addr:housenumber / addr:postcode etc. are free-text in OSM and occasionally
exceed the entity's VARCHAR limits (e.g. housenumber "12-14, 16, 18"), which
aborted the whole --tiled run with a 1406 "Data too long" error. firstTag()
now takes a max length and each field is clamped to its real column size.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Public:
- /escape-rooms: Leaflet + OpenStreetMap map with marker clustering and a
searchable list, both built from a cached slim JSON feed
- /escape-rooms/{id}-{slug}: room detail with a mini-map and user reviews
(rating + title + body, login required)
- /escape-rooms/suggest: visitor submission form with click-to-drop-pin map
and reCAPTCHA v3; new rooms land as pending
Data:
- EscapeRoom / EscapeRoomReview entities (soft-delete), migration
- app:escaperooms:import-osm console command pulls leisure=escape_game from
the Overpass API, upserts by (osmType, osmId), skips locked rows;
--tiled / --area / --bbox options
Admin (new "Website Admin" area, separate from Game Admin):
- src/Website/Controller/AdminEscapeRoomController.php at /admin/escape-rooms:
status-filtered list, new/edit, publish/hide/reject, delete, review removal
- game admin sidebar extracted into a block and cross-linked
Assets: leaflet + leaflet.markercluster; escaperoom-map and escaperoom-detail
Encore entries.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Both refuse to run unless docker/.env names a non-prod COMPOSE_PROJECT_NAME and
scope every compose call to that project.
setup.test.sh: like setup.sh but runs the DB/cache/console steps as www-data and
repairs var/cache|log|sessions ownership at the end, so php-fpm can read its own
compiled cache (bare `docker exec` runs as root -> silent 500s). Never touches
var/volumes/db. Test-appropriate final message (NPM upstream, local curl check).
restart.test.sh: keeps the database by default (--fresh-db to wipe + re-init),
never runs a host-wide docker prune, and only chowns var/cache|log|sessions
(chowning var/volumes/db is what corrupted the MySQL data dir).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
compose.yaml / compose.override.yaml:
- container_name is now ${STACK_NAME:-escapepage}-* (STACK_NAME is a plain var,
not COMPOSE_PROJECT_NAME, so prod keeps its escapepage-* names with no config change)
- every published host port is ${*_PORT:-<current default>}, so prod is unchanged
and a second stack can bind its own (localhost-only) ports
- nginx joins the external nginx_default network so Nginx Proxy Manager can
forward to <stack>-nginx by name
restart.sh:
- scoped to STACK_NAME / COMPOSE_PROJECT_NAME read from docker/.env, so running it
from the test checkout can't touch the prod stack
- host-wide `docker system prune` / `docker builder prune` moved behind --prune-all
Adds docker/.env.test.example and doc/test-environment.md (separate checkout,
env layers, NPM proxy host + Access List IP allowlist, Mercure on test).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Admin can now open a message via a per-row "View" page (which stamps
read_at the first time it's opened), see a sortable Read column on the
list instead of the raw message text, and soft-delete messages via a
Delete button on both the list and detail view. Deleted messages are
excluded from the default list so it doesn't pile up over time.
Adds a public /contact page (name/email/message, protected by the same
reCAPTCHA v3 setup used on registration) that persists submissions to
a new contact_message table, plus a Contact section in the admin panel
to read them. Linked from the site footer.
Adds a /terms page and links it from the site footer and from the
"I agree to the Terms and Conditions" checkbox on registration, which
previously didn't point anywhere.
Composer: ~40 Symfony 7.4.x packages patched to 7.4.17, plus four
majors - doctrine/dbal 3->4, phpdocumentor/reflection-docblock 5->6,
symfony/mercure-bundle 0.3->0.5, symfony/monolog-bundle 3->4. Removed
two doctrine.yaml keys (use_savepoints, report_fields_where_declared)
that DBAL 4 deprecated in favor of fixed defaults.
npm: @symfony/webpack-encore 4->6, which required bumping its peers
webpack-cli 5->6 and sass-loader 14->16 together, plus babel-loader
9->10. Fixes the one high-severity audit finding (RCE in
serialize-javascript, via the old css-minimizer-webpack-plugin). One
moderate finding remains in webpack-notifier's dev-only notification
chain - audit's suggested fix would downgrade it, so left alone; it's
build tooling only, never shipped to users.
Verified: full test suite green, lint:container clean, both `encore
dev` and `npm run build` compile without errors, and the production
CSS output was inspected byte-for-byte to confirm the new SVG-minifier
warnings (on Bootstrap's pre-encoded icon data-URIs) don't corrupt
anything.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Uses simple-datatables (vanilla JS, no jQuery needed) on the Users,
Sessions, Email Log, and Feedback tables, since those grow with real
usage. Left Games (small, admin-curated) and Hints (row order is
meaningful - "checked in order" - a sortable column would mislead)
untouched.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The badge on the homepage and briefing page was static markup, wired
to nothing - flipping a game's status to open in the admin panel had
no effect on it. It's now driven by GameRepository::hasOpenGame() and
only shows while every game is still in development/locked.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
pwd/ls/scan/sudo/rm/cd/cat now work as /pwd, /ls, /scan, /sudo, /rm,
/cd, /cat too, delegating to checkConsoleCommando the same way the
bare chat/verify/decode aliasing already delegates the other
direction.
Hints used to be 11 hardcoded PHP strings/thresholds in
SendMainframeHintsCommand, walked through a fixed if/else chain. They
are now Hint rows (game, condition, thresholdSeconds, message,
sortOrder) manageable at /admin/games/{game}/hints - freely
addable, removable, and reorderable per game, which sets up exactly
what's needed for difficulty-dependent hint timing later (Easy/Medium/
Hard are separate games, so each gets its own hint set).
The condition a hint fires behind (e.g. "hasn't used help yet",
"isn't verified yet") is still a fixed, code-defined check
(HintCondition enum) tied to real game state - a true free-form
condition editor would need a full rules engine, which is out of
scope. What's fully free-form is which hints exist, in what order,
gated behind which of those conditions, with what wording and timing.
SendMainframeHintsCommand now reads hints from the DB: walking them in
sortOrder, it finds the first distinct condition still unresolved for
a player, fires the most-escalated hint sharing that condition whose
threshold has passed, and stops - matching the original "stop at the
first unresolved, dependent step" behavior.
Migration seeds every existing game with the previous hardcoded
hints so behavior doesn't regress on deploy. Two minor fidelity
simplifications from the original hardcoded logic, called out here
since they're easy to miss: the "go to rapports directory" hint no
longer additionally checks whether the player is already in that
folder, and the two chat hints are now independent conditions
(broadcast done / contacted everyone privately) instead of one
combined check - both are edge-case-only behavior changes, not
regressions in the common path.
New game_admin_game_new route/form (reuses AdminGameType, same pattern
as editing a game's total time) so games - including the 3 upcoming
Easy/Medium/Hard difficulty options - can be created through the admin
panel instead of needing a direct DB insert.
Also show each game's total time (in minutes) next to the player
count in the "Create New Session" dropdown, since that's the actual
difficulty signal players are choosing between - number of players
alone didn't convey it.
Difficulty itself needs no new session-level concept: each difficulty
is just a separate Game row with its own TOTAL_TIME setting, which
checkAllPlayersReady() already reads when starting the session's
timer. Hint timing depending on difficulty is separate, upcoming work.
grantRights() is now the only place a player's RightsForPlayer{N}
setting ever gets written. All three previous call sites (verify+cat
after chat tracking, cd+decode after /verify, and the all-players
grant from decoding) now delegate to it instead of duplicating the
same read-modify-persist logic.
The main point: since every grant now flows through one place, it can
notify the specific player over Mercure the moment they receive new
rights, instead of them only finding out by trying a command that
used to be "Unknown command".
Also includes an already-present (uncommitted) tweak allowing bare
chat/verify/decode console input without the leading slash - unrelated
to this change but sitting in the same file, so it's coming along.
- DecodeMessage::PLAYER_3 ("locked up bash files should be removed to
lock it up") was awkward and unclear about what it meant.
- The private-communication hint said "contact each other privately",
which read as any one private message rather than the actual
requirement of messaging every other agent privately.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
rm always glued its argument onto the player's current directory, so
typing an absolute path (e.g. sudo rm /var/arrest/handle.sh) built a
garbage path that matched no file and was rejected as "not allowed" -
even with full rm/sudo rights. cd already special-cased a leading '/'
as absolute; rm now does the same.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Players had no in-game indication of what actually ends the game
before reaching the terminal itself.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Every message append (mainframe broadcasts, lock reveals, the boot
sequence, and responses to your own commands) forced the page to jump
to the bottom. Removed all four window.scrollTo calls so the view
stays put.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Feedback was only ever written to session_setting rows with no way to
view it short of querying the database directly. Adds a FeedbackService
that pulls per-player feedback entries and aggregate averages, backing
two new views:
- /admin/feedback: a full table of every submission (game, session,
player, ratings, comment) plus summary tiles, linked from the admin
sidebar.
- /briefing: a "Field Reports" block with the average difficulty/
entertainment/theme ratings, shown to prospective players. Free-text
comments are deliberately left off this public page since they're
unmoderated player input.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Pressing ArrowUp while the input field is focused now repopulates it
with the last submitted command, cursor placed at the end.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
- testToggleReady: user was never marked verified, so toggleReady()
returned false before doing anything (an isVerified() gate was
added to the service after this test was written).
- testCheckAllPlayersReadyTransitionsStatus: entityManager mock
returned the same SessionSettingRepository for every getRepository()
call, but the service now also fetches a GameSettingRepository for
the session's total-time setting, causing a TypeError. Route the
mock by requested class instead.
- testChatRegeneratesVerifyCodesIfShared: two competing
method('getSetting') stubs were registered without with()
constraints; PHPUnit keeps the first one it sees active for every
call, so the (correct, more complete) willReturnCallback stub was
silently dead code and the regeneration path never actually ran.
Dropped the redundant first stub, and updated the flush() count now
that the real flow (chat tracking + code regeneration) executes.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Decoy names in /var/home and verifyCodes.txt were a hardcoded 4-name
list (Luke, Charles, William, Peter), so a real player registering
under one of those names would collide with it. Now each session
picks 6 decoys from a pool of 10, excluding any name already taken by
a real player, and stores the pick as a session setting.
Likewise the 3 "special report" rapports that get coded messages were
always Doyle, Vega and Lennox. Each session now randomly assigns 3 of
the 20 rapports to that role, and the win screen / mainframe-help
message reference whichever agents were actually picked instead of
the hardcoded names.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Crops the key/globe mark out of the existing logo and wires it into
base.html.twig as favicon.ico plus PNG/apple-touch-icon variants.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Replaces the inline "New here?" link with a proper button and a heading
that more directly asks whether the player is unfamiliar with running
commands on a command line.
Lets players leave the tutorial at any time via a top-left link back to the
game session. Once the tutorial finishes and the input field is disabled,
the button relocates into the input row so it's the obvious next action.
Adds a black-terminal tutorial page at /game/pregame/{session} that walks
players through help, pwd, cd, ls, cat, rm and sudo before the real game
starts. Links to it are shown on both the lobby (waiting for players) and
ready-up screens so players can practice, repeatedly, before starting.
Any /chat {agent-id} {code} was previously exempted only if aimed at
the code's rightful owner; sending to the wrong player still burned
it. Now only an untargeted, open-chat broadcast counts as a leak -
mistargeting via /chat is harmless.
checkAndRegenerateVerifyCodes() regenerated a code on any message
containing it, regardless of who the message was sent to - so sharing
a code exactly as intended, via /chat {agent-id} {code} to the
correct recipient, still burned it immediately, making the puzzle
unsolvable. Now only broadcasts and messages sent to the wrong player
count as a leak.
generateSpecialCode() reused generateRandomString(), whose charset
includes a space. Across a 75-100 character code that made a space
almost certain, and /decode splits its argument on spaces, silently
truncating the code the player typed back in - breaking decoding.
Fixed at generation time rather than at display time, since the
stored value itself was the problem, not just how it's shown.
generateRandomString() is left untouched for its other use (the
"garbage" filler text shown to players who fail to decode), where
spaces are harmless.
Every $hub->publish() call site (chat, hints, security alerts, virus
alerts, game_finished, and the pre-game lobby events) now dispatches
a PushMercureMessageEvent instead of publishing directly. Two
listeners handle it: PublishMercureMessageListener does the actual
Mercure publish exactly as before (same wire format, no client
changes needed), and LogMercureMessageListener appends it to the
activity log of whichever player(s) it was actually delivered to.
A private /chat to one agent only gets logged for that agent, never
broadcast into everyone's transcript - the event carries an explicit
targetScreen (null = everyone) rather than leaving listeners to guess
from the payload shape.
The per-player log format moved from flat text to JSON Lines (one
timestamped, structured entry per line) via a new shared
SessionActivityLogger service, since a flat string can't carry an
event's type or exact payload - both needed for a future feature to
replay a player's full session history, not just their own commands,
on page reload. The admin session log viewer now parses and
formats these entries back into readable lines.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Previously only the cron's timeout sweep broadcast to everyone;
check-finished (called by a player's own client the moment their
local countdown hits zero) just updated the database silently. Now
whichever path notices the timeout first broadcasts - both only act
while the session is still PLAYING, so there's no double broadcast.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The cron (app:hints:check, every minute) now walks each playing
player through a strict dependency chain - help, communication,
verification, navigation, rapports/decode, endgame urgency - and
sends at most one hint per player per run: the first step that's
both unresolved and old enough to nudge about. Later steps genuinely
depend on earlier ones (e.g. /verify isn't usable until a player has
finished contacting everyone), so a player never gets a hint for
something they can't act on yet.
Two bits of new tracking were needed:
- Help command usage wasn't recorded anywhere, so it's now saved to
a new HelpUsedForPlayer{N} session setting.
- "All messages decoded" needed a reliable session-wide signal, so
the rm right (previously granted alongside sudo when player 1
decodes) now comes from player 3's decode instead, making
sudo+scan+rm together mean "the whole team has decoded".
The cron also now finishes sessions whose countdown has run out:
sets the session to LOST and broadcasts the same game_finished
Mercure event the win path already uses, so every connected player
gets pushed to the lost page within a minute of the game actually
ending - not just whichever player's own client-side timer happens
to notice first.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Welcome agent ' + + ' to the mainframe.' had a stray += with nothing
between them, evaluating to NaN. Plugs in the screen variable that
was already sitting in scope right above it.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
checkAllPlayersReady() set the session to PLAYING and cleared/updated its
timer in memory but never flushed, relying on callers to do so. The
toggleReady() caller flushed right after, but GameController::index()'s
lazy catch-up call did not - so if the "everyone ready" transition was
only detected on a page load (e.g. players didn't click ready within the
same 60s window), the terminal would render for that one request from the
in-memory state, letting the game be played entirely through the
unguarded message API, while the database silently kept the session on
'ready' with timer 0 forever. This made sessions invisible to the mainframe
hint cron and the admin "running sessions" count. Moved the flush inside
checkAllPlayersReady() itself so both callers persist reliably.
Also chmod the logrotate configs after COPY in the Dockerfile, since
their on-disk mode ended up group-writable (0664) depending on the
build host's umask, which made logrotate refuse to use them.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Not exploitable today - the only custom auth message data is a
hardcoded resend link - but |raw on a translated exception message is
a latent XSS pattern if a future change ever threads user input
through the auth exception's message data. Twig's default
autoescaping is sufficient here.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The admin panel already checked CSRF tokens on destructive actions,
but the player-facing raw HTML forms (create/join/leave/start
session, toggle ready, lobby chat, feedback) had none - cookie
SameSite=Lax blunts classic cross-site auto-submit attacks but isn't
a substitute for real tokens. Adds matching csrf_token()/
isCsrfTokenValid() checks to all of them.
Also adds login_throttling (5 attempts/15 min) to stop unlimited
password guessing, and a per-user rate limiter (10/min) on the
invite-code join endpoint, since invite codes are only 32-bit and
had no protection against brute-forcing.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
composer audit reported 37 advisories across 15 packages, including
high-severity ones in symfony/security-http. Ran composer update
within the existing 7.4.* constraints - composer audit now reports
zero advisories. Also adds symfony/rate-limiter, needed for login
throttling and invite-code rate limiting in the next commit.
Flex removed a stale, non-functional sendgrid notifier config left
over from before the app switched to Mailgun as part of the sync.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Registration only validated username with NotBlank, so a username
like "../../../../public/x" got concatenated directly into a
filesystem path for both writing (game activity logs) and reading
(admin log viewer) - reachable from the public webroot since public/
is a few directories up from where those logs are stored.
Adds a character-set validator (letters, numbers, underscore, hyphen)
to registration and admin user editing going forward, and sanitizes
at the point of use (Player::getLogFileBasename()) so any
already-stored unsafe username can't escape the log directory either.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The lobby chat was a standalone card sitting above the per-player log
tabs. Folds it into the same tab bar as the first (default-active)
tab instead, so the session log view has one consistent tab strip.
The tab-switching script now toggles by an .admin-tab-panel class
instead of assuming every panel's id starts with "player-", since
that's no longer true.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The chat used to disappear the moment a session left CREATED status.
Sessions now record a finishedAt timestamp when they're won or lost,
and the lobby (with chat) stays reachable via /game/{session} for an
hour afterward instead of immediately redirecting to the win/lose
feedback page. The lobby template shows a distinct "game finished"
header with a link to that feedback page during this window.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Unreferenced by any controller - superseded by
templates/game/admin/sessions/view.html.twig, which is what
GameAdminController::viewSession() actually renders. Confirmed via
grep across src/ and templates/, plus a clean lint:twig and phpunit
run after removal.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Several templates had their own <script> blocks instead of going
through webpack like game1.js already does. Extracted the waiting
page, lobby page, and admin session-log tab scripts into their own
files under assets/, imported from the main app.js entry. Since
app.js is already loaded on every page, each module just reads its
own data-* attributes and no-ops if its target element isn't present
on the current page - same pattern game1.js already uses.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Admins can now see the full lobby chat transcript (who said what,
when) at the top of a session's log view, alongside the existing
per-player terminal logs. Also swaps the log tabs' inline onclick
handler for a data attribute, in prep for moving the tab-switching
script out of the template.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Players used to get bounced back to the dashboard when a session
wasn't full yet. Now they land on a lobby page showing who has
joined, and can chat with each other while waiting - messages are
broadcast live over the existing Mercure hub, and the session
auto-starts (and the lobby notifies everyone) once it fills up.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The sidebar was a fixed 220px flex column that left barely any room
for content on a phone, and tables had no min-width so columns just
squeezed into unreadable slivers instead of scrolling.
Moves the sidebar's layout rules out of inline styles (which media
queries can't override) into real CSS classes, and collapses it into
a horizontally-scrollable pill bar below 768px so the content area
gets the full screen width. Tables now get a sensible min-width so
they scroll horizontally on narrow screens instead of squishing, and
the per-player tab bar on the session log view scrolls too.
Players saw the raw enum value (e.g. "created") in the sessions
table, which doesn't mean anything to them. Adds SessionStatus::label()
mapping each status to a player-facing description like "Waiting for
players".
Lost accidentally in the previous commit; both scripts are invoked
directly (./docker/setup.sh) rather than via bash, so they need +x.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
- Game1 terminal: flesh out the virtual filesystem with a realistic
spread of Linux directories/files (~70 dirs, ~140 files) so it no
longer reads as an obviously small puzzle set, without touching any
win-condition or rapport files.
- Add app:hints:check command + a php-cron container (BusyBox crond)
that nudges players who haven't contacted every teammate 5 minutes
into a session, via a new 'hint' Mercure message type.
- Log the cron command's output to var/log/cron/cron.log and rotate
it (25MB / 90 days) via logrotate, run daily from the same crontab.
- Redirect PHP's error_log and Symfony's prod app/deprecation logs
from stderr-only into var/log/php/*.log (kept alongside stderr),
with the same rotation policy.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
A player's own browser now starts a 60s timer the moment they check
"ready" (with a visible countdown) and proactively tells the server
when it's up via a new expire_ready action - idempotent, so it only
actually clears the status if the deadline has genuinely passed.
Everyone else finds out live through a Mercure player_ready broadcast
that now carries which player and their new ready/not-ready state
(previously the broadcast payload's `ready` flag was always false due
to a stale-variable bug, though nothing consumed it yet).
checkAllPlayersReady() still does the same expiry check server-side
and broadcasts on anyone else's request in the meantime, so a stalled
frontend timer doesn't leave a stale "ready" badge showing forever.
This only affects the pre-PLAYING ready phase: checkAllPlayersReady()
still flips the session to PLAYING and stops touching ready state the
instant everyone is simultaneously ready, so the earlier fix (a reload
should never send an already-started game back to the waiting room)
is unaffected.
Also fixed the ready checkbox itself: unchecking it submitted a POST
without `toggle_ready` in the body (unchecked checkboxes aren't sent),
so un-readying silently did nothing server-side. Added the standard
hidden-fallback-input pattern to fix it.
GameDashboardService::toggleReady() now rejects the toggle if the
user's email isn't verified (mirrors the same gate UserChecker already
applies at login). The waiting-room checkbox is disabled client-side
with an explanatory alert and a link to resend the verification email,
and the controller adds a flash error as a server-side fallback if it
somehow gets submitted anyway.
New /profile route (nav link between Admin and Logout) with two
independent forms, both requiring the current password before
applying a change:
- Change password: standard current/new/repeat flow, same password
strength rules as registration.
- Change email: re-checks the new address isn't already taken (avoids
a 500 from the unique constraint), then marks the account
unverified and re-sends the confirmation email via the existing
EmailVerifier/verify-email flow, same as registration. The current
session stays logged in, but the user needs to verify the new
address before their next login (UserChecker already blocks
unverified accounts).
GameController::index() only special-cased READY; every other status
(including a freshly-created session still waiting for players, and
even an already WON/LOST one) fell straight through to rendering the
live game terminal - which is broken there since screens/rights/pwd
are never initialized before startSession() flips CREATED -> READY.
The dashboard's "Enter Game" button links to this route regardless of
status, so any player clicking it on a CREATED session hit this
directly. Now CREATED redirects back to the dashboard with an
explanatory flash, and WON/LOST redirect to their respective pages
instead of re-rendering a dead terminal.
Fixes the 500 on admin user deletion: deleting a user with an
email_log row (i.e. basically anyone who received any email) hit an
unhandled ForeignKeyConstraintViolationException, since email_log,
reset_password_request and player all have non-nullable FKs to user
with no cascade at the DB level.
Instead of cascading the delete (which would be fine for email_log
and reset_password_request but risky for player - removing a player
row could corrupt other real players' session state), admin delete
now sets a deletedAt timestamp instead of removing the row:
- UserChecker blocks login for deleted users (checkPreAuth).
- EmailLoggerListener rejects the message before send for deleted
recipients (checked at actual send time, not at queue time).
- Added a last_login_at column + a LoginSuccessEvent listener to
populate it, and surfaced both last login and status in the admin
users list.
Added `app:users:purge-deleted`, a command intended to run on a
schedule that permanently removes users who were soft-deleted more
than 3 months ago and never played a game (join to Player via a
NOT EXISTS subquery). For that eventual hard-delete to actually
succeed, email_log and reset_password_request now cascade-delete at
the DB level (migration drops+recreates both FK constraints with ON
DELETE CASCADE) - player intentionally still isn't cascaded, so a
user with game history can never be purged this way even by mistake.
Cancelled donations now show "Donation failed, but thank you for
trying anyway." and successful ones show "Thank you for the
donation!" on the dashboard after redirect.
The audit fix (presumably run with --force) jumped @symfony/webpack-encore
from ^4.6.1 to ^7.1.0, which requires @babel/core ^8.0.0 as a peer -
but @babel/core was left at ^7.25.0, so a clean `npm install` failed
outright with ERESOLVE. Since docker/setup.sh and docker/restart.sh
both run npm install unconditionally, this would have broken every
future deploy/rebuild.
Reverted webpack-encore (and the incidentally-downgraded
webpack-notifier) back to the versions that were actually working,
regenerated package-lock.json, and verified both `npm install` and
`npm run build` succeed cleanly.
Ready status previously expired 60 seconds after a player checked the
box, evaluated per-player against their own timestamp. Unless every
player happened to click ready within the same 60-second window, the
earliest player's readiness would silently expire before the last one
joined, so the session could get stuck on "Waiting for all players to
be ready" indefinitely, especially after a reload re-triggered the
timeout check.
Ready state is now durable: once checked, it stays until the player
unchecks it or the whole group is simultaneously ready, at which point
the session always transitions to PLAYING regardless of how long that
took. session.timer is still only ever set once during that one-way
READY -> PLAYING transition, so a reload never restarts or desyncs the
countdown between players.
Removing all 3 locked files while the timer is still running now marks
the session WON immediately (checked right after every successful rm)
and broadcasts a "game_finished" signal over Mercure so every
connected player gets redirected together, not just the one who
removed the last file. A new /won/{session} route + won.html.twig
mirrors the existing lost flow (victory narrative + the same feedback
form).
The existing timer-expiry path already set LOST but always redirected
to lostUrl regardless of actual status; it now picks won/lost based on
the status the server reports.
Also fixes a pre-existing bug on the lost page (and would-be bug on
the new won page): PlayerService::GetCurrentlyActiveAsPlayer() only
matches players in READY/PLAYING sessions, so by the time a session
has ended it always returned null there, silently breaking the
feedback form. Both pages now look the player up directly via
PlayerRepository instead.
Added a navigatingAway flag so the page's "confirm before leaving"
prompt doesn't block our own win/lose redirects.
Previously the restore check only ran lazily on a player's next
message, so files could stay wrongly-removed for an arbitrary amount
of time after the window expired. The frontend now schedules a
setTimeout (using the server-provided deadline, mirroring the terminal
lock's reveal timer) that pings the backend right at the deadline so
the check runs promptly regardless of player activity. Restored via
data-files-removal-deadline on page load too, so a refresh mid-window
doesn't lose the timer.
Removing one of the 3 AI-virus-protected files now starts a 60-second
window (tracked session-wide via LockedFilesRemovalDeadline). If the
other locked files aren't also removed before it expires, the virus
restores whichever ones were deleted and broadcasts a red warning to
the whole session, forcing players to coordinate the removal instead
of picking them off one at a time.
Also extends the Mercure broadcast payload with an optional 3rd
"messageType" element so pushed messages can render red (virus) or
green (mainframe) instead of always defaulting to green.
Message font dropped from 20px to 14px and the 10px inline margin-bottom
(set from JS on every message) replaced with a 2px CSS margin so the
terminal reads like dense console output instead of spaced-out chat
bubbles. Colors are unchanged.
Connects the previously-disconnected /decode command to per-player
messages: player 1 unlocks sudo for everyone, player 2 unlocks a scan
command that reveals which files the AI virus has locked, player 3
learns those files should be removed.
Also adds a retaliation mechanic: successfully removing a file locks
the player's terminal. The AI virus locks it out in red immediately;
the mainframe reveals a 12-character recovery code in green after 30
seconds, which can be submitted via /unlock to restore access early,
otherwise the terminal auto-recovers after 45 seconds.
Overlay a rotated CSS badge on the homepage hero and briefing page
image instead of baking it into the PNG, so it's easy to remove once
the game ships.
Give the public homepage a themed breach-alert hero and add a /briefing
page with the full mission narrative, both linking into the existing
game dashboard flow.
Adds a Marketing column to the admin users table and a total opt-in
count in the header, so admins can see who signed up for updates on
future projects.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Registration now requires agreeing to the Terms and Conditions and lets
users opt in to hear about future projects. The opt-in is persisted on
the user via a new marketing_opt_in column (migration included).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Admins now see an Admin link in the main navigation, and the admin
section inherits the branded header/footer from layout/site.html.twig
instead of the bare base layout.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
checkIfAllPlayersVerified() read and wrote the EveryoneVerified
setting scoped to whichever player happened to trigger the check
(getSetting(..., $player) / setPlayer($player)), but it's meant to be
a single session-wide flag. getFileContent() correctly reads it as
session-global (no player, filters player IS NULL), so the two never
matched: the "everyone verified" Mercure message still fired (that
code path only checks its own player-scoped copy), but the special
code injection into the Doyle/Vega/Lennox report files never ran
since getFileContent() never found the setting. Store and read it
consistently as session-global.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
getFileContent() built the game1 filesystem path as a relative
string with no leading slash. That only resolves correctly when
PHP's cwd happens to be the project root (e.g. CLI), but under
PHP-FPM/nginx the cwd is nginx's document root (public/), so
file_exists() always failed for real requests even though the file
existed and the in-memory virtual file list (used by ls) said it
should. Use the already-injected $projectDir (%kernel.project_dir%),
matching how the class already builds the session log path.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Every possible path/file in GameResponseService uses a leading slash
(e.g. /var/home/{username}), but a player's initial pwd was seeded as
'var/home/{username}' without one. getAllCurrentFilesInDirectory()
matches entries by comparing getPrevPath() (which always has the
leading slash) against pwd, so a fresh player's ls always came back
empty until their first cd command happened to normalize the format.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
When the last player clicked ready, toggleReady() published a
redundant 'player_ready' event on top of checkAllPlayersReady()'s
'all_ready', and the client reloaded on every message with no guard
and without closing the EventSource. Chrome kept the old page's
script (and its EventSource) alive across the overlapping reload
calls, causing repeated reconnects to the Mercure hub; Firefox
apparently tore the page down fast enough to mask it. Skip the
redundant publish server-side, and make the client reload idempotent
by tracking whether it already fired and closing the EventSource
before reloading.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
docker/.env feeds MERCURE_EXTRA_DIRECTIVES, a Caddyfile-style config
block, via Compose variable substitution. Quoting a space-separated
value there makes Caddy treat both origins as one single malformed
token rather than two arguments, which crash-loops the Mercure
container on startup. Compose's .env parsing for values with spaces
doesn't require quotes (unlike Symfony's Dotenv), so drop them.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Production's MERCURE_CORS_ALLOWED_ORIGINS only allowed
https://escapepage.com, but nginx has no www redirect
(server_name _;), so the site is also reachable at
https://www.escapepage.com. Visitors on the www host got a CORS
error on the Mercure EventSource connection since the Origin header
didn't match the allow-list. Dev's .env already allowed both; bring
docker/.env.dist in line, and fix its stale MERCURE_PUBLIC_URL
(bare domain instead of the mercure. subdomain actually used).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
confirmation_email.html.twig and reset_password/email.html.twig were
plain unstyled HTML with no shared structure. Extract a table-based
layout (templates/emails/layout.html.twig) with header/logo, content
block, and footer, so future transactional emails can extend it
instead of starting from scratch.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Mailer dispatches MessageEvent twice when routed through Messenger:
once when queuing (queued=true) and once on the actual send from the
worker (queued=false). EmailLoggerListener logged on both, creating
two rows per email actually sent. Skip the queued dispatch.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The project now sends mail via Mailgun (symfony/mailgun-mailer), not
SendGrid, so the tracked template should reflect the real transport
in use.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
.env/.env.dev/.env.prod/.env.test are now gitignored, leaving no
tracked reference for what variables a fresh checkout needs. Add a
placeholder-only .env.dist (mirroring docker/.env.dist) to copy from.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
.env, .env.dev, .env.prod, .env.test, and docker/.env contained real
production secrets and were tracked in git despite the Symfony
convention of keeping them local-only. Untrack them and ignore them
going forward; docker/.env.dist stays as a template but now uses
placeholder values instead of live credentials.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
