feat: add PowerDNS reverse DNS (PTR) integration

Introduces an opt-in reverse DNS management subsystem backed by a PowerDNS
Authoritative HTTP API. Runs via a companion WHMCS addon module
(modules/addons/VirtFusionDns) that holds settings and a Test Connection
page; the server module reads those settings from tbladdonmodules and
short-circuits when the addon is absent or disabled, so provisioning is
unaffected for operators who don't use the feature.

Lifecycle hooks:
- createAccount creates PTRs for every assigned IP (forward DNS must
  already resolve to the IP — FCrDNS enforcement)
- renameServer updates only PTRs whose content matched the old hostname,
  preserving client-custom records
- terminateAccount deletes all PTRs before the local state is purged
- TestConnection merges PowerDNS health check with the existing VirtFusion
  check
- A DailyCronJob hook reconciles missing PTRs additive-only (never
  overwrites)

Client UI: new "Reverse DNS" panel on the service overview with one
editable PTR input per assigned IP, per-row status badges, and
forward-DNS rejection on save. Admin services tab gets a parallel
widget with Reconcile (additive) and Reconcile (force reset) buttons.

New subsystem at lib/PowerDns/:
- Client.php    PowerDNS API wrapper (X-API-Key, listZones/getZone/
                patchRRset/notifyZone), auto-NOTIFY on successful PATCH
- Config.php    Loads + decrypts addon settings from tbladdonmodules
- IpUtil.php    PTR-name generation (IPv4 + IPv6), zone matching,
                RFC 2317 classless parsing
- Resolver.php  FCrDNS verification via dns_get_record with CNAME-chain
                following and per-(hostname,ip) caching
- PtrManager.php Orchestrator: syncServer, deleteForServer, listPtrs,
                setPtr, reconcile, reconcileAll

Security hardening helpers added to Module and applied to the rDNS
endpoints:
- requirePost()           HTTP method gate (405 on non-POST mutations)
- requireSameOrigin()     Origin/Referer check against WHMCS host (CSRF
                          defence against cross-site form POST)
- requireServiceStatus()  tblhosting.domainstatus filter (Active for
                          writes, Active+Suspended for reads)

RFC 2317 classless delegations (e.g. 64/64.113.0.203.in-addr.arpa.)
supported with alignment validation: rejects misaligned start addresses
that don't correspond to any real delegation boundary.

PowerDNS zone IDs containing '/' are URL-encoded as '=2F' per the
PowerDNS API convention. PATCH success triggers PUT /zones/{id}/notify
so slaves pick up the SOA-bumped serial immediately.

Includes IPv4 + IPv6 support, per-IP write rate limit (10s), fresh
IP-ownership re-verification on every client write (defends against
stale-ownership after IP reassignment), and audit logging of every
successful edit to the WHMCS module log.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
Prophet731
2026-04-17 21:08:22 -04:00
parent d253bd44e6
commit ad85439dfb
18 changed files with 3312 additions and 21 deletions

View File

@@ -10,8 +10,49 @@ use WHMCS\Database\Capsule;
* server feature methods (power, network, VNC, backup, resource modification,
* self-service billing, traffic, rename, password reset).
*
* Extended by ModuleFunctions (service lifecycle) and ConfigureService (order-time
* operations). Most business logic lives here; subclasses delegate to these methods.
* INHERITANCE SHAPE
* -----------------
* Extended by:
* - ModuleFunctions — service lifecycle (create, suspend, unsuspend, terminate, change package)
* - ConfigureService — order-time operations (package/template discovery, server build init)
*
* Most business logic lives HERE, not in the subclasses. Subclasses are intentionally
* thin — they orchestrate sequences of calls to methods defined on this base, which
* lets us unit-exercise any single feature (e.g. "what happens during rename when
* the VirtFusion API returns 423?") without standing up a full WHMCS lifecycle.
*
* THE resolveServiceContext() PATTERN
* -----------------------------------
* Almost every method follows the same preamble: look up the module table row,
* look up the WHMCS tblhosting row, resolve the control panel credentials, build
* a Curl client with the bearer token. That preamble is consolidated into
* resolveServiceContext() which returns everything as an array or false on any
* missing piece. Every feature method starts with "$ctx = $this->resolveServiceContext($id);
* if (! $ctx) return false;" and can then use $ctx['request'], $ctx['serverId'], etc.
*
* This pattern is the most important abstraction in the module — violating it
* (e.g. reading tblservers directly in a feature method) leads to drift where
* some features handle missing servers gracefully and others don't.
*
* ENDPOINT OUTPUT CONVENTION
* --------------------------
* client.php and admin.php call $this->output() to emit JSON responses. Every
* output() call in a switch case MUST be followed by a `break` — the module
* deliberately does NOT rely on exit() inside output() for flow control because
* that couples the HTTP response format to the control-flow mechanism and makes
* refactoring fragile.
*
* SECURITY HELPERS
* ----------------
* Five guards callers compose in front of sensitive actions:
* - isAuthenticated() — client session required
* - adminOnly() — admin session required
* - requirePost() — HTTP method gate (mutations only)
* - requireSameOrigin() — CSRF origin check
* - requireServiceStatus() — filter by tblhosting.domainstatus
*
* Each exits on failure with the appropriate HTTP status — callers treat them
* as "throw on failure" style assertions rather than having to check return values.
*/
class Module
{
@@ -73,10 +114,23 @@ class Module
/**
* Resolve service context: system service, WHMCS service, control panel, and curl client.
* Returns false if any lookup fails.
*
* This is the most-called method in the module. Every feature action begins
* by calling it, so think of the return value as "everything you need to
* touch VirtFusion for this service":
*
* service — row from mod_virtfusion_direct (has server_id, server_object)
* whmcsService — row from tblhosting (has server, userid, domain, etc.)
* cp — ['url', 'base_url', 'token'] for the VirtFusion API
* request — a fresh Curl instance pre-configured with the bearer token
* serverId — (int) of service.server_id — used in every URL downstream
*
* Returning false on ANY missing piece lets callers write a single
* "if (! $ctx) return false;" check at the top of each feature method
* rather than threading nullability through three separate lookups.
*
* @param int $serviceID
* @return array{service: object, whmcsService: object, cp: array, request: Curl}|false
* @return array{service: object, whmcsService: object, cp: array, request: Curl, serverId: int}|false
*/
protected function resolveServiceContext($serviceID)
{
@@ -328,13 +382,37 @@ class Module
return false;
}
// Capture old hostname + server object from stored state so we can sync rDNS
// after the rename. We read from the cached server_object rather than a fresh
// fetch; this is the hostname the PTR would be set to (if module-managed).
$oldHostname = null;
$serverObject = null;
if (! empty($ctx['service']->server_object)) {
$serverObject = json_decode($ctx['service']->server_object, true);
if (is_array($serverObject)) {
$oldHostname = PowerDns\PtrManager::extractHostname($serverObject);
}
}
$ctx['request']->addOption(CURLOPT_POSTFIELDS, json_encode(['name' => $newName]));
$data = $ctx['request']->patch($ctx['cp']['url'] . '/servers/' . $ctx['serverId'] . '/name');
Log::insert(__FUNCTION__, $ctx['request']->getRequestInfo(), $data);
$httpCode = $ctx['request']->getRequestInfo('http_code');
$success = $httpCode == 200 || $httpCode == 204;
return $httpCode == 200 || $httpCode == 204;
if ($success && $serverObject !== null && PowerDns\Config::isEnabled()) {
// Sync PTRs: only records whose current content equals the old hostname
// will be rewritten; client-customized PTRs are preserved automatically.
// Non-blocking: rDNS failures log but never fail the rename.
try {
(new PowerDns\PtrManager)->syncServer($serverObject, $oldHostname, $newName);
} catch (\Throwable $e) {
Log::insert('PowerDns:renameServer', ['serviceID' => $serviceID], $e->getMessage());
}
}
return $success;
} catch (\Exception $e) {
Log::insert(__FUNCTION__, [], $e->getMessage());
@@ -773,6 +851,26 @@ class Module
/**
* Resolve a WHMCS server record into an API base URL and decrypted Bearer token.
*
* OUTPUT SHAPE
* ------------
* url — full API base like "https://vf.example.com/api/v1". Append
* path components to this for every VirtFusion call.
* base_url — scheme + host only, "https://vf.example.com". Used for SSO
* redirects where we need to hit the panel UI, not the API.
* token — decrypted bearer token. Pass to initCurl() to get an
* authenticated Curl handle.
*
* $any=true is an unusual behaviour: when a WHMCS product doesn't have a
* specific server pinned (allowed if the module is the only VF module on
* the install), we fall back to any enabled VirtFusion server. This mostly
* exists for the "Test Connection" button which doesn't know which server
* to use until after a successful connection. Normal provisioning always
* passes a real server ID.
*
* The token is stored encrypted in tblservers.password and decrypted here
* via WHMCS's global decrypt() — the same encryption key used for addon
* module password fields.
*
* @param int|object $server WHMCS server ID or server object
* @param bool $any When true, fall back to any available server if the given one is not found
* @return array{url: string, base_url: string, token: string}|false
@@ -825,6 +923,164 @@ class Module
$this->output(['success' => false, 'errors' => 'unauthenticated'], true, true, 401);
}
/**
* Enforce POST as the HTTP method. Emits a 405 JSON response and exits otherwise.
*
* WHY THIS EXISTS
* ---------------
* The REST principle says mutations should be POST, and PHP's $_POST / $_GET
* separation means a mutation that reads from $_POST would fail quietly when
* called via GET. But "fail quietly" isn't what we want — an attacker probing
* endpoints via crafted <img src="?action=...&ip=...&ptr=..."> tags shouldn't
* even reach our input-validation code. This gate kills that path with a 405
* before any per-endpoint logic runs.
*
* Combined with requireSameOrigin() below, this closes the most common
* cross-site request forgery vectors (form POST, image GET) without needing
* explicit CSRF tokens threaded through every AJAX call.
*
* @return bool|void
*/
public function requirePost()
{
if (($_SERVER['REQUEST_METHOD'] ?? '') === 'POST') {
return true;
}
$this->output(['success' => false, 'errors' => 'method not allowed'], true, true, 405);
}
/**
* Verify the request's Origin/Referer belongs to this WHMCS install.
*
* THREAT MODEL
* ------------
* A logged-in WHMCS user visits a malicious page. That page makes a POST
* to our rDNS endpoint; because the session cookie is tied to our domain,
* the browser attaches it automatically. Without this check, the attacker
* could silently rewrite the user's PTRs.
*
* The defence: browsers attach an Origin header on cross-origin fetch/XHR
* and a Referer on cross-origin form POST. Those headers carry the
* attacker's origin, not ours — so we compare them against our own
* hostname and reject mismatches with a 403.
*
* This is NOT a full CSRF token scheme. It defends against the common
* cross-site-POST and cross-site-form-submit vectors but a same-site XSS
* that can read the user's DOM could still circumvent it. For that you'd
* need per-request tokens bound to the session — out of scope for the
* current module, but the helper stays here ready to be composed with
* a token check if one's added later.
*
* IMPLEMENTATION
* --------------
* 1. Collect our "known good" host set from HTTP_HOST (what the browser
* connected to) plus the SystemURL host from tblconfiguration (what
* WHMCS thinks its canonical URL is). Behind a reverse proxy these
* can differ; accepting either closes the false-positive gap.
* 2. Parse HTTP_ORIGIN and HTTP_REFERER and pull out their host:port.
* 3. Require at least one of those headers to match.
*
* Fails closed: if we can't determine our own host OR if neither Origin
* nor Referer is present, we reject. A legitimate same-origin AJAX call
* from the module's own JS always sets Origin (fetch API) or Referer
* (form submit), so the "both absent" case only happens with scripted
* non-browser clients — which are exactly who we want to filter out.
*
* @return bool|void true on success; emits 403 JSON and exits otherwise
*/
public function requireSameOrigin()
{
$expected = [];
$host = (string) ($_SERVER['HTTP_HOST'] ?? '');
if ($host !== '') {
$expected[] = strtolower($host);
}
$systemUrl = Database::getSystemUrl();
if ($systemUrl) {
$parsed = parse_url($systemUrl);
if (! empty($parsed['host'])) {
$expected[] = strtolower($parsed['host'] . (isset($parsed['port']) ? ':' . $parsed['port'] : ''));
$expected[] = strtolower($parsed['host']);
}
}
$expected = array_unique(array_filter($expected));
if (empty($expected)) {
// Can't determine our own host; fail closed rather than silently allow.
$this->output(['success' => false, 'errors' => 'cross-origin check failed'], true, true, 403);
}
$origin = (string) ($_SERVER['HTTP_ORIGIN'] ?? '');
$referer = (string) ($_SERVER['HTTP_REFERER'] ?? '');
$candidates = [];
foreach ([$origin, $referer] as $raw) {
if ($raw === '') {
continue;
}
$parsed = parse_url($raw);
if (! empty($parsed['host'])) {
$candidates[] = strtolower($parsed['host'] . (isset($parsed['port']) ? ':' . $parsed['port'] : ''));
$candidates[] = strtolower($parsed['host']);
}
}
if (empty($candidates)) {
$this->output(['success' => false, 'errors' => 'cross-origin check failed (missing origin)'], true, true, 403);
}
foreach ($candidates as $c) {
if (in_array($c, $expected, true)) {
return true;
}
}
Log::insert('csrf:origin-mismatch', ['origin' => $origin, 'referer' => $referer, 'expected' => $expected], 'cross-origin request rejected');
$this->output(['success' => false, 'errors' => 'cross-origin check failed'], true, true, 403);
}
/**
* Ensure the WHMCS service is in a status where client-initiated writes make sense.
*
* tblhosting.domainstatus can be: Active, Suspended, Terminated, Pending,
* Cancelled, Fraud. Not every action makes sense in every status:
* - Reads (rdnsList, serverData) usually allow Active + Suspended so a
* suspended user can still see their current config.
* - Writes (rdnsUpdate, power, etc.) typically require Active only —
* mutating a cancelled service's rDNS has no sensible business meaning.
*
* Pass the allowed set explicitly per endpoint rather than trying to encode
* a global policy here. Some endpoints (admin reconcile) don't call this at
* all because the admin is allowed to touch any service.
*
* Fails with 404 if the service doesn't exist, 400 otherwise — keeping the
* two conditions distinct in the response code helps client-side error
* handling (a 404 usually means "link is stale", a 400 means "not right now").
*
* @param int $serviceID WHMCS service ID
* @param string[] $allowedStatuses Service statuses that permit the operation
* @return bool|void true on success; emits 400/404 JSON and exits otherwise
*/
public function requireServiceStatus(int $serviceID, array $allowedStatuses = ['Active'])
{
$row = Database::getWhmcsService($serviceID);
if (! $row) {
$this->output(['success' => false, 'errors' => 'service not found'], true, true, 404);
}
if (! in_array((string) $row->domainstatus, $allowedStatuses, true)) {
$this->output(
['success' => false, 'errors' => 'service status "' . (string) $row->domainstatus . '" does not permit this action'],
true,
true,
400,
);
}
return true;
}
/**
* Create a pre-configured Curl instance with JSON Accept/Content-Type headers
* and a Bearer token for authenticating against the VirtFusion API.