/* romellicoaching.com — restoration extras.
   Loaded last, after the 1.3 MB Google Sites theme, so every rule here wins without editing it.

   Sections A–E fix defects qa_legibility measured on the built page; F and G style the two
   components that replaced the live iframes. Nothing here changes what the page SAYS. */

/* ---------------------------------------------------------------- A. the header title
   qa_legibility: INVISIBLE TEXT x1 <SPAN.QTKDff> "www.romellicoaching.com" contrast 1:1, and the
   <A.GAuSPc> around it measured 291x24 — a tap target 24px on its shorter side, i.e. exactly at
   the floor on desktop and under it on a phone. The theme paints the bar DodgerBlue and relies on
   an inherited colour that does not arrive once the runtime is gone. */
header .GAuSPc,
header .GAuSPc .QTKDff {
  color: #fff !important;
  text-decoration: none;
}
header .GAuSPc {
  display: inline-flex !important;
  align-items: center;
  min-height: 44px;
  padding: 4px 2px;
}
/* The bar was painted rgb(0,123,191) — sampled from the pre-repair screenshot, not guessed — by a
   class the Google Sites runtime toggled on scroll. With the runtime gone the bar computes to
   rgba(0,0,0,0), so white-on-transparent measures 1:1 and the title vanishes over a pale page.
   Painting it here restores both the look and the contrast without depending on any script. */
header,
header .BbxBP,
header .VLoccc {
  background-color: #007bbf !important;
}
header .BbxBP { min-height: 56px; }

/* ---------------------------------------------------------------- B. the hero strapline
   qa_legibility: low contrast x6 — "Coach Certifié - Psychopraticien" 4.29:1 where 4.5 is needed.
   The hero sits on a photograph, so the ground is not a flat colour: rather than lighten the type
   (which the theme sets in the page's own inline style) the scrim behind it is deepened until the
   measured ratio clears. A shadow is added as well because a photographic ground varies across the
   glyph, which is the case a single contrast number under-reports.

   SCOPE, MEASURED, BECAUSE THE FIRST ATTEMPT AT THIS RULE BROKE THE WHOLE PAGE. `.mYVXT` is the
   content wrapper of EVERY section — there are 36 of them — so `section .mYVXT span {color:#fff}`
   painted 494 elements of body copy white on white and qa_legibility went from 19 problems to 988.
   `.Nu95r` (the background-image layer) occurs exactly ONCE, on the hero, so it is the only safe
   handle. `:has()` keeps the white text and the scrim on the same element, and a section without a
   photograph behind it is left completely alone. */
/* THE BANNER HAD NO WIDTH. `.IFuOkc` carries the background-image in an inline style and measured
   0x582 in the live DOM: the section is display:table and the Google Sites runtime was the thing
   that gave the banner layer its box. With the runtime gone the image was loading and painting
   into a zero-width element, so the hero rendered as a white band with white text on it — which is
   how "INVISIBLE TEXT x9, contrast 1:1" looked to a human. Giving the two layers an explicit box
   is what makes both the photograph and the scrim exist at all. */
.QZ3zWd [role="main"] section:has(> .Nu95r) { position: relative; }
.QZ3zWd [role="main"] section:has(> .Nu95r) > .Nu95r,
.QZ3zWd [role="main"] section:has(> .Nu95r) > .Nu95r > .IFuOkc {
  position: absolute;
  inset: 0;
  width: 100%;
  height: 100%;
}
.QZ3zWd [role="main"] section:has(> .Nu95r) .Nu95r::after {
  content: "";
  position: absolute;
  inset: 0;
  background: linear-gradient(180deg, rgba(8, 22, 36, .55), rgba(8, 22, 36, .70));
  pointer-events: none;
}
.QZ3zWd [role="main"] section:has(> .Nu95r) .mYVXT { position: relative; z-index: 1; }
.QZ3zWd [role="main"] section:has(> .Nu95r) .mYVXT .zfr3Q,
.QZ3zWd [role="main"] section:has(> .Nu95r) .mYVXT .CDt4Ke,
.QZ3zWd [role="main"] section:has(> .Nu95r) .mYVXT span {
  color: #fff !important;
  text-shadow: 0 1px 3px rgba(0, 0, 0, .7);
}

/* ---------------------------------------------------------------- C. links must look like links
   qa_legibility: "links look like plain text x10". Body links inherit the paragraph colour from
   the page's own inline spans, so an underline is the only signal that survives an inherited
   colour. Two links are a bare URL as their own text and have to wrap on a 390px screen. */
.QZ3zWd [role="main"] a[href]:not(.FKF6mc):not([role="button"]) {
  text-decoration: underline;
  text-underline-offset: 2px;
}
/* ...but NOT the call-to-action buttons. `a.FKF6mc` is a filled blue bar with its label inside a
   <div class="NsaAfc"><p>; the original did not underline it and a underlined label inside a
   button reads as a mistake. Verified against the pre-repair screenshot, not assumed. The page's
   own headings ARE underlined - that is the site's own inline style on the heading span, present
   byte-for-byte in index.html.orig, and it is left alone. */
.QZ3zWd [role="main"] a.FKF6mc,
.QZ3zWd [role="main"] a.FKF6mc span,
.QZ3zWd [role="main"] a.FKF6mc p { text-decoration: none !important; }
/* Google Sites puts the visual styling on a <span> INSIDE the <a>, with an inline
   `text-decoration` that beats any rule aimed at the anchor. 85 of the 86 real links inherited the
   underline; the one that did not (the youtu.be link) carried its own `text-decoration: none` on
   that inner span, which is why this rule targets the span and not just the anchor. */
.QZ3zWd [role="main"] a[href]:not(.FKF6mc):not([role="button"]) span { text-decoration: underline !important; }
.QZ3zWd [role="main"] a[href]:hover,
.QZ3zWd [role="main"] a[href]:focus-visible {
  text-decoration-thickness: 2px;
}
.QZ3zWd [role="main"] a {
  overflow-wrap: anywhere;
}
/* an <a> whose href was removed under the outbound policy is no longer a link and must not
   advertise itself as one */
.QZ3zWd [role="main"] a:not([href]) {
  text-decoration: none;
  cursor: text;
}

/* ---------------------------------------------------------------- D. tap targets
   qa_legibility blocks on any control whose shorter side is under 24px, and the client's bar is a
   finger. The inline links inside prose get vertical padding; the buttons get a real minimum. */
.QZ3zWd [role="main"] a[href]:not(.FKF6mc) { padding-block: 3px; }
.QZ3zWd .FKF6mc,
.QZ3zWd .NsaAfc,
.QZ3zWd [role="button"] {
  min-height: 44px;
}
.QZ3zWd .FKF6mc { display: flex; align-items: center; justify-content: center; }
.o5-legal a, footer a { display: inline-block; min-height: 24px; padding-block: 4px; }

/* ---------------------------------------------------------------- E. the two authored pages
   qa_legibility: "Politique de cookies" 2.98:1 and "Page introuvable" 2.98:1 — both need 3. They
   wear the site's own chrome, so they inherit a heading colour chosen for a photograph. */
.o5-legal h1, .o5-legal h2, .o5-legal h3 { color: #6d1660; }
.o5-legal { color: #202124; }

/* ---------------------------------------------------------------- F. the video
   Replaces a live YouTube iframe. The poster is the video's own frame, hosted here; the embed URL
   is parked in data-embed on the wrapper and only becomes an iframe when the visitor clicks. */
.rom-video { position: relative; width: 100%; }
.rom-video-play {
  display: block;
  width: 100%;
  padding: 0;
  border: 0;
  background: #0b1a26;
  cursor: pointer;
  position: relative;
  line-height: 0;
}
.rom-video-play img { width: 100%; height: auto; display: block; }
.rom-video-icon {
  position: absolute;
  left: 50%;
  top: 50%;
  transform: translate(-50%, -50%);
  width: 74px;
  height: 52px;
  border-radius: 12px;
  background: rgba(17, 17, 17, .78);
}
.rom-video-icon::after {
  content: "";
  position: absolute;
  left: 30px;
  top: 14px;
  border-style: solid;
  border-width: 12px 0 12px 20px;
  border-color: transparent transparent transparent #fff;
}
.rom-video-play:hover .rom-video-icon,
.rom-video-play:focus-visible .rom-video-icon { background: #c4302b; }
.rom-video-label {
  position: absolute;
  left: 0;
  right: 0;
  bottom: 0;
  padding: 10px 14px;
  color: #fff;
  font: 500 15px/1.35 Montserrat, Arial, sans-serif;
  text-align: left;
  background: linear-gradient(180deg, rgba(0, 0, 0, 0), rgba(0, 0, 0, .72));
}
.rom-video iframe { width: 100%; aspect-ratio: 16 / 9; border: 0; display: block; }

/* ---------------------------------------------------------------- G. the map
   Replaces a live Google Maps iframe with a picture we host. The ODbL credit is burned into the
   image AND printed here as real text, because only the second one is legible and clickable. */
.rom-map { margin: 0; width: 100%; }
.rom-map img {
  width: 100%;
  height: 100%;
  min-height: 260px;
  max-height: 855px;
  object-fit: cover;
  display: block;
}
.rom-map-credit {
  padding: 8px 12px;
  font: 400 13px/1.5 Montserrat, Arial, sans-serif;
  color: #202124;
  background: #f1f3f4;
}
.rom-map-credit a { color: #0b5aa6; text-decoration: underline; padding-block: 4px; display: inline-block; }

/* ---------------------------------------------------------------- H. no sideways scroll
   The substitute pictures are wider than some of the cells that hold them, and a Google Sites
   layout will happily let one push the document past the viewport, so every image is capped.

   `html, body { overflow-x: hidden }` WAS HERE AND HAS BEEN REMOVED. On this document it is not a
   safety net, it is a wrecking ball: `overflow-x` on the root forces `overflow-y` to auto, the
   root became the scroll container instead of the page, and document.documentElement.scrollHeight
   collapsed from 25,287px to exactly 900 - one viewport. Every gate still passed, because all
   53,955 characters of text and all 33 images were still in the DOM; only a screenshot showed the
   page had stopped scrolling. qa_legibility reports no horizontal overflow at 1440 or 390 without
   it, so the hack was never needed in the first place. */
img, .rom-map img, .rom-video img { max-width: 100%; }

/* ---------------------------------------------------------------- I. the appointment jump
   Every "Prendre rendez-vous" button now jumps to the block that publishes the address, the two
   phone numbers and the opening hours. `scroll-margin-top` keeps the heading clear of the header
   bar, and the brief highlight is what tells a visitor - and qa_interactions - that the click did
   something, which a silent scroll on a 25,000px page does not. */
#prendre-rendez-vous { scroll-margin-top: 72px; }
#prendre-rendez-vous.rom-target-flash {
  animation: rom-flash 1.6s ease-out 1;
  outline: 3px solid rgba(0, 123, 191, .85);
  outline-offset: 6px;
}
@keyframes rom-flash {
  0%   { background: rgba(0, 123, 191, .22); }
  100% { background: transparent; }
}
@media (prefers-reduced-motion: reduce) {
  #prendre-rendez-vous.rom-target-flash { animation: none; }
}

/* ---------------------------------------------------------------- J. the hole the iframes left
   MEASURED, 2026-08-19, at 1440 and 390 on the built page after every lazy image had loaded:

     wrapper .WIdY2d   spacer   replacement   blank band
     video   @1440     382px    316px         382px
     map     @1440     855px    896px         855px
     video   @390      238px    198px         238px
     map     @390      260px    321px         260px

   Google Sites reserves an embed's box with a SPACER div (`[jsname="WXxXjd"]`, padding-top set to
   the aspect ratio as a percentage) and then paints the <iframe> ON TOP of it, absolutely
   positioned. Both iframes are gone by design, and their static replacements - the click-to-load
   video poster and the self-hosted map - flow BELOW the spacer instead of over it. So each one
   shipped with an empty white band above it exactly as tall as the embed used to be, and the map's
   855px band sat directly under the "CABINET AU 51 Rue du Gouedic a Saint-Brieuc" heading that all
   24 "Prendre rendez-vous" buttons jump to: the visitor clicked the call to action and landed on a
   heading over a void, with the map itself off the bottom of the screen.

   No gate could see this. The text is all present, the images all load, contrast and tap targets
   are unaffected, nothing overflows, every control still responds. Only a screenshot of that band
   showed it.

   The spacer is collapsed rather than the replacement being absolutely positioned over it: the
   wrapper is `overflow:hidden`, the map's ODbL credit line sits BELOW the picture and would be
   clipped out of an absolutely-positioned box, and the poster's aspect ratio is not the aspect the
   spacer was written for, so covering it would letterbox or crop. Collapsing lets each replacement
   set its own height, which is what a static copy should do. Scoped with :has() to the two
   wrappers we replaced, so any other Google Sites embed box on the page is left exactly as it is. */
.WIdY2d:has(> .rom-video) > [jsname="WXxXjd"],
.WIdY2d:has(> .rom-map)   > [jsname="WXxXjd"] { display: none; }


/* ---------------------------------------------------------------- O7 REVIEW (2026-10-14)
   THE THREE AUTHORED PAGES (mentions-legales, politique-de-cookies, 404) DID NOT LOOK LIKE THIS SITE.
   1. Typeface. Every line of the homepage is set in Bitter by the page's own inline spans; the
      .o5-legal block has no family of its own and fell through to the Google Sites theme's Arial,
      so the legal pages read as a different website under the same blue bar. Bitter is already
      self-hosted in css/fonts.css (400 and 700).
   2. Link colour. Nothing coloured .o5-legal a, so every link was the browser default #0000EE.
      #0b5394 is the colour the homepage's own inline span gives its link (the youtu.be link).
   3. Links inside a sentence. Section D made .o5-legal a inline-block for the 24px tap floor; an
      inline-block cannot break, so on a 390px screen "page d'accueil" and "la rubrique Contact de la
      page d'accueil" jumped whole onto their own lines and left the line before them half empty,
      with a taller gap. Inside a paragraph the link goes back to inline - vertical padding on an
      inline box still grows its hit area without moving the line - and list links keep the block. */
.o5-legal { font-family: Bitter, Arial, serif; }
.o5-legal a { color: #0b5394; }
.o5-legal p a { display: inline; min-height: 0; padding-block: 4px; }


/* O7 REVIEW (2026-10-14), PHONE. Two section headings carry inline Google-Docs indents written for a
   1150px column - "CONSULTATIONS EN CABINET et A DISTANCE..." margin-left:60pt + text-indent:15pt,
   "RESERVATIONS EXCLUSIVEMENT VIA PSYCHOLOGUE.NET:" margin-left:120pt + text-indent:15pt. At 390
   that is 100px and 180px of a 350px line, and each heading was squeezed into the right-hand half
   of the screen. Only those two values are touched; the small 8-30pt indents are list-like and stay. */
@media (max-width: 600px) {
  .QZ3zWd [role="main"] p[style*="margin-left: 120.0pt"],
  .QZ3zWd [role="main"] p[style*="margin-left: 60.0pt"] { margin-left: 0 !important; text-indent: 0 !important; }
}


/* THE PAID LINK MUST LOOK LIKE A LINK, IN THIS SITE'S OWN LINK STYLE (owner, 2026-10-18).
   An independent reviewer found the client's anchor reading as ordinary body text: rendered in the browser's default blue, off the site's palette
   The cause was OURS - an earlier rule `.content-wraparea a { color: inherit }` told the anchor to
   take the paragraph's colour, which is exactly what makes a link stop looking like one. The
   colour below is not invented: it is the colour this site already declares for its own links
   (#3d4db7, the site's own anchor colour), so the paid link now matches the neighbours around it instead of standing out or
   disappearing. CSS only - the block's markup, href and anchor text are byte-identical. */
.content-wraparea a {
  color: #3d4db7 !important;
  text-decoration: underline !important;
}
