/* =============================================================
 * SUPERFLY — CATEGORY INDEX — looks
 * -------------------------------------------------------------
 * NOW THE SHOP BY CATEGORY PAGE (page 59), not the home page.
 * It was the home page's second band (container sfcat01) until
 * 2026-11-10, when sections 2, 3 and 4 were removed from page 42
 * and this file went with them; the operator asked the same day
 * for the Shop by Category page to take the arrangement instead
 * of its four .sf-bikes--cats tiles. Restored byte-for-byte from
 * ~/sfsections234-backup-20261110-122515 on the box, and enqueued
 * by superfly_category_index_assets() on is_page('shop-by-category').
 * Nothing in here is page-specific — if the home page ever wants it
 * back, widen the enqueue; do not fork the file.
 * -------------------------------------------------------------
 * Markup:   assets/html/category-index.html
 * Movement: assets/js/category-index.js
 *
 * The hover-list arrangement of the three categories, built from
 * docs/research/2026-10-20-category-section-references.md. Read the
 * head of the markup first — it says which figures are real (all of
 * them) and which swatch colours are not (the dots).
 *
 * NAMESPACE. Everything is scoped under .sfidx. The prefixes already
 * taken in this folder are sf-acc, sf-bike(s), sf-cart, sf-cat(s),
 * sf-checkout, sf-dealers, sf-drawer, sf-filters, sf-hero, sf-intro,
 * sf-lights, sf-live, sf-livery, sf-login, sf-maskword, sf-mega,
 * sf-navbar, sf-pdp, sf-pf, sf-sequence, sf-shop, sf-social, sf-sort,
 * sf-step, plus .sfc (categories-flow) and .sfv (the hero) — grep
 * assets/css before adding a fourth section with a fourth prefix.
 *
 * BOX SIZING, AND THE 80px TRAP. Elementor's frontend CSS puts
 * `box-sizing:border-box` on `.elementor *` — every DESCENDANT, never
 * the root. A section that caps its own ROOT therefore measures 80px
 * narrower inside Elementor than outside it (CLAUDE.md has the full
 * story). This file states box-sizing on the root ITSELF, so the two
 * environments agree and no page-id override is needed. Do not remove
 * the first rule.
 *
 * THE REVEAL. Each row ships with an inline `--sfidx-in:0` and the
 * stylesheet reads that number for the clip, the lift and the fade.
 * The script's whole job is to set it to 1 — there is no second
 * source of truth for what "open" looks like, and no tween. Under
 * prefers-reduced-motion the transitions are dropped and the block
 * below forces every row open, so a reader who never triggers the
 * observer still sees a complete list.
 *
 * COLOURS come from theme.json through var(--wp--preset--color--*).
 * accent-1 is #ec1e79 on the live theme and that is the brand magenta
 * (CLAUDE.md → Conventions). The fallbacks here exist only so the
 * local preview renders before WordPress prints its variables; never
 * hardcode the hex anywhere a theme variable would do.
 * ============================================================= */

.sfidx,
.sfidx *{ box-sizing: border-box }

.sfidx{
	--sfidx-accent: var(--wp--preset--color--accent-1, #ec1e79);
	--sfidx-bg:     var(--wp--preset--color--base, #101014);
	--sfidx-ink:    var(--wp--preset--color--contrast, #f2f2f4);
	--sfidx-dim:    rgba(242,242,244,.60);
	--sfidx-faint:  rgba(242,242,244,.34);
	--sfidx-rule:   rgba(242,242,244,.14);

	--sfidx-display: "Inter", system-ui, -apple-system, sans-serif;
	--sfidx-body:    "Inter", system-ui, -apple-system, sans-serif;
	--sfidx-mono:    "Chakra Petch", "Archivo", sans-serif;

	/* NO CAP SINCE 2026-11-17 -- see the long note on .sf-shop__head in
	   shop-archive.css, which this page MUST stay in step with. Both drop
	   it or neither: the two shop pages are only ever right together. */
	--sfidx-wrap:   none;
	/* THE SIDE GUTTER IS THE NAVBAR LOGO'S LEFT EDGE, as close as a
	   fluid value gets to it. The floating pill stops 24px short of
	   the screen and carries up to 20px of padding inside its glass,
	   so the logo sits at 44px on any desktop width — and this section
	   used to start its heading at 72px, 28px further in than the only
	   other thing at the top of the page. Measured on page 59 at
	   1440px: title left 72, logo left 44, and 158px of air above the
	   title. Reduced on the operator's call, 2026-11-10: "why the
	   padding is too much".

	   3.1vw caps at 44 from about 1420px up and falls away below, so
	   it is never WIDER than it was at any width — 43 at 1380, 40 at
	   1280, 32 at 1024, 20 on a phone, against 72 / 69 / 51 / 20
	   before. If navbar.css ever moves --sf-navbar-gutter or
	   --sf-navbar-pad-x, re-measure the logo and re-pitch the 44.

	   NOT a var() read from navbar.css. That file loads on every page
	   and this one on two, so the order happens to work today — but a
	   custom property substitutes at computed-value time on the
	   element where it is DECLARED, and :root tokens crossing files
	   is exactly the trap CLAUDE.md records on the navbar. Two
	   literals that say they belong together beat one that fails
	   silently. */
		/* SUPERSEDED THE SAME DAY by the FAQ geometry above, which is the
	   current state. Kept because it is the one record of what the
	   80px flat ramp measured, and because the two-literals / four-pages
	   argument in it still holds.

	   RAISED 2026-11-16 (operator: "i want you add padding for shop by
	   bike , shop by category , product category , dengan brand page",
	   then 80px from three measured options).

	   Sides went clamp(20px, 3.1vw, 44px) -> clamp(24px, 4.2vw, 80px)
	   and the block padding clamp(40px, 4.4vw, 72px) ->
	   clamp(48px, 5.6vw, 96px). Measured: 24/48 at 375, 43/57 at 1024,
	   60/81 at 1440, 80/96 at 1900.

	   80 IS THE HOME PAGE'S OWN WIDEST BAND. .sfcg carries
	   clamp(20px, 4.2vw, 80px) — sohub's xl:px-20 — so the shop pages
	   now sit on the same line as the category grid rather than on the
	   navbar logo's 44.

	   THE FOUR PAGES MOVE TOGETHER BY CONSTRUCTION, which is the whole
	   reason this is two literals and not four: Shop by Bike and Shop
	   by Category are both .sfidx, and every product-category and brand
	   archive is .sf-shop. The two files were matched value for value
	   earlier today and must stay matched — CHANGE BOTH OR NEITHER.

	   WHAT IT COSTS: the measure on both is 1560 minus twice the
	   gutter, so the content narrows by twice any increase — about 33px
	   at 1440 and 72px at 1900, off a three-column grid. That was put
	   to the operator with the options and taken.

	   AND THE RAIL CLEARANCE RETIRES ITSELF. Both files indent their
	   headings by max(0px, 58px - gutter) to clear the fixed social
	   rail; above a 58px gutter that computes 0 on its own, so from
	   about 1381px up the headings sit flush on the gutter instead of
	   14px inside it. That is the max() doing what it was written to
	   do, not a regression. */
	/* THE SIDES ARE THE FAQ PAGE'S, EXACTLY (operator, 2026-11-16:
	   "added moreee padding , you can open FAQ page , i want like that
	   padding").

	   It is NOT a bigger gutter. The FAQ page — and contact, privacy,
	   terms and About Us with it — is a 1380px box CENTRED in the
	   window with a 40px gutter inside it, so what you read as padding
	   is the centring plus the gutter, and it grows with the screen
	   while the measure stays put. The 80px gutter this replaces was a
	   flat ramp that could never do that.

	   Measured text edge, this against the 80px version it replaces:

	        375   24   (was 24)
	       1024   51   (was 43)
	       1440   65   (was 60)
	       1780  235   (was 179)   <- the operator's own screen
	       1900  295   (was 244)

	   SO THE SHOP IS BACK ON THE HOUSE MEASURE. CLAUDE.md records it
	   leaving --sf-measure earlier today, when the archives were lined
	   up with /shop-by-category/ at 1560. Both families move together
	   here, so those two stay matched with each other AND rejoin the
	   information pages. Change --sf-measure in style.css and all of
	   them follow.

	   TWO DELIBERATE DEVIATIONS FROM THE FAQ, both because the ask was
	   MORE padding and matching it exactly would have given less:

	     - the floor is 24px, not the FAQ's 20, so a phone does not go
	       backwards by 4px;
	     - the TOP padding is left at clamp(48px, 5.6vw, 96px), which is
	       already more than the FAQ's clamp(34px, 5vw, 64px). Only the
	       bottom takes the FAQ's own clamp(60px, 9vw, 120px), which is
	       more than the 96 it replaces.

	   --sf-measure-box IS DECLARED IN style.css, on :root, and that is
	   safe to read from here only because that sheet loads on every
	   page — the cross-file token trap this project records applies to
	   a token declared in a CONDITIONALLY loaded stylesheet. faq.css
	   and dealer-map.css already read it the same way. */
	--sfidx-gutter: var(--sf-gutter);

	/* The one dial worth turning. Dropped from 88px to 38px on
	   2026-10-20: at 88px the category names were BIGGER than the
	   section's own heading (60px), so the page read as three shouted
	   words with a caption, and a reader could not tell the heading
	   was the heading. A row name is a list item, not a title — it
	   stays below --sfidx-title-max. Shrinking it also all but ends
	   the floating picture's overlap with the words. */
	--sfidx-name-max:  38px;
	/* READ BY NOTHING since the title took section 2's stepped ladder
	   on 2026-11-16. Kept, like the other retired tokens in this file,
	   because putting the clamp back is this line plus one rule. */
	--sfidx-title-max: 60px;

	/* The one easing used everywhere in this section. A long tail —
	   it is what makes a row feel like it settles rather than stops. */
	--sfidx-ease: cubic-bezier(.16, 1, .3, 1);

	position: relative;

	/* THE HOUSE CARBON WEAVE, value for value from categories-flow.css and
	   recommend-collage.css, which is where the operator settled it on
	   2026-10-19: every band on the home page sits on ONE surface, or the
	   seam between two of them shows. This section shipped flat by mistake
	   and was the odd one out — measured 2026-10-20, it was the only home
	   section computing background-image:none while its neighbours carried
	   the weave.

	   Both diagonals are needed: the light one at 45deg is the sheen on
	   the tow, the dark one at -45deg is the shadow under it, and with
	   only one of them it reads as corduroy. */
	background-color: var(--sfidx-bg);
	background-image:
		repeating-linear-gradient(45deg, rgba(255, 255, 255, .025) 0 2px, transparent 2px 4px),
		repeating-linear-gradient(-45deg, rgba(0, 0, 0, .35) 0 2px, transparent 2px 4px);
	color: var(--sfidx-ink);
	font-family: var(--sfidx-body);
	/* 158px of air above the heading at 1440px was the other half of
	   the "too much padding" — the floating navbar ends at y=90 and
	   the title did not start until 250. Now about 63px at 1440 and
	   40 on a phone, against 158 and 72. The bottom comes down with
	   it; it is the same value and the footer follows the section. */
	padding-block: clamp(48px, 5.6vw, 96px) clamp(60px, 9vw, 120px);

	/* Full bleed inside Elementor, which otherwise boxes a container
	   at 1140px. The widget should also be set content_width:full —
	   this is the belt to that braces. */
	width: 100%;
}

.sfidx__wrap{
	max-width: var(--sfidx-wrap);
	margin-inline: auto;
	padding-inline: var(--sfidx-gutter);
}

/* ── head ──────────────────────────────────────────────────── */

/* NO MEASURE ON THE HEAD. `ch` resolves against the element it is
   written on, and the head inherits BODY size — so `46ch` here came
   out around 368px and squeezed the 60px headline into four ragged
   lines. A measure belongs on each child, at that child's own size.
   Measured 2026-10-20 at 1440px. */
.sfidx__head{
	margin-bottom: clamp(44px, 6vw, 92px);
	opacity: 0;
	transform: translateY(26px);
	transition: opacity .8s var(--sfidx-ease), transform .8s var(--sfidx-ease);
}

.sfidx__head.is-in{
	opacity: 1;
	transform: none;
}

/* .sfidx__eyebrow was removed 2026-10-20 with the "The range"
   kicker it styled. The heading now opens the section on its own.
   Nothing else used the class. */

.sfidx__title{
	margin: 0;
	max-width: 20ch;
	font-family: var(--sfidx-display);
	/* THE SIZE IS SECTION 2's TOO NOW (operator, 2026-11-16: "i want the
	   font size is same like homepage 112px right for the title section
	   2 in homepage ?").

	   This REVERSES the "everything but the size" answer given an hour
	   earlier, and with it the 2026-10-05 ruling that this heading is a
	   label over a grid rather than the page itself. The cap was 60px.

	   It is a STEPPED LADDER, not a clamp, because that is what
	   .sfcg__title is — 30 / 48 / 60 / 72 / 112 at 768 / 1024 / 1280 /
	   1900 — and the point is that the two compute identically at every
	   width. A clamp fitted to the same endpoints would not: it ramps
	   between them where the ladder jumps.

	   SO ALL SIX VALUES ARE SHARED AGAIN: family, weight, size, leading,
	   tracking, case. CHANGE category-index.css AND category-grid.css
	   WITH THIS FILE, or they drift — the values are literals in each
	   for the usual reason, that a :root token declared in one
	   stylesheet and read from another resolves by load order and fails
	   silently.

	   WHAT IT COSTS, measured: nothing below 1024, where the old clamp
	   was already at or above these steps. Above it the heading nearly
	   doubles — 60 to 72 at 1440 and 60 to 112 at 1900 — and the grid
	   below it moves down by the difference on a two-line title. That
	   was the argument against it in October; it was asked for anyway,
	   with the numbers in front of the operator. */
	font-size: 30px;

	/* SECTION 2's TYPE, EXCEPT FOR THE SIZE (operator, 2026-11-16).
	   Weight, leading, tracking and case are .sfcg__title's value for
	   value; the size ladder is NOT, because these titles are labels
	   over a list rather than the page itself. The reasoning, the two
	   options and the measurements are written up in shop-archive.css
	   against its own h1 — CHANGE BOTH FILES OR NEITHER, since these
	   four values are literals in each and a :root token read across
	   stylesheets resolves by load order and fails silently.

	   700 because the face changed. This was Archivo Black, which has
	   one weight and is black at 400. */
	font-weight: 700;
	line-height: 1;
	letter-spacing: -.075em;
	text-transform: uppercase;
}

@media (min-width:  768px){ .sfidx__title{ font-size:  48px } }
@media (min-width: 1024px){ .sfidx__title{ font-size:  60px } }
@media (min-width: 1280px){ .sfidx__title{ font-size:  72px } }
@media (min-width: 1900px){ .sfidx__title{ font-size: 100px } }

/* The one italic word. <em> is the markup's only emphasis and it
   carries the accent — do not add a second one, the section leans on
   there being exactly one. */
/* THE PINK LINE CARRIES SECTION 2's SHEEN (operator, 2026-11-16:
   "also the color pink text in title please match with effect
   PERFORMANCE text").

   Ported from .sfcg__word in category-grid.css: ink -> magenta -> ink
   clipped to the glyphs and looping for ever, so it reads as a light
   passing through the word rather than as a second colour. Both
   background-clip longhands, or Blink paints a solid magenta band with
   invisible type inside it.

   width: fit-content IS LOAD-BEARING HERE and is NOT in the original.
   .sfcg__word is an inline span inside a line; this one is
   display: block, so without it the background's positioning area is
   the whole title column and the sweep is spread across the measure
   instead of across the words. home-opening.css records the same fix
   for .sfop__line. It must stay display: block so the line still
   breaks after the white half.

   The keyframe is named sfSheen in every file that carries this. CSS
   keyframes are global, and all these copies are identical, so a page
   loading two of them is harmless. */
.sfidx__title em{
	font-style: normal;
	display: inline;
	color: transparent;
	background-image: linear-gradient(90deg,
		var(--sfidx-ink)    0%,
		var(--sfidx-accent) 48%,
		var(--sfidx-ink)   100%);
	background-size: 220% 220%;
	-webkit-background-clip: text;
	background-clip: text;
	animation: sfSheen 5s ease infinite;

	/* the last glyph would otherwise be clipped by the background box */
	margin-right: -.12em;
	padding-right: .12em;
}

@keyframes sfSheen{
	0%   { background-position: 10% 0 }
	50%  { background-position: 91% 100% }
	100% { background-position: 10% 0 }
}


.sfidx__lede{
	margin: 22px 0 0;
	max-width: 54ch;
	font-size: clamp(15px, 1.2vw, 17px);
	line-height: 1.6;
	color: var(--sfidx-dim);
}

/* ── the list ──────────────────────────────────────────────── */

.sfidx__list{
	margin: 0;
	padding: 0;
	list-style: none;
	border-top: 1px solid var(--sfidx-rule);
}

.sfidx__row{
	position: relative;
	border-bottom: 1px solid var(--sfidx-rule);

	/* 0 closed, 1 open. The script sets it; everything below reads it. */
	--sfidx-in: 1;
}

/* NOTHING THAT CLIPS GOES ON .sfidx__row. The row is the element the
   IntersectionObserver watches. An observer measures the target's
   VISIBLE rect, and Chromium applies the target's own `clip-path`
   when computing it — so a row that ships `inset(0 100% 0 0)` can
   report isIntersecting:false while sitting in the middle of the
   screen, and the observer that is supposed to open it never fires.
   The reveal would deadlock on itself.

   HONESTY NOTE, 2026-10-20: this is a defensive restructure, NOT a
   measured one. It could not be confirmed in Claude's browser pane,
   because that pane does not deliver IntersectionObserver callbacks
   at all — a control test there, two identical fixed boxes fully on
   screen with and without the clip, had NEITHER observer fire within
   3s. Whoever next opens this in a real browser: put the clip back on
   the row for one reload and see whether the rows still arrive. If
   they do, this split is unnecessary and the comment should say so.
   Either way it costs nothing, and categories-flow.css happens to be
   arranged the same way (its clip is on an inner element).

   Keep the clip below this line. */

.sfidx__link{
	opacity: var(--sfidx-in);
	transform: translateY(calc((1 - var(--sfidx-in)) * 30px));
	transition:
		opacity   .7s var(--sfidx-ease),
		transform .7s var(--sfidx-ease);
}

.sfidx__inline{
	opacity: var(--sfidx-in);
	transition: opacity .7s var(--sfidx-ease);
}

/* The wipe is on the WORD, not on the row — it is the only part of
   the row worth wiping, and keeping it off the link means the focus
   ring is never clipped (an outline draws outside the border box, so
   even the finished `inset(0 0 0 0)` would have cut it). */
.sfidx__name{
	clip-path: inset(0 calc((1 - var(--sfidx-in)) * 100%) 0 0);
	transition:
		transform .6s var(--sfidx-ease),
		opacity   .45s var(--sfidx-ease),
		clip-path .9s var(--sfidx-ease);
}

/* Staggered arrival. Three rows, so three delays written out rather
   than calculated — it is shorter than the calc and it is obvious
   what happens when a fourth category is added (give it a delay).
   The delay is on the children now, for the reason above. */
.sfidx__row:nth-child(1) .sfidx__link,
.sfidx__row:nth-child(1) .sfidx__name,
.sfidx__row:nth-child(1) .sfidx__inline{ transition-delay: 0s }
.sfidx__row:nth-child(2) .sfidx__link,
.sfidx__row:nth-child(2) .sfidx__name,
.sfidx__row:nth-child(2) .sfidx__inline{ transition-delay: .09s }
.sfidx__row:nth-child(3) .sfidx__link,
.sfidx__row:nth-child(3) .sfidx__name,
.sfidx__row:nth-child(3) .sfidx__inline{ transition-delay: .18s }

/* …but the hover move must not inherit the arrival delay, or the
   name lags a fifth of a second behind the cursor. */
.sfidx__row.is-in .sfidx__name{ transition-delay: 0s }

/* The magenta rule that draws across the row on hover. On the row,
   not the link, so it spans the full measure regardless of padding. */
.sfidx__row::after{
	content: "";
	position: absolute;
	left: 0;
	right: 0;
	bottom: -1px;
	height: 2px;
	background: var(--sfidx-accent);
	transform: scaleX(0);
	transform-origin: left center;
	transition: transform .65s var(--sfidx-ease);
	pointer-events: none;
}

.sfidx__row.is-on::after{ transform: scaleX(1) }

.sfidx__link{
	display: grid;
	grid-template-columns: auto minmax(0, 1fr) auto auto auto;
	align-items: center;
	gap: clamp(18px, 2.6vw, 46px);
	padding-block: clamp(26px, 3.4vw, 46px);
	text-decoration: none;
	color: inherit;
}

.sfidx__link:focus-visible{
	outline: 2px solid var(--sfidx-accent);
	outline-offset: 4px;
}

.sfidx__num{
	font-family: var(--sfidx-mono);
	font-size: 13px;
	font-weight: 700;
	letter-spacing: .16em;
	color: var(--sfidx-faint);
	transition: color .45s var(--sfidx-ease);
}

.sfidx__row.is-on .sfidx__num{ color: var(--sfidx-accent) }

.sfidx__body{ display: block; min-width: 0 }

/* The reveal half of this element — clip-path, and the transition
   that carries all three properties — is declared up with the row,
   next to the note explaining why it cannot live on the row itself.
   Do not add a `transition` here: a second one would win on order
   and silently drop the wipe. */
.sfidx__name{
	display: block;
	font-family: var(--sfidx-display);
	font-size: clamp(22px, 2.9vw, var(--sfidx-name-max));
	/* 700 for the same reason as the title above: Archivo Black was
	   black at 400 and Inter is not. The size, leading and tracking are
	   this element's own and were not part of the section 2 match. */
	font-weight: 700;
	line-height: 1;
	letter-spacing: -.025em;
	text-transform: uppercase;

	/* A descender on a wiped line gets cut by the clip otherwise. */
	padding-bottom: .06em;
}

/* The Format move: the hovered name steps aside, the others recede.
   Both halves matter — dimming alone reads as a bug, moving alone
   reads as a wobble. */
.sfidx__row.is-on .sfidx__name{ transform: translateX(clamp(10px, 1.6vw, 30px)) }
.sfidx__list.is-hovering .sfidx__row:not(.is-on) .sfidx__name{ opacity: .26 }
.sfidx__list.is-hovering .sfidx__row:not(.is-on) .sfidx__sub,
.sfidx__list.is-hovering .sfidx__row:not(.is-on) .sfidx__meta{ opacity: .3 }

.sfidx__sub{
	display: block;
	margin-top: 14px;
	font-family: var(--sfidx-mono);
	font-size: 12.5px;
	font-weight: 600;
	letter-spacing: .1em;
	text-transform: uppercase;
	color: var(--sfidx-faint);
	transition: opacity .45s var(--sfidx-ease);
}

/* ── finish chips ──────────────────────────────────────────── */

/* The whole group is hidden at rest, label included. An earlier pass
   hid only the chips and left "FINISHES" hanging over an empty gap,
   which reads as a section that failed to load rather than as one
   waiting to be hovered. */
.sfidx__chips{
	display: flex;
	flex-wrap: wrap;
	align-items: center;
	gap: 8px;
	max-width: 290px;
	opacity: 0;
	transition: opacity .3s var(--sfidx-ease);
}

.sfidx__row.is-on .sfidx__chips{ opacity: 1 }

.sfidx__chipsLabel{
	width: 100%;
	margin-bottom: 4px;
	font-family: var(--sfidx-mono);
	font-size: 10.5px;
	font-weight: 600;
	letter-spacing: .2em;
	text-transform: uppercase;
	color: var(--sfidx-faint);
}

.sfidx__chip{
	display: inline-flex;
	align-items: center;
	gap: 7px;
	padding: 6px 11px;
	border: 1px solid var(--sfidx-rule);
	border-radius: 999px;
	font-size: 11.5px;
	line-height: 1;
	white-space: nowrap;
	color: var(--sfidx-dim);

	/* Closed state. The script does not touch these — they ride on the
	   row's own .is-on, so a chip row cannot get stuck open. */
	opacity: 0;
	transform: translateY(8px);
	transition: opacity .4s var(--sfidx-ease), transform .4s var(--sfidx-ease);
}

.sfidx__row.is-on .sfidx__chip{ opacity: 1; transform: none }

/* Chips arrive left to right behind the picture, not with it. */
.sfidx__row.is-on .sfidx__chip:nth-of-type(1){ transition-delay: .06s }
.sfidx__row.is-on .sfidx__chip:nth-of-type(2){ transition-delay: .12s }
.sfidx__row.is-on .sfidx__chip:nth-of-type(3){ transition-delay: .18s }
.sfidx__row.is-on .sfidx__chip:nth-of-type(4){ transition-delay: .24s }

/* The dot is a swatch, NOT a control. See the markup's header: these
   hexes are approximations and nothing is wired to them. */
.sfidx__chip i{
	width: 9px;
	height: 9px;
	border-radius: 50%;
	background: var(--sfidx-dot, transparent);
	box-shadow: inset 0 0 0 1px rgba(242,242,244,.22);
	flex: none;
}

/* ── meta + arrow ──────────────────────────────────────────── */

.sfidx__meta{
	display: grid;
	gap: 6px;
	justify-items: end;
	text-align: right;
	font-family: var(--sfidx-mono);
	font-size: 12px;
	letter-spacing: .08em;
	text-transform: uppercase;
	white-space: nowrap;
	transition: opacity .45s var(--sfidx-ease);
}

.sfidx__count{ color: var(--sfidx-dim) }
.sfidx__price{ color: var(--sfidx-ink); font-weight: 600 }

.sfidx__go{
	display: grid;
	place-items: center;
	width: 46px;
	height: 46px;
	border: 1px solid var(--sfidx-rule);
	border-radius: 50%;
	color: var(--sfidx-ink);
	flex: none;
	transition:
		background-color .45s var(--sfidx-ease),
		border-color     .45s var(--sfidx-ease),
		color            .45s var(--sfidx-ease),
		transform        .45s var(--sfidx-ease);
}

.sfidx__go svg{ width: 16px; height: 10px; display: block }

.sfidx__row.is-on .sfidx__go{
	background: var(--sfidx-accent);
	border-color: var(--sfidx-accent);
	color: var(--sfidx-bg);
	transform: translateX(6px);
}

/* ── the floating picture ──────────────────────────────────── */

/* position:fixed, and deliberately a sibling of .sfidx__wrap rather
   than a child — a transformed or filtered ancestor would make
   `fixed` resolve against THAT box and the picture would lag the
   cursor by the section's offset. */
.sfidx__preview{
	position: fixed;
	top: 0;
	left: 0;
	z-index: 40;

	/* A CIRCLE, and a small one — changed 2026-10-20 on the operator's
	   call after seeing it live. It was a 370px 4:5 card, which at a
	   1633px viewport covered most of "SUSPENSIONS" and read as the
	   subject of the section rather than a preview of it. A disc this
	   size reads as something attached to the cursor instead, and the
	   row underneath stays legible. */
	width: clamp(132px, 12vw, 190px);
	aspect-ratio: 1;
	pointer-events: none;
	will-change: transform;

	/* Hidden until the script says otherwise, and hidden for good on
	   any device that never gets a cursor (see the media query at the
	   foot of this file). */
	display: none;
}

/* The script writes `transform` on the OUTER element every frame, so
   the scale-and-fade has to live on an inner one — otherwise the
   per-frame write would clobber the transition. */
/* With the names down to 38px there is real empty row to the right of
   each one, so the disc now mostly travels over background rather than
   over letters. It can still clip a word on a narrow desktop; the ring
   and the shadow are what keep it reading as something above the page,
   and the cursor is never underneath it (see GAP in the script). */
.sfidx__previewInner{
	position: absolute;
	inset: 0;
	overflow: hidden;
	border-radius: 50%;
	background: #111;
	border: 1px solid rgba(242,242,244,.14);
	box-shadow:
		0 24px 50px -12px rgba(0,0,0,.85),
		0 0 0 1px rgba(0,0,0,.4);
	opacity: 0;
	transform: scale(.86);
	transition: opacity .4s var(--sfidx-ease), transform .55s var(--sfidx-ease);
}

.sfidx__preview.is-live .sfidx__previewInner{
	opacity: 1;
	transform: scale(1);
}

/* THE SELECTOR IS DOUBLED ON PURPOSE. Twenty Twenty-Four (and almost
   every theme) ships `img{ height:auto }`, which outranks a single
   class on a plain property and left the shot standing short of its
   box — a black band under the product, visible on the live page
   2026-10-20. Keep the parent in the selector, and keep height
   stated. Round box, so a short image is now a visibly cropped disc. */
.sfidx__preview .sfidx__shot{
	position: absolute;
	inset: 0;
	width: 100%;
	height: 100%;
	min-height: 100%;
	object-fit: cover;
	opacity: 0;
	transform: scale(1.08);
	transition: opacity .45s var(--sfidx-ease), transform .7s var(--sfidx-ease);
}

.sfidx__preview .sfidx__shot.is-on{ opacity: 1; transform: none }

/* The inline picture is the touch-screen half of the same idea. Off
   by default; the media query at the foot turns it on. */
.sfidx__inline{ display: none }

.sfidx__inline img{
	display: block;
	width: 100%;
	height: clamp(180px, 46vw, 300px);
	object-fit: cover;
	border-radius: 2px;
}

/* ── phones and tablets ────────────────────────────────────── */

@media (max-width: 1023px){

	.sfidx__link{
		grid-template-columns: auto minmax(0, 1fr);
		grid-template-areas:
			"num  body"
			".    chips"
			".    meta";
		row-gap: 18px;
		align-items: start;
	}

	.sfidx__num { grid-area: num; padding-top: 6px }
	.sfidx__body{ grid-area: body }
	.sfidx__chips{ grid-area: chips; max-width: none }
	.sfidx__meta{ grid-area: meta; justify-items: start; text-align: left }

	/* The arrow is doing nothing on a touch screen — the whole row is
	   the target. Removed rather than shrunk. */
	.sfidx__go{ display: none }

	/* Same rule on a phone: the name must stay under the heading,
	   which is 28px there. */
	.sfidx__name{ font-size: clamp(19px, 5.4vw, 30px) }
}

/* No cursor to float beside: show the picture in the row instead and
   take the floating one out of the document entirely. `hover:none`
   rather than a width — a touch laptop at 1400px wants this too. */
@media (hover: none), (pointer: coarse){

	.sfidx__preview{ display: none !important }

	.sfidx__inline{
		display: block;
		padding-bottom: clamp(22px, 4vw, 34px);
	}

	/* No hover to arrive on here, so the group and its chips are
	   simply open — both halves, or the label hangs over a gap. */
	.sfidx__chips{ opacity: 1 }
	.sfidx__chip{ opacity: 1; transform: none }
}

@media (hover: hover) and (pointer: fine){
	.sfidx__preview{ display: block }
}

/* ── reduced motion ────────────────────────────────────────── */

/* Everything open, nothing moving. The floating picture goes away
   completely — it is motion that follows the pointer, which is the
   exact thing this setting is asking not to happen. */
@media (prefers-reduced-motion: reduce){

	.sfidx__preview{ display: none !important }

	.sfidx__inline{ display: block; padding-bottom: 24px }

	.sfidx__row{
		--sfidx-in: 1 !important;
		opacity: 1 !important;
		transform: none !important;
		clip-path: none !important;
	}

	.sfidx__head{ opacity: 1 !important; transform: none !important }

	.sfidx__chips,
	.sfidx__chip{ opacity: 1 !important; transform: none !important }

	.sfidx *,
	.sfidx *::before,
	.sfidx *::after{
		transition-duration: .01ms !important;
		transition-delay: 0s !important;
		animation-duration: .01ms !important;
	}
}


/* ── the fixed social rail ─────────────────────────────────── */

/* THE SOCIAL RAIL SITS ON THE ROW NUMBERS, and this was not visible
   while the section was a band of the home page — nobody checked a
   phone against it. `.sfsr` is pinned to the MIDDLE OF THE LEFT EDGE
   of every page at z-index 90, 46px wide and 40px below 782px, and
   the numbers are the leftmost thing in the list.

   Measured on page 59 at 375x812: the rail occupied [0, 323, 40, 489]
   and "01" sat at left 20, right 38 — squarely behind it. Because the
   rail is FIXED and the list is long, this is not one unlucky row:
   every number passes behind it as the page scrolls.

   The indent is on the LINK and the inline picture, not on the row,
   so the hairline rules keep the heading's left edge and the list
   still reads as one block. It is written as a max() against the
   section's own gutter rather than a media query, so it disappears
   on its own the moment the gutter is wide enough to clear the rail
   — 14px at 1440px (44px gutter), 26px at 1024px, 38px at 375px, and
   zero again the moment a gutter of 58 or more is set.

   SO THE NUMBERS SIT 14px INSIDE THE HEADING ON DESKTOP, and that is
   new: the gutter was 72px until 2026-11-10 and this resolved to
   zero, so everything shared one edge. Narrowing the gutter to the
   navbar logo's 44px put the numbers back under the rail, and the
   indent is what takes them out again. The hairline rules still line
   up with the heading, which is the edge the eye actually reads.

   58px = the rail's widest (46px) plus a 12px gap. If --sfsr-w ever
   changes in social-rail.css, change it here and re-measure; the
   rail's own width is not readable from this file. */
.sfidx{
	--sfidx-rail-clear: max(0px, calc(58px - var(--sfidx-gutter)));
}

.sfidx__link,
.sfidx__inline{
	padding-left: var(--sfidx-rail-clear);
}


/* ── a category with no photograph ──────────────────────────────
   Added 2026-11-13, with the conversion to [superfly_category_index].

   WHEN THE ROWS WERE TYPED BY HAND A MISSING PICTURE COULD NOT
   HAPPEN — somebody wrote the <img> or they did not write the row.
   The shortcode builds both lists from one loop over the terms, so
   a category included without art now reaches the page, and the
   script still has to find a shot at that index: it toggles
   shots[index] against rows[index], and a row with no element there
   would shift every later index by one.

   So the shortcode prints an EMPTY shot carrying data-placeholder
   rather than skipping it, and this is what that element looks like
   — the house carbon with a magenta bloom, a deliberate surface
   instead of a hole, and the same tag the category grid and the
   review cards use. WHILE THE TAG IS VISIBLE THE PAGE IS NOT
   FINISHED.

   Set the term's own thumbnail in wp-admin, or drop a file at
   assets/img/ecosystem/<slug>.webp, and all of this falls away with
   nothing to deploy. Verified 2026-11-13 by rendering the band with
   bodykits included: 4 rows, 4 shots, 2 placeholders (its shot and
   its inline), the other three untouched. */
.sfidx__preview .sfidx__shot[data-placeholder],
.sfidx__inline[data-placeholder]{
	background-color: #101014;
	background-image:
		radial-gradient(120% 90% at 78% 112%, rgba(236,30,121,.42) 0%, rgba(236,30,121,0) 58%),
		repeating-linear-gradient( 45deg, rgba(255,255,255,.03) 0 2px, transparent 2px 4px),
		repeating-linear-gradient(-45deg, rgba(0,0,0,.40)       0 2px, transparent 2px 4px);
}

/* The inline half has no <img> to give it a height, and .sfidx__inline
   is otherwise a bare wrapper — so it needs the one its img carries,
   or it collapses to nothing and the tag has no box to sit in. */
.sfidx__inline[data-placeholder]{
	position: relative;
	height: clamp(180px, 46vw, 300px);
	border-radius: 2px;
}

.sfidx__preview .sfidx__shot[data-placeholder]::after,
.sfidx__inline[data-placeholder]::after{
	content: "PLACEHOLDER — ART TO COME";
	position: absolute;
	top: 14px;
	left: 16px;
	padding: 5px 10px;
	border: 1px solid var(--sfidx-accent);
	border-radius: 999px;
	font-family: var(--sfidx-mono);
	font-size: 10px;
	font-weight: 700;
	letter-spacing: .14em;
	color: var(--sfidx-accent);
	white-space: nowrap;
	pointer-events: none;
}


/* ── a row with nothing to sell yet ─────────────────────────────
   Added 2026-11-16, when Shop by Bike stopped hiding the brands
   that hold no products (operator: "Yes, match home").

   Zontes and Modenas are real brand terms with zero products and
   no product_tag of the same slug to fall back on, so the page
   dropped them when it was converted — while the HOME PAGE's bike
   ribbon showed them anyway, knocked back and reading "Coming
   soon". A visitor saw six brands on the home page and four here.

   THE THREE DEVICES ARE bike-ribbon.css's OWN, ported: knock the
   row back, make it not clickable, and say it in words. Keep the
   two files in step if either is re-pitched.

   THE TEST IS STRUCTURAL, NOT A LIST OF NAMES.
   superfly_sfidx_render() omits .sfidx__count when the count is 0,
   so this selector IS the condition — it stops firing on its own
   the day stock is assigned, with nothing to deploy, and it would
   catch a seventh empty brand added in wp-admin. :has() is what
   makes that possible at all: nothing else lets a rule on the row
   depend on a descendant that is missing. A browser without it
   simply gets an ordinary row, which is the right way round for a
   fallback.

   IT CANNOT FIRE ON SHOP BY CATEGORY as things stand — that page's
   `include` names three terms that all carry stock — but it is
   written against the row rather than against the page, so a
   category that empties out gets the same treatment rather than
   printing "0 products". */
.sfidx__row:not(:has(.sfidx__count)) .sfidx__link{
	/* MULTIPLY THE ENTRANCE, NEVER REPLACE IT. `.sfidx__link`'s own
	   opacity IS `var(--sfidx-in)`, the row reveal — a flat .42 here
	   wins over it, so the row skips the entrance and sits there at
	   .42 while its neighbours are still at 0 waiting to rise. It
	   shipped that way for one deploy and read as the two rows
	   having been left on. Composed, they reveal with everything
	   else and land at .42 instead of 1. */
	opacity: calc(var(--sfidx-in) * .42);

	/* Not clickable: the archive behind it is empty, and a dead end
	   is worse than a row you can see is not ready. The data
	   attribute is on the <li>, NOT on this link, so
	   category-index.js's elementFromPoint walk still finds the row
	   and the floating preview keeps working. */
	pointer-events: none;
}

/* The arrow is the one thing that says "go", so it goes entirely
   rather than merely dimming with the rest. */
.sfidx__row:not(:has(.sfidx__count)) .sfidx__go{
	visibility: hidden;
}

/* ...and the meta slot says it in words, where the price would be.
   On ::after rather than in the markup, because the renderer is
   shared with Shop by Category and this is a state of the row, not
   a thing either shortcode should have to print. */
.sfidx__row:not(:has(.sfidx__count)) .sfidx__meta::after{
	content: "Coming soon";
	color: var(--sfidx-accent);
	font-weight: 600;
	letter-spacing: .14em;
}


/* ── the preview wears the home page's own treatment ────────────
   Added 2026-11-16 (operator: "yes match the home page for both").

   Both shop pages draw the front page's photographs now, and until
   this block they drew them RAW — so the same file was painted one
   way on the home page and another here. Measured mean luma, home
   against preview: the category cards 28/43, 33/50, 36/54, and the
   brand badges 13/18, 13/19, 15/22, 17/25.

   THE TWO SETS WANT OPPOSITE CORRECTIONS, which is the whole reason
   superfly_sfidx_render() takes a $variant and prints .sfidx--cats
   or .sfidx--bikes. One rule cannot serve both:

     cats   studio shots on white. .sfcg knocks them back 34% and
            nothing else — they are already bright and separated.

     bikes  near-black castings with a magenta rim. .sfbr LIFTS them
            brightness(1.55) and THEN darkens 55%. The lift is the
            load-bearing half: bike-ribbon.css records that when they
            were left unlifted the complaint was that they read FLAT,
            not that they read dark — the badge and the chrome sank
            into the casting. Raw, this preview was in exactly that
            rejected state.

   THE NUMBERS ARE .sfcg's AND .sfbr's, duplicated as literals rather
   than read across files — the decision CLAUDE.md records for the
   type ladder and the emboss, because a :root token declared in one
   stylesheet and read from another resolves by load order and fails
   silently. Change them with their own band or neither.

   A FLAT BLACK OVERLAY IS EXACTLY brightness(1 - alpha) — both are a
   straight multiply — so the darkening could be folded into the
   filter. It is kept as a real layer because the brand case is lift
   THEN darken, and composing those as one multiply loses the clamp
   at 255 that the rim light actually hits. Faithful beats tidy.

   WHAT IS DELIBERATELY NOT PORTED: .sfcg__veil, the bottom gradient.
   That exists so a card's NAME has something to sit on. The disc
   carries no text, so it would darken the bottom for nothing. */

.sfidx--cats .sfidx__previewInner::before,
.sfidx--cats .sfidx__inline::before{ background: rgba(0, 0, 0, .34) }

.sfidx--bikes .sfidx__previewInner::before,
.sfidx--bikes .sfidx__inline::before{ background: rgba(0, 0, 0, .55) }

.sfidx--cats .sfidx__previewInner::before,
.sfidx--cats .sfidx__inline::before,
.sfidx--bikes .sfidx__previewInner::before,
.sfidx--bikes .sfidx__inline::before{
	content: "";
	position: absolute;
	inset: 0;

	/* Over the shots, which are absolute and carry no z-index of
	   their own. The previewInner clips to its own 50% radius, so
	   this takes the circle's shape for free. */
	z-index: 2;
	pointer-events: none;
	border-radius: inherit;
}

/* THE LIFT IS THE BRAND HALF ONLY, and only on a real photograph —
   a placeholder shot is a <span> carrying the house carbon and a
   magenta bloom, and brightening that washes out the one thing it
   is there to announce. */
.sfidx--bikes img.sfidx__shot:not([data-placeholder]),
.sfidx--bikes .sfidx__inline:not([data-placeholder]) img{
	filter: brightness(1.55);
}

/* AND A PLACEHOLDER IS NEVER SCRIMMED — bike-wall.css settled that:
   darkening the bloom hides the announcement. `:has()` lets the disc
   drop its overlay for exactly the moment a placeholder is the lit
   shot, and go back to normal for every other row. */
.sfidx__previewInner:has(.sfidx__shot[data-placeholder].is-on)::before,
.sfidx__inline[data-placeholder]::before{
	content: none;
}

/* The inline copy is the touch-screen half and needs a box to hang
   the overlay on; it is otherwise a bare wrapper. */
.sfidx__inline{ position: relative }
