Adds sites.root_domain (indexed string, backfilled for existing rows)
computed via a naive "last two labels" heuristic (App\Support\RootDomain)
— good enough for example.com/somesite.es, but wrong for two-part
public suffixes like .co.uk. Doing that correctly needs the Mozilla
Public Suffix List; flagged as a known simplification rather than
built now.
Deliberately not a hard FK: like Backlink.target_domain, a subdomain
can get indexed before its root domain does (or vice versa), so
Site::subdomains()/rootSite() match on the domain string rather than
requiring the relationship to exist at write time.
Site detail page: a root domain shows a paginated "Subdomains"
section (same guest-preview cap as Pages/Backlinks); a subdomain
shows "Subdomain of {root}" linking back, when the root is itself
indexed.
Verified against real data: 21k.ro correctly lists its two known
subdomains (buzau.21k.ro, timisoara.21k.ro), each of which correctly
links back to 21k.ro. Added SubdomainsTest (root lists subdomains,
subdomain links to root, empty state) — full suite (31 tests) passes.
Adds the auth Breeze scaffolds that were missing before: password
reset and email verification, plus a profile page (update name/email,
change password, delete account). User now implements MustVerifyEmail
so verification is actually enforced, not just scaffolded.
This does mean accepting Tailwind v3 (postcss/tailwind.config.js)
over the v4 CSS-first setup the app started with, and Livewire drops
from ^4.3 to ^3.6.4 — both forced by Breeze/Volt's current
compatibility requirements. Restored the Instrument Sans font Breeze's
installer dropped so styling stays close to what we had.
Reconciliation, since Breeze's installer assumes a blank-slate app:
- Deleted the hand-rolled Login/Register Livewire components/views/
test — superseded by Breeze's Volt-based auth pages
(resources/views/livewire/pages/auth/*), which also cover
forgot-password, reset-password, verify-email, and confirm-password.
- Deleted Breeze's own default welcome/dashboard views and its
separate app layout + nav component — kept our existing
components/layouts/app.blade.php as the one real app layout, used
Breeze's minimal layouts/guest.blade.php only for the auth pages.
- Rewrote routes/web.php merging our routes (home, sites, discover)
with routes/auth.php; replaced every hardcoded route('dashboard')
reference (5 Blade files plus, easy to miss, a plain PHP controller
VerifyEmailController) with sites.index.
- Logout: Breeze's Actions\Logout::__invoke() returns void, so wired
directly as a route it silently no-ops instead of redirecting;
fixed with a thin closure that calls it then redirects.
- Fixed Breeze's own generated tests (Auth/AuthenticationTest,
RegistrationTest, EmailVerificationTest, PasswordConfirmationTest)
where they assumed the deleted dashboard route/nav component.
Verified: full suite (28 tests, including Breeze's Registration/
Authentication/PasswordReset/PasswordConfirmation/PasswordUpdate/
EmailVerification/Profile tests plus our own GuestPreviewTest) passes
against the huginn_testing MySQL database. Also smoke-tested for real:
all key pages return 200, /profile redirects guests to /login, guest
preview capping still works, and a real password-reset email was
generated and logged with correct branding and a working link.
Hand-rolled login/register/logout on Laravel's built-in Auth facade
rather than Breeze/Jetstream, which scaffold Tailwind v3 config and
would fight our existing Tailwind v4 CSS-first setup. Login/Register
are Livewire components; logout is a plain POST route (auth-gated,
CSRF-protected).
Routing: "/" is now a public landing page describing the project;
the sites list moved from "/" to "/sites" to make room for it.
Data gating: App\Support\GuestPreview::paginate() forces page 1 and a
5-row page size for unauthenticated visitors on Sites, Discover, and
a site's Pages/Backlinks tables, while total() still reflects the
real match count — authenticated users get the existing full
pagination unchanged. A shared guest-preview-notice component shows
"N of M results, sign in to see all" and only renders when something
is actually being hidden.
Testing: pointed the test suite at a dedicated MySQL database
(huginn_testing) instead of the default in-memory SQLite, since the
existing fulltext index migration isn't SQLite-compatible — this app
is MySQL-specific by now regardless. Added AuthFlowTest (register,
login, wrong-password rejection, logout) and GuestPreviewTest (guest
capped to a preview, authenticated sees everything), all passing.
Also fixed a real bug the tests caught: request()->session() isn't
reliably attached during Livewire's in-process test action calls;
switched Login/Register to the session() helper, which resolves the
session store directly and works in both contexts.
Adds Site/SitePage models and the crawlindex:fetch command, which pulls
CDX index records from Common Crawl and optionally enriches them with
page title/description/keywords/body text extracted from the WARC
archive. CDX/WARC access and the site/page upsert logic live in
CommonCrawlIndexService and WarcRecordFetcher.
m105