grom/benilla-octo
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

the numerals kit goes stock, and a chat link is its own mouse target

Browse Source 
- the status-bar numerals kit (`TextStatusBar.xml` + `.lua`) is the reference's own, and the attempt found a class of silent failure: `TextStatusBar_Initialize()` takes no argument in 1.12 and acts on `this`; ours took an optional bar, our unit frames passed one, and the reference's version ignores it — so the numerals would simply never appear, no error, nothing missing; seventeen of our signatures differ from the reference's that way (Lua drops a surplus argument in silence), written down before their files migrate, and the three "ours narrower" ones — `StaticPopup_Hide(which, data)`, `StaticPopup_FindVisible`, `WorldMapButton_OnClick(button, frame)` — closed now, because that direction bites when a STOCK file arrives and calls our version
- a chat link needs no mouse on its frame: wow-re read the client's ctors as mouse-enabling Button, CheckButton, Slider, ColorSelect and EditBox and NOT the scroll frames; the objection ("something makes a 1.12 chat link clickable") resolved as the engine synthesising one mouse-enabled `CSimpleHyperlinkButton` per link span (`0x7a3240`) — so a link span is a hit-test candidate on its own, the chat window is inert everywhere else, and the wheel is its own predicate rather than a borrowed mouse bit; seven `*ScrollFrame` known divergences retire
- the relic fit rule: `IsValidForSlot`'s slot-17 leg is an EQUALITY — `(InventoryType == RELIC) == hasRelicSlot` — so a relic lights the ranged slot for a druid and a bow does not, on the bag bar and the paper doll alike; closing it half-right would have lit the highlight for a druid dragging a bow
- a chain `.xml` whose `.lua` never loaded: `MoneyInputFrame.xml` had been on the chain without its `.lua`, so the template loaded and all ten `MoneyInputFrame_*` verbs read nil, silently — a gate now fails any chain `.xml` whose `.lua` sibling is neither sourced nor listed; the gap report's LOAD column carries the probe's collision caveat, its fx column stops crying wolf (it follows `<Script file=>`, stops reading `.Field(` as a global, subtracts a file's own locals), and a window is its xml PLUS the lua it sources

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
This commit is contained in:
Sam
2026-09-02 17:01:18 +07:00
co-authored by Claude Fable 5.1
parent 112e82d276
commit feca23de3f
30 changed files with 981 additions and 404 deletions
Download Patch File Download Diff File Expand all files Collapse all files
crates/benilla-app/assets/ui/CharacterFrame.xml
+6 -6
View File
@@ -76,7 +76,7 @@
forced-show path). A real forced-highlight affordance is a `benilla-ui` follow-up, not
blocking this slice — `BenillaPaperDollSlot_LockHighlight`/`_UnlockHighlight` below
tint the slot's Quickslot2 ring instead, flagged at their own definition.
- The relic slot: **restored, and this note used to be wrong** (decision 1785). It read
- The relic slot: **restored, and this note used to be wrong** (decision 1796). It read
"`UnitHasRelicSlot` is vanilla-always-false (there is no relic slot in 1.12), so the
ref's branches are DEAD CODE" — they are not. `ChrClasses.dbc` field 16 is set for
Paladin, Shaman and Druid, and INVSLOT 17 doubles as their relic slot. All four of the
@@ -527,7 +527,7 @@
-- ref PaperDollFrame_OnShow (l.568-585): SetGuild OMITTED (named deferral, this file's
-- header). The relic branch (l.580-584) is transcribed verbatim — a relic class has no
-- ranged weapon and therefore no ammo, so its ammo slot hides. This used to be dropped on
-- the belief that UnitHasRelicSlot is always false; decision 1785.
-- the belief that UnitHasRelicSlot is always false; decision 1796.
function PaperDollFrame_OnShow()
paintRows(paperDollRows)
if UnitHasRelicSlot("player") then
@@ -823,7 +823,7 @@
end
-- ref PaperDollFrame_SetRangedAttack (l.419-466). `hasRelic` is transcribed (it used to
-- be dropped — decision 1785). It is the whole reason a paladin's ranged block reads
-- be dropped — decision 1796). It is the whole reason a paladin's ranged block reads
-- "n/a" rather than the stats of the libram sitting in that slot: `noRanged` is set when
-- there is no ranged texture OR the class has a relic slot, and `SetRangedDamage` /
-- `SetRangedAttackPower` both read it (ref l.482 / l.552), so one flag blanks all three
@@ -1020,7 +1020,7 @@
button.emptyTexture = texture
-- `checkRelic` is the number 1 for the ranged slot alone (a property of the SLOT); the
-- class half is `UnitHasRelicSlot`, and the ref reads the two as a conjunction. Both
-- halves are live — decision 1785.
-- halves are live — decision 1796.
button.checkRelic = checkRelic
button.icon = getglobal(button:GetName() .. "IconTexture")
button.count = getglobal(button:GetName() .. "Count")
@@ -1105,7 +1105,7 @@
else
-- ref l.678-681: the empty ranged slot of a relic class shows the relic plate,
-- not the bow plate. The art ships (`UI-PaperDoll-Slot-Relic.blp`, 3916 bytes,
-- patch.MPQ). Restored — decision 1785.
-- patch.MPQ). Restored — decision 1796.
local emptyTexture = button.emptyTexture
if button.checkRelic and UnitHasRelicSlot("player") then
emptyTexture = "Interface\\Paperdoll\\UI-PaperDoll-Slot-Relic.blp"
@@ -1145,7 +1145,7 @@
-- the doll popped a "Currently Equipped" 0/XX beside a template-full main tooltip).
-- The ref's repair-mode branch (l.757-762) is INTERIM-omitted with the binding's
-- repairCost. checkRelic (l.744-746) is transcribed — it was called TBC-flagged dead
-- weight here, which was wrong (decision 1785): a paladin hovering their empty ranged
-- weight here, which was wrong (decision 1796): a paladin hovering their empty ranged
-- slot reads "Relic", not "Ranged".
function BenillaPaperDollSlot_OnEnter(button)
GameTooltip:SetOwner(button, "ANCHOR_RIGHT")
crates/benilla-app/assets/ui/InspectFrame.xml
+1 -1
View File
@@ -48,7 +48,7 @@
1059/1060): ctrl previews the inspected item on your own body, shift posts its link.
Nothing else happens on a click, which is still the reference's own shape.
- `UnitHasRelicSlot` (ref l.123, l.22) — this note used to say "vanilla-always-false, so
the relic branch can never fire". **That is wrong** (decision 1785): the flag is set
the relic branch can never fire". **That is wrong** (decision 1796): the flag is set
for Paladin, Shaman and Druid, and the binding now exists and answers for any token,
with no `"player"` fast path — inspecting an enemy druid answers 1. The branch is
still not transcribed here, but for a different and honest reason: 1.12's inspect pane
crates/benilla-app/assets/ui/TextStatusBar.xml
-183
View File
@@ -1,183 +0,0 @@
<!--
The shared status-bar readout machinery — the contract 1.12's TextStatusBar.lua/.xml defines
(decision 1082), written here in our own terms. A bar opts in by calling
TextStatusBar_Initialize() in its OnLoad (setting `textLockable`), wiring a FontString via
SetTextStatusBarText(bar, text), and routing its OnEnter/OnLeave/OnValueChanged through the
Show/Hide/Update functions here. The text is "value / max", shown while hovered (a lockShow
refcount) or always when the `statusBarText` cvar is "1" ("Status Text" in the options; our
GetCVar returns nil for a cvar the engine doesn't carry, which reads as OFF — the 1.12
default).
Customers: the XP bar (ActionBar.xml) and, since decision 1143, the player and pet frames'
health/power bars (UnitFrames.xml). `textLockable` is what the statusBarText cvar pins, and
which bars carry it is the REFERENCE's own split, not an omission of ours: 1.12 sets it on the
player's two bars, the pet's two and the XP bar, and on nothing else. The target frame's bars
are deliberately not lockable there — its numerals are hover-only — so ours have the machinery
wired and no text regions yet; the hover half needs mouse-enabled bars we do not have.
-->
<Ui xmlns="http://www.blizzard.com/wow/ui/">
<Script><![CDATA[
-- The readout has exactly three reasons to be on screen, and every function below is some
-- combination of them: the Status Text option pins it (only on a bar that opted in with
-- `textLockable`), the pointer is on the bar (a refcount, so nested hovers balance), or the
-- bar is showing its zero text — a phrase rather than a number, which stands on its own.
--
-- The option test is the one of the three that reads from outside the bar, and it is asked
-- from three places, so it lives here once. It takes the option's VALUE rather than calling
-- GetCVar itself: the CVAR_UPDATE handler is handed the new value along with the event and
-- must not go asking the engine for the one it is being told about.
local function optionPins(bar, setting)
return setting == "1" and bar.textLockable
end
-- `bar` is optional and falls back to `this`, the same shape TextStatusBar_UpdateTextString
-- below already has (and for the same reason): the ref calls this from a bar's own OnLoad,
-- where `this` is the bar, but our unit frames wire their children from ONE shared
-- BenillaUnitFrame_OnLoad that holds each bar as a value, not as `this` (decision 1143).
function TextStatusBar_Initialize(bar)
if not bar then
bar = this
end
bar:RegisterEvent("CVAR_UPDATE")
bar.lockShow = 0
end
-- Hand a bar the FontString its readout lives in. A bar without one is never an error
-- below — it simply has nothing to say, and every entry point here checks for that.
function SetTextStatusBarText(bar, text)
if not bar or not text then
return
end
bar.TextString = text
end
function TextStatusBar_OnEvent(cvar, value)
if event ~= "CVAR_UPDATE" or cvar ~= "STATUS_BAR_TEXT" or not this.TextString then
return
end
-- Switching the option ON pins the readout up at once. Switching it off only puts the
-- readout away when nothing else is holding it there: a hover in progress keeps its own,
-- and gets to hide it when it ends.
if optionPins(this, value) then
this.TextString:Show()
elseif this.lockShow == 0 then
this.TextString:Hide()
end
end
-- Repaint one bar's readout from the value it is currently carrying. `textStatusBar` is
-- optional and falls back to `this` for the same reason TextStatusBar_Initialize's is.
function TextStatusBar_UpdateTextString(textStatusBar)
if not textStatusBar then
textStatusBar = this
end
local readout = textStatusBar.TextString
if not readout then
return
end
local value = textStatusBar:GetValue()
local _, valueMax = textStatusBar:GetMinMaxValues()
-- No ceiling is not the same thing as a bar at zero (below): it means the bar does not
-- apply to this unit at all, so the BAR leaves the screen rather than reading out a
-- number. Its readout is left exactly as it stands and comes back with it.
if valueMax <= 0 then
textStatusBar:Hide()
return
end
textStatusBar:Show()
if value == 0 and textStatusBar.zeroText then
-- Empty, and this bar has a phrase for empty. The phrase displaces the numbers and
-- is shown outright: it is a message, so neither the option nor a hover gates it,
-- and `isZero` is the flag that tells the hover not to take it away again.
textStatusBar.isZero = 1
readout:SetText(textStatusBar.zeroText)
readout:Show()
return
end
textStatusBar.isZero = nil
if textStatusBar.prefix then
readout:SetText(textStatusBar.prefix .. " " .. value .. " / " .. valueMax)
else
readout:SetText(value .. " / " .. valueMax)
end
if optionPins(textStatusBar, GetCVar("statusBarText")) or textStatusBar.lockShow > 0 then
readout:Show()
else
readout:Hide()
end
end
function TextStatusBar_OnValueChanged()
TextStatusBar_UpdateTextString()
end
function SetTextStatusBarTextPrefix(bar, prefix)
if bar and bar.TextString then
bar.prefix = prefix
end
end
function SetTextStatusBarTextZeroText(bar, zeroText)
if bar and bar.TextString then
bar.zeroText = zeroText
end
end
-- One more hold on the readout. `lockShow` counts rather than flags because two things can
-- be holding it open at once: the character sheet takes a hold for as long as it is open
-- (CharacterFrame_OnShow, decision 1146) and a hover on the same bar takes a second one on
-- top. Only the last release puts the readout away, so the hover still balances.
function ShowTextStatusBarText(bar)
if bar and bar.TextString then
bar.lockShow = (bar.lockShow or 0) + 1
bar.TextString:Show()
end
end
-- One hold released. The count floors at none rather than going negative, so an unbalanced
-- release — a frame hidden out from under a hover — cannot strand the readout on screen.
function HideTextStatusBarText(bar)
if bar and bar.TextString then
bar.lockShow = math.max((bar.lockShow or 0) - 1, 0)
-- Releasing a hold only puts the readout away if none of the three reasons is left.
-- The ref reads `this.isZero` for the middle one — a leak of the handler's own frame
-- into a helper that otherwise takes `bar`. Equivalent at every call site (each
-- caller passes its handler's own frame), so it is written as `bar`, which also
-- keeps the function callable outside a handler, where `this` is nil.
if bar.lockShow > 0 or bar.isZero == 1
or optionPins(bar, GetCVar("statusBarText")) then
bar.TextString:Show()
else
bar.TextString:Hide()
end
end
end
]]></Script>
<!--
The opt-in template (ref-TextStatusBar.xml): every script routed through the functions
above. One transcription cut, named: the ref OnEnter's TargetFrame arm (re-anchor the unit
tooltip under the target frame's bars) references a TextStatusBar-inheriting TargetFrame
bar we don't build yet — it is dead until that slice, so it is left out rather than
shipped unreachable. A frame that overrides all five scripts (the XP bar does) takes only
the StatusBar widget type from inheriting this.
-->
<StatusBar name="TextStatusBar" virtual="true">
<Scripts>
<OnLoad>TextStatusBar_Initialize()</OnLoad>
<OnEvent>TextStatusBar_OnEvent(arg1, arg2)</OnEvent>
<OnEnter>
TextStatusBar_UpdateTextString()
ShowTextStatusBarText(this)
if this.tooltipTitle and GameTooltip_AddNewbieTip then
BenillaTooltip_AddNewbieTipFor(this, this.tooltipTitle, 1.0, 1.0, 1.0, this.tooltipText, 1)
end
</OnEnter>
<OnLeave>
HideTextStatusBarText(this)
if GameTooltip then GameTooltip:Hide() end
</OnLeave>
<OnValueChanged>TextStatusBar_OnValueChanged()</OnValueChanged>
</Scripts>
</StatusBar>
</Ui>
crates/benilla-app/assets/ui/UiPanels.xml
+28 -6
View File
@@ -1025,10 +1025,24 @@
return "seconds"
end
function StaticPopup_FindVisible(which)
-- ref StaticPopup.lua l.1414-1426, `data` included. Ours took `which` alone until 1793
-- found the arity drift; the reference's second argument is what makes a `multiple` dialog
-- addressable, and dropping it is silent in both directions (Lua discards a surplus
-- argument without complaint).
--
-- The `info.multiple` gate is the reference's own and is why `data` is not simply an
-- equality: a dialog kind declared `multiple` can have several instances up at once, so it
-- is matched by data; every other kind has at most one, so its data is not consulted at
-- all. An unknown `which` answers nil rather than scanning — the reference's own early out.
function StaticPopup_FindVisible(which, data)
local info = StaticPopupDialogs[which]
if not info then
return nil
end
for index = 1, STATICPOPUP_NUMDIALOGS do
local frame = getglobal("StaticPopup" .. index)
if frame:IsShown() and frame.which == which then
if frame:IsShown() and frame.which == which
and (not info.multiple or frame.data == data) then
return frame
end
end
@@ -1044,11 +1058,19 @@
return nil
end
function StaticPopup_Hide(which)
-- ref StaticPopup.lua l.1703-1710, `data` included (1793). With a `data` the reference
-- hides only the instance carrying it; without one it hides every instance of the kind,
-- which is what every one-argument caller means and gets.
--
-- Note the reference does NOT test `IsShown()` here, where `FindVisible` does — hiding an
-- already-hidden dialog is a no-op, and the missing test is not an oversight to correct:
-- `Hide()` on a hidden frame still runs nothing, and adding the guard would change which
-- frames the loop touches for no visible difference. Transcribed as it is.
function StaticPopup_Hide(which, data)
for index = 1, STATICPOPUP_NUMDIALOGS do
local frame = getglobal("StaticPopup" .. index)
if frame:IsShown() and frame.which == which then
frame:Hide()
local dialog = getglobal("StaticPopup" .. index)
if dialog.which == which and (not data or data == dialog.data) then
dialog:Hide()
end
end
end
crates/benilla-app/assets/ui/UnitFrames.xml
+48 -6
View File
@@ -277,9 +277,9 @@
-- and the only conditional here: 1.12 sets it on the player's two bars, the pet's two
-- and the XP bar, and on nothing else, which is exactly why Status Bar Text pins your
-- own numbers and your pet's but never the target's (decision 1143).
if TextStatusBar_Initialize then -- engine tests load file subsets
TextStatusBar_Initialize(frame.healthbar)
TextStatusBar_Initialize(frame.manabar)
-- `TextStatusBar_Initialize` is NOT called here any more — each bar calls it from its
-- own OnLoad, the reference's way (see the bar declarations below).
if SetTextStatusBarText then -- engine tests load file subsets
SetTextStatusBarText(frame.healthbar, getglobal(name .. "HealthBarText"))
SetTextStatusBarText(frame.manabar, getglobal(name .. "ManaBarText"))
if unit == "player" then
@@ -1265,6 +1265,13 @@
<BarTexture file="Interface\TargetingFrame\UI-StatusBar"/>
<BarColor r="0.0" g="1.0" b="0.0"/>
<Scripts>
<!-- The reference initialises a numerals bar from the BAR'S OWN OnLoad
(PlayerFrame.xml l.300-303, PetFrame.xml l.138/163), never from the parent
with the bar as an argument — `TextStatusBar_Initialize()` takes no
argument in 1.12 and acts on `this`. Ours called it from the parent's
OnLoad, which needed a widened signature and would have broken silently
the day TextStatusBar.xml came off the player's chain (decisions 1793). -->
<OnLoad>if TextStatusBar_Initialize then TextStatusBar_Initialize() end</OnLoad>
<OnEvent>if TextStatusBar_OnEvent then TextStatusBar_OnEvent(arg1, arg2) end</OnEvent>
<OnValueChanged>if TextStatusBar_OnValueChanged then TextStatusBar_OnValueChanged() end</OnValueChanged>
<OnEnter>BenillaUnitFrameBar_OnEnter(self)</OnEnter>
@@ -1281,6 +1288,13 @@
<BarTexture file="Interface\TargetingFrame\UI-StatusBar"/>
<BarColor r="0.0" g="0.0" b="1.0"/>
<Scripts>
<!-- The reference initialises a numerals bar from the BAR'S OWN OnLoad
(PlayerFrame.xml l.300-303, PetFrame.xml l.138/163), never from the parent
with the bar as an argument — `TextStatusBar_Initialize()` takes no
argument in 1.12 and acts on `this`. Ours called it from the parent's
OnLoad, which needed a widened signature and would have broken silently
the day TextStatusBar.xml came off the player's chain (decisions 1793). -->
<OnLoad>if TextStatusBar_Initialize then TextStatusBar_Initialize() end</OnLoad>
<OnEvent>if TextStatusBar_OnEvent then TextStatusBar_OnEvent(arg1, arg2) end</OnEvent>
<OnValueChanged>if TextStatusBar_OnValueChanged then TextStatusBar_OnValueChanged() end</OnValueChanged>
<OnEnter>BenillaUnitFrameBar_OnEnter(self)</OnEnter>
@@ -1609,6 +1623,13 @@
<BarTexture file="Interface\TargetingFrame\UI-StatusBar"/>
<BarColor r="0.0" g="1.0" b="0.0"/>
<Scripts>
<!-- The reference initialises a numerals bar from the BAR'S OWN OnLoad
(PlayerFrame.xml l.300-303, PetFrame.xml l.138/163), never from the parent
with the bar as an argument — `TextStatusBar_Initialize()` takes no
argument in 1.12 and acts on `this`. Ours called it from the parent's
OnLoad, which needed a widened signature and would have broken silently
the day TextStatusBar.xml came off the player's chain (decisions 1793). -->
<OnLoad>if TextStatusBar_Initialize then TextStatusBar_Initialize() end</OnLoad>
<OnEvent>if TextStatusBar_OnEvent then TextStatusBar_OnEvent(arg1, arg2) end</OnEvent>
<OnValueChanged>if TextStatusBar_OnValueChanged then TextStatusBar_OnValueChanged() end</OnValueChanged>
<OnEnter>BenillaUnitFrameBar_OnEnter(self)</OnEnter>
@@ -1625,6 +1646,13 @@
<BarTexture file="Interface\TargetingFrame\UI-StatusBar"/>
<BarColor r="0.0" g="0.0" b="1.0"/>
<Scripts>
<!-- The reference initialises a numerals bar from the BAR'S OWN OnLoad
(PlayerFrame.xml l.300-303, PetFrame.xml l.138/163), never from the parent
with the bar as an argument — `TextStatusBar_Initialize()` takes no
argument in 1.12 and acts on `this`. Ours called it from the parent's
OnLoad, which needed a widened signature and would have broken silently
the day TextStatusBar.xml came off the player's chain (decisions 1793). -->
<OnLoad>if TextStatusBar_Initialize then TextStatusBar_Initialize() end</OnLoad>
<OnEvent>if TextStatusBar_OnEvent then TextStatusBar_OnEvent(arg1, arg2) end</OnEvent>
<OnValueChanged>if TextStatusBar_OnValueChanged then TextStatusBar_OnValueChanged() end</OnValueChanged>
<OnEnter>BenillaUnitFrameBar_OnEnter(self)</OnEnter>
@@ -2265,9 +2293,9 @@
frame.healthbar:SetFrameLevel(level - 1)
frame.manabar:SetFrameLevel(level - 1)
-- The pet's numerals, textLockable like the ref's (PetFrame.xml l.139/164; 1143).
if TextStatusBar_Initialize then -- engine tests load file subsets
TextStatusBar_Initialize(frame.healthbar)
TextStatusBar_Initialize(frame.manabar)
-- `TextStatusBar_Initialize` is NOT called here any more — each bar calls it from its
-- own OnLoad, the reference's way (PetFrame.xml l.138/163).
if SetTextStatusBarText then -- engine tests load file subsets
SetTextStatusBarText(frame.healthbar, PetFrameHealthBarText)
SetTextStatusBarText(frame.manabar, PetFrameManaBarText)
frame.healthbar.textLockable = 1
@@ -2574,6 +2602,13 @@
<BarTexture file="Interface\TargetingFrame\UI-StatusBar"/>
<BarColor r="0" g="1.0" b="0"/>
<Scripts>
<!-- The reference initialises a numerals bar from the BAR'S OWN OnLoad
(PlayerFrame.xml l.300-303, PetFrame.xml l.138/163), never from the parent
with the bar as an argument — `TextStatusBar_Initialize()` takes no
argument in 1.12 and acts on `this`. Ours called it from the parent's
OnLoad, which needed a widened signature and would have broken silently
the day TextStatusBar.xml came off the player's chain (decisions 1793). -->
<OnLoad>if TextStatusBar_Initialize then TextStatusBar_Initialize() end</OnLoad>
<OnEvent>if TextStatusBar_OnEvent then TextStatusBar_OnEvent(arg1, arg2) end</OnEvent>
<OnValueChanged>if TextStatusBar_OnValueChanged then TextStatusBar_OnValueChanged() end</OnValueChanged>
<OnEnter>BenillaUnitFrameBar_OnEnter(self)</OnEnter>
@@ -2587,6 +2622,13 @@
<BarTexture file="Interface\TargetingFrame\UI-StatusBar"/>
<BarColor r="0" g="0" b="1.0"/>
<Scripts>
<!-- The reference initialises a numerals bar from the BAR'S OWN OnLoad
(PlayerFrame.xml l.300-303, PetFrame.xml l.138/163), never from the parent
with the bar as an argument — `TextStatusBar_Initialize()` takes no
argument in 1.12 and acts on `this`. Ours called it from the parent's
OnLoad, which needed a widened signature and would have broken silently
the day TextStatusBar.xml came off the player's chain (decisions 1793). -->
<OnLoad>if TextStatusBar_Initialize then TextStatusBar_Initialize() end</OnLoad>
<OnEvent>if TextStatusBar_OnEvent then TextStatusBar_OnEvent(arg1, arg2) end</OnEvent>
<OnValueChanged>if TextStatusBar_OnValueChanged then TextStatusBar_OnValueChanged() end</OnValueChanged>
<OnEnter>BenillaUnitFrameBar_OnEnter(self)</OnEnter>
crates/benilla-app/assets/ui/WorldMapFrame.xml
+14 -4
View File
@@ -389,12 +389,22 @@
end
-- Left click drills into whatever the host finds at that map position; every other button
-- is the zoom-out shortcut (ref WorldMapButton_OnClick — ours takes no button argument,
-- since nothing but the map button itself routes here).
function WorldMapButton_OnClick(mouseButton)
-- is the zoom-out shortcut. ref WorldMapFrame.lua l.277-296. The SECOND argument is the button the click landed on,
-- defaulting to `this` — restored in 1793's sweep, having been dropped in favour of naming
-- `WorldMapButton` directly.
--
-- Equivalent for 1.12's own caller, which passes exactly that button
-- (`WorldMapFrame.lua:174`) — and not equivalent for anyone else, which is the whole
-- reason the reference takes it: a caller aiming the click at a different map surface was
-- silently redirected to the main one, because Lua discards a surplus argument without
-- complaint.
function WorldMapButton_OnClick(mouseButton, button)
CloseDropDownMenus();
if ( mouseButton == "LeftButton" ) then
ProcessMapClick(cursorMapPosition(WorldMapButton));
if ( not button ) then
button = this
end
ProcessMapClick(cursorMapPosition(button));
else
WorldMapZoomOutButton_OnClick();
end
crates/benilla-app/assets/ui/benilla.toc
+31 -9
View File
@@ -149,6 +149,14 @@ MoneyFrame.xml
# `_SetCopper`/`_ClearFocus`/`_SetTextColor`/`_SetOnvalueChangedFunc` — six of the names below —
# and neither can be sourced until they exist. (Trade still needs four ENGINE verbs beyond this:
# `AddTradeMoney`, `GetCursorMoney`, `GetTradePlayerItemLink`, `GetTradeTargetItemLink`.)
#
# **TWO entries, and the `.lua` was missing.** This `.xml` declares only the template and does not
# source its `.lua`; the reference's toc lists both (l.11-12). Listing the `.xml` alone brought the
# template and none of the ten verbs — the header above has claimed otherwise since the day it was
# written, and a probe of a full manifest load answered `nil` for every one of them. Same shape as
# `TextStatusBar`, found by sweeping every chain entry for it once that one bit;
# `every_chain_xml_brings_its_own_lua` is now a gate so neither can recur.
Interface\FrameXML\MoneyInputFrame.lua
Interface\FrameXML\MoneyInputFrame.xml
# The UIPanel slot manager (decision 0084 §2): script-only, no frames of its own — must load
# before any panel frame (Gossip/Merchant today) so ShowUIPanel/HideUIPanel already exist
@@ -210,10 +218,24 @@ UnitPopup.xml
# (`script/object/movable.rs`), so that note was stale too.
Interface\FrameXML\ItemRef.xml
# The shared status-bar numerals machinery (decision 1082): TextStatusBar_* + the opt-in
# template — before its customers (the XP bar in ActionBar.xml today; the unit-frame
# health/mana numerals are the ref's other users, when that slice lands). The ref's own TOC
# loads TextStatusBar right before its unit frames, same seat as here.
TextStatusBar.xml
# template. The ref's own TOC loads TextStatusBar right before its unit frames, same seat.
#
# ══ THE STATUS-BAR NUMERALS — the reference's own, off the player's chain (1751) ═══════════
# `TextStatusBar_*` + the `TextStatusBar` template. The reference's own toc loads it right
# before its unit frames; this is the same seat.
#
# Nothing of ours inherits the template — it exists for the reference's windows, whose rows
# carry `inherits="TextStatusBar"` and would have found ours. What blocked the swap was a
# signature WE widened (`TextStatusBar_Initialize`), fixed by giving our own bars the OnLoad
# the reference's have.
#
# **TWO entries, because the reference has two** (its toc l.32-33). This `.xml` declares only the
# template and does NOT source its `.lua` — unlike most stock windows, whose `<Script file=>`
# brings their code in. Listing the `.xml` alone loads a template whose OnLoad calls a function
# nothing defined: every `TextStatusBar_*` name reads nil, every guarded call is a silent no-op,
# and the numerals simply never appear. The `.lua` comes first, as it does there.
Interface\FrameXML\TextStatusBar.lua
Interface\FrameXML\TextStatusBar.xml
# ══ THE UNIT-FRAME KIT — the reference's own, off the player's chain (1751, window 22) ══
# `UnitFrame.xml` declares NO frames — it is a `<Script file="UnitFrame.lua"/>` and nothing else —
# so this is a pure addition of thirteen globals: `UnitFrame_Initialize`, `_Update`, `_OnEvent`,
@@ -674,10 +696,10 @@ RaidFrame.xml
# label, our measurement being a frame late; that is what the swap tests.
Interface\FrameXML\GossipFrame.xml
# ══ THE VENDOR WINDOW — the reference's own, off the player's chain (1751) ══════════════════
# It took three things, in this order: `PickupMerchantItem` (the vendor cursor, 1786), the
# It took three things, in this order: `PickupMerchantItem` (the vendor cursor, 1797), the
# `BenillaMoney_*` coin kit moving out to MoneyFrame.xml where eight other windows can reach it
# without loading this one, and `SetMerchantCompareItem` (1791) — which the gap report could not
# see at all until 1787 taught it to read `:Method(` sites, and which would have raised on the
# without loading this one, and `SetMerchantCompareItem` (1802) — which the gap report could not
# see at all until 1798 taught it to read `:Method(` sites, and which would have raised on the
# first vendor-row hover with every check green.
#
# That is why the notes below say "grouping" rather than "reuses that file's helpers": the
@@ -779,11 +801,11 @@ TradeSkillFrame.xml
CraftFrame.xml
# ══ THE LOOT WINDOW — the reference's own, off the player's chain (1751) ════════════════════
# It never needed an engine VERB. It needed a widget TYPE: `<LootButton>` is a real registered
# `CreateFrame` type (decision 1788), and registering it as a plain Button would have loaded the
# `CreateFrame` type (decision 1799), and registering it as a plain Button would have loaded the
# file and left every row dead — 1.12 has no Lua verb for "take slot N".
#
# Its ONE method, `SetSlot`, is also what the gap report's new `method=` column found from the
# call site while wow-re was deriving it from the vtable (1787). Both columns read 0 now, and
# call site while wow-re was deriving it from the vtable (1798). Both columns read 0 now, and
# every verb the stock file calls at runtime — not just at load — was checked by hand against
# the bindings before this line changed.
Interface\FrameXML\LootFrame.xml
crates/benilla-app/src/net.rs
+1 -1
View File
@@ -1149,7 +1149,7 @@ pub(crate) enum ClientCommand {
entry: u32,
count: u8,
},
/// Buy into a **named** container slot (`CMSG_BUY_ITEM_IN_SLOT`, decision 1786) — what a
/// Buy into a **named** container slot (`CMSG_BUY_ITEM_IN_SLOT`, decision 1797) — what a
/// vendor row dragged off the merchant window and dropped into a bag or an equipment slot
/// sends, as against [`ClientCommand::BuyItem`]'s auto-place from a plain row click.
/// `bag_guid` is the container object's guid, or the player's own for the backpack and the
crates/benilla-app/src/ui_char.rs
+20 -1
View File
@@ -565,6 +565,9 @@ fn slot_view(
// `ItemSubClass.dbc` — the count gate below reads its `DisplayFlags` bit 2. `None` (the DBC
// failed to load) leaves every bag at 0, which is the plain-bag answer and the safe one.
sub_classes: Option<&crate::ui_items::ItemSubClasses>,
// The player's own `ChrClasses.dbc` relic flag — `find_equip_slot`'s second half (1803). It
// decides INVSLOT 17 both ways, so it reaches every fit-rule read rather than one of them.
has_relic_slot: bool,
guid: u64,
live_id: u32,
) -> Option<InvSlotView> {
@@ -630,7 +633,7 @@ fn slot_view(
Some(crate::ui_items::item_link_full(
entry, 0, roll, 0, &name, t.quality,
)),
find_equip_slot(t.inventory_type),
find_equip_slot(t.inventory_type, has_relic_slot),
(t.class, t.subclass),
t.placeable_on_action_bar(),
)
@@ -725,7 +728,14 @@ fn inventory_slots(
pending: &PendingItemOps,
names: &mut crate::names::NameCache,
sub_classes: Option<&crate::ui_items::ItemSubClasses>,
classes: Option<&benilla_formats::ChrClasses>,
) -> InventorySlots {
// Resolved once from the player's own class byte, right here — the store is already in scope,
// so nothing new is threaded through the callers for the class half.
let has_relic_slot = store
.0
.unit_class()
.is_some_and(|c| classes.is_some_and(|t| t.has_relic_slot(u32::from(c))));
let mut inv: InventorySlots = Default::default();
let ammo_id = store.0.player_ammo_id().unwrap_or(0);
if ammo_id != 0 {
@@ -786,6 +796,7 @@ fn inventory_slots(
pending,
names,
sub_classes,
has_relic_slot,
guid,
live,
);
@@ -825,6 +836,10 @@ fn bank_bag_slots(
pending,
names,
sub_classes,
// A bank bag slot holds a BAG, and `find_equip_slot(INVTYPE_BAG)` answers the four
// bag slots whatever this says — the flag decides INVSLOT 17 alone, which no item in
// this band can reach. `false` here is provable, not a default.
false,
guid,
BANK_BAG_LIVE_FIRST + i as u32,
);
@@ -941,6 +956,9 @@ pub(crate) fn fire_stat_transitions(
#[allow(clippy::too_many_arguments)]
pub(crate) fn feed_char(
// `ChrClasses.dbc` field 16, for the fit rule's relic half — `ui_items::find_equip_slot`
// (1803). Absent client data reads every class as an ordinary ranged wielder.
classes: Option<Res<crate::chr_classes::ChrClassTable>>,
script: Option<NonSendMut<UiScript>>,
self_q: Query<&ObjectStore, With<SelfPlayer>>,
changed_self: Query<(), (With<SelfPlayer>, Changed<ObjectStore>)>,
@@ -1075,6 +1093,7 @@ pub(crate) fn feed_char(
&pending,
&mut names,
sub_classes.as_deref(),
classes.as_deref().map(|t| &t.0),
);
let bank_bags = bank_bag_slots(
store,
crates/benilla-app/src/ui_items/feed.rs
+22 -1
View File
@@ -553,6 +553,10 @@ fn resolve_slot(
petitions: &mut crate::ui_petition::PetitionState,
now: std::time::Instant,
ui_now: f64,
// The player's own `ChrClasses.dbc` relic flag — `find_equip_slot`'s second half. It decides
// ONE slot (INVSLOT 17) and it decides it both ways, so it has to reach every fit-rule read
// rather than being applied at one of them (1803).
has_relic_slot: bool,
) -> Option<ContainerSlot> {
if guid == 0 {
return None;
@@ -674,7 +678,7 @@ fn resolve_slot(
flags,
already_bound,
enchants: enchant_lines,
equip_slots: find_equip_slot(t.inventory_type),
equip_slots: find_equip_slot(t.inventory_type, has_relic_slot),
bar_placeable: t.placeable_on_action_bar(),
cooldown: t.use_spell.and_then(|u| {
let sd = spells.and_then(|s| s.get(u.spell_id));
@@ -765,6 +769,8 @@ pub(crate) fn resolve_item_locks(
#[allow(clippy::too_many_arguments, clippy::type_complexity)] // the param list IS the input set
pub(crate) fn feed_containers(
script: Option<NonSendMut<UiScript>>,
// `ChrClasses.dbc` field 16, for the fit rule's relic half — see `find_equip_slot` (1803).
classes: Option<Res<crate::chr_classes::ChrClassTable>>,
mut items: ResMut<Items>,
icons: Option<Res<ItemDisplays>>,
// The two item-DBC catalogs, as one param (the 16-SystemParam ceiling): `SpellItemEnchantment`'s
@@ -961,6 +967,16 @@ pub(crate) fn feed_containers(
let transitioned: Vec<(i64, u32)> = std::mem::take(&mut lock_cleared.0);
let mut fresh: HashMap<i64, ContainerState> = HashMap::new();
// The player's own relic flag, resolved once. `find_equip_slot` reads it for INVSLOT 17 and
// reads it BOTH ways, so an absent table (no client data) answers false — every class then
// reads as an ordinary ranged wielder, which is the safe degradation.
let has_relic_slot = player.is_some_and(|p| {
p.0.unit_class().is_some_and(|c| {
classes
.as_deref()
.is_some_and(|t| t.0.has_relic_slot(u32::from(c)))
})
});
if let Some(store) = player {
// Bag 0: the backpack — its slots live directly in the player descriptor.
let mut slots = HashMap::new();
@@ -978,6 +994,7 @@ pub(crate) fn feed_containers(
&mut petitions,
now,
ui_now,
has_relic_slot,
) {
slot.locked = pending.contains(0, u32::from(i) + 1);
slots.insert(u32::from(i) + 1, slot);
@@ -1032,6 +1049,7 @@ pub(crate) fn feed_containers(
&mut petitions,
now,
ui_now,
has_relic_slot,
) {
slot.locked = pending.contains(i64::from(bag), j as u32 + 1);
slots.insert(j as u32 + 1, slot);
@@ -1066,6 +1084,7 @@ pub(crate) fn feed_containers(
&mut petitions,
now,
ui_now,
has_relic_slot,
) {
slot.locked = pending.contains(BANK_CONTAINER, u32::from(i) + 1);
slots.insert(u32::from(i) + 1, slot);
@@ -1116,6 +1135,7 @@ pub(crate) fn feed_containers(
&mut petitions,
now,
ui_now,
has_relic_slot,
) {
slot.locked = pending.contains(bag_id, j as u32 + 1);
slots.insert(j as u32 + 1, slot);
@@ -1154,6 +1174,7 @@ pub(crate) fn feed_containers(
&mut petitions,
now,
ui_now,
has_relic_slot,
) {
slot.locked = pending.contains(KEYRING_CONTAINER, u32::from(i) + 1);
slots.insert(u32::from(i) + 1, slot);
crates/benilla-app/src/ui_items/mod.rs
+105 -59
View File
@@ -928,21 +928,17 @@ pub(super) const INVTYPE_AMMO: u32 = 24;
/// actual equip still round-trips through `SMSG_INVENTORY_CHANGE_FAILURE`
/// (`EQUIP_ERR_CANT_DUAL_WIELD`) if the class can't. Simpler than threading class into every
/// caller for a highlight-only consequence.
/// - **`INVTYPE_RELIC` answers no slots, and this one is a real gap** (decision 1785). It used to
/// be excused on the grounds that the relic slot is vanilla-UI-invisible — that
/// `UnitHasRelicSlot` is always false and no relic slot ever shows on the 1.12 paper doll.
/// **That is false.** `ChrClasses.dbc` field 16 is set for Paladin, Shaman and Druid, and
/// `IsValidForSlot 0x5da1d0`'s `slot == 0x11` leg requires `(InventoryType == 28) ==
/// hasRelicSlot` — so for those three classes INVSLOT 17 takes a relic and refuses a ranged
/// weapon, and for the other six the reverse.
/// - **`INVTYPE_RELIC` is CLOSED** (decision 1803, over the gap 1796 named). `has_relic_slot` is
/// the player's own, read off `ChrClasses.dbc` field 16, and `IsValidForSlot 0x5da1d0`'s
/// `slot == 0x11` leg is an EQUALITY — it takes the slot iff
/// `(InventoryType == 28 RELIC) == hasRelicSlot`. So RELIC and the three ranged types are exact
/// complements on that one slot rather than a union: a druid dragging a bow is offered nothing,
/// a warrior dragging an idol likewise, and a druid dragging an idol is offered INVSLOT 17.
///
/// What remains true is the *consequence*: this function drives the `CURSOR_UPDATE` highlight
/// only, and the equip still round-trips through `SMSG_INVENTORY_CHANGE_FAILURE`. So the gap
/// costs a relic-class player the highlight on their ranged slot when dragging a libram, and
/// over-offers it when dragging a bow. Closing it needs the class threaded through
/// [`crate::ui_char::slot_view`], which is the same threading the dual-wield bullet above
/// declines — one change would close both.
pub(super) fn find_equip_slot(inventory_type: u32) -> Vec<u8> {
/// The dual-wield bullet above is NOT closed by the same threading, which is why it is still
/// here: "knows dual wield" is a SPELL the player has learned, not a fact about their class, so
/// it needs the known-spell set rather than the class byte.
pub(super) fn find_equip_slot(inventory_type: u32, has_relic_slot: bool) -> Vec<u8> {
// Live-API ids (`char_stats::SLOT_INFO`'s own numbering): wire `EQUIPMENT_SLOT_*` + 1. The
// ammo slot is the client's own `GetInventorySlotInfo("AmmoSlot")` == 0 (not a real equip slot;
// ammo loads by entry via `CMSG_SET_AMMO`, decision 0526) — it just names the fit-rule target.
@@ -972,31 +968,36 @@ pub(super) fn find_equip_slot(inventory_type: u32) -> Vec<u8> {
const BAG3: u8 = 23;
match inventory_type {
1 => vec![HEAD], // INVTYPE_HEAD
2 => vec![NECK], // INVTYPE_NECK
3 => vec![SHOULDERS], // INVTYPE_SHOULDERS
4 => vec![BODY], // INVTYPE_BODY (the shirt)
5 | 20 => vec![CHEST], // INVTYPE_CHEST / INVTYPE_ROBE (same slot)
6 => vec![WAIST], // INVTYPE_WAIST
7 => vec![LEGS], // INVTYPE_LEGS
8 => vec![FEET], // INVTYPE_FEET
9 => vec![WRISTS], // INVTYPE_WRISTS
10 => vec![HANDS], // INVTYPE_HANDS
11 => vec![FINGER1, FINGER2], // INVTYPE_FINGER
12 => vec![TRINKET1, TRINKET2], // INVTYPE_TRINKET
13 => vec![MAINHAND, OFFHAND], // INVTYPE_WEAPON (dual-wield simplified, see doc)
14 => vec![OFFHAND], // INVTYPE_SHIELD
15 => vec![RANGED], // INVTYPE_RANGED
16 => vec![BACK], // INVTYPE_CLOAK
17 => vec![MAINHAND], // INVTYPE_2HWEAPON
18 => vec![BAG0, BAG1, BAG2, BAG3], // INVTYPE_BAG
19 => vec![TABARD], // INVTYPE_TABARD
21 => vec![MAINHAND], // INVTYPE_WEAPONMAINHAND
22 => vec![OFFHAND], // INVTYPE_WEAPONOFFHAND
23 => vec![OFFHAND], // INVTYPE_HOLDABLE
24 => vec![AMMO], // INVTYPE_AMMO → the ammo slot (loaded via SET_AMMO)
25 => vec![RANGED], // INVTYPE_THROWN
26 => vec![RANGED], // INVTYPE_RANGEDRIGHT
1 => vec![HEAD], // INVTYPE_HEAD
2 => vec![NECK], // INVTYPE_NECK
3 => vec![SHOULDERS], // INVTYPE_SHOULDERS
4 => vec![BODY], // INVTYPE_BODY (the shirt)
5 | 20 => vec![CHEST], // INVTYPE_CHEST / INVTYPE_ROBE (same slot)
6 => vec![WAIST], // INVTYPE_WAIST
7 => vec![LEGS], // INVTYPE_LEGS
8 => vec![FEET], // INVTYPE_FEET
9 => vec![WRISTS], // INVTYPE_WRISTS
10 => vec![HANDS], // INVTYPE_HANDS
11 => vec![FINGER1, FINGER2], // INVTYPE_FINGER
12 => vec![TRINKET1, TRINKET2], // INVTYPE_TRINKET
13 => vec![MAINHAND, OFFHAND], // INVTYPE_WEAPON (dual-wield simplified, see doc)
14 => vec![OFFHAND], // INVTYPE_SHIELD
// The RANGED slot is the relic slot for Paladin/Shaman/Druid, and `IsValidForSlot
// 0x5da1d0`'s `slot == 0x11` leg is an EQUALITY: it takes the slot iff
// `(InventoryType == 28 RELIC) == hasRelicSlot`. So the three ranged types and RELIC are
// exact complements here, not a union — a druid dragging a bow is offered nothing, and a
// warrior dragging an idol likewise (decisions 1796, 1803).
15 if !has_relic_slot => vec![RANGED], // INVTYPE_RANGED
16 => vec![BACK], // INVTYPE_CLOAK
17 => vec![MAINHAND], // INVTYPE_2HWEAPON
18 => vec![BAG0, BAG1, BAG2, BAG3], // INVTYPE_BAG
19 => vec![TABARD], // INVTYPE_TABARD
21 => vec![MAINHAND], // INVTYPE_WEAPONMAINHAND
22 => vec![OFFHAND], // INVTYPE_WEAPONOFFHAND
23 => vec![OFFHAND], // INVTYPE_HOLDABLE
24 => vec![AMMO], // INVTYPE_AMMO → the ammo slot (loaded via SET_AMMO)
25 | 26 if !has_relic_slot => vec![RANGED], // INVTYPE_THROWN / INVTYPE_RANGEDRIGHT
28 if has_relic_slot => vec![RANGED], // INVTYPE_RELIC — the same slot, the other way
// INVTYPE_NON_EQUIP(0), INVTYPE_QUIVER(27), INVTYPE_RELIC(28, see doc), and anything past
// MAX_INVTYPE(29): not equippable.
_ => Vec::new(),
@@ -1326,30 +1327,75 @@ mod tests {
/// `ItemPrototype::GetAllowedEquipSlots` (`Objects/Item.cpp:577-696`).
#[test]
fn find_equip_slot_matches_the_vmangos_table() {
assert_eq!(find_equip_slot(1), vec![1], "HEAD");
assert_eq!(find_equip_slot(2), vec![2], "NECK");
assert_eq!(find_equip_slot(4), vec![4], "BODY (shirt)");
assert_eq!(find_equip_slot(5), vec![5], "CHEST");
assert_eq!(find_equip_slot(20), vec![5], "ROBE aliases CHEST");
assert_eq!(find_equip_slot(11), vec![11, 12], "FINGER, two slots");
assert_eq!(find_equip_slot(12), vec![13, 14], "TRINKET, two slots");
assert_eq!(find_equip_slot(1, false), vec![1], "HEAD");
assert_eq!(find_equip_slot(2, false), vec![2], "NECK");
assert_eq!(find_equip_slot(4, false), vec![4], "BODY (shirt)");
assert_eq!(find_equip_slot(5, false), vec![5], "CHEST");
assert_eq!(find_equip_slot(20, false), vec![5], "ROBE aliases CHEST");
assert_eq!(
find_equip_slot(13),
find_equip_slot(11, false),
vec![11, 12],
"FINGER, two slots"
);
assert_eq!(
find_equip_slot(12, false),
vec![13, 14],
"TRINKET, two slots"
);
assert_eq!(
find_equip_slot(13, false),
vec![16, 17],
"WEAPON offers both hands (dual-wield simplified)"
);
assert_eq!(find_equip_slot(14), vec![17], "SHIELD -> off hand");
assert_eq!(find_equip_slot(15), vec![18], "RANGED");
assert_eq!(find_equip_slot(16), vec![15], "CLOAK -> back");
assert_eq!(find_equip_slot(17), vec![16], "2HWEAPON -> main hand only");
assert_eq!(find_equip_slot(19), vec![19], "TABARD");
assert_eq!(find_equip_slot(21), vec![16], "WEAPONMAINHAND");
assert_eq!(find_equip_slot(22), vec![17], "WEAPONOFFHAND");
assert_eq!(find_equip_slot(18), vec![20, 21, 22, 23], "BAG");
assert_eq!(find_equip_slot(24), vec![0], "AMMO -> the ammo slot (id 0)");
// Not equippable: no vanilla paper-doll slot, or a named deferral (quiver/relic).
for t in [0u32, 27, 28, 100] {
assert!(find_equip_slot(t).is_empty(), "inventory type {t}");
assert_eq!(find_equip_slot(14, false), vec![17], "SHIELD -> off hand");
assert_eq!(find_equip_slot(15, false), vec![18], "RANGED");
assert_eq!(find_equip_slot(16, false), vec![15], "CLOAK -> back");
assert_eq!(
find_equip_slot(17, false),
vec![16],
"2HWEAPON -> main hand only"
);
assert_eq!(find_equip_slot(19, false), vec![19], "TABARD");
assert_eq!(find_equip_slot(21, false), vec![16], "WEAPONMAINHAND");
assert_eq!(find_equip_slot(22, false), vec![17], "WEAPONOFFHAND");
assert_eq!(find_equip_slot(18, false), vec![20, 21, 22, 23], "BAG");
assert_eq!(
find_equip_slot(24, false),
vec![0],
"AMMO -> the ammo slot (id 0)"
);
// Not equippable for anyone: no vanilla paper-doll slot, or a named deferral (quiver).
for t in [0u32, 27, 100] {
for relic in [false, true] {
assert!(find_equip_slot(t, relic).is_empty(), "inventory type {t}");
}
}
}
/// **INVSLOT 17 is one slot with two meanings, and the flag decides it BOTH ways** —
/// `IsValidForSlot 0x5da1d0`'s `slot == 0x11` leg is an equality, not a union
/// (decisions 1796/1803). A relic class is offered the slot for a libram and refused it for a
/// bow; everyone else the reverse. Getting this half-right — offering relics without
/// withdrawing bows — would light the highlight for a druid dragging a bow, which is the one
/// case the reference is unambiguous about.
#[test]
fn the_ranged_slot_takes_a_relic_or_a_weapon_never_both() {
const RANGED: u8 = 18;
// Paladin / Shaman / Druid.
assert_eq!(find_equip_slot(28, true), vec![RANGED], "RELIC");
for t in [15u32, 25, 26] {
assert!(
find_equip_slot(t, true).is_empty(),
"a relic class is offered no ranged slot for inventory type {t}"
);
}
// Everyone else.
assert!(
find_equip_slot(28, false).is_empty(),
"RELIC, no relic slot"
);
for (t, who) in [(15u32, "RANGED"), (25, "THROWN"), (26, "RANGEDRIGHT")] {
assert_eq!(find_equip_slot(t, false), vec![RANGED], "{who}");
}
}
crates/benilla-app/src/ui_script/exp_bar_tests.rs
+12 -15
View File
@@ -11,27 +11,19 @@ use benilla_ui::script::{QuadContent, ScriptValue, UiScript, UnitState};
fn exp_bar_harness() -> UiScript {
let mut s = UiScript::new().unwrap();
s.set_screen_size(1024.0, 768.0);
// Through the shared both-stores loader, not a disk read off `assets/ui`: `TextStatusBar.xml`
// is the reference's own since 1751, and a hand-rolled reader here would look for it in the
// wrong store — and would leave its `<Script file="TextStatusBar.lua"/>` unresolved besides.
for file in [
"Fonts.xml",
"UIParent.xml",
"GameTooltip.xml",
"TextStatusBar.xml",
"Interface\\FrameXML\\TextStatusBar.lua",
"Interface\\FrameXML\\TextStatusBar.xml",
"Cooldown.xml",
"ActionBar.xml",
] {
let text = std::fs::read_to_string(
std::path::Path::new(env!("CARGO_MANIFEST_DIR"))
.join("assets/ui")
.join(file),
)
.unwrap();
let doc = benilla_ui::framexml::parse(&text).unwrap();
let report = benilla_ui::loader::load(&s, &doc, &|_| None);
assert!(
report.errors.is_empty(),
"{file}: loader errors: {:?}",
report.errors
);
super::test_ui::load_ui(&s, file);
}
s
}
@@ -309,7 +301,12 @@ fn the_xp_bar_numerals_show_on_hover() {
.eval::<bool>("return MainMenuBarExpText:IsShown()")
.unwrap());
s.run("BenillaExpBar_OnLeave(MainMenuExpBar)").unwrap();
// `this` bound, because the engine binds it: the stock `HideTextStatusBarText` reads
// `this.isZero` on one line where every other line reads `bar` (TextStatusBar.lua:95 — the
// reference's own slip, benign there because `this` is always SOME frame during dispatch, and
// a nil index only from bare Lua). Called from the bar's own `<OnLeave>` in play.
s.run("this = MainMenuExpBar BenillaExpBar_OnLeave(MainMenuExpBar) this = nil")
.unwrap();
assert!(
!s.eval::<bool>("return MainMenuBarExpText:IsShown()")
.unwrap(),
crates/benilla-app/src/ui_script/frame_flag_gate.rs
+108 -59
View File
@@ -115,28 +115,19 @@ const MOUSE_HANDLERS: &[&str] = &[
"OnDragStart",
];
/// Widget kinds whose **constructor** enables the mouse, so an element of that tag needs no
/// Whether an element's TAG is a widget kind the constructor mouse-enables — so it needs no
/// `enableMouse` and no handler to be clickable.
///
/// This is deliberately the SAME list [`benilla_ui::widget::WidgetArena::create`] uses, and it has
/// to be: both are one claim about the client's ctors (`CSimpleButton`/`CSimpleEditBox` enable
/// input in theirs; a `Slider` thumb must be draggable; a scroll frame takes the wheel — each
/// carries its own citation there). Reading the engine's list here means a wrong entry is wrong in
/// one place rather than two, and the sweep never reports a divergence that is only the two models
/// disagreeing with each other.
///
/// It is also what answers the chat window: `ChatFrame1` is a `ScrollingMessageFrame`, so the
/// reference's takes the mouse from its ctor even though `FloatingChatFrameTemplate` writes
/// `enableMouse="false"` — which is a no-op, the XML path only ever ENABLES
/// (`scripts-auto-enable.md` §1.3). That is how a 1.12 chat link is clickable at all.
const MOUSE_BY_CTOR: &[&str] = &[
"Button",
"CheckButton",
"EditBox",
"ScrollingMessageFrame",
"ScrollFrame",
"Slider",
];
/// **Answered by the engine, not by a copy of its list.** This used to be a `MOUSE_BY_CTOR` array
/// here whose own comment said it was "deliberately the SAME list `WidgetArena::create` uses… so a
/// wrong entry is wrong in one place rather than two". It was a second list, and it drifted
/// silently the first time the engine's was edited — this sweep then reported six scroll frames as
/// divergences that were only the two models disagreeing with each other, which is precisely what
/// the comment claimed could not happen.
fn tag_mouse_enabled_by_ctor(tag: &str) -> bool {
benilla_ui::script::frame_kind_from_tag(tag)
.is_some_and(benilla_ui::widget::mouse_enabled_by_ctor)
}
/// One accepted difference between benilla and the reference, with the reason it is accepted.
///
@@ -185,6 +176,102 @@ const KNOWN: &[Known] = &[
"the reference bar inherits TextStatusBar, whose OnEnter/OnLeave show the value text; \
ours is a plain StatusBar. Wants the status-text arc, not the flag.",
},
// Seven `*ScrollFrame` mouse entries RETIRED here (1795). They read "our faux scroll pane is a
// Frame with an explicit `<OnMouseWheel>`, not a ScrollFrame" — a divergence that existed only
// because OUR `ScrollFrame` ctor took the mouse and the reference's does not. Correcting the
// ctor list to the client's made both sides agree, and an accepted divergence that has been
// fixed is documentation claiming a defect we do not have.
Known {
frame: "ChatFrame1",
flag: Flag::Mouse,
why: "our chat window takes the mouse and the reference's does not. It was believed a \
`<ScrollingMessageFrame>` takes it BY CONSTRUCTION; wow-re's follow-up round \
(`68987021`) proved that ctor leaves the bit clear, and that a 1.12 chat link is \
clickable through one synthesised `CSimpleHyperlinkButton` child per link span. Our \
hit test now carries that law as a span disjunct, so the LINKS no longer need the \
flag — but ChatFrame.xml still sets it, and a documented chain of choices hangs off \
the old premise (no `SetFrameLevel-1` OnLoad because it would sink the resize grips \
under a mouse-enabled parent; `OnClick` dismisses a held spell, 0843). Unwinding \
that is the chat window's own piece of work, not this flag's.",
},
Known {
frame: "ChatFrame2",
flag: Flag::Mouse,
why: "our chat window takes the mouse and the reference's does not. It was believed a \
`<ScrollingMessageFrame>` takes it BY CONSTRUCTION; wow-re's follow-up round \
(`68987021`) proved that ctor leaves the bit clear, and that a 1.12 chat link is \
clickable through one synthesised `CSimpleHyperlinkButton` child per link span. Our \
hit test now carries that law as a span disjunct, so the LINKS no longer need the \
flag — but ChatFrame.xml still sets it, and a documented chain of choices hangs off \
the old premise (no `SetFrameLevel-1` OnLoad because it would sink the resize grips \
under a mouse-enabled parent; `OnClick` dismisses a held spell, 0843). Unwinding \
that is the chat window's own piece of work, not this flag's.",
},
Known {
frame: "ChatFrame3",
flag: Flag::Mouse,
why: "our chat window takes the mouse and the reference's does not. It was believed a \
`<ScrollingMessageFrame>` takes it BY CONSTRUCTION; wow-re's follow-up round \
(`68987021`) proved that ctor leaves the bit clear, and that a 1.12 chat link is \
clickable through one synthesised `CSimpleHyperlinkButton` child per link span. Our \
hit test now carries that law as a span disjunct, so the LINKS no longer need the \
flag — but ChatFrame.xml still sets it, and a documented chain of choices hangs off \
the old premise (no `SetFrameLevel-1` OnLoad because it would sink the resize grips \
under a mouse-enabled parent; `OnClick` dismisses a held spell, 0843). Unwinding \
that is the chat window's own piece of work, not this flag's.",
},
Known {
frame: "ChatFrame4",
flag: Flag::Mouse,
why: "our chat window takes the mouse and the reference's does not. It was believed a \
`<ScrollingMessageFrame>` takes it BY CONSTRUCTION; wow-re's follow-up round \
(`68987021`) proved that ctor leaves the bit clear, and that a 1.12 chat link is \
clickable through one synthesised `CSimpleHyperlinkButton` child per link span. Our \
hit test now carries that law as a span disjunct, so the LINKS no longer need the \
flag — but ChatFrame.xml still sets it, and a documented chain of choices hangs off \
the old premise (no `SetFrameLevel-1` OnLoad because it would sink the resize grips \
under a mouse-enabled parent; `OnClick` dismisses a held spell, 0843). Unwinding \
that is the chat window's own piece of work, not this flag's.",
},
Known {
frame: "ChatFrame5",
flag: Flag::Mouse,
why: "our chat window takes the mouse and the reference's does not. It was believed a \
`<ScrollingMessageFrame>` takes it BY CONSTRUCTION; wow-re's follow-up round \
(`68987021`) proved that ctor leaves the bit clear, and that a 1.12 chat link is \
clickable through one synthesised `CSimpleHyperlinkButton` child per link span. Our \
hit test now carries that law as a span disjunct, so the LINKS no longer need the \
flag — but ChatFrame.xml still sets it, and a documented chain of choices hangs off \
the old premise (no `SetFrameLevel-1` OnLoad because it would sink the resize grips \
under a mouse-enabled parent; `OnClick` dismisses a held spell, 0843). Unwinding \
that is the chat window's own piece of work, not this flag's.",
},
Known {
frame: "ChatFrame6",
flag: Flag::Mouse,
why: "our chat window takes the mouse and the reference's does not. It was believed a \
`<ScrollingMessageFrame>` takes it BY CONSTRUCTION; wow-re's follow-up round \
(`68987021`) proved that ctor leaves the bit clear, and that a 1.12 chat link is \
clickable through one synthesised `CSimpleHyperlinkButton` child per link span. Our \
hit test now carries that law as a span disjunct, so the LINKS no longer need the \
flag — but ChatFrame.xml still sets it, and a documented chain of choices hangs off \
the old premise (no `SetFrameLevel-1` OnLoad because it would sink the resize grips \
under a mouse-enabled parent; `OnClick` dismisses a held spell, 0843). Unwinding \
that is the chat window's own piece of work, not this flag's.",
},
Known {
frame: "ChatFrame7",
flag: Flag::Mouse,
why: "our chat window takes the mouse and the reference's does not. It was believed a \
`<ScrollingMessageFrame>` takes it BY CONSTRUCTION; wow-re's follow-up round \
(`68987021`) proved that ctor leaves the bit clear, and that a 1.12 chat link is \
clickable through one synthesised `CSimpleHyperlinkButton` child per link span. Our \
hit test now carries that law as a span disjunct, so the LINKS no longer need the \
flag — but ChatFrame.xml still sets it, and a documented chain of choices hangs off \
the old premise (no `SetFrameLevel-1` OnLoad because it would sink the resize grips \
under a mouse-enabled parent; `OnClick` dismisses a held spell, 0843). Unwinding \
that is the chat window's own piece of work, not this flag's.",
},
Known {
frame: "TargetofTargetHealthBar",
flag: Flag::Mouse,
@@ -242,41 +329,6 @@ const KNOWN: &[Known] = &[
//
// The two mail panes are a further step out: ours are flat art with a FontString body, a
// render approximation MailFrame.xml names at the site, so neither side scrolls them.
Known {
frame: "FriendsFrameFriendsScrollFrame",
flag: Flag::Mouse,
why: "our faux scroll pane is a Frame with an explicit <OnMouseWheel>, not a ScrollFrame",
},
Known {
frame: "FriendsFrameIgnoreScrollFrame",
flag: Flag::Mouse,
why: "as FriendsFrameFriendsScrollFrame",
},
Known {
frame: "GuildListScrollFrame",
flag: Flag::Mouse,
why: "as FriendsFrameFriendsScrollFrame",
},
Known {
frame: "WhoListScrollFrame",
flag: Flag::Mouse,
why: "as FriendsFrameFriendsScrollFrame",
},
Known {
frame: "SkillListScrollFrame",
flag: Flag::Mouse,
why: "as FriendsFrameFriendsScrollFrame",
},
Known {
frame: "ReputationListScrollFrame",
flag: Flag::Mouse,
why: "as FriendsFrameFriendsScrollFrame",
},
Known {
frame: "QuestLogListScrollFrame",
flag: Flag::Mouse,
why: "as FriendsFrameFriendsScrollFrame",
},
// ── id ─────────────────────────────────────────────────────────────────────────────────────
//
// Empty, and it is worth saying why rather than leaving a bare header. The reference side of
@@ -482,10 +534,7 @@ fn reference_takes_mouse(name: &str, frames: &HashMap<String, Element>) -> bool
let Some(el) = frames.get(name) else {
return false;
};
if MOUSE_BY_CTOR
.iter()
.any(|t| t.eq_ignore_ascii_case(&el.tag))
{
if tag_mouse_enabled_by_ctor(&el.tag) {
return true;
}
if resolved_attr(name, frames, "enableMouse", 0).is_some_and(|v| v.eq_ignore_ascii_case("true"))
crates/benilla-app/src/ui_script/honor_frame_tests.rs
+2 -1
View File
@@ -71,7 +71,8 @@ fn load_page(s: &UiScript) {
"UiPanels.xml",
"UIParent.xml",
"GameTooltip.xml",
"TextStatusBar.xml",
"Interface\\FrameXML\\TextStatusBar.lua",
"Interface\\FrameXML\\TextStatusBar.xml",
"ScrollTemplates.xml",
"UIPanelTemplates.xml",
"OptionsFrameTemplates.xml",
crates/benilla-app/src/ui_script/options_tests.rs
+2 -1
View File
@@ -728,7 +728,8 @@ fn interface_harness() -> UiScript {
"GameTooltip.xml",
"UIDropDownMenu.xml",
"UnitPopup.xml",
"TextStatusBar.xml",
"Interface\\FrameXML\\TextStatusBar.lua",
"Interface\\FrameXML\\TextStatusBar.xml",
"UnitFrames.xml",
"Cooldown.xml",
"ActionBar.xml",
crates/benilla-app/src/ui_script/reference_ui.rs
+389 -13
View File
@@ -395,13 +395,13 @@ mod tests {
/// manifest in a fresh VM — the same pass [`chain_readiness_report`] makes, run here so the
/// answer is in one table.
///
/// That last one is 1790, and it is the same mistake as the third column one step later.
/// That last one is 1801, and it is the same mistake as the third column one step later.
/// `<LootButton>` and `<TaxiRouteFrame>` are element TAGS: no census of `Name(` or `:Name(`
/// can reach them, so `LootFrame.xml` sat in this report's "needs NO engine work" list while
/// the readiness probe was printing `4 issue(s)` for it in a different table. Both tables were
/// right. Joining them was left to whoever read them, and I read it wrong.
///
/// Until 1787 the method sites were **silently dropped**: a name in neither the engine nor the
/// Until 1798 the method sites were **silently dropped**: a name in neither the engine nor the
/// FrameXML half of `1.12-globals.tsv` was assumed to be a widget method and skipped, on the
/// reasoning that widget methods are not in `_G`. True, and it meant the report could not see
/// a widget method we had *not built*. `MerchantFrame.xml` read `0 engine` and
@@ -410,10 +410,16 @@ mod tests {
/// file loaded, every check passed, and hovering a vendor row would have raised in play. Both
/// instruments were right about what they measure. Neither measured the window.
///
/// **A remaining `<?>` in the `fx=` column is usually a LoadOnDemand addon**, not something to
/// build. `ClassTrainerFrame_Show`, `CraftFrame_Show`, `MacroFrame_SaveMacro`,
/// `InspectFrame_Show`, `TalentFrame_Toggle` and their siblings live in `Blizzard_*` addons the
/// install ships **packed as `.pub`**, so no amount of scanning the extracted FrameXML finds
/// them. They arrive when that addon does, exactly like an `fx=` name with a home.
///
/// One crudeness remains, stated because it decides how to read the output: neither scan can
/// see a name reached through `getglobal`, so both can under-report.
///
/// They no longer over-report on comments. Both counted commented-out calls until 1789 — 1.12
/// They no longer over-report on comments. Both counted commented-out calls until 1800 — 1.12
/// comments out whole blocks, and `PaperDollFrame.lua:754-756`'s `ShowInventorySellCursor` is
/// three commented lines that this report named as that window's last engine gap. A real
/// binding, never called, blocking a window that was not blocked. [`strip_comments`] takes
@@ -457,7 +463,7 @@ mod tests {
.collect();
// Strip what is not code before either scan. Both `Name(` and `:Name(` counted calls
// inside comments until 1789: the round that built `PickupMerchantItem` also asked wow-re
// inside comments until 1800: the round that built `PickupMerchantItem` also asked wow-re
// for `ShowInventorySellCursor`, which this report had named as `PaperDollFrame.xml`'s
// last engine gap — and the answer was that stock `PaperDollFrame.lua:754-756` has the
// call **commented out**, all three lines. A real binding, never called, blocking a window
@@ -604,7 +610,7 @@ mod tests {
.collect()
};
// The instrument's own tripwire: if the probe stops working, the `method=` column goes
// silently empty — which is precisely the failure mode 1787 exists to end.
// silently empty — which is precisely the failure mode 1798 exists to end.
let control: Vec<String> = ["SetPoint", "SetMerchantItem", "BenillaNotAMethod"]
.iter()
.map(|s| (*s).to_string())
@@ -636,10 +642,48 @@ mod tests {
.map(str::to_string)
.collect();
// The `.lua` files a stock `.xml` SOURCES through `<Script file=>`. A window's own sourced
// code is part of the window: its functions are not gaps, and they are where most of its
// `fx=` names would otherwise be looked for.
let sourced_luas = |xml_leaf: &str| -> Vec<String> {
let Some(b) = super::read(&format!("Interface\\FrameXML\\{xml_leaf}")) else {
return Vec::new();
};
let text = String::from_utf8_lossy(&b).into_owned();
let mut out = Vec::new();
for (i, _) in text.match_indices("<Script") {
let rest = &text[i..];
let Some(end) = rest.find("/>").or_else(|| rest.find('>')) else {
continue;
};
let tag = &rest[..end];
if let Some(fi) = tag.find("file=\"") {
let after = &tag[fi + 6..];
if let Some(q) = after.find('"') {
let leaf = after[..q].rsplit(['\\', '/']).next().unwrap_or("");
if leaf.ends_with(".lua") {
out.push(leaf.to_string());
}
}
}
}
out
};
// `function Name(` across the whole corpus, so an fx gap can name the file that holds it.
//
// Each toc `.xml` is scanned together with **every `.lua` it SOURCES**, not just the
// `X.lua` its own name suggests. Guessing missed the common case: `ActionBarFrame.xml`
// sources `ActionButton.lua`, there is no `ActionBarFrame.lua`, and `ActionButton.lua` is
// not a toc line of its own — so the whole `ActionButton_*` family read `<?>`, which says
// "nothing defines this" about five functions a stock file defines and brings with it.
// The difference matters for how the column is read: a name with a home ARRIVES when that
// file does; a `<?>` is something to build.
let mut home: std::collections::HashMap<String, String> = std::collections::HashMap::new();
for f in &stock {
for cand in [f.clone(), format!("{}.lua", &f[..f.len() - 4])] {
let mut cands = vec![f.clone(), format!("{}.lua", &f[..f.len() - 4])];
cands.extend(sourced_luas(f));
for cand in cands {
let Some(bytes) = super::read(&format!("Interface\\FrameXML\\{cand}")) else {
continue;
};
@@ -662,7 +706,16 @@ mod tests {
let mut out = std::collections::HashSet::new();
let mut i = 0;
while i < b.len() {
if b[i].is_ascii_uppercase() && (i == 0 || !super::is_word(b[i - 1])) {
// A GLOBAL call: uppercase, not inside a word, and **not preceded by `.` or
// `:`**. Without that last test `info.UpdateFunc(` and `dialog.OnAccept(` read as
// globals — `.` is not a word character — and a dozen `StaticPopupDialogs` /
// `MoneyTypeInfo` FIELD names showed up as `<?>` gaps. `:` belongs to the method
// scan below; `.` belongs to nobody, since a field call arrives with its table.
let after_field = i > 0 && (b[i - 1] == '.' || b[i - 1] == ':');
if b[i].is_ascii_uppercase()
&& !after_field
&& (i == 0 || !super::is_word(b[i - 1]))
{
let mut j = i;
while j < b.len() && super::is_word(b[j]) {
j += 1;
@@ -696,13 +749,39 @@ mod tests {
if migrated.contains(f) {
continue;
}
// The window IS its xml plus every `.lua` it sources — `ActionBarFrame.xml` sources
// `ActionButton.lua`, and the whole `ActionButton_*` family is that window's own code,
// not a dependency on somebody else's. Gathering only `X.xml` + `X.lua` reported five
// of its own functions as gaps.
let mut text = String::new();
for cand in [f.clone(), format!("{}.lua", &f[..f.len() - 4])] {
let mut parts = vec![f.clone(), format!("{}.lua", &f[..f.len() - 4])];
parts.extend(sourced_luas(f));
for cand in parts {
if let Some(b) = super::read(&format!("Interface\\FrameXML\\{cand}")) {
text.push_str(&String::from_utf8_lossy(&b));
}
}
let text = strip_comments(&text);
// …and the names the file declares as LOCALS, which a bare call resolves to. The
// idiom that needs it is the reference's own dispatch shape:
//
// local OnAccept = StaticPopupDialogs[dialog.which].OnAccept
// if ( OnAccept ) then dontHide = OnAccept(dialog.data, dialog.data2) end
//
// The call is bare, so the global scan sees `OnAccept(` and reports a gap for a name
// that is a table field one line up. `OnAccept`, `OnCancel`, `OnShow`, `OnHide` and
// the three `EditBoxOn*` all arrived that way. A `local` declaration in the same file
// is proof enough: nothing else could be meant.
let locals: std::collections::HashSet<String> = text
.lines()
.filter_map(|l| l.trim_start().strip_prefix("local "))
.map(|r| {
r.chars()
.take_while(|c| c.is_alphanumeric() || *c == '_')
.collect::<String>()
})
.filter(|n| !n.is_empty())
.collect();
let own: std::collections::HashSet<String> = text
.lines()
.filter_map(|l| l.trim_start().strip_prefix("function "))
@@ -715,7 +794,7 @@ mod tests {
let (mut eng, mut fx) = (Vec::new(), Vec::new());
let mut names: Vec<String> = called(&text)
.into_iter()
.filter(|c| !own.contains(c) && !have.contains(c))
.filter(|c| !own.contains(c) && !locals.contains(c) && !have.contains(c))
.collect();
names.sort();
for c in names {
@@ -737,8 +816,30 @@ mod tests {
// while `chain_readiness_report` was printing `4 issue(s)` for it in another table.
//
// Both tables were right. Joining them was left to whoever read them, and I got it
// wrong (1790). So this one runs the load itself — the same fresh-VM-plus-manifest
// wrong (1801). So this one runs the load itself — the same fresh-VM-plus-manifest
// pass the readiness probe does — and reports it in the same row.
// **A `LOAD:` line for a window we ALSO ship is suspect**, and the reason is the same
// one `chain_readiness_report` carries: `publish_global` is non-overwriting (RF-0023),
// so loading the stock file on top of our identically-named one leaves every colliding
// frame's global pointing at OUR frame while the stock file's handlers run against
// THEIRS. A field set on `this` in an OnLoad is then invisible to `getglobal(name)`,
// and the failure looks like a bug in whatever read it back.
//
// Worked example, because it cost an hour: stock `TradeFrame.xml` reported
// `MoneyFrame.xml:525: attempt to index local 'info'`. Nothing was wrong with our
// MoneyFrame — `MoneyFrame_SetType` had set `this.info` correctly, and
// `getglobal(this:GetName())` answered a DIFFERENT table, because our own
// `TradeFrame.xml` already owned `TradeRecipientMoneyFrame`.
//
// So the column is marked, not trusted. A row whose file we do not ship is a real
// load failure; a row whose file we do ship needs the swap attempted to know.
let ours_too = !migrated.contains(f)
&& std::fs::metadata(
std::path::Path::new(env!("CARGO_MANIFEST_DIR"))
.join("assets/ui")
.join(f),
)
.is_ok();
let loads = {
let mut probe = UiScript::new().expect("VM");
probe.set_screen_size(1024.0, 768.0);
@@ -759,8 +860,22 @@ mod tests {
let known = answers(&s, &asked);
let mut meth: Vec<String> = asked.into_iter().filter(|m| !known.contains(m)).collect();
meth.sort();
// A suspect LOAD line does not count as a blocker — it is a question, not an answer.
let load_blockers = if ours_too { 0 } else { loads.len() };
let loads: Vec<String> = loads
.into_iter()
.map(|e| {
let one = e.replace('\n', " ");
let mark = if ours_too {
" (ours too — suspect)"
} else {
""
};
format!("{mark} {}", &one[..one.len().min(150)])
})
.collect();
rows.push((
eng.len() + meth.len() + loads.len(),
eng.len() + meth.len() + load_blockers,
f.clone(),
eng,
fx,
@@ -778,8 +893,7 @@ mod tests {
println!(" method: {}", meth.join(" "));
}
for e in loads {
let one = e.replace('\n', " ");
println!(" LOAD: {}", &one[..one.len().min(150)]);
println!(" LOAD:{e}");
}
if !fx.is_empty() {
println!(" fx: {}", fx.join(" "));
@@ -816,12 +930,103 @@ mod tests {
/// load, which is what this reports: a name ours defines that a chain entry in our own
/// manifest also defines.
///
/// **Two halves, both exact, because both sides DECLARE rather than mention.** The function
/// half is above. The frame half asks the other question a swap has to answer: *can this stock
/// window be added at all* — which is different from "does it load", because a stock window we
/// do not ship under its own name usually has a counterpart of ours under a different one, and
/// both would declare the same frames.
///
/// That half doubles as the map nothing else holds: `ActionBarFrame.xml` is our
/// `ActionBar.xml`, `FloatingChatFrame.xml` is our `ChatFrame.xml`, `PlayerFrame.xml` and
/// `TargetFrame.xml` and `PetFrame.xml` are all our one `UnitFrames.xml`,
/// `MainMenuBarMicroButtons.xml` is our `MicroMenu.xml`, `StaticPopup.xml` is our
/// `UiPanels.xml`. `chain_gap_report` calls several of those unblocked, and they are —
/// individually. They just cannot load beside the file of ours already holding their names.
///
/// Run it before attempting a swap. It predicts which ones will fail without attempting them.
#[test]
#[ignore = "instrument: run by hand before attempting a window swap"]
fn shadowed_reference_functions() {
let _data = benilla_formats::wow_data_or_skip!();
// The FRAME NAMES a document declares — `name="X"` on a widget element, minus the
// `$parent`-relative and `virtual` template forms, which name nothing globally.
//
// The sibling of the function check below and exactly as exact, because both sides
// DECLARE rather than mention. This is the version of the "frame names" idea that works:
// an earlier attempt scanned names a file *referenced* and was pure noise (it found `UI`
// and missed the case it was built for), because a reference can live in any file. A
// declaration cannot.
//
// What it catches: our `UnitFrames.xml` declares `PetFrame`, and so does the stock
// `PetFrame.xml`. Adding that stock file alongside ours would declare the name twice.
// Several stock windows we do not ship under their own name have an equivalent of ours
// under a different one, and `chain_gap_report` calls every one of them unblocked —
// truthfully, because the stock file WOULD load; it just cannot load *beside* ours.
let declares = |text: &str| -> Vec<String> {
let mut out = Vec::new();
for (i, _) in text.match_indices("name=\"") {
let rest = &text[i + 6..];
let Some(end) = rest.find('"') else { continue };
let name = &rest[..end];
if name.starts_with('$') || name.is_empty() {
continue;
}
// A `virtual="true"` element is a template: its name is a registry key, not a
// frame, and two files may legitimately hold the same template name only if one
// replaces the other — which is the same question, so they are reported too.
out.push(name.to_string());
}
out
};
// `function Name(a, b, c)` → the name and its PARAMETER LIST. Arity is the third way our
// files and the reference's can disagree about a name, and the one that fails most
// quietly: `TextStatusBar_Initialize()` takes no argument in 1.12 and acts on `this`, ours
// took an optional bar, and `UnitFrames.xml` calls it with one. Swapping that file stopped
// initialising the unit-frame bars with no error and no missing global — the numerals
// simply never appeared (decision 1793).
//
// **Both directions are silent, and that is the point.** Lua drops extra arguments without
// complaint, so neither an over- nor an under-supplied call raises; they differ only in
// WHO loses information:
//
// * ours WIDER — our callers pass the extra argument and the reference's version drops
// it. Breaks when OUR file is swapped out (the `TextStatusBar` case).
// * ours NARROWER — the reference's callers pass more than ours takes and ours drops it.
// Breaks when a STOCK file is added beside ours and calls our version.
//
// The report names the direction because it says which swap the difference is waiting for,
// not because one of them is safe.
let params_in = |text: &str| -> Vec<(String, usize)> {
let mut out = Vec::new();
for line in text.lines() {
let Some(rest) = line.trim_start().strip_prefix("function ") else {
continue;
};
let name: String = rest
.chars()
.take_while(|c| c.is_alphanumeric() || *c == '_')
.collect();
if name.is_empty() {
continue;
}
let Some(open) = rest.find('(') else { continue };
let Some(close) = rest[open..].find(')') else {
continue;
};
let args = rest[open + 1..open + close].trim();
// `...` is a vararg, which has no fixed arity to compare.
let n = if args.is_empty() || args == "..." {
0
} else {
args.split(',').count()
};
out.push((name, n));
}
out
};
let defined_in = |text: &str| -> Vec<String> {
text.lines()
.filter_map(|l| l.trim_start().strip_prefix("function "))
@@ -841,6 +1046,8 @@ mod tests {
toc.iter().enumerate().map(|(i, f)| (f, i)).collect();
let mut chain_home: std::collections::HashMap<String, (String, usize)> =
std::collections::HashMap::new();
let mut chain_frames: std::collections::HashMap<String, String> =
std::collections::HashMap::new();
for entry in toc.iter().filter(|f| super::is_chain_entry(f)) {
let leaf = entry.rsplit(['\\', '/']).next().unwrap_or(entry);
let mut cands = vec![entry.clone()];
@@ -878,6 +1085,67 @@ mod tests {
}
hits.sort();
hits.dedup();
// The frame names every STOCK window declares — read off the reference's own toc, not off
// our manifest's chain entries. Scanning only what we already load was the first attempt
// and it answered 0 by construction: a stock file we do not load is exactly the one whose
// names could collide, and it was never opened.
let ref_toc = String::from_utf8_lossy(
&super::read("Interface\\FrameXML\\FrameXML.toc").expect("the reference's own toc"),
)
.into_owned();
let mut ref_arity: std::collections::HashMap<String, (usize, String)> =
std::collections::HashMap::new();
for line in ref_toc.lines().map(str::trim) {
if line.is_empty() || line.starts_with('#') || !line.ends_with(".xml") {
continue;
}
for cand in [line.to_string(), format!("{}.lua", &line[..line.len() - 4])] {
let Some(b) = super::read(&format!("Interface/FrameXML/{cand}")) else {
continue;
};
let text = String::from_utf8_lossy(&b).into_owned();
if cand.ends_with(".xml") {
for name in declares(&text) {
chain_frames.entry(name).or_insert_with(|| line.to_string());
}
}
for (name, n) in params_in(&text) {
ref_arity.entry(name).or_insert((n, cand.clone()));
}
}
}
assert!(
chain_frames.contains_key("GameTooltip") && chain_frames.contains_key("PetFrame"),
"the stock frame-name scan found nothing recognisable ({} names) — an empty answer \
here reads as \"no collisions\", which is what the first version of this reported \
for the wrong reason",
chain_frames.len()
);
// …and the frame-name half, over the stock files the manifest does NOT already take off
// the chain. A file of ours whose stock counterpart we already load is a swap that has
// happened; this is about the ones that have not.
let mut frame_hits: Vec<(String, String, String)> = Vec::new();
for entry in toc.iter().filter(|f| !super::is_chain_entry(f)) {
let Ok(text) = std::fs::read_to_string(dir.join(entry)) else {
continue;
};
for name in declares(&text) {
if let Some(home) = chain_frames.get(&name) {
let already = toc
.iter()
.filter(|f| super::is_chain_entry(f))
.any(|f| f.ends_with(home.as_str()));
// A template's name is a registry key rather than a frame, but two files
// holding one is the same question, so it is reported the same way.
if !already {
frame_hits.push((name, entry.clone(), home.clone()));
}
}
}
}
frame_hits.sort();
frame_hits.dedup();
println!("\n=== names ours redefines that a CHAIN entry already defines ===");
println!("{:<36} {:<28} {:<34} winner", "name", "ours", "chain");
@@ -901,6 +1169,114 @@ mod tests {
for (f, n) in &by_file {
println!(" {n:>3} {f}");
}
// The ARITY half: a name we define that the reference also defines, with a different
// parameter count. Not a collision — the two need never both load for this to bite — so it
// is its own section rather than a column.
let mut arity: Vec<(String, String, usize, usize, String)> = Vec::new();
for entry in toc.iter().filter(|f| !super::is_chain_entry(f)) {
let Ok(text) = std::fs::read_to_string(dir.join(entry)) else {
continue;
};
for (name, ours_n) in params_in(&text) {
if let Some((ref_n, home)) = ref_arity.get(&name) {
if *ref_n != ours_n {
arity.push((name, entry.clone(), ours_n, *ref_n, home.clone()));
}
}
}
}
arity.sort();
arity.dedup();
println!(
"\n=== {} signatures of ours differ in ARITY from the reference's ===",
arity.len()
);
println!("{:<34} {:<26} ours ref direction", "name", "ours");
for (name, ours, a, b, home) in &arity {
// Which swap this one is waiting for — see the note above; both are silent.
let dir = if a > b {
"ours WIDER — bites when OUR file goes"
} else {
"ours NARROWER — bites when THEIRS arrives"
};
println!("{name:<34} {ours:<26} {a:>4} {b:>3} {dir} — ref in {home}");
}
// The frame half, grouped the other way — by the STOCK file, because that is the unit of
// the question it answers: "can this stock window be added?"
let mut by_stock: std::collections::BTreeMap<&String, std::collections::BTreeSet<&String>> =
std::collections::BTreeMap::new();
for (_, ours, home) in &frame_hits {
by_stock.entry(home).or_default().insert(ours);
}
println!(
"\n=== {} FRAME-NAME collisions: {} stock windows we do not load already have their \
names declared by a file of ours ===",
frame_hits.len(),
by_stock.len()
);
for (stock, ours) in &by_stock {
let n = frame_hits.iter().filter(|(_, _, h)| h == *stock).count();
let mine: Vec<&str> = ours.iter().map(|s| s.as_str()).collect();
println!(" {n:>3} {stock:<34} vs {}", mine.join(", "));
}
}
/// **A chain `.xml` that does not source its own `.lua` needs TWO manifest lines**, and
/// getting it wrong is silent.
///
/// Most stock windows pull their code in with `<Script file="X.lua"/>`, so naming the `.xml`
/// brings both. A few do not — `TextStatusBar.xml` and `MoneyInputFrame.xml` declare only a
/// template, and the reference's own toc lists their `.lua` on the preceding line (l.32-33,
/// l.11-12). Name the `.xml` alone and the template loads against nothing: every global that
/// file was supposed to define reads nil, every guarded call becomes a no-op, and there is no
/// error anywhere.
///
/// Both were live. `TextStatusBar` was caught by three tests that happened to assert on the
/// numerals; `MoneyInputFrame` was caught only by sweeping for the shape afterwards, and its
/// manifest header had claimed for months to bring "the ten `MoneyInputFrame_*` verbs" while a
/// full-manifest probe answered nil for all of them.
///
/// A gate rather than an instrument, because the answer should always be zero and the failure
/// mode is invisible.
#[test]
fn every_chain_xml_brings_its_own_lua() {
let _data = benilla_formats::wow_data_or_skip!();
let toc = &super::super::addons::Addon::builtin().toc.files;
let listed: std::collections::HashSet<&str> = toc
.iter()
.map(|f| f.rsplit(['\\', '/']).next().unwrap_or(f))
.collect();
let mut orphans = Vec::new();
for entry in toc.iter().filter(|f| super::is_chain_entry(f)) {
let leaf = entry.rsplit(['\\', '/']).next().unwrap_or(entry);
let Some(stem) = leaf.strip_suffix(".xml") else {
continue;
};
let lua = format!("{stem}.lua");
// No sibling in the archive means there is nothing to miss.
if super::read(&format!("Interface/FrameXML/{lua}")).is_none() {
continue;
}
let xml = super::read(&entry.replace('\\', "/"))
.unwrap_or_else(|| panic!("{entry}: not in the chain"));
let text = String::from_utf8_lossy(&xml);
let sourced = text
.to_ascii_lowercase()
.contains(&format!("file=\"{}\"", lua.to_ascii_lowercase()));
if !sourced && !listed.contains(lua.as_str()) {
orphans.push(format!(
"{leaf} does not source {lua}, and {lua} is not a manifest entry"
));
}
}
assert!(
orphans.is_empty(),
"a chain window whose code never loads — silent, every global it defines reads nil:\n {}",
orphans.join("\n ")
);
}
/// A path is a chain entry; a bare name is ours. The one-line rule the manifest rests on.
crates/benilla-app/src/ui_script/reputation_frame_tests.rs
+2 -1
View File
@@ -87,7 +87,8 @@ fn load_page(s: &UiScript) {
"UiPanels.xml",
"UIParent.xml",
"GameTooltip.xml",
"TextStatusBar.xml",
"Interface\\FrameXML\\TextStatusBar.lua",
"Interface\\FrameXML\\TextStatusBar.xml",
"Cooldown.xml",
"ActionBar.xml",
"ScrollTemplates.xml",
crates/benilla-app/src/ui_script/static_popup_tests.rs
+65
View File
@@ -191,3 +191,68 @@ fn the_death_countdown_text_rerenders_each_tick() {
);
assert!(s.errors().is_empty(), "script errors: {:?}", s.errors());
}
/// **`StaticPopup_Hide`'s and `StaticPopup_FindVisible`'s second argument** (decisions 1793).
///
/// Both took `which` alone until the arity scan found the drift. The reference takes `(which,
/// data)`, and the difference is invisible without a test: Lua discards a surplus argument without
/// complaint, so a stock caller's `StaticPopup_Hide("CONFIRM_BATTLEFIELD_ENTRY", i)` — the shape
/// `BattlefieldFrame.lua:120` uses — would have hidden every instance of that dialog instead of
/// the one it named, and said nothing.
///
/// The `multiple` gate is the reference's own and is why `data` is not a plain equality: a kind
/// declared `multiple` can have several instances up at once and is matched by data; every other
/// kind has at most one, so its data is never consulted.
#[test]
fn hide_and_find_address_one_instance_by_data_only_for_a_multiple_dialog() {
let s = setup();
s.run(
r#"
StaticPopupDialogs["T_MULTI"] = { text = "multi", button1 = "Ok", multiple = 1, timeout = 0 }
StaticPopupDialogs["T_ONE"] = { text = "one", button1 = "Ok", timeout = 0 }
a = StaticPopup_Show("T_MULTI") a.data = "alpha"
b = StaticPopup_Show("T_MULTI") b.data = "beta"
c = StaticPopup_Show("T_ONE") c.data = "gamma"
"#,
)
.unwrap();
assert!(s.errors().is_empty(), "{:?}", s.errors());
// A `multiple` kind is addressed by data — both ways round, so this cannot pass by finding
// whichever instance happens to be first.
for (data, want) in [("alpha", "a"), ("beta", "b")] {
assert!(
s.eval::<bool>(&format!(
"return StaticPopup_FindVisible(\"T_MULTI\", {data:?}) == {want}"
))
.unwrap(),
"FindVisible picks the instance carrying {data:?}"
);
}
// …and a data nothing carries finds nothing, rather than the first instance.
assert!(s
.eval::<bool>(r#"return StaticPopup_FindVisible("T_MULTI", "nobody") == nil"#)
.unwrap());
// Hiding by data takes ONE instance and leaves its sibling up.
s.run(r#"StaticPopup_Hide("T_MULTI", "alpha")"#).unwrap();
assert!(s
.eval::<bool>("return not a:IsShown() and b:IsShown()")
.unwrap());
// No data hides every instance of the kind — which is what every one-argument caller means.
s.run(r#"StaticPopup_Hide("T_MULTI")"#).unwrap();
assert!(s.eval::<bool>("return not b:IsShown()").unwrap());
// A NON-multiple kind ignores data entirely: the reference gates the comparison on
// `info.multiple`, so a wrong data still finds it.
assert!(s
.eval::<bool>(r#"return StaticPopup_FindVisible("T_ONE", "wrong") == c"#)
.unwrap());
// An unknown kind answers nil rather than scanning — the reference's own early out.
assert!(s
.eval::<bool>(r#"return StaticPopup_FindVisible("T_NOPE") == nil"#)
.unwrap());
assert!(s.errors().is_empty(), "{:?}", s.errors());
}
crates/benilla-app/src/ui_script/unit_frame_tests.rs
+13 -5
View File
@@ -17,7 +17,8 @@ fn load_unit_frames(s: &UiScript) {
load_xml(s, "UIParent.xml");
// The bars' numerals machinery (decision 1082), which the manifest loads immediately ahead of
// UnitFrames.xml and which every bar's OnLoad wires into since 1143.
load_xml(s, "TextStatusBar.xml");
load_xml(s, "Interface\\FrameXML\\TextStatusBar.lua");
load_xml(s, "Interface\\FrameXML\\TextStatusBar.xml");
load_xml(s, "GameTooltip.xml");
load_xml(s, "UIDropDownMenu.xml");
load_xml(s, "UnitPopup.xml");
@@ -1516,8 +1517,13 @@ fn status_bar_text_paints_the_player_numerals_but_not_the_targets() {
.unwrap();
assert!(shown(&s, "TargetFrameHealthBarText"));
assert_eq!(text(&s, "TargetFrameHealthBarText"), "50 / 100");
s.run("BenillaUnitFrameBar_OnLeave(TargetFrameHealthBar)")
.unwrap();
// `this` bound, as the engine binds it for a bar's own `<OnLeave>`: the stock
// `HideTextStatusBarText` reads `this.isZero` on one line where every other reads `bar`
// (TextStatusBar.lua:95 — the reference's own slip; a nil index only from bare Lua).
s.run(
"this = TargetFrameHealthBar BenillaUnitFrameBar_OnLeave(TargetFrameHealthBar) this = nil",
)
.unwrap();
assert!(
!shown(&s, "TargetFrameHealthBarText"),
"and go away again — the lockShow refcount balances"
@@ -1535,8 +1541,10 @@ fn status_bar_text_paints_the_player_numerals_but_not_the_targets() {
shown(&s, "PlayerFrameHealthBarText"),
"hover shows them even with the option off"
);
s.run("BenillaUnitFrameBar_OnLeave(PlayerFrameHealthBar)")
.unwrap();
s.run(
"this = PlayerFrameHealthBar BenillaUnitFrameBar_OnLeave(PlayerFrameHealthBar) this = nil",
)
.unwrap();
assert!(!shown(&s, "PlayerFrameHealthBarText"));
s.run("SetCVar(\"statusBarText\", \"1\", \"STATUS_BAR_TEXT\")")
.unwrap();
crates/benilla-formats/src/chr_classes.rs
+1 -1
View File
@@ -49,7 +49,7 @@
//! (`0x54240e`/`0x542446`), so the patch copy is the only one the client can read. Read the base
//! archive alone — or an early-vanilla 16-column struct — and the flag simply is not there, which
//! is how "the relic slot post-dates 1.12" became a settled belief in this codebase and stayed one
//! across two decision records. It is false; see decision 1785. [`CHR_CLASSES_FIELDS`] is what
//! across two decision records. It is false; see decision 1796. [`CHR_CLASSES_FIELDS`] is what
//! keeps us on the patch copy, and it is load-bearing, not defensive.
//!
//! The flag is engine-enforced, not a UI conceit: `IsValidForSlot 0x5da1d0`'s `slot == 0x11` leg
crates/benilla-protocol/src/messages/opcode.rs
+1 -1
View File
@@ -782,7 +782,7 @@ pub const SMSG_SELL_ITEM: u16 = 0x01A1; // 417
pub const CMSG_BUY_ITEM: u16 = 0x01A2; // 418
/// Buy into a **specific** bag slot. This is what a vendor row dropped out of the merchant
/// cursor sends (`PickupMerchantItem` mode 5 → the container/doll drop), as against
/// `CMSG_BUY_ITEM`'s auto-place, which is what clicking the row sends. Both ship (decision 1786).
/// `CMSG_BUY_ITEM`'s auto-place, which is what clicking the row sends. Both ship (decision 1797).
pub const CMSG_BUY_ITEM_IN_SLOT: u16 = 0x01A3; // 419
pub const SMSG_BUY_ITEM: u16 = 0x01A4; // 420
pub const SMSG_BUY_FAILED: u16 = 0x01A5; // 421
crates/benilla-ui/src/script/char_stats.rs
+1 -1
View File
@@ -1115,7 +1115,7 @@ pub(super) fn install(lua: &Lua) -> mlua::Result<()> {
// a property of the SLOT, not of the player. It pairs with `UnitHasRelicSlot("player")`,
// which is the class half, and stock reads the two as a conjunction
// (`PaperDollFrame.lua:680`/`:744`). This comment used to say `UnitHasRelicSlot` is a
// later-era concept, always false in vanilla; it is not — decision 1785.
// later-era concept, always false in vanilla; it is not — decision 1796.
g.set(
"GetInventorySlotInfo",
lua.create_function(|lua, name: String| {
crates/benilla-ui/src/script/mod.rs
+1
View File
@@ -101,6 +101,7 @@ mod model;
mod modelframe;
mod net_stats;
mod object;
pub use object::frame_kind_from_tag;
mod party;
mod pet;
mod petition;
crates/benilla-ui/src/script/object.rs
+9 -1
View File
@@ -192,12 +192,20 @@ pub(super) fn draw_layer_name(l: DrawLayer) -> &'static str {
}
}
/// A widget tag or `CreateFrame` type string → its [`FrameKind`], case- and separator-insensitive
/// ([`enum_token`]). Public because it is the ONE mapping: `benilla-app`'s `frame_flag_gate` sweep
/// needs it to ask the engine whether a tag is mouse-enabled by construction, and a second copy
/// there is exactly what drifted.
pub fn frame_kind_from_tag(s: &str) -> Option<FrameKind> {
frame_kind_from_str(s)
}
fn frame_kind_from_str(s: &str) -> Option<FrameKind> {
Some(match enum_token(s).as_str() {
"FRAME" => FrameKind::Frame,
"BUTTON" => FrameKind::Button,
"CHECKBUTTON" => FrameKind::CheckButton,
// A real registered type (`0x4959a6`), not an alias for Button — decision 1788. Registering
// A real registered type (`0x4959a6`), not an alias for Button — decision 1799. Registering
// it AS a Button would load `LootFrame.xml` and leave every row dead, because 1.12 has no
// Lua verb for "take slot N": `LootSlot` is the bind-confirm continuation, hard-wired to
// `flag = 1`.
crates/benilla-ui/src/script/pointer.rs
+53 -3
View File
@@ -50,7 +50,7 @@ use crate::order;
use crate::widget::FrameHandle;
use super::clip::{effective_clip, scroll_clip_sources};
use super::{button, cursor, editbox, event, UiScript};
use super::{button, cursor, editbox, event, Model, UiScript};
/// The double-click interval — **300 ms, VERIFIED off the bytes**, a hardcoded instruction
/// immediate: `0x77937b 81 f9 2c 01 00 00 cmp ecx, 0x12c`, comparing `now − [this+0x334]` inside
@@ -127,6 +127,15 @@ pub(super) fn install(lua: &mlua::Lua) -> mlua::Result<()> {
Ok(())
}
/// Whether `(x, y)` falls inside one of this frame's hyperlink spans — the engine half of the
/// reference's per-span `CSimpleHyperlinkButton` children.
fn link_span_hit(model: &Model, fh: FrameHandle, x: f32, y: f32) -> bool {
model
.link_spans
.get(&fh)
.is_some_and(|spans| spans.iter().any(|(r, _, _)| point_in_rect(*r, x, y)))
}
impl UiScript {
/// Hit-test the cursor at `(x, y)` and return the **handle** of the captured frame (the
/// topmost-drawn mouse-enabled, effective-visible frame whose rect contains the point **and**
@@ -146,7 +155,22 @@ impl UiScript {
let sorted = order::traversal(&model.arena);
let scroll_sources = scroll_clip_sources(&model);
order::hit_test(&sorted, |fh| {
model.arena.is_mouse_enabled(fh)
// Mouse-enabled, **or inside one of this frame's hyperlink spans**.
//
// That second half is how the reference makes a chat link clickable on a frame that
// takes no mouse at all. `CSimpleMessageScrollFrame`'s ctor leaves `[+0xcc] = 0`, so
// `ChatFrame1` is never in the hit-test index — instead the engine synthesises one
// mouse-enabled `CSimpleHyperlinkButton` child PER SPAN (`0x7a3240`, born through the
// `CSimpleButton` ctor) and routes the three hyperlink scripts back to the parent
// (wow-re `mouse-enable-law.md` + the follow-up round `68987021`).
//
// We hold the spans as rects on the parent rather than as child frames, so the same
// law expresses as a disjunct here. What it buys is everything the reference gets from
// the split: the chat window is a hit candidate only over a LINK, so `GetMouseFocus()`
// answers what is behind it elsewhere, its `movable="true"` +
// `<TitleRegion setAllPoints="true"/>` stay inert so it drags by its tab and not its
// body, and a hover over plain text reassigns to whatever is underneath.
(model.arena.is_mouse_enabled(fh) || link_span_hit(&model, fh, x, y))
&& model.resolved.get(&fh).is_some_and(|r| {
point_in_rect(inset_rect(*r, model.arena.hit_rect_insets(fh)), x, y)
})
@@ -155,6 +179,26 @@ impl UiScript {
})
}
/// [`Self::hit_test`] over the frames that take the WHEEL — either bit, since a mouse-enabled
/// frame with an `<OnMouseWheel>` is a candidate for both. Same rect and clip rules; only the
/// enable predicate differs.
fn hit_test_wheel(&self, x: f32, y: f32) -> Option<u32> {
let model = self.model_ref();
let sorted = order::traversal(&model.arena);
let scroll_sources = scroll_clip_sources(&model);
let fh = order::hit_test(&sorted, |fh| {
(model.arena.is_mouse_wheel_enabled(fh)
|| model.arena.is_mouse_enabled(fh)
|| link_span_hit(&model, fh, x, y))
&& model.resolved.get(&fh).is_some_and(|r| {
point_in_rect(inset_rect(*r, model.arena.hit_rect_insets(fh)), x, y)
})
&& effective_clip(&model, &scroll_sources, fh)
.is_none_or(|c| point_in_rect(c, x, y))
})?;
model.frame_to_id.get(&fh).copied()
}
/// Hit-test the cursor at `(x, y)` and return the **id** of the captured frame, or `None` —
/// [`Self::hit_test_frame`] with the id lookup every caller in the input path wants. The id is
/// the same one used across this host (e.g. the layout [`crate::layout::Handle`]); the app can
@@ -749,7 +793,13 @@ impl UiScript {
/// [`UiScript::errors`].
pub fn mouse_wheel(&mut self, x: f32, y: f32, delta: f32) {
let target: Option<u32> = {
let hit = self.hit_test(x, y);
// **Wheel-enabled OR mouse-enabled**, not mouse-enabled alone. The reference's wheel
// dispatcher `0x7664f0` walks its OWN index and calls the hit probe directly, ignoring
// hover, capture and the mouse bit — `EnableMouse` is not a prerequisite
// (wow-re `mouse-enable-law.md` §6). Ours went through the mouse hit test, which
// coupled the two: correcting the ctor list to the reference's (ScrollFrame is inert
// on the mouse bit) silently stopped every scroll pane from taking the wheel.
let hit = self.hit_test_wheel(x, y);
let model = self.model_ref();
hit.and_then(|id| {
let mut h = *model.id_to_frame.get(&id)?;
crates/benilla-ui/src/script/unit/bindings.rs
+1 -1
View File
@@ -599,7 +599,7 @@ pub(in crate::script) fn install(lua: &Lua) -> mlua::Result<()> {
// **This shipped absent for a long time, on a claim that was simply false** — that the relic
// slot post-dates 1.12 and `UnitHasRelicSlot` is a later-era verb. The base `dbc.MPQ` copy of
// `ChrClasses.dbc` really does have no field 16; the `patch.MPQ` copy the client actually
// reads has it, set for Paladin, Shaman and Druid. Decision 1785. Stock `PaperDollFrame.lua`
// reads has it, set for Paladin, Shaman and Druid. Decision 1796. Stock `PaperDollFrame.lua`
// calls this **unconditionally** at l.429 and l.580 — so while it was missing, the character
// sheet raised `attempt to call global 'UnitHasRelicSlot' (a nil value)` for *every* class,
// not just the three.
crates/benilla-ui/src/script/unit/mod.rs
+1 -1
View File
@@ -173,7 +173,7 @@ pub struct UnitState {
/// which is the nil leg.
///
/// True for exactly Paladin, Shaman and Druid in 1.12 (Libram, Totem, Idol). It was believed
/// here for a long time that no 1.12 class had one; see decision 1785 for how that survived.
/// here for a long time that no 1.12 class had one; see decision 1796 for how that survived.
pub has_relic_slot: bool,
/// The unit's sex on the `UnitSex` scale: `2` male, `3` female (`1` = neuter/unknown). `0`
/// (the unfilled default) reports as nil — the API's "can't tell", like [`Self::reaction`].
crates/benilla-ui/src/script/unit/tests.rs
+1 -1
View File
@@ -1295,7 +1295,7 @@ fn unit_is_party_leader_ors_two_legs_and_answers_one_when_solo() {
/// `UnitHasRelicSlot` — the number 1 or nil, per token, never a boolean.
///
/// This shipped **absent** for months on the belief that the relic slot post-dates 1.12, which is
/// false (decision 1785). Stock `PaperDollFrame.lua` calls it unconditionally at l.429 and l.580,
/// false (decision 1796). Stock `PaperDollFrame.lua` calls it unconditionally at l.429 and l.580,
/// so while it was missing the character sheet raised `attempt to call global` for every class —
/// which is why the nil-global case is asserted here too, not just the answer.
#[test]
crates/benilla-ui/src/widget/kinds/mod.rs
+1 -1
View File
@@ -48,7 +48,7 @@ pub enum FrameKind {
///
/// Its whole Lua surface of its own is **one method, `SetSlot(index)`** (table `0x847ce4`,
/// count 1 read off the registrar's own `mov edx,1`) — 1-based in, 0-based stored, no returns.
/// `LootFrame.lua:94` is its only caller. Decision 1788; wow-re `e5338caf`.
/// `LootFrame.lua:94` is its only caller. Decision 1799; wow-re `e5338caf`.
LootButton,
CheckButton,
EditBox,
crates/benilla-ui/src/widget/mod.rs
+42 -21
View File
@@ -447,6 +447,47 @@ impl Default for WidgetArena {
}
}
/// Whether a widget of this kind is **mouse-enabled by its constructor**, needing neither an
/// `enableMouse` attribute nor an auto-enabling `<Scripts>` handler to be clickable.
///
/// One list, because there are two readers: this arena, and `benilla-app`'s `frame_flag_gate`
/// sweep, which compares our frames against a model of the reference's XML. That sweep used to
/// carry its own copy with a comment claiming it was "deliberately the SAME list" — it was not,
/// and it drifted silently the first time this one was edited. A wrong entry should be wrong in
/// one place.
///
/// **Read off the client's ctors** (wow-re `ui/scratch/mouse-enable-law.md` `774f7eb6`, and the
/// follow-up `68987021`). `[frame+0xcc]` is a four-bit input mask — char, keyboard, MOUSE, wheel —
/// zeroed by the base `CSimpleFrame` ctor `0x769090` and set by exactly five subclass ctors:
/// Button `0x778771`, CheckButton (through Button), Slider `0x789467`, ColorSelect `0x78b2b5`,
/// and EditBox `0x779ced`, which takes char+keyboard+mouse together because it must take a click
/// to focus. `Minimap` is an engine frame born `0x4` (`0x4edc38`), and ours is a `FrameKind`.
///
/// **`ScrollFrame` and `ScrollingMessageFrame` are NOT here, and that was the hard one.** They
/// were, on the reasoning that a scroll frame "takes the wheel in its ctor" — the wheel is a
/// different bit, and no ctor writes it at all. Against that stood a recorded observation:
/// `ChatFrame1` is a `ScrollingMessageFrame`, and something has to make a 1.12 chat link
/// clickable. The follow-up round settled it: nothing about the frame does. The engine synthesises
/// one mouse-enabled `CSimpleHyperlinkButton` child PER LINK SPAN (`0x7a3240`, born through the
/// `CSimpleButton` ctor) and routes the hyperlink scripts back to the parent. We hold spans as
/// rects rather than child frames, so `script::pointer`'s hit test carries the same law as a
/// disjunct: the chat window is a candidate over a link and inert everywhere else.
pub fn mouse_enabled_by_ctor(kind: FrameKind) -> bool {
matches!(
kind,
FrameKind::Button
| FrameKind::CheckButton
| FrameKind::EditBox
// A Slider's thumb must be draggable: every scrollbar is a UIPanelScrollBarTemplate
// Slider that declares no `enableMouse` yet is draggable in-game (decision 0250).
| FrameKind::Slider
| FrameKind::ColorSelect
| FrameKind::Minimap
// `LootButton` extends `CSimpleButton` and inherits its ctor's bit (decision 1799).
| FrameKind::LootButton
)
}
impl WidgetArena {
/// A new, empty arena in **1.12 mode** (alpha does not propagate).
pub fn new() -> WidgetArena {
@@ -602,27 +643,7 @@ impl WidgetArena {
effective_alpha,
shown,
effective_visible: shown && parent_visible,
// A button/editbox is mouse-enabled by construction (the client's CSimpleButton and
// CSimpleEditBox enable input in their ctors — an XML button needs no enableMouse attr
// to be clickable, and an editbox must take clicks to focus, RF-0082 §1).
mouse_enabled: matches!(
kind,
FrameKind::Button
| FrameKind::CheckButton
| FrameKind::EditBox
// The scroll frame takes the wheel in its ctor (msgframe-runtime.md) — an XML
// ScrollingMessageFrame is wheel-interactive with no enableMouse attr, exactly
// like a Button is clickable without one.
| FrameKind::ScrollingMessageFrame
// A plain ScrollFrame likewise takes the wheel/drag by construction (decision
// 0112) — an XML ScrollFrame is scroll-interactive with no enableMouse attr,
// the same rationale as the ScrollingMessageFrame above.
| FrameKind::ScrollFrame
// A Slider takes the mouse in its ctor too (the thumb must be draggable): every
// scrollbar is a UIPanelScrollBarTemplate Slider that declares no enableMouse yet
// is draggable in-game (decision 0250), so the enablement is by construction.
| FrameKind::Slider
),
mouse_enabled: mouse_enabled_by_ctor(kind),
// The two kinds whose ctor takes the WHEEL rather than the mouse generally — the same
// by-construction argument as above, narrowed to the flag it is actually about
// (decision 1198): a ScrollingMessageFrame and a ScrollFrame are wheel-interactive