/* cogitator/datasheet.css v104 */ /* DevAgent24 - 2026-07-16 What: v76 — Removed the .ds--tyranids / .ds--astartes faction colour token classes and their static #ffffff --ds-faction-color-text. The datasheet root now carries a NEUTRAL default (--banner charcoal); the real accent and its contrast-derived ink arrive as inline custom properties from faction-colours.js (factions.accent_color). Also repointed the two static #ffffff declarations that sit directly on the faction gradient (.ds-unit-name, .ds-category-badge) at --ds-faction-color-text. Why: Only two factions had token classes, so the other 26 fell through to .ds--astartes and rendered Space Marine navy. Navy-as-default is what made the bug invisible for Space Marines; the fallback is now neutral and no faction colour is a static CSS value anywhere in this file. */ /* DevAgent23 - 2026-07-16 What: v75 — Added the .ds-cv-* (list-convention text) rules, ported verbatim from codex-col3.css v3's transitional block. Home for them here — the .ds-* namespace owner, loaded on the Army Builder AND Codex pages — so ArmyBuilderCol2.js v77's new WARGEAR OPTIONS / WARGEAR ABILITIES ConventionTextBlock styles correctly (codex-col3.css is NOT loaded on the builder page). Same move DevAgent22 made for .ds-equip-* in v74. Why: Builder v77 renders wargear_options_text / wargear_abilities_text via ConventionTextBlock; those .ds-cv-* classes existed only in codex-col3.css, so the builder panel was unstyled without them. Note: codex-col3.css v3 still carries its own identical .ds-cv-* copy; values match, so the duplicate is harmless. Stripping it there is a later codex-col3.css tidy (same as the still-duplicated .ds-equip-*). */ /* DevAgent22 - 2026-07-15 v74 — Added the .ds-equip-* equipment-pill rules (type-coloured solid scheme, matching army-view-col2.css v8 / codex-col3.css v3). These previously lived only in codex-col3.css (transitional home), which the Builder page does not load — so the Builder datasheet's pills were unstyled. This is their permanent .ds- home. NOTE: codex-col3.css still carries an identical copy; it can be stripped in a follow-up once confirmed army-builder-col2.css is loaded on the codex page too (per codex-col3.css v1's "loaded globally" note). */ /* Army Builder Column 2 — Unit Datasheet panel styles */ /* DevAgent10 - 2026-06-01 What: v73 — Datasheet header image reworked for transparent-PNG art: the photo is widened, switched to object-fit:contain (no crop) and right-anchored so wide models extend left across the faction gradient, behind the stats (z-index 0). A left-fading mask dissolves the model into the gradient. Tunable levers: .ds-unit-photo width, mask stop %, .ds-header gradient stops, .ds-header__content padding-right. */ /* DevAgent10 - 2026-06-01 What: v72 — Wargear option bullets (GW style): top-level options use a filled square marker, sub-options a hollow circle. Lists switched from flex to block so list markers render (flex items suppress markers). */ /* DevAgent10 - 2026-06-01 What: v71 — Review feedback on v70: .ds-stat__inv-label-cell now align-self: center so "INVULNERABLE SAVE" (and the optional note beneath it) is vertically centred against the inv shield. The stat grid's align-items:end was bottom-aligning the cell, dropping a single-line label below the shield's centre. */ /* DevAgent10 - 2026-06-01 What: v70 — Review feedback on v69: 1. Ability group header now uses .ds-banner (full width). New .ds-ability-group__body padded container; .ds-ability-group__header rule removed. 2. .ds-stat__inv-label-cell stacks label + footnote, vertically centred against the inv shield; .ds-stat__inv-footnote restyled to sit under the label. 4. .ds-leader-block__desc italic. */ /* DevAgent32 - 2026-07-19 What: v104 — .ds-ability__desc gets list rules, so an ability that says "select one of the following:" shows its options as an indented bulleted list. Why: Upstream ships REAL MARKUP for these. Bladeguard Veteran Squad's description (waha_datasheet_abilities, datasheet 000000071, line 2) is a lead-in sentence, a
, then . UL, LI and B are all in codex-col3's ALLOWED_TAGS, so the structure already survived the sanitiser intact — nothing was being dropped and no text was being lost. ⚠ The flattening is base.css:107, `ul, ol { list-style: none }`, plus the `*` box-model reset zeroing padding — so the options arrived structurally correct, unmarked and unindented. Exactly the cause fixed for the wargear options in v78 and for the popup body in v81, and fixed the same way: markers declared locally rather than by widening the allow-list to admit upstream's inline style. ⚠ CSS ONLY. codex-col3.js is not involved in this fix and the rules text is untouched — this changes how the options LOOK, never what they SAY. */ /* DevAgent29 - 2026-07-19 What: v103 — .ds-equip-pill gets vertical-align: middle. Why: An inline-flex box aligns on its BASELINE by default, which puts the pill's inner text on the sentence's baseline and pushes the pill's padding and border below it — so the pill sat low against the words either side of it. `middle` aligns the pill's own centre to the middle of the surrounding lowercase instead, which is what reads as centred. */ /* DevAgent29 - 2026-07-19 What: v102 — .ds-equip-pill__qty goes to full opacity and a slightly larger size. Why: It now appears on every pill (codex-col3 v29 shows x1 as well as x2), so it has to read at a glance rather than sit back as a footnote. */ /* DevAgent29 - 2026-07-19 What: v101 — stripes narrowed again, 6px → 4px. vertical-align removed from .ds-equip-pill__qty. Why: Jérôme review. The align nudge is no longer needed: codex-col3 v25 emits a plain lowercase 'x', which sits on the baseline by construction, rather than U+00D7. */ /* DevAgent29 - 2026-07-19 What: v100 — dual-pill stripes narrowed 9px → 6px; .ds-equip-pill__qty gets vertical-align: -0.08em. Why: Jérôme review. ⚠ The '×' is U+00D7 MULTIPLICATION SIGN, which is centred on the mathematical axis — higher than the baseline the digit beside it sits on — so it read as raised. This nudges it into line rather than replacing the glyph with a lowercase 'x', which would sit correctly but read as a letter next to a number. */ /* DevAgent29 - 2026-07-19 What: v99 — the dual pill's rule is raised to .ds-equip-pill.ds-equip-pill--dual and uses the `background` SHORTHAND. Adds .ds-equip-pill__qty. Why: v98 relied on SOURCE ORDER to beat .ds-equip-pill's background shorthand, and it rendered as an empty white pill — white text on the base rule's fallback surface, gradient absent. Order-dependence is the wrong mechanism for this: the rule is now more SPECIFIC (0,2,0 vs 0,1,0), so it wins wherever it sits and survives any future reshuffle of this file. .ds-equip-pill__qty carries the '×2' that upstream writes as a bare number before the weapon name. */ /* DevAgent29 - 2026-07-19 What: v98 — the dual pill becomes diagonal STRIPES, and its rule MOVES to after .ds-equip-pill. Why: Two problems, one of them invisible. v97 put --pill-c: #4a6a48 on the dual modifier purely as a border colour and set a split gradient — but the modifier rules sit ABOVE the base rule in this file, and .ds-equip-pill's `background: var(--pill-c)` SHORTHAND then reset the gradient. The pill rendered as flat #4a6a48, a muddy green that was never a chosen colour. ⚠ Any modifier that needs its own background must come AFTER .ds-equip-pill, or the shorthand wins on source order. The three single-colour modifiers are unaffected — they only set --pill-c, which is exactly what the shorthand consumes. Stripes rather than a two-part split: both colours read at pill size, where a 50/50 split just looked like a gradient artefact. */ /* DevAgent29 - 2026-07-19 What: v97 — .ds-equip-pill--dual, for a weapon that appears in BOTH the ranged and the melee table. Split fill: ranged green into melee brown. Why: A name carried by both tables is the old dual-listed defect class — the pill used to point at one of them, arbitrarily, and nothing on screen showed the choice had been made. The split fill states it, and the type badge reads 'R/M' rather than picking a side. ⚠ Do NOT "simplify" this to a single colour. The ambiguity is the information. */ /* DevAgent29 - 2026-07-19 What: v96 — .ds-wargear-options__item--note loses its marker and its indent. Why: Those rows (button='*') are NOTES about the options above them, not options themselves — "* This model can only be equipped with two ranged weapons if one of them is a Pistol". A bullet implied it was another choice, and the indent aligned it with the choices it qualifies. The asterisk is already in the published text, so the row needs no marker of its own. */ /* DevAgent29 - 2026-07-19 What: v95 — REVERTS v94. .ds-keywords-row is a 7fr/4fr grid again and .ds-keywords-row--faction is removed. Why: The keyword split was meant for the Col1 filter panel, not the datasheet. The published sheet sets the two side by side. */ /* DevAgent29 - 2026-07-19 What: v94 — .ds-keywords-row goes from a 7fr/4fr two-column grid to a single full-width block, and .ds-keywords-row--faction takes a lighter internal divider. Why: KEYWORDS and FACTION KEYWORDS are now two stacked sections (codex-col3 v19). The keyword list is much the longer of the two and was confined to 7/11ths of the width while a one-word faction list held the rest. The faction divider is deliberately lighter than the block's own top border so the two read as one region, not two. */ /* DevAgent29 - 2026-07-19 What: v93 — RENAMED from army-builder-col2.css. Content is unchanged; only the filename and this header differ. Version lineage continues (v92 → v93), following the precedent set when DevAgent28 renamed the CodexCol* files. Why: The datasheet is not a column. It renders at Army Builder col 2, Army Viewer col 2, and Codex col 3 — so a name carrying a column position was wrong on two of the three surfaces, and reading the Codex datasheet out of a file called "army builder col 2" was actively misleading. 158 of the rules here are .ds-*: this file IS the datasheet component's stylesheet, and the three columns are its consumers. ⚠ Load position in index.html is UNCHANGED (still between army-builder-col1.css and army-builder-col3.css). Moving it would reorder the cascade, and this rename is deliberately not that change. Note: the .dsv-* twin, army-view-col2.css, still describes itself as a copy to "keep in step" with this file. Merging the two namespaces into one component is a later session's work; this rename is a precondition for it, not the thing itself. */ /* DevAgent29 - 2026-07-19 What: v92 — .ds-weapon-tags: text-transform uppercase. Why: Upstream casing is transcription variance, not convention. Of 6,568 wargear rows carrying tags, 6,165 are all-lowercase, 364 all-uppercase and ~39 mixed — one datasheet can show 'pistol', 'PISTOL' and 'PISTOl' on three different weapons. GW prints them in capitals, so the 94% case did not match the published sheet either. ⚠ Fixed HERE, not in the payload or the DB. The mirror keeps what Wahapedia published; the transform flattens all three casings for display, and takes the 'PISTOl' typo with it. Do not normalise the source data to achieve this. */ /* DevAgent29 - 2026-07-19 What: v91 — keywords abandon font-variant-caps for explicit two-tier markup, and switch to --font-heading. Why: ⚠ SYNTHESISED small caps scale at a ratio fixed by the browser. Big Shoulders Display carries no true small-cap glyphs, so the reduced letters sat well below cap height and NO CSS property could raise them — that is why v89/v90 could not fix it by size or weight. codex-col3 v16 now emits .ds-keyword__initial and .ds-keyword__rest, and the ratio is set here: 0.86em, close to cap height rather than the browser's ~0.75. --font-heading is condensed, so it is both narrower and lighter than the body face at the same size, which was the other half of the ask. The remainder is uppercased HERE, not in the payload. */ /* DevAgent29 - 2026-07-19 What: v90 — keyword values: font-size --text-sm → --text-base, and an explicit font-weight: 400. Why: Jérôme review — thinner and taller. The weight is stated rather than inherited because the small caps here are SYNTHESISED (Big Shoulders Display carries no true small-cap glyphs), and a synthesised small cap is a scaled-down capital that reads heavier than the text around it. Larger size plus an explicit lighter weight corrects both at once. */ /* DevAgent29 - 2026-07-19 What: v89 — keyword values render in SMALL CAPS, as GW sets them: the initial letter full height, the rest as reduced capitals. Why: Jérôme review. ⚠ This is TYPOGRAPHIC, not textual. The payload keeps upstream's own casing ("Adeptus Astartes") and font-variant-caps does the rest. text-transform: uppercase would defeat it entirely — with every letter already a capital there is no lowercase left to reduce, and the effect disappears. Do not "normalise" the keyword text to fix a casing inconsistency; the mixed casing IS what makes this render. .ds-keywords__label keeps its own uppercase: it is a heading, not a keyword, and small-caps on already-capital text is a no-op anyway. */ /* DevAgent29 - 2026-07-19 What: v88 — the popup cap gains an absolute ceiling: min(560px, calc(100vh - 160px)). Why: v86's viewport-only cap worked, but 100vh is whatever the screen happens to be — on a tall desktop window it resolved to ~1,080px and the popup was still far taller than anything readable. 560px is sized for the target device (iPad 10 landscape, ~680px usable after chrome) and holds on any larger screen; the vh term still shrinks it further on anything shorter. */ /* DevAgent29 - 2026-07-19 What: v87 — the top-level wargear option marker is DRAWN as a 7px box instead of set as a glyph. Why: v85 used U+25AA BLACK SMALL SQUARE, whose rendered size is decided by the font and came out smaller than the nested circle beneath it — the opposite of the intended hierarchy. A background box is exact, font-independent, and guaranteed larger than the ~6px nested marker. */ /* DevAgent29 - 2026-07-19 What: v86 — .ds-popup gets a height cap and a scrolling body. Why: The popup now carries Wahapedia's full published rule text. The CORE Leader ability is ~4KB of markup and rendered at its natural height, running off the bottom of the screen with no way to reach the end. Flex column: the header is fixed, the body scrolls. The cap is calc(100vh - 160px) — iPad 10 landscape is 820 CSS px tall, so this leaves roughly 660px after browser chrome and the app header, and adapts down on anything smaller. ⚠ min-height: 0 on the body is load-bearing, not decoration: a flex child will not shrink below its content height without it, and the overflow would never engage. */ /* DevAgent29 - 2026-07-19 What: v85 — wargear option markers swapped: the top-level option takes a SQUARE, the nested choice list takes a CIRCLE. v79 had them the other way round. Why: Matches the published sheet, and the nested level matches what upstream itself declares — the