kicad/include/hashtables.h

145 lines
4.8 KiB
C
Raw Normal View History

#ifndef HASHTABLES_H_
#define HASHTABLES_H_
/*
* This program source code file is part of KiCad, a free EDA CAD application.
*
* Copyright (C) 2012 SoftPLC Corporation, Dick Hollenbeck <dick@softplc.com>
* Copyright (C) 2012 KiCad Developers, see CHANGELOG.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 2
* of the License, or (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, you may find one here:
* http://www.gnu.org/licenses/old-licenses/gpl-2.0.html
* or you may search the http://www.gnu.org website for the version 2 license,
* or you may write to the Free Software Foundation, Inc.,
* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
*/
#include <base_struct.h>
#include <wx/string.h>
// Three strategies for providing a portable hashtable are given.
// C++, boost, and wx, in that order. C++ solution is no good for mingw.
// So boost seems best for all platforms.
#if 0 // C++ std::unordered_map, maybe in the future
#include <unordered_map>
/// Map a C string to an integer. Used in DSNLEXER.
typedef std::unordered_map< std::string, int > KEYWORD_MAP;
/// Map a C string to an EDA_RECT.
/// The key is the classname of the derived wxformbuilder dialog.
typedef std::unordered_map< std::string, EDA_RECT > RECT_MAP;
#elif 1 // boost::unordered_map
// fix a compile bug at line 97 of boost/detail/container_fwd.hpp
#define BOOST_DETAIL_TEST_FORCE_CONTAINER_FWD
#include <boost/unordered_map.hpp>
// see http://www.boost.org/doc/libs/1_49_0/doc/html/boost/unordered_map.html
/// Equality test for "const char*" type used in very specialized KEYWORD_MAP below
struct iequal_to : std::binary_function< const char*, const char*, bool >
{
bool operator()( const char* x, const char* y ) const
{
return !strcmp( x, y );
}
};
/// Very fast and efficient hash function for "const char*" type, used in specialized
/// KEYWORD_MAP below.
/// taken from: http://www.boost.org/doc/libs/1_53_0/libs/unordered/examples/fnv1.hpp
struct fnv_1a
{
/* not used, std::string is too slow:
std::size_t operator()( std::string const& text ) const
{
std::size_t hash = 2166136261u;
for( std::string::const_iterator it = text.begin(), end = text.end();
it != end; ++it )
{
hash ^= *it;
hash *= 16777619;
}
return hash;
}
*/
std::size_t operator()( const char* it ) const
{
std::size_t hash = 2166136261u;
for( ; *it; ++it )
{
2013-07-09 05:18:03 +00:00
hash ^= (unsigned char) *it;
hash *= 16777619;
}
return hash;
}
};
Changed the way of looking up NETINFO_ITEM using net names (using boost::unordered_map). Added a hash function (wxString) for that. Introduced NETINFO_ITEM::GetNetItem( wxString ). BOARD::FindNet() uses the map. Net codes are updated upon net list update. (BOARD::ReplaceNetlist()) Added in some places (mostly class_board.cpp) pad->SetNet() calls to synchronize net codes. On creation of NETINFO_LIST, the first NETINFO_ITEM is added (the unconnected items net). Removed COMPONENT_NET::m_netNumber, as it was not used anywhere. Added an assert to D_PAD::GetNetname(), checking if net code and net name is consistent for unconnected pads. Added an assert for NETINFO_LIST::AppendNet() to assure that appended nets are unique. It seems that at this point: - Updating net lists works fine. The only difference between the file ouput is that after changes it contains empty nets as well. - Nets are not saved in the lexical order. Still, net names and net codes are properly assigned to all items in the .kicad_pcb file. It is going to be addressed in the next commit. I believe it should not create any problems, as pads are sorted by their net names anyway (NETINFO_LIST::buildPadsFullList()) Performed tests: - Created a blank PCB, saved as pic_programmer.kicad_pcb (from demos folder). Updated net lists. .kicad_pcb file (comparing to the results from master branch) differ with net order (as mentioned before), net codes and timestamps. - Removed some of components from the above .kicad_pcb file and updated net lists. Modules reappeared. .kicad_pcb file differs in the same way as described above. - Trying to change a pad net name (via properties dialog) results in assert being fired. It is done on purpose (as there is a call to GetNetname() and net name and net code do not match). This will not happen after the next commit. - Prepared a simple project (starting with schematics). Imported net list, changed schematic, reimported net list - changes are applied. - Eagle & KiCad legacy boards seem to load without any problem.
2014-01-10 17:04:07 +00:00
/// Hash function for wxString, counterpart of std::string hash
struct WXSTRING_HASH : std::unary_function<wxString, std::size_t>
{
std::size_t operator()( const wxString& aString ) const
{
std::size_t hash = 2166136261u;
for( wxString::const_iterator it = aString.begin(); it != aString.end(); ++it )
{
unsigned ch = static_cast<unsigned>( *it );
hash ^= ch;
Changed the way of looking up NETINFO_ITEM using net names (using boost::unordered_map). Added a hash function (wxString) for that. Introduced NETINFO_ITEM::GetNetItem( wxString ). BOARD::FindNet() uses the map. Net codes are updated upon net list update. (BOARD::ReplaceNetlist()) Added in some places (mostly class_board.cpp) pad->SetNet() calls to synchronize net codes. On creation of NETINFO_LIST, the first NETINFO_ITEM is added (the unconnected items net). Removed COMPONENT_NET::m_netNumber, as it was not used anywhere. Added an assert to D_PAD::GetNetname(), checking if net code and net name is consistent for unconnected pads. Added an assert for NETINFO_LIST::AppendNet() to assure that appended nets are unique. It seems that at this point: - Updating net lists works fine. The only difference between the file ouput is that after changes it contains empty nets as well. - Nets are not saved in the lexical order. Still, net names and net codes are properly assigned to all items in the .kicad_pcb file. It is going to be addressed in the next commit. I believe it should not create any problems, as pads are sorted by their net names anyway (NETINFO_LIST::buildPadsFullList()) Performed tests: - Created a blank PCB, saved as pic_programmer.kicad_pcb (from demos folder). Updated net lists. .kicad_pcb file (comparing to the results from master branch) differ with net order (as mentioned before), net codes and timestamps. - Removed some of components from the above .kicad_pcb file and updated net lists. Modules reappeared. .kicad_pcb file differs in the same way as described above. - Trying to change a pad net name (via properties dialog) results in assert being fired. It is done on purpose (as there is a call to GetNetname() and net name and net code do not match). This will not happen after the next commit. - Prepared a simple project (starting with schematics). Imported net list, changed schematic, reimported net list - changes are applied. - Eagle & KiCad legacy boards seem to load without any problem.
2014-01-10 17:04:07 +00:00
hash *= 16777619;
}
return hash;
}
};
/**
* Type KEYWORD_MAP
* is a hashtable made of a const char* and an int. Note that use of this
* type outside very specific circumstances is foolish since there is no storage
* provided for the actual C string itself. This type assumes use with type KEYWORD
* that is created by CMake and that table creates *constant* storage for C strings
* (and pointers to those C strings). Here we are only interested in the C strings
* themselves and only the pointers are duplicated within the hashtable.
* If the strings were not constant and fixed, this type would not work.
* Also note that normally a hashtable (i.e. unordered_map) using a const char* key
* would simply compare the 32 bit or 64 bit pointers themselves, rather than
* the C strings which they are known to point to in this context.
* I force the latter behavior by supplying both "hash" and "equality" overloads
* to the hashtable (unordered_map) template.
* @author Dick Hollenbeck
*/
typedef boost::unordered_map< const char*, int, fnv_1a, iequal_to > KEYWORD_MAP;
/// Map a std::string to an EDA_RECT.
/// The key is the classname of the derived wxformbuilder dialog.
typedef boost::unordered_map< std::string, EDA_RECT > RECT_MAP;
#endif
#endif // HASHTABLES_H_