/* ---------------------------------------------------------------------
   96-dogs.css -- the Dogs layer (Phase 2.3, docs/dog-kennel-specification.md)
   ---------------------------------------------------------------------

   WHY THIS FILE EXISTS. inc/enqueue.php glob-loads assets/css/*.css and
   sort()s by filename, so this file loads after every other layer, 90-misc
   and 95-search included. That makes it the one place where the dog pages can
   settle a shared rule without editing the stylesheet that rule lives in --
   80-pets.css interleaves dogs, shop-category and product-detail selectors,
   and 90-misc.css:96 is literally one rule shared by the pet detail layout
   and the shop category layout. Editing either to suit dogs edits two other
   surfaces at the same time.

   THE PRICE OF LOADING LAST, AND THE RULE THAT PAYS IT. A bare selector here
   outranks nothing on specificity but wins every tie on source order, so it
   would silently repaint whatever else happens to match -- /shop/, a product
   page, /pets/ or the guides. So: EVERY selector in this file is scoped to a
   dogs-only root. Today that root is `.pet-dogs-page` in every single rule
   below -- the /{lang}/pets/dogs/ category page, emitted by
   scripts/update-pet-category-pages.php. Nothing here is scoped to
   `.pet-breed-detail-page`: the four breed pages already have a
   dogs-exclusive stylesheet of their own in 82-breed-gallery.css, and no
   shared rule has yet needed overriding for them from out here.
   `.pet-breed-detail-page` stays the second admissible root if that day
   comes; `.pet-pet-detail-page` is NOT one -- six other pet pages carry it.
   Nothing here may be written bare, however obviously dogs-only the class
   name reads.

   WHAT BELONGS HERE. Overrides of a rule that lives in a file this stream
   does not own, and the two components the dogs page grew in Phase 2.3 whose
   natural home (80-pets.css) is shared. Breed-page layout stays in
   82-breed-gallery.css, which is dogs-exclusive already.

   Each override below names the rule it beats and the file:line it lives in,
   so a later hardening pass can consolidate it instead of guessing.
   --------------------------------------------------------------------- */

/* STATUS UNDER ADR-20260906T090000Z (Accepted 2026-09-06): TEMPORARY, NOT
   RETIRED.

   This band exists for a concurrent implementation phase and is meant to empty
   into the layers that own its rules. Eleven rules survive here. Eight style
   classes no other stylesheet styles -- they are dogs-only components sitting
   in a stream layer, and their destination is the pets surfaces. Three override
   a component something else also styles: .pet-info-chip (40-sections.css),
   .pet-specialty-card-photos (80-pets.css) and .pet-help-steps-list li
   (40-sections.css); those are decided one rule at a time, never in a batch,
   because merging two of them back would repaint pages this file does not own.

   The file is not emptied for neatness. A rule with no proven narrower owner
   stays late, and that is correct rather than tolerated. */

/* THE FOUR TRUST CHIPS, under the hero of /{lang}/pets/dogs/.
   The owner fixed this set at exactly four strings on 2026-09-02; the
   builder's comment carries the list and the reason. */
.pet-dogs-page .pet-dogs-trust {
  margin: 1.25rem auto 0;
  max-width: 1180px;
}
/* NOT .pet-section-jump-row, although the declarations below look like its.
   .pet-section-jump-row (85-section-jump.css:57) is a documented shared
   component, and reusing it here was considered and rejected on behaviour,
   not on taste: below 780px (85-section-jump.css:102) that component stops
   being a wrapping row and becomes a horizontally scrolling one --
   `flex-wrap: nowrap`, `overflow-x: auto`, `scroll-behavior: smooth`, plus a
   `padding-inline` and `scroll-padding-inline` measured in
   --pet-section-jump-gutter, a variable declared on its `.pet-section-jump`
   parent, which is also where the matching negative margin that pays for that
   padding lives. Wearing the class without that parent imports the gutter
   padding and the scroller and gets neither the bleed nor the sticky nav they
   exist for. Four non-interactive chips under a hero must wrap onto a second
   line on a phone, not hide the fourth one off-screen behind a scroll nothing
   signals.

   What is left after that is a flex-row-of-pills plus a list reset the jump
   row does not carry at all (it is a <div>; this is a <ul>). If a genuinely
   unscoped "row of chips" primitive is ever factored out of
   85-section-jump.css -- the wrapping row alone, with the scroller behind a
   modifier -- this block should join it and stop existing. */
.pet-dogs-page .pet-dogs-trust-list {
  display: flex;
  flex-wrap: wrap;
  gap: .6rem;
  list-style: none;
  margin: 0;
  padding: 0;
}
/* OVERRIDE (scope, not value): .pet-pets-page .pet-info-chip
   -- 40-sections.css:128.
   The site already has this chip, and these are the same chip, so they carry
   the same class rather than a second one. But that rule is scoped to
   .pet-pets-page: on the pets hub the chip is an absolutely positioned badge
   pinned inside a card. It does not reach this page at all, so without this
   block the four chips render as unstyled list items.

   The declarations are therefore the hub chip's own, minus its positioning --
   a chip in a row is in flow and needs none of it -- and minus its shadow.
   Three deliberate differences from the hub's values:

     - NO box-shadow. The hub chip is an intentionally floating control: it is
       absolutely positioned over a photograph, and the shadow is what lifts
       it off the picture underneath. In normal flow it is none of the four
       uses DS §13 permits (dropdown/mega-menu, modal/dialog, intentionally
       floating control, temporary overlay), and it would add a second
       elevation level to a page that already carries `box-shadow: 0 16px
       34px` on the specialty cards -- DS §13's "multiple competing elevation
       levels on one screen" verbatim. It also made four non-interactive pills
       look more pressable than the actually-interactive .pet-filter-chip
       quick-jump row further down the same page.
     - the type goes from .78rem to .85rem, because 12.5px is a label riding a
       picture and this is a line a visitor is meant to read;
     - the padding grows with it.

   FOR THE HARDENING PASS: the fix is to lift the hub rule's appearance out of
   .pet-pets-page into an unscoped .pet-info-chip and leave the absolute
   positioning AND the shadow behind a modifier (.pet-info-chip-overlay or the
   like) -- the shadow belongs to the overlay case, not to the chip. This
   block then deletes down to the two font/padding lines -- and note that it
   would at that moment also need `position: static`, which is deliberately
   absent today because nothing here sets `position` and a declaration that
   changes nothing is noise. */
.pet-dogs-page .pet-dogs-trust .pet-info-chip {
  background: var(--pet-surface);
  border-radius: 999px;
  color: var(--pet-primary);
  font-size: .85rem;
  font-weight: 800;
  padding: .45rem .95rem;
}

/* THE BREEDS ROW OF FOUR, AND THE PRE-ORDER BAND.
   The reference composes the kennel as a row of four breed cards with the
   pre-order offer as a band beneath them, and the specification adopts both.
   Production already had the right CONTENT -- five specialty cards, four of
   which link 1:1 to the four real breed pages -- laid out as three-up, which
   put the Doberman and the pre-order card alone on a second row. This is that
   same content in the reference's arrangement. No card is added, removed or
   relinked, and the quick-jump group still lists the same five chips. */

/* OVERRIDE: .pet-specialty-grid-links article { flex: 0 1 calc((100% - 2rem) / 3) }
   -- 80-pets.css:126.
   Four across instead of three, so the four breed pages read as one row.
   `gap: 1rem` comes from .pet-specialty-grid (80-pets.css:58), which the same
   element also carries, hence three gaps rather than two in the arithmetic.

   Bounded to "wider than the two-up breakpoint" on purpose: 90-misc.css:261
   already takes this grid to two-up at <=980px and 90-misc.css:272 to one-up
   at <=560px, and those are the right numbers for a card this size. A fourth
   column on a tablet would paint each breed card about 170px wide. Widening
   the breakpoint would also reach /pets/rodents/, which shares this grid and
   is not this stream's.

   WHY `not all and (max-width: 980px)` AND NOT `min-width: 981px`. Viewport
   widths are not integers. `min-width: 981px` against 90-misc's `max-width:
   980px` leaves every fractional width between them -- 980.5px, which browser
   zoom and a fractional device pixel ratio both produce -- matching NEITHER
   rule, so the grid fell back to the un-overridden three-up in that band; and
   a three-up dogs grid is precisely the layout this block exists to replace,
   with the Doberman and the pre-order card sharing a second row. Negating the
   other rule's own condition is the only form that cannot leave a hole: it is
   true at exactly the widths `(max-width: 980px)` is false, by construction,
   whatever the fraction and whatever the media type. If 90-misc.css's 980px
   ever moves, this must move with it -- it is one breakpoint written twice,
   and writing the second copy as a negation is what stops the two from
   drifting apart by a rounding error again. */
@media not all and (max-width: 980px) {
  .pet-dogs-page .pet-specialty-grid-links article {
    flex-basis: calc((100% - 3rem) / 4);
  }
}
/* OVERRIDE: the same rule, and 90-misc.css:261 / 90-misc.css:272 with it.
   The fifth card is the pre-order offer, and a band is what makes it read as
   a different KIND of thing from the four breeds above it rather than a fifth
   breed we forgot to photograph. Full width at every viewport, so it is
   declared outside the media query above and beats all three of those rules
   on specificity (0-3-0 against 0-1-1) rather than on source order. */
.pet-dogs-page .pet-specialty-grid-links .pet-dogs-preorder-card {
  flex-basis: 100%;
}
/* OVERRIDE: .pet-specialty-card-photos { grid-template-columns: repeat(3, 1fr) }
   -- 80-pets.css:178, for this one card only.
   Three equal fractions of whatever card they are in is correct for a card
   sized by the row. The band is not sized by the row: it IS the row, ~1180px,
   so its content box is ~1137px and the same three tracks paint ~375px each,
   while a breed card at four-up is ~283px wide, ~240px of content, and paints
   ~76px per thumbnail. Five times the linear size and twenty-five times the
   area -- the card for the breeds Pet Town does NOT stock became the largest
   photography in the section and read as its hero, inverting the hierarchy
   the row of four exists to create. (The Doberman card's lone frame is
   already capped at 200px by 80-pets.css:184 for the same reason.)

   Capping the strip rather than the card keeps the band a band: it still
   spans the row, still carries its heading, its text and its contact-form CTA
   at full width. 380px over three tracks and two .4rem gaps is ~122px a
   thumbnail -- larger than the ~76px of a four-up breed card, smaller than
   the ~176px the same breed strip paints at the two-up breakpoint and than
   the Doberman's 200px -- so it sits inside the band the four cards already
   occupy instead of above it. Below ~380px of available width the cap stops
   biting and the shared 1fr tracks shrink exactly as they do in a breed card,
   so the phone layout is untouched. No `margin-inline: auto`: the strip stays
   on the card's inline start, which follows the writing direction on its own
   and needs no [dir="rtl"] rule. */
.pet-dogs-page .pet-specialty-grid-links .pet-dogs-preorder-card .pet-specialty-card-photos {
  max-width: 380px;
}

/* THE FIVE-STEP PURCHASE JOURNEY.
   Reuses the numbered-step list the pets hub already ships --
   .pet-help-steps / .pet-help-steps-list, 40-sections.css:146-150 -- which is
   unscoped and so reaches this page as it stands, counter badge included.
   Only the two things a five-step strip needs differently are stated here.

   (The block this styles is not the hub's "how we help" list returning to
   this page under a new name: different process, different copy, no
   id="how-we-help". The builder's comment on $dogs_journey has the detail.) */

/* OVERRIDE: .pet-help-steps { margin: 2.5rem 0 0 } -- 40-sections.css:146.
   On the hub that rule sits directly in the page container and needs no width
   of its own. Here the page container is --pet-container-brand (1280px) while
   every block this section follows -- .pet-shop-specialties (80-pets.css:22)
   and .pet-pet-gallery (80-pets.css:234) -- is 1180px and centred. Without
   this the strip would be the one section on the page 100px wider than the
   ones above it, hanging 50px past both edges of the gallery. */
.pet-dogs-page .pet-dogs-journey {
  margin: clamp(2rem, 4vw, 3.5rem) auto 0;
  max-width: 1180px;
}

/* NO COLUMN-COUNT OVERRIDE, DELIBERATELY. The shared
   `grid-template-columns: repeat(auto-fit, minmax(240px, 1fr))`
   (40-sections.css:148) runs here unchanged, at every width.

   This file used to force `repeat(5, minmax(0, 1fr))` above 1024px so all
   five steps stayed on one line and the handover step did not read as an
   afterthought on a row of its own. It is gone because it could not deliver
   that at any width this strip actually gets. The strip is capped at 1180px
   by the rule above, so five tracks with the inherited `gap: 1rem` and the
   shared step's `padding: 1.4rem 1.6rem` leave (1180 - 4x16)/5 - 51.2 = 172px
   of text column at the widest the strip is ever painted, and 131px at a
   1024px viewport. Both are below the 240px floor the shared component itself
   declares, and below the ~150px the comment on the rule immediately below
   already names as too narrow for a label and a sentence.

   1024px was a regression point rather than an improvement point, too: at
   1023px auto-fit gives three columns of ~314px and puts the long handover
   step on a row where it stretches to the full width; crossing to 1024px made
   the strip TALLER and every column NARROWER at the same time, so dragging a
   window wider made the layout worse.

   What auto-fit does instead, and what this page accepts: four columns at
   1180px, with the handover step alone on the second row stretched across it
   -- which is the only step long enough to want that space, and the one the
   /terms/ link underneath is about. The five-on-one-line composition is not
   reachable from a 1180px strip without taking the text below the shared
   component's own minimum, so it is not attempted here. */

/* OVERRIDE: .pet-help-steps-list li { align-items: center; display: flex }
   -- 40-sections.css:149.
   Kept, and kept unconditionally, because the reason for it is the CONTENT of
   a step and not the width of the strip. A hub step is one short sentence
   beside its badge, so a row is right there. Each step here is two parts -- a
   bold label with a sentence under it -- and a row spends the same 56px (a
   40px badge plus the 1rem gap) on the badge whatever the column is: 23% of
   the narrowest column auto-fit will ever produce here (240px, ~189px after
   the step's own padding, leaving ~133px of text) and still 19% of the ~283px
   four-up column at 1180px. At the narrow end a row wraps the label itself
   against the badge; at the wide end it indents a two-part block for nothing.
   Stacking gives the full column width to both parts at both ends, so there
   is no width at which the row form is the better one and no media query is
   needed to say so.

   flex-start rather than the inherited centre so the badge sits on the reading
   edge above its text; it follows the writing direction on its own, so the
   Hebrew page needs no [dir="rtl"] rule for it. */
.pet-dogs-page .pet-help-steps-list li {
  align-items: flex-start;
  flex-direction: column;
}
.pet-dogs-page .pet-dogs-journey-step {
  display: grid;
  gap: .3rem;
}
.pet-dogs-page .pet-dogs-journey-step b {
  font-weight: 800;
}
.pet-dogs-page .pet-dogs-journey-note {
  color: var(--pet-muted);
  font-size: .85rem;
  margin: 1rem 0 0;
}
