wrolpi/app/src/themes/tokens.css
Roland 5873e65e8c
review: the color-scheme half of this was already there
Acting on review of #644.

The largest finding came out of a suggestion rather than being in it.  The
reviewer asked for coverage of the `color-scheme: dark` rule, on the grounds
that dropping it would leave night's track wrong with no failing test.  Written
and passing -- then mutated, by pointing the selector at a type that does not
exist, and the tests stayed green.

Mantine's own stylesheet puts `color-scheme: var(--mantine-color-scheme)` on
`:root`, and ThemeProvider drives it with `forceColorScheme`.  The property is
inherited, so a range slider on a dark theme has had a dark track all along and
the rule added here changed no pixel.  It is deleted, and the claim about "the
UA's pale slab" is out of the comments and the gallery copy with it.

The tests stay, re-aimed at what actually supplies it: breaking
`forceColorScheme` fails all three, which is worth guarding because it lives in
another module and every accent assertion passes without it.

The other suggestions:

- The gallery slider misrepresented the control it stands for -- it invented an
  "Error" step, dropped Trace and ran 0..4, where Settings maps 1..5 onto
  Critical / Warning / Info / Debug / Trace.  A gallery whose purpose is showing
  the real control is worse than useless when it shows a different one.  Keyed
  to match `levelNameMap` exactly, and it starts at Info so both halves of the
  track are visible.

- The "never leaves it at the UA default" test restated the four above it:
  `toRgb` throws on anything unresolved, and the computed value of
  `accent-color: auto` is the keyword, so those already fail on an unstyled
  slider.  Replaced with a constraint they do not cover -- `accent-color` is
  inherited, so a text field beside the slider must still read `auto`, which is
  what says the accent comes from this rule and not from a tinted document.

- The comment implied native checkboxes and radios are on screen the way
  sliders are.  They are not: those are Mantine's and Progress is a div, so
  three of the four selectors are a guard rather than a fix.  Said so.

Nits: four product sliders, not five -- the fifth is the sample itself.  And
the progressive-enhancement limit is now stated where the rule is, including
what a complete fix would take.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-08 18:05:54 -06:00

440 lines
17 KiB
CSS

/*
* WROLPi theme tokens.
*
* Every color in the interface resolves through these custom properties, so a
* theme is a table of values and components never branch on the current mode.
* See ui-design.md for the design rules and ui-migration-plan.md for the
* migration this is part of.
*
* Four built-in themes:
* light - default
* dark - default when the OS prefers dark
* night - night-vision preserving; red hues ONLY. Status is encoded by
* brightness, danger by dashed borders. Explicit choice only.
* amber - monochrome amber terminal palette. Explicit choice only.
*
* `data-theme` is stamped on <html> by an inline script in index.html before
* first paint, then maintained by ThemeProvider.
*/
/* Light is also the fallback when `data-theme` is missing or unrecognized. */
:root,
html[data-theme="light"] {
--bg: #f3f4f3;
--panel: #ffffff;
--head: #e4e6e5;
--border: #8f9698;
/* The rule between table rows: softer than --border, which reads as a grid. */
--table-line: #eceeed;
--text: #22282a;
--muted: #5f686a;
--red: #b23a3a;
--amber: #95661a;
--orange: #b05a24;
--yellow: #96821c;
--olive: #6f7a24;
--green: #3f7a2e;
--teal: #2a7d78;
--blue: #20629c;
--violet: #5c4fa8;
--purple: #7b3f9c;
--pink: #b03a72;
--brown: #7a5233;
--grey: #6a7072;
--black: #22282a;
--white: #ffffff;
/* Text drawn on top of a filled color button. */
--btn-text: #ffffff;
/* Switch thumbs and slider handles. */
--knob: #ffffff;
/* Roles: light has hues to spend, so each role is simply the hue that means it. */
--neutral: var(--grey);
--info: var(--blue);
--success: var(--green);
--warning: var(--amber);
--danger: var(--red);
}
/*
* Semantic roles.
*
* What a color MEANS, separate from what hue it happens to be. A call site asks
* for `--danger`, not `--red`, and each theme decides how danger looks.
*
* This matters because two of the four themes have no hues to spend. Night is red
* and nothing else; amber is amber and nothing else. There, a role is a step on a
* BRIGHTNESS ramp -- and it has to be, because `--orange` in night is byte-identical
* to `--text` and `--yellow` is byte-identical to `--amber`, so a call site choosing
* between hue names there is choosing between synonyms. That is why all four toast
* kinds looked the same in night, and why a warning and an error were the same pixel.
*
* In EVERY theme:
* - the five roles resolve to five distinct colors;
* - every role except neutral clears 3:1 against the panel. Neutral sits below the
* rest deliberately -- for `pending` and disabled, being dim IS the signal.
*
* In the MONOCHROME themes only (night and amber):
* - the order neutral < info < success < warning < danger is strictly increasing in
* contrast against the panel, so severity reads as brightness with no hue to help;
* - BOTH warning and danger reach that theme's own `--text` brightness. Anything that
* means "look at this" has to be louder than the prose around it, and with no hue to
* carry the signal, louder means brighter. Warning used to sit just under `--text`
* (4.91 against 4.98 in night), so a machine under load rendered marginally quieter
* than a healthy one.
*
* Getting warning and danger above `--text` costs a little saturation, and it is worth
* knowing exactly how little: red contributes only 21% of luminance, so in night nothing
* above ~4.9:1 can be a pure hue -- `#ff0000` itself is only 4.99:1 there. `--text` is
* already `#e04a4a`, carrying 74 of green and blue. The new `--warning` `#f54040` carries
* 64, LESS than the text beside it; only `--danger` `#ff5757` goes further, at 87. The
* theme's red-only character survives; a pure-hue ramp was never available.
*
* Those last two are NOT true of light and dark and must not be made true. There a role
* is a hue, and the contrast order is an accident of which hues were chosen -- blue is
* darker than green on white, so light "fails" a brightness ramp while looking perfectly
* correct. Reordering light's hues to satisfy a ramp would be changing the design to
* satisfy a test.
*
* `ui-layout.cy.js` asserts exactly this split, per theme, in a real browser.
*
* `--primary` is deliberately NOT one of these. "The color of a primary button" and
* "the color that means informational" are different jobs, and a user who recolors
* their buttons should not thereby recolor their messages.
*/
:root {
--primary: var(--blue);
}
html[data-theme="dark"] {
--bg: #1c1f21;
--panel: #26292b;
--head: #141617;
--border: #4a5052;
/* The rule between table rows: softer than --border, which reads as a grid. */
--table-line: #343a3c;
--text: #e6e8e6;
--muted: #93999b;
--red: #e06c6c;
--amber: #d6a656;
--orange: #dd8a55;
--yellow: #d6c156;
--olive: #b5bd63;
--green: #7fb069;
--teal: #63bdb6;
--blue: #5c9dd6;
--violet: #9a8fe0;
--purple: #b884d6;
--pink: #dd7fae;
--brown: #b08a68;
--grey: #93999b;
--black: #0d0f10;
--white: #e6e8e6;
--btn-text: #1c1f21;
--knob: var(--text);
/* Roles: as light -- a hue each. See the note above the :root block. */
--neutral: var(--grey);
--info: var(--blue);
--success: var(--green);
--warning: var(--amber);
--danger: var(--red);
}
/*
* Night: red hues only. A single non-red pixel defeats the purpose of the
* mode, so the whole palette collapses onto one red ramp ordered by
* brightness — dim for structural roles, brightest reserved for danger.
*/
html[data-theme="night"] {
--bg: #0a0000;
--panel: #160404;
--head: #000000;
--border: #5c1010;
/* The rule between table rows: softer than --border, which reads as a grid. */
--table-line: #3a0d0d;
--text: #e04a4a;
--muted: #7a2525;
--red: #ff3b3b;
--amber: #d94040;
--orange: #e04a4a;
--yellow: #d94040;
--olive: #5c1a1a;
--green: #b03030;
--teal: #963030;
--blue: #8a2828;
--violet: #7a2929;
--purple: #6e2020;
--pink: #c04848;
--brown: #451212;
--grey: #7a2525;
--black: #2a0808;
--white: #ff8a8a;
--btn-text: #0a0000;
--knob: var(--text);
/* Roles: one hue, five brightnesses; saturation climbs with severity too. */
--neutral: #992c2c;
--info: #c02929;
--success: #e02323;
--warning: #f54040;
--danger: #ff5757;
}
/* Amber: a monochrome amber terminal palette, same flat style as the rest. */
html[data-theme="amber"] {
--bg: #0d0600;
--panel: #1a0f00;
--head: #000000;
--border: #5c3a08;
/* The rule between table rows: softer than --border, which reads as a grid. */
--table-line: #3d2708;
--text: #ff9500;
--muted: #8a5810;
--red: #ffc040;
--amber: #e08a20;
--orange: #f09428;
--yellow: #e08a20;
--olive: #6e4610;
--green: #c97a10;
--teal: #8a581c;
--blue: #a86618;
--violet: #7a4c14;
--purple: #6e4410;
--pink: #d08430;
--brown: #452708;
--grey: #8a5810;
--black: #241504;
--white: #ffc875;
--btn-text: #0d0600;
--knob: var(--text);
/* Amber assigns the mono face to the body role; that is its character. */
--font-body: var(--font-system-mono);
/* Roles: scaled to amber's own headroom (--text is 8.6:1, night's is 5.0:1). */
--neutral: #815825;
--info: #b06e1f;
--success: #d37c11;
--warning: #fa9f31;
--danger: #ffb760;
}
/* ------------------------------------------------------------------------- */
/* Shell */
/* ------------------------------------------------------------------------- */
/*
* Base element styling.
*
* A framework stylesheet used to supply all of this, and removing it took the lot with
* it -- links lost their color and gained the browser's underline and visited purple, and
* every padded element silently changed size when `box-sizing: border-box` went away.
* None of it belonged to a framework; it belongs here, expressed in tokens.
*/
/* ------------------------------------------------------------------------- */
/* Interface scale */
/* ------------------------------------------------------------------------- */
/*
* How large the interface is, overall.
*
* Measured against the build users had before this library, at 1280px, the drift was not
* one number: its headings were h1 28 / h2 24 / h3 18 / h4 15 / h5 14 against our
* 26 / 21 / 17 / 14 / 13, and its video card a fixed 290px with an 18px title against
* our 238px with a 13px one -- while our BODY text and chrome went the other way,
* a 16px base against its 14 and 42px buttons against 36. No per-component nudge
* reproduces what a user comparing the two sees, which is the whole page.
*
* Ten percent is not a fudge factor: our heading scale times 1.1 is
* 28.6 / 23.1 / 18.7 / 15.4 / 14.3, which is the old scale to within a pixel at
* every step.
*
* The lever is the root font-size, and everything that should grow with it is written
* in `rem`. That is Mantine's own mechanism -- its size declarations are
* `calc(Xrem * var(--mantine-scale))`, and `theme.scale` exists to CANCEL a customised
* root font-size, so it must stay at 1 here.
*
* A percentage rather than a px value, so a user who has set a larger default font in
* their browser keeps it, scaled, instead of having it silently overridden.
*
* `zoom: 1.1` on the root was tried first and rejected. It is one line and it works,
* but it scales at paint time: every 1px hairline becomes 1.1px and softens on a 1x
* display, and viewport units do not shrink to compensate, so `100vh` paints 110% of
* the screen -- a 95vw modal overflowed sideways and the CBZ viewer's fullscreen
* overflowed downwards. Rem scales real lengths, leaves borders crisp and leaves
* viewport units alone.
*
* Borders, outlines and shadows are deliberately NOT in rem. A hairline is a hairline
* at any text size; ui-scale.test.js enforces both halves of that.
*/
:root {
--ui-scale: 1.1;
}
html {
box-sizing: border-box;
font-size: calc(100% * var(--ui-scale));
}
*,
*::before,
*::after {
box-sizing: inherit;
}
body {
background: var(--bg);
color: var(--text);
font-family: var(--font-body);
}
/*
* Links take the primary token rather than the browser's blue, and carry no underline
* until hovered -- a great many of them wrap cards, posters, statistics and icons, where
* a permanent underline is noise. `--blue` remaps per theme, so night gets a dim red
* link and amber an amber one without another rule.
*/
a {
color: var(--blue);
text-decoration: none;
}
a:hover {
text-decoration: underline;
}
/* A link wrapping a card, poster or icon carries no text of its own to underline. */
a:has(img, svg, .wrolpi-card-icon):hover {
text-decoration: none;
}
/*
* Native controls the browser paints itself.
*
* A range slider is drawn by the UA, and with nothing said about it the UA uses its own
* accent -- a fixed blue no theme can reach. Four are in the product: the log level on
* Settings, and the duration sliders in the food, water and ration calculators.
*
* Night is where that is not merely inconsistent. The theme exists so a user reading in
* the dark keeps their adaptation and every token is a red, so a slider arriving in the
* browser's blue was the brightest non-red thing on the page -- on a control the user drags
* back and forth while watching it.
*
* `--blue` for the same reason links take it: it remaps per theme, so night gets the dim
* red and amber the amber without a second rule.
*
* Checkbox, radio and `progress` are the other three types `accent-color` paints, and are
* listed for completeness rather than because they are on screen: the app's checkboxes and
* radios are Mantine's, which draw their own box from tokens, and `Progress` is a div. So
* those three are a guard for the day a native one appears, not a fix for anything visible
* today -- the slider is the one that was actually wrong.
*
* `accent-color` reaches the thumb and the FILLED half of the track and nothing else. The
* rest of the track comes from the UA palette that `color-scheme` selects -- which is
* already dark on the dark themes, because Mantine's stylesheet puts
* `color-scheme: var(--mantine-color-scheme)` on `:root` and ThemeProvider drives it with
* `forceColorScheme`. That property is inherited, so a control needs nothing of its own; a
* rule here restating it was verified to change no pixel and is not present.
*
* Progressive enhancement: `accent-color` wants Safari 15.4+ / Chrome 93+ / Firefox 92+, and
* an older client keeps the UA blue, which is what it had before this rule. A complete fix
* would mean `::-webkit-slider-runnable-track` and `::-moz-range-track` rules of our own.
*/
input[type="range"],
input[type="checkbox"],
input[type="radio"],
progress {
accent-color: var(--blue);
}
a:focus-visible {
outline: 2px solid var(--blue);
outline-offset: 2px;
}
html[data-theme="night"] a:focus-visible,
html[data-theme="amber"] a:focus-visible {
outline-color: var(--text);
}
/* ------------------------------------------------------------------------- */
/* Media filtering */
/* ------------------------------------------------------------------------- */
/*
* Videos, images, PDFs, embedded pages, and the map canvas cannot consume
* tokens, so a theme can pass them through an SVG color matrix instead. Night
* maps luminance onto the red channel so nothing a user opens can break their
* dark adaptation; amber tints media to match the interface.
*
* These rules key on `data-media-filter`, not `data-theme`, because filtering is
* a per-theme setting the user controls rather than a property of the theme's
* palette: night is filtered unless turned off, amber is unfiltered unless
* turned on. An empty or absent attribute filters nothing, so light and dark —
* which offer no filter — match no rule at all. ThemeProvider stamps the
* attribute, and index.html stamps it before first paint (an unfiltered flash of
* thumbnails would cost the user their dark adaptation before React even loads).
*
* The selectors target element types rather than an opt-in class on purpose:
* filtering has to fail safe. A library of thousands of files reaches the
* screen through many components, and one that forgot to opt in would undo half
* an hour of dark adaptation. `.media` stays available for a wrapper that needs
* filtering as a unit (a player with custom controls, say), and
* `.night-unfiltered` opts a single element out — for media whose real colors
* are the point, like the side-by-side comparison on /theme-sample.
*
* Only leaf media is filtered, never an app-level ancestor: a CSS filter
* creates a containing block, which would break `position: fixed` descendants
* like modals and the nav bar. These elements have no such descendants — an
* iframe's contents are a separate document with their own viewport.
*
* Filter definitions live in <MediaFilterDefs/> (mounted once by App).
*/
html[data-media-filter="night-red"] :is(img, video, canvas, iframe, embed, object, .media):not(.night-unfiltered) {
filter: url(#wrolpi-night-red);
}
html[data-media-filter="amber-mono"] :is(img, video, canvas, iframe, embed, object, .media):not(.night-unfiltered) {
filter: url(#wrolpi-amber-mono);
}
/*
* A `.media` wrapper filters its whole subtree as one unit, so the leaf media inside
* it must not be filtered a second time. Two passes of a luminance-to-red matrix is
* not the same as one: the second pass takes the luminance of an already-red pixel
* (0.2126 of it) and the image comes out markedly darker than everything around it.
*
* The audio player's poster is the case that found this — the poster sits under
* custom control glyphs, so the wrapper is filtered to keep the two tinted together.
*
* The trailing `:not(.night-unfiltered)` is here for specificity, not for its meaning.
* The rule above scores (0,3,1) — an attribute, a `.media` inside `:is()`, and a `:not()`
* — so this reset had to reach three class-level components to win it; without the extra
* one it scored (0,2,2) and silently lost every time, which is exactly what it did until
* a browser test measured the leaf and found it filtered twice. jsdom cannot catch this:
* it never resolves `filter` at all.
*/
html[data-media-filter] .media :is(img, video, canvas, iframe, embed, object):not(.night-unfiltered) {
filter: none;
}
/*
* A filter on an ancestor stops applying once the browser promotes a video to
* native fullscreen, so the filter is restated on the fullscreened element.
*/
html[data-media-filter="night-red"] :fullscreen:not(.night-unfiltered) {
filter: url(#wrolpi-night-red);
}
html[data-media-filter="amber-mono"] :fullscreen:not(.night-unfiltered) {
filter: url(#wrolpi-amber-mono);
}