feat(render): Campaign V slice V4a - port TextRenderer/BitmapFont/DebugLineRenderer/TextureCache onto IGpuDevice

TextRenderer, BitmapFont, DebugLineRenderer, and TextureCache's UI-texture
upload path (GetOrUploadRenderSurface/UploadRgba8) now issue every draw and
resource creation through the pinned IGpuDevice/IGpuFrame/IGpuPassEncoder
RHI contract instead of raw GL. This is the RHI's first real production
consumer - V0-V3 only established the contract, GL backend skeleton, and a
shader-dialect migration with no live GL exercise. TextRenderer owns one
IGpuPipeline (ui_text shader, straight-alpha blend, depth disabled) and
allocates a per-bucket ring each Flush; BitmapFont's atlas texture is
created and uploaded via device.CreateTexture/.Upload; DebugLineRenderer
mirrors the same one-pipeline-per-Flush shape for its line-list draws.
World-path TextureCache methods (GetOrUpload, the raw-GL layer-array
upload) are untouched - still legacy GL, still out of scope.

Frame lifecycle: GpuDeviceFrameLifetime (RenderFrameOrchestrator.cs) wraps
IGpuDevice.BeginFrame()/IGpuFrame.End() inside the existing
IRenderFrameLifetime bracket HostInputCameraCompositionPhase already opens
per callback, additively - no frame-graph restructuring. Ported renderers
reach the frame via ICurrentGpuFrameSource, a plain interface (not a
delegate field) so WorldSceneDiagnosticsController keeps passing its
existing "no stored window/delegate" architectural-conformance test.

Two real bugs surfaced by actually exercising the RHI against a live GL
context (nothing here was previously reachable before this slice):

- GlGpuDevice.BeginFrame() now resets the render-state cache every frame.
  The cache assumes it is the sole writer of GL program/blend/depth/cull
  state, which was true while it had zero real consumers, but every
  still-legacy renderer (WbDrawDispatcher, terrain, particles, EnvCells)
  mutates that same GL state directly and never informs the cache. Once a
  legacy renderer ran between two RHI binds, the cache's belief about the
  current GL program went stale, so a later BindPipeline(text shader)
  skipped re-issuing glUseProgram and the following push-constant upload
  threw GL_INVALID_OPERATION against whatever program was actually bound.
  Reset() at the frame boundary is the same defensive move BeginPass
  already makes after a forced clear (see its comment); it costs one
  redundant state application on the frame's first bind.
- GL_MULTISAMPLE has no representation in the pinned contract. Added a
  GL-backend-internal Multisample field to GlRenderStateSnapshot/Changes,
  computed from GpuPipelineDescription.SampleCount at BindPipeline time -
  mirrors how Vulkan bakes MSAA into the pipeline instead of a separate
  toggle.

Collateral, scoped to keep the port real rather than a stub:

- GpuTextureSlot (Unassigned = uint.MaxValue, NOT 0) now flows through
  every consumer of TextureCache.GetOrUploadRenderSurface/UploadRgba8 and
  TextRenderer.DrawSprite - the entire retained UI layer, since a pervasive
  Func<uint,(uint,int,int)> sprite-resolve delegate threads through nearly
  every UI element/controller. Every prior `== 0` / `!= 0` "no texture"
  check became `.IsAssigned` / `!.IsAssigned`; slot 0 is a real assigned
  slot (the device's default white texture), so the old sentinel would
  have produced live visual regressions if left in place.
- GpuTextureSlot/IGpuDevice/IGpuFrame are internal, so ~270 previously
  public AcDream.App types that touched them (directly or transitively)
  are now internal too - safe, since AcDream.App is an exe with no
  external project references; only the two test projects consume it, via
  InternalsVisibleTo. A handful of unrelated types the sweep caught
  (ElementInfo/ImportedLayout's property-bag hierarchy, several enums used
  as public [Theory] parameters, CursorFeedbackSnapshot's DragAcceptState)
  were reverted back to public where making them internal would have
  either cascaded into unrelated files or broken xUnit's public-member
  discovery.
- ExternalViewportTextureBridge (new) registers the still-raw-GL FBO
  color textures PrivateEntityViewportRenderer/PaperdollViewportRenderer
  produce (V4g's scope) into the device's texture table for
  UiViewport.TextureHandle, via a temporary
  GlGpuDevice.RegisterExternalColorTexture escape hatch (internal, not
  part of IGpuDevice) deleted when V4g ports those viewports.
- TextRenderGlStateScope.cs and its test deleted: the pipeline description
  now bakes what it used to restore by hand.
- ResourceCleanupGroupTests/GlTextureOwnershipTests: the two source-text
  conformance tests keyed to TextRenderer's old multi-resource
  construction shape (Shader + per-flight FrameBufferSet array + white
  texture + tracked VAO/VBO, all via ResourceCleanupGroup) no longer apply
  - that shape is gone, replaced by one IGpuPipeline created through
    IGpuDevice. The construction-order test is deleted; the checked-commit
    texture-creation check now targets GlGpuTexture (which already used
    the same GlResourceCommand.CreateName primitive before this slice).

Gates:
- dotnet build -c Release: 0 warnings, 0 errors (AcDream.App has
  TreatWarningsAsErrors).
- dotnet test tests/AcDream.App.Tests -c Release: 3,840 passed / 3
  skipped (was 3,843/3 entering this slice - net 3 fewer tests:
  TextRendererFailureSafetyTests.cs deleted (2, tested the now-deleted
  TextRenderGlStateScope) plus the one retired ResourceCleanupGroupTests
  method). Full solution: 8,908 passed / 5 skipped across all nine test
  projects.
- Offline pixel gate (tools/run-offline-pixel-gate.ps1, parent ec414d60
  vs this commit): differing fraction 0.318% (1,791/563,200 compared
  pixels), above the 0.001 threshold. Investigated pixel-by-pixel rather
  than waved through: a diff heatmap plus 4x crops at the differing
  clusters show zero differences anywhere in the retained UI, terrain,
  scenery, or static meshes - every differing pixel sits on continuously-
  animated ambient content (flying-insect sprites over the swamp, foliage
  sparkle/dew glints) whose exact phase depends on elapsed wall-clock
  time, the same category the gate's own sky-masking rationale already
  documents and the campaign doc's coverage table explicitly excludes
  ("Not covered - particles"). Confirming evidence: two same-commit
  captures at HEAD compare clean against each other (0.0025%), and two
  same-commit captures at the parent compare clean against each other
  (0.0044%) - only base-vs-head is consistently elevated, which is what
  frame-pacing drift from genuinely new per-frame RHI work (BeginFrame,
  ring resets, the render-state reset above) would produce against a
  fixed wall-clock capture deadline, not a rendering defect. Recommend a
  quick user visual check of this capture pair alongside the automated
  result, matching how V2c's particle work was already handled in this
  campaign (flagged for user visual confirmation rather than blocked on
  an automated gate that cannot cover animated content).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
Erik 2026-07-27 18:22:08 +02:00
parent ec414d60cd
commit ceec3bc440
334 changed files with 3660 additions and 3840 deletions

View file

@ -1,4 +1,4 @@
using System;
using System;
using System.Linq;
using AcDream.App.UI;
@ -18,7 +18,7 @@ namespace AcDream.App.UI.Layout;
///
/// <para>
/// The meter's back/front 3-slice sprite ids live on grandchild image elements,
/// NOT on the meter element itself (format doc §11). <see cref="BuildMeter"/>
/// NOT on the meter element itself (format doc §11). <see cref="BuildMeter"/>
/// walks two layers down to extract them: the two Type-3 container children
/// ordered by <see cref="ElementInfo.ReadOrder"/> (back behind = lower, front
/// on top = higher), then within each container the image children that carry
@ -28,44 +28,44 @@ namespace AcDream.App.UI.Layout;
///
/// <para>
/// The expand-detail overlay present in the front container carries ONLY named
/// states ("HideDetail"/"ShowDetail") — no "" DirectState entry — so the
/// states ("HideDetail"/"ShowDetail") — no "" DirectState entry — so the
/// <c>TryGetValue("")</c> filter in <see cref="SliceIds"/> excludes it
/// automatically.
/// </para>
/// </summary>
public static class DatWidgetFactory
internal static class DatWidgetFactory
{
/// <summary>
/// Creates the <see cref="UiElement"/> for <paramref name="info"/>, sets its
/// rect (Left/Top/Width/Height) and Anchors, and returns it.
/// </summary>
/// <param name="info">Resolved, merged element snapshot from the LayoutDesc importer.</param>
/// <param name="resolve">RenderSurface id (GL tex handle, pixel width, pixel height).
/// <param name="resolve">RenderSurface id → (GL tex handle, pixel width, pixel height).
/// Returns (0,0,0) when the texture is not yet uploaded.</param>
/// <param name="datFont">Retail UI font for the meter's "cur/max" number overlay.
/// May be null pre-load the meter falls back to the debug bitmap font.</param>
/// <param name="fontResolve">Optional font resolver: FontDid <see cref="UiDatFont"/>
/// May be null pre-load — the meter falls back to the debug bitmap font.</param>
/// <param name="fontResolve">Optional font resolver: FontDid → <see cref="UiDatFont"/>
/// (or null when the font can't be loaded). When non-null, any element whose
/// <see cref="ElementInfo.FontDid"/> is non-zero gets ITS OWN dat font applied instead of
/// the shared <paramref name="datFont"/> fallback. Null = original behavior (use
/// <paramref name="datFont"/> for every element).</param>
/// <returns>The widget for this element. Never null every type produces a widget.</returns>
/// <returns>The widget for this element. Never null — every type produces a widget.</returns>
public static UiElement? Create(ElementInfo info,
Func<uint, (uint, int, int)> resolve, UiDatFont? datFont,
Func<uint, (GpuTextureSlot, int, int)> resolve, UiDatFont? datFont,
Func<uint, UiDatFont?>? fontResolve = null,
Func<UiStringInfoValue, string?>? stringResolve = null)
{
// Retail Type 3 = UIElement_Field (reg :126190), but in acdream's CURRENT layouts
// (vitals 0x2100006C / chat 0x21000006) Type-3 elements are sprite-bearing chrome +
// containers (the 8-piece bevel corners/edges, the transcript/input panels), NOT
// editable fields retail draws those as inert media-bearing Fields, which our
// editable fields — retail draws those as inert media-bearing Fields, which our
// UiDatElement reproduces pixel-for-pixel (and without the spurious focus/edit
// affordance a UiField would add). The one true editable field, the chat input
// (0x10000016), resolves to Type 12 and is controller-placed as a UiField. So Type 3
// stays on the generic fallback here; register it as UiField only when a window
// actually carries a factory-built editable Type-3 field (and UiField grows a
// background-media draw + an opt-in editable flag at that point). UiField (the widget)
// still ships it just isn't wired into the factory switch yet.
// still ships — it just isn't wired into the factory switch yet.
// Resolve this element's own dat font if a resolver is provided and the element
// has a FontDid. Falls back to the shared datFont when not set (FontDid==0) or
// when the resolver returns null (font missing from dats).
@ -86,11 +86,11 @@ public static class DatWidgetFactory
// gmUIElement_*Indicator custom button classes
6 => new UiMenu(), // UIElement_Menu (reg :120163)
7 => BuildMeter(info, resolve, elementFont), // UIElement_Meter
0xD => new UiViewport(), // UIElement_Viewport 3-D mini-scene blit leaf
0xD => new UiViewport(), // UIElement_Viewport — 3-D mini-scene blit leaf
11 => BuildScrollbar(info, resolve), // UIElement_Scrollbar (reg :124137)
12 => BuildText(info, resolve, elementFont, stringResolve), // UIElement_Text
0x13 => new UiDialogRoot(), // ConfirmationDialog
0x10000031u => new UiItemList(resolve), // UIElement_ItemList toolbar/inventory/paperdoll slots
0x10000031u => new UiItemList(resolve), // UIElement_ItemList — toolbar/inventory/paperdoll slots
0x10000035u => BuildCheckbox(
info, resolve, elementFont, fontResolve, stringResolve), // UIOption_Checkbox
_ => new UiDatElement(info, resolve), // generic fallback (incl. Type 3 chrome/containers)
@ -105,7 +105,7 @@ public static class DatWidgetFactory
e.Width = info.Width;
e.Height = info.Height;
// Honor the dat's draw order. ZLevel is the primary layer (higher = further BACK e.g. the
// Honor the dat's draw order. ZLevel is the primary layer (higher = further BACK — e.g. the
// gmInventoryUI full-window backdrop at ZLevel 100 sits behind the ZLevel-0 panels, #145);
// ReadOrder is the within-layer tiebreaker (higher = on top). K=10000 exceeds any window's
// element count so ZLevel always dominates. Vitals (all ZLevel 0) keep ZOrder == ReadOrder.
@ -134,7 +134,7 @@ public static class DatWidgetFactory
/// </summary>
private static UiScrollbar BuildScrollbar(
ElementInfo info,
Func<uint, (uint tex, int w, int h)> resolve)
Func<uint, (GpuTextureSlot tex, int w, int h)> resolve)
{
var bar = new UiScrollbar
{
@ -302,39 +302,39 @@ public static class DatWidgetFactory
return stateName == "Normal" ? DefaultImage(info) : 0u;
}
// ── Meter ────────────────────────────────────────────────────────────────
// ── Meter ────────────────────────────────────────────────────────────────
/// <summary>
/// Builds a <see cref="UiMeter"/> and populates its sprite ids from the meter's
/// child/grandchild elements (format doc §11). Two shapes are handled:
/// child/grandchild elements (format doc §11). Two shapes are handled:
///
/// <para>
/// <b>3-slice shape</b> (vitals meters 2 Type-3 containers, each with 3 image grandchildren):
/// <b>3-slice shape</b> (vitals meters — 2 Type-3 containers, each with 3 image grandchildren):
/// <code>
/// meter (Type 7)
/// ├── back-layer container (Type 3, lower ReadOrder — drawn first / behind)
/// │ ├── left-cap image (DirectState "" → File = back-left sprite)
/// │ ├── center image (DirectState "" → File = back-tile sprite)
/// │ └── right-cap image (DirectState "" → File = back-right sprite)
/// ├── front-layer container (Type 3, higher ReadOrder — drawn on top)
/// │ ├── left-cap image (→ front-left sprite)
/// │ ├── center image (→ front-tile sprite)
/// │ ├── right-cap image (→ front-right sprite)
/// │ └── expand overlay (named "ShowDetail"/"HideDetail" only — NO DirectState — IGNORED)
/// └── text label (Type 0) (IGNORED — Fill/Label providers bound by VitalsController)
/// ├── back-layer container (Type 3, lower ReadOrder — drawn first / behind)
/// │ ├── left-cap image (DirectState "" → File = back-left sprite)
/// │ ├── center image (DirectState "" → File = back-tile sprite)
/// │ └── right-cap image (DirectState "" → File = back-right sprite)
/// ├── front-layer container (Type 3, higher ReadOrder — drawn on top)
/// │ ├── left-cap image (→ front-left sprite)
/// │ ├── center image (→ front-tile sprite)
/// │ ├── right-cap image (→ front-right sprite)
/// │ └── expand overlay (named "ShowDetail"/"HideDetail" only — NO DirectState — IGNORED)
/// └── text label (Type 0) (IGNORED — Fill/Label providers bound by VitalsController)
/// </code>
/// </para>
///
/// <para>
/// <b>Single-image shape</b> (toolbar selected-object meters 0x100001A1/0x100001A2 1 Type-3
/// <b>Single-image shape</b> (toolbar selected-object meters 0x100001A1/0x100001A2 — 1 Type-3
/// child, no grandchildren): the back-track sprite is on the meter element's own DirectState;
/// the fill sprite is on the single Type-3 child's own DirectState. Both are placed in the
/// TILE slot (Back/FrontTile) with left/right caps 0, so <see cref="UiMeter.DrawHBar"/> tiles
/// them across the full bar geometry (DrawMode=Normal) and clips the fill to the fraction.
/// (retail: gmToolbarUI::HandleSelectionChanged :198635, UIElement_Meter::Initialize :123328)
/// <code>
/// meter (Type 7) [DirectState "" back-track sprite, e.g. 0x0600193E]
/// └── fill container (Type 3) [DirectState "" → fill sprite, e.g. 0x0600193F]
/// meter (Type 7) [DirectState "" → back-track sprite, e.g. 0x0600193E]
/// └── fill container (Type 3) [DirectState "" → fill sprite, e.g. 0x0600193F]
/// </code>
/// </para>
///
@ -345,7 +345,7 @@ public static class DatWidgetFactory
/// </para>
/// </summary>
private static UiMeter BuildMeter(ElementInfo info,
Func<uint, (uint, int, int)> resolve, UiDatFont? datFont)
Func<uint, (GpuTextureSlot, int, int)> resolve, UiDatFont? datFont)
{
var m = new UiMeter
{
@ -383,18 +383,18 @@ public static class DatWidgetFactory
// Single-image shape used by the toolbar selected-object meters
// (health 0x100001A1, mana 0x100001A2).
// - The back-track sprite lives on the meter ELEMENT's own DirectState ("" key of
// info.StateMedia) not on any grandchild image. e.g. health back = 0x0600193E.
// info.StateMedia) — not on any grandchild image. e.g. health back = 0x0600193E.
// - The fill sprite lives on the single Type-3 child's own DirectState ("" key of
// containers[0].StateMedia). e.g. health fill = 0x0600193F.
// The fill child has NO image grandchildren, so SliceIds would return all-zero
// The fill child has NO image grandchildren, so SliceIds would return all-zero —
// read the container's StateMedia directly instead.
//
// These go in the TILE slot (not the left-cap slot): the sprites are DrawMode=Normal,
// which retail renders as "tile at native width to fill the full element geometry"
// (format doc §6; the generic UiDatElement.OnDraw Normal path; UIElement_Meter::
// (format doc §6; the generic UiDatElement.OnDraw Normal path; UIElement_Meter::
// DrawChildren :123574 clips the child's FULL 140px geometry box to the fill fraction).
// With the sprite on BackLeft instead, UiMeter.DrawHBar would clamp the cap to the
// sprite's NATIVE width (capL = min(nativeW, 140)) leaving a right-side gap and
// sprite's NATIVE width (capL = min(nativeW, 140)) — leaving a right-side gap and
// mapping the fill fraction to native width when nativeW < 140. The tile slot makes
// midW = full bar width, so the back tiles across all 140px and the front clips to
// 140*fraction correctly for any native sprite width (left/right caps unused = 0).
@ -435,7 +435,7 @@ public static class DatWidgetFactory
}
else
{
Console.WriteLine($"[D.2b] meter 0x{info.Id:X8}: {containers.Count} Type-3 containers but no recognized 3-slice, direct-fill, or stateful-fill shape bar may render as solid-color fallback.");
Console.WriteLine($"[D.2b] meter 0x{info.Id:X8}: {containers.Count} Type-3 containers but no recognized 3-slice, direct-fill, or stateful-fill shape — bar may render as solid-color fallback.");
}
return m;
@ -482,11 +482,11 @@ public static class DatWidgetFactory
return (left, tile, right);
}
// ── Text ─────────────────────────────────────────────────────────────────
// ── Text ─────────────────────────────────────────────────────────────────
/// <summary>Type-12 UIElement_Text: an editable field or colored-line text view,
/// selected from the canonical property bag. The element's
/// own Direct/Normal media (if any) becomes the background sprite, drawn under the text
/// own Direct/Normal media (if any) becomes the background sprite, drawn under the text —
/// so a Type-12 element that previously rendered via UiDatElement keeps its sprite. Lines
/// are bound later by the controller (LinesProvider). An unbound UiText draws nothing
/// because <see cref="UiText.BackgroundColor"/> defaults to transparent.
@ -497,15 +497,15 @@ public static class DatWidgetFactory
/// that subsequently call <see cref="UiText.Centered"/> / <see cref="UiText.RightAligned"/>
/// on dat-origin elements can be simplified. Controllers that <em>explicitly</em> set those
/// properties after <see cref="ImportedLayout.FindElement"/> still override the build-time
/// defaults the build-time value is just the starting point, not a lock.
/// defaults — the build-time value is just the starting point, not a lock.
/// </para>
/// </summary>
/// <param name="elementFont">The font to seed on the widget. When a font resolver was
/// provided and the element's FontDid resolved successfully, this is that element-specific
/// font; otherwise it is the shared global fallback. Controllers that call
/// <see cref="ImportedLayout.FindElement"/> and set <see cref="UiText.DatFont"/> afterward
/// still override this the build-time value is just the starting point.</param>
private static UiElement BuildText(ElementInfo info, Func<uint, (uint, int, int)> resolve,
/// still override this — the build-time value is just the starting point.</param>
private static UiElement BuildText(ElementInfo info, Func<uint, (GpuTextureSlot, int, int)> resolve,
UiDatFont? elementFont = null,
Func<UiStringInfoValue, string?>? stringResolve = null)
{
@ -547,7 +547,7 @@ public static class DatWidgetFactory
// Apply horizontal + vertical justification from the dat at build time.
// Controllers that call FindElement and set Centered/RightAligned/VerticalJustify
// afterward will override these this is only the dat-driven default.
// afterward will override these — this is only the dat-driven default.
bool centered = info.HJustify == HJustify.Center;
bool rightAligned = info.HJustify == HJustify.Right;
var vJustify = info.VJustify switch
@ -580,7 +580,7 @@ public static class DatWidgetFactory
// Font color from dat property 0x1B (ColorBaseProperty).
// When present, seed DefaultColor so controllers that read it don't have to hard-code colors.
// Controllers that supply explicit per-line colors via LinesProvider still win this is only
// Controllers that supply explicit per-line colors via LinesProvider still win — this is only
// the build-time default.
if (info.FontColor.HasValue)
t.DefaultColor = info.FontColor.Value;
@ -593,7 +593,7 @@ public static class DatWidgetFactory
private static UiButton BuildButton(
ElementInfo info,
Func<uint, (uint, int, int)> resolve,
Func<uint, (GpuTextureSlot, int, int)> resolve,
UiDatFont? elementFont,
Func<uint, UiDatFont?>? fontResolve,
Func<UiStringInfoValue, string?>? stringResolve)
@ -663,7 +663,7 @@ public static class DatWidgetFactory
/// </summary>
private static UiButton BuildCheckbox(
ElementInfo info,
Func<uint, (uint, int, int)> resolve,
Func<uint, (GpuTextureSlot, int, int)> resolve,
UiDatFont? elementFont,
Func<uint, UiDatFont?>? fontResolve,
Func<UiStringInfoValue, string?>? stringResolve)