_X:
        case PADSTACK_COMPONENT::TYPE_OBLONG_Y:
            layerCuProps->shape.shape = PAD_SHAPE::OVAL;
            layerCuProps->shape.size = scaleSize( VECTOR2I{ padComp.m_W, padComp.m_H } );
            layerCuProps->shape.offset = scale( VECTOR2I{ padComp.m_X3, padComp.m_X4 } );
            break;
        case PADSTACK_COMPONENT::TYPE_ROUNDED_RECTANGLE:
        {
            layerCuProps->shape.shape = PAD_SHAPE::ROUNDRECT;
            layerCuProps->shape.size = scaleSize( VECTOR2I{ padComp.m_W, padComp.m_H } );
            layerCuProps->shape.offset = scale( VECTOR2I{ padComp.m_X3, padComp.m_X4 } );
 
            int minDim = std::min( std::abs( padComp.m_W ), std::abs( padComp.m_H ) );
 
            if( padComp.m_Z1.has_value() && padComp.m_Z1.value() > 0 && minDim > 0 )
                layerCuProps->shape.round_rect_radius_ratio = padComp.m_Z1.value() / (double) minDim;
            else
                layerCuProps->shape.round_rect_radius_ratio = 0.25;
 
            break;
        }
        case PADSTACK_COMPONENT::TYPE_CHAMFERED_RECTANGLE:
        {
            layerCuProps->shape.shape = PAD_SHAPE::CHAMFERED_RECT;
            layerCuProps->shape.size = scaleSize( VECTOR2I{ padComp.m_W, padComp.m_H } );
            layerCuProps->shape.offset = scale( VECTOR2I{ padComp.m_X3, padComp.m_X4 } );
 
            int minDim = std::min( std::abs( padComp.m_W ), std::abs( padComp.m_H ) );
 
            if( padComp.m_Z1.has_value() && padComp.m_Z1.value() > 0 && minDim > 0 )
                layerCuProps->shape.chamfered_rect_ratio = padComp.m_Z1.value() / (double) minDim;
            else
                layerCuProps->shape.chamfered_rect_ratio = 0.25;
 
            layerCuProps->shape.chamfered_rect_positions = RECT_CHAMFER_ALL;
            break;
        }
        case PADSTACK_COMPONENT::TYPE_OCTAGON:
        {
            // Approximate octagon as a round rectangle with ~29.3% corner radius
            // (tan(22.5°) ≈ 0.414, half of that as ratio ≈ 0.207, but visually 0.293 is closer)
            layerCuProps->shape.shape = PAD_SHAPE::CHAMFERED_RECT;
            layerCuProps->shape.size = scaleSize( VECTOR2I{ padComp.m_W, padComp.m_H } );
            layerCuProps->shape.offset = scale( VECTOR2I{ padComp.m_X3, padComp.m_X4 } );
            layerCuProps->shape.chamfered_rect_ratio = 1.0 - 1.0 / sqrt( 2.0 );
            layerCuProps->shape.chamfered_rect_positions = RECT_CHAMFER_ALL;
            break;
        }
        case PADSTACK_COMPONENT::TYPE_SHAPE_SYMBOL:
        {
            // Custom shape defined by a 0x28 polygon. Walk the shape's segments and build
            // a polygon primitive for this pad.
            const BLK_0x28_SHAPE* shapeData = expectBlockByKey<BLK_0x28_SHAPE>( padComp.m_StrPtr );
 
            if( !shapeData )
            {
                wxLogTrace( traceAllegroBuilder,
                            "Padstack %s: SHAPE_SYMBOL on layer %zu has no 0x28 shape at %#010x",
                            padStackName, i, padComp.m_StrPtr );
                break;
            }
 
            SHAPE_LINE_CHAIN outline = buildSegmentChain( shapeData->m_FirstSegmentPtr );
 
            if( outline.PointCount() >= 3 )
            {
                outline.SetClosed( true );
 
                layerCuProps->shape.shape = PAD_SHAPE::CUSTOM;
                layerCuProps->shape.anchor_shape = PAD_SHAPE::CIRCLE;
 
                // Anchor size based on the shape's bounding box center
                BOX2I bbox = outline.BBox();
                int anchorSize = static_cast<int>(
                        std::min( bbox.GetWidth(), bbox.GetHeight() ) / 4 );
 
                if( anchorSize < 1 )
                    anchorSize = 1;
 
                layerCuProps->shape.size = VECTOR2I( anchorSize, anchorSize );
 
                auto poly = std::make_shared<PCB_SHAPE>( nullptr, SHAPE_T::POLY );
                poly->SetPolyShape( SHAPE_POLY_SET( outline ) );
                poly->SetFilled( true );
                poly->SetWidth( 0 );
                layerCuProps->custom_shapes.push_back( poly );
            }
            else
            {
                wxLogTrace( traceAllegroBuilder,
                            "Padstack %s: SHAPE_SYMBOL on layer %zu produced only %d points",
                            padStackName, i, outline.PointCount() );
            }
 
            break;
        }
        case PADSTACK_COMPONENT::TYPE_PENTAGON:
        {
            layerCuProps->shape.shape = PAD_SHAPE::CUSTOM;
            layerCuProps->shape.anchor_shape = PAD_SHAPE::CIRCLE;
            layerCuProps->shape.offset = scale( VECTOR2I{ padComp.m_X3, padComp.m_X4 } );
 
            const int w = std::max( padComp.m_W, 300 );
            const int h = std::max( padComp.m_H, 220 );
 
            SHAPE_LINE_CHAIN outline;
            auto             S = [&]( int x, int y )
            {
                return scale( VECTOR2I{ x, y } );
            };
 
            // Regular pentagon with flat bottom edge
            outline.Append( S( 0, -h / 2 ) );
            outline.Append( S( w / 2, -h / 6 ) );
            outline.Append( S( w / 3, h / 2 ) );
            outline.Append( S( -w / 3, h / 2 ) );
            outline.Append( S( -w / 2, -h / 6 ) );
            outline.SetClosed( true );
 
            BOX2I bbox = outline.BBox();
            int   anchorSize = static_cast<int>(
                    std::min( bbox.GetWidth(), bbox.GetHeight() ) / 7 );
 
            if( anchorSize < 1 )
                anchorSize = 1;
 
            layerCuProps->shape.size = VECTOR2I( anchorSize, anchorSize );
 
            auto poly = std::make_shared<PCB_SHAPE>( nullptr, SHAPE_T::POLY );
            poly->SetPolyShape( SHAPE_POLY_SET( outline ) );
            poly->SetFilled( true );
            poly->SetWidth( 0 );
            layerCuProps->custom_shapes.push_back( poly );
            break;
        }
        default:
            m_reporter.Report(
                    wxString::Format( "Padstack %s: unhandled copper pad shape type %d on layer %zu",
                                      padStackName, static_cast<int>( padComp.m_Type ), i ),
                    RPT_SEVERITY_WARNING );
            break;
        }
 
        if( antiPadComp.m_Type != PADSTACK_COMPONENT::TYPE_NULL)
        {
            if( antiPadComp.m_Type != padComp.m_Type )
            {
                wxLogTrace( traceAllegroBuilder, "Padstack %s: copper layer %zu antipad shape %d "
                            "differs from pad shape %d",
                            padStackName, i, antiPadComp.m_Type, padComp.m_Type );
            }
 
            int clearanceX = scale( ( antiPadComp.m_W - padComp.m_W ) / 2 );
            int clearanceY = scale( ( antiPadComp.m_H - padComp.m_H ) / 2 );
 
            if( clearanceX && clearanceX != clearanceY )
            {
                wxLogTrace( traceAllegroBuilder, "Padstack %s: copper layer %zu unequal antipad "
                            "clearance X=%d Y=%d",
                            padStackName, i, clearanceX, clearanceY );
            }
 
            if( antiPadComp.m_X3 != 0 || antiPadComp.m_X4 != 0 )
            {
                wxLogTrace( traceAllegroBuilder, "Padstack %s: copper layer %zu antipad offset "
                            "%d, %d",
                            padStackName, i, antiPadComp.m_X3, antiPadComp.m_X4 );
            }
 
            layerCuProps->clearance = clearanceX;
        }
 
        if( thermalComp.m_Type != PADSTACK_COMPONENT::TYPE_NULL && !thermalGap.has_value() )
        {
            // The thermal gap is the clearance between the pad copper and the surrounding zone.
            // We derive it from the antipad-to-pad size difference.
            if( antiPadComp.m_Type != PADSTACK_COMPONENT::TYPE_NULL && padComp.m_W > 0 )
            {
                int gap = scale( ( antiPadComp.m_W - padComp.m_W ) / 2 );
 
                if( gap > 0 )
                    thermalGap = gap;
            }
 
            wxLogTrace( traceAllegroBuilder,
                        "Padstack %s: thermal relief type=%d, gap=%snm",
                        padStackName, thermalComp.m_Type,
                        thermalGap.has_value() ? wxString::Format( "%d", thermalGap.value() )
                                               : wxString( "N/A" ) );
        }
 
        // Padstack-level keepouts (antipad relief geometry) are handled via the
        // antipad slot in copperLayers above. Board-level keepouts use BLK_0x34.
    }
 
    // We have now constructed a list of copper props. We can determine the PADSTACK mode now and assign the shapes
    PADSTACK padStack( &aFp );
 
    if( copperLayers.size() == 0 )
    {
        // SMD aperture PAD or something?
    }
    else
    {
        const auto layersEqual = [&](size_t aFrom, size_t aTo) -> bool
        {
            bool eq = true;
            for( size_t i = aFrom + 1; i < aTo; ++i )
            {
                if( !copperLayers[i - 1] || !copperLayers[i] || *copperLayers[i - 1] != *copperLayers[i] )
                {
                    eq = false;
                    break;
                }
            }
            return eq;
        };
 
        for(size_t i = 0; i < copperLayers.size(); ++i )
        {
            wxLogTrace( traceAllegroBuilder, "  Layer %zu: %s", i, copperLayers[i] ? "present" : "null" );
        }
 
        padStack.SetLayerSet( PAD::PTHMask() );
 
        if( copperLayers.front() && copperLayers.back() && layersEqual( 0, copperLayers.size() ) )
        {
            wxLogTrace( traceAllegroBuilder, "  Using NORMAL padstack mode (all layers identical)" );
            padStack.SetMode( PADSTACK::MODE::NORMAL );
            PADSTACK::COPPER_LAYER_PROPS& layerProps = padStack.CopperLayer( F_Cu );
            layerProps = *copperLayers.front();
        }
        else if( copperLayers.front() && copperLayers.back()
                 && layersEqual( 1, copperLayers.size() - 1 ) )
        {
            wxLogTrace( traceAllegroBuilder, "  Using FRONT_INNER_BACK padstack mode (inner layers identical)" );
            padStack.SetMode( PADSTACK::MODE::FRONT_INNER_BACK );
            padStack.CopperLayer( F_Cu ) = *copperLayers.front();
            padStack.CopperLayer( B_Cu ) = *copperLayers.back();
 
            // May be B_Cu if layers = 2, but that's OK
            if( copperLayers.size() > 2 && copperLayers[1] )
                padStack.CopperLayer( In1_Cu ) = *copperLayers[1];
        }
        else
        {
            wxLogTrace( traceAllegroBuilder, "  Using CUSTOM padstack mode (layers differ)" );
            padStack.SetMode( PADSTACK::MODE::CUSTOM );
 
            for( size_t i = 0; i < copperLayers.size(); ++i )
            {
                if( !copperLayers[i] )
                    continue;
 
                PCB_LAYER_ID layer = nthCopperLayerId( static_cast<int>( i ),
                                                       static_cast<int>( copperLayers.size() ) );
 
                padStack.CopperLayer( layer ) = *copperLayers[i];
            }
        }
    }
 
    // The drill/slot dimensions are extracted in priority order:
    //
    // 1. V172+ m_SlotAndUnknownArr[0] and [3] hold the true slot outline dimensions (X and Y).
    //    For routed slots (round drill bit + routing path), m_DrillArr only has the bit diameter,
    //    while m_SlotAndUnknownArr has the full slot envelope.
    //
    // 2. V172+ m_DrillArr[4] (width) and m_DrillArr[7] (height, 0 for round).
    //    These match m_SlotAndUnknownArr for punched oblong drills but only have the bit
    //    diameter for routed slots.
    //
    // 3. Pre-V172 m_Drill field (drill diameter, always round).
    int drillW = 0;
    int drillH = 0;
 
    if( std::holds_alternative<BLK_0x1C_PADSTACK::HEADER_v16x>( aPadstack.m_Header ) )
    {
        const auto& hdr16x = std::get<BLK_0x1C_PADSTACK::HEADER_v16x>( aPadstack.m_Header );
 
        if( hdr16x.m_SlotY > 0 )
        {
            drillW = scale( static_cast<int>( hdr16x.m_SlotX ) );
            drillH = scale( static_cast<int>( hdr16x.m_SlotY ) );
        }
        else
        {
            drillW = scale( static_cast<int>( aPadstack.GetDrillSize() ) );
        }
    }
    else
    {
        const auto& hdr17x = std::get<BLK_0x1C_PADSTACK::HEADER_v17x>( aPadstack.m_Header );
 
        if( hdr17x.m_SlotY > 0 )
        {
            drillW = scale( static_cast<int>( hdr17x.m_SlotX ) );
            drillH = scale( static_cast<int>( hdr17x.m_SlotY ) );
        }
        else
        {
            drillW = scale( static_cast<int>( hdr17x.m_DrillSize ) );
        }
    }
 
    if( drillH == 0 )
        drillH = drillW;
 
    // Allegro stores slot dimensions as (primary, secondary) regardless of orientation,
    // not as (X, Y). Compare the first copper layer pad's aspect ratio to determine if
    // the drill needs to be rotated 90 degrees.
    if( drillW != drillH && aPadstack.GetLayerCount() > 0 )
    {
        size_t firstCopperIdx = aPadstack.m_NumFixedCompEntries;
        const ALLEGRO::PADSTACK_COMPONENT& firstPadComp =
                aPadstack.m_Components[firstCopperIdx + BLK_0x1C_PADSTACK::LAYER_COMP_SLOT::PAD];
 
        bool padIsTaller = ( std::abs( firstPadComp.m_H ) > std::abs( firstPadComp.m_W ) );
        bool drillIsTaller = ( drillH > drillW );
 
        if( padIsTaller != drillIsTaller )
            std::swap( drillW, drillH );
    }
 
    bool isSmd = ( drillW == 0 ) || ( aPadstack.GetLayerCount() == 1 );
 
    if( isSmd )
    {
        padStack.Drill().size = VECTOR2I( 0, 0 );
    }
    else
    {
        padStack.Drill().size = VECTOR2I( drillW, drillH );
 
        if( drillW != drillH )
            padStack.Drill().shape = PAD_DRILL_SHAPE::OBLONG;
        else
            padStack.Drill().shape = PAD_DRILL_SHAPE::CIRCLE;
    }
 
    std::unique_ptr<PAD> pad = std::make_unique<PAD>( &aFp );
    pad->SetPadstack( padStack );
    pad->SetNumber( aPadName );
    pad->SetNetCode( aNetcode );
 
    if( isSmd )
    {
        pad->SetAttribute( PAD_ATTRIB::SMD );
        pad->SetLayerSet( PAD::SMDMask() );
    }
    else if( aPadstack.IsPlated() )
    {
        pad->SetAttribute( PAD_ATTRIB::PTH );
        pad->SetLayerSet( PAD::PTHMask() );
    }
    else
    {
        pad->SetAttribute( PAD_ATTRIB::NPTH );
        pad->SetLayerSet( PAD::UnplatedHoleMask() );
    }
 
    if( thermalGap.has_value() )
        pad->SetThermalGap( thermalGap.value() );
 
    padItems.push_back( std::move( pad ) );
 
    // Now, for each technical layer, we see if we can include it into the existing padstack, or if we need to add
    // it as a standalone pad
    for( size_t i = 0; i < aPadstack.m_NumFixedCompEntries; ++i )
    {
        const ALLEGRO::PADSTACK_COMPONENT& psComp = aPadstack.m_Components[i];

        if( psComp.m_Type == PADSTACK_COMPONENT::TYPE_NULL)
            continue;
 
        // All fixed slots are technical layers (solder mask, paste mask, film mask,
        // assembly variant, etc). Custom mask expansion extraction is not yet implemented;
        // KiCad's default pad-matches-mask behavior applies.
        wxLogTrace( traceAllegroBuilder,
                    "Fixed padstack slot %zu: type=%d, W=%d, H=%d",
                    i, static_cast<int>( psComp.m_Type ), psComp.m_W, psComp.m_H );
    }
 
    return padItems;
}
 
 
std::unique_ptr<FOOTPRINT> BOARD_BUILDER::buildFootprint( const BLK_0x2D_FOOTPRINT_INST& aFpInstance )
{
    auto fp = std::make_unique<FOOTPRINT>( &m_board );
 
    const BLK_0x07_COMPONENT_INST* fpInstData = getFpInstRef( aFpInstance );
 
    const bool backSide = ( aFpInstance.m_Layer != 0 );
 
    wxLogTrace( traceAllegroBuilder, "Building footprint from 0x2D block key %#010x", aFpInstance.m_Key );
 
    wxString refDesStr;
    if( fpInstData )
    {
        refDesStr = m_brdDb.GetString( fpInstData->m_RefDesStrPtr );
 
        if( refDesStr.IsEmpty() )
        {
            // Does this happen even when there's an 0x07 block?
            m_reporter.Report( wxString::Format( "Empty ref des for 0x2D key %#010x", aFpInstance.m_Key ),
                               RPT_SEVERITY_WARNING );
        }
    }
 
    // We may update the PCB_FIELD layer if it's specified explicitly (e.g. with font size and so on),
    // but if not, set the refdes at least, but make it invisible
    fp->SetReference( refDesStr );
    fp->GetField( FIELD_T::REFERENCE )->SetVisible( false );
 
    wxLogTrace( traceAllegroBuilder, "  Footprint reference: '%s'", refDesStr );
 
    const VECTOR2I  fpPos = scale( VECTOR2I{ aFpInstance.m_CoordX, aFpInstance.m_CoordY } );
 
    {
        EDA_ANGLE rotation = fromMillidegrees( aFpInstance.m_Rotation );
 
        if( backSide )
            rotation = ANGLE_180 - rotation;
 
        fp->SetPosition( fpPos );
        fp->SetOrientation( rotation );
    }
 
    // Allegro stores placed instance data in board-absolute form: bottom-side
    // components already have shapes on bottom layers with bottom-side positions.
    // Allegro stores placed footprints in board-absolute form with final layers.
    // KiCad stores footprints in canonical front-side form and uses Flip() to
    // mirror both positions and layers to the back side.
    //
    // Move back-layer items to their front-side counterpart so that fp->Flip()
    // consistently mirrors positions AND layers for all children. Without this,
    // bottom-side footprints would have their back-layer graphics double-flipped
    // to the front.
    //
    // Even if there isn't a layer flip, the postions still need to be flipped.
    const auto canonicalizeLayer = [backSide, fpPos]( BOARD_ITEM* aItem )
    {
        if( backSide )
            aItem->Flip( fpPos, FLIP_DIRECTION::LEFT_RIGHT );
    };
 
    TYPED_LL_WALKER<BLK_0x14_GRAPHIC> graphicsWalker{ aFpInstance.m_GraphicPtr, aFpInstance.m_Key, m_brdDb,
                                                      MISMATCH_POLICY::REPORT };
    graphicsWalker.SetMismatchReporter(
            [this]( uint8_t aType, const BLOCK_BASE& )
            {
                m_reporter.Report( wxString::Format( "Unexpected type in graphics list: %#04x", aType ),
                                   RPT_SEVERITY_WARNING );
            } );
 
    for( const BLK_0x14_GRAPHIC& graphics : graphicsWalker )
    {
        std::vector<std::unique_ptr<PCB_SHAPE>> shapes = buildShapes( graphics, *fp );
 
        for( std::unique_ptr<PCB_SHAPE>& shape : shapes )
        {
            canonicalizeLayer( shape.get() );
            fp->Add( shape.release() );
        }
    }
 
    bool valueFieldSet = false;
 
    TYPED_LL_WALKER<BLK_0x30_STR_WRAPPER> textWalker{ aFpInstance.m_TextPtr, aFpInstance.m_Key, m_brdDb,
                                                      MISMATCH_POLICY::LOG_TRACE };
 
    for( const BLK_0x30_STR_WRAPPER& strWrapper : textWalker )
    {
        std::unique_ptr<PCB_TEXT> text = buildPcbText( strWrapper, *fp );
 
        if( !text )
            continue;
 
        canonicalizeLayer( text.get() );
 
        const uint8_t textClass = strWrapper.m_Layer.m_Class;
        const uint8_t textSubclass = strWrapper.m_Layer.m_Subclass;
 
        bool isSilk = ( textSubclass == LAYER_INFO::SUBCLASS::SILKSCREEN_TOP
                        || textSubclass == LAYER_INFO::SUBCLASS::SILKSCREEN_BOTTOM );
        bool isAssembly = ( textSubclass == LAYER_INFO::SUBCLASS::ASSEMBLY_TOP
                            || textSubclass == LAYER_INFO::SUBCLASS::ASSEMBLY_BOTTOM );
 
        if( textClass == LAYER_INFO::CLASS::REF_DES && isSilk )
        {
            // Visible silkscreen refdes updates the built-in REFERENCE field.
            PCB_FIELD* const refDes = fp->GetField( FIELD_T::REFERENCE );
 
            // KiCad netlisting requires non-digit + digit annotation.
            if( !text->GetText().IsEmpty() && !wxIsalpha( text->GetText()[0] ) )
                text->SetText( wxString( "UNK" ) + text->GetText() );
 
            *refDes = PCB_FIELD( *text, FIELD_T::REFERENCE );
        }
        else if( textClass == LAYER_INFO::CLASS::REF_DES && isAssembly )
        {
            // Assembly refdes becomes a user field with the KiCad reference variable
            PCB_FIELD* field = new PCB_FIELD( *text, FIELD_T::USER, wxS( "Reference" ) );
            field->SetText( wxS( "${REFERENCE}" ) );
            field->SetVisible( false );
            fp->Add( field, ADD_MODE::APPEND );
        }
        else if( textClass == LAYER_INFO::CLASS::COMPONENT_VALUE && isAssembly )
        {
            if( !valueFieldSet )
            {
                // First COMPONENT_VALUE on assembly updates the built-in VALUE field
                PCB_FIELD* valField = fp->GetField( FIELD_T::VALUE );
                *valField = PCB_FIELD( *text, FIELD_T::VALUE );
                valField->SetVisible( false );
                valueFieldSet = true;
            }
            else
            {
                PCB_FIELD* field = new PCB_FIELD( *text, FIELD_T::USER, wxS( "Component Value" ) );
                field->SetVisible( false );
                fp->Add( field, ADD_MODE::APPEND );
            }
        }
        else if( textClass == LAYER_INFO::CLASS::DEVICE_TYPE )
        {
            PCB_FIELD* field = new PCB_FIELD( *text, FIELD_T::USER, wxS( "Device Type" ) );
            field->SetVisible( !isAssembly );
            fp->Add( field, ADD_MODE::APPEND );
        }
        else if( textClass == LAYER_INFO::CLASS::TOLERANCE )
        {
            PCB_FIELD* field = new PCB_FIELD( *text, FIELD_T::USER, wxS( "Tolerance" ) );
            field->SetVisible( isSilk );
            fp->Add( field, ADD_MODE::APPEND );
        }
        else if( textClass == LAYER_INFO::CLASS::USER_PART_NUMBER )
        {
            PCB_FIELD* field = new PCB_FIELD( *text, FIELD_T::USER, wxS( "User Part Number" ) );
            field->SetVisible( isSilk );
            fp->Add( field, ADD_MODE::APPEND );
        }
        else if( textClass == LAYER_INFO::CLASS::COMPONENT_VALUE && isSilk )
        {
            PCB_FIELD* field = new PCB_FIELD( *text, FIELD_T::USER, wxS( "Component Value" ) );
            field->SetVisible( true );
            fp->Add( field, ADD_MODE::APPEND );
        }
        else
        {
            fp->Add( text.release() );
        }
    }
 
    // Assembly drawing
    LL_WALKER assemblyWalker{ aFpInstance.m_AssemblyPtr, aFpInstance.m_Key, m_brdDb };
 
    for( const BLOCK_BASE* assemblyBlock : assemblyWalker )
    {
        std::vector<std::unique_ptr<BOARD_ITEM>> shapes = buildGraphicItems( *assemblyBlock, *fp );
 
        for( std::unique_ptr<BOARD_ITEM>& item : shapes )
        {
            canonicalizeLayer( item.get() );
            fp->Add( item.release() );
        }
    }
 
    // Areas (courtyards, etc)
    LL_WALKER areaWalker{ aFpInstance.m_AreasPtr, aFpInstance.m_Key, m_brdDb };
 
    // Probably don't need this as we can't have filled zone, but it's cheap
    ZONE_FILL_HANDLER zoneFillHandler;
 
    for( const BLOCK_BASE* areaBlock : areaWalker )
    {
        std::optional<LAYER_INFO> layerInfo = tryLayerFromBlock( *areaBlock );
 
        if( layerInfo.has_value() && layerIsZone( *layerInfo ) )
        {
            // Zone within a footprint - we can handle keepouts at least
            std::unique_ptr<ZONE> zone = buildZone( *areaBlock, {}, zoneFillHandler );
            if( zone )
            {
                canonicalizeLayer( zone.get() );
                fp->Add( zone.release() );
            }
        }
        else
        {
            std::vector<std::unique_ptr<BOARD_ITEM>> shapes = buildGraphicItems( *areaBlock, *fp );
 
            for( std::unique_ptr<BOARD_ITEM>& item : shapes )
            {
                canonicalizeLayer( item.get() );
 
                // If we find shapes in the areas list, they are (presumably) filled.
                // Maybe there's a flag to look at rather than just assuming this?
                if( item->Type() == PCB_SHAPE_T )
                {
                    PCB_SHAPE& shape = static_cast<PCB_SHAPE&>( *item );
 
                    // But in KiCad, courtyards are usually not filled even if they come in as "areas"
                    if( shape.GetLayer() != F_CrtYd && shape.GetLayer() != B_CrtYd )
                    {
                        shape.SetFilled( true );
                    }
                }
 
                fp->Add( item.release() );
            }
        }
    }
 
    // Find the pads
    TYPED_LL_WALKER<BLK_0x32_PLACED_PAD> padWalker{ aFpInstance.m_FirstPadPtr, aFpInstance.m_Key, m_brdDb,
                                                    MISMATCH_POLICY::THROW, PadGetNextInFootprint };
    for( const BLK_0x32_PLACED_PAD& placedPadInfo : padWalker )
    {
        const BLK_0x04_NET_ASSIGNMENT* netAssignment =
                expectBlockByKey<BLK_0x04_NET_ASSIGNMENT>( placedPadInfo.m_NetPtr );
        const BLK_0x0D_PAD* padInfo = expectBlockByKey<BLK_0x0D_PAD>( placedPadInfo.m_PadPtr );
 
        if( !padInfo )
            continue;
 
        const BLK_0x1C_PADSTACK* padStack = expectBlockByKey<BLK_0x1C_PADSTACK>( padInfo->m_PadStack );
 
        if( !padStack )
            continue;
 
        int netCode = NETINFO_LIST::UNCONNECTED;
 
        if( netAssignment )
        {
            auto netIt = m_netCache.find( netAssignment->m_Net );
            if( netIt != m_netCache.end() )
                netCode = netIt->second->GetNetCode();
        }
 
        const wxString padName = m_brdDb.GetString( padInfo->m_NameStrId );
 
        // 0x0D coordinates and rotation are in the footprint's local (unrotatesinced) space.
        // Use SetFPRelativePosition/Orientation to let KiCad handle the transform to
        // board-absolute coordinates (rotating by FP orientation and adding FP position).
        VECTOR2I  padLocalPos = scale( VECTOR2I{ padInfo->m_CoordsX, padInfo->m_CoordsY } );
        EDA_ANGLE padLocalRot = fromMillidegrees( padInfo->m_Rotation );
 
        // Unlike other items, pads in "canonical front side form" - a normal pad is on F.Cu already,
        // but the positions, like all the other items, are in board-absolute form, so we need to pre-transform
        // so that the final footprint flip puts them in the right place.
        if ( backSide )
        {
            RotatePoint( padLocalPos, ANGLE_180 );
            padLocalRot += ANGLE_180;
        }
 
        std::vector<std::unique_ptr<BOARD_ITEM>> padItems = buildPadItems( *padStack, *fp, padName, netCode );
 
        for( std::unique_ptr<BOARD_ITEM>& item : padItems )
        {
            if( item->Type() == PCB_PAD_T )
            {
                PAD* pad = static_cast<PAD*>( item.get() );
                pad->SetFPRelativeOrientation( padLocalRot );
            }
 
            item->SetFPRelativePosition( padLocalPos );
            fp->Add( item.release() );
        }
    }
 
    // Flip AFTER adding all children so that graphics, text, and pads all get
    // their layers and positions mirrored correctly for bottom-layer footprints.
    // We have carefully constructed a front-side canonical form by applying
    // pre-transforms to compensate for the coming Flip().
    if( backSide )
    {
        fp->Flip( fpPos, FLIP_DIRECTION::LEFT_RIGHT );
    }
 
    return fp;
}
 
std::vector<std::unique_ptr<BOARD_ITEM>> BOARD_BUILDER::buildTrack( const BLK_0x05_TRACK& aTrackBlock, int aNetCode )
{
    std::vector<std::unique_ptr<BOARD_ITEM>> items;
 
    // Anti-etch tracks are thermal relief patterns generated by Allegro.
    // These are handled by pad-level thermal relief properties instead.
    if( aTrackBlock.m_Layer.m_Class == LAYER_INFO::CLASS::ANTI_ETCH )
    {
        wxLogTrace( traceAllegroBuilder, "Skipping ANTI_ETCH track (class=%#04x, subclass=%#04x)",
                    aTrackBlock.m_Layer.m_Class, aTrackBlock.m_Layer.m_Subclass );
        return items;
    }
 
    const PCB_LAYER_ID layer = getLayer( aTrackBlock.m_Layer );
 
    LL_WALKER segWalker{ aTrackBlock.m_FirstSegPtr, aTrackBlock.m_Key, m_brdDb };
    for( const BLOCK_BASE* block : segWalker )
    {
        const uint8_t segType = block->GetBlockType();
 
        switch( segType )
        {
        case 0x15:
        case 0x16:
        case 0x17:
        {
            const BLK_0x15_16_17_SEGMENT& segInfo = BlockDataAs<BLK_0x15_16_17_SEGMENT>( *block );
 
            VECTOR2I start{ segInfo.m_StartX, segInfo.m_StartY };
            VECTOR2I end{ segInfo.m_EndX, segInfo.m_EndY };
            int      width = static_cast<int>( segInfo.m_Width );
 
            std::unique_ptr<PCB_TRACK> seg = std::make_unique<PCB_TRACK>( &m_board );
 
            seg->SetNetCode( aNetCode );
            seg->SetLayer( layer );
 
            seg->SetStart( scale( start ) );
            seg->SetEnd( scale( end ) );
            seg->SetWidth( scale( width ) );
 
            items.push_back( std::move( seg ) );
            break;
        }
        case 0x01:
        {
            const BLK_0x01_ARC& arcInfo = BlockDataAs<BLK_0x01_ARC>( *block );
 
            VECTOR2I start = scale( { arcInfo.m_StartX, arcInfo.m_StartY } );
            VECTOR2I end = scale( { arcInfo.m_EndX, arcInfo.m_EndY } );
            VECTOR2I c = scale( KiROUND( VECTOR2D{ arcInfo.m_CenterX, arcInfo.m_CenterY } ) );
            int      width = scale( static_cast<int>( arcInfo.m_Width ) );
 
            bool clockwise = ( arcInfo.m_SubType & 0x40 ) != 0;
 
            EDA_ANGLE startAngle( start - c );
            EDA_ANGLE endAngle( end - c );
            startAngle.Normalize();
            endAngle.Normalize();
 
            EDA_ANGLE angle = endAngle - startAngle;
 
            if( clockwise && angle < ANGLE_0 )
                angle += ANGLE_360;
 
            if( !clockwise && angle > ANGLE_0 )
                angle -= ANGLE_360;
 
            VECTOR2I mid = start;
            RotatePoint( mid, c, -angle / 2.0 );
 
            std::unique_ptr<PCB_ARC> arc = std::make_unique<PCB_ARC>( &m_board );
 
            arc->SetNetCode( aNetCode );
            arc->SetLayer( layer );
 
            arc->SetStart( start );
            arc->SetMid( mid );
            arc->SetEnd( end );
            arc->SetWidth( width );
 
            items.push_back( std::move( arc ) );
            break;
        }
        default:
            wxLogTrace( traceAllegroBuilder, "Unhandled segment type in track: %#04x", segType );
            break;
        }
    }
    return items;
}
 

static bool parseViaSpanFromName( const wxString& aName, int aSpanLen, int aTotalCu, int& aBegin,
                                  int& aEnd )
{
    // Bounded digit runs keep std::stoi in range; the non-digit boundaries stop a span from
    // matching inside a longer number such as a date or dimension.
    static const std::regex spanRe( R"((?:^|\D)(\d{1,2})-(\d{1,2})(?:\D|$))" );
 
    const std::string name = aName.ToStdString();
 
    for( std::sregex_iterator it( name.begin(), name.end(), spanRe ), last; it != last; ++it )
    {
        const int begin = std::stoi( ( *it )[1].str() );
        const int end = std::stoi( ( *it )[2].str() );
 
        if( begin >= 1 && begin < end && end <= aTotalCu && ( end - begin + 1 ) == aSpanLen )
        {
            aBegin = begin;
            aEnd = end;
            return true;
        }
    }
 
    return false;
}
 
 
std::optional<std::pair<int, int>> BOARD_BUILDER::resolveViaSpan( const BLK_0x1C_PADSTACK& aPadstack,
                                                                  const VECTOR2I& aViaCenterRaw,
                                                                  int aTotalCu, bool& aResolved )
{
    // Only a span read from the padstack or a fully connected via is "resolved" and cacheable; an
    // undetermined via stays unresolved so a better-connected sibling can still settle the padstack.
    aResolved = true;
 
    // The padstack stores only how many layers the via crosses, not where it starts. A span
    // reaching the whole stack is a through via.
    const int spanLen = static_cast<int>( aPadstack.GetLayerCount() );
 
    if( spanLen <= 1 || spanLen >= aTotalCu )
        return std::nullopt;
 
    // The name is the only signal that separates stacked microvias, whose shared location makes
    // their connected copper ambiguous, so prefer it.
    const wxString name = m_brdDb.GetString( aPadstack.m_PadStr );
    int            begin = 0;
    int            end = 0;
 
    if( parseViaSpanFromName( name, spanLen, aTotalCu, begin, end ) )
        return std::make_pair( begin - 1, end - 1 );
 
    // Otherwise the tracks meeting at the via centre bound its span. Collect them on first need so
    // boards with no name-less blind/buried vias never pay for the extra walk.
    if( !m_trackEndpointsCollected )
    {
        collectTrackEndpointCopper();
        m_trackEndpointsCollected = true;
    }
 
    auto it = m_trackEndpointCopper.find( packPoint( aViaCenterRaw.x, aViaCenterRaw.y ) );
 
    if( it != m_trackEndpointCopper.end() )
    {
        const auto& [minCu, maxCu] = it->second;
 
        // Accept only when the connected layer range matches the padstack's layer count.
        if( maxCu > minCu && ( maxCu - minCu + 1 ) == spanLen )
            return std::make_pair( minCu, maxCu );
    }
 
    // Neither source determined the span; leave it a through via rather than guess.
    aResolved = false;
 
    m_reporter.Report(
            wxString::Format( "Via padstack '%s' crosses %d of %d copper layers but its span could "
                              "not be determined; importing as a through via.",
                              name, spanLen, aTotalCu ),
            RPT_SEVERITY_WARNING );
 
    return std::nullopt;
}
 
 
void BOARD_BUILDER::collectTrackEndpointCopper()
{
    const int totalCu = m_board.GetCopperLayerCount();
 
    auto record = [this]( int32_t aX, int32_t aY, int aCopper )
    {
        auto [it, inserted] = m_trackEndpointCopper.try_emplace( packPoint( aX, aY ), aCopper, aCopper );
 
        if( !inserted )
        {
            it->second.first = std::min( it->second.first, aCopper );
            it->second.second = std::max( it->second.second, aCopper );
        }
    };
 
    TYPED_LL_WALKER<BLK_0x1B_NET> netWalker{ m_brdDb.m_Header->m_LL_0x1B_Nets, m_brdDb,
                                             MISMATCH_POLICY::SKIP };
 
    for( const BLK_0x1B_NET& net : netWalker )
    {
        TYPED_LL_WALKER<BLK_0x04_NET_ASSIGNMENT> assignmentWalker{ net.m_Assignment, net.m_Key,
                                                                   m_brdDb, MISMATCH_POLICY::SKIP };
 
        for( const BLK_0x04_NET_ASSIGNMENT& assign : assignmentWalker )
        {
            LL_WALKER connWalker{ assign.m_ConnItem, assign.m_Key, m_brdDb };
 
            for( const BLOCK_BASE* connItemBlock : connWalker )
            {
                if( connItemBlock->GetBlockType() != 0x05 )
                    continue;
 
                const BLK_0x05_TRACK& track = BlockDataAs<BLK_0x05_TRACK>( *connItemBlock );
 
                // Only etch tracks carry a copper-layer index in their subclass.
                if( track.m_Layer.m_Class != LAYER_INFO::CLASS::ETCH )
                    continue;
 
                const int copper = track.m_Layer.m_Subclass;
 
                if( copper < 0 || copper >= totalCu )
                    continue;
 
                LL_WALKER segWalker{ track.m_FirstSegPtr, track.m_Key, m_brdDb };
 
                for( const BLOCK_BASE* segBlock : segWalker )
                {
                    switch( segBlock->GetBlockType() )
                    {
                    case 0x15:
                    case 0x16:
                    case 0x17:
                    {
                        const auto& seg = BlockDataAs<BLK_0x15_16_17_SEGMENT>( *segBlock );
                        record( seg.m_StartX, seg.m_StartY, copper );
                        record( seg.m_EndX, seg.m_EndY, copper );
                        break;
                    }
                    case 0x01:
                    {
                        // Arc tracks join vias at their endpoints just as line segments do.
                        const auto& arc = BlockDataAs<BLK_0x01_ARC>( *segBlock );
                        record( arc.m_StartX, arc.m_StartY, copper );
                        record( arc.m_EndX, arc.m_EndY, copper );
                        break;
                    }
                    default:
                        break;
                    }
                }
            }
        }
    }
 
    wxLogTrace( traceAllegroBuilder, "Collected %zu track endpoints for via span resolution",
                m_trackEndpointCopper.size() );
}
 
 
std::unique_ptr<BOARD_ITEM> BOARD_BUILDER::buildVia( const BLK_0x33_VIA& aViaData, int aNetCode )
{
    VECTOR2I viaPos{ aViaData.m_CoordsX, aViaData.m_CoordsY };
 
    const BLK_0x1C_PADSTACK* viaPadstack = expectBlockByKey<BLK_0x1C_PADSTACK>( aViaData.m_Padstack );
 
    if( !viaPadstack )
        return nullptr;
 
    std::unique_ptr<PCB_VIA> via = std::make_unique<PCB_VIA>( &m_board );
    via->SetPosition( scale( viaPos ) );
    via->SetNetCode( aNetCode );
 
    // The via size comes from the first copper layer's pad component.
    int viaWidth = 0;
 
    if( viaPadstack->GetLayerCount() > 0 )
    {
        const ALLEGRO::PADSTACK_COMPONENT& padComp =
                viaPadstack->m_Components[viaPadstack->m_NumFixedCompEntries
                                         + BLK_0x1C_PADSTACK::LAYER_COMP_SLOT::PAD];
 
        if( padComp.m_Type != PADSTACK_COMPONENT::TYPE_NULL )
            viaWidth = scale( padComp.m_W );
    }
 
    // The span is a padstack property, so resolve it once per padstack key and cache it.
    const int totalCu = m_board.GetCopperLayerCount();
 
    std::optional<std::pair<int, int>> span;
 
    if( auto cacheIt = m_viaSpanCache.find( viaPadstack->m_Key ); cacheIt != m_viaSpanCache.end() )
    {
        span = cacheIt->second;
    }
    else
    {
        bool resolved = false;
        span = resolveViaSpan( *viaPadstack, viaPos, totalCu, resolved );
 
        // Cache only a resolved span, so a better-connected instance can still settle one this
        // instance could not.
        if( resolved )
            m_viaSpanCache.emplace( viaPadstack->m_Key, span );
    }
 
    if( span )
    {
        // Set the type before the layer pair: SanitizeLayers() snaps a THROUGH via back to
        // F_Cu/B_Cu and would discard the span. Reaching an outer layer makes it BLIND, else BURIED.
        const bool touchesOuter = ( span->first == 0 ) || ( span->second == totalCu - 1 );
 
        via->SetViaType( touchesOuter ? VIATYPE::BLIND : VIATYPE::BURIED );
        via->SetLayerPair( nthCopperLayerId( span->first, totalCu ),
                           nthCopperLayerId( span->second, totalCu ) );
    }
    else
    {
        via->SetViaType( VIATYPE::THROUGH );
        via->SetTopLayer( F_Cu );
        via->SetBottomLayer( B_Cu );
    }
 
    int viaDrill = scale( viaPadstack->GetDrillSize() );
 
    if( viaDrill == 0 )
    {
        viaDrill = viaWidth / 2;
        const wxString& padstackName = m_brdDb.GetString( viaPadstack->m_PadStr );
        wxLogTrace( traceAllegroBuilder, "Via at (%d, %d): no drill in padstack '%s' key %#010x, using fallback %d",
                    aViaData.m_CoordsX, aViaData.m_CoordsY, padstackName, viaPadstack->m_Key, viaDrill );
    }
 
    if( viaWidth <= 0 )
    {
        const wxString& padstackName = m_brdDb.GetString( viaPadstack->m_PadStr );
        wxLogTrace( traceAllegroBuilder,
                    "Via at (%d, %d) in padstack '%s' key %#010x has no valid pad component, using drill-based "
                    "fallback (%d * 2)",
                    aViaData.m_CoordsX, aViaData.m_CoordsY, padstackName, viaPadstack->m_Key, viaDrill );
        viaWidth = viaDrill * 2;
    }
 
    via->SetWidth( F_Cu, viaWidth );
    via->SetDrill( viaDrill );
 
    return via;
}
 
 
void BOARD_BUILDER::createTracks()
{
    wxLogTrace( traceAllegroBuilder, "Creating tracks, vias, and other routed items" );
 
    std::vector<BOARD_ITEM*> newItems;
 
    // We need to walk this list again - we could do this all in createNets, but this seems tidier.
    TYPED_LL_WALKER<BLK_0x1B_NET> netWalker{ m_brdDb.m_Header->m_LL_0x1B_Nets, m_brdDb, MISMATCH_POLICY::REPORT };
    netWalker.SetMismatchReporter(
            [this]( uint8_t aType, const BLOCK_BASE& aBlock )
            {
                reportUnexpectedBlockType( aType, BLK_0x1B_NET::BLOCK_TYPE_CODE, 0, aBlock.GetOffset(), "Net" );
            } );
 
    for( const BLK_0x1B_NET& net : netWalker )
    {
        auto netIt = m_netCache.find( net.m_Key );
 
        if( netIt == m_netCache.end() )
            continue;
 
        const int netCode = netIt->second->GetNetCode();
 
        TYPED_LL_WALKER<BLK_0x04_NET_ASSIGNMENT> assignmentWalker{ net.m_Assignment, net.m_Key, m_brdDb,
                                                                   MISMATCH_POLICY::REPORT };
        assignmentWalker.SetMismatchReporter(
                [this]( uint8_t aType, const BLOCK_BASE& aBlock )
                {
                    reportUnexpectedBlockType( aType, 0x04, 0, aBlock.GetOffset(), "Net assignment" );
                } );
 
        for( const BLK_0x04_NET_ASSIGNMENT& assign : assignmentWalker )
        {
            // Walk the 0x05/0x32/... list
            LL_WALKER connWalker{ assign.m_ConnItem, assign.m_Key, m_brdDb };
            for( const BLOCK_BASE* connItemBlock : connWalker )
            {
                const uint8_t connType = connItemBlock->GetBlockType();
 
                // One connected item can be multiple KiCad objects, e.g.
                // 0x05 track -> list of segments/arcs
                std::vector<std::unique_ptr<BOARD_ITEM>> newItemList;
 
                switch( connType )
                {
                // Track
                case 0x05:
                {
                    const BLK_0x05_TRACK& trackData = BlockDataAs<BLK_0x05_TRACK>( *connItemBlock );
                    newItemList = buildTrack( trackData, netCode );
                    break;
                }
                case 0x33:
                {
                    const BLK_0x33_VIA& viaData = BlockDataAs<BLK_0x33_VIA>( *connItemBlock );
                    newItemList.push_back( buildVia( viaData, netCode ) );
                    break;
                }
                case 0x32:
                {
                    // This is a pad in a footprint - we don't need to handle this here, as we do all the footprint
                    // pads, connected or not, in the footprint step.
                    break;
                }
                case 0x28:
                {
                    // 0x28 shapes on the net chain are computed copper fills.
                    // Collect them for teardrop and polygon import.
                    const BLK_0x28_SHAPE& fillShape = BlockDataAs<BLK_0x28_SHAPE>( *connItemBlock );
 
                    PCB_LAYER_ID fillLayer = getLayer( fillShape.m_Layer );
 
                    if( fillLayer != UNDEFINED_LAYER )
                        m_zoneFillShapes[fillShape.m_Key] = { &fillShape, netCode, fillLayer };
 
                    break;
                }
                case 0x2E:
                default:
                {
                    wxLogTrace( traceAllegroBuilder, "  Unhandled connected item code: %#04x",
                                (int) connType );
                }
                }
 
                for( std::unique_ptr<BOARD_ITEM>& newItem : newItemList )
                {
                    newItems.push_back( newItem.get() );
                    m_board.Add( newItem.release(), ADD_MODE::BULK_APPEND );
                }
            }
        }
    }
 
    m_board.FinalizeBulkAdd( newItems );
 
    m_trackEndpointCopper.clear();
 
    wxLogTrace( traceAllegroBuilder, "Finished creating %zu track/via items", newItems.size() );
}
 
 
void BOARD_BUILDER::createBoardShapes()
{
    wxLogTrace( traceAllegroBuilder, "Creating shapes" );
 
    // Walk through LL_0x24_0x28 which contains rectangles (0x24) and shapes (0x28)
    const LL_WALKER shapeWalker( m_brdDb.m_Header->m_LL_0x24_0x28, m_brdDb );
    int             blockCount = 0;
 
    std::vector<std::unique_ptr<BOARD_ITEM>> newItems;
 
    for( const BLOCK_BASE* block : shapeWalker )
    {
        blockCount++;
 
        switch( block->GetBlockType() )
        {
        case 0x24:
        {
            const BLK_0x24_RECT& rectData = BlockDataAs<BLK_0x24_RECT>( *block );
 
            // These are zones, we don't handle them here
            if( layerIsZone( rectData.m_Layer ) )
                continue;
 
            std::unique_ptr<PCB_SHAPE> rectShape = buildRect( rectData, m_board );
            newItems.push_back( std::move( rectShape ) );
            break;
        }
        case 0x28:
        {
            const BLK_0x28_SHAPE& shapeData = BlockDataAs<BLK_0x28_SHAPE>( *block );
 
            // These are zones, we don't handle them here
            if( layerIsZone( shapeData.m_Layer ) )
                continue;
 
            std::vector<std::unique_ptr<PCB_SHAPE>> shapeItems = buildPolygonShapes( shapeData, m_board );
 
            for( auto& shapeItem : shapeItems )
                newItems.push_back( std::move( shapeItem ) );
            break;
        }
        default:
        {
            wxLogTrace( traceAllegroBuilder, "  Unhandled block type in outline walker: %#04x", block->GetBlockType() );
            break;
        }
        }
    }
 
    wxLogTrace( traceAllegroBuilder, "  Found %d shape blocks", blockCount, newItems.size() );
    blockCount = 0;
 
    LL_WALKER outline2Walker( m_brdDb.m_Header->m_LL_Shapes, m_brdDb );
    for( const BLOCK_BASE* block : outline2Walker )
    {
        blockCount++;
 
        switch( block->GetBlockType() )
        {
        case 0x0E:
        {
            const BLK_0x0E_RECT& rectData = BlockDataAs<BLK_0x0E_RECT>( *block );
 
            if( layerIsZone( rectData.m_Layer ) )
                continue;
 
            std::unique_ptr<PCB_SHAPE> rectShape = buildRect( rectData, m_board );
            newItems.push_back( std::move( rectShape ) );
            break;
        }
        case 0x24:
        {
            const BLK_0x24_RECT& rectData = BlockDataAs<BLK_0x24_RECT>( *block );
 
            if( layerIsZone( rectData.m_Layer ) )
                continue;
 
            std::unique_ptr<PCB_SHAPE> rectShape = buildRect( rectData, m_board );
            newItems.push_back( std::move( rectShape ) );
            break;
        }
        case 0x28:
        {
            const BLK_0x28_SHAPE& shapeData = BlockDataAs<BLK_0x28_SHAPE>( *block );
 
            if( layerIsZone( shapeData.m_Layer ) )
                continue;
 
            std::vector<std::unique_ptr<PCB_SHAPE>> shapeItems = buildPolygonShapes( shapeData, m_board );
 
            for( auto& shapeItem : shapeItems )
                newItems.push_back( std::move( shapeItem ) );
            break;
        }
        default:
        {
            wxLogTrace( traceAllegroBuilder, "  Unhandled block type in outline walker: %#04x", block->GetBlockType() );
            break;
        }
        }
    }
 
    wxLogTrace( traceAllegroBuilder, "  Found %d outline items in m_LL_Shapes", blockCount );
    blockCount = 0;
 
    TYPED_LL_WALKER<BLK_0x14_GRAPHIC> graphicContainerWalker( m_brdDb.m_Header->m_LL_0x14, m_brdDb,
                                                              MISMATCH_POLICY::LOG_TRACE );
    for( const BLK_0x14_GRAPHIC& graphicContainer : graphicContainerWalker )
    {
        blockCount++;
 
        if( layerIsZone( graphicContainer.m_Layer ) )
            continue;
 
        std::vector<std::unique_ptr<PCB_SHAPE>> graphicItems = buildShapes( graphicContainer, m_board );
 
        for( auto& item : graphicItems )
            newItems.push_back( std::move( item ) );
    }
 
    wxLogTrace( traceAllegroBuilder, "  Found %d graphic container items", blockCount );
 
    std::vector<BOARD_ITEM*> addedItems;
    for( std::unique_ptr<BOARD_ITEM>& item : newItems )
    {
        addedItems.push_back( item.get() );
        m_board.Add( item.release(), ADD_MODE::BULK_APPEND );
    }
 
    m_board.FinalizeBulkAdd( addedItems );
 
    wxLogTrace( traceAllegroBuilder, "Created %zu board shapes", addedItems.size() );
}
 
 
const SHAPE_LINE_CHAIN& BOARD_BUILDER::buildSegmentChain( uint32_t aStartKey ) const
{
    auto cacheIt = m_segChainCache.find( aStartKey );
 
    if( cacheIt != m_segChainCache.end() )
        return cacheIt->second;
 
    SHAPE_LINE_CHAIN& outline = m_segChainCache[aStartKey];
    uint32_t          currentKey = aStartKey;
 
    // Safety limit to prevent infinite loops on corrupt data
    static constexpr int MAX_CHAIN_LENGTH = 50000;
    int visited = 0;
 
    while( currentKey != 0 && visited < MAX_CHAIN_LENGTH )
    {
        const BLOCK_BASE* block = m_brdDb.GetObjectByKey( currentKey );
 
        if( !block )
            break;
 
        visited++;
 
        switch( block->GetBlockType() )
        {
        case 0x01:
        {
            const auto& arc = BlockDataAs<BLK_0x01_ARC>( *block );
            VECTOR2I    start = scale( { arc.m_StartX, arc.m_StartY } );
            VECTOR2I    end = scale( { arc.m_EndX, arc.m_EndY } );
            VECTOR2I    center = scale( KiROUND( VECTOR2D{ arc.m_CenterX, arc.m_CenterY } ) );
 
            if( start == end )
            {
                SHAPE_ARC shapeArc( center, start, ANGLE_360 );
                outline.Append( shapeArc );
            }
            else
            {
                bool clockwise = ( arc.m_SubType & 0x40 ) != 0;
 
                EDA_ANGLE startAngle( start - center );
                EDA_ANGLE endAngle( end - center );
                startAngle.Normalize();
                endAngle.Normalize();
 
                EDA_ANGLE arcAngle = endAngle - startAngle;
 
                if( clockwise && arcAngle < ANGLE_0 )
                    arcAngle += ANGLE_360;
 
                if( !clockwise && arcAngle > ANGLE_0 )
                    arcAngle -= ANGLE_360;
 
                VECTOR2I mid = start;
                RotatePoint( mid, center, -arcAngle / 2.0 );
 
                SHAPE_ARC shapeArc( start, mid, end, 0 );
                outline.Append( shapeArc );
            }
 
            currentKey = arc.m_Next;
            break;
        }
        case 0x15:
        case 0x16:
        case 0x17:
        {
            const auto& seg = BlockDataAs<BLK_0x15_16_17_SEGMENT>( *block );
            VECTOR2I start = scale( { seg.m_StartX, seg.m_StartY } );
 
            if( outline.PointCount() == 0 || outline.CLastPoint() != start )
                outline.Append( start );
 
            VECTOR2I end = scale( { seg.m_EndX, seg.m_EndY } );
            outline.Append( end );
            currentKey = seg.m_Next;
            break;
        }
        default:
            currentKey = 0;
            break;
        }
    }
 
    return outline;
}
 
 
SHAPE_LINE_CHAIN BOARD_BUILDER::buildOutline( const BLK_0x0E_RECT& aRect ) const
{
    SHAPE_LINE_CHAIN outline;
 
    VECTOR2I topLeft = scale( VECTOR2I{ aRect.m_Coords[0], aRect.m_Coords[1] } );
    VECTOR2I botRight = scale( VECTOR2I{ aRect.m_Coords[2], aRect.m_Coords[3] } );
    VECTOR2I topRight{ botRight.x, topLeft.y };
    VECTOR2I botLeft{ topLeft.x, botRight.y };
 
    outline.Append( topLeft );
    outline.Append( topRight );
    outline.Append( botRight );
    outline.Append( botLeft );
 
    const EDA_ANGLE angle = fromMillidegrees( aRect.m_Rotation );
    outline.Rotate( angle, topLeft );
 
    return outline;
}
 
 
SHAPE_LINE_CHAIN BOARD_BUILDER::buildOutline( const BLK_0x24_RECT& aRect ) const
{
    SHAPE_LINE_CHAIN outline;
 
    VECTOR2I topLeft = scale( VECTOR2I{ aRect.m_Coords[0], aRect.m_Coords[1] } );
    VECTOR2I botRight = scale( VECTOR2I{ aRect.m_Coords[2], aRect.m_Coords[3] } );
    VECTOR2I topRight{ botRight.x, topLeft.y };
    VECTOR2I botLeft{ topLeft.x, botRight.y };
 
    outline.Append( topLeft );
    outline.Append( topRight );
    outline.Append( botRight );
    outline.Append( botLeft );
 
    const EDA_ANGLE angle = fromMillidegrees( aRect.m_Rotation );
    outline.Rotate( angle, topLeft );
 
    return outline;
}
 
 
SHAPE_LINE_CHAIN BOARD_BUILDER::buildOutline( const BLK_0x28_SHAPE& aShape ) const
{
    SHAPE_LINE_CHAIN outline;
    const LL_WALKER  segWalker{ aShape.m_FirstSegmentPtr, aShape.m_Key, m_brdDb };
 
    for( const BLOCK_BASE* segBlock : segWalker )
    {
        switch( segBlock->GetBlockType() )
        {
        case 0x01:
        {
            const auto& arc = BlockDataAs<BLK_0x01_ARC>( *segBlock );
            VECTOR2I    start = scale( { arc.m_StartX, arc.m_StartY } );
            VECTOR2I    end = scale( { arc.m_EndX, arc.m_EndY } );
            VECTOR2I    center = scale( KiROUND( VECTOR2D{ arc.m_CenterX, arc.m_CenterY } ) );
 
            if( start == end )
            {
                SHAPE_ARC shapeArc( center, start, ANGLE_360 );
                outline.Append( shapeArc );
            }
            else
            {
                bool clockwise = ( arc.m_SubType & 0x40 ) != 0;
 
                EDA_ANGLE startAngle( start - center );
                EDA_ANGLE endAngle( end - center );
                startAngle.Normalize();
                endAngle.Normalize();
 
                EDA_ANGLE arcAngle = endAngle - startAngle;
 
                if( clockwise && arcAngle < ANGLE_0 )
                    arcAngle += ANGLE_360;
 
                if( !clockwise && arcAngle > ANGLE_0 )
                    arcAngle -= ANGLE_360;
 
                VECTOR2I mid = start;
                RotatePoint( mid, center, -arcAngle / 2.0 );
 
                SHAPE_ARC shapeArc( start, mid, end, 0 );
                outline.Append( shapeArc );
            }
 
            break;
        }
        case 0x15:
        case 0x16:
        case 0x17:
        {
            const auto& seg = BlockDataAs<BLK_0x15_16_17_SEGMENT>( *segBlock );
            VECTOR2I    start = scale( { seg.m_StartX, seg.m_StartY } );
 
            if( outline.PointCount() == 0 || outline.CLastPoint() != start )
                outline.Append( start );
 
            VECTOR2I end = scale( { seg.m_EndX, seg.m_EndY } );
            outline.Append( end );
            break;
        }
        default:
            wxLogTrace( traceAllegroBuilder, "    Unhandled segment type in shape outline: %#04x",
                        segBlock->GetBlockType() );
            break;
        }
    }
 
    return outline;
}
 
 
SHAPE_POLY_SET BOARD_BUILDER::shapeToPolySet( const BLK_0x28_SHAPE& aShape ) const
{
    SHAPE_POLY_SET   polySet;
    SHAPE_LINE_CHAIN outline = buildSegmentChain( aShape.m_FirstSegmentPtr );
 
    if( outline.PointCount() < 3 )
    {
        wxLogTrace( traceAllegroBuilder, "  Not enough points for polygon (%d)", outline.PointCount() );
        return polySet;
    }
 
    outline.SetClosed( true );
    polySet.AddOutline( outline );
 
    // Walk 0x34 KEEPOUT chain from m_Ptr4 for holes
    uint32_t holeKey = aShape.m_FirstKeepoutPtr;
 
    while( holeKey != 0 )
    {
        const BLOCK_BASE* holeBlock = m_brdDb.GetObjectByKey( holeKey );
 
        if( !holeBlock || holeBlock->GetBlockType() != 0x34 )
            break;
 
        const auto& keepout = BlockDataAs<BLK_0x34_KEEPOUT>( *holeBlock );
 
        SHAPE_LINE_CHAIN holeOutline = buildSegmentChain( keepout.m_FirstSegmentPtr );
 
        if( holeOutline.PointCount() >= 3 )
        {
            holeOutline.SetClosed( true );
            polySet.AddHole( holeOutline );
        }
 
        holeKey = keepout.m_Next;
    }
 
    return polySet;
}
 
 
SHAPE_POLY_SET ALLEGRO::BOARD_BUILDER::tryBuildZoneShape( const BLOCK_BASE& aBlock )
{
    SHAPE_POLY_SET polySet;
 
    switch( aBlock.GetBlockType() )
    {
    case 0x0E:
    {
        const auto& rectData = BlockDataAs<BLK_0x0E_RECT>( aBlock );
 
        SHAPE_LINE_CHAIN chain( buildOutline( rectData ) );
        chain.SetClosed( true );
 
        polySet = SHAPE_POLY_SET( chain );
        break;
    }
    case 0x14:
    {
        const auto& graphicContainer = BlockDataAs<BLK_0x14_GRAPHIC>( aBlock );
 
        SHAPE_LINE_CHAIN chain = buildSegmentChain( graphicContainer.m_SegmentPtr );
        chain.SetClosed( true );
 
        polySet = SHAPE_POLY_SET( chain );
        break;
    }
    case 0x24:
    {
        const auto& rectData = BlockDataAs<BLK_0x24_RECT>( aBlock );
 
        SHAPE_LINE_CHAIN chain( buildOutline( rectData ) );
        chain.SetClosed( true );
 
        polySet = SHAPE_POLY_SET( chain );
        break;
    }
    case 0x28:
    {
        const auto& shapeData = BlockDataAs<BLK_0x28_SHAPE>( aBlock );
        polySet = shapeToPolySet( shapeData );
        break;
    }
    default:
        wxLogTrace( traceAllegroBuilder, "  Unhandled block type in tryBuildZoneShape: %#04x", aBlock.GetBlockType() );
    }
 
    return polySet;
}
 
 
std::unique_ptr<ZONE> BOARD_BUILDER::buildZone( const BLOCK_BASE&                     aBoundaryBlock,
                                                const std::vector<const BLOCK_BASE*>& aRelatedBlocks,
                                                ZONE_FILL_HANDLER&                    aZoneFillHandler )
{
    int              netCode = NETINFO_LIST::UNCONNECTED;
    const LAYER_INFO layerInfo = expectLayerFromBlock( aBoundaryBlock );
 
    bool isCopperZone = ( layerInfo.m_Class == LAYER_INFO::CLASS::ETCH
                          || layerInfo.m_Class == LAYER_INFO::CLASS::BOUNDARY );
 
    PCB_LAYER_ID layer = UNDEFINED_LAYER;
 
    if( isCopperZone )
    {
        // BOUNDARY shares the ETCH layer list, so resolve subclass via ETCH class
        if( layerInfo.m_Class == LAYER_INFO::CLASS::BOUNDARY )
        {
            LAYER_INFO etchLayer{};
            etchLayer.m_Class = LAYER_INFO::CLASS::ETCH;
            etchLayer.m_Subclass = layerInfo.m_Subclass;
            layer = getLayer( etchLayer );
        }
        else
        {
            layer = getLayer( layerInfo );
        }
    }
    else
    {
        layer = F_Cu;
    }
 
    if( isCopperZone && layer == UNDEFINED_LAYER )
    {
        wxLogTrace( traceAllegroBuilder, "  Skipping shape on layer %#02x:%#02x - unmapped copper layer",
                    layerInfo.m_Class, layerInfo.m_Subclass );
        return nullptr;
    }
 
    const SHAPE_POLY_SET zoneShape = tryBuildZoneShape( aBoundaryBlock );
 
    if( zoneShape.OutlineCount() != 1 )
    {
        wxLogTrace( traceAllegroBuilder, "  Skipping zone with type %#04x, key %#010x - failed to build outline",
                    aBoundaryBlock.GetBlockType(), aBoundaryBlock.GetKey() );
        return nullptr;
    }
 
    auto zone = std::make_unique<ZONE>( &m_board );
    zone->SetHatchStyle( ZONE_BORDER_DISPLAY_STYLE::NO_HATCH );
 
    if( isCopperZone )
    {
        zone->SetLayer( layer );
        zone->SetFillMode( ZONE_FILL_MODE::POLYGONS );
    }
    else
    {
        LSET layerSet = m_layerMapper->GetRuleAreaLayers( layerInfo );
 
        bool isRouteKeepout = ( layerInfo.m_Class == LAYER_INFO::CLASS::ROUTE_KEEPOUT );
        bool isViaKeepout = ( layerInfo.m_Class == LAYER_INFO::CLASS::VIA_KEEPOUT );
        bool isPackageKeepout = ( layerInfo.m_Class == LAYER_INFO::CLASS::PACKAGE_KEEPOUT );
        bool isRouteKeepin = ( layerInfo.m_Class == LAYER_INFO::CLASS::ROUTE_KEEPIN );
        bool isPackageKeepin = ( layerInfo.m_Class == LAYER_INFO::CLASS::PACKAGE_KEEPIN );
 
        zone->SetIsRuleArea( true );
        zone->SetLayerSet( layerSet );
        zone->SetDoNotAllowTracks( isRouteKeepout );
        zone->SetDoNotAllowVias( isViaKeepout );
        zone->SetDoNotAllowZoneFills( isRouteKeepout || isViaKeepout );
        zone->SetDoNotAllowPads( false );
        zone->SetDoNotAllowFootprints( isPackageKeepout );
 
        // Zones don't have native keepin functions, so we leave a note for the user here
        // Later, we could consider adding a custom DRC rule for this (or KiCad could add native keepin
        // zone support)
        if( isRouteKeepin )
            zone->SetZoneName( "Route Keepin" );
        else if( isPackageKeepin )
            zone->SetZoneName( "Package Keepin" );
    }
 
    SHAPE_POLY_SET combinedFill;
 
    for( const BLOCK_BASE* block : aRelatedBlocks )
    {
        if( !block )
            continue;
 
        switch( block->GetBlockType() )
        {
        case 0x1B:
        {
            const auto it = m_netCache.find( block->GetKey() );
 
            if( it != m_netCache.end() )
            {
                wxLogTrace( traceAllegroBuilder, "  Resolved BOUNDARY %#010x -> net '%s' (code %d)",
                            aBoundaryBlock.GetKey(), it->second->GetNetname(), it->second->GetNetCode() );
 
                netCode = it->second->GetNetCode();
            }
            else
            {
                m_reporter.Report( wxString::Format( "Could not find net key %#010x in cache for BOUNDARY %#010x",
                                                     block->GetKey(), aBoundaryBlock.GetKey() ),
                                   RPT_SEVERITY_WARNING );
            }
            break;
        }
        case 0x28:
        {
            const BLK_0x28_SHAPE& shapeData = BlockDataAs<BLK_0x28_SHAPE>( *block );
 
            SHAPE_POLY_SET fillPolySet = shapeToPolySet( shapeData );
            combinedFill.Append( fillPolySet );
            m_usedZoneFillShapes.emplace( block->GetKey() );
            break;
        }
        default: break;
        }
    }
 
    // Set net code AFTER layer assignment. SetNetCode checks IsOnCopperLayer() and
    // forces net=0 if the zone isn't on a copper layer yet.
    zone->SetNetCode( netCode );
 
    for( const SHAPE_LINE_CHAIN& chain : zoneShape.CPolygon( 0 ) )
        zone->AddPolygon( chain );
 
    // Add zone fills
    if( isCopperZone && !combinedFill.IsEmpty() )
    {
        // We don't do this here, though it feels like we should. We collect the
        // information for batch processing later on (which is conceptually
        // easier to parallelise compared to this function).  But there is room
        // for improvement by threading more of this work: shapeToPolySet
        // accounts for about 40% of the remaining single-threaded time in the
        // building process.
 
        // combinedFill.ClearArcs();
        // zoneOutline.ClearArcs();
        // combinedFill.BooleanIntersection( zoneOutline );
        // zone->SetFilledPolysList( layer, combinedFill );
 
        zone->SetIsFilled( true );
        zone->SetNeedRefill( false );
 
        // Poke these relevant context in here for batch processing later on
        aZoneFillHandler.QueuePolygonForZone( *zone, std::move( combinedFill ), layer );
    }
 
    return zone;
}
 
 
std::vector<const BLOCK_BASE*> BOARD_BUILDER::getShapeRelatedBlocks( const BLK_0x28_SHAPE& aShape ) const
{
    // Follow pointer chain: BOUNDARY.TablePtr -> 0x2C TABLE -> Ptr1 -> 0x37 -> m_Ptrs
    std::vector<const BLOCK_BASE*> ret;
    uint32_t                       tableKey = aShape.GetTablePtr();
 
    if( tableKey == 0 )
        return ret;
 
    const BLK_0x2C_TABLE* tbl = expectBlockByKey<BLK_0x2C_TABLE>( tableKey );
 
    if( !tbl )
        return ret;
 
    const BLK_0x37_PTR_ARRAY* ptrArray = expectBlockByKey<BLK_0x37_PTR_ARRAY>( tbl->m_Ptr1 );
 
    if( !ptrArray || ptrArray->m_Count == 0 )
        return ret;
 
    const size_t count = std::min( std::min( ptrArray->m_Count, ptrArray->m_Capacity ), 100u );
    ret.resize( count );
 
    for( size_t i = 0; i < count; i++ )
        ret[i] = m_brdDb.GetObjectByKey( ptrArray->m_Ptrs[i] );
 
    return ret;
}
 
 
void BOARD_BUILDER::createBoardText()
{
    TYPED_LL_WALKER<BLK_0x30_STR_WRAPPER> textWalker( m_brdDb.m_Header->m_LL_0x03_0x30, m_brdDb );
    int textCount = 0;
 
    for( const BLK_0x30_STR_WRAPPER& strWrapper : textWalker )
    {
        std::unique_ptr<PCB_TEXT> text = buildPcbText( strWrapper, m_board );
 
        if( !text )
            continue;
 
        // If the text is referenced from a group, it's not board-level text,
        // and we'll pick up up while iterating the group elsewhere.
        if( strWrapper.GetGroupPtr() != 0 )
        {
            // In a group
            continue;
        }
 
        wxLogTrace( traceAllegroBuilder, "  Board text '%s' on layer %s at (%d, %d)",
                    text->GetText(), m_board.GetLayerName( text->GetLayer() ),
                    text->GetPosition().x, text->GetPosition().y );
 
        m_board.Add( text.release(), ADD_MODE::APPEND );
        textCount++;
    }
 
    wxLogTrace( traceAllegroBuilder, "Created %d board-level text objects", textCount );
}
 
 
template <std::derived_from<BOARD_ITEM> T>
void BulkAddToBoard( BOARD& aBoard, std::vector<std::unique_ptr<T>>&& aItems )
{
    std::vector<BOARD_ITEM*> rawPointers;
    rawPointers.reserve( aItems.size() );
 
    for( std::unique_ptr<T>& item : aItems )
    {
        rawPointers.push_back( item.get() );
        aBoard.Add( item.release(), ADD_MODE::BULK_APPEND );
    }
 
    aBoard.FinalizeBulkAdd( rawPointers );
}
 
 
void BOARD_BUILDER::createZones()
{
    wxLogTrace( traceAllegroBuilder, "Creating zones from m_LL_Shapes and m_LL_0x24_0x28" );
 
    std::vector<std::unique_ptr<ZONE>> newZones;
    std::vector<std::unique_ptr<ZONE>> keepoutZones;
 
    ZONE_FILL_HANDLER zoneFillHandler;
 
    // Walk m_LL_Shapes to find BOUNDARY shapes (zone outlines).
    // BOUNDARY shapes use class 0x15 with copper layer subclass indices.
    const LL_WALKER shapeWalker( m_brdDb.m_Header->m_LL_Shapes, m_brdDb );
 
    for( const BLOCK_BASE* block : shapeWalker )
    {
        std::unique_ptr<ZONE> zone;
        LAYER_INFO            layerInfo{};
 
        switch( block->GetBlockType() )
        {
        case 0x0e:
        {
            const BLK_0x0E_RECT& rectData = BlockDataAs<BLK_0x0E_RECT>( *block );
 
            if( !layerIsZone( rectData.m_Layer ) )
                continue;
 
            zone = buildZone( *block, {}, zoneFillHandler );
            layerInfo = rectData.m_Layer;
            break;
        }
        case 0x24:
        {
            const BLK_0x24_RECT& rectData = BlockDataAs<BLK_0x24_RECT>( *block );
 
            if( !layerIsZone( rectData.m_Layer ) )
                continue;
 
            zone = buildZone( *block, {}, zoneFillHandler );
            layerInfo = rectData.m_Layer;
            break;
        }
        case 0x28:
        {
            const BLK_0x28_SHAPE& shapeData = BlockDataAs<BLK_0x28_SHAPE>( *block );
 
            if( !layerIsZone( shapeData.m_Layer ) )
                continue;
 
            zone = buildZone( *block, getShapeRelatedBlocks( shapeData ), zoneFillHandler );
            layerInfo = shapeData.m_Layer;
            break;
        }
        default:
        {
            wxLogTrace( traceAllegroBuilder, "Unhandled block type in zone shape walker: %#04x, key: %#010x",
                        block->GetBlockType(), block->GetKey() );
            break;
        }
        }
 
        if( zone )
        {
            wxLogTrace( traceAllegroBuilder, "  Zone %#010x net=%d layer=%s class=%#04x:%#04x", block->GetKey(),
                        zone->GetNetCode(), m_board.GetLayerName( zone->GetFirstLayer() ), layerInfo.m_Class,
                        layerInfo.m_Subclass );
 
            newZones.push_back( std::move( zone ) );
        }
    }
 
    // Walk m_LL_0x24_0x28 for keepout/in shapes
    const LL_WALKER keepoutWalker( m_brdDb.m_Header->m_LL_0x24_0x28, m_brdDb );
 
    for( const BLOCK_BASE* block : keepoutWalker )
    {
        std::unique_ptr<ZONE> zone;
 
        switch( block->GetBlockType() )
        {
        case 0x24:
        {
            const BLK_0x24_RECT& rectData = BlockDataAs<BLK_0x24_RECT>( *block );
 
            if( !layerIsZone( rectData.m_Layer ) )
                continue;
 
            wxLogTrace( traceAllegroBuilder, "  Processing %s rect %#010x", layerInfoDisplayName( rectData.m_Layer ),
                        rectData.m_Key );
 
            zone = buildZone( *block, {}, zoneFillHandler );
            break;
        }
        case 0x28:
        {
            const BLK_0x28_SHAPE& shapeData = BlockDataAs<BLK_0x28_SHAPE>( *block );
 
            if( !layerIsZone( shapeData.m_Layer ) )
                continue;
 
            wxLogTrace( traceAllegroBuilder, "  Processing %s shape %#010x", layerInfoDisplayName( shapeData.m_Layer ),
                        shapeData.m_Key );
 
            zone = buildZone( *block, {}, zoneFillHandler );
            break;
        }
        default:
            break;
        }
 
        if( zone )
        {
            newZones.push_back( std::move( zone ) );
        }
    }
 
    // Deal with all the collected zone fill polygons now, all at once
    zoneFillHandler.ProcessPolygons( true );
 
    int originalCount = newZones.size();
 
    // Merge zones with identical polygons and same net into multi-layer zones.
    // Allegro often defines the same zone outline on multiple copper layers (e.g.
    // a ground pour spanning all layers). KiCad represents this as a single zone
    // with multiple fill layers.
    //
    // Rule areas (keepouts) can also merge.
    std::vector<std::unique_ptr<ZONE>> mergedZones = MergeZonesWithSameOutline( std::move( newZones ) );
    int                                mergedCount = mergedZones.size();
 
    int keepoutCount = 0;
    int boundaryCount = 0;
 
    for( const std::unique_ptr<ZONE>& zone : mergedZones )
    {
        if( zone->GetIsRuleArea() )
            keepoutCount++;
        else
            boundaryCount++;
    }
 
    BulkAddToBoard( m_board, std::move( mergedZones ) );
 
    wxLogTrace( traceAllegroBuilder, "Created %d zone outlines and %d keepout areas (%d merged away), ", boundaryCount,
                keepoutCount, originalCount - mergedCount );
}
 
 
void BOARD_BUILDER::createTables()
{
    wxLogTrace( traceAllegroBuilder, "Creating tables from m_LL_0x2C" );
 
    TYPED_LL_WALKER<BLK_0x2C_TABLE> tableWalker( m_brdDb.m_Header->m_LL_0x2C, m_brdDb );
    for( const BLK_0x2C_TABLE& tableData : tableWalker )
    {
        if( tableData.m_SubType != BLK_0x2C_TABLE::SUBTYPE::SUBTYPE_GRAPHICAL_GROUP )
        {
            // 0x2c tables can have lots of subtypes. Only 0x110 seems useful to iterate in this way for now.
            continue;
        }
 
        const wxString& tableName = m_brdDb.GetString( tableData.m_StringPtr );
 
        std::vector<std::unique_ptr<BOARD_ITEM>> newItems;
 
        LL_WALKER keyTableWalker{ tableData.m_Ptr1, tableData.m_Key, m_brdDb };
 
        for( const BLOCK_BASE* keyTable : keyTableWalker )
        {
            wxLogTrace( traceAllegroBuilder, "  Table '%s' (key %#010x, table block key %#010x)", tableName,
                        tableData.m_Key, tableData.m_Ptr1 );
 
            if( !keyTable )
            {
                wxLogTrace( traceAllegroBuilder, "    Key table pointer %#010x is invalid", tableData.m_Ptr1 );
                continue;
            }
 
            switch( keyTable->GetBlockType() )
            {
            case 0x37:
            {
                const BLK_0x37_PTR_ARRAY& ptrArray = BlockDataAs<BLK_0x37_PTR_ARRAY>( *keyTable );
 
                uint32_t count = std::min( ptrArray.m_Count, static_cast<uint32_t>( ptrArray.m_Ptrs.size() ) );
 
                wxLogTrace( traceAllegroBuilder, "    Pointer array with %zu entries", static_cast<size_t>( count ) );
 
                for( uint32_t ptrIndex = 0; ptrIndex < count; ptrIndex++ )
                {
                    uint32_t ptrKey = ptrArray.m_Ptrs[ptrIndex];
 
                    if( ptrKey == 0 )
                        continue;
 
                    const BLOCK_BASE* entryBlock = m_brdDb.GetObjectByKey( ptrKey );
 
                    if( !entryBlock )
                    {
                        wxLogTrace( traceAllegroBuilder, "      Entry pointer %#010x is invalid", ptrKey );
                        continue;
                    }
 
                    for( std::unique_ptr<BOARD_ITEM>& newItem : buildGraphicItems( *entryBlock, m_board ) )
                    {
                        newItems.push_back( std::move( newItem ) );
                    }
                }
 
                break;
            }
            case 0x3c:
            {
                const BLK_0x3C_KEY_LIST& keyList = BlockDataAs<BLK_0x3C_KEY_LIST>( *keyTable );
 
                wxLogTrace( traceAllegroBuilder, "    Key list with %zu entries",
                            static_cast<size_t>( keyList.m_NumEntries ) );
                break;
            }
            default:
            {
                wxLogTrace( traceAllegroBuilder, "    Table has unhandled key table type %#04x",
                            keyTable->GetBlockType() );
                break;
            }
            }
        }
 
        if( newItems.size() > 0 )
        {
            wxLogTrace( traceAllegroBuilder, "  Creating group '%s' with %zu items", tableName, newItems.size() );
 
            std::unique_ptr<PCB_GROUP> group = std::make_unique<PCB_GROUP>( &m_board );
            group->SetName( tableName );
 
            for( const auto& item : newItems )
                group->AddItem( item.get() );
 
            newItems.push_back( std::move( group ) );
 
            BulkAddToBoard( m_board, std::move( newItems ) );
        }
    }
}
 
 
void BOARD_BUILDER::applyZoneFills()
{
    if( m_zoneFillShapes.empty() )
        return;
 
    PROF_TIMER fillTimer;
 
    wxLogTrace( traceAllegroBuilder, "Applying zone fill polygons from %zu collected fills",
                m_zoneFillShapes.size() );
 
 
    // Unmatched ETCH shapes are either standalone copper polygons or dynamic copper
    // (teardrops/fillets). On V172+ boards, m_Unknown2 bit 12 (0x1000) marks auto-generated
    // dynamic copper that maps to KiCad teardrop zones. Shapes without this flag are genuine
    // standalone copper imported as filled PCB_SHAPE.
    int copperShapeCount = 0;
    int teardropCount = 0;
 
    for( const auto& [fillKey, fill] : m_zoneFillShapes )
    {
        if( m_usedZoneFillShapes.contains( fillKey ) )
            continue;
 
        SHAPE_POLY_SET polySet = shapeToPolySet( *fill.shape );
        polySet.Simplify();
 
        for( const SHAPE_POLY_SET::POLYGON& poly : polySet.CPolygons() )
        {
            SHAPE_POLY_SET fractured( poly );
            fractured.Fracture( /* aSimplify */ false );
 
            const bool isDynCopperShape = ( fill.shape->m_Unknown2.value_or( 0 ) & 0x1000 ) != 0;
 
            if( isDynCopperShape )
            {
                auto zone = std::make_unique<ZONE>( &m_board );
 
                zone->SetTeardropAreaType( TEARDROP_TYPE::TD_VIAPAD );
                zone->SetLayer( fill.layer );
                zone->SetNetCode( fill.netCode );
                zone->SetLocalClearance( 0 );
                zone->SetPadConnection( ZONE_CONNECTION::FULL );
                zone->SetIslandRemovalMode( ISLAND_REMOVAL_MODE::NEVER );
                zone->SetHatchStyle( ZONE_BORDER_DISPLAY_STYLE::INVISIBLE_BORDER );
 
                for( const SHAPE_LINE_CHAIN& chain : poly )
                    zone->AddPolygon( chain );
 
                zone->SetFilledPolysList( fill.layer, fractured );
                zone->SetIsFilled( true );
                zone->SetNeedRefill( false );
                zone->CalculateFilledArea();
 
                m_board.Add( zone.release(), ADD_MODE::APPEND );
                teardropCount++;
            }
            else
            {
                auto shape = std::make_unique<PCB_SHAPE>( &m_board, SHAPE_T::POLY );
                shape->SetPolyShape( fractured );
                shape->SetFilled( true );
                shape->SetLayer( fill.layer );
                shape->SetNetCode( fill.netCode );
                shape->SetStroke( STROKE_PARAMS( 0, LINE_STYLE::SOLID ) );
 
                m_board.Add( shape.release(), ADD_MODE::APPEND );
                copperShapeCount++;
            }
        }
    }
 
    wxLogTrace( traceAllegroPerf, wxT( "    applyZoneFills unmatched loop: %.3f ms (%d shapes, %d teardrops)" ), //format:allow
                fillTimer.msecs( true ), copperShapeCount, teardropCount );
}
 
 
void BOARD_BUILDER::enablePadTeardrops()
{
    std::unordered_map<int, std::vector<ZONE*>> teardropsByNet;
 
    for( ZONE* zone : m_board.Zones() )
    {
        if( zone->IsTeardropArea() )
            teardropsByNet[zone->GetNetCode()].push_back( zone );
    }
 
    if( teardropsByNet.empty() )
        return;
 
    int padCount = 0;
    int viaCount = 0;
 
    for( FOOTPRINT* fp : m_board.Footprints() )
    {
        for( PAD* pad : fp->Pads() )
        {
            auto it = teardropsByNet.find( pad->GetNetCode() );
 
            if( it == teardropsByNet.end() )
                continue;
 
            for( ZONE* tdZone : it->second )
            {
                if( !pad->IsOnLayer( tdZone->GetLayer() ) )
                    continue;
 
                if( tdZone->Outline()->Contains( pad->GetPosition() ) )
                {
                    pad->SetTeardropsEnabled( true );
                    padCount++;
                    break;
                }
            }
        }
    }
 
    for( PCB_TRACK* track : m_board.Tracks() )
    {
        if( track->Type() != PCB_VIA_T )
            continue;
 
        PCB_VIA* via = static_cast<PCB_VIA*>( track );
        auto it = teardropsByNet.find( via->GetNetCode() );
 
        if( it == teardropsByNet.end() )
            continue;
 
        for( ZONE* tdZone : it->second )
        {
            if( !via->IsOnLayer( tdZone->GetLayer() ) )
                continue;
 
            if( tdZone->Outline()->Contains( via->GetPosition() ) )
            {
                via->SetTeardropsEnabled( true );
                viaCount++;
                break;
            }
        }
    }
 
    wxLogTrace( traceAllegroBuilder, "Enabled teardrops on %d pads and %d vias", padCount, viaCount );
}
 
 
bool BOARD_BUILDER::BuildBoard()
{
    wxLogTrace( traceAllegroBuilder, "Starting BuildBoard() - Phase 2 of Allegro import" );
    wxLogTrace( traceAllegroBuilder, "  Format version: %d (V172+ = %s)",
                static_cast<int>( m_brdDb.m_FmtVer ),
                ( m_brdDb.m_FmtVer >= FMT_VER::V_172 ) ? "yes" : "no" );
    wxLogTrace( traceAllegroBuilder, "  Allegro version string: %.60s",
                m_brdDb.m_Header->m_AllegroVersion.data() );
 
    if( m_progressReporter )
    {
        m_progressReporter->AddPhases( 4 );
        m_progressReporter->AdvancePhase( _( "Constructing caches" ) );
        m_progressReporter->KeepRefreshing();
    }
 
    PROF_TIMER buildTimer;
 
    wxLogTrace( traceAllegroBuilder, "Caching font definitions and setting up layers" );
    cacheFontDefs();
    wxLogTrace( traceAllegroPerf, wxT( "  cacheFontDefs: %.3f ms" ), buildTimer.msecs( true ) ); //format:allow
 
    setupLayers();
    wxLogTrace( traceAllegroPerf, wxT( "  setupLayers: %.3f ms" ), buildTimer.msecs( true ) ); //format:allow
 
    if( m_progressReporter )
    {
        m_progressReporter->AdvancePhase( _( "Creating nets" ) );
        m_progressReporter->KeepRefreshing();
    }
 
    createNets();
    wxLogTrace( traceAllegroPerf, wxT( "  createNets: %.3f ms" ), buildTimer.msecs( true ) ); //format:allow
 
    if( m_progressReporter )
    {
        m_progressReporter->AdvancePhase( _( "Creating tracks" ) );
        m_progressReporter->KeepRefreshing();
    }
 
    createTracks();
    wxLogTrace( traceAllegroPerf, wxT( "  createTracks: %.3f ms" ), buildTimer.msecs( true ) ); //format:allow
 
    if( m_progressReporter )
        m_progressReporter->KeepRefreshing();
 
    createBoardShapes();
    wxLogTrace( traceAllegroPerf, wxT( "  createBoardShapes: %.3f ms" ), buildTimer.msecs( true ) ); //format:allow
 
    createBoardText();
    wxLogTrace( traceAllegroPerf, wxT( "  createBoardText: %.3f ms" ), buildTimer.msecs( true ) ); //format:allow
 
    createZones();
    wxLogTrace( traceAllegroPerf, wxT( "  createZones: %.3f ms" ), buildTimer.msecs( true ) ); //format:allow
 
    createTables();
    wxLogTrace( traceAllegroPerf, wxT( "  createTables: %.3f ms" ), buildTimer.msecs( true ) ); //format:allow
 
    if( m_progressReporter )
        m_progressReporter->KeepRefreshing();
 
    applyZoneFills();
    wxLogTrace( traceAllegroPerf, wxT( "  applyZoneFills: %.3f ms" ), buildTimer.msecs( true ) ); //format:allow
 
    applyConstraintSets();
    wxLogTrace( traceAllegroPerf, wxT( "  applyConstraintSets: %.3f ms" ), buildTimer.msecs( true ) ); //format:allow
 
    applyNetConstraints();
    wxLogTrace( traceAllegroPerf, wxT( "  applyNetConstraints: %.3f ms" ), buildTimer.msecs( true ) ); //format:allow
 
    applyMatchGroups();
    wxLogTrace( traceAllegroPerf, wxT( "  applyMatchGroups: %.3f ms" ), buildTimer.msecs( true ) ); //format:allow
 
    if( m_progressReporter )
    {
        m_progressReporter->AdvancePhase( _( "Converting footprints" ) );
        m_progressReporter->KeepRefreshing();
    }
 
    TYPED_LL_WALKER<BLK_0x2B_FOOTPRINT_DEF> fpWalker( m_brdDb.m_Header->m_LL_0x2B, m_brdDb );
    std::vector<BOARD_ITEM*>                bulkAddedItems;
 
    THROTTLE refreshThrottle( std::chrono::milliseconds( 100 ) );
 
    for( const BLK_0x2B_FOOTPRINT_DEF& fpBlock : fpWalker )
    {
        TYPED_LL_WALKER<BLK_0x2D_FOOTPRINT_INST> instWalker( fpBlock.m_FirstInstPtr, fpBlock.m_Key, m_brdDb,
                                                             MISMATCH_POLICY::REPORT );
        instWalker.SetMismatchReporter(
                [this, &fpBlock]( uint8_t aType, const BLOCK_BASE& )
                {
                    m_reporter.Report( wxString::Format( "Unexpected object of type %#04x found in footprint %#010x",
                                                         aType, fpBlock.m_Key ),
                                       RPT_SEVERITY_ERROR );
                } );
 
        for( const BLK_0x2D_FOOTPRINT_INST& inst : instWalker )
        {
            std::unique_ptr<FOOTPRINT> fp = buildFootprint( inst );
 
            if( fp )
            {
                bulkAddedItems.push_back( fp.get() );
                m_board.Add( fp.release(), ADD_MODE::BULK_APPEND, true );
            }
            else
            {
                m_reporter.Report( wxString::Format( "Failed to construct footprint for 0x2D key %#010x", inst.m_Key ),
                                   RPT_SEVERITY_ERROR );
            }
 
            if( m_progressReporter && refreshThrottle.Ready() )
                m_progressReporter->KeepRefreshing();
        }
    }
 
    wxLogTrace( traceAllegroPerf, wxT( "  convertFootprints (%zu footprints): %.3f ms" ), //format:allow
                bulkAddedItems.size(), buildTimer.msecs( true ) );
 
    if( !bulkAddedItems.empty() )
        m_board.FinalizeBulkAdd( bulkAddedItems );
 
    wxLogTrace( traceAllegroPerf, wxT( "  FinalizeBulkAdd: %.3f ms" ), buildTimer.msecs( true ) ); //format:allow
    wxLogTrace( traceAllegroBuilder, "Converted %zu footprints", bulkAddedItems.size() );
 
    enablePadTeardrops();
    wxLogTrace( traceAllegroPerf, wxT( "  enablePadTeardrops: %.3f ms" ), buildTimer.msecs( true ) ); //format:allow
    wxLogTrace( traceAllegroPerf, wxT( "  Phase 2 total: %.3f ms" ), buildTimer.msecs() ); //format:allow
 
    wxLogTrace( traceAllegroBuilder, "Board construction completed successfully" );
    return true;
}