Add autobuyer page capture, browser extension, and desktop clicker

Builds the pipeline the autobuyer needs: see the page, find an element,
click it.

extension/ — MV3 Chromium extension. Polls /api/autobuyer/status and,
while on, scrapes the target tab's HTML and posts it back. Also serves
locate requests: focuses the window, scrolls the element into view, and
reports its position. host_permissions is scoped to tradeify plus
localhost so it cannot read other sites — an empty target pattern would
otherwise capture whatever tab happened to be active, including banking
or mail.

app/api/autobuyer/ — status toggle, capture store, and the locate request
queue. CORS is open because the extension's origin changes every time an
unpacked extension is reloaded.

app/autobuyer/page.tsx — ON switch, source view (default) and a rendered
view. The render uses sandbox="allow-scripts" without allow-same-origin:
the page's own JS is needed because sites ship content at opacity:0 and
fade it in, but the frame must not reach the dashboard's same-origin API
routes, which serve firm credentials.

clicker/ — Python CLI. Asks the extension where a selector is, adds the
element rect to the window's screen position and the browser chrome
height to get desktop coordinates, then clicks with a human motion model
(curved path, eased velocity, occasional overshoot, dwell before press).
Raises the browser application first, since macOS consumes a click on an
unfocused window rather than delivering it.

Refuses to click when the element is covered by an overlay, when the
coordinates fall off-screen, or when the browser cannot be confirmed
frontmost.

Verified: API round-trips, capture pruning, locate claim-once semantics,
motion geometry and timing, and focus activation — the last two against
stubs, since pyautogui and pyobjc are not installed here. NOT verified
end to end: Chrome is still running a stale build of the extension, so a
locate request has never completed against a real page and no real click
has been sent.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
This commit is contained in:
Brandon Li
2026-08-27 16:15:18 -05:00
co-authored by Claude Opus 5
parent 24a54c6642
commit 54221bbc0c
19 changed files with 1752 additions and 2 deletions
+271 -2
View File
@@ -1,11 +1,280 @@
'use client';
import { useCallback, useEffect, useRef, useState } from 'react';
interface ElementPosition {
index: number;
tag: string;
id: string | null;
text: string;
visible: boolean;
inViewport: boolean;
viewport: { x: number; y: number; width: number; height: number };
page: { x: number; y: number };
screen: { x: number; y: number };
}
interface ViewportMetrics {
scrollX: number;
scrollY: number;
innerWidth: number;
innerHeight: number;
screenX: number;
screenY: number;
chromeHeight: number;
devicePixelRatio: number;
}
interface Capture {
url: string;
title: string;
html: string;
viewport: Partial<ViewportMetrics>;
elements: ElementPosition[];
capturedAt: number;
}
const POLL_MS = 2000;
/** Point relative URLs (stylesheets, images, fonts) at the origin the snapshot came
* from, otherwise the render comes out unstyled. The HTML parser hoists a leading
* <base> into <head>, so prepending works even if the markup has no explicit head. */
function withBaseTag(html: string, url: string): string {
const base = `<base href="${url.replace(/"/g, '&quot;')}">`;
const head = html.match(/<head\b[^>]*>/i);
if (!head || head.index === undefined) return base + html;
const at = head.index + head[0].length;
return html.slice(0, at) + base + html.slice(at);
}
export default function AutoBuyer() {
const [enabled, setEnabled] = useState(false);
const [capture, setCapture] = useState<Capture | null>(null);
const [busy, setBusy] = useState(false);
const [error, setError] = useState<string | null>(null);
const [view, setView] = useState<'rendered' | 'source'>('source');
// The extension posts a new capture every few seconds, and re-feeding srcDoc would
// reload the frame that often. So the rendered view shows a snapshot pinned when you
// opened it — clicking Rendered again re-pins to the latest capture.
const [pinned, setPinned] = useState<Capture | null>(null);
// Tracks the newest capture we already hold, so the poll can skip re-downloading it.
const lastAtRef = useRef(0);
const poll = useCallback(async () => {
try {
const s = await fetch('/api/autobuyer/status', { cache: 'no-store' }).then((r) => r.json());
setEnabled(!!s.enabled);
if (!s.enabled) return;
const url = lastAtRef.current
? `/api/autobuyer/capture?since=${lastAtRef.current}`
: '/api/autobuyer/capture';
const data = await fetch(url, { cache: 'no-store' }).then((r) => r.json());
if (data.unchanged || !data.capture) return;
lastAtRef.current = data.capture.capturedAt;
setCapture(data.capture);
setError(null);
} catch (err: any) {
setError(err?.message ?? 'Poll failed');
}
}, []);
useEffect(() => {
poll();
const id = setInterval(poll, POLL_MS);
return () => clearInterval(id);
}, [poll]);
async function toggle() {
setBusy(true);
try {
const res = await fetch('/api/autobuyer/status', {
method: 'PATCH',
headers: { 'Content-Type': 'application/json' },
body: JSON.stringify({ enabled: !enabled }),
});
const data = await res.json();
setEnabled(!!data.enabled);
setError(null);
} catch (err: any) {
setError(err?.message ?? 'Failed to toggle');
} finally {
setBusy(false);
}
}
function showRendered() {
setView('rendered');
if (capture) setPinned(capture);
}
async function clearCapture() {
await fetch('/api/autobuyer/capture', { method: 'DELETE' });
lastAtRef.current = 0;
setCapture(null);
setPinned(null);
}
const stale = capture ? Date.now() - capture.capturedAt > 15000 : false;
return (
<div className="min-h-screen bg-slate-50 p-8">
<div className="max-w-7xl mx-auto">
<h1 className="text-2xl font-bold text-slate-900 mb-6">AutoBuyer</h1>
<div className="bg-white border border-slate-200 rounded-xl shadow-sm p-8 text-center text-slate-400">
Coming soon
{/* ── Capture switch ── */}
<div className="bg-white border border-slate-200 rounded-xl shadow-sm mb-6">
<div className="flex items-center justify-between px-4 py-3">
<div>
<p className="text-sm font-medium text-slate-800">Page capture</p>
<p className="text-xs text-slate-400 mt-0.5">
While on, the AutoFirmer Capture extension posts the target tab&apos;s HTML here every few seconds
</p>
</div>
<button
onClick={toggle}
disabled={busy}
className={`text-sm px-4 py-1.5 rounded-lg font-medium transition-colors disabled:opacity-50 ${
enabled
? 'bg-green-500 hover:bg-green-600 text-white'
: 'bg-slate-200 hover:bg-slate-300 text-slate-700'
}`}
>
{busy ? '…' : enabled ? 'ON' : 'OFF'}
</button>
</div>
</div>
{error && (
<div className="mb-6 rounded-lg border border-red-200 bg-red-50 px-4 py-2 text-sm text-red-700">
{error}
</div>
)}
{/* ── Latest capture ── */}
<h2 className="text-sm font-semibold uppercase tracking-wider text-slate-500 mb-3">
Latest capture
</h2>
{!enabled ? (
<div className="bg-white border border-slate-200 rounded-xl shadow-sm p-8 text-center text-slate-400">
Capture is off
</div>
) : !capture ? (
<div className="bg-white border border-slate-200 rounded-xl shadow-sm p-8 text-center text-slate-400">
Waiting for the extension open a page in Chrome with the extension installed
</div>
) : (
<div className="bg-white border border-slate-200 rounded-xl shadow-sm overflow-hidden">
<div className="flex flex-wrap items-baseline gap-x-4 gap-y-1 px-4 py-3 border-b border-slate-100">
<span className="text-sm font-medium text-slate-800">{capture.title || '(untitled)'}</span>
<span className="text-xs font-mono text-slate-400 break-all">{capture.url}</span>
<span className="ml-auto flex items-center gap-3 text-xs text-slate-400">
<span>{(capture.html.length / 1024).toFixed(1)} KB</span>
<span className={stale ? 'text-amber-600' : ''}>
{new Date(capture.capturedAt).toLocaleTimeString()}
{stale && ' (stale)'}
</span>
<span className="flex rounded-md border border-slate-200 overflow-hidden">
{(['rendered', 'source'] as const).map((v) => (
<button
key={v}
onClick={() => (v === 'rendered' ? showRendered() : setView('source'))}
className={`px-2 py-0.5 capitalize transition-colors ${
view === v
? 'bg-slate-700 text-white'
: 'text-slate-500 hover:bg-slate-50'
}`}
>
{v}
</button>
))}
</span>
<button
onClick={clearCapture}
className="text-slate-400 hover:text-red-500 transition-colors"
>
Clear
</button>
</span>
</div>
{capture.viewport?.innerWidth != null && (
<div className="px-4 py-2 border-b border-slate-100 flex flex-wrap gap-x-5 gap-y-1 text-xs font-mono text-slate-500">
<span>viewport {capture.viewport.innerWidth}×{capture.viewport.innerHeight}</span>
<span>scroll {Math.round(capture.viewport.scrollX ?? 0)},{Math.round(capture.viewport.scrollY ?? 0)}</span>
<span>window@screen {capture.viewport.screenX},{capture.viewport.screenY}</span>
<span>chrome {capture.viewport.chromeHeight}px</span>
<span>dpr {capture.viewport.devicePixelRatio}</span>
</div>
)}
{capture.elements?.length > 0 && (
<div className="px-4 py-3 border-b border-slate-100">
<p className="text-xs uppercase tracking-wider text-slate-400 mb-2">
Matched elements ({capture.elements.length})
</p>
<div className="overflow-x-auto">
<table className="text-xs font-mono text-slate-600">
<thead className="text-slate-400">
<tr>
<th className="text-left pr-4 font-normal">tag</th>
<th className="text-left pr-4 font-normal">text</th>
<th className="text-right pr-4 font-normal">window x,y</th>
<th className="text-right pr-4 font-normal">size</th>
<th className="text-right pr-4 font-normal">page x,y</th>
<th className="text-right pr-4 font-normal">screen x,y</th>
<th className="text-left font-normal">vis</th>
</tr>
</thead>
<tbody>
{capture.elements.map((el) => (
<tr key={el.index} className="border-t border-slate-50">
<td className="pr-4 py-1">{el.tag}{el.id ? `#${el.id}` : ''}</td>
<td className="pr-4 py-1 max-w-xs truncate">{el.text}</td>
<td className="pr-4 py-1 text-right">{Math.round(el.viewport.x)},{Math.round(el.viewport.y)}</td>
<td className="pr-4 py-1 text-right">{Math.round(el.viewport.width)}×{Math.round(el.viewport.height)}</td>
<td className="pr-4 py-1 text-right">{Math.round(el.page.x)},{Math.round(el.page.y)}</td>
<td className="pr-4 py-1 text-right">{Math.round(el.screen.x)},{Math.round(el.screen.y)}</td>
<td className="py-1">{el.inViewport ? '✓' : el.visible ? 'off-screen' : 'hidden'}</td>
</tr>
))}
</tbody>
</table>
</div>
</div>
)}
{view === 'rendered' ? (
<>
{pinned && pinned.capturedAt !== capture.capturedAt && (
<div className="px-4 py-1.5 border-b border-slate-100 text-xs text-amber-600">
Frozen snapshot from {new Date(pinned.capturedAt).toLocaleTimeString()} click Rendered again to refresh
</div>
)}
{/* allow-scripts WITHOUT allow-same-origin: the frame runs the
page's own JS inside an opaque origin, so it cannot reach this
dashboard's DOM, storage, or same-origin API routes — /api/firms
serves firm credentials. Granting both flags together is what
would let a frame drop its own sandbox.
Scripts are needed because most sites ship content at opacity:0
and fade it in with JS — blocked, they render blank. */}
<iframe
sandbox="allow-scripts"
srcDoc={withBaseTag((pinned ?? capture).html, (pinned ?? capture).url)}
title="Captured page"
className="block w-full h-[70vh] bg-white"
/>
</>
) : (
<pre className="max-h-[70vh] overflow-auto bg-slate-900 text-slate-200 text-xs font-mono p-4 whitespace-pre-wrap break-all">
{capture.html}
</pre>
)}
</div>
)}
</div>
</div>
);