Skip to main content

makeover_webview/
lib.rs

1//! The webview renderer for [`makeover_layout`].
2//!
3//! <!-- wiki: makeover-webview -->
4//!
5//! # The renderer that needs no palette
6//!
7//! `makeover-immediate` and `makeover-tui` both take a `Palette`, because egui
8//! and a terminal need an actual colour before they can put anything on
9//! screen. A webview does not: `var(--surface-raised)` *is* the late binding,
10//! and the browser resolves it against whatever `themes.js` last wrote onto
11//! `:root`.
12//!
13//! So this crate emits text naming intents, and never learns a colour. It is
14//! the deferral rule with no adapter in the way, and it is why the webview was
15//! always the wrong renderer to derive a vocabulary from: it can express
16//! anything, so it never pushes back.
17//!
18//! # Phase A: the stylesheet
19//!
20//! This module emits component CSS and no markup, deliberately. GoingsOn has
21//! 145 `innerHTML` sites and Balanced Breakfast 175 `createElement` sites, so
22//! moving markup is a migration where adopting a generated stylesheet is not.
23//! The apps keep every line of their markup and gain the classes.
24//!
25//! It is not a deletion either, which this header claimed until the measurement
26//! came in. Adoption across goingson removed 49 declarations net and *added* 25
27//! lines: a rule loses its depth declarations and gains a variant selector next
28//! to it, so the file stays the same size. What phase A moves is where depth is
29//! defined, not how much CSS exists. Numbers and method in the wiki note under
30//! "The deletion test, run".
31//!
32//! The bevel properties are byte-identical to what both apps already
33//! hand-write, which is asserted below.
34//!
35//! # Phase B: the markup, one description at a time
36//!
37//! [`form`] renders [`makeover_layout::Field`], which is the half of phase B
38//! whose description is settled. It emits strings, because both apps
39//! interpolate their fields into larger string-built forms and returning nodes
40//! would rewrite those too. It owns its own escaping, on the reasoning in that
41//! module: a Rust encoder can cover element text and attribute values with one
42//! function, where the apps need four and have to choose correctly at every
43//! call site.
44//!
45//! [`facet`] renders `makeover_layout::Facet`: a dimension a set is narrowed by,
46//! and the one phase-B emitter whose markup an app is not keeping, because the
47//! markup it replaces was two mechanisms rather than one. A tag's selection and
48//! a tag's browse position were separate state on MNW's discover page, which is
49//! why every filter row there carries a tick box *and* a chevron; one gesture
50//! doing both is what lets the second one go.
51//!
52//! [`list`] is the other half: column tracks, the narrowing rules, and the cell
53//! containers a row is made of. It stops at the cell boundary and does not
54//! render what goes inside one, on the reasoning in that module. So phase B is
55//! now the frame around content in both directions, and what an app still owns
56//! is the content itself.
57//!
58//! # What phase A settled, and what it costs
59//!
60//! Decided 2026-07-29 against goingson's `styles.css` rather than against a
61//! component list. The useful finding there was that `.btn` (line 644),
62//! `.card` (768) and `.tag, .badge` (882) each hand-write the same
63//! composition, so three quarters of phase A is one rule with several names.
64//!
65//! Two of the four decisions change how goingson looks, and adoption should
66//! not be described as a pure deletion:
67//!
68//! - **Pressed carries its fill.** [`interactive_rules`] emits
69//!   [`Depth::pressed`] whole. goingson presses to `--surface-sunken` today and
70//!   will press to `--surface-well`, and hovers to `--surface-overlay` today
71//!   and will hover to `--hover-surface`. Since `surface-well` inverts by theme
72//!   where `surface-sunken` does not, a dark theme presses *lighter* than it
73//!   hovers. That falls out of `makeover`'s own derivation, which says outright
74//!   that `surface-sunken` cannot serve as a well, so if it reads wrong the
75//!   answer is there and not here.
76//! - **Badges go flat.** See [`token_rules`].
77//!
78//! The other two: the progress trough is renderer-local and the scrollbar
79//! track was dropped ([`component_rules`]), and no class prefix ships by
80//! default, so adoption means deleting the app's hand-written rule in the same
81//! commit that adds the generated one. `.card`, `.badge` and the tab classes
82//! all already exist in goingson, and while both rules exist the cascade order
83//! decides which wins. That is the one real risk in adopting this, and it is
84//! why the migration lands per component rather than in one commit.
85//!
86//! # 0.10.0: the states this crate used to leave to its consumers
87//!
88//! [`interactive_rules`] emitted hover and pressed and stopped, because
89//! `makeover-layout` modelled no interaction state. Focus and disabled were
90//! therefore unsayable, and every app completed the primitive from outside the
91//! only way that works: by out-specifying a rule it does not own. goingson
92//! carries 19 such rules and the MNW server 21, and the three focus rings do
93//! not match each other.
94//!
95//! That also blocked the cascade-layer work outright. An app that declares
96//! `@layer` puts its own rules in a named layer, and unlayered declarations
97//! outrank every named layer regardless of specificity, so all of those
98//! overrides lose in the commit that adopts layers. They cannot simply be
99//! deleted, because they are the only thing supplying the missing states.
100//! Emitting the states here is what turns that adoption into a deletion.
101//!
102//! Four states now, in emission order, and the order is load-bearing: they are
103//! all specificity (0,2,0), so disabled beats hover by coming last and by
104//! nothing else. Nothing here reaches for `:not(:disabled)`, which would raise
105//! a selector this crate will shortly be wrapping in its own layer.
106//!
107//! Hover additionally sits inside a capability query now. `makeover-touch`
108//! answers whether a fingertip has hover and `makeover-geometry` spells the
109//! condition; this crate asks and does not decide. goingson's section 60 exists
110//! solely to take the hover state back on touch, which is a fight it should
111//! never have been handed.
112//!
113//! # 0.11.0: the layer contract
114//!
115//! [`stylesheet`] emits into the `makeover` cascade layer ([`CSS_LAYER`], which
116//! lives in `makeover-geometry` because that is the one crate every CSS emitter
117//! in the family already depends on). `makeover-geometry` 0.6.0 does the same
118//! for `geometry.css`.
119//!
120//! The cascade resolves origin and importance, then layer, then specificity,
121//! then source order, and **unlayered normal declarations outrank every named
122//! layer**. So before this, an app that declared `@layer base, components,
123//! responsive` put every rule it owns into a named layer and lost all of them to
124//! this unlayered file, regardless of specificity and regardless of loading
125//! last. Nothing errors when that happens: the CSS is valid, the minifier is
126//! happy, and buttons and badges look subtly wrong.
127//!
128//! That is why the layer belongs here rather than in each app. An app cannot fix
129//! it from its own stylesheet, because the fix is to layer the file it does not
130//! own.
131//!
132//! **What it flips**, and the reason each app wants a look when it bumps the
133//! pin: a generated rule that currently beats an app rule by being more specific
134//! stops beating it. The direction is always "the app wins", which is what the
135//! apps already assume, but a hand-written rule an app thought was dead can come
136//! back to life.
137//!
138//! An app should declare the order once, or the layer's position is decided by
139//! whichever generated file the browser happens to see first:
140//!
141//! ```css
142//! @layer makeover, base, components, responsive;
143//! ```
144//!
145//! [`in_css_layer`] is re-exported for an app that assembles its own stylesheet
146//! from this crate's pieces. goingson builds `tables.css` in its own `build.rs`
147//! out of [`list::narrowing_css`] and [`list::grid_template_columns`], and those
148//! rules are as generated as the ones here, so they belong in the same layer and
149//! this crate cannot put them there on the app's behalf.
150//!
151//! # 0.12.0: the ring gets its own width
152//!
153//! [`focus_rule`] reused [`Emit::border_width`] and emitted a 1px ring. That was
154//! an implementation convenience dressed as consistency with the invalid-field
155//! ring: a bevel and a focus indicator answer different questions, and only one
156//! of them has to be noticed from across a desk.
157//!
158//! Caught while adopting 0.11.0 into goingson, by the check the adoption tasks
159//! ask for. Every consumer had already written its own ring and all three chose
160//! at least 2px: the MNW server 2px across 10 rules, Balanced Breakfast 2px,
161//! goingson 2px on three rules and 3px on the one covering twelve selectors. The
162//! design system was the only thing in the tree saying 1px, so deleting the app
163//! rules in favour of it would have thinned the focus indicator everywhere.
164//!
165//! [`Emit::focus_width`] now carries it, defaulting to `2px`, and the offset is
166//! the same magnitude with its sign off the depth. Both values are the measured
167//! consensus rather than a new opinion.
168//!
169//! # 0.57.0: a suggestion's second line, in place of a reason it cannot be picked
170//!
171//! `1fcf2e9b`, ruled by Max 2026-08-21. `Outcome::Suggestions` now carries
172//! `Candidate` rather than `Choice`, and a candidate has no `unavailable`: a
173//! suggestion that cannot be picked is a row a route should not have offered.
174//! What it has instead is a `detail`, the line that tells it from a row reading
175//! the same -- "Format" under audio, software, writing and video.
176//!
177//! So `.form-suggestion-why` is gone and `.form-suggestion-detail` is here, and
178//! the difference is not only the name. `why` was drawn in the disabled token,
179//! which said the row was refusing. A detail orients rather than refuses, so it
180//! is `--content-muted`, which is what every other secondary line in this crate
181//! reads as.
182//!
183//! No consumer is affected: the only emitter is `quasi-webview`, which moves in
184//! the same cascade.
185//!
186//! # 0.56.0: an interval, as the group MNW already wrote by hand
187//!
188//! [`makeover_layout::FieldKind::Interval`] emits a `role="group"` named by the
189//! field's label, holding one `<input type="number">` per end. That is not a
190//! shape invented here: MNW's discover sidebar has a `role="group"` with
191//! `aria-labelledby` over `min_price` and `max_price` today, written by hand
192//! because nothing in the description could say the two boxes were one
193//! question. The markup is what the measurement found, and this emits it.
194//!
195//! - **The group carries the error and the descriptions**, on the split
196//!   [`makeover_layout::FieldKind::Radio`] already uses here: what is wrong is
197//!   the answer, and a crossed interval is not the fault of either end.
198//! - **Both boxes take the whole extent.** `min`, `max` and `step` describe the
199//!   axis, so they are written twice. The crossing rule is not emitted, because
200//!   HTML has no attribute for it and the description does not carry it: it
201//!   comes back as an error on the group, like every other refusal.
202//! - **Which end is which is `aria-label` and nothing more.** The description
203//!   states direction structurally, by which member holds which name, and never
204//!   in words. Visible Min and Max captions are a page's own and reach the
205//!   group through [`form::Filling::trailing`].
206//!
207//! [`form::Value::Between`] is the second value. A separator inside one string
208//! would have made this crate the owner of a delimiter that either end could
209//! contain.
210//!
211//! # 0.55.0: a number's unit, as adjacent text
212//!
213//! `makeover-layout` 0.33.0's `Field::unit`. HTML has no unit attribute and
214//! inventing one would be markup nothing reads, so it is a `<span>` after the
215//! control -- and it is named in `aria-describedby` rather than left as
216//! decoration, because a number and what it is measured in are one fact and
217//! reading the first without the second is reading it wrong.
218//!
219//! This is the renderer the member costs something. egui already had somewhere
220//! better than the label to put a unit; here the two placements are equally
221//! adjacent, and what is bought is that the unit is a value a consumer can read
222//! back rather than a suffix on a label it would have to parse.
223//!
224//! # 0.54.0: a curve this renderer can carry, and one it cannot
225//!
226//! `makeover-layout` 0.32.0's `Curve`. A range takes its granularity from the
227//! curve now (`Field::curve.step()`), every other kind keeps `Field::step`, and
228//! `Curve::Linear` emits exactly what it emitted before.
229//!
230//! **A constant-ratio curve still emits a linear track, and that is a
231//! shortfall rather than a decision.** HTML has no logarithmic range input, so
232//! honouring one here means either shipping JS that maps the thumb position to
233//! a value -- renderer-side app code, which is the thing this stack exists to
234//! delete -- or changing what the control submits from a value to a fraction,
235//! and then something has to map it on the way back in. That something is not
236//! this crate: the MNW server reads these forms with its own handlers, so a
237//! fraction arriving where a value is expected would be silent. Filed rather
238//! than guessed at.
239//!
240//! What is not lost: the extent, the granularity, and the value's own units.
241//! What is lost is resolution at the small end, which is exactly the reason an
242//! app asked for a ratio in the first place. No consumer is affected today --
243//! every described range on this path is linear.
244//!
245//! # 0.52.0: a markdown field gets the preview it was permitted
246//!
247//! 0.50.0 marked a [`makeover_layout::FieldKind::Rich`] field with
248//! `data-format="markdown"` and stopped there, and nothing read the mark: no
249//! rule, no markup, no selector anywhere in the tree. So a described editor was
250//! a plain textarea carrying an attribute, and MNW converting its four
251//! hand-written section editors onto the member would have *lost* the
252//! Write/Preview pair one of them already has. A permission taken and not spent
253//! turns every conversion into a regression.
254//!
255//! [`form::editor_rules`] and the chrome beside it are the spending. The pair is
256//! a segmented control, so it takes the depth, the focus ring and the chosen
257//! state from rules that already existed; the preview pane is a well, because it
258//! stands where the control stood. Both are gated on the attribute rather than
259//! on a class, which is what the attribute was for.
260//!
261//! **This crate still renders no markdown.** The pane arrives empty and is
262//! filled by whatever binds the editor, which is where the host's sanitiser
263//! already is. A converter here would have moved that guarantee into a crate
264//! with no view of the host's content-security posture.
265//!
266//! # 0.47.0: a range, a chooser's ghost text, and an option that cannot be
267//! picked yet
268//!
269//! `makeover-layout` 0.28.0's three form findings, all of them cheap here and
270//! none of them cheap in the app that found them.
271//!
272//! - `FieldKind::Range` emits `<input type="range">`, and `Field::step` emits
273//!   `step`. The step is emitted only when the description carries one: the
274//!   browser's own default is `step="1"`, which is what a description means by
275//!   saying nothing, and is also what turns a 0-to-1 threshold into a
276//!   two-position control.
277//! - A select with nothing chosen emits a disabled, selected, valueless first
278//!   option carrying `Field::placeholder`. HTML has no placeholder attribute on
279//!   `<select>`; this is the idiom, and `required` keeps working through it
280//!   because the option's value is empty.
281//! - `Choice::unavailable` emits `disabled` plus the reason. Where it goes
282//!   differs by control and the difference is forced: a radio group gets a
283//!   `.form-option-reason` span beside the label, and a `<select>` option has
284//!   room for no element at all, so the reason runs into its text.
285//!
286//! # 0.25.0: a cell says what it holds
287//!
288//! 0.23.0 gave a table its layout and left every cell the same. One `.cell`
289//! carried the whole thing, so a cell holding text and a cell holding a button
290//! were one class and one content colour, and a control in a cell was painted
291//! as text. That is the drift [`RowPart::intent`](makeover_layout::RowPart)
292//! has prevented for list rows since 0.2.0 and prevented for nothing here.
293//!
294//! makeover-layout 0.14.0's [`CellPart`](makeover_layout::CellPart) names the
295//! four things a cell holds, and [`table_rules`] turns them into
296//! `.cell-value`, `.cell-tokens`, `.cell-actions` and `.cell-link`. Only the
297//! first takes a colour: a token carries its own tone, an action is a control
298//! rather than text, and a link takes the action colour from the anchor it is.
299//!
300//! The colour going on `.cell-value` rather than on `.cell` is the fix rather
301//! than an implementation detail. On the container it cascades into the parts
302//! that are not text, which is the bug said in one rule.
303//!
304//! [`list::Cell::part`] is `Option<CellPart>` here, where it was
305//! `Option<RowPart>`. A table cell borrowing the list row's vocabulary was the
306//! drift with a type on it: the two answer different questions, and only one of
307//! them was ever about a cell.
308//!
309//! # 0.23.0: a table lays itself out, and a row shows its controls
310//!
311//! Three things a description could say and this renderer had no rule for,
312//! found together by rendering the MNW server's SSH-keys settings tab through
313//! `quasi` and preferring the hand-written Askama original.
314//!
315//! **A table had no layout at all.** [`list::narrowing_css`] emits the track
316//! list, and it has to be called with the columns, so it works where the
317//! columns are known at build time: goingson builds `tables.css` in its own
318//! `build.rs` and is untouched. A table a description produced knows its
319//! columns at render time, and the rules would have had to travel with the
320//! markup: a `<style>` element per table, which needs `style-src
321//! 'unsafe-inline'` that the MNW server is working to drop, or the head, which
322//! an htmx fragment swap does not carry. [`table_rules`] lays a table out with
323//! `display: table` instead, which aligns columns across rows knowing nothing
324//! about how many there are. [`Priority`](makeover_layout::Priority) hiding
325//! moves from a generated rule per dropped column to one rule per drop class,
326//! and [`list::column_classes`] is what puts those classes on a cell. A header
327//! row emitted by a renderer's own code should call it too, or the header and
328//! the body disagree about which column just dropped.
329//!
330//! **A destructive button had nowhere for its tone to land.** `.button` carried
331//! no tone, on the reading that a control's colour is its surface. Every
332//! consumer had written the danger rule itself. It joins the badge in taking
333//! the four tones as colour, off `data-tone`.
334//!
335//! **A list rendered as a bulleted list**, because nothing here reset the `ul`
336//! a renderer emits for one.
337//!
338//! `RowPart::revealed_on_hover` also stops being honoured, which its own doc
339//! sanctioned: a renderer decides. It was retired outright in makeover-layout
340//! 0.13.0, once this had been its only consumer for a release. The rule hid a
341//! row's actions until hover,
342//! and every escape it grew was a report that hiding was wrong for somebody:
343//! `focus-within` for the keyboard, the capability gate for a fingertip with no
344//! way to unhide. What survived hid the controls from pointer users alone, who
345//! are the ones scanning a list to learn what can be done to a row.
346//!
347//! # 0.17.0: the depth classes stop being controls
348//!
349//! [`depth_rules`] gave `.raised` the whole interactive set. A depth is a
350//! statement about shape, so that left the vocabulary with no raised surface
351//! that is merely an object, and an app wanting one had two moves: write its
352//! own class from tokens, or take a control class and cancel the control half.
353//! goingson took the second, in three variants over sixteen elements
354//! (`.card--static` at 14 call sites, `.card--muted` at 2, `.card--shell` at 1),
355//! each re-asserting the resting fill and bevel on `:hover` and `:active`.
356//!
357//! Measured before changing it: `.raised` is emitted into goingson, Balanced
358//! Breakfast and the MNW server, and none of the three has a single call site.
359//! The states were unasked-for everywhere at once, and dropping them costs no
360//! migration anywhere.
361//!
362//! `.card` and `.button` are unchanged. They are the same depth *and* controls,
363//! and they take their states from [`surface_rules`], which is where a state
364//! belongs: on the thing that claims to answer a pointer.
365//!
366//! # Substitution, three ways
367//!
368//! `Fill::Well` has no colour on makeover before 2.3.0, and each renderer
369//! answers that differently, which is the evidence that dropping
370//! `Fill::fallback` from the description was right:
371//!
372//! - `makeover-immediate` substitutes the page in Rust.
373//! - `makeover-tui` refuses to substitute and draws an edge instead, because a
374//!   terminal would quantise the two together.
375//! - here, CSS already has the mechanism: `var(--surface-well,
376//!   var(--surface-page))` falls back in the browser, and nothing in Rust
377//!   decides anything.
378
379#![forbid(unsafe_code)]
380
381pub mod facet;
382pub mod figure;
383pub mod form;
384pub mod list;
385pub mod meter;
386pub mod placeholder;
387pub mod vocabulary;
388
389/// A render of every emitter, scraped for the classes it wrote.
390///
391/// Test-only, and the guard behind [`vocabulary::names`]. See the module's own
392/// header for why the check renders rather than reads the source.
393#[cfg(test)]
394mod corpus;
395
396use crate::list::{cell_part_class, part_class};
397use makeover_geometry::{Density, SizeClass};
398// Re-exported rather than redefined. An app assembling its own stylesheet out
399// of this crate's pieces needs the same layer name, and most such apps depend
400// on this crate and not on `makeover-geometry` directly: goingson builds
401// `tables.css` in its own build.rs from [`list::narrowing_css`], and those
402// rules are as generated as the ones here.
403pub use makeover_geometry::{CSS_LAYER, in_css_layer};
404
405/// This crate's version, as the generated stylesheet reports it.
406///
407/// A consumer whose lockfile still pins an old `makeover-webview` gets a
408/// well-formed sheet with components missing and no error anywhere, so the
409/// emitter has to name itself in what it writes. Read by
410/// `makeover_build::layout_css` through [`stylesheet`].
411pub const VERSION: &str = env!("CARGO_PKG_VERSION");
412use makeover_layout::{
413    Bevel, CellPart, Depth, Fallback, Fill, Flow, Intent, RowPart, Selector, Sort, State, Token,
414    Tone,
415};
416use makeover_touch::Affordance;
417use std::fmt::Write as _;
418
419/// How the emitted CSS is shaped.
420#[derive(Debug, Clone, Copy, PartialEq, Eq)]
421pub struct Emit {
422    /// Bevel thickness, as a CSS length.
423    ///
424    /// A value, so it arrives from the caller: border widths belong to
425    /// `makeover-geometry` and will come from there once it carries them.
426    pub border_width: &'static str,
427    /// Focus ring thickness, as a CSS length.
428    ///
429    /// Separate from [`border_width`](Self::border_width), which it reused
430    /// until 0.12.0. That reuse was an implementation convenience dressed as
431    /// consistency, and it emitted a 1px ring: a bevel and a focus indicator
432    /// are answering different questions, and only one of them has to be
433    /// noticed from across a desk.
434    ///
435    /// The default is the measured consensus rather than a new opinion. Every
436    /// consumer had already written its own ring and all three chose at least
437    /// 2px: the MNW server 2px across 10 rules, Balanced Breakfast 2px,
438    /// goingson 2px on three rules and 3px on the one covering twelve
439    /// selectors. The design system was the only thing in the tree saying 1px.
440    pub focus_width: &'static str,
441    /// Prefix for emitted class names, without the leading dot.
442    pub class_prefix: &'static str,
443}
444
445impl Default for Emit {
446    fn default() -> Self {
447        Self {
448            border_width: "1px",
449            focus_width: "2px",
450            class_prefix: "",
451        }
452    }
453}
454
455/// A string as CSS escapes, for a `content` value.
456///
457/// `\u{25B2}` becomes `\25B2`. Emitted escaped rather than literally so the
458/// stylesheet is ASCII whatever the description spells: a `content` string is
459/// read by whatever encoding the consumer serves the file as, and a caret that
460/// depends on that is a caret that works on one machine.
461///
462/// Terminated by the closing quote at every site here. A CSS hex escape takes
463/// up to six digits and ends at the first character that cannot be one, so an
464/// escape followed by more text would need a space that these do not.
465fn css_escape(text: &str) -> String {
466    text.chars().fold(String::new(), |mut out, c| {
467        let _ = write!(out, "\\{:X}", c as u32);
468        out
469    })
470}
471
472/// The CSS custom property holding a bevel's composition.
473#[must_use]
474pub fn bevel_var(bevel: Bevel) -> &'static str {
475    match bevel {
476        Bevel::Raised => "--bevel-raised",
477        Bevel::Inset => "--bevel-inset",
478    }
479}
480
481/// A `var()` reference to a fill intent, with the browser's own fallback where
482/// the intent may be absent.
483///
484/// The fallback is CSS syntax, not a decision made here. That is the whole
485/// difference between this renderer and the other two.
486#[must_use]
487pub fn fill_var(fill: Fill) -> String {
488    match fill {
489        Fill::Well => format!("var(--{}, var(--{}))", fill.token(), Fill::Page.token()),
490        other => format!("var(--{})", other.token()),
491    }
492}
493
494/// The two-tone edge as a `box-shadow` value.
495///
496/// Two inset shadows, one per corner pair: the light one offset down and
497/// right so it lands on the top and left edges, the dark one the other way.
498/// The same assignment `makeover-immediate` draws with polylines and
499/// `makeover-tui` draws with box-drawing characters.
500#[must_use]
501pub fn bevel_shadow(bevel: Bevel, opts: &Emit) -> String {
502    let (top_left, bottom_right) = bevel.edges();
503    let w = opts.border_width;
504    format!(
505        "inset {w} {w} 0 var(--{}), inset -{w} -{w} 0 var(--{})",
506        top_left.token(),
507        bottom_right.token()
508    )
509}
510
511/// The custom properties both bevels resolve through.
512///
513/// Emitted as properties rather than inlined into every rule because that is
514/// what the apps already do, and because a consumer that wants the edge
515/// without the fill reads the property directly.
516#[must_use]
517pub fn bevel_properties(opts: &Emit) -> String {
518    let mut css = String::new();
519    for bevel in [Bevel::Raised, Bevel::Inset] {
520        let _ = writeln!(
521            css,
522            "    {}: {};",
523            bevel_var(bevel),
524            bevel_shadow(bevel, opts)
525        );
526    }
527    css.push_str(ELEVATION_PROPERTY);
528    css
529}
530
531/// The cast shadow of a surface that floats over the page.
532///
533/// Composed here for the reason the bevel pair is: `makeover` derives the tone,
534/// this crate owns the geometry, and neither has to know the other's numbers.
535///
536/// **Only for a surface that overlays the page.** A menu, a toast, a popover, a
537/// dropdown. A surface *in* the page takes `.raised` and its bevel, and a rule
538/// that reaches for this on a card or a plate has renamed a literal rather than
539/// replaced it.
540///
541/// Two lengths rather than one, because a single blur reads as a smudge at
542/// plate size and as a halo at menu size. The offset is small and downward: a
543/// Platinum-era menu sits just off the page rather than hovering above it.
544const ELEVATION_PROPERTY: &str =
545    "    --elevation-overlay: 0 2px 4px var(--elevation), 0 8px 24px var(--elevation);\n";
546
547/// The class name for a depth.
548#[must_use]
549pub fn depth_class(depth: Depth, opts: &Emit) -> Option<String> {
550    let name = match depth {
551        Depth::Flat => return None,
552        Depth::Raised => "raised",
553        Depth::Well => "well",
554        Depth::Sunken => "sunken",
555        // A depth added to the description since this renderer was last
556        // built. No class, on the same footing as Flat: emitting a name
557        // whose rule body we cannot write would put a class in the markup
558        // that the stylesheet never defines.
559        _ => return None,
560    };
561    Some(format!("{}{name}", opts.class_prefix))
562}
563
564/// A prefixed class name.
565///
566/// Public since 0.27.0, for the renderers that emit markup this crate does not.
567/// A screen renderer writing `class="row"` has to prefix it the way the
568/// stylesheet half does or a prefixed app gets rules matching everything except
569/// the elements that renderer wrote, and the failure is invisible: the CSS
570/// stays valid and one element is unstyled. quasi-webview carried a byte
571/// identical copy of this function until it could call this one.
572#[must_use]
573pub fn class(name: &str, opts: &Emit) -> String {
574    let mut out = String::with_capacity(opts.class_prefix.len() + name.len());
575    push_class(&mut out, name, opts);
576    out
577}
578
579/// A prefixed class name, written into a buffer the caller already has.
580///
581/// The form the emitters use, and the reason it exists is [`escape_into`]'s:
582/// every class on every element went through a `format!` before 0.40.0,
583/// including the default case where the prefix is empty and the answer is the
584/// argument. A described table row carried roughly eighty transient
585/// allocations, and this and the escaper were most of them.
586///
587/// [`class`] stays for callers holding a name rather than a buffer.
588///
589/// [`escape_into`]: crate::form::escape_into
590pub fn push_class(out: &mut String, name: &str, opts: &Emit) {
591    out.push_str(opts.class_prefix);
592    out.push_str(name);
593}
594
595/// The class an option of a selector carries, which is what the rules key off.
596///
597/// Named for the option and not for the group: [`selector_rules`] styles the
598/// thing that gets picked, so `Selector::Tabs` is `tab` and not `tabs`. The
599/// distinction is not pedantry. quasi-webview spelled these `tabs`, `segmented`
600/// and `option`, put `toggle` on the wrapping element rather than on the
601/// buttons inside it, and every described selector in that renderer came out
602/// with no depth, no focus ring and no chosen state, while the toggle group got
603/// a bevel meant for its buttons.
604///
605/// The chosen option additionally carries `chosen`, the same way a latched chip
606/// carries `latched`. That name is this crate's too; there is no reason for a
607/// caller to spell it, and [`selector_rules`] is where it is written down.
608#[must_use]
609pub fn option_class(selector: Selector) -> &'static str {
610    match selector {
611        Selector::Tabs => "tab",
612        Selector::Segmented => "segment",
613        Selector::Toggle => "toggle",
614    }
615}
616
617/// The fill and edge declarations for a depth, as a rule body.
618///
619/// Empty for [`Depth::Flat`], which has neither and inherits what it sits on.
620/// Callers lean on the emptiness to skip the rule rather than emit a class that
621/// sets nothing: a class that sets no properties is a class that means "I
622/// thought about this", which is what comments are for.
623///
624/// The two halves are emitted independently because [`Depth::Sunken`] has a
625/// fill and no bevel. Requiring both, which this did before makeover-layout
626/// 0.3.0, silently dropped the fill for exactly that case. Independent does not
627/// mean unpaired: both halves still come off one `Depth`, so they cannot
628/// disagree about what the region is.
629#[must_use]
630pub fn depth_declarations(depth: Depth) -> String {
631    let mut css = String::new();
632    if let Some(fill) = depth.fill() {
633        let _ = writeln!(css, "    background: {};", fill_var(fill));
634    }
635    if let Some(bevel) = depth.bevel() {
636        let _ = writeln!(css, "    box-shadow: var({});", bevel_var(bevel));
637    }
638    css
639}
640
641/// One rule giving a selector a depth, or nothing when the depth declares
642/// nothing.
643#[must_use]
644pub fn depth_rule(selector: &str, depth: Depth) -> String {
645    let body = depth_declarations(depth);
646    if body.is_empty() {
647        return String::new();
648    }
649    format!(".{selector} {{\n{body}}}\n")
650}
651
652/// The media condition a hover rule has to sit inside, or `None` if hover is
653/// unconditional.
654///
655/// Two crates answer this and neither answer is made here. `makeover-touch`
656/// owns *whether* hover exists at a density, and `makeover-geometry` owns how
657/// that capability is spelled as a media condition. Asking both is what stops
658/// this renderer minting a third opinion, which is what all three apps did:
659/// goingson sniffed the user agent, Balanced Breakfast used `(hover: none)`
660/// alone, and the MNW server had no gate at all.
661///
662/// [`SizeClass`] is required by [`Affordance::available`] and ignored by this
663/// member, which reports as much through `reads_size`. Passing Compact is not
664/// a claim about width; the test below pins that every class agrees.
665fn hover_condition() -> Option<&'static str> {
666    if Affordance::Hover.available(Density::Touch, SizeClass::Compact) {
667        // A fingertip grew a hover state. Nothing to gate, and this renderer
668        // should not invent a reason to gate anyway.
669        None
670    } else {
671        Some(Density::Pointer.media_condition())
672    }
673}
674
675/// Put a rule inside a media query, or leave it alone.
676fn gated(condition: Option<&str>, rule: &str) -> String {
677    let Some(condition) = condition else {
678        return rule.to_string();
679    };
680    let mut css = format!("@media {condition} {{\n");
681    for line in rule.lines() {
682        // Blank lines stay blank. Indenting one leaves trailing whitespace,
683        // which is the sort of thing a formatter later reverts and calls a diff.
684        if line.is_empty() {
685            css.push('\n');
686        } else {
687            let _ = writeln!(css, "    {line}");
688        }
689    }
690    css.push_str("}\n");
691    css
692}
693
694/// The keyboard focus ring, placed by the depth it lands on.
695///
696/// This is the webview's **focus ring** and nothing more. **Reach** and
697/// **focus** are both the browser's — the document decides what is reachable
698/// and `:focus-visible` decides which reached thing wears the ring — and no
699/// description states either. The three terms are defined once in
700/// `makeover_layout`'s crate header, "Reach, focus and the focus ring".
701///
702/// One ring for the whole system, because a focus ring's job is to be
703/// recognised and three apps having three of them is the failure. What varies
704/// is where it sits, and that comes off [`Depth`] rather than off a per-
705/// component choice: a well takes the ring inside its own edge, and anything
706/// standing proud of the page takes it outside.
707///
708/// `outline` rather than the composed `box-shadow` the invalid-field ring at
709/// [`field_rules`] uses, and deliberately the one place the two rings are built
710/// differently. A `box-shadow` ring has to restate the bevel beside it, because
711/// `box-shadow` is not additive and a lone ring silently drops the well out
712/// from under the element. That restatement is a second copy of the depth,
713/// living in a different function from the first, and it is exactly the
714/// duplication `Depth` exists to prevent. `outline` occupies its own property,
715/// so the bevel survives untouched and there is nothing to keep in agreement.
716/// They render the same: both are a flush ring one border-width wide.
717#[must_use]
718pub fn focus_rule(selector: &str, depth: Depth, opts: &Emit) -> String {
719    let w = opts.focus_width;
720    // Same magnitude either way, and only the sign comes off the depth. Both
721    // values are what the consumers had already converged on independently:
722    // 2px out is what all three wrote, and 2px in is the MNW server's own
723    // answer for the one inset ring it had.
724    let offset = match depth.bevel() {
725        // Inside the well, clear of its edge rather than painted over it.
726        Some(Bevel::Inset) => format!("calc(-1 * {w})"),
727        // Raised, or no edge at all. Outside, standing off by its own width.
728        _ => w.to_string(),
729    };
730    // The token by name. It is `makeover`'s, derived from the action colour,
731    // and reaching it through a description member was a second path to the
732    // same variable for as long as one existed.
733    format!(
734        ".{selector}:focus-visible {{\n    outline: {w} solid var(--focus-ring);\n    outline-offset: {offset};\n}}\n"
735    )
736}
737
738/// Present, visible, and not answering.
739///
740/// Matches the ARIA attribute as well as the pseudo-class, because `:disabled`
741/// only matches form elements and half the things this crate emits are not
742/// one: a `div` carrying `.chip` or `.tab` can never be `:disabled`. Keying on
743/// the accessible state is the pattern [`field_rules`] already establishes for
744/// `aria-invalid`, on the reasoning that one fact read by both the styling and
745/// the accessibility tree cannot drift from itself.
746///
747/// The rest depth is re-asserted rather than assumed, because this rule has to
748/// beat the hover and pressed rules above it. It does that on source order at
749/// equal specificity, not by out-specifying them: every rule this function's
750/// caller emits is (0,2,0), and adding a `:not(:disabled)` anywhere would raise
751/// one of them and have to be unpicked when this output moves inside its own
752/// cascade layer.
753#[must_use]
754pub fn disabled_rule(selector: &str, depth: Depth) -> String {
755    format!(
756        ".{selector}:disabled,\n.{selector}[aria-disabled=\"true\"] {{\n{}    color: var(--{});\n    cursor: not-allowed;\n}}\n",
757        depth_declarations(depth),
758        State::Disabled.token()
759    )
760}
761
762/// Every state a selector that answers a click implies: hover, pressed, focus
763/// and disabled, in that order.
764///
765/// Order is the whole cascade mechanism here. All four selectors are
766/// specificity (0,2,0), so disabled wins over hover and pressed by coming last
767/// and by nothing else.
768///
769/// Pressed emits [`Depth::pressed`] in full, fill and edge together. Emitting
770/// only the edge is what left goingson hand-writing `background:
771/// var(--surface-sunken)` on three separate rules, and a fill that does not
772/// travel with its edge is precisely the disagreement `Depth` exists to make
773/// unrepresentable. So the pressed fill comes from the description
774/// (`--surface-well`) rather than from whatever each app reached for.
775///
776/// Hover has no member in the description and is renderer policy: a terminal
777/// and an immediate-mode painter have no hover to express. It resolves against
778/// `--hover-surface`, which `makeover` already derives and which nothing
779/// consumed until now. What it *is* gated on is capability, via
780/// [`hover_condition`]. Before that gate existed the apps each wrote their own:
781/// goingson's section 60 exists solely to take back the hover state this
782/// function had just handed it, by out-specifying a rule it does not own.
783///
784/// `depth` is the selector's **rest** depth, used to place the focus ring and
785/// to restore the surface under a disabled control. The pressed rule keeps
786/// inverting from [`Depth::Raised`] regardless, which is what every caller got
787/// before this parameter existed: a tab's unchosen depth is
788/// [`Depth::Sunken`], and `Sunken.pressed()` is `Sunken`, so deriving the press
789/// from the rest depth would leave a tab with no press at all.
790#[must_use]
791pub fn interactive_rules(selector: &str, depth: Depth, opts: &Emit) -> String {
792    let mut css = gated(
793        hover_condition(),
794        &format!(".{selector}:hover {{\n    background: var(--hover-surface);\n}}\n"),
795    );
796    css.push_str(&depth_rule(
797        &format!("{selector}:active"),
798        Depth::Raised.pressed(),
799    ));
800    css.push_str(&focus_rule(selector, depth, opts));
801    css.push_str(&disabled_rule(selector, depth));
802    css
803}
804
805/// One rule per depth: its fill and its edge, together.
806///
807/// A depth and nothing else. `.raised` says a surface sits on what is behind
808/// it, which is a statement about the shape and not about what happens when a
809/// pointer arrives, so it emits no hover, press, focus or disabled rule. The
810/// named surfaces are where interaction lives: `.card` and `.button` are the
811/// same depth *and* controls, and they get their states from
812/// [`surface_rules`].
813///
814/// This class carried the interactive set until 0.17.0, which left the
815/// vocabulary with no raised surface that is merely an object. Consumers that
816/// needed one took a control class and cancelled half of it instead: sixteen
817/// elements in goingson across three `.card--*` variants, each re-asserting the
818/// resting fill and bevel on `:hover` and `:active`. Nothing anywhere used
819/// `.raised` itself, so the states were unasked-for in every consumer at once.
820#[must_use]
821pub fn depth_rules(opts: &Emit) -> String {
822    let mut css = String::new();
823    for depth in [Depth::Raised, Depth::Well] {
824        let Some(class) = depth_class(depth, opts) else {
825            continue;
826        };
827        css.push_str(&depth_rule(&class, depth));
828    }
829    css
830}
831
832/// The three surfaces that are a depth with a name.
833///
834/// `button` and `card` are both [`Depth::Raised`], and `field` is a
835/// [`Depth::Well`] because that is the reading `Depth`'s own documentation
836/// gives a text field. Their bodies come out identical by construction rather
837/// than by hand: three hand-written copies in goingson's stylesheet is what
838/// phase A deletes, and generating them from one call is what stops them
839/// drifting apart again.
840fn surface_rules(opts: &Emit) -> String {
841    let mut css = String::new();
842    for name in ["button", "card"] {
843        let c = class(name, opts);
844        css.push_str(&depth_rule(&c, Depth::Raised));
845        css.push_str(&interactive_rules(&c, Depth::Raised, opts));
846    }
847
848    let field = class("field", opts);
849    css.push_str(&depth_rule(&field, Depth::Well));
850
851    // A field takes focus and refuses input like everything else here, and got
852    // neither until now, which is why all three apps hand-write a focus ring
853    // for it and no two of them match. No hover or pressed: a text field does
854    // not light up under the pointer and does not invert when clicked, so the
855    // two states `interactive_rules` would add are the two it does not have.
856    css.push_str(&focus_rule(&field, Depth::Well, opts));
857    css.push_str(&disabled_rule(&field, Depth::Well));
858
859    // Keyed on the ARIA attribute rather than on a class, so the visual state
860    // and the accessible state cannot drift apart: there is one fact and both
861    // read it. goingson already drove its invalid styling this way and was
862    // right to; the `.invalid` class this emitted before 0.5.0 was a second
863    // place to forget.
864    //
865    // The ring composes *after* the bevel rather than replacing it. box-shadow
866    // is not additive, so a lone ring silently dropped the well out from under
867    // an invalid field. Flat and unlit: this edge is saying "wrong", and
868    // lighting one side would have it say "raised" at the same time.
869    let _ = writeln!(
870        css,
871        ".{field}[aria-invalid=\"true\"] {{\n    box-shadow: var({}), 0 0 0 {} var(--danger);\n}}",
872        bevel_var(Bevel::Inset),
873        opts.border_width
874    );
875    css
876}
877
878/// Badges and chips.
879///
880/// The one place phase A changes how goingson looks rather than only where its
881/// rules live. [`Token::Badge`] is [`Depth::Flat`], so a badge emits no fill
882/// and no edge at all, where goingson ships `.tag, .badge` as a single rule
883/// carrying the raised bevel. Splitting that means reading every call site to
884/// decide which of the two it always was.
885///
886/// What a badge does carry is a [`Tone`], the intent family it shares with
887/// notices and nothing else. Neutral is the bare class rather than a variant,
888/// because it is the absence of a status and not a status called "none".
889/// Text that goes somewhere.
890///
891/// The one inline control, and the vocabulary had no word for it until a
892/// description layer needed one. A table cell has carried `cell-link` since
893/// 0.14.0's part list, deliberately unruled because the cell's own rule covers
894/// it; what was missing is the same thing outside a table, which is what a
895/// described run holds when a sentence contains a link.
896///
897/// Colour and underline only. Whether a link is inline in a sentence or sitting
898/// on its own line is the app's layout, and how much room it takes is
899/// `makeover-geometry`'s. What is here is the pair of signals that say "this
900/// goes somewhere" and nothing that says where it sits.
901///
902/// The visited arm is deliberately absent. A link inside an app points at the
903/// app's own screens, which the user is expected to have been to, so painting
904/// them differently marks almost everything and distinguishes nothing.
905fn link_rules(opts: &Emit) -> String {
906    let mut css = String::new();
907    let link = class("link", opts);
908
909    let _ = writeln!(
910        css,
911        ".{link} {{\n    color: var(--action);\n    \
912         text-decoration: underline;\n}}"
913    );
914    // The hover step is the same one every other control takes, and it is a
915    // colour rather than a surface: a link has no box to raise.
916    let _ = writeln!(
917        css,
918        "@media (hover: hover) and (pointer: fine) {{\n    .{link}:hover \
919         {{\n        color: var(--action-hover);\n    }}\n}}"
920    );
921    let _ = writeln!(
922        css,
923        ".{link}:focus-visible {{\n    outline: {} solid var(--focus-ring);\n    \
924         outline-offset: 2px;\n}}",
925        opts.focus_width
926    );
927    // A link is often a `<button>` rather than an `<a>`: a renderer picks the
928    // element from the method, so a link that writes is a button that has to
929    // stop looking like one.
930    let _ = writeln!(
931        css,
932        "button.{link} {{\n    background: none;\n    border: none;\n    \
933         padding: 0;\n    font: inherit;\n    cursor: pointer;\n}}"
934    );
935    css
936}
937
938fn token_rules(opts: &Emit) -> String {
939    let mut css = String::new();
940
941    // No `depth_rule` call here, deliberately: `Token::Badge.depth(_)` is Flat,
942    // and a label with an edge says it can be pressed.
943    let badge = class("badge", opts);
944    // `content-muted` literally, not `Tone::Neutral.token()`. What makes a
945    // badge quiet is `Token::Badge` answering no click, which this crate holds
946    // and `Tone` genuinely does not know. Routing it through Neutral put the
947    // claim where the evidence was not, and the bill arrived on the figure
948    // value: it took the same muting from the same call and read as its own
949    // caption. Neutral answers `content` from makeover-layout 0.36.0.
950    let _ = writeln!(css, ".{badge} {{\n    color: var(--content-muted);\n}}");
951    for tone in [Tone::Info, Tone::Success, Tone::Warning, Tone::Danger] {
952        let _ = writeln!(
953            css,
954            ".{badge}[data-tone=\"{0}\"] {{\n    color: var(--{0});\n}}",
955            tone.token()
956        );
957    }
958
959    // A button carries the four tones a badge does. It had none, on the reading
960    // that a control's colour is its surface rather than its text, and that
961    // reading has one hole big enough to matter: the button that destroys
962    // something. Every consumer had written that rule itself, and a description
963    // that says `Tone::Danger` on an act had nowhere for it to land.
964    //
965    // Colour and not a fill, matching the badge. A red surface is a decision
966    // about emphasis that belongs to an app's own layer, and two of them
967    // fighting is worse than neither.
968    let button = class("button", opts);
969    for tone in [Tone::Info, Tone::Success, Tone::Warning, Tone::Danger] {
970        let _ = writeln!(
971            css,
972            ".{button}[data-tone=\"{0}\"] {{\n    color: var(--{0});\n}}",
973            tone.token()
974        );
975    }
976
977    // A chip holds itself down, which is `Depth::pressed` arrived at
978    // independently by two apps. `removable` is a remove affordance, so it is
979    // markup and waits for phase B.
980    let chip = class("chip", opts);
981    let unlatched = Token::Chip { removable: false };
982    css.push_str(&depth_rule(&chip, unlatched.depth(false)));
983    css.push_str(&interactive_rules(&chip, unlatched.depth(false), opts));
984    css.push_str(&depth_rule(
985        &format!("{chip}.latched"),
986        unlatched.depth(true),
987    ));
988    css
989}
990
991/// The three selectors, each named by what it picks.
992///
993/// A tab comes *forward* to join the pane it opens, which is why
994/// [`Selector::Tabs`] chooses [`Depth::Raised`] where a segment and a toggle
995/// are held in. That is the folder semantic, and it is the whole reason the
996/// three are not one member with a flag.
997///
998/// [`Selector::abutting`] is not emitted: whether the options touch is
999/// spacing, and spacing is `makeover-geometry`'s question to answer.
1000///
1001/// Both states emit as of makeover-layout 0.3.0. Before it the description
1002/// named only the chosen option, so an unchosen one fell through to
1003/// [`Depth::Flat`] and nothing was drawn for it, which left goingson's tab
1004/// strip hand-writing the recess that makes its chosen tab read as forward.
1005fn selector_rules(opts: &Emit) -> String {
1006    let mut css = String::new();
1007    for selector in [Selector::Tabs, Selector::Segmented, Selector::Toggle] {
1008        let c = class(option_class(selector), opts);
1009        css.push_str(&depth_rule(&c, selector.unchosen()));
1010        css.push_str(&interactive_rules(&c, selector.unchosen(), opts));
1011        css.push_str(&depth_rule(&format!("{c}.chosen"), selector.chosen()));
1012    }
1013    css
1014}
1015
1016/// The parts of a list row.
1017///
1018/// The list is written out rather than derived because `RowPart` is
1019/// `#[non_exhaustive]` as of makeover-layout 0.9.0, so there is nothing to
1020/// iterate. A member added upstream emits no rule until it is named here, which
1021/// is the trade `non_exhaustive` makes: a silent gap instead of a build break.
1022/// [`part_class`] carries the same list and the same obligation.
1023fn row_rules(opts: &Emit) -> String {
1024    let mut css = String::new();
1025
1026    // The container the rows sit in. Three zeroes and a keyword, all of them
1027    // undoing a browser default the description never asked for: a described
1028    // list of SSH keys is not a bulleted list, and it rendered as one because
1029    // nothing here said otherwise. Not a size: there is no magnitude in it,
1030    // which is the line this crate holds.
1031    let _ = writeln!(
1032        css,
1033        ".{} {{\n    list-style: none;\n    margin: 0;\n    padding: 0;\n}}",
1034        class("list", opts)
1035    );
1036
1037    for part in [
1038        RowPart::Primary,
1039        RowPart::Secondary,
1040        RowPart::Meta,
1041        RowPart::Actions,
1042        RowPart::Tokens,
1043        RowPart::Proportion,
1044    ] {
1045        let c = class(part_class(part), opts);
1046
1047        // Actions carry controls rather than text, and `RowPart::intent` says
1048        // so by returning the same intent inheriting already gives. Pinning it
1049        // would be louder than saying nothing. Tokens answer alike, for their
1050        // own reason: each token carries its own tone, and a colour on the
1051        // strip would fight the things sitting in it. A proportion is the same
1052        // case again: the meter inside carries the tone.
1053        if !matches!(
1054            part,
1055            RowPart::Actions | RowPart::Tokens | RowPart::Proportion
1056        ) {
1057            let _ = writeln!(css, ".{c} {{\n    color: var(--{});\n}}", part.intent());
1058        }
1059
1060        // A row's actions are shown at rest. `RowPart::revealed_on_hover` said
1061        // otherwise and was not honoured here from 0.23.0; makeover-layout
1062        // 0.13.0 retired the method, so there is no longer a description saying
1063        // one thing and a renderer doing another.
1064        //
1065        // The rule was `opacity: 0` gated to pointer devices, revealed on
1066        // `:hover` and on `:focus-within`. Each escape it needed was a report
1067        // that hiding was wrong for somebody: `focus-within` because tabbing
1068        // could never reach an action; the gate because a fingertip had no way
1069        // to unhide, which both webview apps had already hand-written
1070        // `opacity: 1` to undo. What was left was a control hidden from
1071        // exactly one group: people using a pointer, who are also the group
1072        // scanning a list to find out what can be done to a row.
1073        //
1074        // A settings screen is where that reads worst: the whole reason to be
1075        // on it is to remove a key, and the button doing so was invisible
1076        // until pointed at. A table cell's actions were never hidden, so the
1077        // two arrangements now agree.
1078    }
1079
1080    // A part that may take two lines. `Flow::Tight` gets no rule: one line is
1081    // what a run already does, and restating it here would put a declaration on
1082    // every part in every row to say nothing.
1083    //
1084    // This is the shape both webview apps had already written by hand and
1085    // commented -- Balanced Breakfast on a feed row's title, goingson on a
1086    // problem's body -- which is the whole argument for the description
1087    // carrying it. `-webkit-` prefixed and unprefixed together: the prefixed
1088    // trio is what every engine actually implements, and `line-clamp` is the
1089    // standard property landing behind it.
1090    let _ = writeln!(
1091        css,
1092        ".{} {{\n    display: -webkit-box;\n    -webkit-box-orient: vertical;\n    -webkit-line-clamp: {lines};\n    line-clamp: {lines};\n    overflow: hidden;\n}}",
1093        class("row-relaxed", opts),
1094        lines = Flow::Relaxed.lines()
1095    );
1096
1097    css
1098}
1099
1100/// A row of things that share their space, and what each fallback gets here.
1101///
1102/// Ruling: wiki `layout-room-and-fallback`, Max 2026-08-18. The bug it answers
1103/// is goingson's `styles.css:702`, which pinned every `.page-header` over the
1104/// pill strip with `position: absolute` and so took the toolbar out of flow:
1105/// zero width contributed to the row it shared, nothing able to collide with
1106/// it, and therefore nothing preventing the collision. Rule 1 of the ruling is
1107/// that every described member is in flow, and these rules are how that is kept
1108/// rather than asked for.
1109///
1110/// # The floor, which is most of the fix
1111///
1112/// `.run > *` gets `min-width: min-content`. That is the derived minimum the
1113/// ruling asks for, in this renderer's own unit and stated by the browser
1114/// rather than by anybody: a member cannot be squeezed narrower than what is
1115/// in it, so members in one flow push each other instead of overlapping. It
1116/// costs no query and no number, and it is what fixes all four measured widths
1117/// whichever fallback the group declared.
1118///
1119/// # What each fallback gets, exactly
1120///
1121/// [`Fallback::Wrap`] is `flex-wrap: wrap`, which is exact. The browser wraps
1122/// the run when the members no longer fit, deciding that from their own
1123/// intrinsic widths, which is the derived minimum doing the whole job.
1124///
1125/// [`Fallback::Stack`] is wrap plus `flex: 1 1 max-content` on the members, so
1126/// a member that cannot sit beside its sibling takes a line of its own and
1127/// fills it. For the two-member run this was ruled on -- a tab strip and a
1128/// band -- that is precisely "a row becomes a column".
1129///
1130/// [`Fallback::Shed`] and [`Fallback::Menu`] get wrap, and this renderer is
1131/// honouring less than the description says. **CSS cannot express either one
1132/// without breaking the ruling's own first constraint.** Both need to know that
1133/// the run is out of room in order to take a member out of it, a container
1134/// query is the only construct that can ask, and `@container` compares against
1135/// a `<length>` -- there is no `@container (inline-size < min-content)`. So
1136/// every honest spelling of Shed here needs an authored breakpoint, which is
1137/// the thing the ruling exists to forbid, and the dishonest ones are worse: a
1138/// clamped height clips by document order rather than by [`Priority`], and
1139/// `display: none` under a viewport `@media` is the `nth-child(n+5)` bug the
1140/// vocabulary replaced.
1141///
1142/// Wrapping is the right thing to do instead. It keeps every member reachable,
1143/// which is the property that was actually broken -- goingson's new-contact
1144/// button left the viewport entirely at 560 -- and it keeps rule 1. A renderer
1145/// answering with less than was described is precedented and deliberate here:
1146/// [`makeover_layout::Region::Columns`] says a terminal stacking a board's
1147/// columns is honouring the description rather than degrading it.
1148///
1149/// The real mechanism needs the shed members to have somewhere to go, which is
1150/// markup and belongs to quasi-webview: an overflow control is a member of the
1151/// run, and the description does not yet say that a member *is* one. Filed
1152/// rather than guessed at.
1153fn run_rules(opts: &Emit) -> String {
1154    let run = class("run", opts);
1155    let mut css = String::new();
1156
1157    // `flex-wrap: nowrap` is stated rather than left to the default, because
1158    // the fallbacks below are read as overrides of this line and a reader
1159    // should not have to know which way flexbox leans to see that.
1160    //
1161    // No gap. Spacing between members is the app's, the same way this crate
1162    // states no margins anywhere else; a gap here would be a size, and the one
1163    // hardcoded size in the mechanism is makeover-geometry's contact patch.
1164    let _ = writeln!(
1165        css,
1166        ".{run} {{\n    display: flex;\n    flex-wrap: nowrap;\n    align-items: center;\n}}"
1167    );
1168
1169    // The derived minimum, and the whole reason a member can no longer be
1170    // overlapped. `min-width: auto` is flexbox's default for a flex item and is
1171    // *not* the same thing: auto lets an item be compressed below its content
1172    // in a nowrap run, which is how a toolbar ends up drawn over a tab strip
1173    // even without anything leaving the flow.
1174    let _ = writeln!(css, ".{run} > * {{\n    min-width: min-content;\n}}");
1175
1176    for fallback in [
1177        Fallback::Wrap,
1178        Fallback::Stack,
1179        Fallback::Shed,
1180        Fallback::Menu,
1181    ] {
1182        let name = fallback_class(fallback);
1183        let c = class(name, opts);
1184        let _ = writeln!(css, ".{c} {{\n    flex-wrap: wrap;\n}}");
1185        if matches!(fallback, Fallback::Stack) {
1186            let _ = writeln!(css, ".{c} > * {{\n    flex: 1 1 max-content;\n}}");
1187        }
1188    }
1189
1190    css
1191}
1192
1193/// Every class [`fallback_class`] can return, plus the run itself.
1194///
1195/// [`ROW_PART_CLASSES`](crate::list::ROW_PART_CLASSES)'s reasoning and the same
1196/// obligation: a `match` over a `#[non_exhaustive]` enum cannot be enumerated
1197/// from outside, so the list sits beside it and a test holds the two together.
1198/// `run` is in it because it is emitted in its own right rather than only as a
1199/// fallback's fallback.
1200pub const RUN_CLASSES: &[&str] = &["run", "run-wrap", "run-stack", "run-shed", "run-menu"];
1201
1202/// The class a run carries for what it does when it is tight.
1203///
1204/// A run always carries `.run` as well, so an unrecognised fallback -- the enum
1205/// is `#[non_exhaustive]` -- lands as a plain nowrap row with the min-content
1206/// floor still under it. That is the safe failure: every member in flow and
1207/// none overlapped, which is the property, with only the rearrangement missing.
1208#[must_use]
1209pub fn fallback_class(fallback: Fallback) -> &'static str {
1210    match fallback {
1211        Fallback::Wrap => "run-wrap",
1212        Fallback::Stack => "run-stack",
1213        Fallback::Shed => "run-shed",
1214        Fallback::Menu => "run-menu",
1215        _ => "run",
1216    }
1217}
1218
1219/// The progress trough these rules fill.
1220///
1221/// This was renderer-local chrome with nothing behind it until makeover-layout
1222/// 0.10.0, which is the unusual order: the tones below were emitted for every
1223/// bar in the tree while the only way to describe one was to concatenate the
1224/// numbers into a heading. `Meter` is the word that arrived late, and
1225/// [`meter::meter_html`](crate::meter::meter_html) is what now fills these.
1226///
1227/// The rules stay a superset of what a description can ask for. An app drawing
1228/// its own bar keeps these classes, which is what the four goingson grew
1229/// independently were adopted onto.
1230///
1231/// The trough is a [`Depth::Well`], the same reading a text field gets:
1232/// something with its content down inside it.
1233fn progress_rules(opts: &Emit) -> String {
1234    let progress = class("progress", opts);
1235    // `progress-fill` rather than a bare `fill`: an unprefixed build claims
1236    // these names in the app's own stylesheet, and `.fill` is grabby enough to
1237    // catch things that have nothing to do with progress. goingson already
1238    // calls it `.progress-fill`, so this is also the name that deletes.
1239    let fill = class("progress-fill", opts);
1240    let mut css = depth_rule(&progress, Depth::Well);
1241
1242    // The untoned bar is `--action`, not [`Tone::Neutral`]. That is the one
1243    // place this differs from the badge rules, and deliberately: a badge with
1244    // no status is a muted label, while a bar with no status is still
1245    // reporting progress, and `content-muted` would read as disabled.
1246    let _ = writeln!(
1247        css,
1248        ".{progress} > .{fill} {{\n    background: var(--action);\n}}"
1249    );
1250
1251    // A bar can be saying something, same as a badge: goingson colours subtask
1252    // progress as success and an over-estimate as danger, which is real
1253    // information rather than decoration. Emitting the tones is what lets that
1254    // survive adoption instead of staying hand-written.
1255    for tone in [Tone::Info, Tone::Success, Tone::Warning, Tone::Danger] {
1256        let _ = writeln!(
1257            css,
1258            ".{progress} > .{fill}[data-tone=\"{0}\"] {{\n    background: var(--{0});\n}}",
1259            tone.token()
1260        );
1261    }
1262    css
1263}
1264
1265/// What a wait looks like, for the attribute that has been describing one to
1266/// nobody.
1267///
1268/// Wiki `loading-and-progress-standard`, phase 2. `data-awaiting` has been
1269/// emitted since the vocabulary gained `Awaiting`, with
1270/// `data-awaiting="determinate"` and a `data-awaiting-amount` beside it or
1271/// `data-awaiting="indeterminate"` alone, and until 0.60.0 **nothing in the tree
1272/// styled either**. Determinate and indeterminate rendered identically, which is
1273/// to say not at all. This is the half that was missing.
1274///
1275/// # Why it is keyed on `aria-busy` and not on the attribute alone
1276///
1277/// `data-awaiting` is a fact about the control: pressing this waits. It is true
1278/// when the page is painted and it stays true. Whether a wait is *running* is
1279/// true only between two events, so it is the binder's to set, and `aria-busy`
1280/// is the standard spelling of it — announced as well as drawn, which a class
1281/// of our own would not be.
1282///
1283/// That also keeps this crate out of any one client library's vocabulary.
1284/// `quasi-webview` sets `aria-busy` from htmx's request events; a host driving
1285/// the same markup another way sets it the same way and gets the same drawing.
1286///
1287/// # The two drawings
1288///
1289/// One pseudo-element either way, so no renderer has to emit an extra node.
1290///
1291/// Indeterminate is the activity mark of rule 2: a small square that blinks.
1292/// Determinate is a trough with a fill, drawn as a single gradient whose stop is
1293/// `--awaiting-share`, a plain number from 0 to 1 that the binder sets from
1294/// bytes it has actually watched land. A determinate control with nothing
1295/// setting the share draws an empty trough rather than a full one, which is the
1296/// honest reading: the size is known and the delivery is not.
1297///
1298/// **What the bar may not do**, from rule 1 and from `Awaiting`'s own docs:
1299/// what is done over what there is, and never a remaining time, an arrival time
1300/// or a rate extrapolated forward. Nothing here can express one, which is
1301/// deliberate — the only input is a share of a measured payload.
1302///
1303/// # Sizes, and the deferral rule
1304///
1305/// `progress_rules` emits the tones and never the width, because the width is
1306/// not this crate's to know. A pseudo-element has no intrinsic size at all, so
1307/// the same treatment would render nothing anywhere. Both sizes are therefore
1308/// custom properties with defaults: an app that wants a different mark sets
1309/// `--awaiting-mark` and `--awaiting-bar` once, and one that says nothing gets a
1310/// mark that is visible.
1311///
1312/// # The cadence, and what happens without it
1313///
1314/// `--cadence-activity` comes from `makeover-timing` through `makeover-build`,
1315/// and is a half-period, so a full cycle is twice it. It is used bare rather
1316/// than with a fallback: a number written here would be a second heartbeat for
1317/// a mark three renderers draw.
1318///
1319/// A sheet assembled without the time axis leaves `animation-duration` invalid,
1320/// which resolves to `0s`, which runs no animation and leaves the base style —
1321/// a lit, still mark. That is also exactly what the reduced-motion block does,
1322/// since it sets the cadence to `0ms`. Both fall out of one rule because the
1323/// base style is lit and the keyframes do the dimming, which is the ordering
1324/// `makeover_timing::reduced_motion_css` asks its consumers for by name.
1325fn awaiting_rules(opts: &Emit) -> String {
1326    let mut css = String::new();
1327
1328    // Dimming rather than lighting, so a zero-length animation leaves a lit
1329    // mark rather than a blank one. See `makeover_timing::reduced_motion_css`.
1330    css.push_str(
1331        "@keyframes makeover-activity {\n    \
1332         0%, 49.99% {\n        background: var(--action);\n    }\n    \
1333         50%, 100% {\n        background: var(--surface-sunken);\n    }\n\
1334         }\n",
1335    );
1336
1337    // Nothing is drawn until something is waiting. `content` on the base rule
1338    // rather than on the busy one keeps the box the same box across the
1339    // transition, so a mark appearing does not reflow the line it is in.
1340    let _ = writeln!(
1341        css,
1342        "[data-awaiting]::after {{\n    \
1343         content: \"\";\n    \
1344         display: none;\n    \
1345         margin-inline-start: var(--awaiting-gap, 0.5ch);\n    \
1346         vertical-align: baseline;\n\
1347         }}"
1348    );
1349
1350    let _ = writeln!(
1351        css,
1352        "[data-awaiting][aria-busy=\"true\"]::after {{\n    \
1353         display: inline-block;\n    \
1354         inline-size: var(--awaiting-mark, 0.5em);\n    \
1355         block-size: var(--awaiting-mark, 0.5em);\n    \
1356         background: var(--action);\n    \
1357         opacity: 1;\n    \
1358         animation: makeover-activity calc(var(--cadence-activity) * 2) \
1359         step-end infinite;\n\
1360         }}"
1361    );
1362
1363    // The measured half. A wider box, no blink, and a gradient whose stop is
1364    // the share: the fill and the trough in one paint, so the markup stays one
1365    // pseudo-element on both branches.
1366    //
1367    // `--awaiting-share` unset is an empty trough, not a full one. A bar that
1368    // read full because nobody was counting would be the confidently-wrong
1369    // drawing rule 1 exists to forbid.
1370    //
1371    // The trough takes an edge for the reason a well does: a bar at zero share
1372    // is otherwise a rectangle of the surface it sits on, which is nothing at
1373    // all. `border_width` rather than a literal, the way `depth_rule` and
1374    // `track_rules` write theirs.
1375    let _ = writeln!(
1376        css,
1377        "[data-awaiting=\"determinate\"][aria-busy=\"true\"]::after {{\n    \
1378         inline-size: var(--awaiting-bar, 6em);\n    \
1379         animation: none;\n    \
1380         outline: {} solid var(--border);\n    \
1381         outline-offset: -{};\n    \
1382         background: linear-gradient(\n        \
1383         to inline-end,\n        \
1384         var(--action) 0 calc(var(--awaiting-share, 0) * 100%),\n        \
1385         var(--surface-sunken) 0\n    \
1386         );\n\
1387         }}",
1388        opts.border_width, opts.border_width
1389    );
1390
1391    css
1392}
1393
1394/// A strip of figures, and the two spans inside each one.
1395///
1396/// Colour only, which is the deferral rule applied to a component that badly
1397/// wants to break it. A figure reads as a figure because the value is set large
1398/// over a small caption, and that is a size: `makeover-geometry` answers how
1399/// much space and this crate answers what the thing is. Emitting `font-size`
1400/// here would be this crate naming a value, which is the one thing it is defined
1401/// by not doing, and `progress_rules` is the precedent — it emits the tones and
1402/// never the width, because the width is not its to know.
1403///
1404/// So the arrangement and the type scale are the app's, and what is generated is
1405/// the part an app cannot get right by itself: which of the two spans carries
1406/// the tone.
1407fn figure_rules(opts: &Emit) -> String {
1408    let figure = class("figure", opts);
1409    let value = class("figure-value", opts);
1410    let caption = class("figure-caption", opts);
1411    let change = class("figure-change", opts);
1412    let mut css = String::new();
1413
1414    // `content` literally. A figure's value is the thing itself, at full
1415    // weight -- wiki `three-tone-convention` classes it "active, emphasised",
1416    // and both other renderers already draw it that way (makeover-tui
1417    // `piece.rs:391` bold, makeover-immediate `widget.rs:244`). It reached
1418    // here through `Tone::Neutral.token()` and came out `content-muted`, so
1419    // the headline number sat at the colour of its own caption.
1420    let _ = writeln!(
1421        css,
1422        ".{figure} > .{value} {{\n    color: var(--content);\n}}"
1423    );
1424    let _ = writeln!(
1425        css,
1426        ".{figure} > .{caption} {{\n    color: var(--content-muted);\n}}"
1427    );
1428    let _ = writeln!(
1429        css,
1430        ".{figure} > .{change} {{\n    color: var(--content-muted);\n}}"
1431    );
1432
1433    // A toned figure tones one part and never the caption. The caption is the
1434    // noun and stays muted.
1435    //
1436    // Which part depends on whether there is a change, and that is the whole of
1437    // what 0.13.0 changed here. A figure with a delta is an ordinary number that
1438    // has moved in a direction worth reading, so the delta takes the colour and
1439    // the number stays plain; a figure without one has nowhere else to put it.
1440    // `:has` is what lets one attribute mean both, and the alternative was the
1441    // emitter deciding by writing the attribute onto a different element, which
1442    // leaves two elements able to disagree about a figure's one meaning.
1443    for tone in [Tone::Info, Tone::Success, Tone::Warning, Tone::Danger] {
1444        let _ = writeln!(
1445            css,
1446            ".{figure}[data-tone=\"{0}\"] > .{change} {{\n    color: var(--{0});\n}}",
1447            tone.token()
1448        );
1449        let _ = writeln!(
1450            css,
1451            ".{figure}[data-tone=\"{0}\"]:not(:has(> .{change})) > .{value} \
1452             {{\n    color: var(--{0});\n}}",
1453            tone.token()
1454        );
1455    }
1456    css
1457}
1458
1459/// A picture, its frame and its caption.
1460///
1461/// `makeover_layout::Image` arrived at 0.21.0, found by trying to describe MNW's
1462/// carousel and discovering nothing named a picture.
1463///
1464/// # The frame is a border, and this is the one place a bevel is wrong
1465///
1466/// 0.36.0 emitted [`Depth::Raised`] here through [`depth_rule`], the same call
1467/// `button` and `card` make. It was wrong, and MNW's landing page is where it
1468/// showed: **the frames had no visible edge at all.**
1469///
1470/// `Depth::Raised`'s edge is `--bevel-raised`, which is an *inset* shadow — a
1471/// 1px light run at the top-left and a dark one at the bottom-right, drawn
1472/// **inside** the element's box. On a button or a card that box is a surface
1473/// this crate owns, so an inset edge reads as the surface catching the light.
1474/// On a picture it is drawn on top of the picture, over whatever pixels the
1475/// image happens to have at its border. MNW's screenshots are light-on-light
1476/// parchment, so the light half landed on a light image and the frame
1477/// disappeared.
1478///
1479/// **You cannot bevel a surface you do not own.** A picture's content is the
1480/// app's, arrives at request time, and can be any colour, so its edge has to
1481/// sit *outside* the content rather than on it. That is a border.
1482///
1483/// The fill stays, and it is not decoration: it is what shows through a
1484/// transparent PNG and what stands in the frame's place while the image is
1485/// still loading.
1486///
1487/// This is a real limit on [`Depth`] rather than a special case. Every other
1488/// consumer of a depth draws its own surface; a picture is the first member
1489/// whose surface belongs to someone else.
1490///
1491/// # No size
1492///
1493/// `width: 100%` and nothing else. How large a picture is depends on the box it
1494/// was put in, which is the app's arrangement and `makeover-geometry`'s scales,
1495/// and a renderer that picked one would be answering for every consumer at
1496/// once. This is where `figure_rules` landed for the same reason.
1497fn picture_rules(opts: &Emit) -> String {
1498    let picture = class("picture", opts);
1499    let img = class("picture-img", opts);
1500    let caption = class("picture-caption", opts);
1501    let mut css = String::new();
1502
1503    // Block, or an inline image sits on the text baseline and carries a
1504    // descender's worth of space under it that no app ever wants and every app
1505    // deletes by hand. The border is the frame; see the type docs for why it is
1506    // not the bevel every other surface here gets.
1507    let _ = writeln!(
1508        css,
1509        ".{img} {{\n    display: block;\n    width: 100%;\n    height: auto;\n    \
1510         background: var(--{});\n    border: {} solid var(--border);\n}}",
1511        Fill::Raised.token(),
1512        opts.border_width
1513    );
1514
1515    // The two fits that need a rule. `Fit::Natural` emits no attribute at all,
1516    // so it is the bare rule above and needs nothing here.
1517    let _ = writeln!(
1518        css,
1519        ".{img}[data-fit=\"cover\"] {{\n    height: 100%;\n    object-fit: cover;\n}}"
1520    );
1521    let _ = writeln!(
1522        css,
1523        ".{img}[data-fit=\"contain\"] {{\n    height: 100%;\n    object-fit: contain;\n}}"
1524    );
1525
1526    // A caption reads back one step, which is `figure-caption`'s answer and the
1527    // same claim: it says what the thing above it is, and it is not the thing.
1528    let _ = writeln!(
1529        css,
1530        ".{picture} > .{caption} {{\n    color: var(--content-muted);\n}}"
1531    );
1532
1533    css
1534}
1535
1536/// A region showing one child at a time, and the chrome that moves between them.
1537///
1538/// [`makeover_layout::Showing`] lets a description say that a region holds
1539/// several children and shows some of them. A renderer derives its own chrome
1540/// from that, which is what stops every renderer growing a `match` on a widget
1541/// name; these are the rules the derived chrome needs.
1542///
1543/// # Why the default is every child, and the enhancement takes them away
1544///
1545/// The controls are a lie until something binds them. A prev button rendered
1546/// into a document with no script is a control that looks live and answers
1547/// nothing, and the reader it lies to is exactly the one who cannot see the
1548/// other children either — the collapsing and the moving are the same half.
1549///
1550/// So the rules run in the direction the enhancement does. Nothing here hides a
1551/// child and nothing here shows a control. Whatever binds the region sets
1552/// `data-ready` on it, and that is what collapses the stack to one and reveals
1553/// the row that moves it. A reader with no script gets every child in order and
1554/// no controls, which is more content rather than less, and a reader with
1555/// script gets a settled page rather than a stack that jumps to one frame after
1556/// load.
1557///
1558/// MNW proved this shape by hand — a `<noscript>` stylesheet opening its
1559/// carousel back out — and it is here rather than there because the property is
1560/// the description's, not one app's.
1561///
1562/// # Not spacing
1563///
1564/// The row's gaps are `makeover-geometry`'s question and are absent for
1565/// [`row_rules`]'s reason. What is here is `display`, which carries no
1566/// magnitude, and the muted readout, which is the same claim `picture-caption`
1567/// makes: it says where you are among the children and it is not one of them.
1568fn showing_rules(opts: &Emit) -> String {
1569    let controls = class("showing", opts);
1570    let position = class("showing-position", opts);
1571    let frame = class("showing-frame", opts);
1572    let mut css = String::new();
1573
1574    // Hidden until something binds it, which is the whole argument above.
1575    let _ = writeln!(css, ".{controls} {{\n    display: none;\n}}");
1576    // Block, and nothing about how the three sit in it. A button and a span are
1577    // inline already, so they make a row without this crate saying so, and
1578    // saying so is where `align-items` and a gap would follow -- both spacing,
1579    // both `makeover-geometry`'s, and `row_rules` refuses them for the same
1580    // reason.
1581    let _ = writeln!(
1582        css,
1583        "[data-ready] > .{controls} {{\n    display: block;\n}}"
1584    );
1585
1586    // A child is in flow until the region is bound, and then only the current
1587    // one is. `.current` is a modifier for the reason `.chosen` and `.latched`
1588    // are: one name for the state, set by whoever knows it.
1589    let _ = writeln!(
1590        css,
1591        "[data-ready] > .{frame}:not(.current) {{\n    display: none;\n}}"
1592    );
1593
1594    // Reads back one step. `picture-caption`'s rule and its reason.
1595    let _ = writeln!(css, ".{position} {{\n    color: var(--content-muted);\n}}");
1596
1597    css
1598}
1599
1600/// A time axis: the container, its gridlines and ruler, and the placed things.
1601///
1602/// `makeover-layout` 0.24.0's [`Track`](makeover_layout::Track), and the first
1603/// member here whose whole point is *position*. That makes the magnitude line
1604/// this crate keeps worth restating rather than assuming.
1605///
1606/// # Where the numbers come from
1607///
1608/// Every value that varies per item is a custom property the caller sets
1609/// inline, and every rule here reads one. `--track-at` and `--track-for` are
1610/// percentages of the span, which
1611/// [`Track::fraction`](makeover_layout::Track::fraction) computes once so three
1612/// renderers cannot disagree about it. Nothing here knows a pixel.
1613///
1614/// That is what lets the stylesheet stay static while the items move: an entry
1615/// carries `style="--track-at: 37.5%; --track-for: 4.166%"` and the rule below
1616/// turns it into a box. The alternative was emitting a rule per item, which is
1617/// a stylesheet that grows with the data.
1618///
1619/// **The height of the track is the app's**, not this crate's. 96 quarter-hour
1620/// slots at some slot height is a size, and a size is `makeover-geometry`'s
1621/// question -- the same refusal `figure_rules` and `placeholder_rules` make.
1622/// Percentages need a resolved height above them, so `.track` gets
1623/// `position: relative` and nothing else; the app says how tall a day is.
1624///
1625/// # Overlap
1626///
1627/// Two things at the same time is intrinsic to an axis and has no analogue in a
1628/// list. The description does not declare it -- `Placement::overlaps` derives it
1629/// from the times -- so what arrives here is a lane index and a lane count, and
1630/// the rule divides the width. A renderer that would rather stack them ignores
1631/// both properties and the defaults give it one full-width lane.
1632fn track_rules(opts: &Emit) -> String {
1633    let track = class("track", opts);
1634    let slot = class("track-slot", opts);
1635    let tick = class("track-tick", opts);
1636    let entry = class("track-entry", opts);
1637    let mut css = String::new();
1638
1639    // The positioning context every entry resolves against, and the whole of
1640    // what this crate says about the container. No height: see the doc above.
1641    let _ = writeln!(css, ".{track} {{\n    position: relative;\n}}");
1642
1643    // Gridlines and the ruler are the axis reading itself back, which is
1644    // `picture-caption` and `showing-position`'s claim: about the thing rather
1645    // than one of the things.
1646    //
1647    // `border_width` rather than a literal, the way `depth_rule` writes its
1648    // edge. A gridline is the one place a timeline would most naturally reach
1649    // for a hardcoded 1px, and a hardcoded 1px is this crate naming a size.
1650    let _ = writeln!(
1651        css,
1652        ".{slot} {{\n    border-top: {} solid var(--border);\n}}",
1653        opts.border_width
1654    );
1655    let _ = writeln!(css, ".{tick} {{\n    color: var(--content-muted);\n}}");
1656
1657    // The one rule that does real work. Top and height are the placement;
1658    // left and width are the lane. Both lane properties default so an entry
1659    // that names neither is full width, which is the common case and the one a
1660    // renderer gets for free.
1661    let _ = writeln!(
1662        css,
1663        ".{entry} {{\n    \
1664         position: absolute;\n    \
1665         top: var(--track-at, 0%);\n    \
1666         height: var(--track-for, 100%);\n    \
1667         left: calc(var(--track-lane, 0) / var(--track-lanes, 1) * 100%);\n    \
1668         width: calc(100% / var(--track-lanes, 1));\n\
1669         }}"
1670    );
1671
1672    css
1673}
1674
1675/// A region's stand-in, and the header of a table that can be reordered.
1676///
1677/// Both are 0.12.0 members and both are colour and affordance only, which is
1678/// where `figure_rules` landed after trying to emit a type scale. How much room
1679/// a stand-in gets is a size — goingson has the same one at three, as
1680/// `--compact`, `--dashboard` and `--padded` — and a size is
1681/// `makeover-geometry`'s question.
1682///
1683/// The caret is the one thing here that is neither colour nor affordance, and it
1684/// is a renderer's own expression rather than a value the description named:
1685/// `aria-sort` is what the table actually says, and this turns it into something
1686/// visible for everyone not using a screen reader. A terminal draws its own; an
1687/// immediate-mode painter draws its own.
1688fn state_rules(opts: &Emit) -> String {
1689    let placeholder = class("placeholder", opts);
1690    let text = class("placeholder-text", opts);
1691    let heading = class("table-heading", opts);
1692    let mut css = String::new();
1693
1694    let _ = writeln!(
1695        css,
1696        ".{placeholder} > .{text} {{\n    color: var(--content-muted);\n}}"
1697    );
1698    // Only the failure is toned. An empty list is the normal state of a new
1699    // install, and `Readiness::tone` is what says so.
1700    let _ = writeln!(
1701        css,
1702        ".{placeholder}[data-tone=\"{0}\"] > .{text} {{\n    color: var(--{0});\n}}",
1703        Tone::Danger.token()
1704    );
1705
1706    // A header that reorders the table is a control, and the pointer is the
1707    // only part of saying so that is not the app's own type and spacing.
1708    let _ = writeln!(
1709        css,
1710        ".{heading}[data-sortable] {{\n    cursor: pointer;\n}}"
1711    );
1712    // The caret carries its own leading space, the way `makeover-tui` and
1713    // `makeover-immediate` both write `" \u{25B2}"`. It used to be emitted bare,
1714    // and both apps that adopted the vocabulary had to put the gap back in their
1715    // own stylesheets on the same afternoon -- each having to work out first that
1716    // app CSS outranks this crate's cascade layer, so adding the space the
1717    // obvious way, as `content`, silently wins over the glyph and leaves the
1718    // heading with no caret at all. A consumer should not have to know that, and
1719    // with the space emitted here there is nothing left for one to add.
1720    //
1721    // Three states, three tones, on the convention in wiki
1722    // `three-tone-convention`. A column in force is `content`; a column offering
1723    // to reorder and not doing it now is `content-secondary`, because it still
1724    // answers a press; a column that is not sortable emits no caret at all and
1725    // takes nothing. The idle arm used to hide its glyph and reserve the box,
1726    // which cost a reflow-free press and said nothing. It draws now, and the
1727    // reservation stops being a thing to get right.
1728    //
1729    // The glyph is `Sort::glyph`, escaped rather than written: a CSS `content`
1730    // string cannot carry the character literally through this file's own
1731    // escaping, and spelling it here as well would put the third copy back that
1732    // `makeover-layout` 0.27.5 exists to remove.
1733    let _ = writeln!(
1734        css,
1735        ".{heading}[data-sortable]::after \
1736         {{\n    content: \" {}\";\n    color: var(--content-secondary);\n}}",
1737        css_escape(Sort::Ascending.glyph())
1738    );
1739    for direction in [Sort::Ascending, Sort::Descending] {
1740        let _ = writeln!(
1741            css,
1742            ".{heading}[aria-sort=\"{}\"]::after \
1743             {{\n    content: \" {}\";\n    color: var(--content);\n}}",
1744            direction.as_str(),
1745            css_escape(direction.glyph())
1746        );
1747    }
1748    css
1749}
1750
1751/// The frame a table sits in.
1752///
1753/// # Why this is a CSS table and not the grid the rest of the module assumes
1754///
1755/// A grid row needs `grid-template-columns`, which has to name every column in
1756/// order, so it cannot be written without knowing the columns. That is what
1757/// [`list::narrowing_css`] is for, and it works: goingson builds `tables.css` in
1758/// its own `build.rs` out of it, and nothing here changes that.
1759///
1760/// It does not work for a table a *description* produced. Those columns are
1761/// known at render time rather than at build time, and the rules would have to
1762/// travel with the markup: a `<style>` element per table, which needs
1763/// `style-src 'unsafe-inline'` and so blocks the MNW server's standing plan to
1764/// drop it. The head is not an escape: a table swapped in by htmx after a
1765/// delete arrives as a fragment with no head at all.
1766///
1767/// A CSS table aligns its columns across rows knowing nothing about how many
1768/// there are, so there is no track list to emit and nothing per-table to carry.
1769/// The cost is that a described table cannot take a per-column fixed length,
1770/// which costs nothing today: [`Sizing`](list::Sizing) is looked up by column
1771/// name and a description carries no lengths to put in it, so every track a
1772/// described table could ask for is already content, fill or auto.
1773///
1774/// [`Priority`] hiding is unchanged in kind. It moves from a generated
1775/// `display: none` per dropped column to one rule per drop class, which is the
1776/// same fact addressed by class rather than by cutoff, and still never by
1777/// position.
1778fn table_rules(opts: &Emit) -> String {
1779    let table = class("table", opts);
1780    let head = class("table-head", opts);
1781    let row = class("table-row", opts);
1782    let heading = class("table-heading", opts);
1783    let cell = class("cell", opts);
1784    let mut css = String::new();
1785
1786    let _ = writeln!(
1787        css,
1788        ".{table} {{\n    display: table;\n    width: 100%;\n}}"
1789    );
1790    let _ = writeln!(css, ".{head},\n.{row} {{\n    display: table-row;\n}}");
1791    // Scoped under `.{table}` rather than keyed on the classes alone. The
1792    // table model is what a cell takes *by being in a table*, and only there:
1793    // goingson's task headings are `.table-heading` inside a CSS grid, so the
1794    // unscoped rule reached them, and the sort caret and the sortable cursor
1795    // -- which `state_rules` keys on `.table-heading` and rightly still does
1796    // -- came with a `display` the grid had to blockify away.
1797    //
1798    // It costs no markup anywhere: quasi-webview always nests the heading
1799    // inside the table (`quasi-webview/src/node.rs:1591`, `:1605`, `:1615`).
1800    let _ = writeln!(
1801        css,
1802        ".{table} .{heading},\n.{table} .{cell} {{\n    display: table-cell;\n}}"
1803    );
1804
1805    // A content column shrinks to what is in it. `width: 1%` is how a CSS table
1806    // is told that: auto layout hands the slack to the columns that asked for
1807    // room, and a column asking for almost none gets what it needs and no more.
1808    // The `nowrap` is what stops it being given less by wrapping.
1809    let _ = writeln!(
1810        css,
1811        ".{} {{\n    white-space: nowrap;\n    width: 1%;\n}}",
1812        class("cell-content", opts)
1813    );
1814
1815    // A fixed column has no length to be fixed to. The description carries none
1816    // and `Sizing` is not reachable from here, so it behaves as content: the
1817    // honest answer to a width nobody supplied, and the same one `Sizing::track`
1818    // gives it.
1819    let _ = writeln!(
1820        css,
1821        ".{} {{\n    white-space: nowrap;\n}}",
1822        class("cell-fixed", opts)
1823    );
1824
1825    // Optional columns go at the narrowest class, secondary ones go with them,
1826    // which is the cutoff walk `kept_at` describes said as two media queries.
1827    // Essential columns have no rule at all, because never dropping is what not
1828    // being mentioned already means.
1829    for (size, drops) in [
1830        (
1831            SizeClass::Compact,
1832            &["cell-drops-first", "cell-drops-next"][..],
1833        ),
1834        (SizeClass::Medium, &["cell-drops-first"][..]),
1835    ] {
1836        let selectors: Vec<String> = drops
1837            .iter()
1838            .map(|drop| format!(".{}", class(drop, opts)))
1839            .collect();
1840        css.push_str(&gated(
1841            Some(&size.media_condition()),
1842            &format!("{} {{\n    display: none;\n}}\n", selectors.join(",\n")),
1843        ));
1844    }
1845
1846    // What is inside a cell, which the table side could not say until
1847    // makeover-layout 0.14.0. Every cell was one `.cell` and one content
1848    // colour, so a button in a cell was painted as text -- the drift
1849    // `RowPart::intent` has prevented for list rows since 0.2.0 and prevented
1850    // for nothing here.
1851    //
1852    // The colour goes on `.cell-value` rather than on `.cell`, and that
1853    // placement is the whole fix. On the container it would cascade into the
1854    // tokens and the controls sitting beside the text, which is the bug said
1855    // in one rule; on the part that is text, it reaches text and stops.
1856    for part in [
1857        CellPart::Value,
1858        CellPart::Tokens,
1859        CellPart::Actions,
1860        CellPart::Link,
1861    ] {
1862        // Three of the four inherit, each for its own reason: a token carries
1863        // its own tone, an action is a control rather than text, and a link
1864        // takes the action colour from the anchor it is. `CellPart::intent`
1865        // says so by answering with the intent inheriting already gives, and
1866        // pinning that would be louder than saying nothing.
1867        //
1868        // Written as a skip-list rather than as a match on Value, so a member
1869        // added upstream gets its intent emitted rather than being silently
1870        // dropped. That is the same trade `part_class`'s fallback makes: land
1871        // plainly, never land as nothing.
1872        if !matches!(part, CellPart::Tokens | CellPart::Actions | CellPart::Link) {
1873            let _ = writeln!(
1874                css,
1875                ".{} {{\n    color: var(--{});\n}}",
1876                class(cell_part_class(part), opts),
1877                part.intent()
1878            );
1879        }
1880    }
1881
1882    css
1883}
1884
1885/// The component layer: every named thing phase A emits.
1886///
1887/// No scrollbar track. It was on the phase A list and came off: eight lines of
1888/// `::-webkit-scrollbar` with no shape a terminal or an immediate-mode painter
1889/// would want handed to it, so it stays with the apps.
1890#[must_use]
1891pub fn component_rules(opts: &Emit) -> String {
1892    let mut css = String::new();
1893    css.push_str(&surface_rules(opts));
1894    css.push_str(&link_rules(opts));
1895    css.push_str(&token_rules(opts));
1896    css.push_str(&selector_rules(opts));
1897    css.push_str(&row_rules(opts));
1898    css.push_str(&run_rules(opts));
1899    css.push_str(&progress_rules(opts));
1900    css.push_str(&awaiting_rules(opts));
1901    css.push_str(&figure_rules(opts));
1902    css.push_str(&picture_rules(opts));
1903    css.push_str(&showing_rules(opts));
1904    css.push_str(&track_rules(opts));
1905    css.push_str(&state_rules(opts));
1906    css.push_str(&table_rules(opts));
1907    css.push_str(&facet::facet_rules(opts));
1908    css.push_str(&form::editor_rules(opts));
1909    css.push_str(&form::suggestion_rules(opts));
1910    css.push_str(&form::unit_rules(opts));
1911    css.push_str(&form::note_rules(opts));
1912    css
1913}
1914
1915/// The whole phase-A stylesheet: properties, depth rules and components, in
1916/// [`CSS_LAYER`], under a generated-file banner.
1917///
1918/// The banner sits outside the layer, because a comment participates in no
1919/// cascade and a reader opening the file should see what it is before seeing
1920/// an at-rule.
1921#[must_use]
1922pub fn stylesheet(opts: &Emit) -> String {
1923    let body = in_css_layer(&format!(
1924        ":root {{\n{}}}\n\n{}\n{}",
1925        bevel_properties(opts),
1926        depth_rules(opts),
1927        component_rules(opts)
1928    ));
1929    // Counted from the body rather than through `vocabulary::names`, which
1930    // calls back into here.
1931    let classes = vocabulary::classes_in_css(&body).len();
1932    let version = VERSION;
1933    format!(
1934        "/* Generated by makeover-webview {version} from makeover-layout, \
1935         {classes} classes.\n   \
1936         Do not edit. The version and the count are here because a stale\n   \
1937         lockfile fails silently: an older emitter writes a well-formed sheet\n   \
1938         with components missing, and nothing else in the file says so. If\n   \
1939         this version trails what the manifest asks for, re-resolve.\n\n   \
1940         Depth is a fill and an edge together; naming them apart is what let\n   \
1941         them disagree. See the crate's README and wiki note makeover-layout.\n\n   \
1942         Everything below is in the `{CSS_LAYER}` cascade layer. Declare the\n   \
1943         order once in your own stylesheet, or this layer's position is decided\n   \
1944         by whichever generated file the browser happens to see first:\n\n   \
1945         @layer {CSS_LAYER}, base, components, responsive; */\n{body}"
1946    )
1947}
1948
1949#[cfg(test)]
1950mod tests {
1951    use super::*;
1952    use makeover_layout::Edge;
1953
1954    #[test]
1955    fn every_fallback_class_is_one_a_checker_knows_about() {
1956        // The obligation ROW_PART_CLASSES carries, for the same reason: a class
1957        // this crate can write and the vocabulary list does not carry is
1958        // invisible to the dead-vocabulary seal and to the overlap check both.
1959        for fallback in [
1960            Fallback::Wrap,
1961            Fallback::Stack,
1962            Fallback::Shed,
1963            Fallback::Menu,
1964        ] {
1965            assert!(
1966                RUN_CLASSES.contains(&fallback_class(fallback)),
1967                "{fallback:?} is missing from RUN_CLASSES"
1968            );
1969        }
1970        let names = crate::vocabulary::names(&Emit::default());
1971        for name in RUN_CLASSES {
1972            assert!(names.contains(*name), "{name} is not in the vocabulary");
1973        }
1974    }
1975
1976    #[test]
1977    fn a_run_gives_every_member_a_floor_it_cannot_be_squeezed_below() {
1978        // The whole of what stops the overlap, and it is not a fallback: it
1979        // applies to every run whatever the group declared. flexbox's default
1980        // min-width is auto, which lets an item be compressed below its own
1981        // content in a nowrap row, and that is how a toolbar is drawn over a
1982        // tab strip even with nothing out of flow.
1983        let css = run_rules(&Emit::default());
1984        assert!(css.contains(".run > * {\n    min-width: min-content;\n}"));
1985        // No number anywhere in it. The minimum is derived by the browser from
1986        // what the members contain, which is the ruling's own requirement.
1987        assert!(!css.contains("px"));
1988        assert!(!css.contains("rem"));
1989        assert!(!css.contains("@media"));
1990    }
1991
1992    #[test]
1993    fn room_is_never_asked_of_the_viewport() {
1994        // The 913 case: a window in SizeClass::Expanded holding a group out of
1995        // room. A viewport query answers about the window and would be wrong
1996        // about the group, which is why the table's @media walk is not the
1997        // precedent this follows.
1998        let css = run_rules(&Emit::default());
1999        for size in [SizeClass::Compact, SizeClass::Medium] {
2000            assert!(!css.contains(&size.media_condition()));
2001        }
2002    }
2003
2004    #[test]
2005    fn every_fallback_lands_as_a_class_and_an_unknown_one_lands_plainly() {
2006        let css = run_rules(&Emit::default());
2007        for fallback in [
2008            Fallback::Wrap,
2009            Fallback::Stack,
2010            Fallback::Shed,
2011            Fallback::Menu,
2012        ] {
2013            let class = fallback_class(fallback);
2014            assert!(css.contains(&format!(".{class} {{")), "{class} unemitted");
2015        }
2016        // Stack is the one that also says what a member does with the line it
2017        // took, which is what separates it from wrapping.
2018        assert!(css.contains(".run-stack > * {\n    flex: 1 1 max-content;\n}"));
2019    }
2020
2021    #[test]
2022    fn the_emitted_bevel_matches_what_the_apps_already_hand_write() {
2023        // Balanced Breakfast's styles.css, verbatim. Adoption has to be a
2024        // deletion, not a redesign, or nobody will take it.
2025        let opts = Emit::default();
2026        assert_eq!(
2027            bevel_shadow(Bevel::Raised, &opts),
2028            "inset 1px 1px 0 var(--bevel-light), inset -1px -1px 0 var(--bevel-dark)"
2029        );
2030        assert_eq!(
2031            bevel_shadow(Bevel::Inset, &opts),
2032            "inset 1px 1px 0 var(--bevel-dark), inset -1px -1px 0 var(--bevel-light)"
2033        );
2034    }
2035
2036    #[test]
2037    fn no_colour_ever_reaches_the_output() {
2038        let css = stylesheet(&Emit::default());
2039        assert!(!css.contains('#'), "a hex literal escaped into the CSS");
2040        assert!(
2041            !css.contains("rgb"),
2042            "a colour function escaped into the CSS"
2043        );
2044        // Every colour is named, never resolved.
2045        assert!(css.contains("var(--surface-raised)"));
2046        assert!(css.contains("var(--bevel-light)"));
2047    }
2048
2049    #[test]
2050    fn a_well_falls_back_through_css_rather_than_through_rust() {
2051        assert_eq!(
2052            fill_var(Fill::Well),
2053            "var(--surface-well, var(--surface-page))"
2054        );
2055        // Nothing else needs one.
2056        assert_eq!(fill_var(Fill::Raised), "var(--surface-raised)");
2057        assert_eq!(fill_var(Fill::Page), "var(--surface-page)");
2058    }
2059
2060    #[test]
2061    fn raised_and_well_do_not_collapse_onto_each_other() {
2062        let css = depth_rules(&Emit::default());
2063        assert!(css.contains(".raised {"));
2064        assert!(css.contains(".well {"));
2065        assert!(css.contains("var(--bevel-raised)"));
2066        assert!(css.contains("var(--bevel-inset)"));
2067    }
2068
2069    /// The cast shadow is composed here from the tone `makeover` derives, so
2070    /// neither crate has to hold the other's numbers.
2071    ///
2072    /// It is a `:root` property and deliberately not a depth class. There is no
2073    /// `Depth::Overlay` in the description layer, and adding one would be a
2074    /// claim about what a screen means rather than about how it is painted;
2075    /// until something asks for it, a consumer names the property on the rule
2076    /// for the menu or the toast it already has.
2077    #[test]
2078    fn the_cast_shadow_is_a_root_property_not_a_depth() {
2079        let css = bevel_properties(&Emit::default());
2080        assert!(css.contains("--elevation-overlay:"));
2081        assert!(css.contains("var(--elevation)"));
2082        assert!(
2083            !depth_rules(&Emit::default()).contains("elevation"),
2084            "elevation is not a depth class"
2085        );
2086    }
2087
2088    #[test]
2089    fn the_cascade_carries_the_pressed_state() {
2090        let css = surface_rules(&Emit::default());
2091        // The one thing this renderer gets free that the other two resolve by
2092        // hand, eighteen call sites deep in audiofiles' case. Asserted on a
2093        // named surface: pressing belongs to the control, not to the depth.
2094        assert!(css.contains(".card:active {"));
2095        assert!(css.contains(".button:active {"));
2096    }
2097
2098    #[test]
2099    fn the_depth_class_is_a_surface_and_not_a_control() {
2100        let css = depth_rules(&Emit::default());
2101        // The static surface the vocabulary was missing. Sixteen goingson
2102        // elements wore .card and cancelled its hover and press to get this,
2103        // because a raised object that is not pressable had no other spelling.
2104        for state in [":hover", ":active", ":focus-visible", ":disabled"] {
2105            assert!(
2106                !css.contains(&format!(".raised{state}")),
2107                "the depth class claimed {state}: {css}"
2108            );
2109        }
2110        assert!(css.contains("var(--bevel-raised)"), "still raised: {css}");
2111    }
2112
2113    #[test]
2114    fn pressing_moves_the_fill_and_not_only_the_edge() {
2115        // The decision-1 guard, and the regression that mattered: emitting the
2116        // bevel flip alone is what left goingson hand-writing `background:
2117        // var(--surface-sunken)` on .btn, .card and .tag/.badge alike, so none
2118        // of the three could be deleted.
2119        let pressed = interactive_rules("button", Depth::Raised, &Emit::default());
2120        assert!(pressed.contains(".button:active {"));
2121        assert!(
2122            pressed.contains("background: var(--surface-well, var(--surface-page))"),
2123            "pressed dropped its fill: {pressed}"
2124        );
2125        assert!(pressed.contains("box-shadow: var(--bevel-inset)"));
2126    }
2127
2128    #[test]
2129    fn pressed_takes_its_fill_from_the_description_not_from_the_app() {
2130        // goingson presses to --surface-sunken. The description says a pressed
2131        // raised region reads as a well, and makeover says outright that
2132        // surface-sunken cannot serve as one, so the app is the thing that
2133        // moves.
2134        //
2135        // Scoped to the pressed rules rather than to the whole sheet: since
2136        // makeover-layout 0.3.0 an unchosen tab is legitimately
2137        // --surface-sunken, so the token appearing somewhere in the output no
2138        // longer means the app's choice leaked in.
2139        let css = stylesheet(&Emit::default());
2140        let mut checked = 0;
2141        for rule in css.split("}\n") {
2142            if !rule.contains(":active") {
2143                continue;
2144            }
2145            checked += 1;
2146            assert!(
2147                !rule.contains("surface-sunken"),
2148                "a pressed rule took the app's fill: {rule}"
2149            );
2150        }
2151        assert!(checked > 0, "no pressed rules found to check");
2152        assert_eq!(
2153            Depth::Raised.pressed().fill(),
2154            Some(Fill::Well),
2155            "the description changed under us"
2156        );
2157    }
2158
2159    #[test]
2160    fn the_whole_stylesheet_is_emitted_in_the_family_layer() {
2161        // The point of 0.11.0. Unlayered normal declarations outrank every
2162        // named layer, so an app declaring `@layer base, components` loses
2163        // every rule it owns to this file until this file is layered too.
2164        let css = stylesheet(&Emit::default());
2165        assert!(css.contains(&format!("@layer {CSS_LAYER} {{")));
2166
2167        // Exactly one layer block, and nothing outside it but the banner.
2168        assert_eq!(css.matches("@layer").count(), 2, "banner names it once");
2169        let opened = css.find("@layer makeover {").expect("layer opens");
2170        for (i, line) in css.lines().enumerate() {
2171            let before_layer = css.lines().take(i).map(str::len).sum::<usize>() < opened;
2172            if before_layer || line.is_empty() {
2173                continue;
2174            }
2175            assert!(
2176                line.starts_with("    ") || line == "}" || line.starts_with("   "),
2177                "line outside the layer: {line:?}"
2178            );
2179        }
2180    }
2181
2182    #[test]
2183    fn the_generated_sheet_carries_no_trailing_whitespace() {
2184        // A checked-in generated file that a formatter wants to rewrite is a
2185        // diff every time somebody saves it.
2186        let css = stylesheet(&Emit::default());
2187        for (i, line) in css.lines().enumerate() {
2188            assert_eq!(line, line.trim_end(), "trailing whitespace on line {i}");
2189        }
2190    }
2191
2192    #[test]
2193    fn the_banner_tells_an_app_how_to_order_the_layer() {
2194        // Without a declared order the layer's position depends on which
2195        // generated file the browser sees first, which is not a contract.
2196        let css = stylesheet(&Emit::default());
2197        assert!(css.contains("@layer makeover, base, components, responsive;"));
2198        // And the banner is outside the layer, not a rule inside it.
2199        assert!(css.starts_with("/* Generated by makeover-webview"));
2200    }
2201
2202    #[test]
2203    fn the_banner_names_the_emitter_so_a_stale_pin_is_visible_on_sight() {
2204        // A consumer whose lockfile pins an old version gets a well-formed
2205        // sheet with components missing and no error. balanced_breakfast ran
2206        // on 657 bytes from a 0.1.0 emitter while its manifest asked for
2207        // 0.5.1, and the only way it surfaced was diffing two apps' generated
2208        // files. The version and the count are what the file says instead.
2209        let css = stylesheet(&Emit::default());
2210        let banner = css.lines().next().unwrap();
2211        assert!(
2212            banner.contains(VERSION),
2213            "{banner} does not name the emitter"
2214        );
2215        let classes = vocabulary::classes_in_css(&css).len();
2216        assert!(classes > 0);
2217        assert!(
2218            banner.contains(&format!("{classes} classes")),
2219            "{banner} does not carry the class count"
2220        );
2221    }
2222
2223    #[test]
2224    fn a_primitive_owns_every_state_it_implies() {
2225        // The whole point of 0.10.0. Anything emitting a hover rule owes the
2226        // other three, or the consuming app supplies them by out-specifying a
2227        // rule it does not own: 19 such rules in goingson, 21 in the MNW
2228        // server, and three focus rings that do not match.
2229        let css = stylesheet(&Emit::default());
2230        for selector in ["button", "card", "chip", "tab", "segment", "toggle"] {
2231            assert!(css.contains(&format!(".{selector}:hover {{")), "{selector}");
2232            assert!(
2233                css.contains(&format!(".{selector}:active {{")),
2234                "{selector}"
2235            );
2236            assert!(
2237                css.contains(&format!(".{selector}:focus-visible {{")),
2238                "{selector} has no focus ring"
2239            );
2240            assert!(
2241                css.contains(&format!(".{selector}:disabled,")),
2242                "{selector} has no disabled state"
2243            );
2244        }
2245    }
2246
2247    #[test]
2248    fn a_field_takes_focus_and_refuses_input_without_taking_a_hover() {
2249        // A text field does not light up under the pointer, so it gets the two
2250        // states it has and not the two it does not.
2251        let css = stylesheet(&Emit::default());
2252        assert!(css.contains(".field:focus-visible {"));
2253        assert!(css.contains(".field:disabled,"));
2254        assert!(!css.contains(".field:hover {"));
2255        assert!(!css.contains(".field:active {"));
2256    }
2257
2258    #[test]
2259    fn disabled_is_emitted_after_hover_so_source_order_settles_it() {
2260        // Every one of these selectors is specificity (0,2,0), so nothing but
2261        // order decides which wins. A disabled button taking the hover fill is
2262        // the exact bug goingson's `.button:disabled:hover` was written to fix,
2263        // and the reason it had to reach (0,3,0) to do it.
2264        let css = interactive_rules("button", Depth::Raised, &Emit::default());
2265        let hover = css.find(":hover").expect("hover");
2266        let active = css.find(":active").expect("active");
2267        let focus = css.find(":focus-visible").expect("focus");
2268        let disabled = css.find(":disabled").expect("disabled");
2269        assert!(hover < active && active < focus && focus < disabled);
2270
2271        // And it restores the surface, or the hover fill survives underneath.
2272        let tail = &css[disabled..];
2273        assert!(tail.contains("background: var(--surface-raised)"));
2274    }
2275
2276    #[test]
2277    fn a_disabled_state_reaches_things_that_cannot_be_disabled() {
2278        // `:disabled` matches form elements only, and a chip is a div. Keying
2279        // on the ARIA attribute too is the pattern the invalid field already
2280        // set: one fact, read by the styling and the accessibility tree alike.
2281        let css = disabled_rule("chip", Depth::Raised);
2282        assert!(css.contains(".chip:disabled,"));
2283        assert!(css.contains(".chip[aria-disabled=\"true\"]"));
2284        assert!(css.contains("cursor: not-allowed"));
2285    }
2286
2287    #[test]
2288    fn the_focus_ring_does_not_disturb_the_bevel_it_lands_on() {
2289        // `outline` has its own property, so unlike the invalid ring there is
2290        // no bevel to restate beside it and nothing to keep in agreement.
2291        let opts = Emit::default();
2292        let css = focus_rule("button", Depth::Raised, &opts);
2293        assert!(css.contains("outline: 2px solid var(--focus-ring)"));
2294        assert!(!css.contains("box-shadow"), "the ring restated the bevel");
2295    }
2296
2297    #[test]
2298    fn a_well_takes_the_ring_inside_and_a_raised_surface_outside() {
2299        // One ring, placed by depth. The offset comes off `Depth::bevel` and
2300        // not off a per-component choice, which is what gave three apps three
2301        // different rings.
2302        let opts = Emit::default();
2303        assert!(focus_rule("field", Depth::Well, &opts).contains("outline-offset: calc(-1 * 2px)"));
2304        assert!(focus_rule("button", Depth::Raised, &opts).contains("outline-offset: 2px"));
2305        // Nothing to sit inside of, so it sits outside.
2306        assert!(focus_rule("badge", Depth::Sunken, &opts).contains("outline-offset: 2px"));
2307
2308        // And the ring is not the bevel. Reusing border_width emitted a 1px
2309        // ring that every consumer had already overridden.
2310        assert_ne!(opts.focus_width, opts.border_width);
2311    }
2312
2313    #[test]
2314    fn hover_is_gated_on_capability_and_the_keyboard_path_is_not() {
2315        // goingson's section 60 exists only to take back the hover state this
2316        // crate handed it. Gating at the source is what deletes that section
2317        // in all three apps rather than having each fight for it.
2318        let css = stylesheet(&Emit::default());
2319        let condition = format!("@media {}", Density::Pointer.media_condition());
2320        assert!(css.contains(&condition));
2321
2322        // What is gated is every hover state the surfaces carry. The row's
2323        // actions used to be the other half of this test and are not gated any
2324        // more, because they are not hidden any more: a rule that reveals
2325        // nothing needs no capability answer.
2326        let gated: Vec<&str> = css.lines().filter(|line| line.contains(":hover")).collect();
2327        assert!(!gated.is_empty(), "{css}");
2328        for line in gated {
2329            let indent = line.len() - line.trim_start().len();
2330            assert!(indent > 4, "an ungated hover rule: {line}");
2331        }
2332        assert!(!css.contains(".row:hover"), "{css}");
2333    }
2334
2335    #[test]
2336    fn the_capability_answer_is_asked_for_and_not_assumed() {
2337        // Both halves come from the crates that own them. If `makeover-touch`
2338        // ever says a fingertip has hover, this stops gating on its own.
2339        assert!(!Affordance::Hover.available(Density::Touch, SizeClass::Compact));
2340        assert!(Affordance::Hover.available(Density::Pointer, SizeClass::Compact));
2341        assert_eq!(hover_condition(), Some(Density::Pointer.media_condition()));
2342
2343        // And the size class passed to that call is not a claim about width.
2344        assert!(Affordance::Hover.reads_density());
2345        for size in [SizeClass::Compact, SizeClass::Medium, SizeClass::Expanded] {
2346            assert!(!Affordance::Hover.available(Density::Touch, size));
2347        }
2348    }
2349
2350    #[test]
2351    fn hover_resolves_against_the_token_makeover_already_derives() {
2352        let css = interactive_rules("card", Depth::Raised, &Emit::default());
2353        assert!(css.contains(".card:hover {"));
2354        assert!(css.contains("background: var(--hover-surface)"));
2355        // Not the app's choice, which was --surface-overlay.
2356        assert!(!css.contains("surface-overlay"));
2357    }
2358
2359    #[test]
2360    fn a_badge_gets_no_edge_and_no_fill() {
2361        // Decision 2, and the one visible redesign in phase A. Token::Badge is
2362        // Flat: an edge on a label says it can be pressed.
2363        let css = token_rules(&Emit::default());
2364        let badge = css
2365            .lines()
2366            .skip_while(|l| !l.starts_with(".badge {"))
2367            .take_while(|l| !l.starts_with('}'))
2368            .collect::<Vec<_>>()
2369            .join("\n");
2370        assert!(!badge.contains("box-shadow"), "badge kept an edge: {badge}");
2371        assert!(!badge.contains("background"), "badge kept a fill: {badge}");
2372        assert_eq!(Token::Badge.depth(false), Depth::Flat);
2373        assert_eq!(Token::Badge.depth(true), Depth::Flat);
2374    }
2375
2376    #[test]
2377    fn a_badge_carries_a_tone_and_neutral_is_the_bare_class() {
2378        let css = token_rules(&Emit::default());
2379        // Neutral is the absence of a status, not a status named "none".
2380        assert!(css.contains(".badge {\n    color: var(--content-muted);"));
2381        assert!(!css.contains("data-tone=\"content-muted\""));
2382        for tone in ["info", "success", "warning", "danger"] {
2383            assert!(
2384                css.contains(&format!(".badge[data-tone=\"{tone}\"]")),
2385                "missing tone {tone}"
2386            );
2387            assert!(css.contains(&format!("color: var(--{tone})")));
2388        }
2389    }
2390
2391    #[test]
2392    fn a_chip_is_raised_and_latches_into_a_well() {
2393        let css = token_rules(&Emit::default());
2394        assert!(css.contains(".chip {"));
2395        assert!(css.contains(".chip.latched {"));
2396        assert!(css.contains(".chip:active {"));
2397        // The whole difference from a badge: it answers a click.
2398        assert!(Token::Chip { removable: false }.interactive());
2399        assert!(!Token::Badge.interactive());
2400    }
2401
2402    #[test]
2403    fn only_a_tab_comes_forward_when_chosen() {
2404        // The folder semantic. Collapsing the three selectors would lose it.
2405        let css = selector_rules(&Emit::default());
2406        assert!(css.contains(".tab.chosen {"));
2407        assert!(css.contains(".segment.chosen {"));
2408        assert!(css.contains(".toggle.chosen {"));
2409        assert_eq!(Selector::Tabs.chosen(), Depth::Raised);
2410        assert_eq!(Selector::Segmented.chosen(), Depth::Well);
2411        assert_eq!(Selector::Toggle.chosen(), Depth::Well);
2412
2413        let tab = css
2414            .lines()
2415            .skip_while(|l| !l.starts_with(".tab.chosen {"))
2416            .take_while(|l| !l.starts_with('}'))
2417            .collect::<Vec<_>>()
2418            .join("\n");
2419        assert!(
2420            tab.contains("var(--bevel-raised)"),
2421            "tab was held in: {tab}"
2422        );
2423    }
2424
2425    #[test]
2426    fn an_unchosen_tab_recedes_without_looking_picked() {
2427        let css = selector_rules(&Emit::default());
2428        // Recessed by colour and given no edge. An edge would make every option
2429        // look picked; flat would leave the chosen one nothing to come forward
2430        // from, which is the gap makeover-layout 0.3.0 closed.
2431        assert!(
2432            css.contains(".tab {\n    background: var(--surface-sunken);\n}"),
2433            "unchosen tab is not recessed: {css}"
2434        );
2435        assert_eq!(Selector::Tabs.unchosen(), Depth::Sunken);
2436        assert!(css.contains(".tab:hover {"));
2437    }
2438
2439    #[test]
2440    fn a_segment_stands_up_so_the_chosen_one_can_be_held_in() {
2441        // The inverse of the tab, and why the three selectors are not one
2442        // member with a flag.
2443        let css = selector_rules(&Emit::default());
2444        assert!(css.contains(".segment {\n    background: var(--surface-raised);"));
2445        assert_eq!(Selector::Segmented.unchosen(), Depth::Raised);
2446        assert_eq!(Selector::Segmented.chosen(), Depth::Well);
2447    }
2448
2449    #[test]
2450    fn a_rows_actions_are_shown_at_rest() {
2451        let css = row_rules(&Emit::default());
2452
2453        // The hover reveal is gone, and with it every escape it needed. What
2454        // it hid was hidden from pointer users alone, who are the ones
2455        // scanning a list to learn what can be done to a row.
2456        assert!(!css.contains("opacity"), "{css}");
2457        assert!(!css.contains("pointer-events"), "{css}");
2458        assert!(!css.contains(":hover"), "{css}");
2459        assert!(!css.contains(":focus-within"), "{css}");
2460
2461        // Nor is it hidden any other way. `display: none` would reflow the row
2462        // and `visibility: hidden` would take the actions out of the focus
2463        // order; the point is that neither is reached for.
2464        assert!(!css.contains("display: none"), "{css}");
2465        assert!(!css.contains("visibility:"), "{css}");
2466    }
2467
2468    #[test]
2469    fn a_figures_tone_lands_on_the_delta_when_there_is_one() {
2470        // 0.13.0. The delta is the part that reads as good or bad; the number
2471        // itself is an ordinary fact. A figure with no delta has nowhere else to
2472        // put the colour, so the value takes it, and `:has` is what lets one
2473        // attribute mean both without the emitter choosing an element.
2474        let css = stylesheet(&Emit::default());
2475
2476        assert!(
2477            css.contains(".figure[data-tone=\"success\"] > .figure-change"),
2478            "{css}"
2479        );
2480        assert!(
2481            css.contains(
2482                ".figure[data-tone=\"success\"]:not(:has(> .figure-change)) > .figure-value"
2483            ),
2484            "{css}"
2485        );
2486        // The caption is the noun and never takes the tone.
2487        assert!(
2488            !css.contains("[data-tone=\"success\"] > .figure-caption"),
2489            "{css}"
2490        );
2491    }
2492
2493    #[test]
2494    fn the_three_text_parts_take_their_intents_and_actions_inherits() {
2495        let css = row_rules(&Emit::default());
2496        assert!(css.contains(".row-primary {\n    color: var(--content);"));
2497        assert!(css.contains(".row-secondary {\n    color: var(--content-secondary);"));
2498        assert!(css.contains(".row-meta {\n    color: var(--content-muted);"));
2499        // Actions carry controls, not text. Pinning the colour it would inherit
2500        // anyway is louder than saying nothing.
2501        assert!(!css.contains(".row-actions {\n    color:"));
2502    }
2503
2504    #[test]
2505    fn the_token_strip_takes_no_colour_of_its_own() {
2506        // makeover-layout 0.9.0. A token carries its own tone, so a colour on
2507        // the strip would be a rule fighting the things sitting in it -- the
2508        // same reasoning as actions, reached for a different reason.
2509        let css = row_rules(&Emit::default());
2510        assert!(!css.contains(".row-tokens {\n    color:"));
2511    }
2512
2513    #[test]
2514    fn an_unknown_row_part_renders_plainly_rather_than_failing_to_build() {
2515        // What `#[non_exhaustive]` bought and what it cost. `part_class` can no
2516        // longer be exhaustive, so a member added upstream lands as a bare
2517        // class with no rule instead of stopping the build. Asserting the
2518        // fallback exists is what keeps it from being written as `unreachable!`
2519        // by someone who reads the match as closed.
2520        assert_eq!(part_class(RowPart::Tokens), "row-tokens");
2521        assert_eq!(part_class(RowPart::Meta), "row-meta");
2522    }
2523
2524    #[test]
2525    fn a_link_takes_the_action_colour_the_theme_actually_defines() {
2526        // `--action-primary` shipped here for months and no theme has ever
2527        // defined it, so every `.link` dropped its colour declaration outright
2528        // and fell back to inherited text. Nothing caught it because the sheet
2529        // is valid CSS either way; MNW's no-undefined-token lint is what found
2530        // it, 2026-08-14. The hover arm two lines below was always `--action-hover`,
2531        // which is what makes the typo legible in hindsight.
2532        let css = link_rules(&Emit::default());
2533        assert!(css.contains("color: var(--action);"));
2534        assert!(!css.contains("--action-primary"));
2535        assert!(css.contains("color: var(--action-hover);"));
2536    }
2537
2538    #[test]
2539    fn the_progress_trough_is_a_well() {
2540        let css = progress_rules(&Emit::default());
2541        assert!(css.contains(".progress {"));
2542        assert!(css.contains("box-shadow: var(--bevel-inset)"));
2543        assert!(css.contains(".progress > .progress-fill {"));
2544        assert!(css.contains("background: var(--action)"));
2545        // A bare `.fill` would catch things that have nothing to do with
2546        // progress once the sheet lands unprefixed.
2547        assert!(!css.contains("> .fill "));
2548    }
2549
2550    #[test]
2551    fn a_progress_bar_can_carry_a_tone_and_defaults_to_action() {
2552        let css = progress_rules(&Emit::default());
2553        // Untoned is --action, not Tone::Neutral's content-muted: a bar with no
2554        // status is still reporting progress, and muted would read as disabled.
2555        assert!(css.contains(".progress > .progress-fill {\n    background: var(--action);"));
2556        assert!(!css.contains("progress-fill {\n    color: var(--content-muted)"));
2557        for tone in ["info", "success", "warning", "danger"] {
2558            assert!(
2559                css.contains(&format!(".progress > .progress-fill[data-tone=\"{tone}\"]")),
2560                "missing progress tone {tone}"
2561            );
2562        }
2563        // goingson's two live cases, which is why the tones are emitted at all.
2564        assert!(css.contains("[data-tone=\"success\"] {\n    background: var(--success);"));
2565        assert!(css.contains("[data-tone=\"danger\"] {\n    background: var(--danger);"));
2566    }
2567
2568    #[test]
2569    fn no_scrollbar_track_is_emitted() {
2570        // Decision 3's negative half. It was on the phase A list and came off;
2571        // this is what stops it drifting back in.
2572        let css = stylesheet(&Emit::default());
2573        assert!(!css.contains("scrollbar"));
2574        assert!(!css.contains("::-webkit"));
2575    }
2576
2577    #[test]
2578    fn an_invalid_field_is_ringed_without_being_lit() {
2579        let css = surface_rules(&Emit::default());
2580        assert!(css.contains(".field {"));
2581        // The ARIA attribute, not a class: one fact, read by both the visual
2582        // and the accessible state, so they cannot drift.
2583        assert!(css.contains(".field[aria-invalid=\"true\"] {"));
2584        assert!(!css.contains(".field.invalid"));
2585        // A flat ring: this edge says "wrong", and a two-tone bevel would have
2586        // it say "raised" at the same time.
2587        assert!(css.contains("0 0 0 1px var(--danger)"));
2588    }
2589
2590    #[test]
2591    fn an_invalid_field_keeps_the_well_underneath_it() {
2592        // box-shadow is not additive. A lone ring replaces the bevel and drops
2593        // the well out from under the field, which is what this emitted before
2594        // 0.5.0 and is the whole reason the rule composes.
2595        let css = surface_rules(&Emit::default());
2596        let invalid = css
2597            .lines()
2598            .skip_while(|l| !l.starts_with(".field[aria-invalid"))
2599            .take_while(|l| !l.starts_with('}'))
2600            .collect::<Vec<_>>()
2601            .join("\n");
2602        assert!(
2603            invalid.contains("var(--bevel-inset)"),
2604            "the well was dropped: {invalid}"
2605        );
2606        assert!(invalid.contains("var(--danger)"));
2607    }
2608
2609    #[test]
2610    fn button_and_card_come_out_identical_by_construction() {
2611        // The duplication phase A deletes. They are the same composition, so
2612        // the only honest way to emit both is from one call.
2613        let opts = Emit::default();
2614        let css = surface_rules(&opts);
2615        assert_eq!(
2616            depth_declarations(Depth::Raised),
2617            depth_declarations(Depth::Raised)
2618        );
2619        assert!(css.contains(".button {"));
2620        assert!(css.contains(".card {"));
2621        assert_eq!(
2622            interactive_rules("button", Depth::Raised, &Emit::default()).replace("button", "card"),
2623            interactive_rules("card", Depth::Raised, &Emit::default())
2624        );
2625    }
2626
2627    #[test]
2628    fn a_prefix_reaches_the_component_classes_too() {
2629        let opts = Emit {
2630            class_prefix: "mo-",
2631            ..Emit::default()
2632        };
2633        let css = stylesheet(&opts);
2634        for name in [
2635            "mo-button",
2636            "mo-card",
2637            "mo-field",
2638            "mo-badge",
2639            "mo-chip",
2640            "mo-tab",
2641            "mo-row-primary",
2642            "mo-progress",
2643            "mo-progress-fill",
2644        ] {
2645            assert!(css.contains(&format!(".{name}")), "unprefixed: {name}");
2646        }
2647        // The bare names must be gone entirely, or a prefixed build still
2648        // collides with the app's own stylesheet.
2649        assert!(!css.contains(".button {"));
2650        assert!(!css.contains(".card {"));
2651        assert!(!css.contains(".badge {"));
2652    }
2653
2654    #[test]
2655    fn the_class_a_renderer_puts_on_an_option_is_the_one_the_rules_key_off() {
2656        // `option_class` is the contract a screen renderer writes markup
2657        // against, and the rules below are the other half of it. They come off
2658        // one mapping now, so this asserts the mapping is the one that reaches
2659        // the stylesheet rather than that two lists still agree.
2660        let css = stylesheet(&Emit::default());
2661        for selector in [Selector::Tabs, Selector::Segmented, Selector::Toggle] {
2662            let name = option_class(selector);
2663            assert!(css.contains(&format!(".{name} {{")), "{name}: {css}");
2664            assert!(css.contains(&format!(".{name}.chosen {{")), "{name}: {css}");
2665        }
2666        assert_eq!(option_class(Selector::Tabs), "tab");
2667    }
2668
2669    #[test]
2670    fn the_caret_brings_its_own_gap_and_is_the_glyph_the_description_names() {
2671        let css = stylesheet(&Emit::default());
2672
2673        // The space is inside the glyph, which is what the other two renderers
2674        // write. Emitted bare, every consumer has to add it back, and the
2675        // obvious way to add it -- `content` in an app stylesheet, which is
2676        // unlayered and so outranks this sheet -- deletes the caret instead.
2677        assert!(css.contains("content: \" \\25B2\";"), "{css}");
2678        assert!(css.contains("content: \" \\25BC\";"), "{css}");
2679        assert!(!css.contains("content: \"\\2"), "{css}");
2680
2681        // The arrows this renderer used to draw alone are gone. Composition
2682        // rather than agreement: the glyph comes from `Sort::glyph`, so a
2683        // fourth spelling cannot appear here without appearing everywhere.
2684        assert!(!css.contains("2191") && !css.contains("2193"), "{css}");
2685        assert!(css.contains(&css_escape(Sort::Ascending.glyph())), "{css}");
2686
2687        // Three states, three tones. An idle sortable heading draws its caret
2688        // now rather than reserving a hidden box for it, so there is no
2689        // visibility to order and no reflow left to guard against; what
2690        // separates the states is the colour, and the sorted arms come after
2691        // the idle one because the specificity is the same.
2692        let idle = css
2693            .find(".table-heading[data-sortable]::after")
2694            .expect("the idle caret is emitted");
2695        let sorted = css
2696            .find(".table-heading[aria-sort=\"ascending\"]::after")
2697            .expect("the ascending caret is emitted");
2698        assert!(idle < sorted, "{css}");
2699        assert!(
2700            css[idle..sorted].contains("color: var(--content-secondary);"),
2701            "{css}"
2702        );
2703        assert!(css[sorted..].contains("color: var(--content);"), "{css}");
2704        assert!(!css.contains("visibility: hidden;"), "{css}");
2705    }
2706
2707    #[test]
2708    fn a_destructive_button_has_somewhere_for_its_tone_to_land() {
2709        let css = component_rules(&Emit::default());
2710
2711        for tone in [Tone::Info, Tone::Success, Tone::Warning, Tone::Danger] {
2712            assert!(
2713                css.contains(&format!(".button[data-tone=\"{}\"]", tone.token())),
2714                "{css}"
2715            );
2716        }
2717        // Colour, not a fill. A red surface is an app's decision about emphasis.
2718        assert!(!css.contains(".button[data-tone=\"danger\"] {\n    background"));
2719    }
2720
2721    #[test]
2722    fn the_table_model_is_scoped_to_the_table_but_the_caret_is_not() {
2723        // goingson's task headings are `.table-heading` inside a CSS grid.
2724        // They want the sort caret and the sortable cursor; they do not want
2725        // `display: table-cell`, which a grid item blockifies away anyway.
2726        // Scoping the one and not the other is what separates them.
2727        let css = component_rules(&Emit::default());
2728
2729        assert!(
2730            css.contains(".table .table-heading,\n.table .cell {\n    display: table-cell;"),
2731            "{css}"
2732        );
2733        assert!(
2734            !css.contains(".table-heading,\n.cell {\n    display: table-cell;"),
2735            "the table model is still unscoped: {css}"
2736        );
2737
2738        let states = state_rules(&Emit::default());
2739        assert!(
2740            states.contains(".table-heading[aria-sort"),
2741            "the caret got scoped along with the model: {states}"
2742        );
2743    }
2744
2745    #[test]
2746    fn a_figure_value_is_the_thing_itself_and_a_badge_is_quiet() {
2747        // Both used to read `Tone::Neutral.token()`, which answered
2748        // `content-muted`, so the headline number sat at the colour of its own
2749        // caption. Neutral answers `content` now; each site states its own
2750        // claim rather than borrowing one from the status axis.
2751        let css = component_rules(&Emit::default());
2752
2753        assert!(
2754            css.contains(".figure > .figure-value {\n    color: var(--content);"),
2755            "{css}"
2756        );
2757        assert!(
2758            css.contains(".figure > .figure-caption {\n    color: var(--content-muted);"),
2759            "{css}"
2760        );
2761        assert!(
2762            css.contains(".badge {\n    color: var(--content-muted);"),
2763            "{css}"
2764        );
2765    }
2766
2767    #[test]
2768    fn a_described_list_is_not_a_bulleted_list() {
2769        let css = component_rules(&Emit::default());
2770        assert!(css.contains(".list {\n    list-style: none;"), "{css}");
2771    }
2772
2773    #[test]
2774    fn a_table_lays_itself_out_without_being_told_its_columns() {
2775        // The whole point of the CSS table. A described table's columns are
2776        // known at render time, so anything the stylesheet has to be told about
2777        // them would have to travel with the markup.
2778        let css = component_rules(&Emit::default());
2779
2780        assert!(css.contains(".table {\n    display: table;"), "{css}");
2781        assert!(css.contains("display: table-row;"), "{css}");
2782        assert!(css.contains("display: table-cell;"), "{css}");
2783        assert!(!css.contains("grid-template-columns"), "{css}");
2784    }
2785
2786    #[test]
2787    fn a_control_in_a_cell_is_not_painted_as_text() {
2788        // The point of makeover-layout 0.14.0's CellPart, and the table-side
2789        // twin of `the_three_text_parts_take_their_intents_and_actions_inherits`
2790        // above. One `.cell` and one content colour meant a button in a cell
2791        // inherited it.
2792        let css = table_rules(&Emit::default());
2793
2794        assert!(
2795            css.contains(".cell-value {\n    color: var(--content);"),
2796            "{css}"
2797        );
2798        assert!(!css.contains(".cell-actions {\n    color:"), "{css}");
2799        assert!(!css.contains(".cell-tokens {\n    color:"), "{css}");
2800        assert!(!css.contains(".cell-link {\n    color:"), "{css}");
2801
2802        // The colour is on the part that is text, never on the container. On
2803        // `.cell` it would cascade into the three parts that are not text,
2804        // which is the bug written as one rule.
2805        assert!(!css.contains(".cell {\n    color:"), "{css}");
2806    }
2807
2808    #[test]
2809    fn an_unknown_cell_part_renders_plainly_rather_than_failing_to_build() {
2810        // `part_class`'s obligation, taken on for the table side too. CellPart
2811        // is `#[non_exhaustive]`, so a member added upstream must land as a
2812        // bare class rather than as a build that stops.
2813        assert_eq!(cell_part_class(CellPart::Value), "cell-value");
2814        assert_eq!(cell_part_class(CellPart::Actions), "cell-actions");
2815    }
2816
2817    #[test]
2818    fn a_column_drops_by_its_priority_and_never_by_its_position() {
2819        let css = component_rules(&Emit::default());
2820
2821        // Optional goes at the narrowest class and secondary goes with it,
2822        // which is `kept_at`'s cutoff walk said as two queries.
2823        let compact = css
2824            .find(&format!("@media {}", SizeClass::Compact.media_condition()))
2825            .expect("a compact query");
2826        let medium = css
2827            .find(&format!("@media {}", SizeClass::Medium.media_condition()))
2828            .expect("a medium query");
2829        assert!(css[compact..].contains(".cell-drops-next"), "{css}");
2830        assert!(!css[medium..].contains(".cell-drops-next"), "{css}");
2831
2832        // Essential columns are never mentioned, because not being mentioned is
2833        // already what never dropping means.
2834        assert!(!css.contains(".cell-keeps"), "{css}");
2835
2836        // And nothing counts. `nth-child` is the bug the priority vocabulary
2837        // exists to end.
2838        assert!(!css.contains("nth-child"), "{css}");
2839    }
2840
2841    #[test]
2842    fn a_track_places_by_custom_property_and_never_by_a_size() {
2843        let css = track_rules(&Emit::default());
2844
2845        // Placement arrives from the caller, computed once by Track::fraction.
2846        // If either of these becomes a literal, three renderers have started
2847        // disagreeing about where 09:30 is.
2848        assert!(css.contains("top: var(--track-at"), "{css}");
2849        assert!(css.contains("height: var(--track-for"), "{css}");
2850
2851        // Overlap lanes default so an entry naming neither is full width.
2852        assert!(css.contains("--track-lane, 0"), "{css}");
2853        assert!(css.contains("--track-lanes, 1"), "{css}");
2854
2855        // The refusal that matters. A slot height here would be this crate
2856        // deciding how tall a quarter of an hour is, which is the thing
2857        // makeover-geometry owns and the reason `.track` gets no height at all.
2858        assert!(!css.contains("height: var(--track-slot"), "{css}");
2859        for size in ["px", "rem", "em", "vh"] {
2860            let bare = css
2861                .lines()
2862                .filter(|l| !l.contains("var(--"))
2863                .any(|l| l.contains(size));
2864            assert!(!bare, "track_rules named a {size} outside a var(): {css}");
2865        }
2866    }
2867
2868    #[test]
2869    fn a_relaxed_part_clamps_and_a_tight_one_says_nothing() {
2870        let css = stylesheet(&Emit::default());
2871        assert!(css.contains(".row-relaxed {"));
2872        assert!(css.contains("-webkit-line-clamp: 2;"));
2873        // The count is `Flow`'s, not this crate's. If the tier ever means three
2874        // lines, this fails here rather than in an app.
2875        assert!(css.contains(&format!("line-clamp: {};", Flow::Relaxed.lines())));
2876        // Tight gets no rule at all: one line is what a run already does, and a
2877        // class per part saying so is a declaration that changes nothing.
2878        assert!(!css.contains("row-tight"));
2879        assert_eq!(crate::list::flow_class(Flow::Relaxed), Some("row-relaxed"));
2880        assert_eq!(crate::list::flow_class(Flow::Tight), None);
2881        // Emitted, therefore checkable: an app's dead-vocabulary seal and the
2882        // overlap check both read `vocabulary::names`, so a class the renderer
2883        // can write and that list does not carry is invisible to both.
2884        assert!(crate::vocabulary::names(&Emit::default()).contains("row-relaxed"));
2885    }
2886
2887    #[test]
2888    fn the_two_kinds_of_wait_stop_rendering_identically() {
2889        // The state `d43ea1c5` fixes: the emitter had been writing
2890        // `data-awaiting` for months and nothing styled either value, so a
2891        // measured wait and an unmeasured one drew the same nothing.
2892        let css = stylesheet(&Emit::default());
2893        assert!(css.contains("[data-awaiting]::after"), "{css}");
2894        assert!(
2895            css.contains("[data-awaiting][aria-busy=\"true\"]::after"),
2896            "the mark is drawn only while something is actually waiting"
2897        );
2898        assert!(
2899            css.contains("[data-awaiting=\"determinate\"][aria-busy=\"true\"]::after"),
2900            "the measured half is its own drawing"
2901        );
2902    }
2903
2904    #[test]
2905    fn the_blink_takes_its_cadence_and_never_names_one() {
2906        // Three renderers draw this mark. A number written here would be a
2907        // second heartbeat for one wait.
2908        let css = stylesheet(&Emit::default());
2909        assert!(
2910            css.contains("calc(var(--cadence-activity) * 2)"),
2911            "a half-period doubled, not a literal"
2912        );
2913        assert!(!css.contains("500ms"), "{css}");
2914    }
2915
2916    #[test]
2917    fn a_bar_nobody_is_counting_is_empty_rather_than_full() {
2918        // Rule 1. The share defaults to zero, so a determinate control with no
2919        // binder watching bytes draws a trough. A default of 1 would be the
2920        // confidently-wrong drawing the rule exists to forbid.
2921        let css = stylesheet(&Emit::default());
2922        assert!(css.contains("var(--awaiting-share, 0)"), "{css}");
2923    }
2924
2925    #[test]
2926    fn motion_off_leaves_the_mark_lit_because_the_keyframes_do_the_dimming() {
2927        // `makeover_timing::reduced_motion_css` sets `--cadence-activity: 0ms`,
2928        // and a zero-length animation leaves the element in its base style
2929        // rather than at its last keyframe. So the base has to be the lit one.
2930        // The inverted spelling would blank the mark for a reader who asked for
2931        // less motion, which answers a request nobody made.
2932        let css = stylesheet(&Emit::default());
2933        let busy = css
2934            .split("[data-awaiting][aria-busy=\"true\"]::after {")
2935            .nth(1)
2936            .expect("the busy rule");
2937        let busy = busy.split('}').next().expect("its body");
2938        assert!(
2939            busy.contains("background: var(--action);"),
2940            "the base state is lit: {busy}"
2941        );
2942    }
2943
2944    #[test]
2945    fn the_whole_sheet_still_names_every_colour() {
2946        // The crate's founding property, asserted over the component layer and
2947        // not only the primitives.
2948        let css = stylesheet(&Emit::default());
2949        assert!(!css.contains('#'));
2950        assert!(!css.contains("rgb"));
2951        for line in css.lines() {
2952            // Declarations only: a selector or an at-rule can carry a colon of
2953            // its own (`:root`, `:hover`, `@media (hover: hover)`) and declares
2954            // nothing. Keyed on the trailing semicolon rather than on leading
2955            // indentation, which only ever worked as a proxy for nesting depth
2956            // and stopped when the sheet gained a cascade layer around it.
2957            let trimmed = line.trim();
2958            if !trimmed.ends_with(';') {
2959                continue;
2960            }
2961            let Some((_, value)) = trimmed.split_once(": ") else {
2962                continue;
2963            };
2964            if value.contains("var(--") {
2965                continue;
2966            }
2967            // Everything left has to be a keyword, a number or a
2968            // caller-supplied length, never a colour.
2969            //
2970            // The length arm is what the comment above always claimed and the
2971            // list never covered: `border_width` arrives from `Emit` and lands
2972            // bare in the focus ring's offset, where the bevel had only ever
2973            // used it inside an `inset` shadow.
2974            let opts = Emit::default();
2975            assert!(
2976                value.contains("inset")
2977                    || value.contains(opts.border_width)
2978                    || value.contains(opts.focus_width)
2979                    // 0.12.0's two: a sortable header is a control and says so
2980                    // with the pointer, and the caret is this renderer's own
2981                    // expression of `aria-sort`. Neither is a colour, which is
2982                    // what this test is actually about, and neither is a size,
2983                    // which is the other thing this crate must not name. The
2984                    // leading space inside the glyph is the same thing the other
2985                    // two renderers write into theirs, so it is part of the
2986                    // caret rather than spacing this crate decided on.
2987                    || value
2988                        .trim_start_matches('"')
2989                        .trim_start()
2990                        .starts_with("\\2")
2991                    || matches!(
2992                        value.trim_end_matches(';'),
2993                        "0" | "1"
2994                            // The wait's mark and bar, 0.60.0. An empty
2995                            // `content` is what brings a pseudo-element into
2996                            // existence with nothing in it, and `step-end`
2997                            // is an easing: the blink is two states, not a
2998                            // slide between them. Neither is a colour, and
2999                            // neither is a magnitude.
3000                            | "\"\""
3001                            // Where the mark sits on the line it joins.
3002                            // Alignment is structure, the same way `display`
3003                            // is, and `baseline` is the initial value said out
3004                            // loud so a host stylesheet cannot leave it
3005                            // wherever an earlier rule put it.
3006                            | "baseline"
3007                            | "inline-block"
3008                            | "none"
3009                            | "auto"
3010                            | "not-allowed"
3011                            | "pointer"
3012                            // The caret's reserved box. Visibility is presence,
3013                            // not magnitude and not colour.
3014                            | "hidden"
3015                            | "visible"
3016                            // The link's two signals. `underline` is a line and
3017                            // `inherit` defers to whatever the app set, so
3018                            // neither names a colour or a magnitude.
3019                            | "underline"
3020                            | "inherit"
3021                            // The table frame. `display` is structure and not a
3022                            // size; `nowrap` is what makes a content column
3023                            // content. The two widths are the awkward pair and
3024                            // they are still not sizes: `100%` is "all of
3025                            // whatever you were given" and `1%` is the CSS
3026                            // table idiom for "shrink to fit", which is a
3027                            // behaviour spelled as a number because CSS has no
3028                            // keyword for it. Neither names a magnitude, which
3029                            // is the thing this crate leaves to
3030                            // makeover-geometry.
3031                            | "table"
3032                            | "table-row"
3033                            | "table-cell"
3034                            | "nowrap"
3035                            | "100%"
3036                            | "1%"
3037                            // The time axis, 0.42.0. `position` is the one
3038                            // property whose whole job is where a thing sits,
3039                            // which is exactly what this crate spent its life
3040                            // refusing to say -- so it is worth being exact
3041                            // about why these two are not that refusal
3042                            // breaking.
3043                            //
3044                            // Neither names a magnitude. `relative` says the
3045                            // track is what its entries resolve against, and
3046                            // `absolute` says an entry is placed rather than
3047                            // flowed. *Where* each entry lands is
3048                            // `--track-at` and `--track-for`, custom
3049                            // properties the caller sets from
3050                            // `Track::fraction`, and they are skipped by the
3051                            // `var(--` arm above like every other value this
3052                            // crate refuses to decide.
3053                            //
3054                            // The rule that would break the refusal is a slot
3055                            // height, and there is none: the track's height is
3056                            // the app's, so the percentages have something to
3057                            // resolve against and this crate still never says
3058                            // how tall a day is.
3059                            | "relative"
3060                            | "absolute"
3061                            // A run's five, 0.49.0. `flex`, `wrap` and
3062                            // `center` are structure and alignment, the same
3063                            // reading `table` gets: which way members are laid
3064                            // out and how they line up, never how much of
3065                            // anything.
3066                            //
3067                            // The two intrinsic keywords are the interesting
3068                            // pair and they are the opposite of a size. A
3069                            // magnitude is a number somebody chose;
3070                            // `min-content` and `max-content` are the browser
3071                            // being asked what the members themselves come to,
3072                            // which is the derived minimum the room ruling
3073                            // requires and the reason no breakpoint appears
3074                            // anywhere in these rules. `1 1 max-content` is
3075                            // grow, shrink and that basis, so its two digits
3076                            // are ratios rather than lengths.
3077                            | "flex"
3078                            | "wrap"
3079                            | "center"
3080                            | "min-content"
3081                            | "1 1 max-content"
3082                            // A picture's three, 0.36.0. `block` is structure
3083                            // for the reason `table` is: an inline image sits
3084                            // on the baseline and carries a descender's worth
3085                            // of space under it, which is a fact about
3086                            // replaced elements rather than a size this crate
3087                            // chose. `cover` and `contain` are `Fit`'s two
3088                            // named members reaching CSS unchanged, which is
3089                            // an intent arriving rather than a value being
3090                            // picked.
3091                            | "block"
3092                            | "cover"
3093                            | "contain"
3094                            // A relaxed part's three, and the third is the
3095                            // awkward one. `-webkit-box` and `vertical` are
3096                            // structure: they say the part is a box of lines
3097                            // stacked downward, which is the only way CSS lets
3098                            // anyone ask for a clamp at all.
3099                            //
3100                            // `2` is a count of lines, not a length. The
3101                            // distinction this crate holds is between naming a
3102                            // magnitude -- a padding, a height, a font size,
3103                            // all of which belong to makeover-geometry -- and
3104                            // naming how many of something there are. A line's
3105                            // height is still the app's, so two lines is
3106                            // whatever two of the app's lines come to, and
3107                            // nothing here decides how tall that is. It is also
3108                            // not a value picked here: it is `Flow::Relaxed`'s
3109                            // own answer arriving unchanged, the same way
3110                            // `cover` and `contain` are `Fit`'s.
3111                            | "-webkit-box"
3112                            | "vertical"
3113                            | "2"
3114                    ),
3115                "unrecognised literal value: {line}"
3116            );
3117        }
3118    }
3119
3120    #[test]
3121    fn flat_emits_nothing_at_all() {
3122        assert_eq!(depth_class(Depth::Flat, &Emit::default()), None);
3123        assert!(!depth_rules(&Emit::default()).contains("flat"));
3124    }
3125
3126    #[test]
3127    fn a_prefix_namespaces_every_class() {
3128        let opts = Emit {
3129            class_prefix: "mo-",
3130            ..Emit::default()
3131        };
3132        let css = depth_rules(&opts);
3133        assert!(css.contains(".mo-raised {"));
3134        assert!(css.contains(".mo-well {"));
3135        assert!(!css.contains(".raised {"));
3136    }
3137
3138    #[test]
3139    fn the_border_width_is_the_callers() {
3140        let opts = Emit {
3141            border_width: "2px",
3142            ..Emit::default()
3143        };
3144        assert!(bevel_shadow(Bevel::Raised, &opts).contains("inset 2px 2px 0"));
3145    }
3146
3147    #[test]
3148    fn edges_agree_with_the_description() {
3149        // Not a tautology: it is the guard that a CSS-shaped convenience never
3150        // quietly reverses which side is lit.
3151        let (tl, br) = Bevel::Raised.edges();
3152        assert_eq!(tl.token(), Edge::Light.token());
3153        assert_eq!(br.token(), Edge::Dark.token());
3154    }
3155}