mirror/OpenMW
1
0
Fork 0
mirror of https://gitlab.com/OpenMW/openmw.git synced 2025-01-26 18:35:20 +00:00
Code Issues Projects Releases Wiki Activity
OpenMW/apps/openmw/mwworld/esmstore.cpp

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

750 lines
26 KiB
C++
Raw Normal View History

move ESMStore to MWWorld
2012-10-01 19:17:04 +04:00
#include "esmstore.hpp"
Use stable sort+unique to collect RefIDs for ESMStore records counting The idea is to avoid std::map lookup for each CellRef. Instead generate a sequence of added and removed RefNums into a vector then order them by RefNum using a stable sort that preserves relative order of elements with the same RefNum. RefIDs are stored in a different vector to avoid std::string move ctor calls when swapping elements while sorting. Reversed iteration over added and removed RefNums for each unique RefNum is an equivalent to what map-based algorithm produces. The main benefit from sorting a vector is a data locality that means less cache misses for each access. Reduces ESMStore::countRecords perf cycles by 25%.
2021-04-06 23:59:58 +02:00
#include <algorithm>
Load LuaScriptsCfg from both *.omwscripts and *.omwaddon files.
2021-09-26 15:21:46 +02:00
#include <fstream>
gets rid of the macros, we use a tuple instead, with a mechanism to assign an index to each type. so the tuple is only defined in the cpp, but we can still have template functions in the header that can ge tthe index with the type
2022-09-03 11:08:23 +02:00
#include <tuple>
Main program now stores ESM data in memory
2010-05-17 17:35:42 +02:00
Use new logging system for game itself
2018-08-14 23:05:43 +04:00
#include <components/debug/debuglog.hpp>
With the records include removed from store.hpp, need to include the relevant files accross the codebase.Lots of touched files, but very little done
2022-09-05 19:35:15 +02:00
#include <components/esm/records.hpp>
move most of the files from esm to esm3, keep common code in esm; this is make space for a future with esm4 esm typo esm typo
2022-01-22 15:58:41 +01:00
#include <components/esm3/esmreader.hpp>
#include <components/esm3/esmwriter.hpp>
Limit the number of simultaneously open not actively used content files Use LRU cache for ESMReaders. When cache capacity is reached close least recently used ESMReader. Remember the file name if a reader was open. Once the reader requested again open the file if there is stored name for it. Put released ESMReader to the back of the free items list. Close ESMReader's from the front of the free items list. Cached item can be used only by one client at the same time. If the same item is requested twice exception is thrown. This should never happen in practice. If this happens need to fix the client logic. It's allowed to go over the capacity limit when requesting different readers. Ideally this should never happen but there will be system error anyway signalizing about too many open files. Need to fix client logic in this case. All places that were using a vector of ESMReaders now using the cache. Cache is local for each use case and there is no need for a thread safety.
2022-06-01 22:53:18 +02:00
#include <components/esm3/readerscache.hpp>
Load LuaScriptsCfg from both *.omwscripts and *.omwaddon files.
2021-09-26 15:21:46 +02:00
#include <components/loadinglistener/loadinglistener.hpp>
#include <components/lua/configuration.hpp>
Use stable sort+unique to collect RefIDs for ESMStore records counting The idea is to avoid std::map lookup for each CellRef. Instead generate a sequence of added and removed RefNums into a vector then order them by RefNum using a stable sort that preserves relative order of elements with the same RefNum. RefIDs are stored in a different vector to avoid std::string move ctor calls when swapping elements while sorting. Reversed iteration over added and removed RefNums for each unique RefNum is an equivalent to what map-based algorithm produces. The main benefit from sorting a vector is a data locality that means less cache misses for each access. Reduces ESMStore::countRecords perf cycles by 25%.
2021-04-06 23:59:58 +02:00
#include <components/misc/algorithm.hpp>
With the records include removed from store.hpp, need to include the relevant files accross the codebase.Lots of touched files, but very little done
2022-09-05 19:35:15 +02:00
The esm4 reader logic is mutualised to avoid copy pasting code, readerutils gives functions that take visitors as params to decide how a record must be handled Check encoder exists, and get value of stateless encoder. fixes code formatting conventions Fixed output of record with RefId also fixed readTypedRecord and readRecord to have the proper return types Check if the type has a sRecordId
2022-12-30 11:28:19 +01:00
#include <components/esm4/common.hpp>
Initial commit to load ESM4 Some data is actually loaded and store in ESM Store Any new ESM4 will go through the same code path and be automatically sent to the right store
2022-12-28 20:48:25 +01:00
#include <components/esm4/loadcell.hpp>
Load light models in the engine. Doesn't emit light
2023-02-06 20:22:17 +01:00
#include <components/esm4/loadligh.hpp>
Initial commit to load ESM4 Some data is actually loaded and store in ESM Store Any new ESM4 will go through the same code path and be automatically sent to the right store
2022-12-28 20:48:25 +01:00
#include <components/esm4/loadrefr.hpp>
#include <components/esm4/loadstat.hpp>
#include <components/esm4/reader.hpp>
The esm4 reader logic is mutualised to avoid copy pasting code, readerutils gives functions that take visitors as params to decide how a record must be handled Check encoder exists, and get value of stateless encoder. fixes code formatting conventions Fixed output of record with RefId also fixed readTypedRecord and readRecord to have the proper return types Check if the type has a sRecordId
2022-12-30 11:28:19 +01:00
#include <components/esm4/readerutils.hpp>
Show groundcover loading progress
2022-06-03 18:59:08 +02:00
#include <components/esmloader/load.hpp>
Remove some unused code, include cleanup
2015-01-25 01:53:20 +01:00
Mutate base records when adding/removing spells
2020-07-28 08:33:28 +02:00
#include "../mwmechanics/spelllist.hpp"
implement additem/removeitem for non-unique actors
2020-07-26 11:07:18 +02:00
namespace
{
Use stable sort+unique to collect RefIDs for ESMStore records counting The idea is to avoid std::map lookup for each CellRef. Instead generate a sequence of added and removed RefNums into a vector then order them by RefNum using a stable sort that preserves relative order of elements with the same RefNum. RefIDs are stored in a different vector to avoid std::string move ctor calls when swapping elements while sorting. Reversed iteration over added and removed RefNums for each unique RefNum is an equivalent to what map-based algorithm produces. The main benefit from sorting a vector is a data locality that means less cache misses for each access. Reduces ESMStore::countRecords perf cycles by 25%.
2021-04-06 23:59:58 +02:00
struct Ref
{
ESM::RefNum mRefNum;
std::size_t mRefID;
Ref(ESM::RefNum refNum, std::size_t refID)
: mRefNum(refNum)
, mRefID(refID)
{
}
};
constexpr std::size_t deletedRefID = std::numeric_limits<std::size_t>::max();
Initial commit: In ESM structures, replace the string members that are RefIds to other records, to a new strong type The strong type is actually just a string underneath, but this will help in the future to have a distinction so it's easier to search and replace when we use an integer ID Slowly going through all the changes to make, still hundreds of errors a lot of functions/structures use std::string or stringview to designate an ID. So it takes time Continues slowly replacing ids. There are technically more and more compilation errors I have good hope that there is a point where the amount of errors will dramatically go down as all the main functions use the ESM::RefId type Continue moving forward, changes to the stores slowly moving along Starting to see the fruit of those changes. still many many error, but more and more Irun into a situation where a function is sandwiched between two functions that use the RefId type. More replacements. Things are starting to get easier I can see more and more often the issue is that the function is awaiting a RefId, but is given a string there is less need to go down functions and to fix a long list of them. Still moving forward, and for the first time error count is going down! Good pace, not sure about topics though, mId and mName are actually the same thing and are used interchangeably Cells are back to using string for the name, haven't fixed everything yet. Many other changes Under the bar of 400 compilation errors. more good progress <100 compile errors! More progress Game settings store can use string for find, it was a bit absurd how every use of it required to create refId from string some more progress on other fronts Mostly game settings clean one error opened a lot of other errors. Down to 18, but more will prbably appear only link errors left?? Fixed link errors OpenMW compiles, and launches, with some issues, but still!
2022-09-25 13:17:09 +02:00
void readRefs(const ESM::Cell& cell, std::vector<Ref>& refs, std::vector<ESM::RefId>& refIDs,
std::set<ESM::RefId>& keyIDs, ESM::ReadersCache& readers)
implement additem/removeitem for non-unique actors
2020-07-26 11:07:18 +02:00
{
restores countRecords optimisations (#3210) With this PR we restore @elsid 's optimisations of countRecords we have unintentionally discarded in PR #3197. In addition, we give it a more appropriate name and add comments concerning its peculiar background.
2021-11-02 13:46:41 +00:00
// TODO: we have many similar copies of this code.
implement additem/removeitem for non-unique actors
2020-07-26 11:07:18 +02:00
for (size_t i = 0; i < cell.mContextList.size(); i++)
{
Limit the number of simultaneously open not actively used content files Use LRU cache for ESMReaders. When cache capacity is reached close least recently used ESMReader. Remember the file name if a reader was open. Once the reader requested again open the file if there is stored name for it. Put released ESMReader to the back of the free items list. Close ESMReader's from the front of the free items list. Cached item can be used only by one client at the same time. If the same item is requested twice exception is thrown. This should never happen in practice. If this happens need to fix the client logic. It's allowed to go over the capacity limit when requesting different readers. Ideally this should never happen but there will be system error anyway signalizing about too many open files. Need to fix client logic in this case. All places that were using a vector of ESMReaders now using the cache. Cache is local for each use case and there is no need for a thread safety.
2022-06-01 22:53:18 +02:00
const std::size_t index = static_cast<std::size_t>(cell.mContextList[i].index);
const ESM::ReadersCache::BusyItem reader = readers.get(index);
cell.restore(*reader, i);
implement additem/removeitem for non-unique actors
2020-07-26 11:07:18 +02:00
ESM::CellRef ref;
bool deleted = false;
Limit the number of simultaneously open not actively used content files Use LRU cache for ESMReaders. When cache capacity is reached close least recently used ESMReader. Remember the file name if a reader was open. Once the reader requested again open the file if there is stored name for it. Put released ESMReader to the back of the free items list. Close ESMReader's from the front of the free items list. Cached item can be used only by one client at the same time. If the same item is requested twice exception is thrown. This should never happen in practice. If this happens need to fix the client logic. It's allowed to go over the capacity limit when requesting different readers. Ideally this should never happen but there will be system error anyway signalizing about too many open files. Need to fix client logic in this case. All places that were using a vector of ESMReaders now using the cache. Cache is local for each use case and there is no need for a thread safety.
2022-06-01 22:53:18 +02:00
while (cell.getNextRef(*reader, ref, deleted))
implement additem/removeitem for non-unique actors
2020-07-26 11:07:18 +02:00
{
if (deleted)
Use stable sort+unique to collect RefIDs for ESMStore records counting The idea is to avoid std::map lookup for each CellRef. Instead generate a sequence of added and removed RefNums into a vector then order them by RefNum using a stable sort that preserves relative order of elements with the same RefNum. RefIDs are stored in a different vector to avoid std::string move ctor calls when swapping elements while sorting. Reversed iteration over added and removed RefNums for each unique RefNum is an equivalent to what map-based algorithm produces. The main benefit from sorting a vector is a data locality that means less cache misses for each access. Reduces ESMStore::countRecords perf cycles by 25%.
2021-04-06 23:59:58 +02:00
refs.emplace_back(ref.mRefNum, deletedRefID);
implement additem/removeitem for non-unique actors
2020-07-26 11:07:18 +02:00
else if (std::find(cell.mMovedRefs.begin(), cell.mMovedRefs.end(), ref.mRefNum)
== cell.mMovedRefs.end())
{
Add key flag on load
2022-09-30 12:16:45 +02:00
if (!ref.mKey.empty())
keyIDs.insert(std::move(ref.mKey));
Use stable sort+unique to collect RefIDs for ESMStore records counting The idea is to avoid std::map lookup for each CellRef. Instead generate a sequence of added and removed RefNums into a vector then order them by RefNum using a stable sort that preserves relative order of elements with the same RefNum. RefIDs are stored in a different vector to avoid std::string move ctor calls when swapping elements while sorting. Reversed iteration over added and removed RefNums for each unique RefNum is an equivalent to what map-based algorithm produces. The main benefit from sorting a vector is a data locality that means less cache misses for each access. Reduces ESMStore::countRecords perf cycles by 25%.
2021-04-06 23:59:58 +02:00
refs.emplace_back(ref.mRefNum, refIDs.size());
refIDs.push_back(std::move(ref.mRefID));
implement additem/removeitem for non-unique actors
2020-07-26 11:07:18 +02:00
}
}
}
Use stable sort+unique to collect RefIDs for ESMStore records counting The idea is to avoid std::map lookup for each CellRef. Instead generate a sequence of added and removed RefNums into a vector then order them by RefNum using a stable sort that preserves relative order of elements with the same RefNum. RefIDs are stored in a different vector to avoid std::string move ctor calls when swapping elements while sorting. Reversed iteration over added and removed RefNums for each unique RefNum is an equivalent to what map-based algorithm produces. The main benefit from sorting a vector is a data locality that means less cache misses for each access. Reduces ESMStore::countRecords perf cycles by 25%.
2021-04-06 23:59:58 +02:00
for (const auto& [value, deleted] : cell.mLeasedRefs)
implement additem/removeitem for non-unique actors
2020-07-26 11:07:18 +02:00
{
if (deleted)
Use stable sort+unique to collect RefIDs for ESMStore records counting The idea is to avoid std::map lookup for each CellRef. Instead generate a sequence of added and removed RefNums into a vector then order them by RefNum using a stable sort that preserves relative order of elements with the same RefNum. RefIDs are stored in a different vector to avoid std::string move ctor calls when swapping elements while sorting. Reversed iteration over added and removed RefNums for each unique RefNum is an equivalent to what map-based algorithm produces. The main benefit from sorting a vector is a data locality that means less cache misses for each access. Reduces ESMStore::countRecords perf cycles by 25%.
2021-04-06 23:59:58 +02:00
refs.emplace_back(value.mRefNum, deletedRefID);
implement additem/removeitem for non-unique actors
2020-07-26 11:07:18 +02:00
else
{
Add key flag on load
2022-09-30 12:16:45 +02:00
if (!value.mKey.empty())
keyIDs.insert(std::move(value.mKey));
Use stable sort+unique to collect RefIDs for ESMStore records counting The idea is to avoid std::map lookup for each CellRef. Instead generate a sequence of added and removed RefNums into a vector then order them by RefNum using a stable sort that preserves relative order of elements with the same RefNum. RefIDs are stored in a different vector to avoid std::string move ctor calls when swapping elements while sorting. Reversed iteration over added and removed RefNums for each unique RefNum is an equivalent to what map-based algorithm produces. The main benefit from sorting a vector is a data locality that means less cache misses for each access. Reduces ESMStore::countRecords perf cycles by 25%.
2021-04-06 23:59:58 +02:00
refs.emplace_back(value.mRefNum, refIDs.size());
refIDs.push_back(value.mRefID);
implement additem/removeitem for non-unique actors
2020-07-26 11:07:18 +02:00
}
}
}
Also run NPC validation for modified base records
2021-03-23 21:07:57 +01:00
Initial commit: In ESM structures, replace the string members that are RefIds to other records, to a new strong type The strong type is actually just a string underneath, but this will help in the future to have a distinction so it's easier to search and replace when we use an integer ID Slowly going through all the changes to make, still hundreds of errors a lot of functions/structures use std::string or stringview to designate an ID. So it takes time Continues slowly replacing ids. There are technically more and more compilation errors I have good hope that there is a point where the amount of errors will dramatically go down as all the main functions use the ESM::RefId type Continue moving forward, changes to the stores slowly moving along Starting to see the fruit of those changes. still many many error, but more and more Irun into a situation where a function is sandwiched between two functions that use the RefId type. More replacements. Things are starting to get easier I can see more and more often the issue is that the function is awaiting a RefId, but is given a string there is less need to go down functions and to fix a long list of them. Still moving forward, and for the first time error count is going down! Good pace, not sure about topics though, mId and mName are actually the same thing and are used interchangeably Cells are back to using string for the name, haven't fixed everything yet. Many other changes Under the bar of 400 compilation errors. more good progress <100 compile errors! More progress Game settings store can use string for find, it was a bit absurd how every use of it required to create refId from string some more progress on other fronts Mostly game settings clean one error opened a lot of other errors. Down to 18, but more will prbably appear only link errors left?? Fixed link errors OpenMW compiles, and launches, with some issues, but still!
2022-09-25 13:17:09 +02:00
const ESM::RefId& getDefaultClass(const MWWorld::Store<ESM::Class>& classes)
Also run NPC validation for modified base records
2021-03-23 21:07:57 +01:00
{
auto it = classes.begin();
if (it != classes.end())
Replace empty NPC class with fallback
2022-06-06 10:59:26 +02:00
return it->mId;
throw std::runtime_error("List of NPC classes is empty!");
}
std::vector<ESM::NPC> getNPCsToReplace(const MWWorld::Store<ESM::Faction>& factions,
create a specific esm reader function for RefID to avoid allocation for string and then again for RefId Fixed some types removed useless header applied clang format fixed compile tests fixed clang tidy, and closer to logic before this MR Removed hardcoded refids unless there is a returned value we don't use static RefIds can use == between RefId and hardcoded string Fix clang format Fixed a few instances where std::string was used, when only const std::string& was needed removed unused variable
2022-10-18 09:26:55 +02:00
const MWWorld::Store<ESM::Class>& classes, const std::unordered_map<ESM::RefId, ESM::NPC>& npcs)
Replace empty NPC class with fallback
2022-06-06 10:59:26 +02:00
{
// Cache first class from store - we will use it if current class is not found
Initial commit: In ESM structures, replace the string members that are RefIds to other records, to a new strong type The strong type is actually just a string underneath, but this will help in the future to have a distinction so it's easier to search and replace when we use an integer ID Slowly going through all the changes to make, still hundreds of errors a lot of functions/structures use std::string or stringview to designate an ID. So it takes time Continues slowly replacing ids. There are technically more and more compilation errors I have good hope that there is a point where the amount of errors will dramatically go down as all the main functions use the ESM::RefId type Continue moving forward, changes to the stores slowly moving along Starting to see the fruit of those changes. still many many error, but more and more Irun into a situation where a function is sandwiched between two functions that use the RefId type. More replacements. Things are starting to get easier I can see more and more often the issue is that the function is awaiting a RefId, but is given a string there is less need to go down functions and to fix a long list of them. Still moving forward, and for the first time error count is going down! Good pace, not sure about topics though, mId and mName are actually the same thing and are used interchangeably Cells are back to using string for the name, haven't fixed everything yet. Many other changes Under the bar of 400 compilation errors. more good progress <100 compile errors! More progress Game settings store can use string for find, it was a bit absurd how every use of it required to create refId from string some more progress on other fronts Mostly game settings clean one error opened a lot of other errors. Down to 18, but more will prbably appear only link errors left?? Fixed link errors OpenMW compiles, and launches, with some issues, but still!
2022-09-25 13:17:09 +02:00
const ESM::RefId& defaultCls = getDefaultClass(classes);
Also run NPC validation for modified base records
2021-03-23 21:07:57 +01:00
// Validate NPCs for non-existing class and faction.
// We will replace invalid entries by fixed ones
std::vector<ESM::NPC> npcsToReplace;
Use const references when possible in for loops No need for useless copies.
2021-05-15 19:50:01 +02:00
for (const auto& npcIter : npcs)
Also run NPC validation for modified base records
2021-03-23 21:07:57 +01:00
{
Clean up MSVC warnings
2021-04-17 10:53:47 +04:00
ESM::NPC npc = npcIter.second;
Also run NPC validation for modified base records
2021-03-23 21:07:57 +01:00
bool changed = false;
Initial commit: In ESM structures, replace the string members that are RefIds to other records, to a new strong type The strong type is actually just a string underneath, but this will help in the future to have a distinction so it's easier to search and replace when we use an integer ID Slowly going through all the changes to make, still hundreds of errors a lot of functions/structures use std::string or stringview to designate an ID. So it takes time Continues slowly replacing ids. There are technically more and more compilation errors I have good hope that there is a point where the amount of errors will dramatically go down as all the main functions use the ESM::RefId type Continue moving forward, changes to the stores slowly moving along Starting to see the fruit of those changes. still many many error, but more and more Irun into a situation where a function is sandwiched between two functions that use the RefId type. More replacements. Things are starting to get easier I can see more and more often the issue is that the function is awaiting a RefId, but is given a string there is less need to go down functions and to fix a long list of them. Still moving forward, and for the first time error count is going down! Good pace, not sure about topics though, mId and mName are actually the same thing and are used interchangeably Cells are back to using string for the name, haven't fixed everything yet. Many other changes Under the bar of 400 compilation errors. more good progress <100 compile errors! More progress Game settings store can use string for find, it was a bit absurd how every use of it required to create refId from string some more progress on other fronts Mostly game settings clean one error opened a lot of other errors. Down to 18, but more will prbably appear only link errors left?? Fixed link errors OpenMW compiles, and launches, with some issues, but still!
2022-09-25 13:17:09 +02:00
const ESM::RefId& npcFaction = npc.mFaction;
Also run NPC validation for modified base records
2021-03-23 21:07:57 +01:00
if (!npcFaction.empty())
{
const ESM::Faction* fact = factions.search(npcFaction);
if (!fact)
{
Log(Debug::Verbose) << "NPC '" << npc.mId << "' (" << npc.mName << ") has nonexistent faction '"
<< npc.mFaction << "', ignoring it.";
Replace ESM::RefId::sEmpty by default constructed RefId where possible Static const is only required to provide a reference or a pointer when it is not possible with default constructed temporary.
2023-02-17 19:20:29 +01:00
npc.mFaction = ESM::RefId();
Also run NPC validation for modified base records
2021-03-23 21:07:57 +01:00
npc.mNpdt.mRank = 0;
changed = true;
}
}
Initial commit: In ESM structures, replace the string members that are RefIds to other records, to a new strong type The strong type is actually just a string underneath, but this will help in the future to have a distinction so it's easier to search and replace when we use an integer ID Slowly going through all the changes to make, still hundreds of errors a lot of functions/structures use std::string or stringview to designate an ID. So it takes time Continues slowly replacing ids. There are technically more and more compilation errors I have good hope that there is a point where the amount of errors will dramatically go down as all the main functions use the ESM::RefId type Continue moving forward, changes to the stores slowly moving along Starting to see the fruit of those changes. still many many error, but more and more Irun into a situation where a function is sandwiched between two functions that use the RefId type. More replacements. Things are starting to get easier I can see more and more often the issue is that the function is awaiting a RefId, but is given a string there is less need to go down functions and to fix a long list of them. Still moving forward, and for the first time error count is going down! Good pace, not sure about topics though, mId and mName are actually the same thing and are used interchangeably Cells are back to using string for the name, haven't fixed everything yet. Many other changes Under the bar of 400 compilation errors. more good progress <100 compile errors! More progress Game settings store can use string for find, it was a bit absurd how every use of it required to create refId from string some more progress on other fronts Mostly game settings clean one error opened a lot of other errors. Down to 18, but more will prbably appear only link errors left?? Fixed link errors OpenMW compiles, and launches, with some issues, but still!
2022-09-25 13:17:09 +02:00
const ESM::RefId& npcClass = npc.mClass;
Replace empty NPC class with fallback
2022-06-06 10:59:26 +02:00
const ESM::Class* cls = classes.search(npcClass);
if (!cls)
Also run NPC validation for modified base records
2021-03-23 21:07:57 +01:00
{
Replace empty NPC class with fallback
2022-06-06 10:59:26 +02:00
Log(Debug::Verbose) << "NPC '" << npc.mId << "' (" << npc.mName << ") has nonexistent class '"
<< npc.mClass << "', using '" << defaultCls << "' class as replacement.";
npc.mClass = defaultCls;
changed = true;
Also run NPC validation for modified base records
2021-03-23 21:07:57 +01:00
}
if (changed)
npcsToReplace.push_back(npc);
}
return npcsToReplace;
}
Improve save cleaning
2021-07-12 20:51:13 +02:00
// Custom enchanted items can reference scripts that no longer exist, this doesn't necessarily mean the base item no
// longer exists however. So instead of removing the item altogether, we're only removing the script.
refactors MWWorld::Store maps (#3197) With this PR we clean up `MWWorld::Store` maps according to the approach of PR #3184.
2021-10-30 17:27:57 +00:00
template <class MapT>
void removeMissingScripts(const MWWorld::Store<ESM::Script>& scripts, MapT& items)
Improve save cleaning
2021-07-12 20:51:13 +02:00
{
for (auto& [id, item] : items)
{
if (!item.mScript.empty() && !scripts.search(item.mScript))
{
Replace ESM::RefId::sEmpty by default constructed RefId where possible Static const is only required to provide a reference or a pointer when it is not possible with default constructed temporary.
2023-02-17 19:20:29 +01:00
item.mScript = ESM::RefId();
Improve save cleaning
2021-07-12 20:51:13 +02:00
Log(Debug::Verbose) << "Item '" << id << "' (" << item.mName << ") has nonexistent script '"
<< item.mScript << "', ignoring it.";
}
}
}
implement additem/removeitem for non-unique actors
2020-07-26 11:07:18 +02:00
}
move ESMStore to MWWorld
2012-10-01 19:17:04 +04:00
namespace MWWorld
Main program now stores ESM data in memory
2010-05-17 17:35:42 +02:00
{
Initial commit: In ESM structures, replace the string members that are RefIds to other records, to a new strong type The strong type is actually just a string underneath, but this will help in the future to have a distinction so it's easier to search and replace when we use an integer ID Slowly going through all the changes to make, still hundreds of errors a lot of functions/structures use std::string or stringview to designate an ID. So it takes time Continues slowly replacing ids. There are technically more and more compilation errors I have good hope that there is a point where the amount of errors will dramatically go down as all the main functions use the ESM::RefId type Continue moving forward, changes to the stores slowly moving along Starting to see the fruit of those changes. still many many error, but more and more Irun into a situation where a function is sandwiched between two functions that use the RefId type. More replacements. Things are starting to get easier I can see more and more often the issue is that the function is awaiting a RefId, but is given a string there is less need to go down functions and to fix a long list of them. Still moving forward, and for the first time error count is going down! Good pace, not sure about topics though, mId and mName are actually the same thing and are used interchangeably Cells are back to using string for the name, haven't fixed everything yet. Many other changes Under the bar of 400 compilation errors. more good progress <100 compile errors! More progress Game settings store can use string for find, it was a bit absurd how every use of it required to create refId from string some more progress on other fronts Mostly game settings clean one error opened a lot of other errors. Down to 18, but more will prbably appear only link errors left?? Fixed link errors OpenMW compiles, and launches, with some issues, but still!
2022-09-25 13:17:09 +02:00
using IDMap = std::unordered_map<ESM::RefId, int>;
mids and mStaticIds moved to mStoreImp renamed recordid to storeid
2022-07-20 19:28:06 +02:00
ESMStore: no longer necessary to include the type definitions in the header For now there still needs to declare all the getters, but a macro may make that easier, or an different method entierly
2022-07-03 11:47:46 +02:00
struct ESMStoreImp
{
The tuple types declaration is back in the header, but there is no need for implementation details to be known for any of the types
2022-09-05 18:38:59 +02:00
ESMStore::StoreTuple mStores;
Store refactor: Now way easier to create new stores, a good chunk of it is automated, only simple macros are used now The case of indexedStores remains an issue, because they can't be stored with the rest because they don't inherit of store base
2022-07-03 18:11:20 +02:00
rename StoreBase =>DynamicStore, and create new class Storebase all stores inherit from base class StoreBase.Storebase is just an empty interface class
2022-09-02 23:46:37 +02:00
std::map<ESM::RecNameInts, DynamicStore*> mRecNameToStore;
Store refactor: Now way easier to create new stores, a good chunk of it is automated, only simple macros are used now The case of indexedStores remains an issue, because they can't be stored with the rest because they don't inherit of store base
2022-07-03 18:11:20 +02:00
mids and mStaticIds moved to mStoreImp renamed recordid to storeid
2022-07-20 19:28:06 +02:00
// Lookup of all IDs. Makes looking up references faster. Just
// maps the id name to the record type.
IDMap mIds;
IDMap mStaticIds;
gets rid of the macros, we use a tuple instead, with a mechanism to assign an index to each type. so the tuple is only defined in the cpp, but we can still have template functions in the header that can ge tthe index with the type
2022-09-03 11:08:23 +02:00
template <typename T>
function doesn't need to return int
2022-09-06 23:30:51 +02:00
static void assignStoreToIndex(ESMStore& stores, Store<T>& store)
Store refactor: Now way easier to create new stores, a good chunk of it is automated, only simple macros are used now The case of indexedStores remains an issue, because they can't be stored with the rest because they don't inherit of store base
2022-07-03 18:11:20 +02:00
{
Type index type error fixed
2022-09-03 16:36:53 +02:00
const std::size_t storeIndex = ESMStore::getTypeIndex<T>();
removes useless make_tuple, and changes type of typeIndex to size_t
2022-09-03 13:32:44 +02:00
if (stores.mStores.size() <= storeIndex)
stores.mStores.resize(storeIndex + 1);
Fixed naming convention problems removed 2 macros. - One is replaced by the underlying code - The second one ise replaced by a template function removed tabulation used as indentation used getWritable instead of a const_cast used for( val : cont) loop removed useless getId function
2022-07-19 17:58:21 +02:00
gets rid of the macros, we use a tuple instead, with a mechanism to assign an index to each type. so the tuple is only defined in the cpp, but we can still have template functions in the header that can ge tthe index with the type
2022-09-03 11:08:23 +02:00
assert(&store == &std::get<Store<T>>(stores.mStoreImp->mStores));
stores.mStores[storeIndex] = &store;
various fixes fixed naming convention replaced std::string by string_view when possible removed unused function and member varaible replaced type::value by type_v set default destructor in cpp function getTypeIndex => getnextindex
2022-09-06 23:10:58 +02:00
if constexpr (std::is_convertible_v<Store<T>*, DynamicStore*>)
Very convincing experiment, by associatingg a compiletime index to the esm record, it becomes possible to automate the loop that creates them it will also be possible to associate AT COMPILETIME the RecNameInt to the esm type, which in the same manner will automatically populate mESM3RecordToStore
2022-07-23 16:41:42 +02:00
{
gets rid of the macros, we use a tuple instead, with a mechanism to assign an index to each type. so the tuple is only defined in the cpp, but we can still have template functions in the header that can ge tthe index with the type
2022-09-03 11:08:23 +02:00
stores.mDynamicStores.push_back(&store);
removes recname type, we use what is already included in ESM types
2022-09-02 23:36:41 +02:00
constexpr ESM::RecNameInts recName = T::sRecordId;
Binds at compile time esm struct and RecNameInts to automatically populate mESM3RecordToStore fewer possible mistakes now, one macro takes all the information to create all the stores and maps from RecName to Store
2022-07-24 12:37:27 +02:00
if constexpr (recName != ESM::REC_INTERNAL_PLAYER)
{
gets rid of the macros, we use a tuple instead, with a mechanism to assign an index to each type. so the tuple is only defined in the cpp, but we can still have template functions in the header that can ge tthe index with the type
2022-09-03 11:08:23 +02:00
stores.mStoreImp->mRecNameToStore[recName] = &store;
Binds at compile time esm struct and RecNameInts to automatically populate mESM3RecordToStore fewer possible mistakes now, one macro takes all the information to create all the stores and maps from RecName to Store
2022-07-24 12:37:27 +02:00
}
}
gets rid of the macros, we use a tuple instead, with a mechanism to assign an index to each type. so the tuple is only defined in the cpp, but we can still have template functions in the header that can ge tthe index with the type
2022-09-03 11:08:23 +02:00
}
Initial commit to load ESM4 Some data is actually loaded and store in ESM Store Any new ESM4 will go through the same code path and be automatically sent to the right store
2022-12-28 20:48:25 +01:00
template <typename T>
Applied review advice better deals with templated functions meant for esm3 used by esm4s. simplified loading of esm4 and dealing with unhandled record types.
2023-01-02 19:20:03 +01:00
static bool typedReadRecordESM4(ESM4::Reader& reader, Store<T>& store)
Initial commit to load ESM4 Some data is actually loaded and store in ESM Store Any new ESM4 will go through the same code path and be automatically sent to the right store
2022-12-28 20:48:25 +01:00
{
auto recordType = static_cast<ESM4::RecordTypes>(reader.hdr().record.typeId);
ESM::RecNameInts esm4RecName = static_cast<ESM::RecNameInts>(ESM::esm4Recname(recordType));
The esm4 reader logic is mutualised to avoid copy pasting code, readerutils gives functions that take visitors as params to decide how a record must be handled Check encoder exists, and get value of stateless encoder. fixes code formatting conventions Fixed output of record with RefId also fixed readTypedRecord and readRecord to have the proper return types Check if the type has a sRecordId
2022-12-30 11:28:19 +01:00
if constexpr (std::is_convertible_v<Store<T>*, DynamicStore*> && HasRecordId<T>::value)
Initial commit to load ESM4 Some data is actually loaded and store in ESM Store Any new ESM4 will go through the same code path and be automatically sent to the right store
2022-12-28 20:48:25 +01:00
{
if constexpr (ESM::isESM4Rec(T::sRecordId))
{
if (T::sRecordId == esm4RecName)
{
reader.getRecordData();
T value;
value.load(reader);
store.insertStatic(value);
Applied review advice better deals with templated functions meant for esm3 used by esm4s. simplified loading of esm4 and dealing with unhandled record types.
2023-01-02 19:20:03 +01:00
return true;
Initial commit to load ESM4 Some data is actually loaded and store in ESM Store Any new ESM4 will go through the same code path and be automatically sent to the right store
2022-12-28 20:48:25 +01:00
}
}
}
Applied review advice better deals with templated functions meant for esm3 used by esm4s. simplified loading of esm4 and dealing with unhandled record types.
2023-01-02 19:20:03 +01:00
return false;
Initial commit to load ESM4 Some data is actually loaded and store in ESM Store Any new ESM4 will go through the same code path and be automatically sent to the right store
2022-12-28 20:48:25 +01:00
}
The esm4 reader logic is mutualised to avoid copy pasting code, readerutils gives functions that take visitors as params to decide how a record must be handled Check encoder exists, and get value of stateless encoder. fixes code formatting conventions Fixed output of record with RefId also fixed readTypedRecord and readRecord to have the proper return types Check if the type has a sRecordId
2022-12-30 11:28:19 +01:00
static bool readRecord(ESM4::Reader& reader, ESMStore& store)
Initial commit to load ESM4 Some data is actually loaded and store in ESM Store Any new ESM4 will go through the same code path and be automatically sent to the right store
2022-12-28 20:48:25 +01:00
{
Static references are created as refcells, nothing displayed yet.
2023-01-23 23:51:45 +01:00
return std::apply(
[&reader](auto&... x) { return (typedReadRecordESM4(reader, x) || ...); }, store.mStoreImp->mStores);
Initial commit to load ESM4 Some data is actually loaded and store in ESM Store Any new ESM4 will go through the same code path and be automatically sent to the right store
2022-12-28 20:48:25 +01:00
}
ESMStore: no longer necessary to include the type definitions in the header For now there still needs to declare all the getters, but a macro may make that easier, or an different method entierly
2022-07-03 11:47:46 +02:00
};
Initial commit: In ESM structures, replace the string members that are RefIds to other records, to a new strong type The strong type is actually just a string underneath, but this will help in the future to have a distinction so it's easier to search and replace when we use an integer ID Slowly going through all the changes to make, still hundreds of errors a lot of functions/structures use std::string or stringview to designate an ID. So it takes time Continues slowly replacing ids. There are technically more and more compilation errors I have good hope that there is a point where the amount of errors will dramatically go down as all the main functions use the ESM::RefId type Continue moving forward, changes to the stores slowly moving along Starting to see the fruit of those changes. still many many error, but more and more Irun into a situation where a function is sandwiched between two functions that use the RefId type. More replacements. Things are starting to get easier I can see more and more often the issue is that the function is awaiting a RefId, but is given a string there is less need to go down functions and to fix a long list of them. Still moving forward, and for the first time error count is going down! Good pace, not sure about topics though, mId and mName are actually the same thing and are used interchangeably Cells are back to using string for the name, haven't fixed everything yet. Many other changes Under the bar of 400 compilation errors. more good progress <100 compile errors! More progress Game settings store can use string for find, it was a bit absurd how every use of it required to create refId from string some more progress on other fronts Mostly game settings clean one error opened a lot of other errors. Down to 18, but more will prbably appear only link errors left?? Fixed link errors OpenMW compiles, and launches, with some issues, but still!
2022-09-25 13:17:09 +02:00
int ESMStore::find(const ESM::RefId& id) const
mids and mStaticIds moved to mStoreImp renamed recordid to storeid
2022-07-20 19:28:06 +02:00
{
IDMap::const_iterator it = mStoreImp->mIds.find(id);
if (it == mStoreImp->mIds.end())
{
return 0;
}
return it->second;
}
Initial commit: In ESM structures, replace the string members that are RefIds to other records, to a new strong type The strong type is actually just a string underneath, but this will help in the future to have a distinction so it's easier to search and replace when we use an integer ID Slowly going through all the changes to make, still hundreds of errors a lot of functions/structures use std::string or stringview to designate an ID. So it takes time Continues slowly replacing ids. There are technically more and more compilation errors I have good hope that there is a point where the amount of errors will dramatically go down as all the main functions use the ESM::RefId type Continue moving forward, changes to the stores slowly moving along Starting to see the fruit of those changes. still many many error, but more and more Irun into a situation where a function is sandwiched between two functions that use the RefId type. More replacements. Things are starting to get easier I can see more and more often the issue is that the function is awaiting a RefId, but is given a string there is less need to go down functions and to fix a long list of them. Still moving forward, and for the first time error count is going down! Good pace, not sure about topics though, mId and mName are actually the same thing and are used interchangeably Cells are back to using string for the name, haven't fixed everything yet. Many other changes Under the bar of 400 compilation errors. more good progress <100 compile errors! More progress Game settings store can use string for find, it was a bit absurd how every use of it required to create refId from string some more progress on other fronts Mostly game settings clean one error opened a lot of other errors. Down to 18, but more will prbably appear only link errors left?? Fixed link errors OpenMW compiles, and launches, with some issues, but still!
2022-09-25 13:17:09 +02:00
int ESMStore::findStatic(const ESM::RefId& id) const
mids and mStaticIds moved to mStoreImp renamed recordid to storeid
2022-07-20 19:28:06 +02:00
{
IDMap::const_iterator it = mStoreImp->mStaticIds.find(id);
if (it == mStoreImp->mStaticIds.end())
{
return 0;
}
return it->second;
}
ESMStore: no longer necessary to include the type definitions in the header For now there still needs to declare all the getters, but a macro may make that easier, or an different method entierly
2022-07-03 11:47:46 +02:00
ESMStore::ESMStore()
{
fix compile
2022-09-03 20:04:31 +02:00
mStoreImp = std::make_unique<ESMStoreImp>();
fixed naming convention issue
2022-09-06 14:27:53 +02:00
std::apply([this](auto&... x) { (ESMStoreImp::assignStoreToIndex(*this, x), ...); }, mStoreImp->mStores);
ESMStore: no longer necessary to include the type definitions in the header For now there still needs to declare all the getters, but a macro may make that easier, or an different method entierly
2022-07-03 11:47:46 +02:00
mDynamicCount = 0;
Store refactor: Now way easier to create new stores, a good chunk of it is automated, only simple macros are used now The case of indexedStores remains an issue, because they can't be stored with the rest because they don't inherit of store base
2022-07-03 18:11:20 +02:00
getWritable<ESM::Pathgrid>().setCells(getWritable<ESM::Cell>());
ESMStore: no longer necessary to include the type definitions in the header For now there still needs to declare all the getters, but a macro may make that easier, or an different method entierly
2022-07-03 11:47:46 +02:00
}
various fixes fixed naming convention replaced std::string by string_view when possible removed unused function and member varaible replaced type::value by type_v set default destructor in cpp function getTypeIndex => getnextindex
2022-09-06 23:10:58 +02:00
ESMStore::~ESMStore() = default;
ESMStore: no longer necessary to include the type definitions in the header For now there still needs to declare all the getters, but a macro may make that easier, or an different method entierly
2022-07-03 11:47:46 +02:00
Store: moved all the template specialization to its own heaper file, included where it's needed in the esm store a function is defined in the cpp file to not rely on the knowledge of store.hpp in the header file
2022-07-03 14:40:37 +02:00
void ESMStore::clearDynamic()
{
gets rid of the macros, we use a tuple instead, with a mechanism to assign an index to each type. so the tuple is only defined in the cpp, but we can still have template functions in the header that can ge tthe index with the type
2022-09-03 11:08:23 +02:00
for (const auto& store : mDynamicStores)
Fixed naming convention problems removed 2 macros. - One is replaced by the underlying code - The second one ise replaced by a template function removed tabulation used as indentation used getWritable instead of a const_cast used for( val : cont) loop removed useless getId function
2022-07-19 17:58:21 +02:00
store->clearDynamic();
Store: moved all the template specialization to its own heaper file, included where it's needed in the esm store a function is defined in the cpp file to not rely on the knowledge of store.hpp in the header file
2022-07-03 14:40:37 +02:00
movePlayerRecord();
}
just replace esmstore, inconsistent
2012-11-05 18:09:14 +04:00
static bool isCacheableRecord(int id)
{
code cleanup
2022-09-03 19:25:15 +02:00
switch (id)
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
{
code cleanup
2022-09-03 19:25:15 +02:00
case ESM::REC_ACTI:
case ESM::REC_ALCH:
case ESM::REC_APPA:
case ESM::REC_ARMO:
case ESM::REC_BOOK:
case ESM::REC_CLOT:
case ESM::REC_CONT:
case ESM::REC_CREA:
case ESM::REC_DOOR:
case ESM::REC_INGR:
case ESM::REC_LEVC:
case ESM::REC_LEVI:
case ESM::REC_LIGH:
case ESM::REC_LOCK:
case ESM::REC_MISC:
case ESM::REC_NPC_:
case ESM::REC_PROB:
case ESM::REC_REPA:
case ESM::REC_STAT:
case ESM::REC_WEAP:
case ESM::REC_BODY:
Static references are created as refcells, nothing displayed yet.
2023-01-23 23:51:45 +01:00
case ESM::REC_STAT4:
Load light models in the engine. Doesn't emit light
2023-02-06 20:22:17 +01:00
case ESM::REC_LIGH4:
code cleanup
2022-09-03 19:25:15 +02:00
return true;
break;
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
}
just replace esmstore, inconsistent
2012-11-05 18:09:14 +04:00
return false;
}
Implement ignored records
2022-04-16 16:28:39 +02:00
void ESMStore::load(ESM::ESMReader& esm, Loading::Listener* listener, ESM::Dialogue*& dialogue)
Main program now stores ESM data in memory
2010-05-17 17:35:42 +02:00
{
Show groundcover loading progress
2022-06-03 18:59:08 +02:00
if (listener != nullptr)
listener->setProgressRange(::EsmLoader::fileProgress);
Refactored loading screen - Add loading progress for data files, global map, terrain - Refactored and improved cell loading progress
2013-08-27 15:48:13 +02:00
Do not assert() for invalid land data in plugins (Bug #3037) The resizing of LTEX store to the correct number of plugins was done in the load() method, but the load method won't be called if a plugin contains LAND records but doesn't contain LTEX records. For such plugins the Store<ESM::LandTexture>::search() function would then fail an assertion.
2015-11-27 21:40:36 +01:00
// Land texture loading needs to use a separate internal store for each plugin.
refactors parentFileIndices (#3211) This PR aims to start addressing `ESM` design issues that have silenced errors we incorporated into groundcover `ESM` loading approaches. - We move the resolution of `parentFileIndices` from `ESMStore` to `ESMReader` as suggested in a `TODO` comment. - We improve a highly misleading comment which downplayed the significance of `parentFileIndices`. - We document important preconditions. - We move a user facing error message to the highest level and improve its context. - We remove an inappropriate `setGlobalReaderList` method. We now pass this reader list into the method that requires it. - We remove a thoroughly pointless optimisation of `Store<ESM::LandTexture>`'s construction that has unnecessarily depended on `getGlobalReaderList`. There should be no functional changes for `master`, but this PR should remove an issue blocking PR #3208.
2021-11-03 08:02:06 +00:00
// We set the number of plugins here so we can properly verify if valid plugin
// indices are being passed to the LandTexture Store retrieval methods.
Store refactor: Now way easier to create new stores, a good chunk of it is automated, only simple macros are used now The case of indexedStores remains an issue, because they can't be stored with the rest because they don't inherit of store base
2022-07-03 18:11:20 +02:00
getWritable<ESM::LandTexture>().resize(esm.getIndex() + 1);
- Add some updated files missing from last commit. - Move plugin dependency test from esmreader.cpp to esmstpre.cpp; fixes crash in omwlauncher.
2012-11-25 17:19:29 +01:00
read info records
2010-08-06 15:19:39 +02:00
// Loop through all records
while (esm.hasMoreRecs())
Implement ignored records
2022-04-16 16:28:39 +02:00
{
move ESMStore to MWWorld
2012-10-01 19:17:04 +04:00
ESM::NAME n = esm.getRecName();
read info records
2010-08-06 15:19:39 +02:00
esm.getRecHeader();
Implement ignored records
2022-04-16 16:28:39 +02:00
if (esm.getRecordFlags() & ESM::FLAG_Ignored)
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
{
Implement ignored records
2022-04-16 16:28:39 +02:00
esm.skipRecord();
continue;
}
Main program now stores ESM data in memory
2010-05-17 17:35:42 +02:00
read info records
2010-08-06 15:19:39 +02:00
// Look up the record type.
Fixes a bug with es3overrideRecord, esm3InsertStatic and esm3StoreInsert This also comes with a change to the mapping from esm3 record name to Store to be more direct with a pointer and the creation of a map from store pointer to esm3 record type
2022-07-20 19:04:31 +02:00
ESM::RecNameInts recName = static_cast<ESM::RecNameInts>(n.toInt());
Rename member to be clearer and more generic
2022-09-02 23:39:38 +02:00
const auto& it = mStoreImp->mRecNameToStore.find(recName);
just replace esmstore, inconsistent
2012-11-05 18:09:14 +04:00
Rename member to be clearer and more generic
2022-09-02 23:39:38 +02:00
if (it == mStoreImp->mRecNameToStore.end())
Fixes a bug with es3overrideRecord, esm3InsertStatic and esm3StoreInsert This also comes with a change to the mapping from esm3 record name to Store to be more direct with a pointer and the creation of a map from store pointer to esm3 record type
2022-07-20 19:04:31 +02:00
{
if (recName == ESM::REC_INFO)
Merge DialInfo objects by subrecord instead of overwriting the object Fixes #1360
2014-05-31 00:36:37 +02:00
{
if (dialogue)
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
{
Merge DialInfo objects by subrecord instead of overwriting the object Fixes #1360
2014-05-31 00:36:37 +02:00
dialogue->readInfo(esm, esm.getIndex() != 0);
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
}
else
{
Use new logging system for game itself
2018-08-14 23:05:43 +04:00
Log(Debug::Error) << "Error: info record without dialog";
add project file to content file list when running OpenMW from OpenCS
2014-09-13 20:48:24 +02:00
esm.skipRecord();
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
}
Merge DialInfo objects by subrecord instead of overwriting the object Fixes #1360
2014-05-31 00:36:37 +02:00
}
else if (n.toInt() == ESM::REC_MGEF)
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
{
Store refactor: Now way easier to create new stores, a good chunk of it is automated, only simple macros are used now The case of indexedStores remains an issue, because they can't be stored with the rest because they don't inherit of store base
2022-07-03 18:11:20 +02:00
getWritable<ESM::MagicEffect>().load(esm);
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
}
Do not use union to access FIXED_STRING<4> as int https://en.cppreference.com/w/cpp/language/union: > It's undefined behavior to read from the member of the union that wasn't most recently written.
2021-10-17 02:52:22 +02:00
else if (n.toInt() == ESM::REC_SKIL)
Merge DialInfo objects by subrecord instead of overwriting the object Fixes #1360
2014-05-31 00:36:37 +02:00
{
Use new logging system for game itself
2018-08-14 23:05:43 +04:00
getWritable<ESM::Skill>().load(esm);
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
}
Use new logging system for game itself
2018-08-14 23:05:43 +04:00
else if (n.toInt() == ESM::REC_FILT || n.toInt() == ESM::REC_DBGP)
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
{
Use new logging system for game itself
2018-08-14 23:05:43 +04:00
// ignore project file only records
read info records
2010-08-06 15:19:39 +02:00
esm.skipRecord();
}
Load LuaScriptsCfg from both *.omwscripts and *.omwaddon files.
2021-09-26 15:21:46 +02:00
else if (n.toInt() == ESM::REC_LUAL)
{
ESM::LuaScriptsCfg cfg;
cfg.load(esm);
Lua scripts configuration in omwaddon
2022-05-20 21:47:13 +02:00
cfg.adjustRefNums(esm);
Load LuaScriptsCfg from both *.omwscripts and *.omwaddon files.
2021-09-26 15:21:46 +02:00
mLuaContent.push_back(std::move(cfg));
}
add project file to content file list when running OpenMW from OpenCS
2014-09-13 20:48:24 +02:00
else
{
Reduce a bit the code bloat in hot paths/small functions There is no need to bloat hot/small functions with a ton of function calls and involving locales for simple error messages.
2021-06-10 12:55:18 +02:00
throw std::runtime_error("Unknown record: " + n.toString());
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
}
read info records
2010-08-06 15:19:39 +02:00
}
just replace esmstore, inconsistent
2012-11-05 18:09:14 +04:00
else
Remove NAME handling from MWWorld::ESMStore
2015-07-08 21:26:20 +03:00
{
Fixes a bug with es3overrideRecord, esm3InsertStatic and esm3StoreInsert This also comes with a change to the mapping from esm3 record name to Store to be more direct with a pointer and the creation of a map from store pointer to esm3 record type
2022-07-20 19:04:31 +02:00
RecordId id = it->second->load(esm);
Load/read methods in MWWorld::Store return a pair (record ID, deleted flag)
2015-07-12 15:20:22 +03:00
if (id.mIsDeleted)
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
{
Fixes a bug with es3overrideRecord, esm3InsertStatic and esm3StoreInsert This also comes with a change to the mapping from esm3 record name to Store to be more direct with a pointer and the creation of a map from store pointer to esm3 record type
2022-07-20 19:04:31 +02:00
it->second->eraseStatic(id.mId);
Remove NAME handling from MWWorld::ESMStore
2015-07-08 21:26:20 +03:00
continue;
Recognize DELE subrecords at the end of the record (Fixes #1414)
2014-06-07 19:21:37 +02:00
}
Do not use union to access FIXED_STRING<4> as int https://en.cppreference.com/w/cpp/language/union: > It's undefined behavior to read from the member of the union that wasn't most recently written.
2021-10-17 02:52:22 +02:00
if (n.toInt() == ESM::REC_DIAL)
{
Store refactor: Now way easier to create new stores, a good chunk of it is automated, only simple macros are used now The case of indexedStores remains an issue, because they can't be stored with the rest because they don't inherit of store base
2022-07-03 18:11:20 +02:00
dialogue = const_cast<ESM::Dialogue*>(getWritable<ESM::Dialogue>().find(id.mId));
just replace esmstore, inconsistent
2012-11-05 18:09:14 +04:00
}
else
{
Replace zeroes and nulls by nullptrs
2020-11-13 11:39:47 +04:00
dialogue = nullptr;
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
}
just replace esmstore, inconsistent
2012-11-05 18:09:14 +04:00
}
Show groundcover loading progress
2022-06-03 18:59:08 +02:00
if (listener != nullptr)
listener->setProgress(::EsmLoader::fileProgress * esm.getFileOffset() / esm.getFileSize());
read info records
2010-08-06 15:19:39 +02:00
}
Main program now stores ESM data in memory
2010-05-17 17:35:42 +02:00
}
move ESMStore to MWWorld
2012-10-01 19:17:04 +04:00
Applies review advice.
2023-01-04 12:20:02 +01:00
void ESMStore::loadESM4(ESM4::Reader& reader)
Initial commit to load ESM4 Some data is actually loaded and store in ESM Store Any new ESM4 will go through the same code path and be automatically sent to the right store
2022-12-28 20:48:25 +01:00
{
The esm4 reader logic is mutualised to avoid copy pasting code, readerutils gives functions that take visitors as params to decide how a record must be handled Check encoder exists, and get value of stateless encoder. fixes code formatting conventions Fixed output of record with RefId also fixed readTypedRecord and readRecord to have the proper return types Check if the type has a sRecordId
2022-12-30 11:28:19 +01:00
auto visitorRec = [this](ESM4::Reader& reader) { return ESMStoreImp::readRecord(reader, *this); };
ESM4::ReaderUtils::readAll(reader, visitorRec, [](ESM4::Reader&) {});
Initial commit to load ESM4 Some data is actually loaded and store in ESM Store Any new ESM4 will go through the same code path and be automatically sent to the right store
2022-12-28 20:48:25 +01:00
}
Initial commit: In ESM structures, replace the string members that are RefIds to other records, to a new strong type The strong type is actually just a string underneath, but this will help in the future to have a distinction so it's easier to search and replace when we use an integer ID Slowly going through all the changes to make, still hundreds of errors a lot of functions/structures use std::string or stringview to designate an ID. So it takes time Continues slowly replacing ids. There are technically more and more compilation errors I have good hope that there is a point where the amount of errors will dramatically go down as all the main functions use the ESM::RefId type Continue moving forward, changes to the stores slowly moving along Starting to see the fruit of those changes. still many many error, but more and more Irun into a situation where a function is sandwiched between two functions that use the RefId type. More replacements. Things are starting to get easier I can see more and more often the issue is that the function is awaiting a RefId, but is given a string there is less need to go down functions and to fix a long list of them. Still moving forward, and for the first time error count is going down! Good pace, not sure about topics though, mId and mName are actually the same thing and are used interchangeably Cells are back to using string for the name, haven't fixed everything yet. Many other changes Under the bar of 400 compilation errors. more good progress <100 compile errors! More progress Game settings store can use string for find, it was a bit absurd how every use of it required to create refId from string some more progress on other fronts Mostly game settings clean one error opened a lot of other errors. Down to 18, but more will prbably appear only link errors left?? Fixed link errors OpenMW compiles, and launches, with some issues, but still!
2022-09-25 13:17:09 +02:00
void ESMStore::setIdType(const ESM::RefId& id, ESM::RecNameInts type)
moved a bit of code back in the header to avoid many template specialization
2022-09-06 14:19:55 +02:00
{
no more getIdType, only setIdType
2022-09-07 18:45:15 +02:00
mStoreImp->mIds[id] = type;
moved a bit of code back in the header to avoid many template specialization
2022-09-06 14:19:55 +02:00
}
Load LuaScriptsCfg from both *.omwscripts and *.omwaddon files.
2021-09-26 15:21:46 +02:00
ESM::LuaScriptsCfg ESMStore::getLuaScriptsCfg() const
{
Make sure all paths are passed as std::filesystem::path instead of std::string where possible.
2022-06-19 13:28:33 +02:00
ESM::LuaScriptsCfg cfg;
for (const LuaContent& c : mLuaContent)
Load LuaScriptsCfg from both *.omwscripts and *.omwaddon files.
2021-09-26 15:21:46 +02:00
{
if (std::holds_alternative<std::filesystem::path>(c))
{
// *.omwscripts are intentionally reloaded every time when `getLuaScriptsCfg` is called.
// It is important for the `reloadlua` console command.
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
try
{
Make sure all paths are passed as std::filesystem::path instead of std::string where possible.
2022-06-19 13:28:33 +02:00
auto file = std::ifstream(std::get<std::filesystem::path>(c));
Load LuaScriptsCfg from both *.omwscripts and *.omwaddon files.
2021-09-26 15:21:46 +02:00
std::string fileContent(std::istreambuf_iterator<char>(file), {});
LuaUtil::parseOMWScripts(cfg, fileContent);
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
}
Load LuaScriptsCfg from both *.omwscripts and *.omwaddon files.
2021-09-26 15:21:46 +02:00
catch (std::exception& e)
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
{
Load LuaScriptsCfg from both *.omwscripts and *.omwaddon files.
2021-09-26 15:21:46 +02:00
Log(Debug::Error) << e.what();
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
}
}
else
{
Load LuaScriptsCfg from both *.omwscripts and *.omwaddon files.
2021-09-26 15:21:46 +02:00
const ESM::LuaScriptsCfg& addition = std::get<ESM::LuaScriptsCfg>(c);
cfg.mScripts.insert(cfg.mScripts.end(), addition.mScripts.begin(), addition.mScripts.end());
}
}
return cfg;
}
Limit the number of simultaneously open not actively used content files Use LRU cache for ESMReaders. When cache capacity is reached close least recently used ESMReader. Remember the file name if a reader was open. Once the reader requested again open the file if there is stored name for it. Put released ESMReader to the back of the free items list. Close ESMReader's from the front of the free items list. Cached item can be used only by one client at the same time. If the same item is requested twice exception is thrown. This should never happen in practice. If this happens need to fix the client logic. It's allowed to go over the capacity limit when requesting different readers. Ideally this should never happen but there will be system error anyway signalizing about too many open files. Need to fix client logic in this case. All places that were using a vector of ESMReaders now using the cache. Cache is local for each use case and there is no need for a thread safety.
2022-06-01 22:53:18 +02:00
void ESMStore::setUp()
just replace esmstore, inconsistent
2012-11-05 18:09:14 +04:00
{
mids and mStaticIds moved to mStoreImp renamed recordid to storeid
2022-07-20 19:28:06 +02:00
mStoreImp->mIds.clear();
Minor fix
2015-12-11 16:59:13 +01:00
ESMStore: fill mIds in setUp()
2015-12-11 15:55:45 +01:00
std::map<ESM::RecNameInts, DynamicStore*>::iterator storeIt = mStoreImp->mRecNameToStore.begin();
for (; storeIt != mStoreImp->mRecNameToStore.end(); ++storeIt)
{
Fixes a bug with es3overrideRecord, esm3InsertStatic and esm3StoreInsert This also comes with a change to the mapping from esm3 record name to Store to be more direct with a pointer and the creation of a map from store pointer to esm3 record type
2022-07-20 19:04:31 +02:00
storeIt->second->setUp();
ESMStore: fill mIds in setUp()
2015-12-11 15:55:45 +01:00
if (isCacheableRecord(storeIt->first))
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
{
Initial commit: In ESM structures, replace the string members that are RefIds to other records, to a new strong type The strong type is actually just a string underneath, but this will help in the future to have a distinction so it's easier to search and replace when we use an integer ID Slowly going through all the changes to make, still hundreds of errors a lot of functions/structures use std::string or stringview to designate an ID. So it takes time Continues slowly replacing ids. There are technically more and more compilation errors I have good hope that there is a point where the amount of errors will dramatically go down as all the main functions use the ESM::RefId type Continue moving forward, changes to the stores slowly moving along Starting to see the fruit of those changes. still many many error, but more and more Irun into a situation where a function is sandwiched between two functions that use the RefId type. More replacements. Things are starting to get easier I can see more and more often the issue is that the function is awaiting a RefId, but is given a string there is less need to go down functions and to fix a long list of them. Still moving forward, and for the first time error count is going down! Good pace, not sure about topics though, mId and mName are actually the same thing and are used interchangeably Cells are back to using string for the name, haven't fixed everything yet. Many other changes Under the bar of 400 compilation errors. more good progress <100 compile errors! More progress Game settings store can use string for find, it was a bit absurd how every use of it required to create refId from string some more progress on other fronts Mostly game settings clean one error opened a lot of other errors. Down to 18, but more will prbably appear only link errors left?? Fixed link errors OpenMW compiles, and launches, with some issues, but still!
2022-09-25 13:17:09 +02:00
std::vector<ESM::RefId> identifiers;
Fixes a bug with es3overrideRecord, esm3InsertStatic and esm3StoreInsert This also comes with a change to the mapping from esm3 record name to Store to be more direct with a pointer and the creation of a map from store pointer to esm3 record type
2022-07-20 19:04:31 +02:00
storeIt->second->listIdentifier(identifiers);
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
Initial commit: In ESM structures, replace the string members that are RefIds to other records, to a new strong type The strong type is actually just a string underneath, but this will help in the future to have a distinction so it's easier to search and replace when we use an integer ID Slowly going through all the changes to make, still hundreds of errors a lot of functions/structures use std::string or stringview to designate an ID. So it takes time Continues slowly replacing ids. There are technically more and more compilation errors I have good hope that there is a point where the amount of errors will dramatically go down as all the main functions use the ESM::RefId type Continue moving forward, changes to the stores slowly moving along Starting to see the fruit of those changes. still many many error, but more and more Irun into a situation where a function is sandwiched between two functions that use the RefId type. More replacements. Things are starting to get easier I can see more and more often the issue is that the function is awaiting a RefId, but is given a string there is less need to go down functions and to fix a long list of them. Still moving forward, and for the first time error count is going down! Good pace, not sure about topics though, mId and mName are actually the same thing and are used interchangeably Cells are back to using string for the name, haven't fixed everything yet. Many other changes Under the bar of 400 compilation errors. more good progress <100 compile errors! More progress Game settings store can use string for find, it was a bit absurd how every use of it required to create refId from string some more progress on other fronts Mostly game settings clean one error opened a lot of other errors. Down to 18, but more will prbably appear only link errors left?? Fixed link errors OpenMW compiles, and launches, with some issues, but still!
2022-09-25 13:17:09 +02:00
for (auto& record : identifiers)
mStoreImp->mIds[record] = storeIt->first;
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
}
ESMStore: fill mIds in setUp()
2015-12-11 15:55:45 +01:00
}
terrainbased objectpaging Signed-off-by: Bret Curtis <psi29a@gmail.com>
2019-06-13 13:37:00 +00:00
mids and mStaticIds moved to mStoreImp renamed recordid to storeid
2022-07-20 19:28:06 +02:00
if (mStoreImp->mStaticIds.empty())
for (const auto& [k, v] : mStoreImp->mIds)
Initial commit: In ESM structures, replace the string members that are RefIds to other records, to a new strong type The strong type is actually just a string underneath, but this will help in the future to have a distinction so it's easier to search and replace when we use an integer ID Slowly going through all the changes to make, still hundreds of errors a lot of functions/structures use std::string or stringview to designate an ID. So it takes time Continues slowly replacing ids. There are technically more and more compilation errors I have good hope that there is a point where the amount of errors will dramatically go down as all the main functions use the ESM::RefId type Continue moving forward, changes to the stores slowly moving along Starting to see the fruit of those changes. still many many error, but more and more Irun into a situation where a function is sandwiched between two functions that use the RefId type. More replacements. Things are starting to get easier I can see more and more often the issue is that the function is awaiting a RefId, but is given a string there is less need to go down functions and to fix a long list of them. Still moving forward, and for the first time error count is going down! Good pace, not sure about topics though, mId and mName are actually the same thing and are used interchangeably Cells are back to using string for the name, haven't fixed everything yet. Many other changes Under the bar of 400 compilation errors. more good progress <100 compile errors! More progress Game settings store can use string for find, it was a bit absurd how every use of it required to create refId from string some more progress on other fronts Mostly game settings clean one error opened a lot of other errors. Down to 18, but more will prbably appear only link errors left?? Fixed link errors OpenMW compiles, and launches, with some issues, but still!
2022-09-25 13:17:09 +02:00
mStoreImp->mStaticIds.emplace(k, v);
terrainbased objectpaging Signed-off-by: Bret Curtis <psi29a@gmail.com>
2019-06-13 13:37:00 +00:00
Store refactor: Now way easier to create new stores, a good chunk of it is automated, only simple macros are used now The case of indexedStores remains an issue, because they can't be stored with the rest because they don't inherit of store base
2022-07-03 18:11:20 +02:00
getWritable<ESM::Skill>().setUp();
getWritable<ESM::MagicEffect>().setUp();
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
;
Store refactor: Now way easier to create new stores, a good chunk of it is automated, only simple macros are used now The case of indexedStores remains an issue, because they can't be stored with the rest because they don't inherit of store base
2022-07-03 18:11:20 +02:00
getWritable<ESM::Attribute>().setUp();
getWritable<ESM::Dialogue>().setUp();
implement additem/removeitem for non-unique actors
2020-07-26 11:07:18 +02:00
}
Limit the number of simultaneously open not actively used content files Use LRU cache for ESMReaders. When cache capacity is reached close least recently used ESMReader. Remember the file name if a reader was open. Once the reader requested again open the file if there is stored name for it. Put released ESMReader to the back of the free items list. Close ESMReader's from the front of the free items list. Cached item can be used only by one client at the same time. If the same item is requested twice exception is thrown. This should never happen in practice. If this happens need to fix the client logic. It's allowed to go over the capacity limit when requesting different readers. Ideally this should never happen but there will be system error anyway signalizing about too many open files. Need to fix client logic in this case. All places that were using a vector of ESMReaders now using the cache. Cache is local for each use case and there is no need for a thread safety.
2022-06-01 22:53:18 +02:00
void ESMStore::validateRecords(ESM::ReadersCache& readers)
Use stable sort+unique to collect RefIDs for ESMStore records counting The idea is to avoid std::map lookup for each CellRef. Instead generate a sequence of added and removed RefNums into a vector then order them by RefNum using a stable sort that preserves relative order of elements with the same RefNum. RefIDs are stored in a different vector to avoid std::string move ctor calls when swapping elements while sorting. Reversed iteration over added and removed RefNums for each unique RefNum is an equivalent to what map-based algorithm produces. The main benefit from sorting a vector is a data locality that means less cache misses for each access. Reduces ESMStore::countRecords perf cycles by 25%.
2021-04-06 23:59:58 +02:00
{
validate();
Add key flag on load
2022-09-30 12:16:45 +02:00
countAllCellRefsAndMarkKeys(readers);
Add NPC validation to esmstore (bug #2772)
2018-06-09 20:47:17 +04:00
}
Add key flag on load
2022-09-30 12:16:45 +02:00
void ESMStore::countAllCellRefsAndMarkKeys(ESM::ReadersCache& readers)
Add NPC validation to esmstore (bug #2772)
2018-06-09 20:47:17 +04:00
{
Fixed naming convention problems removed 2 macros. - One is replaced by the underlying code - The second one ise replaced by a template function removed tabulation used as indentation used getWritable instead of a const_cast used for( val : cont) loop removed useless getId function
2022-07-19 17:58:21 +02:00
// TODO: We currently need to read entire files here again.
// We should consider consolidating or deferring this reading.
if (!mRefCount.empty())
return;
std::vector<Ref> refs;
Initial commit: In ESM structures, replace the string members that are RefIds to other records, to a new strong type The strong type is actually just a string underneath, but this will help in the future to have a distinction so it's easier to search and replace when we use an integer ID Slowly going through all the changes to make, still hundreds of errors a lot of functions/structures use std::string or stringview to designate an ID. So it takes time Continues slowly replacing ids. There are technically more and more compilation errors I have good hope that there is a point where the amount of errors will dramatically go down as all the main functions use the ESM::RefId type Continue moving forward, changes to the stores slowly moving along Starting to see the fruit of those changes. still many many error, but more and more Irun into a situation where a function is sandwiched between two functions that use the RefId type. More replacements. Things are starting to get easier I can see more and more often the issue is that the function is awaiting a RefId, but is given a string there is less need to go down functions and to fix a long list of them. Still moving forward, and for the first time error count is going down! Good pace, not sure about topics though, mId and mName are actually the same thing and are used interchangeably Cells are back to using string for the name, haven't fixed everything yet. Many other changes Under the bar of 400 compilation errors. more good progress <100 compile errors! More progress Game settings store can use string for find, it was a bit absurd how every use of it required to create refId from string some more progress on other fronts Mostly game settings clean one error opened a lot of other errors. Down to 18, but more will prbably appear only link errors left?? Fixed link errors OpenMW compiles, and launches, with some issues, but still!
2022-09-25 13:17:09 +02:00
std::set<ESM::RefId> keyIDs;
std::vector<ESM::RefId> refIDs;
Add key flag on load
2022-09-30 12:16:45 +02:00
Store<ESM::Cell> Cells = get<ESM::Cell>();
Fixed naming convention problems removed 2 macros. - One is replaced by the underlying code - The second one ise replaced by a template function removed tabulation used as indentation used getWritable instead of a const_cast used for( val : cont) loop removed useless getId function
2022-07-19 17:58:21 +02:00
for (auto it = Cells.intBegin(); it != Cells.intEnd(); ++it)
Add key flag on load
2022-09-30 12:16:45 +02:00
readRefs(*it, refs, refIDs, keyIDs, readers);
Store refactor: Now way easier to create new stores, a good chunk of it is automated, only simple macros are used now The case of indexedStores remains an issue, because they can't be stored with the rest because they don't inherit of store base
2022-07-03 18:11:20 +02:00
for (auto it = Cells.extBegin(); it != Cells.extEnd(); ++it)
Add key flag on load
2022-09-30 12:16:45 +02:00
readRefs(*it, refs, refIDs, keyIDs, readers);
Fixed naming convention problems removed 2 macros. - One is replaced by the underlying code - The second one ise replaced by a template function removed tabulation used as indentation used getWritable instead of a const_cast used for( val : cont) loop removed useless getId function
2022-07-19 17:58:21 +02:00
const auto lessByRefNum = [](const Ref& l, const Ref& r) { return l.mRefNum < r.mRefNum; };
std::stable_sort(refs.begin(), refs.end(), lessByRefNum);
Use stable sort+unique to collect RefIDs for ESMStore records counting The idea is to avoid std::map lookup for each CellRef. Instead generate a sequence of added and removed RefNums into a vector then order them by RefNum using a stable sort that preserves relative order of elements with the same RefNum. RefIDs are stored in a different vector to avoid std::string move ctor calls when swapping elements while sorting. Reversed iteration over added and removed RefNums for each unique RefNum is an equivalent to what map-based algorithm produces. The main benefit from sorting a vector is a data locality that means less cache misses for each access. Reduces ESMStore::countRecords perf cycles by 25%.
2021-04-06 23:59:58 +02:00
const auto equalByRefNum = [](const Ref& l, const Ref& r) { return l.mRefNum == r.mRefNum; };
const auto incrementRefCount = [&](const Ref& value) {
if (value.mRefID != deletedRefID)
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
{
Initial commit: In ESM structures, replace the string members that are RefIds to other records, to a new strong type The strong type is actually just a string underneath, but this will help in the future to have a distinction so it's easier to search and replace when we use an integer ID Slowly going through all the changes to make, still hundreds of errors a lot of functions/structures use std::string or stringview to designate an ID. So it takes time Continues slowly replacing ids. There are technically more and more compilation errors I have good hope that there is a point where the amount of errors will dramatically go down as all the main functions use the ESM::RefId type Continue moving forward, changes to the stores slowly moving along Starting to see the fruit of those changes. still many many error, but more and more Irun into a situation where a function is sandwiched between two functions that use the RefId type. More replacements. Things are starting to get easier I can see more and more often the issue is that the function is awaiting a RefId, but is given a string there is less need to go down functions and to fix a long list of them. Still moving forward, and for the first time error count is going down! Good pace, not sure about topics though, mId and mName are actually the same thing and are used interchangeably Cells are back to using string for the name, haven't fixed everything yet. Many other changes Under the bar of 400 compilation errors. more good progress <100 compile errors! More progress Game settings store can use string for find, it was a bit absurd how every use of it required to create refId from string some more progress on other fronts Mostly game settings clean one error opened a lot of other errors. Down to 18, but more will prbably appear only link errors left?? Fixed link errors OpenMW compiles, and launches, with some issues, but still!
2022-09-25 13:17:09 +02:00
ESM::RefId& refId = refIDs[value.mRefID];
Use stable sort+unique to collect RefIDs for ESMStore records counting The idea is to avoid std::map lookup for each CellRef. Instead generate a sequence of added and removed RefNums into a vector then order them by RefNum using a stable sort that preserves relative order of elements with the same RefNum. RefIDs are stored in a different vector to avoid std::string move ctor calls when swapping elements while sorting. Reversed iteration over added and removed RefNums for each unique RefNum is an equivalent to what map-based algorithm produces. The main benefit from sorting a vector is a data locality that means less cache misses for each access. Reduces ESMStore::countRecords perf cycles by 25%.
2021-04-06 23:59:58 +02:00
++mRefCount[std::move(refId)];
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
}
};
Fixed naming convention problems removed 2 macros. - One is replaced by the underlying code - The second one ise replaced by a template function removed tabulation used as indentation used getWritable instead of a const_cast used for( val : cont) loop removed useless getId function
2022-07-19 17:58:21 +02:00
Misc::forEachUnique(refs.rbegin(), refs.rend(), equalByRefNum, incrementRefCount);
Add key flag on load
2022-09-30 12:16:45 +02:00
auto& store = getWritable<ESM::Miscellaneous>().mStatic;
fixed Topic selected event Fix rebase
2022-10-14 00:41:59 +02:00
for (const auto& id : keyIDs)
Add key flag on load
2022-09-30 12:16:45 +02:00
{
auto it = store.find(id);
if (it != store.end())
it->second.mData.mFlags |= ESM::Miscellaneous::Key;
}
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
}
Add NPC validation to esmstore (bug #2772)
2018-06-09 20:47:17 +04:00
Initial commit: In ESM structures, replace the string members that are RefIds to other records, to a new strong type The strong type is actually just a string underneath, but this will help in the future to have a distinction so it's easier to search and replace when we use an integer ID Slowly going through all the changes to make, still hundreds of errors a lot of functions/structures use std::string or stringview to designate an ID. So it takes time Continues slowly replacing ids. There are technically more and more compilation errors I have good hope that there is a point where the amount of errors will dramatically go down as all the main functions use the ESM::RefId type Continue moving forward, changes to the stores slowly moving along Starting to see the fruit of those changes. still many many error, but more and more Irun into a situation where a function is sandwiched between two functions that use the RefId type. More replacements. Things are starting to get easier I can see more and more often the issue is that the function is awaiting a RefId, but is given a string there is less need to go down functions and to fix a long list of them. Still moving forward, and for the first time error count is going down! Good pace, not sure about topics though, mId and mName are actually the same thing and are used interchangeably Cells are back to using string for the name, haven't fixed everything yet. Many other changes Under the bar of 400 compilation errors. more good progress <100 compile errors! More progress Game settings store can use string for find, it was a bit absurd how every use of it required to create refId from string some more progress on other fronts Mostly game settings clean one error opened a lot of other errors. Down to 18, but more will prbably appear only link errors left?? Fixed link errors OpenMW compiles, and launches, with some issues, but still!
2022-09-25 13:17:09 +02:00
int ESMStore::getRefCount(const ESM::RefId& id) const
Add NPC validation to esmstore (bug #2772)
2018-06-09 20:47:17 +04:00
{
Initial commit: In ESM structures, replace the string members that are RefIds to other records, to a new strong type The strong type is actually just a string underneath, but this will help in the future to have a distinction so it's easier to search and replace when we use an integer ID Slowly going through all the changes to make, still hundreds of errors a lot of functions/structures use std::string or stringview to designate an ID. So it takes time Continues slowly replacing ids. There are technically more and more compilation errors I have good hope that there is a point where the amount of errors will dramatically go down as all the main functions use the ESM::RefId type Continue moving forward, changes to the stores slowly moving along Starting to see the fruit of those changes. still many many error, but more and more Irun into a situation where a function is sandwiched between two functions that use the RefId type. More replacements. Things are starting to get easier I can see more and more often the issue is that the function is awaiting a RefId, but is given a string there is less need to go down functions and to fix a long list of them. Still moving forward, and for the first time error count is going down! Good pace, not sure about topics though, mId and mName are actually the same thing and are used interchangeably Cells are back to using string for the name, haven't fixed everything yet. Many other changes Under the bar of 400 compilation errors. more good progress <100 compile errors! More progress Game settings store can use string for find, it was a bit absurd how every use of it required to create refId from string some more progress on other fronts Mostly game settings clean one error opened a lot of other errors. Down to 18, but more will prbably appear only link errors left?? Fixed link errors OpenMW compiles, and launches, with some issues, but still!
2022-09-25 13:17:09 +02:00
auto it = mRefCount.find(id);
implement additem/removeitem for non-unique actors
2020-07-26 11:07:18 +02:00
if (it == mRefCount.end())
return 0;
Fixed naming convention problems removed 2 macros. - One is replaced by the underlying code - The second one ise replaced by a template function removed tabulation used as indentation used getWritable instead of a const_cast used for( val : cont) loop removed useless getId function
2022-07-19 17:58:21 +02:00
return it->second;
Add NPC validation to esmstore (bug #2772)
2018-06-09 20:47:17 +04:00
}
Add spell effect argument validation (bug #4927)
2019-03-31 18:54:12 +03:00
void ESMStore::validate()
{
Fixed naming convention problems removed 2 macros. - One is replaced by the underlying code - The second one ise replaced by a template function removed tabulation used as indentation used getWritable instead of a const_cast used for( val : cont) loop removed useless getId function
2022-07-19 17:58:21 +02:00
auto& npcs = getWritable<ESM::NPC>();
Add spell effect argument validation (bug #4927)
2019-03-31 18:54:12 +03:00
std::vector<ESM::NPC> npcsToReplace
= getNPCsToReplace(getWritable<ESM::Faction>(), getWritable<ESM::Class>(), npcs.mStatic);
Erase effects with invalid indices from spells (bug #5050)
2019-05-31 11:23:58 +03:00
for (const ESM::NPC& npc : npcsToReplace)
Add spell effect argument validation (bug #4927)
2019-03-31 18:54:12 +03:00
{
Store refactor: Now way easier to create new stores, a good chunk of it is automated, only simple macros are used now The case of indexedStores remains an issue, because they can't be stored with the rest because they don't inherit of store base
2022-07-03 18:11:20 +02:00
npcs.eraseStatic(npc.mId);
Fix typos in invalid spell effect warnings, clarify them
2019-10-17 23:45:27 +03:00
npcs.insertStatic(npc);
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
}
Fix typos in invalid spell effect warnings, clarify them
2019-10-17 23:45:27 +03:00
// Validate spell effects for invalid arguments
std::vector<ESM::Spell> spellsToReplace;
Erase effects with invalid indices from spells (bug #5050)
2019-05-31 11:23:58 +03:00
auto& spells = getWritable<ESM::Spell>();
for (ESM::Spell spell : spells)
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
{
Erase effects with invalid indices from spells (bug #5050)
2019-05-31 11:23:58 +03:00
if (spell.mEffects.mList.empty())
Add spell effect argument validation (bug #4927)
2019-03-31 18:54:12 +03:00
continue;
bool changed = false;
auto iter = spell.mEffects.mList.begin();
while (iter != spell.mEffects.mList.end())
{
Erase effects with invalid indices from spells (bug #5050)
2019-05-31 11:23:58 +03:00
const ESM::MagicEffect* mgef = getWritable<ESM::MagicEffect>().search(iter->mEffectID);
if (!mgef)
Add spell effect argument validation (bug #4927)
2019-03-31 18:54:12 +03:00
{
Fix typos in invalid spell effect warnings, clarify them
2019-10-17 23:45:27 +03:00
Log(Debug::Verbose) << "Spell '" << spell.mId << "' has an invalid effect (index "
<< iter->mEffectID << ") present. Dropping the effect.";
iter = spell.mEffects.mList.erase(iter);
Add spell effect argument validation (bug #4927)
2019-03-31 18:54:12 +03:00
changed = true;
continue;
}
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
Add spell effect argument validation (bug #4927)
2019-03-31 18:54:12 +03:00
if (mgef->mData.mFlags & ESM::MagicEffect::TargetSkill)
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
{
Erase effects with invalid indices from spells (bug #5050)
2019-05-31 11:23:58 +03:00
if (iter->mAttribute != -1)
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
{
Erase effects with invalid indices from spells (bug #5050)
2019-05-31 11:23:58 +03:00
iter->mAttribute = -1;
Log(Debug::Verbose)
<< ESM::MagicEffect::effectIdToString(iter->mEffectID) << " effect of spell '" << spell.mId
Fix typos in invalid spell effect warnings, clarify them
2019-10-17 23:45:27 +03:00
<< "' has an attribute argument present. Dropping the argument.";
Add spell effect argument validation (bug #4927)
2019-03-31 18:54:12 +03:00
changed = true;
}
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
}
Add spell effect argument validation (bug #4927)
2019-03-31 18:54:12 +03:00
else if (mgef->mData.mFlags & ESM::MagicEffect::TargetAttribute)
{
Erase effects with invalid indices from spells (bug #5050)
2019-05-31 11:23:58 +03:00
if (iter->mSkill != -1)
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
{
Erase effects with invalid indices from spells (bug #5050)
2019-05-31 11:23:58 +03:00
iter->mSkill = -1;
Log(Debug::Verbose)
<< ESM::MagicEffect::effectIdToString(iter->mEffectID) << " effect of spell '" << spell.mId
Fix typos in invalid spell effect warnings, clarify them
2019-10-17 23:45:27 +03:00
<< "' has a skill argument present. Dropping the argument.";
Add spell effect argument validation (bug #4927)
2019-03-31 18:54:12 +03:00
changed = true;
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
}
}
Erase effects with invalid indices from spells (bug #5050)
2019-05-31 11:23:58 +03:00
else if (iter->mSkill != -1 || iter->mAttribute != -1)
Add spell effect argument validation (bug #4927)
2019-03-31 18:54:12 +03:00
{
Erase effects with invalid indices from spells (bug #5050)
2019-05-31 11:23:58 +03:00
iter->mSkill = -1;
iter->mAttribute = -1;
Log(Debug::Verbose) << ESM::MagicEffect::effectIdToString(iter->mEffectID) << " effect of spell '"
Fix typos in invalid spell effect warnings, clarify them
2019-10-17 23:45:27 +03:00
<< spell.mId << "' has argument(s) present. Dropping the argument(s).";
Add spell effect argument validation (bug #4927)
2019-03-31 18:54:12 +03:00
changed = true;
}
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
Erase effects with invalid indices from spells (bug #5050)
2019-05-31 11:23:58 +03:00
++iter;
Add spell effect argument validation (bug #4927)
2019-03-31 18:54:12 +03:00
}
Erase effects with invalid indices from spells (bug #5050)
2019-05-31 11:23:58 +03:00
Add spell effect argument validation (bug #4927)
2019-03-31 18:54:12 +03:00
if (changed)
Erase effects with invalid indices from spells (bug #5050)
2019-05-31 11:23:58 +03:00
spellsToReplace.emplace_back(spell);
Add spell effect argument validation (bug #4927)
2019-03-31 18:54:12 +03:00
}
for (const ESM::Spell& spell : spellsToReplace)
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
{
Add spell effect argument validation (bug #4927)
2019-03-31 18:54:12 +03:00
spells.eraseStatic(spell.mId);
spells.insertStatic(spell);
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
}
Add spell effect argument validation (bug #4927)
2019-03-31 18:54:12 +03:00
}
void ESMStore::movePlayerRecord()
{
various fixes fixed naming convention replaced std::string by string_view when possible removed unused function and member varaible replaced type::value by type_v set default destructor in cpp function getTypeIndex => getnextindex
2022-09-06 23:10:58 +02:00
auto& npcs = getWritable<ESM::NPC>();
create a specific esm reader function for RefID to avoid allocation for string and then again for RefId Fixed some types removed useless header applied clang format fixed compile tests fixed clang tidy, and closer to logic before this MR Removed hardcoded refids unless there is a returned value we don't use static RefIds can use == between RefId and hardcoded string Fix clang format Fixed a few instances where std::string was used, when only const std::string& was needed removed unused variable
2022-10-18 09:26:55 +02:00
auto player = npcs.find(ESM::RefId::stringRefId("Player"));
various fixes fixed naming convention replaced std::string by string_view when possible removed unused function and member varaible replaced type::value by type_v set default destructor in cpp function getTypeIndex => getnextindex
2022-09-06 23:10:58 +02:00
npcs.insert(*player);
Add spell effect argument validation (bug #4927)
2019-03-31 18:54:12 +03:00
}
ESMStore: no longer necessary to include the type definitions in the header For now there still needs to declare all the getters, but a macro may make that easier, or an different method entierly
2022-07-03 11:47:46 +02:00
Also run NPC validation for modified base records
2021-03-23 21:07:57 +01:00
void ESMStore::validateDynamic()
{
Fixed naming convention problems removed 2 macros. - One is replaced by the underlying code - The second one ise replaced by a template function removed tabulation used as indentation used getWritable instead of a const_cast used for( val : cont) loop removed useless getId function
2022-07-19 17:58:21 +02:00
auto& npcs = getWritable<ESM::NPC>();
Store refactor: Now way easier to create new stores, a good chunk of it is automated, only simple macros are used now The case of indexedStores remains an issue, because they can't be stored with the rest because they don't inherit of store base
2022-07-03 18:11:20 +02:00
auto& scripts = getWritable<ESM::Script>();
Fixed naming convention problems removed 2 macros. - One is replaced by the underlying code - The second one ise replaced by a template function removed tabulation used as indentation used getWritable instead of a const_cast used for( val : cont) loop removed useless getId function
2022-07-19 17:58:21 +02:00
std::vector<ESM::NPC> npcsToReplace
= getNPCsToReplace(getWritable<ESM::Faction>(), getWritable<ESM::Class>(), npcs.mDynamic);
Also run NPC validation for modified base records
2021-03-23 21:07:57 +01:00
for (const ESM::NPC& npc : npcsToReplace)
Fixed naming convention problems removed 2 macros. - One is replaced by the underlying code - The second one ise replaced by a template function removed tabulation used as indentation used getWritable instead of a const_cast used for( val : cont) loop removed useless getId function
2022-07-19 17:58:21 +02:00
npcs.insert(npc);
Improve save cleaning
2021-07-12 20:51:13 +02:00
Store refactor: Now way easier to create new stores, a good chunk of it is automated, only simple macros are used now The case of indexedStores remains an issue, because they can't be stored with the rest because they don't inherit of store base
2022-07-03 18:11:20 +02:00
removeMissingScripts(scripts, getWritable<ESM::Armor>().mDynamic);
removeMissingScripts(scripts, getWritable<ESM::Book>().mDynamic);
removeMissingScripts(scripts, getWritable<ESM::Clothing>().mDynamic);
removeMissingScripts(scripts, getWritable<ESM::Weapon>().mDynamic);
Improve save cleaning
2021-07-12 20:51:13 +02:00
Store refactor: Now way easier to create new stores, a good chunk of it is automated, only simple macros are used now The case of indexedStores remains an issue, because they can't be stored with the rest because they don't inherit of store base
2022-07-03 18:11:20 +02:00
removeMissingObjects(getWritable<ESM::CreatureLevList>());
removeMissingObjects(getWritable<ESM::ItemLevList>());
Improve save cleaning
2021-07-12 20:51:13 +02:00
}
Static references are created as refcells, nothing displayed yet.
2023-01-23 23:51:45 +01:00
// Leveled lists can be modified by scripts. This removes items that no longer exist (presumably because the
// plugin was removed) from modified lists
Improve save cleaning
2021-07-12 20:51:13 +02:00
template <class T>
void ESMStore::removeMissingObjects(Store<T>& store)
{
Workaround clang and actually erase from the vector
2021-07-13 21:18:24 +02:00
for (auto& entry : store.mDynamic)
Improve save cleaning
2021-07-12 20:51:13 +02:00
{
auto first = std::remove_if(entry.second.mList.begin(), entry.second.mList.end(), [&](const auto& item) {
if (!find(item.mId))
{
Workaround clang and actually erase from the vector
2021-07-13 21:18:24 +02:00
Log(Debug::Verbose) << "Leveled list '" << entry.first << "' has nonexistent object '" << item.mId
<< "', ignoring it.";
Improve save cleaning
2021-07-12 20:51:13 +02:00
return true;
}
return false;
});
Workaround clang and actually erase from the vector
2021-07-13 21:18:24 +02:00
entry.second.mList.erase(first, entry.second.mList.end());
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
}
Improve save cleaning
2021-07-12 20:51:13 +02:00
}
Also run NPC validation for modified base records
2021-03-23 21:07:57 +01:00
save and load dynamic records
2013-12-07 13:17:28 +01:00
int ESMStore::countSavedGameRecords() const
{
Fixes #1234: Store dynamic record counter in savegame to prevent name clashes
2014-05-11 00:32:22 +02:00
return 1 // DYNA (dynamic name counter)
Store refactor: Now way easier to create new stores, a good chunk of it is automated, only simple macros are used now The case of indexedStores remains an issue, because they can't be stored with the rest because they don't inherit of store base
2022-07-03 18:11:20 +02:00
+ get<ESM::Potion>().getDynamicSize() + get<ESM::Armor>().getDynamicSize()
+ get<ESM::Book>().getDynamicSize() + get<ESM::Class>().getDynamicSize()
+ get<ESM::Clothing>().getDynamicSize() + get<ESM::Enchantment>().getDynamicSize()
+ get<ESM::NPC>().getDynamicSize() + get<ESM::Spell>().getDynamicSize()
+ get<ESM::Weapon>().getDynamicSize() + get<ESM::CreatureLevList>().getDynamicSize()
+ get<ESM::ItemLevList>().getDynamicSize() + get<ESM::Creature>().getDynamicSize()
+ get<ESM::Container>().getDynamicSize();
save and load dynamic records
2013-12-07 13:17:28 +01:00
}
Fixes #1172: Added basic loading/saving progress bar The progress is not particularly accurate. It simply uses the current / total number of records written/read as indication. Cell records are currently the largest by far, but there is a good chance that could be optimized using a change tracking system.
2014-04-28 11:29:57 +02:00
void ESMStore::write(ESM::ESMWriter& writer, Loading::Listener& progress) const
save and load dynamic records
2013-12-07 13:17:28 +01:00
{
Fixes #1234: Store dynamic record counter in savegame to prevent name clashes
2014-05-11 00:32:22 +02:00
writer.startRecord(ESM::REC_DYNA);
writer.startSubRecord("COUN");
writer.writeT(mDynamicCount);
writer.endRecord("COUN");
writer.endRecord(ESM::REC_DYNA);
Store refactor: Now way easier to create new stores, a good chunk of it is automated, only simple macros are used now The case of indexedStores remains an issue, because they can't be stored with the rest because they don't inherit of store base
2022-07-03 18:11:20 +02:00
get<ESM::Potion>().write(writer, progress);
get<ESM::Armor>().write(writer, progress);
get<ESM::Book>().write(writer, progress);
get<ESM::Class>().write(writer, progress);
get<ESM::Clothing>().write(writer, progress);
get<ESM::Enchantment>().write(writer, progress);
get<ESM::NPC>().write(writer, progress);
get<ESM::Spell>().write(writer, progress);
get<ESM::Weapon>().write(writer, progress);
get<ESM::CreatureLevList>().write(writer, progress);
get<ESM::ItemLevList>().write(writer, progress);
get<ESM::Creature>().write(writer, progress);
get<ESM::Container>().write(writer, progress);
save and load dynamic records
2013-12-07 13:17:28 +01:00
}
Fixes a bug with es3overrideRecord, esm3InsertStatic and esm3StoreInsert This also comes with a change to the mapping from esm3 record name to Store to be more direct with a pointer and the creation of a map from store pointer to esm3 record type
2022-07-20 19:04:31 +02:00
bool ESMStore::readRecord(ESM::ESMReader& reader, uint32_t type_id)
save and load dynamic records
2013-12-07 13:17:28 +01:00
{
Fixes a bug with es3overrideRecord, esm3InsertStatic and esm3StoreInsert This also comes with a change to the mapping from esm3 record name to Store to be more direct with a pointer and the creation of a map from store pointer to esm3 record type
2022-07-20 19:04:31 +02:00
ESM::RecNameInts type = (ESM::RecNameInts)type_id;
save and load dynamic records
2013-12-07 13:17:28 +01:00
switch (type)
{
case ESM::REC_ALCH:
case ESM::REC_ARMO:
case ESM::REC_BOOK:
case ESM::REC_CLAS:
case ESM::REC_CLOT:
case ESM::REC_ENCH:
case ESM::REC_SPEL:
case ESM::REC_WEAP:
Implement leveled list script functions (Fixes #1546)
2014-12-17 01:05:32 +01:00
case ESM::REC_LEVI:
case ESM::REC_LEVC:
Rename member to be clearer and more generic
2022-09-02 23:39:38 +02:00
mStoreImp->mRecNameToStore[type]->read(reader);
Disallow inserting containers, creatures, and npcs from the save game not present in content files
2021-03-22 22:29:10 +01:00
return true;
case ESM::REC_NPC_:
Mutate base records when editing AI settings (#2798)
2020-06-02 21:59:37 +02:00
case ESM::REC_CREA:
Save and load container records; fixes #5668
2020-10-29 13:55:24 +01:00
case ESM::REC_CONT:
Rename member to be clearer and more generic
2022-09-02 23:39:38 +02:00
mStoreImp->mRecNameToStore[type]->read(reader, true);
save and load dynamic records
2013-12-07 13:17:28 +01:00
return true;
Fixes #1234: Store dynamic record counter in savegame to prevent name clashes
2014-05-11 00:32:22 +02:00
case ESM::REC_DYNA:
reader.getSubNameIs("COUN");
reader.getHT(mDynamicCount);
return true;
save and load dynamic records
2013-12-07 13:17:28 +01:00
default:
return false;
}
}
Mutate base records when editing AI settings (#2798)
2020-06-02 21:59:37 +02:00
void ESMStore::checkPlayer()
{
setUp();
create a specific esm reader function for RefID to avoid allocation for string and then again for RefId Fixed some types removed useless header applied clang format fixed compile tests fixed clang tidy, and closer to logic before this MR Removed hardcoded refids unless there is a returned value we don't use static RefIds can use == between RefId and hardcoded string Fix clang format Fixed a few instances where std::string was used, when only const std::string& was needed removed unused variable
2022-10-18 09:26:55 +02:00
const ESM::NPC* player = get<ESM::NPC>().find(ESM::RefId::stringRefId("Player"));
Mutate base records when editing AI settings (#2798)
2020-06-02 21:59:37 +02:00
Store refactor: Now way easier to create new stores, a good chunk of it is automated, only simple macros are used now The case of indexedStores remains an issue, because they can't be stored with the rest because they don't inherit of store base
2022-07-03 18:11:20 +02:00
if (!get<ESM::Race>().find(player->mRace) || !get<ESM::Class>().find(player->mClass))
Mutate base records when editing AI settings (#2798)
2020-06-02 21:59:37 +02:00
throw std::runtime_error("Invalid player record (race or class unavailable");
}
Initial commit: In ESM structures, replace the string members that are RefIds to other records, to a new strong type The strong type is actually just a string underneath, but this will help in the future to have a distinction so it's easier to search and replace when we use an integer ID Slowly going through all the changes to make, still hundreds of errors a lot of functions/structures use std::string or stringview to designate an ID. So it takes time Continues slowly replacing ids. There are technically more and more compilation errors I have good hope that there is a point where the amount of errors will dramatically go down as all the main functions use the ESM::RefId type Continue moving forward, changes to the stores slowly moving along Starting to see the fruit of those changes. still many many error, but more and more Irun into a situation where a function is sandwiched between two functions that use the RefId type. More replacements. Things are starting to get easier I can see more and more often the issue is that the function is awaiting a RefId, but is given a string there is less need to go down functions and to fix a long list of them. Still moving forward, and for the first time error count is going down! Good pace, not sure about topics though, mId and mName are actually the same thing and are used interchangeably Cells are back to using string for the name, haven't fixed everything yet. Many other changes Under the bar of 400 compilation errors. more good progress <100 compile errors! More progress Game settings store can use string for find, it was a bit absurd how every use of it required to create refId from string some more progress on other fronts Mostly game settings clean one error opened a lot of other errors. Down to 18, but more will prbably appear only link errors left?? Fixed link errors OpenMW compiles, and launches, with some issues, but still!
2022-09-25 13:17:09 +02:00
std::pair<std::shared_ptr<MWMechanics::SpellList>, bool> ESMStore::getSpellList(const ESM::RefId& id) const
Mutate base records when adding/removing spells
2020-07-28 08:33:28 +02:00
{
auto result = mSpellListCache.find(id);
std::shared_ptr<MWMechanics::SpellList> ptr;
if (result != mSpellListCache.end())
ptr = result->second.lock();
if (!ptr)
{
int type = find(id);
ptr = std::make_shared<MWMechanics::SpellList>(id, type);
if (result != mSpellListCache.end())
result->second = ptr;
else
mSpellListCache.insert({ id, ptr });
return { ptr, false };
}
return { ptr, true };
}
ESMStore: no longer necessary to include the type definitions in the header For now there still needs to declare all the getters, but a macro may make that easier, or an different method entierly
2022-07-03 11:47:46 +02:00
template <>
const ESM::Cell* ESMStore::insert<ESM::Cell>(const ESM::Cell& cell)
{
Store refactor: Now way easier to create new stores, a good chunk of it is automated, only simple macros are used now The case of indexedStores remains an issue, because they can't be stored with the rest because they don't inherit of store base
2022-07-03 18:11:20 +02:00
return getWritable<ESM::Cell>().insert(cell);
ESMStore: no longer necessary to include the type definitions in the header For now there still needs to declare all the getters, but a macro may make that easier, or an different method entierly
2022-07-03 11:47:46 +02:00
}
template <>
const ESM::NPC* ESMStore::insert<ESM::NPC>(const ESM::NPC& npc)
{
Apply clang-format to code base
2022-09-22 21:26:05 +03:00
Fixed naming convention problems removed 2 macros. - One is replaced by the underlying code - The second one ise replaced by a template function removed tabulation used as indentation used getWritable instead of a const_cast used for( val : cont) loop removed useless getId function
2022-07-19 17:58:21 +02:00
auto& npcs = getWritable<ESM::NPC>();
create a specific esm reader function for RefID to avoid allocation for string and then again for RefId Fixed some types removed useless header applied clang format fixed compile tests fixed clang tidy, and closer to logic before this MR Removed hardcoded refids unless there is a returned value we don't use static RefIds can use == between RefId and hardcoded string Fix clang format Fixed a few instances where std::string was used, when only const std::string& was needed removed unused variable
2022-10-18 09:26:55 +02:00
if (npc.mId == "Player")
ESMStore: no longer necessary to include the type definitions in the header For now there still needs to declare all the getters, but a macro may make that easier, or an different method entierly
2022-07-03 11:47:46 +02:00
{
Fixed naming convention problems removed 2 macros. - One is replaced by the underlying code - The second one ise replaced by a template function removed tabulation used as indentation used getWritable instead of a const_cast used for( val : cont) loop removed useless getId function
2022-07-19 17:58:21 +02:00
return npcs.insert(npc);
ESMStore: no longer necessary to include the type definitions in the header For now there still needs to declare all the getters, but a macro may make that easier, or an different method entierly
2022-07-03 11:47:46 +02:00
}
Initial commit: In ESM structures, replace the string members that are RefIds to other records, to a new strong type The strong type is actually just a string underneath, but this will help in the future to have a distinction so it's easier to search and replace when we use an integer ID Slowly going through all the changes to make, still hundreds of errors a lot of functions/structures use std::string or stringview to designate an ID. So it takes time Continues slowly replacing ids. There are technically more and more compilation errors I have good hope that there is a point where the amount of errors will dramatically go down as all the main functions use the ESM::RefId type Continue moving forward, changes to the stores slowly moving along Starting to see the fruit of those changes. still many many error, but more and more Irun into a situation where a function is sandwiched between two functions that use the RefId type. More replacements. Things are starting to get easier I can see more and more often the issue is that the function is awaiting a RefId, but is given a string there is less need to go down functions and to fix a long list of them. Still moving forward, and for the first time error count is going down! Good pace, not sure about topics though, mId and mName are actually the same thing and are used interchangeably Cells are back to using string for the name, haven't fixed everything yet. Many other changes Under the bar of 400 compilation errors. more good progress <100 compile errors! More progress Game settings store can use string for find, it was a bit absurd how every use of it required to create refId from string some more progress on other fronts Mostly game settings clean one error opened a lot of other errors. Down to 18, but more will prbably appear only link errors left?? Fixed link errors OpenMW compiles, and launches, with some issues, but still!
2022-09-25 13:17:09 +02:00
const ESM::RefId id = ESM::RefId::stringRefId("$dynamic" + std::to_string(mDynamicCount++));
fixed made by @ZehMatt in 6abb96250fb750a71cf05e7684ac6abf0a22c75d and cdcf1393fcab8521644feede483316f3b3d352b0
2022-09-05 18:04:10 +02:00
if (npcs.search(id) != nullptr)
ESMStore: no longer necessary to include the type definitions in the header For now there still needs to declare all the getters, but a macro may make that easier, or an different method entierly
2022-07-03 11:47:46 +02:00
{
Initial commit: In ESM structures, replace the string members that are RefIds to other records, to a new strong type The strong type is actually just a string underneath, but this will help in the future to have a distinction so it's easier to search and replace when we use an integer ID Slowly going through all the changes to make, still hundreds of errors a lot of functions/structures use std::string or stringview to designate an ID. So it takes time Continues slowly replacing ids. There are technically more and more compilation errors I have good hope that there is a point where the amount of errors will dramatically go down as all the main functions use the ESM::RefId type Continue moving forward, changes to the stores slowly moving along Starting to see the fruit of those changes. still many many error, but more and more Irun into a situation where a function is sandwiched between two functions that use the RefId type. More replacements. Things are starting to get easier I can see more and more often the issue is that the function is awaiting a RefId, but is given a string there is less need to go down functions and to fix a long list of them. Still moving forward, and for the first time error count is going down! Good pace, not sure about topics though, mId and mName are actually the same thing and are used interchangeably Cells are back to using string for the name, haven't fixed everything yet. Many other changes Under the bar of 400 compilation errors. more good progress <100 compile errors! More progress Game settings store can use string for find, it was a bit absurd how every use of it required to create refId from string some more progress on other fronts Mostly game settings clean one error opened a lot of other errors. Down to 18, but more will prbably appear only link errors left?? Fixed link errors OpenMW compiles, and launches, with some issues, but still!
2022-09-25 13:17:09 +02:00
const std::string msg = "Try to override existing record '" + id.getRefIdString() + "'";
ESMStore: no longer necessary to include the type definitions in the header For now there still needs to declare all the getters, but a macro may make that easier, or an different method entierly
2022-07-03 11:47:46 +02:00
throw std::runtime_error(msg);
}
ESM::NPC record = npc;
record.mId = id;
Fixed naming convention problems removed 2 macros. - One is replaced by the underlying code - The second one ise replaced by a template function removed tabulation used as indentation used getWritable instead of a const_cast used for( val : cont) loop removed useless getId function
2022-07-19 17:58:21 +02:00
ESM::NPC* ptr = npcs.insert(record);
mids and mStaticIds moved to mStoreImp renamed recordid to storeid
2022-07-20 19:28:06 +02:00
mStoreImp->mIds[ptr->mId] = ESM::REC_NPC_;
ESMStore: no longer necessary to include the type definitions in the header For now there still needs to declare all the getters, but a macro may make that easier, or an different method entierly
2022-07-03 11:47:46 +02:00
return ptr;
}
move ESMStore to MWWorld
2012-10-01 19:17:04 +04:00
} // end namespace
Reference in New Issue Copy Permalink