// This file is part of the 64k demo project.
// It implements the Renderer3D class.

#include "3d/renderer.h"
#include <iostream>
#include <cstring>

// Simple Cube Geometry (Triangle list)
// 36 vertices
static const float kCubeVertices[] = {
    // Front face
    -1.0, -1.0,  1.0,
     1.0, -1.0,  1.0,
     1.0,  1.0,  1.0,
    -1.0, -1.0,  1.0,
     1.0,  1.0,  1.0,
    -1.0,  1.0,  1.0,
    // Back face
    -1.0, -1.0, -1.0,
    -1.0,  1.0, -1.0,
     1.0,  1.0, -1.0,
    -1.0, -1.0, -1.0,
     1.0,  1.0, -1.0,
     1.0, -1.0, -1.0,
    // Top face
    -1.0,  1.0, -1.0,
    -1.0,  1.0,  1.0,
     1.0,  1.0,  1.0,
    -1.0,  1.0, -1.0,
     1.0,  1.0,  1.0,
     1.0,  1.0, -1.0,
    // Bottom face
    -1.0, -1.0, -1.0,
     1.0, -1.0, -1.0,
     1.0, -1.0,  1.0,
    -1.0, -1.0, -1.0,
     1.0, -1.0,  1.0,
    -1.0, -1.0,  1.0,
    // Right face
     1.0, -1.0, -1.0,
     1.0,  1.0, -1.0,
     1.0,  1.0,  1.0,
     1.0, -1.0, -1.0,
     1.0,  1.0,  1.0,
     1.0, -1.0,  1.0,
    // Left face
    -1.0, -1.0, -1.0,
    -1.0, -1.0,  1.0,
    -1.0,  1.0,  1.0,
    -1.0, -1.0, -1.0,
    -1.0,  1.0,  1.0,
    -1.0,  1.0, -1.0,
};

static const char* kShaderCode = R"(
struct GlobalUniforms {
    view_proj: mat4x4<f32>,
    camera_pos: vec3<f32>,
    time: f32,
};

struct ObjectData {
    model: mat4x4<f32>,
    color: vec4<f32>,
    params: vec4<f32>,
};

struct ObjectsBuffer {
    objects: array<ObjectData>,
};

@group(0) @binding(0) var<uniform> globals: GlobalUniforms;
@group(0) @binding(1) var<storage, read> object_data: ObjectsBuffer;

struct VertexOutput {
    @builtin(position) position: vec4<f32>,
    @location(0) local_pos: vec3<f32>,
    @location(1) color: vec4<f32>,
};

@vertex
fn vs_main(@builtin(vertex_index) vertex_index: u32, 
           @builtin(instance_index) instance_index: u32) -> VertexOutput {
    
    // Hardcoded cube vertices (similar to C++ array but in shader for simplicity if desired, 
    // but here we might assume a vertex buffer or just generate logic. 
    // For this demo, let's use the buffer-less approach for vertices if we want to save space, 
    // but we have a C++ array. Let's just generate a cube on the fly from index?)
    // Actually, let's map the C++ kCubeVertices to a vertex buffer or use a hardcoded array here.
    // For 64k size, hardcoded in shader is good.
    
    var pos = array<vec3<f32>, 36>(
        vec3(-1.0, -1.0,  1.0), vec3( 1.0, -1.0,  1.0), vec3( 1.0,  1.0,  1.0),
        vec3(-1.0, -1.0,  1.0), vec3( 1.0,  1.0,  1.0), vec3(-1.0,  1.0,  1.0),
        vec3(-1.0, -1.0, -1.0), vec3(-1.0,  1.0, -1.0), vec3( 1.0,  1.0, -1.0),
        vec3(-1.0, -1.0, -1.0), vec3( 1.0,  1.0, -1.0), vec3( 1.0, -1.0, -1.0),
        vec3(-1.0,  1.0, -1.0), vec3(-1.0,  1.0,  1.0), vec3( 1.0,  1.0,  1.0),
        vec3(-1.0,  1.0, -1.0), vec3( 1.0,  1.0,  1.0), vec3( 1.0,  1.0, -1.0),
        vec3(-1.0, -1.0, -1.0), vec3( 1.0, -1.0, -1.0), vec3( 1.0, -1.0,  1.0),
        vec3(-1.0, -1.0, -1.0), vec3( 1.0, -1.0,  1.0), vec3(-1.0, -1.0,  1.0),
        vec3( 1.0, -1.0, -1.0), vec3( 1.0,  1.0, -1.0), vec3( 1.0,  1.0,  1.0),
        vec3( 1.0, -1.0, -1.0), vec3( 1.0,  1.0,  1.0), vec3( 1.0, -1.0,  1.0),
        vec3(-1.0, -1.0, -1.0), vec3(-1.0, -1.0,  1.0), vec3(-1.0,  1.0,  1.0),
        vec3(-1.0, -1.0, -1.0), vec3(-1.0,  1.0,  1.0), vec3(-1.0,  1.0, -1.0)
    );

    let p = pos[vertex_index];
    let obj = object_data.objects[instance_index];
    
    // Model -> World -> Clip
    let world_pos = obj.model * vec4<f32>(p, 1.0);
    let clip_pos = globals.view_proj * world_pos;

    var out: VertexOutput;
    out.position = clip_pos;
    out.local_pos = p;
    out.color = obj.color;
    return out;
}

@fragment
fn fs_main(in: VertexOutput) -> @location(0) vec4<f32> {
    // Simple wireframe-ish effect using barycentric coords logic? 
    // Or just check proximity to edge of local cube?
    let d = abs(in.local_pos);
    let edge_dist = max(max(d.x, d.y), d.z);
    
    // Mix object color with edge highlight
    var col = in.color.rgb;
    if (edge_dist > 0.95) {
        col = vec3<f32>(1.0, 1.0, 1.0); // White edges
    } else {
        // Simple shading
        let normal = normalize(cross(dpdx(in.local_pos), dpdy(in.local_pos)));
        let light = normalize(vec3<f32>(0.5, 1.0, 0.5));
        let diff = max(dot(normal, light), 0.2);
        col = col * diff;
    }
    
    return vec4<f32>(col, 1.0);
}
)";

void Renderer3D::init(WGPUDevice device, WGPUQueue queue, WGPUTextureFormat format) {
  device_ = device;
  queue_ = queue;
  format_ = format;

  create_default_resources();
  create_pipeline();
}

void Renderer3D::shutdown() {
  if (pipeline_) wgpuRenderPipelineRelease(pipeline_);
  if (bind_group_) wgpuBindGroupRelease(bind_group_);
  if (global_uniform_buffer_) wgpuBufferRelease(global_uniform_buffer_);
  if (object_storage_buffer_) wgpuBufferRelease(object_storage_buffer_);
  if (depth_view_) wgpuTextureViewRelease(depth_view_);
  if (depth_texture_) wgpuTextureRelease(depth_texture_);
}

void Renderer3D::resize(int width, int height) {
  if (width == width_ && height == height_) return;
  
  width_ = width;
  height_ = height;

  if (depth_view_) wgpuTextureViewRelease(depth_view_);
  if (depth_texture_) wgpuTextureRelease(depth_texture_);

  WGPUTextureDescriptor desc = {};
  desc.usage = WGPUTextureUsage_RenderAttachment;
  desc.dimension = WGPUTextureDimension_2D;
  desc.size = {(uint32_t)width, (uint32_t)height, 1};
  desc.format = WGPUTextureFormat_Depth24Plus; // Common depth format
  desc.mipLevelCount = 1;
  desc.sampleCount = 1;
  
  depth_texture_ = wgpuDeviceCreateTexture(device_, &desc);
  
  WGPUTextureViewDescriptor view_desc = {};
  view_desc.format = WGPUTextureFormat_Depth24Plus;
  view_desc.dimension = WGPUTextureViewDimension_2D;
  view_desc.aspect = WGPUTextureAspect_DepthOnly;
  view_desc.arrayLayerCount = 1;
  view_desc.mipLevelCount = 1;
  
  depth_view_ = wgpuTextureCreateView(depth_texture_, &view_desc);
}

void Renderer3D::create_default_resources() {
  // Uniform Buffer
  global_uniform_buffer_ = gpu_create_buffer(device_, sizeof(GlobalUniforms),
      WGPUBufferUsage_Uniform | WGPUBufferUsage_CopyDst, nullptr).buffer;

  // Storage Buffer
  size_t storage_size = sizeof(ObjectData) * kMaxObjects;
  object_storage_buffer_ = gpu_create_buffer(device_, storage_size,
      WGPUBufferUsage_Storage | WGPUBufferUsage_CopyDst, nullptr).buffer;
}

void Renderer3D::create_pipeline() {
  // Bind Group Layout
  WGPUBindGroupLayoutEntry entries[2] = {};
  
  // Binding 0: Globals (Uniform)
  entries[0].binding = 0;
  entries[0].visibility = WGPUShaderStage_Vertex | WGPUShaderStage_Fragment;
  entries[0].buffer.type = WGPUBufferBindingType_Uniform;
  entries[0].buffer.minBindingSize = sizeof(GlobalUniforms);

  // Binding 1: Object Data (Storage)
  entries[1].binding = 1;
  entries[1].visibility = WGPUShaderStage_Vertex | WGPUShaderStage_Fragment;
  entries[1].buffer.type = WGPUBufferBindingType_ReadOnlyStorage;
  entries[1].buffer.minBindingSize = sizeof(ObjectData) * kMaxObjects;

  WGPUBindGroupLayoutDescriptor bgl_desc = {};
  bgl_desc.entryCount = 2;
  bgl_desc.entries = entries;
  WGPUBindGroupLayout bgl = wgpuDeviceCreateBindGroupLayout(device_, &bgl_desc);

  // Bind Group
  WGPUBindGroupEntry bg_entries[2] = {};
  bg_entries[0].binding = 0;
  bg_entries[0].buffer = global_uniform_buffer_;
  bg_entries[0].size = sizeof(GlobalUniforms);
  
  bg_entries[1].binding = 1;
  bg_entries[1].buffer = object_storage_buffer_;
  bg_entries[1].size = sizeof(ObjectData) * kMaxObjects;

  WGPUBindGroupDescriptor bg_desc = {};
  bg_desc.layout = bgl;
  bg_desc.entryCount = 2;
  bg_desc.entries = bg_entries;
  bind_group_ = wgpuDeviceCreateBindGroup(device_, &bg_desc);

  // Pipeline Layout
  WGPUPipelineLayoutDescriptor pl_desc = {};
  pl_desc.bindGroupLayoutCount = 1;
  pl_desc.bindGroupLayouts = &bgl;
  WGPUPipelineLayout pipeline_layout = wgpuDeviceCreatePipelineLayout(device_, &pl_desc);

  // Shader Code
  const char* shader_source = kShaderCode;

  // Shader Module
#if defined(DEMO_CROSS_COMPILE_WIN32)
  WGPUShaderModuleWGSLDescriptor wgsl_desc = {};
  wgsl_desc.chain.sType = WGPUSType_ShaderModuleWGSLDescriptor;
  wgsl_desc.code = shader_source;
  
  WGPUShaderModuleDescriptor shader_desc = {};
  shader_desc.nextInChain = (const WGPUChainedStruct*)&wgsl_desc.chain;
#else
  WGPUShaderSourceWGSL wgsl_desc = {};
  wgsl_desc.chain.sType = WGPUSType_ShaderSourceWGSL;
  wgsl_desc.code = {shader_source, strlen(shader_source)};
  
  WGPUShaderModuleDescriptor shader_desc = {};
  shader_desc.nextInChain = (const WGPUChainedStruct*)&wgsl_desc.chain;
#endif

  WGPUShaderModule shader_module = wgpuDeviceCreateShaderModule(device_, &shader_desc);

  // Depth Stencil State
  WGPUDepthStencilState depth_stencil = {};
  depth_stencil.format = WGPUTextureFormat_Depth24Plus;
  depth_stencil.depthWriteEnabled = WGPUOptionalBool_True;
  depth_stencil.depthCompare = WGPUCompareFunction_Less;
  
  // Render Pipeline
  WGPURenderPipelineDescriptor desc = {};
  desc.layout = pipeline_layout;
  
  // Vertex
  desc.vertex.module = shader_module;
#if defined(DEMO_CROSS_COMPILE_WIN32)
  desc.vertex.entryPoint = "vs_main";
#else
  desc.vertex.entryPoint = {"vs_main", 7};
#endif

  // Fragment
  WGPUColorTargetState color_target = {};
  color_target.format = format_;
  color_target.writeMask = WGPUColorWriteMask_All;
  
  WGPUFragmentState fragment = {};
  fragment.module = shader_module;
#if defined(DEMO_CROSS_COMPILE_WIN32)
  fragment.entryPoint = "fs_main";
#else
  fragment.entryPoint = {"fs_main", 7};
#endif
  fragment.targetCount = 1;
  fragment.targets = &color_target;
  desc.fragment = &fragment;

  desc.primitive.topology = WGPUPrimitiveTopology_TriangleList;
  desc.primitive.cullMode = WGPUCullMode_Back;
  desc.primitive.frontFace = WGPUFrontFace_CCW;
  
  desc.depthStencil = &depth_stencil;
  desc.multisample.count = 1;
  desc.multisample.mask = 0xFFFFFFFF;

  pipeline_ = wgpuDeviceCreateRenderPipeline(device_, &desc);
  
  wgpuBindGroupLayoutRelease(bgl);
  wgpuPipelineLayoutRelease(pipeline_layout);
  wgpuShaderModuleRelease(shader_module);
}

void Renderer3D::update_uniforms(const Scene& scene, const Camera& camera, float time) {
  // Update Globals
  GlobalUniforms globals;
  globals.view_proj = camera.get_projection_matrix() * camera.get_view_matrix();
  globals.camera_pos = camera.position;
  globals.time = time;
  wgpuQueueWriteBuffer(queue_, global_uniform_buffer_, 0, &globals, sizeof(GlobalUniforms));

  // Update Objects
  std::vector<ObjectData> obj_data;
  obj_data.reserve(scene.objects.size());
  for (const auto& obj : scene.objects) {
    ObjectData data;
    data.model = obj.get_model_matrix();
    data.color = obj.color;
    // data.params = ...
    obj_data.push_back(data);
    if (obj_data.size() >= kMaxObjects) break;
  }
  
  if (!obj_data.empty()) {
    wgpuQueueWriteBuffer(queue_, object_storage_buffer_, 0, obj_data.data(), obj_data.size() * sizeof(ObjectData));
  }
}

void Renderer3D::render(const Scene& scene, const Camera& camera, float time,
                        WGPUTextureView target_view, WGPUTextureView depth_view_opt) {
  update_uniforms(scene, camera, time);

  WGPUTextureView depth_view = depth_view_opt ? depth_view_opt : depth_view_;
  if (!depth_view) return; // Should have been created by resize

  WGPURenderPassColorAttachment color_attachment = {};
  gpu_init_color_attachment(color_attachment, target_view);
  color_attachment.clearValue = {0.05, 0.05, 0.1, 1.0}; // Dark blue-ish background

  WGPURenderPassDepthStencilAttachment depth_attachment = {};
  depth_attachment.view = depth_view;
  depth_attachment.depthLoadOp = WGPULoadOp_Clear;
  depth_attachment.depthStoreOp = WGPUStoreOp_Store;
  depth_attachment.depthClearValue = 1.0f;

  WGPURenderPassDescriptor pass_desc = {};
  pass_desc.colorAttachmentCount = 1;
  pass_desc.colorAttachments = &color_attachment;
  pass_desc.depthStencilAttachment = &depth_attachment;

  WGPUCommandEncoder encoder = wgpuDeviceCreateCommandEncoder(device_, nullptr);
  WGPURenderPassEncoder pass = wgpuCommandEncoderBeginRenderPass(encoder, &pass_desc);

  wgpuRenderPassEncoderSetPipeline(pass, pipeline_);
  wgpuRenderPassEncoderSetBindGroup(pass, 0, bind_group_, 0, nullptr);
  
  // Draw all objects (Instance Count = object count)
  // Vertex Count = 36 (Cube)
  uint32_t instance_count = (uint32_t)std::min((size_t)kMaxObjects, scene.objects.size());
  if (instance_count > 0) {
    wgpuRenderPassEncoderDraw(pass, 36, instance_count, 0, 0);
  }

  wgpuRenderPassEncoderEnd(pass);
  WGPUCommandBuffer commands = wgpuCommandEncoderFinish(encoder, nullptr);
  wgpuQueueSubmit(queue_, 1, &commands);

  wgpuRenderPassEncoderRelease(pass);
  wgpuCommandBufferRelease(commands);
  wgpuCommandEncoderRelease(encoder);
}