fix missed domains

postal-router.php

@@ -34,7 +34,7 @@ declare(strict_types=1);
  * Matching is case-insensitive.
  */
 $ROUTES = [
-    'example.com'       => 'https://example.com/mail-events.php',
+    'example.com'       => 'https://example.com/wp-json/fluent-crm/v2/public/bounce_handler/postalserver/handle/fcrm_xxxxxxxx',
 ];
 
 // Log file. Set to null to disable logging entirely.
@@ -91,59 +91,132 @@ function wh_respond(int $code, array $body): void
 }
 
 /**
- * Pull the sending domain out of whatever shape the payload has.
+ * Postal versions differ in how they wrap events. Older ones POST the event
- * Postal nests the mail details under "message" for essentially every
+ * object directly; newer ones wrap it as
- * event type (MessageSent, MessageDeliveryFailed, MessageHeld, MessageBounced,
+ *   {"event":"MessageSent","timestamp":...,"uuid":"...","payload":{...}}
- * MessageLinkClicked, MessageLoaded, DomainDNSError, ...), but we also scan a
+ * This returns the inner event object to work with, whatever the shape.
- * couple of top-level keys just in case.
  */
-function wh_extract_domain(array $payload): ?string
+function wh_unwrap(array $payload): array
 {
-    $candidates = [];
+    foreach (['payload', 'data'] as $key) {
-
+        if (isset($payload[$key]) && is_array($payload[$key])) {
-    $message = $payload['message'] ?? null;
+            // Merge so top-level keys (event, uuid) stay reachable too.
-    if (is_array($message)) {
+            return array_merge($payload, $payload[$key]);
-        // Preferred: the envelope sender.
-        if (!empty($message['from'])) {
-            $candidates[] = $message['from'];
-        }
-        // Fallback: Postal generates message_id as <token>@<sending domain>.
-        if (!empty($message['message_id'])) {
-            $candidates[] = $message['message_id'];
         }
     }
 
-    // Some events (e.g. bounce payloads) carry the original message separately.
+    return $payload;
-    $original = $payload['original_message'] ?? null;
+}
-    if (is_array($original)) {
+
-        if (!empty($original['from'])) {
+/** Pull a domain out of "user@domain.tld", "Name <user@domain.tld>" or "domain.tld". */
-            $candidates[] = $original['from'];
+function wh_domain_from(string $value): ?string
+{
+    $value = trim($value);
+
+    if ($value === '') {
+        return null;
+    }
+
+    if (preg_match('/@([A-Za-z0-9.-]+\.[A-Za-z]{2,})/', $value, $m)) {
+        return strtolower(rtrim($m[1], '.'));
+    }
+
+    if (preg_match('/^[A-Za-z0-9.-]+\.[A-Za-z]{2,}$/', $value)) {
+        return strtolower($value);
+    }
+
+    return null;
+}
+
+/**
+ * Recursively hunt for the first usable value under any of $keys, at any depth.
+ * Used as a safety net when Postal nests things somewhere unexpected.
+ */
+function wh_deep_find(array $data, array $keys, int $depth = 0): ?string
+{
+    if ($depth > 6) {
+        return null;
+    }
+
+    foreach ($keys as $wanted) {
+        if (!empty($data[$wanted]) && is_string($data[$wanted])) {
+            $domain = wh_domain_from($data[$wanted]);
+            if ($domain !== null) {
+                return $domain;
             }
-        if (!empty($original['message_id'])) {
-            $candidates[] = $original['message_id'];
         }
     }
 
-    // Domain-level events (e.g. DNS error webhooks).
+    foreach ($data as $value) {
-    foreach (['domain', 'from'] as $key) {
+        if (is_array($value)) {
-        if (!empty($payload[$key]) && is_string($payload[$key])) {
+            $found = wh_deep_find($value, $keys, $depth + 1);
-            $candidates[] = $payload[$key];
+            if ($found !== null) {
+                return $found;
+            }
         }
     }
 
-    foreach ($candidates as $candidate) {
+    return null;
-        if (!is_string($candidate) || $candidate === '') {
+}
-            continue;
+
+/**
+ * Work out the sending domain.
+ *
+ * Order matters: message.from is the real sending domain. message_id is only a
+ * fallback, because Postal stamps it with the server's return-path domain
+ * (e.g. ...@rp.postal2.dimail.hu), which is NOT the customer domain.
+ */
+function wh_extract_domain(array $payload, string $rawBody): ?string
+{
+    $event = wh_unwrap($payload);
+
+    $ordered = [];
+
+    foreach (['message', 'original_message'] as $section) {
+        if (isset($event[$section]) && is_array($event[$section])) {
+            foreach (['from', 'sender', 'mail_from', 'from_address'] as $key) {
+                if (!empty($event[$section][$key]) && is_string($event[$section][$key])) {
+                    $ordered[] = $event[$section][$key];
                 }
-        // Handles "user@domain.tld", "Name <user@domain.tld>" and bare "domain.tld".
-        if (preg_match('/@([A-Za-z0-9.-]+\.[A-Za-z]{2,})/', $candidate, $m)) {
-            return strtolower($m[1]);
             }
-        if (preg_match('/^[A-Za-z0-9.-]+\.[A-Za-z]{2,}$/', trim($candidate))) {
-            return strtolower(trim($candidate));
         }
     }
 
+    // Domain-level events (DNS errors etc.) and flat variants.
+    foreach (['domain', 'from', 'sender'] as $key) {
+        if (!empty($event[$key]) && is_string($event[$key])) {
+            $ordered[] = $event[$key];
+        }
+    }
+
+    foreach ($ordered as $candidate) {
+        $domain = wh_domain_from($candidate);
+        if ($domain !== null) {
+            return $domain;
+        }
+    }
+
+    // Safety net 1: deep scan for a sender-ish key anywhere in the structure.
+    $deep = wh_deep_find($event, ['from', 'sender', 'mail_from', 'from_address', 'domain'], 0);
+    if ($deep !== null) {
+        return $deep;
+    }
+
+    // Safety net 2: message_id, accepting that it may be a return-path domain.
+    foreach (['message', 'original_message'] as $section) {
+        if (!empty($event[$section]['message_id']) && is_string($event[$section]['message_id'])) {
+            $domain = wh_domain_from($event[$section]['message_id']);
+            if ($domain !== null) {
+                return $domain;
+            }
+        }
+    }
+
+    // Safety net 3: scrape the raw body for a "from" field we failed to reach.
+    if (preg_match('/"(?:from|sender|mail_from)"\s*:\s*"([^"]+)"/i', $rawBody, $m)) {
+        return wh_domain_from($m[1]);
+    }
+
     return null;
 }
 
@@ -247,13 +320,17 @@ if (json_last_error() !== JSON_ERROR_NONE || !is_array($payload)) {
     wh_respond(400, ['status' => 'error', 'message' => 'Invalid JSON']);
 }
 
-$event  = (string) ($payload['event'] ?? $payload['status'] ?? 'unknown');
+$unwrapped = wh_unwrap($payload);
-$domain = wh_extract_domain($payload);
+$event  = (string) ($unwrapped['event'] ?? $unwrapped['status'] ?? 'unknown');
+$domain = wh_extract_domain($payload, $rawBody);
 
 if ($domain === null) {
+    // Log the WHOLE body here - if extraction ever fails again, this is the
+    // only way to see what shape Postal actually sent.
     wh_log('warning', 'No sending domain found, dropping event', [
         'event'      => $event,
-        'excerpt' => substr($rawBody, 0, 400),
+        'top_keys'   => array_keys($payload),
+        'full_body'  => $rawBody,
     ]);
     // 200 on purpose: this is not a transient failure, a retry would be
     // identical and we'd still have nothing to route on.
@@ -269,7 +346,11 @@ foreach ($ROUTES as $configuredDomain => $url) {
 }
 
 if ($target === null) {
-    wh_log('info', 'No route configured, skipping', ['domain' => $domain, 'event' => $event]);
+    wh_log('info', 'No route configured, skipping', [
+        'domain' => $domain,
+        'event'  => $event,
+        'known'  => array_keys($ROUTES),
+    ]);
     wh_respond(200, ['status' => 'ignored', 'reason' => 'no route for ' . $domain]);
 }