bsnes/ruby/input/joypad/directinput.cpp

218 lines
7.9 KiB
C++
Raw Normal View History

#ifndef RUBY_INPUT_JOYPAD_DIRECTINPUT
#define RUBY_INPUT_JOYPAD_DIRECTINPUT
auto CALLBACK DirectInput_EnumJoypadsCallback(const DIDEVICEINSTANCE* instance, void* p) -> BOOL;
auto CALLBACK DirectInput_EnumJoypadAxesCallback(const DIDEVICEOBJECTINSTANCE* instance, void* p) -> BOOL;
auto CALLBACK DirectInput_EnumJoypadEffectsCallback(const DIDEVICEOBJECTINSTANCE* instance, void* p) -> BOOL;
struct InputJoypadDirectInput {
Input& input;
InputJoypadDirectInput(Input& input) : input(input) {}
struct Joypad {
shared_pointer<HID::Joypad> hid{new HID::Joypad};
LPDIRECTINPUTDEVICE8 device = nullptr;
LPDIRECTINPUTEFFECT effect = nullptr;
uint32_t pathID = 0;
uint16_t vendorID = 0;
uint16_t productID = 0;
bool isXInputDevice = false;
};
vector<Joypad> joypads;
uintptr_t handle = 0;
LPDIRECTINPUT8 context = nullptr;
LPDIRECTINPUTDEVICE8 device = nullptr;
bool xinputAvailable = false;
unsigned effects = 0;
auto assign(shared_pointer<HID::Joypad> hid, unsigned groupID, unsigned inputID, int16_t value) -> void {
auto& group = hid->group(groupID);
if(group.input(inputID).value() == value) return;
input.doChange(hid, groupID, inputID, group.input(inputID).value(), value);
group.input(inputID).setValue(value);
}
auto poll(vector<shared_pointer<HID::Device>>& devices) -> void {
for(auto& jp : joypads) {
if(FAILED(jp.device->Poll())) jp.device->Acquire();
DIJOYSTATE2 state;
if(FAILED(jp.device->GetDeviceState(sizeof(DIJOYSTATE2), &state))) continue;
for(unsigned n = 0; n < 4; n++) {
assign(jp.hid, HID::Joypad::GroupID::Axis, 0, state.lX);
assign(jp.hid, HID::Joypad::GroupID::Axis, 1, state.lY);
assign(jp.hid, HID::Joypad::GroupID::Axis, 2, state.lZ);
assign(jp.hid, HID::Joypad::GroupID::Axis, 3, state.lRx);
assign(jp.hid, HID::Joypad::GroupID::Axis, 4, state.lRy);
assign(jp.hid, HID::Joypad::GroupID::Axis, 5, state.lRz);
unsigned pov = state.rgdwPOV[n];
int16_t xaxis = 0;
int16_t yaxis = 0;
if(pov < 36000) {
if(pov >= 31500 || pov <= 4500) yaxis = -32768;
if(pov >= 4500 && pov <= 13500) xaxis = +32767;
if(pov >= 13500 && pov <= 22500) yaxis = +32767;
if(pov >= 22500 && pov <= 31500) xaxis = -32768;
}
assign(jp.hid, HID::Joypad::GroupID::Hat, n * 2 + 0, xaxis);
assign(jp.hid, HID::Joypad::GroupID::Hat, n * 2 + 1, yaxis);
}
for(unsigned n = 0; n < 128; n++) {
assign(jp.hid, HID::Joypad::GroupID::Button, n, (bool)state.rgbButtons[n]);
}
devices.append(jp.hid);
}
}
auto rumble(uint64_t id, bool enable) -> bool {
for(auto& jp : joypads) {
if(jp.hid->id() != id) continue;
if(jp.effect == nullptr) continue;
if(enable) jp.effect->Start(1, 0);
else jp.effect->Stop();
return true;
}
return false;
}
auto init(uintptr_t handle, LPDIRECTINPUT8 context, bool xinputAvailable) -> bool {
this->handle = handle;
this->context = context;
this->xinputAvailable = xinputAvailable;
context->EnumDevices(DI8DEVCLASS_GAMECTRL, DirectInput_EnumJoypadsCallback, (void*)this, DIEDFL_ATTACHEDONLY);
return true;
}
auto term() -> void {
for(auto& jp : joypads) {
jp.device->Unacquire();
if(jp.effect) jp.effect->Release();
jp.device->Release();
}
joypads.reset();
context = nullptr;
}
auto initJoypad(const DIDEVICEINSTANCE* instance) -> bool {
Joypad jp;
jp.vendorID = instance->guidProduct.Data1 >> 0;
jp.productID = instance->guidProduct.Data1 >> 16;
jp.isXInputDevice = false;
if(auto device = rawinput.find(jp.vendorID, jp.productID)) {
jp.isXInputDevice = device().isXInputDevice;
}
//Microsoft has intentionally imposed artificial restrictions on XInput devices when used with DirectInput
//a) the two triggers are merged into a single axis, making uniquely distinguishing them impossible
//b) rumble support is not exposed
//thus, it's always preferred to let the XInput driver handle these joypads
//but if the driver is not available (XInput 1.3 does not ship with stock Windows XP), fall back on DirectInput
if(jp.isXInputDevice && xinputAvailable) return DIENUM_CONTINUE;
if(FAILED(context->CreateDevice(instance->guidInstance, &device, 0))) return DIENUM_CONTINUE;
jp.device = device;
device->SetDataFormat(&c_dfDIJoystick2);
device->SetCooperativeLevel((HWND)handle, DISCL_NONEXCLUSIVE | DISCL_BACKGROUND);
effects = 0;
device->EnumObjects(DirectInput_EnumJoypadAxesCallback, (void*)this, DIDFT_ABSAXIS);
device->EnumObjects(DirectInput_EnumJoypadEffectsCallback, (void*)this, DIDFT_FFACTUATOR);
jp.hid->setRumble(effects > 0);
DIPROPGUIDANDPATH property;
memset(&property, 0, sizeof(DIPROPGUIDANDPATH));
property.diph.dwSize = sizeof(DIPROPGUIDANDPATH);
property.diph.dwHeaderSize = sizeof(DIPROPHEADER);
property.diph.dwObj = 0;
property.diph.dwHow = DIPH_DEVICE;
device->GetProperty(DIPROP_GUIDANDPATH, &property.diph);
string devicePath = (const char*)utf8_t(property.wszPath);
Update to v094r12 release. byuu says: Changelog: * added driver selection * added video scale + aspect correction settings * added A/V sync + audio mute settings * added configuration file * fixed compilation bugs under Windows and Linux * fixed window sizing * removed HSU1 * the system menu stays as "System", because "Game Boy Advance" was too long a string for the smallest scale size * some more stuff You guys probably won't be ecstatic about the video sizing options, but it's basically your choice of 1x, 2x or 4x scale with optional aspect correction. 3x was intentionally skipped because it looks horrible on hires SNES games. The window is resized and recentered upon loading games. The window doesn't resize otherwise. I never really liked the way v094 always left you with black screen areas and left you with off-centered window positions. I might go ahead and add the pseudo-fullscreen toggle that will jump into 4x mode (respecting your aspect setting.) Short-term: * add input port changing support * add other input types (mouse-based, etc) * add save states * add cheat codes * add timing configuration (video/audio sync) * add hotkeys (single state) We can probably do a new release once the short-term items are completed. Long-term: * add slotted cart loader (SGB, BSX, ST) * add DIP switch selection window (NSS) * add cheat code database * add state manager * add overscan masking Not planned: * video color adjustments (will allow emulated color vs raw color; but no more sliders) * pixel shaders * ananke integration (will need to make a command-line version to get my games in) * fancy audio adjustment controls (resampler, latency, volume) * input focus settings * relocating game library (not hard, just don't feel like it) * localization support (not enough users) * window geometry memory * anything else not in higan v094
2015-03-03 10:14:49 +00:00
jp.pathID = Hash::CRC32(devicePath.data(), devicePath.size()).value();
jp.hid->setID((uint64_t)jp.pathID << 32 | jp.vendorID << 16 | jp.productID << 0);
if(jp.hid->rumble()) {
//disable auto-centering spring for rumble support
DIPROPDWORD property;
memset(&property, 0, sizeof(DIPROPDWORD));
property.diph.dwSize = sizeof(DIPROPDWORD);
property.diph.dwHeaderSize = sizeof(DIPROPHEADER);
property.diph.dwObj = 0;
property.diph.dwHow = DIPH_DEVICE;
property.dwData = false;
device->SetProperty(DIPROP_AUTOCENTER, &property.diph);
Update to higan and icarus v095r15 release. r13 and r14 weren't posted as individual releases, but their changelogs were posted. byuu says about r13: I'm not going to be posting WIPs for r13 and above for a while. The reason is that I'm working on the major manifest overhaul I've discussed previously on the icarus subforum. I'm recreating my boards database from scratch using the map files and the new map analyzer. The only games that will load are ones I've created board definitions for, and updated sfc/cartridge/markup.cpp to parse. Once I've finished all the boards, then I'll update the heuristics. Then finally, I'll sync the syntax changes over to the fc, gb, gba cores. Once that's done, I'll start posting WIPs again, along with a new build of icarus. But I'll still post changelogs as I work through things. Changelog (r13): - preservation: created new database-builder tool (merges region-specific databases with boards) - icarus: support new, external database format (~/.config/icarus/Database/(Super Famicom.bml, ...) - added 1A3B-(10,11,12); 1A3B-20 byuu says about r14: r14 work: I successfully created mappings for every board used in the US set. I also updated icarus' heuristics to use the new mappings, and created ones there for the boards that are only in the JP set. Then I patched icarus to support pulling games out of the database when it's used on a game folder to generate a manifest file. Then I updated a lot of code in higan/sfc to support the new mapping syntax. sfc/cartridge/markup.cpp is about half the size it used to be with the new mappings, and I was able to kill off both map/id and map/select entirely. Then I updated all four emulated systems (and both subsystems) to use "board" as the root node, and harmonized their syntax (made them all more consistent with each other.) Then I added a manifest viewer to the tools window+menu. It's kind of an advanced user feature, but oh well. No reason to coddle people when the feature is very useful for developers. The viewer will show all manifests in order when you load multi-cart games as well. Still not going to call any syntax 100% done right now, but thankfully with the new manifest-free folders, nobody will have to do anything to use the new format. Just download the new version and go. The Super Famicom Event stuff is currently broken (CC92/PF94 boards). That's gonna be fun to support. byuu says about r15: EDIT: small bug in icarus with heuristics. Edit core/super-famicom.cpp line 27: if(/*auto*/ markup = cartridge.markup) { Gotta remove that "auto" so that it returns valid markup. Resolved the final concerns I had with the new manifest format. Right now there are two things that are definitely broken: MCC (BS-X Town cart) and Event (CC '92 and PF'94). And there are a few things that are untested: SPC7110, EpsonRTC, SharpRTC, SDD1+RAM, SufamiTurbo, BS-X slotted carts.
2015-12-19 08:52:34 +00:00
DWORD dwAxes[2] = {(DWORD)DIJOFS_X, (DWORD)DIJOFS_Y};
LONG lDirection[2] = {0, 0};
DICONSTANTFORCE force;
force.lMagnitude = DI_FFNOMINALMAX; //full force
DIEFFECT effect;
memset(&effect, 0, sizeof(DIEFFECT));
effect.dwSize = sizeof(DIEFFECT);
effect.dwFlags = DIEFF_CARTESIAN | DIEFF_OBJECTOFFSETS;
effect.dwDuration = INFINITE;
effect.dwSamplePeriod = 0;
effect.dwGain = DI_FFNOMINALMAX;
effect.dwTriggerButton = DIEB_NOTRIGGER;
effect.dwTriggerRepeatInterval = 0;
effect.cAxes = 2;
effect.rgdwAxes = dwAxes;
effect.rglDirection = lDirection;
effect.lpEnvelope = 0;
effect.cbTypeSpecificParams = sizeof(DICONSTANTFORCE);
effect.lpvTypeSpecificParams = &force;
effect.dwStartDelay = 0;
device->CreateEffect(GUID_ConstantForce, &effect, &jp.effect, NULL);
}
for(unsigned n = 0; n < 6; n++) jp.hid->axes().append(n);
for(unsigned n = 0; n < 8; n++) jp.hid->hats().append(n);
for(unsigned n = 0; n < 128; n++) jp.hid->buttons().append(n);
joypads.append(jp);
return DIENUM_CONTINUE;
}
auto initAxis(const DIDEVICEOBJECTINSTANCE* instance) -> bool {
DIPROPRANGE range;
memset(&range, 0, sizeof(DIPROPRANGE));
range.diph.dwSize = sizeof(DIPROPRANGE);
range.diph.dwHeaderSize = sizeof(DIPROPHEADER);
range.diph.dwHow = DIPH_BYID;
range.diph.dwObj = instance->dwType;
range.lMin = -32768;
range.lMax = +32767;
device->SetProperty(DIPROP_RANGE, &range.diph);
return DIENUM_CONTINUE;
}
auto initEffect(const DIDEVICEOBJECTINSTANCE* instance) -> bool {
effects++;
return DIENUM_CONTINUE;
}
};
auto CALLBACK DirectInput_EnumJoypadsCallback(const DIDEVICEINSTANCE* instance, void* p) -> BOOL {
return ((InputJoypadDirectInput*)p)->initJoypad(instance);
}
auto CALLBACK DirectInput_EnumJoypadAxesCallback(const DIDEVICEOBJECTINSTANCE* instance, void* p) -> BOOL {
return ((InputJoypadDirectInput*)p)->initAxis(instance);
}
auto CALLBACK DirectInput_EnumJoypadEffectsCallback(const DIDEVICEOBJECTINSTANCE* instance, void* p) -> BOOL {
return ((InputJoypadDirectInput*)p)->initEffect(instance);
}
#endif