feat(web): SvelteKit PWA with the Nordic design system

The interface for phase 1: sign-in and first-run setup, Today, Upcoming,
project views, sharing, Cmd+K search and the keyboard shortcuts.

Quick add highlights what the parser understood underneath the text as
you type — a mirrored layer behind a transparent input, sliced by byte
offset because the server counts bytes and JavaScript counts UTF-16
units, which disagree on exactly the Danish letters this app is full of.

Every local action is optimistic and reconciled afterwards: ticking a
task off strikes it through and animates it out before the request
leaves, and rolls back with a message if the server disagrees.

The Content-Security-Policy is computed at startup by hashing the inline
scripts in the built index.html. Both of them change on every frontend
build, so a hard-coded hash would go stale silently and show up only as
a blank page in production. The alternative, 'unsafe-inline', would drop
the one protection that makes an injected script in a task title inert.

Today is derived from the store by date rather than from the ids that
came back with the first request — filtering against a captured list
meant a task added through quick add never appeared until a reload.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
This commit is contained in:
Kristian Wind
2026-08-17 00:03:20 +02:00
co-authored by Claude Opus 5
parent dff86dd23d
commit 12a9597919
25 changed files with 4776 additions and 15 deletions
+265
View File
@@ -0,0 +1,265 @@
<script>
/**
* The quick-add box.
*
* What makes it worth its own component is the highlighting: as you type, the
* parts the parser understood are tinted underneath the text. That is the only
* way "i morgen kl 10 p1" stops feeling like guesswork — you can see it being
* read before you commit to it.
*
* The tinting is a mirrored layer sitting exactly behind a transparent input.
* A contenteditable would let us style the text directly and would also break
* autocorrect, IME input and undo, which is a bad trade for a box people type
* one line into.
*/
import { api } from '$lib/api.js';
import { app } from '$lib/stores.svelte.js';
let { projectId = undefined, onadded = undefined, autofocus = false } = $props();
let text = $state('');
let spans = $state([]);
let submitting = $state(false);
let input;
let previewController = null;
let previewTimer = null;
/**
* Parsing runs on the server, so it is debounced and each request cancels the
* one before it. Without the cancel, a fast typist has six parses in flight and
* the highlighting flickers between whichever happens to land last.
*/
function schedulePreview(value) {
clearTimeout(previewTimer);
previewController?.abort();
if (!value.trim()) {
spans = [];
return;
}
previewTimer = setTimeout(async () => {
previewController = new AbortController();
try {
const result = await api.quickAddPreview(value, previewController.signal);
spans = result.spans ?? [];
} catch {
// An aborted or failed preview just means no highlighting; the text
// is still perfectly submittable.
}
}, 120);
}
$effect(() => {
schedulePreview(text);
});
$effect(() => {
if (autofocus && input) input.focus();
});
/** Splits the raw text into highlighted and plain runs, by byte offset. */
let segments = $derived.by(() => {
if (!spans.length) return [{ text, kind: null }];
// The server counts bytes; JavaScript counts UTF-16 units. For "køb" those
// disagree, and using one for the other paints the highlight over the wrong
// characters — so the string is converted and sliced as bytes.
const encoder = new TextEncoder();
const decoder = new TextDecoder();
const bytes = encoder.encode(text);
const out = [];
let cursor = 0;
for (const span of [...spans].sort((a, b) => a.start - b.start)) {
if (span.start > cursor) {
out.push({ text: decoder.decode(bytes.slice(cursor, span.start)), kind: null });
}
out.push({ text: decoder.decode(bytes.slice(span.start, span.end)), kind: span.kind });
cursor = span.end;
}
if (cursor < bytes.length) {
out.push({ text: decoder.decode(bytes.slice(cursor)), kind: null });
}
return out;
});
async function submit(event) {
event?.preventDefault();
const value = text.trim();
if (!value || submitting) return;
submitting = true;
// Cleared before the response: the box has to be ready for the next thought
// immediately, and a failure puts the text back rather than blocking on it.
text = '';
spans = [];
const task = await app.quickAdd(value, projectId);
submitting = false;
if (task) onadded?.(task);
else text = value;
input?.focus();
}
function onkeydown(event) {
if (event.key === 'Escape') {
text = '';
spans = [];
input?.blur();
return;
}
// Submitting on Enter explicitly rather than relying on the browser's
// implicit form submission: that only fires when the form has a submit
// button, and this one is hidden while the field is empty. Tying the
// primary interaction of the app to a conditional element is not a
// dependency worth having.
if (event.key === 'Enter' && !event.isComposing) {
event.preventDefault();
submit();
}
}
</script>
<form class="quickadd" onsubmit={submit}>
<span class="plus" aria-hidden="true">+</span>
<div class="field">
<!-- The mirror sits behind the input and must match it exactly: same font,
same padding, same wrapping. Any difference shows up as highlighting
that drifts out of alignment as the line grows. -->
<div class="mirror" aria-hidden="true">
{#each segments as segment}
{#if segment.kind}
<mark data-kind={segment.kind}>{segment.text}</mark>
{:else}
<span>{segment.text}</span>
{/if}
{/each}
</div>
<input
bind:this={input}
bind:value={text}
{onkeydown}
type="text"
placeholder="Tilføj en opgave — prøv “betal moms i morgen kl 10 p1 #Firma”"
aria-label="Ny opgave"
autocomplete="off"
spellcheck="false"
/>
</div>
{#if text.trim()}
<button type="submit" class="submit" disabled={submitting}>Tilføj</button>
{/if}
</form>
<style>
.quickadd {
display: flex;
align-items: center;
gap: var(--s3);
padding: var(--s3) var(--s2);
border-bottom: 1px solid var(--line);
}
.plus {
flex: none;
width: 20px;
height: 20px;
display: grid;
place-items: center;
color: var(--ink-faint);
font-size: var(--text-lg);
line-height: 1;
}
.quickadd:focus-within .plus {
color: var(--accent);
}
.field {
position: relative;
flex: 1;
min-width: 0;
}
/* Both layers share every metric that affects text layout. */
.mirror,
.field input {
font: inherit;
font-size: var(--text-base);
line-height: 1.5;
letter-spacing: normal;
padding: 0;
border: 0;
white-space: pre-wrap;
overflow-wrap: anywhere;
}
.mirror {
position: absolute;
inset: 0;
pointer-events: none;
color: transparent;
}
/* The tint is behind the glyphs, not on them: colouring the text itself would
fight the reading of the sentence, while a wash behind it reads as "this bit
was understood" without changing the words. */
.mirror mark {
background: var(--accent-sunken);
color: transparent;
border-radius: var(--radius-sm);
box-shadow: 0 0 0 1px color-mix(in srgb, var(--accent) 25%, transparent);
}
.mirror mark[data-kind='priority'] {
background: color-mix(in srgb, var(--p1) 18%, transparent);
box-shadow: 0 0 0 1px color-mix(in srgb, var(--p1) 30%, transparent);
}
.mirror mark[data-kind='project'] {
background: color-mix(in srgb, var(--p3) 18%, transparent);
box-shadow: 0 0 0 1px color-mix(in srgb, var(--p3) 30%, transparent);
}
.mirror mark[data-kind='label'] {
background: color-mix(in srgb, var(--p2) 18%, transparent);
box-shadow: 0 0 0 1px color-mix(in srgb, var(--p2) 30%, transparent);
}
.field input {
position: relative;
width: 100%;
background: transparent;
color: var(--ink);
outline: none;
}
.field input::placeholder {
color: var(--ink-faint);
}
.submit {
flex: none;
background: var(--accent);
color: var(--accent-ink);
font-size: var(--text-sm);
font-weight: 550;
padding: var(--s1) var(--s3);
border-radius: var(--radius);
transition: background var(--fast) var(--ease);
}
.submit:hover:not(:disabled) {
background: var(--accent-hover);
}
.submit:disabled {
opacity: 0.5;
cursor: default;
}
</style>