send_fcm.php هي نقطة كل الرسائل والمكالمات بين الراكب والسائق (التطبيقان
يستدعيانها مباشرة)، ولم تكن تحتوي أي error_log إطلاقاً — بخلاف FcmService
التي تطبع [FCM_DEBUG] وتخدم دورة حياة الرحلة فقط.
النتيجة أن فشل الرسائل كان غير قابل للتشخيص: لا سبب، ولا رد Google، ولا
حتى معرفة إن كان الطلب وصل. وقد وعدت المستخدم بقراءة [FCM_DEBUG] لهذا
المسار وهو وعد خاطئ — لا وجود له هنا.
أُضيفت ثلاث نقاط:
- نتيجة الإرسال: category + طرف من التوكن + http + طول التوكن + رد Google
عند الفشل. طول التوكن مقصود: توكن FCM الحقيقي ~163 محرفاً، والمشفّر في
جدول tokens 216 — فيُكشف أي blob مشفّر من سطر واحد.
- رفض 403: التطبيقان لا يرسلان x-api-key، فلحظة ضبط FCM_INTERNAL_API_KEY
تموت كل الرسائل بينهما بصمت. الآن يُسجَّل السبب صريحاً.
- رفض 400 على target فارغ: يعني أن المُرسِل لا يملك توكن الطرف الآخر
(tokenPassenger أو driverToken لم يصله في حمولة القبول).
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
الموضعان كانا يقرآن مفتاحاً لا يُكتب في أي مكان في المشروع:
$redisLocation->hGetAll("driver:location:$driverId")
الكاتب الفعلي هو معالج الدفعات في loction_server/driver_socket.php، وهو
يكتب hmset على driver:profile:{id} و driver:public:{id} معاً بنفس الحقول
(lat/lng/heading/speed/status/updated_at). و driver:public له TTL 86400
بينما driver:profile له 900 فقط، فالعام هو الأنسب للقراءة.
الأثر: كانت حمولة القبول تصل الراكب بلا إحداثيات أولية للسائق، فلا يظهر
الماركر إلا بعد أول تحديث موقع من السوكيت أو الـ polling.
أحد الموضعين ملف getRideOrderID.php الذي أضفته في f66db7db — نسخت النمط
من acceptRide.php فنقلت الخطأ معه.
مثبَّت على الإنتاج: redis-cli --scan --pattern 'driver:*' أرجع
driver:public:<id> فقط، ولا شيء باسم driver:location.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
تطبيق الراكب يطلب /ride/rides/getRideOrderID.php منذ البداية، والموجود على
السيرفر getRideOrderIDNew.php فقط ⇒ 404 صامت في getUpdatedRideForDriverApply.
هذا يفسّر عرضين ظنّاهما منفصلين:
- الـ polling يسبق الـ FCM فيضبط _isAcceptanceProcessed=true ثم يأخذ 404،
فيُرفض بعدها payload الـ FCM الذي يحمل driver_info كاملاً
("Already processed") ⇒ الرحلة مقبولة بلا معلومات سائق.
- ولأن driverToken يبقى فارغاً، ترد send_fcm.php بـ 400 Missing: target
⇒ رسائل الراكب للسائق لا تُرسَل إطلاقاً. (الاتجاه المعاكس كان يعمل بعد
إصلاح getRideWaiting.php — نفس خطأ التشفير معكوساً.)
getRideOrderIDNew.php لا يصلح بديلاً: داخلي عبر get_connect.php →
validateInternalKey فلا يستطيع التطبيق مناداته، ولا يرجع
ratingCount/completedRides/driverTier.
النقطة الجديدة:
- connect.php (JWT)، والراكب من الـ JWT فقط لا من الطلب (حماية IDOR).
- نفس استعلام acceptRide.php وشكل رده حرفياً حتى يقرأه
_fillDriverDataLocally بنفس المفاتيح.
- فك تشفير driverToken.token — بلا ذلك يصل التطبيق blob يستخدمه كـ FCM
target فيرفضه FCM بـ 400.
- تفادي تصادم المفاتيح: getUpdatedRideForDriverApply يقرأ
passengerName + last_name كاسم الراكب، فلقب السائق نُقل إلى
driver_last_name واسم الراكب الحقيقي يُرجَّع في مكانه.
- ride DB هو المرجع مع fallback على primary، ورحلة بلا سائق ترد success
لا failure.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
سوكيت الراكب (سبب عدم ظهور معلومات السائق عند القبول):
- تطبيق الراكب كان يرسل id فقط بلا jwt، و passenger_socket.php يرفض أي
اتصال بلا jwt ⇒ الراكب لا ينضم لغرفته أبداً ولا يستلم ride_status_change
ولا driver_location_update. تظهر حالة القبول عبر الـ polling فقط بينما
driver_info يصل بالسوكيت وحده. (سوكيت السائق يعتبر الـ jwt اختيارياً،
ومن هنا جاء التباين بين التطبيقين.)
- cancelled_by_driver كان يسقط من switch حالات الراكب فيبقى معلّقاً بعد
إلغاء السائق.
- حماية socket (late) من القراءة قبل التهيئة عند الانسحاب بلا jwt.
الإشعارات والرسائل (سبب "مرات توصل ومرات لا"):
- جدول tokens يخزّن توكن الراكب مشفّراً، و getRideWaiting.php كان يرجعه
بلا فك تشفير ⇒ من يقبل من قائمة السوق يحمل blob مشفّراً يستخدمه كـ FCM
target فيرفضه FCM بـ 400: لا إشعار قبول ولا رسائل. ومن يقبل من الـ
dispatch/FCM يحمل نصاً صريحاً فتعمل. الفرق كان في طريقة القبول.
- acceptRide.php يحلّ التوكن من القاعدة دائماً ولا يثق بالعميل (أصحّ أمنياً).
- market_new_ride كان لا يحمل passengerId ولا الإحداثيات فتصل "null"؛
أُضيفت بلا أي PII لأن الحمولة تُبَثّ لكل سائق قريب لا للفائز فقط.
- send_fcm.php: مهلة على OAuth (كان يعلّق حتى مهلة PHP فتُسقط الرسالة
بصمت)، توحيد ding→default لأندرويد، وحقن title/body/tone في data
مطابقةً لـ FcmService.
- تطبيق السائق يقرأ title/body من data أولاً مثل الراكب، ولا يعرض فقاعة
فارغة للرسائل الصامتة.
السوكيت والإعدادات:
- forwardLocationToPassengerSocket كان يقرأ lat/lng والحمولة فيها
latitude/longitude ⇒ المسافة تخرج ضخمة والـ throttle معطّل تماماً
فيُعاد التوجيه مع كل نبضة GPS.
- notifyPassengerOnRideServer كان يرجع null بصمت مطلق عند حجب العنوان.
- العنوان الافتراضي لسيرفر الموقع كان nginx/loction_server/driver_socket.php
وهو ديمون Workerman لا يُخدَم عبر nginx ⇒ صار socket_driver:2021.
(LOCATION_API_URL بقي على nginx لأن api_get_nearby.php سكربت عادي.)
- ride_server/passenger_socket.php (النسخة التي يشغّلها Docker) كانت ناقصة
كل كود مواصلاتي الموجود في passenger_server/ ⇒ نُقل مع REDIS_HOST.
- .env.example: ALLOWED_SOCKET_URLS يغطّي أسماء حاويات Docker، وإضافة
PASSENGER_SOCKET_INTERNAL_URL.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Broadcasts never reached anyone. The internal FCM call defaulted to
127.0.0.1, which inside the php container is the php container itself — the
web server runs in a separate nginx container, reachable by service name on
the Compose network. Every send failed the curl and returned a generic 502.
The default now points at nginx, the URL is overridable via FCM_INTERNAL_URL,
and the error carries the actual reason and target instead of a bare status.
Transit organisations were a read-only count table with nothing to act on.
The module now supports the operations an admin actually needs:
- open an organisation for its counts, routes, recent trips and admins
- create one, including the founding administrator create.php requires
- edit city, contact details, contract status and trial end, with a
confirmation when the contract changes since suspending cuts off service
- add an administrator, and enable or disable an existing one
Verified end to end against the endpoints' real payload shapes, including
that an incomplete create form is rejected before any request is sent.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Eight queries across the v2 modules counted only status = 'Finished' and so
reported zero on live data, where the current ride pipeline writes
'completed': realtime revenue for today and yesterday, financial stats,
settlements, driver scorecard, driver ranking, and both revenue queries. All
now match either spelling.
Growth and Advanced Analytics rendered through the generic shape-detecting
renderer, which produced raw tables that said little. Both now have purpose-
built views:
- Growth: totals, 30-day joins, and a two-series daily chart. growth.php only
returns days that had signups, so the series is expanded to a continuous
30-day axis with explicit zeros — plotting the returned rows directly would
hide the gaps and make a quiet month look like steady growth. A caption
states how many days actually had a signup.
- Analytics: revenue summary tiles, a daily revenue trend, and the captain
ranking, with a note explaining that platform share is what remains after
the captain's cut.
Null aggregates render as "—" rather than 0.00, and markers are drawn only on
days with a value so a flat zero line stays readable.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
- Customer-service notes joined to the account by comparing encrypted phone
columns. Both notes tables now carry phone_key, written when a note is
saved, and the three joins match on it.
- The email_verifications join was comparing a plaintext column against an
encrypted one, so it never matched and `verified` was always NULL in both
passenger and driver sign-in. It is now resolved in PHP against the
decrypted address, which fixes a pre-existing bug rather than only
preparing for GCM.
- auth/sendVerifyEmail.php built all three of its statements by interpolating
the request values into SQL. Any caller could inject through the email or
token field. Now parameterised.
- serviceapp/register.php duplicate detection consults the users indexes and
writes them with the row.
Sweep confirms no join or lookup compares two encrypted columns any more.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Storing the verification phone as a keyed HMAC fixed OTP lookups but broke
every query that joined those tables back to the account, because
phone_verification*.phone_number no longer holds the same value as
driver.phone / passengers.phone. Six joins were affected, and four of them
feed the `verified` flag that the rider and driver apps check at sign-in — so
this was already failing under the current CBC mode, not only after a switch
to GCM.
Accounts now carry phone_key, computed exactly as otpPhoneKey() does, and the
joins match on it. It is written at registration for both apps and populated
for existing rows by the backfill.
The backfill also covers the columns added for the remaining lookups:
users.email_bidx/phone_bidx and driver.national_bidx, which were migrated but
never populated, and honours a per-field prefix so phone_key reproduces
otpPhoneKey's exact output.
Insert column/value counts verified with a paren-aware parser after editing.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Completes the set of queries that matched a freshly encrypted value against a
stored one, which only works while encryption is deterministic. Each keeps its
original comparison and adds an index comparison in the same WHERE, so nothing
changes today.
- passenger sign-in by email, service-staff sign-in, Firebase token lookup
- driver lookup by phone and by national number
- admin ride lookup and ride monitor (both tables)
- nabeh: driver status, user resolution, ride history, complaint submission
transit_org_admins lives in the transit database and has no index column, so
login there falls back to decrypting the small set of active admins and
comparing normalised numbers.
Schema: adds users.email_bidx/phone_bidx and driver.national_bidx with their
indexes.
Verified that every :*_bidx placeholder introduced is actually bound — an
unbound one is a fatal error at request time, not a silent miss.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Storing the verification phone as a keyed HMAC made OTP lookups independent
of the encryption mode, but the hash is one-way — and customer service reads
those same rows to chase people who requested a code and never finished
registering. That workflow would have lost the number entirely.
The verification tables now carry both forms: phone_number holds the lookup
key, and a new phone_enc column holds the encrypted number, which is
decryptable when a human needs to call.
The two follow-up queries also compared the verification row against the
driver/passengers tables and the notes tables by matching ciphertext, which
only ever worked because encryption was deterministic. Under GCM every number
would have looked unregistered and every note would have disappeared. Both now
read the number from phone_enc and match on normalised plaintext, so they are
correct under either mode.
Rows written before phone_enc existed are skipped rather than shown without a
number.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The verification tables (token_verification*, phone_verification*) use the
phone number as a lookup key: written when the code is sent, read when it is
checked. Storing it encrypted worked only because encryptData() is
deterministic — under AES-GCM the two sides would produce different
ciphertexts and no code would ever verify, locking every user out of
registration and OTP sign-in.
otpPhoneKey() stores a keyed HMAC of the normalised number instead. No schema
change is needed since the column is textual, local and international formats
now resolve to the same key, and the value cannot be reversed without the
pepper. It falls back to the previous behaviour when no pepper is configured.
Applied to both sides of every affected flow — request/verify, and the driver
and passenger send/verify pairs — including the OTP value itself where it is
compared by equality rather than decrypted. auth/otp/verify.php already
decrypts the token before comparing, so it needed no change there.
Also adds ENCRYPTION_MODE to EncryptionHelper: encryptData() writes GCM when
set to 'gcm', CBC otherwise. Verified in both directions — rows written under
CBC stay readable after switching, and rows written under GCM stay readable
after rolling back — so the switch is reversible by an environment variable.
The admin console's own OTP is unaffected: it keys the table by the stored
ciphertext read from adminUser, identical on both sides.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
These are the paths that must stop depending on deterministic encryption
before storage can move to AES-GCM. Each keeps its original ciphertext
comparison in the same statement, so behaviour is unchanged today and no
account becomes unreachable during the transition.
Lookups:
- auth/login.php — passenger sign-in matched the raw value against the
encrypted column, which only works because encryptData() is CBC with a
fixed IV.
- auth/passenger/register.php and auth/driver/register.php — duplicate
detection. Without the index these would stop detecting existing accounts
under GCM and allow the same phone to register twice.
Writes now populate the index in the same statement as the value:
- both registration paths write phone/email/name indexes with the row;
driver indexes are computed before the encryption pass, since the raw
values are unavailable afterwards.
- passenger profile update and admin driver update refresh the index when
the underlying field changes. For the composite name index the untouched
half is read back from the row.
Adds --audit to the backfill script: recomputes every index from its
encrypted value and reports missing or stale entries. Drift here is silent
by nature — it surfaces only when a real search fails.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Confirms a value is findable through the blind index after backfilling,
without opening the console or the database. Prints matched row ids only.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Hamza-Ayed