Add a test suite, and fix the three dead-data bugs it found

There was no test file -- CLAUDE.md described the harness in prose and
referred to an "orphan-tag check" that did not exist. test.lua now stubs
the game globals and exercises the real engine, database and catalog tab
builders. It found three bugs on its first run:

* four_kind was the only theme in the catalog that could never be
  discovered. Theme discovery needs a giver and a wanter on two DIFFERENT
  fielded jokers, and The Family was its sole member -- so the Hands bar
  and the overall Progress bar could never reach 100%. DNA is the real
  enabler (copying a card is how you stack duplicate ranks, the only
  vanilla route to a reliable Four of a Kind), so it now gives the
  matching-hand tags as a pure enabler.

* hand_level was given by Space Joker and Burnt Joker and wanted by
  nobody, so it could never contribute to a score and its "want" tooltip
  text was unreachable. Nothing in vanilla rewards hand levels; the tag
  is dropped rather than faked.

* Invisible Joker had no tags at all, so it scored 0 against the entire
  game. Selling it duplicates one of your Jokers -- the same "stronger
  target, better payoff" logic Blueprint and Brainstorm already model --
  so it now wants copy_target.

The suite guards both directions of the tag contract (no orphan wants, no
dead gives), theme reachability, blurb widths, the inert-joker list, and
the card materialize fix from the previous commit.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
funman300
2026-07-14 11:15:10 -07:00
parent f8bba13f28
commit 07cada9e49
4 changed files with 447 additions and 36 deletions
+41 -29
View File
@@ -10,38 +10,42 @@ and pulses strongly recommended cards. Pure Lua, no build step.
## Commands
Syntax check everything:
Syntax check, then run the test suite — **always both, after every edit**:
```sh
luac -p main.lua synergies.lua config.lua
luac -p main.lua synergies.lua config.lua test.lua
lua test.lua
```
There is no test framework. Test the engine standalone by stubbing the game globals
before loading `main.lua` (the game itself cannot be launched headlessly here):
`test.lua` stubs the handful of game globals `main.lua` touches (the game cannot be
launched headlessly here) and then exercises the real engine, the real database and
the real catalog tab builders. It is the only safety net; there is no other harness.
It enforces, among other things:
```sh
lua - <<'EOF'
SMODS = { load_file = function(f) return loadfile(f) end,
current_mod = { config = dofile('config.lua') } }
Card = { generate_UIBox_ability_table = function() end, update = function() end }
G = { UIT = {ROOT='ROOT', R='R', T='T'},
C = { UI = {TEXT_DARK='dark'}, GREEN='green', CLEAR='clear' },
STAGES = {RUN=1}, FUNCS = {} }
dofile('main.lua')
print(JCA.score('j_baron', 'j_mime')) -- pairwise score
-- simulate owned jokers via G.jokers.cards; each card needs
-- { config = { center = { key=..., name=..., set='Joker' } } }
EOF
```
- Score tiers: famous explicit pair + tags = 6, hand-type/suit cluster = 4, single
tag match = 2, generic mult+xmult complement = +1 on top, unrelated = 0.
- **No orphan wants** — a `wants` tag nothing gives is a payoff with no enabler.
- **No dead gives** — a non-engine tag given but never wanted can never change a
score, so it is invisible dead weight. (This caught `hand_level`, given by Space
and Burnt Joker and wanted by nobody.)
- **Every theme is discoverable** — theme discovery needs a giver and a wanter on two
*different* fielded jokers, so a theme with fewer than two members is an
achievement no player can ever earn and pins the Progress tab under 100% forever.
(This caught `four_kind`, whose only member was The Family.)
- `theme_info` and `THEME_TABS` cover exactly the same tag set.
- Pair/clash/caution keys all exist; every blurb fits its ~34-char row.
- Every card the catalog pages emplace is materialized (see the UI notes below).
- Jokers with no tags at all are listed in `EXPECTED_INERT` on purpose — a joker
landing there by accident scores 0 against the entire game and says nothing.
Expected score tiers: famous explicit pair + tags ≈ 6, hand-type/suit cluster = 4,
single tag match = 2, generic mult+xmult complement = 1, unrelated = 0.
To poke at the engine by hand, copy the stub block from the top of `test.lua`.
## In-game verification (this machine)
- The mod is **symlinked** into the live Mods folder:
`~/.local/share/Balatro/Mods/JokerComboAdvisor -> ~/Documents/Balatro Mod`.
Edits here go live on next game launch — do not copy files anywhere.
`~/.local/share/Balatro/Mods/JokerComboAdvisor -> ~/Documents/JokerComboAdvisor`
(this repo). Edits here go live on next game launch — do not copy files anywhere.
Lovely reads that folder as `C:\users\steamuser\AppData\Roaming\Balatro\Mods`.
- Game: Steam/Proton at `/mnt/games/Steam/steamapps/common/Balatro` (note: second
Steam library, not the default paths). Lovely injector + Steamodded 1.0.0-beta
are already installed; mods are managed via Balatro Mod Manager.
@@ -96,10 +100,16 @@ Three-layer split; `JCA` is the single global namespace:
per page; pagers swap the `tab_contents` UIBox from an option-cycle
callback — the same pattern as SMODS's achievements tab. Because pages
contain live Card/CardArea objects, tab definitions must be rebuilt on
every call — never cache the node trees. The same tabs open in-run via
a "Combos" HUD button (wrapped `create_UIBox_HUD`, injected into the
`button_area` node under Run Info/Options); the pagers work there
because vanilla `create_tabs` also names its body `tab_contents`.
every call — never cache the node trees. Cards must be `start_materialize`d
as they are emplaced (`card_row_node`), exactly as every vanilla collection
page does — that call *is* the appear animation, and without it cards pop
in with no animation on tab open and page switch. Only the first card of a
page build plays the sound (vanilla's `i>1 or j>1` silence flag); tab defs
call `begin_page()` so the counter spans the page, not one row. The same
tabs open in-run via a "Combos" HUD button (wrapped `create_UIBox_HUD`,
injected into the `button_area` node under Run Info/Options); the pagers
work there because vanilla `create_tabs` also names its body
`tab_contents`.
- Discovery: `CardArea:emplace` hook calls `JCA.check_discoveries` when a
joker lands in `G.jokers`; famous pairs fielded together persist in
`config.discovered` (saved immediately via `SMODS.save_mod_config`) and
@@ -132,9 +142,11 @@ Three-layer split; `JCA` is the single global namespace:
- **Cluster tags** (hand types `pair`/`straight`/`flush`/…, suits, `planet`,
`spectral_gen`): put the same tag in *both* `gives` and `wants` of every member
so members mutually reinforce (2+2 = 4). Pure enablers (e.g. Four Fingers,
Smeared, Pareidolia) only `give`; pure payoffs only `want`.
- Every `wants` tag must be given by at least one joker — the standalone test's
orphan-tag check enforces this idea; re-run it after database edits.
Smeared, Pareidolia, DNA) only `give`; pure payoffs only `want`.
- A tag needs **both sides to exist, on at least two different jokers**: something
must give it, something must want it, or it is dead data that can never change a
score — and if it is also a theme, an achievement nobody can earn. `lua test.lua`
enforces both directions; run it after every `synergies.lua` edit.
- Tooltips teach: `JCA.explain(hovered, partner)` picks one reason per partner
— famous-pair blurb (third element of each `PAIRS` entry), else the first
`TAG_ORDER` match phrased from `TAG_TEXT` (give/want/both variants), else
+1 -1
View File
@@ -267,7 +267,7 @@ local TAG_LABEL = {
faces = 'face cards', retrig_scoring = 'played-card retriggers',
retrig_held = 'held-card retriggers', hand_size = 'hand size',
hands_up = 'extra hands', discards_up = 'discards',
probability = 'listed odds', money = 'money', hand_level = 'hand levels',
probability = 'listed odds', money = 'money',
tarot_gen = 'Tarot cards', planet = 'Planet cards',
spectral_gen = 'Spectral cards', gold_gen = 'Gold cards',
stone_gen = 'Stone cards', enhance_gen = 'card enhancing',
+10 -6
View File
@@ -114,7 +114,7 @@ j('j_fibonacci', {'mult'}, {'retrig_scoring'})
j('j_steel_joker', {'xmult'})
j('j_hack', {'retrig_scoring'})
j('j_pareidolia', {'faces'})
j('j_space', {'hand_level'}, {'probability'})
j('j_space', {}, {'probability'})
j('j_burglar', {'hands_up'})
j('j_blackboard', {'xmult'}, {'spades','clubs'})
j('j_sixth_sense', {'spectral_gen'}, {'spectral_gen'})
@@ -127,9 +127,12 @@ j('j_vampire', {'xmult'}, {'enhance_gen'})
j('j_hologram', {'xmult','copy_target'}, {'card_gen'})
j('j_cartomancer', {'tarot_gen'})
j('j_astronomer', {'planet'}, {'planet'})
j('j_burnt', {'hand_level'}, {'discards_up'})
j('j_burnt', {}, {'discards_up'})
j('j_bootstraps', {'mult'}, {'money'})
j('j_dna', {'card_gen'})
-- DNA copies a played card into the deck, which is how you stack duplicate
-- ranks -- the only vanilla route to a reliable Four of a Kind. Pure enabler:
-- it gives the matching-hand tags without wanting them.
j('j_dna', {'card_gen','pair','three_kind','four_kind'})
j('j_midas_mask', {'gold_gen','enhance_gen'}, {'faces'})
j('j_photograph', {'xmult','copy_target'}, {'faces','retrig_scoring'})
j('j_stone', {'chips'}, {'stone_gen'})
@@ -176,7 +179,9 @@ j('j_family', {'xmult','four_kind'}, {'four_kind'})
j('j_order', {'xmult','straight'}, {'straight'})
j('j_tribe', {'xmult','flush'}, {'flush'})
j('j_stuntman', {'chips'})
j('j_invisible', {})
-- Sell Invisible Joker to duplicate one of your Jokers: the same "the stronger
-- the target, the better" logic as Blueprint and Brainstorm.
j('j_invisible', {}, {'copy_target'})
j('j_blueprint', {}, {'copy_target'})
j('j_brainstorm', {}, {'copy_target'})
j('j_drivers_license', {'xmult'}, {'enhance_gen'})
@@ -258,7 +263,6 @@ local TAG_TEXT = {
discards_up = {give = 'grants extra discards', want = 'feeds on discards'},
probability = {give = 'improves listed odds', want = 'relies on listed odds'},
money = {give = 'earns money', want = 'scales with your money'},
hand_level = {give = 'levels up poker hands', want = 'rewards hand levels'},
tarot_gen = {give = 'creates Tarot cards', want = 'rewards Tarot use'},
planet = {both = 'shares the Planet economy'},
spectral_gen = {both = 'shares the Spectral theme'},
@@ -288,7 +292,7 @@ local TAG_ORDER = {
'retrig_held', 'retrig_scoring', 'faces',
'gold_gen', 'stone_gen', 'enhance_gen', 'card_gen',
'tarot_gen', 'spectral_gen', 'planet',
'probability', 'discards_up', 'hands_up', 'hand_size', 'hand_level',
'probability', 'discards_up', 'hands_up', 'hand_size',
'money', 'sell_value', 'joker_gen',
'hearts', 'diamonds', 'spades', 'clubs',
'pair', 'two_pair', 'three_kind', 'four_kind', 'straight', 'flush',
+395
View File
@@ -0,0 +1,395 @@
-- Joker Combo Advisor - standalone test suite.
--
-- lua test.lua
--
-- The game cannot be launched headlessly, so this stubs the handful of Balatro
-- globals main.lua touches and then exercises the real engine, the real
-- database, and the real catalog tab builders. It covers three things:
--
-- * ENGINE - scoring tiers, partner aggregation, the sell advisor.
-- * DATA - the invariants that keep synergies.lua honest. A tag that
-- nothing wants, or a theme too small to ever be discovered, is
-- silent dead weight in the game; only a check catches it.
-- * UI - the catalog pages build, and every card they place animates.
--
-- Run it after every synergies.lua edit.
--------------------------------------------------------------------------------
-- Test harness
--------------------------------------------------------------------------------
local pass, fail = 0, 0
local function ok(cond, what, detail)
if cond then
pass = pass + 1
else
fail = fail + 1
print((' FAIL %s%s'):format(what, detail and ('\n ' .. detail) or ''))
end
return cond
end
local function eq(got, want, what)
return ok(got == want, what, ('expected %s, got %s'):format(tostring(want), tostring(got)))
end
local function section(name) print('\n== ' .. name .. ' ==') end
-- Sorted key list, so failure messages are stable and diffable.
local function sorted_keys(t)
local keys = {}
for k in pairs(t) do keys[#keys + 1] = k end
table.sort(keys)
return keys
end
--------------------------------------------------------------------------------
-- Game global stubs (only what main.lua actually reaches for)
--------------------------------------------------------------------------------
-- G.C is full of colour tables; auto-vivify so any colour lookup works.
local function auto()
return setmetatable({}, {__index = function(t, k)
local v = setmetatable({}, getmetatable(t))
rawset(t, k, v)
return v
end})
end
SMODS = {
load_file = function(f) return loadfile(f) end,
current_mod = {config = dofile('config.lua')},
save_mod_config = function() end,
}
G = {
UIT = setmetatable({}, {__index = function(_, k) return k end}),
C = auto(),
STAGES = {RUN = 1},
STAGE = 1,
FUNCS = {},
ROOM = {T = {x = 0, y = 0, w = 20, h = 12}},
CARD_W = 1, CARD_H = 1.4,
P_CARDS = {empty = {}},
P_CENTERS = {},
SETTINGS = {profile = 1},
PROFILES = {{all_unlocked = false}},
GAME = {},
}
-- Cards the catalog pages build, so the UI section can inspect them.
local emplaced, materialized, sounds = {}, {}, 0
CardArea = function(x, y, w, h, cfg)
return {
T = {x = x, y = y, w = w, h = h}, config = cfg, cards = {},
emplace = function(self, c)
self.cards[#self.cards + 1] = c
emplaced[#emplaced + 1] = c
end,
}
end
Card = setmetatable({
generate_UIBox_ability_table = function() end,
update = function() end,
}, {__call = function(_, x, y, w, h, front, center)
local card
card = {
T = {x = x, y = y, w = w, h = h},
config = {center = center},
ability = {set = 'Joker'},
children = {}, states = {hover = {}},
juice_up = function() end,
start_materialize = function(_, _, silent)
materialized[card] = true
if not silent then sounds = sounds + 1 end
end,
}
return card
end})
UIBox = function(args) return {def = args and args.definition, remove = function() end} end
create_option_cycle = function() return {n = 'R', config = {}, nodes = {}} end
localize = function() error('no localization outside the game') end
local data = dofile('synergies.lua')
dofile('main.lua')
-- Every joker the database knows needs a center for the UI to render it.
for key in pairs(JCA.db) do
G.P_CENTERS[key] = {key = key, name = key, set = 'Joker', rarity = 1}
end
-- Engine tags carry no teaching text and are deliberately never "wanted" --
-- the generic mult/xmult complement handles them instead.
local ENGINE_TAGS = {mult = true, chips = true, xmult = true}
-- Put jokers on the board, as G.jokers.cards holding real center references.
local function field(...)
local cards = {}
for _, key in ipairs({...}) do
cards[#cards + 1] = {config = {center = {key = key, name = key, set = 'Joker'}}}
end
G.jokers = {cards = cards, config = {card_limit = #cards}}
return cards
end
--------------------------------------------------------------------------------
section('Engine: scoring tiers')
--------------------------------------------------------------------------------
eq(JCA.score('j_baron', 'j_mime'), 6, 'famous pair + shared tags = 6')
eq(JCA.score('j_blueprint', 'j_brainstorm'), 4, 'famous pair alone = 4')
eq(JCA.score('j_jolly', 'j_sly'), 4, 'hand-type cluster (2 + 2) = 4')
-- Hack gives retrig_scoring, which The Idol wants; neither side triggers the
-- generic complement (Hack adds no Mult or Chips), so this is a clean 2.
eq(JCA.score('j_hack', 'j_idol'), 2, 'single tag match = 2')
eq(JCA.score('j_8_ball', 'j_scary_face'), 0, 'unrelated jokers = 0')
-- The generic complement stacks on top of a tag match: Gluttonous gives clubs
-- (Blackboard wants them) AND feeds Blackboard's xMult, so 2 + 1.
eq(JCA.score('j_gluttenous_joker', 'j_blackboard'), 3, 'tag match + generic complement = 3')
eq(JCA.score('j_not_a_real_joker', 'j_baron'), 0, 'unknown key (other mods) scores 0')
eq(JCA.score('j_not_a_real_joker', 'j_also_fake'), 0, 'two unknown keys score 0')
ok(select(2, JCA.score('j_gluttenous_joker', 'j_blackboard')),
'generic +Mult x xMult complement flags itself')
ok(not select(2, JCA.score('j_joker', 'j_stencil')),
'no_generic opts Joker Stencil out of the generic complement')
--------------------------------------------------------------------------------
section('Engine: partners, recommendation, sell advisor')
--------------------------------------------------------------------------------
field('j_mime', 'j_hack')
local partners, total = JCA.partners_for({config = {center = {key = 'j_baron', set = 'Joker'}}})
ok(#partners >= 1, 'Baron finds a partner on a Mime board')
eq(partners[1].key, 'j_mime', 'strongest partner sorts first')
ok(total >= JCA.config.threshold, 'Baron clears the threshold next to Mime')
-- An unrelated board names no partners. The total can still be 1 (Baron's xMult
-- against Scary Face's chips is the generic complement), which is exactly why
-- the threshold is even -- see below.
field('j_8_ball', 'j_scary_face')
local nobody, low = JCA.partners_for({config = {center = {key = 'j_baron', set = 'Joker'}}})
eq(#nobody, 0, 'no partners named on an unrelated board')
ok(low < JCA.config.threshold, 'an unrelated board stays under the threshold')
-- The generic +1 must never be able to trigger a recommendation on its own:
-- tag and pair scores are even and the threshold is even, so a pulsing card
-- always has a nameable partner.
ok(JCA.config.threshold % 2 == 0, 'threshold is even, so the generic +1 alone cannot recommend')
field('j_baron', 'j_mime') -- 2 jokers, board not full
eq(JCA.weakest_link(), nil, 'no sell advice below 3 jokers')
local board = field('j_baron', 'j_mime', 'j_8_ball')
eq(JCA.weakest_link(), board[3], 'flags the joker with the least synergy')
G.jokers.config.card_limit = 5
eq(JCA.weakest_link(), nil, 'no sell advice while slots are free')
field('j_8_ball', 'j_8_ball', 'j_8_ball')
eq(JCA.weakest_link(), nil, 'no sell advice when every joker scores the same')
--------------------------------------------------------------------------------
section('Data: tag integrity')
--------------------------------------------------------------------------------
local givers, wanters, members = {}, {}, {}
for key, e in pairs(JCA.db) do
for t in pairs(e.gives) do
givers[t] = (givers[t] or 0) + 1
members[t] = members[t] or {}
members[t][key] = true
end
for t in pairs(e.wants) do
wanters[t] = (wanters[t] or 0) + 1
members[t] = members[t] or {}
members[t][key] = true
end
end
-- A wanted tag nobody gives is a payoff with no enabler.
for _, t in ipairs(sorted_keys(wanters)) do
ok(givers[t], ('tag %q is wanted (%dx) and given by someone'):format(t, wanters[t]))
end
-- The mirror image, and the one that actually bit us: a tag given but never
-- wanted can never add to a score, so it is invisible dead weight.
for _, t in ipairs(sorted_keys(givers)) do
if not ENGINE_TAGS[t] then
ok(wanters[t], ('tag %q is given (%dx) and wanted by someone'):format(t, givers[t]),
'nothing wants it, so it can never affect a score')
end
end
for _, t in ipairs(sorted_keys(givers)) do
if not ENGINE_TAGS[t] then
ok(data.tag_text[t], ('tag %q has TAG_TEXT for the tooltip'):format(t))
end
end
local in_order = {}
for _, t in ipairs(data.tag_order) do in_order[t] = true end
for _, t in ipairs(sorted_keys(members)) do
if not ENGINE_TAGS[t] then
ok(in_order[t], ('tag %q appears in TAG_ORDER'):format(t))
end
end
for _, t in ipairs(data.tag_order) do
ok(members[t] or ENGINE_TAGS[t], ('TAG_ORDER tag %q is used by at least one joker'):format(t))
end
--------------------------------------------------------------------------------
section('Data: themes are reachable')
--------------------------------------------------------------------------------
-- The catalog tabs and the discovery loop must agree on the theme set, or a
-- theme is either untrackable or unshowable.
for _, t in ipairs(sorted_keys(data.theme_info)) do
ok(JCA.theme_names[t], ('theme %q has a tab entry in THEME_TABS'):format(t))
end
for _, t in ipairs(sorted_keys(JCA.theme_names)) do
ok(data.theme_info[t], ('THEME_TABS tag %q has a theme_info sentence'):format(t))
end
-- Discovery needs a giver and a wanter on two DIFFERENT jokers (main.lua:480).
-- A theme that cannot satisfy that is an achievement no player can ever earn,
-- and it holds the Progress tab below 100% forever.
for _, t in ipairs(sorted_keys(data.theme_info)) do
local n = 0
for _ in pairs(members[t] or {}) do n = n + 1 end
ok(givers[t] and wanters[t] and n >= 2,
('theme %q is discoverable'):format(t),
('givers=%d wanters=%d distinct jokers=%d - needs >=1 giver, >=1 wanter, >=2 jokers')
:format(givers[t] or 0, wanters[t] or 0, n))
end
for _, t in ipairs(sorted_keys(data.theme_info)) do
ok(#data.theme_info[t] <= 90, ('theme %q blurb fits (<=90 chars)'):format(t),
('%d chars'):format(#data.theme_info[t]))
end
--------------------------------------------------------------------------------
section('Data: pairs, clashes and cautions')
--------------------------------------------------------------------------------
local BLURB_MAX = 34 -- name + blurb must fit one tooltip row
for _, p in ipairs(data.pairs) do
local label = ('pair %s|%s'):format(p[1], p[2])
ok(JCA.db[p[1]], label .. ': first key is a known joker')
ok(JCA.db[p[2]], label .. ': second key is a known joker')
ok(p[1] ~= p[2], label .. ': pairs two different jokers')
ok(p[3] and #p[3] > 0, label .. ': has a blurb')
ok(not p[3] or #p[3] <= BLURB_MAX, label .. ': blurb fits one row',
p[3] and ('%d chars: %q'):format(#p[3], p[3]))
end
local seen = {}
for _, p in ipairs(data.pairs) do
local a, b = p[1], p[2]
local k = a < b and (a .. '|' .. b) or (b .. '|' .. a)
ok(not seen[k], ('pair %s is listed only once'):format(k))
seen[k] = true
end
for _, c in ipairs(data.clashes or {}) do
local label = ('clash %s|%s'):format(c[1], c[2])
ok(JCA.db[c[1]] and JCA.db[c[2]], label .. ': both keys are known jokers')
ok(#c[3] <= BLURB_MAX, label .. ': warning fits one row',
('%d chars: %q'):format(#c[3], c[3]))
-- A clash is a trap, not a combo. Listing a pair as both would be incoherent.
local k = c[1] < c[2] and (c[1] .. '|' .. c[2]) or (c[2] .. '|' .. c[1])
ok(not seen[k], label .. ': is not also listed as a famous pair')
end
for _, key in ipairs(sorted_keys(data.cautions or {})) do
ok(JCA.db[key], ('caution %s: is a known joker'):format(key))
ok(#data.cautions[key] <= BLURB_MAX, ('caution %s: fits one row'):format(key),
('%d chars: %q'):format(#data.cautions[key], data.cautions[key]))
end
--------------------------------------------------------------------------------
section('Data: joker coverage')
--------------------------------------------------------------------------------
local n_jokers, inert = 0, {}
for _, key in ipairs(sorted_keys(JCA.db)) do
n_jokers = n_jokers + 1
local e = JCA.db[key]
if not next(e.gives) and not next(e.wants) then inert[#inert + 1] = key end
end
eq(n_jokers, 150, 'all 150 vanilla jokers are in the database')
-- Some jokers really have no combo (Mr. Bones, Credit Card...). That is a
-- legitimate answer, but it should be a deliberate list, not an accident:
-- anything new showing up here scores 0 against the entire game.
local EXPECTED_INERT = {
j_chaos = true, -- free reroll each shop; helps no specific build
j_chicot = true, -- disables the boss blind
j_credit_card = true, -- go negative on money
j_diet_cola = true, -- free Double Tag
j_luchador = true, -- sell to disable the boss blind
j_mr_bones = true, -- prevents death
j_ring_master = true, -- duplicates in the shop pool
}
for _, key in ipairs(inert) do
ok(EXPECTED_INERT[key], ('joker %s is inert on purpose'):format(key),
'it has no gives and no wants, so it scores 0 against every joker')
end
for _, key in ipairs(sorted_keys(EXPECTED_INERT)) do
ok(JCA.db[key] and not next(JCA.db[key].gives) and not next(JCA.db[key].wants),
('%s is still listed as intentionally inert'):format(key),
'it now has tags - drop it from EXPECTED_INERT')
end
--------------------------------------------------------------------------------
section('Data: public API')
--------------------------------------------------------------------------------
JCA.register('j_mod_test', {'mult', 'bogus_tag'}, {'retrig_scoring'})
ok(JCA.db.j_mod_test, 'JCA.register adds a joker')
ok(JCA.db.j_mod_test.gives.mult, 'known tags survive registration')
ok(not JCA.db.j_mod_test.gives.bogus_tag, 'unknown tags are dropped, not fatal')
ok(JCA.score('j_mod_test', 'j_hack') >= 2, 'a registered joker scores against vanilla')
local before = JCA.score('j_mod_test', 'j_baron')
JCA.register_pair('j_mod_test', 'j_baron', 'test blurb')
eq(JCA.score('j_mod_test', 'j_baron') - before, 4, 'JCA.register_pair adds the +4 pair bonus')
JCA.db.j_mod_test = nil -- keep the UI section on vanilla data only
--------------------------------------------------------------------------------
section('UI: catalog pages build and animate')
--------------------------------------------------------------------------------
-- Cards must materialize as they are emplaced, or a page switch pops them in
-- with no animation. Only the first card of a page may play the sound.
local function build(label, fn)
emplaced, materialized, sounds = {}, {}, 0
local built, err = pcall(fn)
if not ok(built, ('page %s builds'):format(label), tostring(err)) then return end
local missing = 0
for _, c in ipairs(emplaced) do
if not materialized[c] then missing = missing + 1 end
end
eq(missing, 0, ('page %s materializes every one of its %d cards'):format(label, #emplaced))
ok(sounds <= 1, ('page %s plays at most one materialize sound'):format(label),
('%d cards played it'):format(sounds))
end
G.jokers = nil -- the catalog is reachable from the main menu, outside a run
for _, tab in ipairs(SMODS.current_mod.extra_tabs()) do
build(tab.label, tab.tab_definition_function)
end
-- And the same on a page switch, which is what the pager callbacks do.
G.OVERLAY_MENU = {get_UIE_by_ID = function()
return {config = {object = {remove = function() end}},
UIBox = {recalculate = function() end}}
end}
for _, name in ipairs({'Combos', 'Engine', 'Economy', 'Hands'}) do
build(name .. ' page 2', function()
G.FUNCS['jca_page_' .. name]{cycle_config = {current_option = 2}}
end)
end
--------------------------------------------------------------------------------
print(('\n%d passed, %d failed'):format(pass, fail))
os.exit(fail == 0 and 0 or 1)