AWCharacterCreator/README.md

Arcane Wasteland Character Creator

Website for creating characters for the Arcane Wasteland TTRPG, with a searchable rulebook and export to PDF or Word.

• Frontend: React (Vite), 8-step wizard, searchable rules page
• Backend: FastAPI, SQLite, python-docx and reportlab for the export
• Everything runs in one container

Run it on the server

docker compose up -d --build

Then open http://localhost:8090 (the port is bound to the machine itself; see DEPLOY.md for running it on a server with its own user, nginx and HTTPS). Users, characters and presets are stored in ./db/aw.db.

Login and accounts

Everything (the whole API, rules, characters) is behind a login. The only public things are the login page itself and the app's JavaScript.

1. Copy .env.example to .env and set AW_REGISTRATION_CODE. People need this code once to create an account; with an empty code registration is disabled.
2. Open the site, choose "Create account", enter a username, a password (min. 8 characters) and the registration code.
3. Sessions last 30 days (HttpOnly cookie, stored server-side). Logging out or changing the password ends them.

Wrong logins and wrong registration codes are throttled (10 failures per address, 8 per username, 5 wrong codes per address and 30 per hour for the whole server), so a short code is not trivially guessable, but a long one is still better. Passwords are stored as scrypt hashes.

HTTPS is a must when the site is reachable from the internet, otherwise passwords travel in plain text. Put the container behind a reverse proxy (Caddy, nginx, Traefik) that terminates TLS, pass the Host header through, and set AW_TRUST_PROXY=1 in .env so the app sees the real client address and marks the cookie Secure.

Admin tasks (list users, reset a password, delete a user):

docker compose exec awcc python -m app.admin list
docker compose exec awcc python -m app.admin passwd <username>
docker compose exec awcc python -m app.admin delete <username>

Characters and presets made before accounts existed are adopted automatically by the account that logs in from the same browser (the old secret keys are used once and then removed).

Edit the rules

All rule data lives in backend/data/rules.json: attributes, derived-stat formulas, skills, professions, classes, talents, mutations, implants, conditions, equipment and the texts of the rules page. Edit the file and reload the page; no rebuild is needed because the folder is mounted into the container.

Entries with "placeholder": true are placeholders and show a badge in the app. To remove the badge, fill in the entry and delete that line.

Note: Derived-stat formulas are in rules.json (derived[].formula = list of attributes whose mods are summed). Special rules that cannot be expressed there (profession attribute bonus, skill maximum, higher-of-two-attributes for skills, carry capacity) are mirrored in frontend/src/calc.js and backend/app/derive.py. When changing them, change both files.

Rules presets

Presets are named sets of creation numbers (attribute points, skill pool, talents, proficiency picks, unlock cost, caps ...). They are defined in rules.json under presets: each has an id, name, desc and a creation object that overrides keys of the global creation block (an empty object = the standard values). default_preset is the one new characters start with. The list of values that can be edited per character is tunables.

In the creator, the Concept page only has a "Rules preset" selector. Presets are created, edited, duplicated and deleted on the Presets tab. Built-in presets (from rules.json) are read-only; user presets are stored in the database (/api/presets) and visible to everyone, but only the browser that created one holds the key to change or delete it. The chosen numbers are copied into the character, so shared links and exports use the same numbers even if the preset is changed or deleted later. "Start over" keeps the chosen preset.

Skills, proficiency and expertise

• creation.advanced_attr_percent: share of the attribute that counts for advanced skills (default 50, rounded down; 100 = full attribute like basic skills).
• creation.advanced_unlock_cost: pool points needed to unlock an advanced skill (free if the character's profession grants proficiency in it; proficiency picks do not unlock a skill).
• creation.free_proficiencies: free proficiency picks every character gets. Talents can add more with "proficiency_picks": N (see "Versatile Training").
• Proficiency is +proficiency_bonus; two picks on one skill (a profession proficiency counts as one) give Expertise, +expertise_bonus.

Equipment

equipment lists every item with id, name, cat (category), stat (damage / armor / type), kg and note. There are no starting-kit rules: players pick any items (with an amount) on the Equipment page and the total weight is checked against the carry capacity. Weights are approximate; for ranges in the source document the upper value is used.

Development

# Backend
cd backend && pip install -r requirements.txt
uvicorn app.main:app --reload --port 8000

# Frontend (second terminal, proxies /api to port 8000)
cd frontend && npm install && npm run dev

Saving and sharing

Characters are saved in your account ("My characters" page). Only the owner can see a character, change it or delete it. On "My characters" the Share button lets you add usernames (e.g. your DM): they get read-only access, see the character under "Shared with me" and can open it through /#creator/<ID>. Saving a viewed character creates their own copy. Use the JSON backup (Summary page) to move a character between servers.

Hover (or tap / focus) the dotted-underlined labels and the small "i" icons to see where a number comes from, e.g. how a skill's target number and maximum are derived, or what a derived stat is made of. The explanations are generated from the rules, so they follow presets and rules.json changes.

While creating, the full character sheet stays visible next to the wizard. Edits are shown as a difference (old → new, added / removed items) until you press Apply (or Apply & next); Discard drops them.

Each wizard page has a Reset this page button that clears only that page's data, plus Start over for the whole character.

Character sheet (A4)

The Summary page and the sheet tabs in the lobbies show the character as A4 pages: page 1 has name, attributes, derived stats and all skills (proficiency dots, attribute, points, total), page 2 has talents, mutations and implants, equipment, background and a ruled notes area. On the Summary page, Print / PDF (A4) opens just the pages in a clean window; choose "Save as PDF" in the print dialog (margins none, background graphics on if your browser asks). Hover or tap a value to see how it is made up (not printed). Light / Dark (in the lobby it sits above the sheet) switches the colours of the sheet on screen (remembered in the browser; printed pages are always light). The Word and PDF downloads on the Summary page are separate exports with their own layout.

Rules editor

Admins get an Editor tab to add, change and delete items, item categories, skills, skill groups and professions with forms. Admins are the usernames in AW_ADMINS (comma separated, set in .env); without that setting only the first account ever created is admin.

• Changes are saved in db/rules_overrides.json (next to the database, so it is part of your backup). For each edited list it replaces the list from rules.json; the shipped file is not touched. A previous copy is kept as rules_overrides.json.bak. "Reset to shipped" removes the override of one list.
• New entries get their id from the name (lowercase, _). Ids of existing entries cannot be changed, because characters refer to them.
• The server checks every reference: an item needs an existing category, a skill a group and attributes, a profession existing skills. Removing a category, group or skill that is still used is refused. Removing an item, skill or profession that saved characters use asks for confirmation first (their data stays, the entry disappears from the sheets).
• Talents, mutations, implants, classes and the rules text are still edited in rules.json. Players see changes after reloading the page.

Lobbies

The Lobbies tab groups a campaign. Whoever creates a lobby is its DM and gets an 8-character join code. Players join with the code and pick one of their saved characters (they can switch it later or leave).

The lobby page is built for desktop with three columns: the character sheet on the left (collapsible with the arrow), the Player Area in the middle and the chat at the right edge of the window.

• Chat (right): live chat for everyone, plus whispers to a single person (dropdown next to the message field). Whispers are only visible to sender and recipient, the DM does not see other people's whispers. The same feed shows the event log (joins, leaves, character switches, renames, new codes, can be hidden), dice rolls and skill checks. Names are shown in the player's colour. Everything in the lobby (chat, dice, difficulty, colours, sheets) updates live over a WebSocket (/api/lobbies/{id}/ws); it only sends a short "something changed" note and the data is fetched through the normal API, so the visibility rules (whispers, hidden numbers) stay in one place. If the socket is blocked or drops, the page reconnects and meanwhile polls every few seconds. The chat keeps the last 1000 entries per lobby and allows 500 characters per message and 30 messages/rolls per minute.
• Character bar (top of the Player Area): one card per player with the character, class and the player's colour. The DM clicks a card to show that character's sheet on the left (always the current saved version, refreshed every 15 s) and can enter a personal bonus / malus per character in the small number field. Players see their own sheet only and just a short card of the others. (The bar is also meant for initiative and enemies later.)
• Skill and attribute checks: players click an attribute or an unlocked skill on their own sheet. The server rolls a d100 (roll under): target = attribute or skill total + the DM's difficulty + the character's personal modifier. 01-05 is always a critical success, 96-100 always a critical failure. The chat and the table show the roll, the target and by how much it succeeded or failed. Combat is not handled specially: roll the normal skill and let the DM adjust with the modifiers.
• Difficulty (bar under the characters): the DM picks a level from the rules table (Trivial +60 ... Hellish -60) or a custom number. It applies to all checks until changed. With Hide numbers from players the players only see their roll and success / failure, not the target, difficulty or margin.
• Dice (below the table): d4 ... d100 with a count (1 to 10 dice). The server rolls, the dice tumble into the middle of the table in the colour of whoever threw them, and the result goes into the chat. With a person selected as recipient in the chat dropdown it is a private roll. DM rolls have three modes: Mystery (default: players see grey dice with "?" but not the result), Hidden (only the DM sees it) and Public.
• Own dice: with this switch on, a click on a die or a skill asks for the result of your real dice instead of rolling. The server only checks the range and marks the roll as "own dice" in the chat.
• Colours: every player has a different colour (lobby menu -> Your colour); the DM is gold.
• Lobby menu (top right): join code, rename, remove players, colour, switch character, leave or delete the lobby.

Maps, tokens and line of sight

• Maps tab (top navigation): all your maps with preview. Upload PNG, JPEG or WebP (up to 12 MB, 30 maps each; the server checks and re-encodes every image as WebP, max 4096 px per side, into db/maps/). Edit opens the map editor, Share gives other users access by username (like characters: they can put the map on their own table or Copy it, but not change it), Duplicate copies a map.
• Map editor: grid (pixels per square, token sizes are counted in squares), enemies and NPCs placed in advance (name, size, colour, Hidden from players, duplicate), and walls and doors for line of sight: tools Wall (click points, chains lines), Room (two corners), Door (two ends), Erase, Undo last; new lines snap to existing corners, optionally to grid corners. An optional sight range in squares limits how far characters see (0 = unlimited).
• Show on table (lobby, DM, Lobby actions): shows one of your or a shared map; the prepared tokens are copied into the lobby, so moving them during play never changes the map. Doors start closed. Every player has a token of their own. Without walls the players drag it onto the map themselves; with walls (line of sight) the DM places it from the tray.
• At the table: pan (drag the background), zoom (wheel, pinch, buttons) and Grid are personal. Players move only their own token; the DM moves everything, edits NPC tokens (click one), adds NPCs (+ NPC) and clicks doors to open or close them (Walls shows the layout in the full view).
• Line of sight is strict: the server calculates what each character can see from the walls and closed doors, cuts the map image to that area (everything else is black, only that piece is ever sent) and leaves out all other tokens. Walls are never sent to players. The DM chooses Full view or View as (exactly what that player gets) and can switch Sight on/off for the table.

Players only see a short card of the other players, never their sheets. Guessing join codes is rate limited. Deleting a character takes it out of the lobbies (the player can pick another); deleting a lobby does not delete characters.

API

All /api/* routes except /api/auth/me, /api/auth/register and /api/auth/login need the session cookie.

Method	Path	Purpose
GET	/api/auth/me	Current user (or null)
POST	/api/auth/register / login / logout / password	Accounts
GET	/api/rules	Rule data including presets
GET / POST	/api/characters	Own and shared-with-me characters / create one
GET	/api/characters/{id}	Read (owner, shared users, or the DM of a lobby it is in)
POST / DELETE	/api/characters/{id}/shares[/{username}]	Share / unshare (owner only)
PUT / DELETE	/api/characters/{id}	Change / delete (owner only)
GET / POST	/api/lobbies	Own and joined lobbies / create one (you are the DM)
POST	/api/lobbies/join	Join with {code, character_id}
GET / PUT / DELETE	/api/lobbies/{id}	Members and cards (DM sees all ids) / rename / delete (DM only)
POST	/api/lobbies/{id}/code	New join code (DM only)
GET / POST	/api/maps	Your maps and the ones shared with you / upload (raw image body, ?name=)
GET / PUT / DELETE	/api/maps/{id}	Map with tokens and walls / change (name, grid, sight, tokens, walls) / delete
POST / DELETE	/api/maps/{id}/shares[/{username}]	Share with a user / stop sharing
POST	/api/maps/{id}/copy	Your own editable copy
GET	/api/maps/{id}/thumb	Preview image
GET / PUT	/api/lobbies/{id}/map	Map and tokens as the viewer may see them (?as=<player> for the DM) / DM: show map (map_id, null clears)
GET	/api/lobbies/{id}/map/view	The map image for this viewer, cut to their line of sight (X-Crop = x,y,w,h)
PUT	/api/lobbies/{id}/doors/{wall} , /sight	DM: open / close a door; line of sight on / off
POST / PUT / DELETE	/api/lobbies/{id}/tokens[/{tid}]	Add NPC token (DM) / move (own token or DM) and edit (DM) / remove (DM)
WebSocket	/api/lobbies/{id}/ws	Live notes {t: messages / lobby / sheet / map} for lobby members (cookie login, same origin)
GET / POST	/api/lobbies/{id}/messages	Chat and event log (?after=<id> to poll) / send, optional to = whisper
POST	/api/lobbies/{id}/rolls	Roll {sides, count, to?, manual?, mode?} (d4-d20, d100; up to 10 dice). manual = results of real dice, mode (DM only) = hidden/mystery/public
POST	/api/lobbies/{id}/checks	d100 check {kind: skill|attr, id, manual?} for the caller's lobby character
PUT	/api/lobbies/{id}/difficulty	{label, mod, hidden} (DM only)
PUT	/api/lobbies/{id}/members/{username}/mod	Personal bonus / malus {mod} (DM only)
PUT / DELETE	/api/lobbies/{id}/me	Switch your character and / or {color} / leave
DELETE	/api/lobbies/{id}/members/{username}	Remove a player (DM only)
POST	/api/presets	Create a preset
PUT / DELETE	/api/presets/{id}	Change / delete (owner only)
POST	/api/export/pdf or /docx	Character JSON in the body, file in the response