EL LEVERT JR

Romanian version of Leveret by Neeve

Installation • Features • Commands • CLI • WebSocket • API • Configuration • Startup • Importing • Amogus

Installation

Warning

This project only supports Node.js 20.x and 21.x.

Node.js 21.7.3 is the final supported version.

Node.js 22 and newer are not supported.

Use Node.js 20.19.6 if you plan on enabling the inspector features.

Install the bot with:

npm install --omit=optional

Optional runtime support:

• vm2 support depends on the optional vm2 package being available
• external language eval support depends on the separate external VM setup being installed on the machine

Configuration

This section is for bot setup. It covers credentials, feature toggles, interface settings, and the main config files.

Credentials

You can provide bot credentials in any of these ways:

• config/auth.json
• config/auth.env
• existing environment variables

Supported environment variable names:

• LEVERET_TOKEN
• LEVERET_OWNER

If config/auth.json exists, the bot uses that file for token and owner.

If there is no config/auth.json, the bot looks for config/auth.env and then for the matching environment variables.

config/auth.json example:

{
    "token": "bot token",
    "owner": "your discord id"
}

config/auth.env example:

LEVERET_TOKEN=your bot token
LEVERET_OWNER=your discord id

Main bot config

config/config.json controls the main bot behavior, including command prefixes, limits, enabled features, logging, and paths.

Important groups of settings:

• Discord text interface: cmdPrefix
• CLI interface: enableCliCommands, cliCmdPrefix
• WebSocket interface: enableWebsocket, websocketPort
• Eval modes and script network access: enableEval, enableVM2, enableOtherLangs, allowLocalhostRequests, allowFilesystemRequests
• Inspector and debugging: enableInspector, enableUserInspector, timeout and port settings
• Tags and quotas: maxTagNameLength, tagNameRegex, maxTagSize (per-tag limits for text, script, and binary), maxQuota, maxTagCount
• Reply and output limits: outCharLimit, outLineLimit, embedCharLimit, embedLineLimit
• Major features: enablePreviews, enableSed, enablePermissions, enableReminders
• Logging and file paths: log files, log level, database path, command paths, Discord log settings

Reactions config

config/reactions.json controls automatic emoji reactions.

At minimum, it can simply disable the feature:

{
    "enableReacts": false
}

If reactions are enabled, the file defines the bracket reactions and the word-triggered reactions:

{
    "enableReacts": true,
    "parens": {
        "left": ["left parenthesis emoji ids"],
        "right": ["right parenthesis emoji ids"]
    },
    "funnyWords": [
        {
            "word(s)": "word or [word1, word2]",
            "react(s)": "emoji or [emoji1, emoji2]"
        }
    ]
}

Interface and feature notes

• If enableCliCommands is off, the local CLI section in this README does not apply.
• If enableWebsocket is off, the WebSocket API is not available.
• If enableEval is off, eval commands are not available.
• If enableVM2 is off, VM2 eval and VM2 tag execution are not available.
• If enableOtherLangs is off, c, cpp, and py eval modes are not available.
• If allowLocalhostRequests is off (the default), script HTTP requests (request / fetch) cannot connect to localhost, loopback, or private network IP addresses.
• If allowFilesystemRequests is off (the default), script HTTP requests cannot access local files through file:// URLs.
• If inspector support is off, debug flows described in this README are not available.

Features

• Discord text commands with a configurable prefix
• Tags with pipelines and streaming, aliases, quotas, search, raw dump, ownership, and script support
• Snapshot-based tag and permission audit history with revision reversion
• JavaScript eval, optional vm2, optional other languages, and inspector debugging
• Permission groups with configurable levels
• Reminders delivered by DM
• Message link previews, sed-style replacements, and reaction triggers
• CLI commands for local testing and tag/script emulation
• WebSocket command API for external tools and editor integration

Updating the bot

Run the following commands to update the bot with the latest code:

git fetch origin
git reset --soft origin/main

TODO

• Extract responses into a lang file
• Rewrite vm2 runner
• Add cabal (banning)
• Add the option to save tag attachments locally
• Add support for slash commands (perhaps)
• Add support for other databases
• Add caching
• Add different behaviour when replying to messages

Commands

All Discord commands use the bot's text-command prefix. By default that prefix is %.

Argument notation used below:

• (value) means required
• [value] means optional
• "quoted text" means keep the text together as one argument

1. help

%help

Shows the commands you can currently use.

If a command has its own help text, you can usually ask for it with:

%(command) -help

2. ping

%ping

Checks whether the bot is responsive and shows latency information.

3. version

%version

Shows the version of the bot that is currently running.

4. uptime

%uptime

Shows how long the bot has been running and when it started.

5. tag / t

Tags are saved snippets you can call by name. They are useful for canned replies, reusable snippets, aliases, and script-based responses.

Who uses this:

• everyone for normal tag use
• tag owners for creating and maintaining their own tags
• moderators for moderator-only actions such as info and set_type

Common usage:

%tag (name) [args]

Runs the tag named (name). Any extra text is passed to the tag as its arguments.

Main actions:

• add (name) [body] creates a tag
• edit (name) [new_body] replaces a tag's body
• delete (name) removes a tag
• rename (name) (new_name) renames a tag
• hide (name) hides a tag from search, suggestions, and fullsearch
• unhide (name) unhides a hidden tag (binary tags cannot be unhidden)
• raw (name) shows the stored body exactly as saved
• info (name) [raw] shows stored tag details for moderators
• owner (name) shows the owner
• list [user] lists your tags or another user's tags
• search (text) [max_results / all] searches tag names
• fullsearch (query) [max_results / all] searches inside tag bodies
• random (prefix) picks a random matching tag
• count [all/new/script/me/user] shows counts by scope or type
• leaderboard (count/size/usage) [limit] shows tag leaderboards
• quota shows how much tag space you are using
• dump [inline/full] [spaces] exports tags. full sends a JSON file
• alias (name) (target) [args] creates an alias-style tag
• chown (name) (new_owner) transfers ownership
• set_type (name) (type|version) changes script metadata for moderators
• audit [tag_name] [revision_id] [--options] shows recent tag changes or a full revision diff for moderators
• audit_clear [tag_name] [revision_id] [end_revision_id] [--options] clears tag audit history and resets the revision sequence when fully cleared (bot owner only)
• revert <name> [revision_id] restores a previous tag state. Tag owners can revert their own latest actions within one hour; moderators can restore arbitrary revisions.
• stream <pipeline> (or pipe) executes a pipeline of tags and stream operators

Examples:

• %tag audit shows the full audit log across all tags
• %tag audit 42 shows the before/after diff for revision 42 directly
• %tag audit example lists recent revisions for example
• %tag audit example 42 shows the before/after diff for revision 42
• %tag audit --operation update --limit 20 lists recent tag updates
• %tag audit_clear clears all tag audit history and resets the revision sequence back to 0 (bot owner only)
• %tag audit_clear example clears history for one tag
• %tag audit_clear example 42 45 clears an inclusive revision-ID range for one tag
• %tag audit_clear example --from "last week" --to "yesterday" clears a date range for one tag
• %tag stream source_tag > upper runs source_tag and converts the result to uppercase
• %tag stream echo hello world | upper pipes echoed text through the upper operator
• %tag stream source_tag | echo prefix $ suffix uses $ to interpolate piped input
• %tag alias shout stream echo_tag $ | upper creates an alias tag using a pipeline

Tag audit options:

• --user or --actor filters by the user who made the change
• --operation or --op filters by import, create, update, delete, or revert
• --from and --to filter by a date or time range
• --page and --limit control pagination

Audit results are shown in a compact paginated embed. Providing a revision ID shows the tracked fields before and after that change. --from and --to accept natural-language dates and times.

Tag stream / pipeline:

%tag stream (step1) > (step2) > ... > (stepN) %tag pipe (step1) | (step2) | ... | (stepN)

Executes a sequence of steps chained together, passing the output of each step into the next. Both > and | can be used as step delimiters. Up to 30 steps are supported in a single pipeline.

Each step can be the name of an existing tag or a built-in stream operator.

Argument handling and the $ placeholder:

• If a step's arguments contain $, $ is replaced by the piped output from the previous step (for example, source_tag | echo prefix $ suffix).
• If a step does not contain $, the piped output is prepended to the step's arguments.
• Quotes ("..." or '...') keep delimiters like > and | within arguments from splitting the pipeline.
• Tags can alias to stream pipelines (for example, %tag alias shout stream echo_tag $ | upper), which allows calling %t shout custom_input.

Stream operators:

• echo [text]: Outputs the specified text or forwards the piped input. When text contains $, it replaces $ with the piped input.
• unembed [url]: Extracts and stringifies text from embeds. Accepts JSON embed objects/arrays as input, a Discord message URL in arguments or input, or extracts embeds from the current or referenced message.
• unescape: Removes backslash escape characters (for example, \* to * or \_ to _).
• trim: Trims leading and trailing whitespace from the input.
• lower: Converts text to lowercase.
• upper: Converts text to uppercase.
• head [count / -n count]: Returns the first n lines of input (defaults to 10 lines).
• tail [count / -n count]: Returns the last n lines of input (defaults to 10 lines).

What users should know:

• Tag names are normalized to lowercase.
• Tag names and bodies are validated before the tag is saved.
• You can usually edit, rename, delete, or transfer only your own tags unless you have elevated permissions.
• If a tag name is wrong, the bot may suggest similar tags.

Tag body input:

• Plain text creates a normal text tag.
• A code block creates a script tag.
• Starting the body with vm2 creates a vm2 script tag.
• An attached image is stored as an image URL tag.
• An attached text file is stored as script content.
• An attached binary file (MIME application/octet-stream or extension .bin) is stored as a binary tag. Binary tags cannot be executed directly, are hidden by default, and can be retrieved via raw, fetch, or dump.
• For add and edit, admins can also give a local file path instead of inline body text. Absolute paths and file:// URLs must exist. Relative paths are only treated as files when the file exists.
• If a saved tag uses a Discord-hosted attachment URL, the bot warns that the media can disappear if the original source message is deleted.

6. eval

Eval commands run code through one of the bot's script environments.

Who uses this:

• people allowed to use eval features on the server
• admins when file-path input is used

Base usage:

%eval (script)

Runs JavaScript in the default eval environment. You can pass inline code, a code block, or an attached text file.

Modes:

• langs lists the available eval languages
• vm2 (script) runs the script in the VM2 environment
• c (script), cpp (script), py (script) run code in those languages when that support is available

What users should know:

• simple text is sent back as text
• objects and arrays are formatted for you
• empty output is rejected
• some modes may be unavailable on a given bot depending on how that bot is set up

Path input:

• Admins can give a local file path instead of inline script text.
• Absolute paths and file:// URLs must point to a real file.
• Relative paths are only treated as file input when the file exists.

Debugging:

• %eval debug (script) starts a debug session when user-facing inspector support is enabled on the bot.

7. perm / p

Permission groups decide who can use higher-privilege bot features.

Who uses this:

• moderators and admins
• the bot owner for the highest-level changes

Base usage:

%perm (subcommand) ...

Main actions:

• list shows groups and levels
• check (user) shows a user's permissions
• add (group_name) (user) adds a user to a group
• remove (group_name) (user) removes a user from a group
• remove_all (user) removes all groups from a user
• add_group (group_name) (level) creates a group
• remove_group (group_name) deletes a group
• update_group (group_name) [new_name/unchanged] [new_level/unchanged] updates a group
• audit [subject] [revision_id] [--options] shows permission group and membership changes or a full revision diff for permission admins
• audit_clear [subject] [revision_id] [end_revision_id] [--options] clears permission audit history for the bot owner
• revert group (group_name) [revision_id] restores a group revision for permission admins
• revert membership (user_id/group_name) [revision_id] restores a membership revision for permission admins

Examples:

• %perm audit moderators lists revisions for the moderators group
• %perm audit moderators 12345 shows the before/after diff for revision 12345
• %perm audit user-id/moderators lists revisions for one membership
• %perm audit --target permission_group --operation update lists group updates
• %perm audit_clear clears all permission audit history (bot owner only)
• %perm audit_clear moderators clears history for one group or membership
• %perm audit_clear moderators 42 45 clears an inclusive revision-ID range for one subject
• %perm audit_clear --from "last week" --to "yesterday" clears a date range

Permission audit options:

• --user or --actor filters by the administrator who made the change
• --operation or --op filters by import, create, update, delete, or revert
• --target filters by permission_group or permission_user
• --from and --to filter by a date or time range
• --page and --limit control pagination

Audit results show who changed a group or membership, when it happened, and which fields changed.

What users should know:

• The special owner group cannot be edited or removed.
• You cannot add yourself to a group above your own level.
• User arguments may be accepted as a mention, ID, username, or tag depending on the command.

8. reminder

Reminders let users schedule a DM for a future time.

Who uses this:

• everyone who can use reminder features on the server

Base usage:

%reminder (subcommand) ...

Main actions:

• add (date) "message" creates a reminder
• list shows your reminders
• remove (index) removes one reminder
• remove_all removes all reminders

What users should know:

• The date is parsed in natural language, so inputs like tomorrow 18:00 or in 2 hours usually work.
• The message should be quoted when it contains spaces.
• Dates in the past are rejected.
• remove uses a 1-based index from your list.
• Reminders are sent to you by DM.

9. Utility commands

These are general-purpose helper commands for calculations and lookups.

overclock / oc / oceu

Calculates GregTech recipe overclocking, voltage requirements, duration, and parallel throughput.

Syntax:

%oc [mode] <EU> <duration> [base_chance] [chance_bonus] [parallel] [amperage] [--options]
%oc ebf <EU> <duration> <recipe_heat> <coil_heat> [parallel] [amperage] [--options]

Note: <> denotes required arguments, [] denotes optional arguments. Use - to skip an optional positional argument.

Modes:

• standard: Normal overclocking (2x speed, 4x power). Default mode.
• ebf: Electric Blast Furnace mode; calculates heat difference overclocking (each 900K above required heat gives an extra 0.95x duration multiplier).
• lcr: Large Chemical Reactor / perfect overclocking (4x speed, 4x power).
• ce: Emulates GTCE overclocking (2.8x speed, chance doubling per overclock).
• macerator: GTCE macerator mode with special chanced output handling.

Arguments:

• [mode]: Machine processing mode (standard, ebf, lcr, ce, macerator).
• <EU>: Base EU/t cost of the recipe.
• <duration>: Base recipe duration in seconds or ticks (e.g. 20, 30s, 1200t).
• [base_chance]: Base chanced output percentage (1-100).
• [chance_bonus]: Chance increase percentage per voltage tier.
• [parallel]: Number of items or batch operations run in parallel.
• [amperage]: Available amperage (default: 1).
• <recipe_heat>: Minimum heat capacity required by the recipe in Kelvin (EBF mode only).
• <coil_heat>: Total coil heat available in Kelvin (EBF mode only).

Options:

• --tape: Applies maintenance/config hatch speed boost (0.9x duration).
• --subtick: Enables subtick overclocking (doubles parallel processing once duration reaches 1t).
• --extra: Displays extra high-tier voltages (UHV through MAX).
• --rf: Treats input EU as RF cost (divides by 4).
• --time <multiplier>: Multiplies recipe duration by a factor.
• --eu <multiplier>: Multiplies recipe EU/t cost by a factor.
• --voltage <tier> or --<tier>: Filters output table to only the specified voltage tier (e.g. --iv).
• --rates: Displays item production rates (recipes/s) in the table.
• --input <multiplier>, --output <multiplier>: Multiplies input consumption and output production rates.
• --count <number>: Sets recipe batch count.
• --tick: Displays durations in ticks only.
• --text: Outputs plain text table instead of Discord embed.
• --auto: Automatically selects the optimal voltage tier.
• --bulk: Calculates multiple recipes at once, one per line.

Examples:

• %oc 120 20 (overclock 120 EU/t for 20s)
• %oc 480 30s 100 10 4 2 --iv (480 EU/t, 30s, 4 parallel, 2A, filter to IV)
• %oc ebf 1920 60.3s 3600 5200 4 1 (EBF mode at 1920 EU/t, 3600K recipe heat, 5200K coil heat)
• %oc lcr 2000 15s --rates --subtick (LCR perfect overclocking with rates and subtick)

Other utilities

• convert <value> <from_unit> <to_unit...>: Converts between supported units (metric, imperial, digital, temp) and displays the conversion chain.
• cleanroomcalc / crc <width>x<height>x<length>: Calculates required filter casings, plascrete, and construction blocks for a GregTech cleanroom of given dimensions (e.g. 5x5x5, 9x7x9).
• stoik <reactants> -> <products>: Balances chemical equations and verifies reaction stoichiometry.

10. Maintenance commands

These are higher-privilege operational commands for moderators, admins, or the bot owner:

• reload_commands
• restart
• stop
• admin_eval

Command usage example

Other responses

1. Previews

The bot can respond to message links with an embed of the message and/or the first attachment when present.

If the link's sender cannot read the original message, the bot does not preview it, which prevents leaking private channels.

Previews can also be generated for message links inside tag output.

Example:

2. Sed replace

The following syntax:

sed/regex/replace/flags

can be used to replace a pattern in a previous message with another pattern.

When replying, only the referenced message is considered. Otherwise, the bot searches backward for the first matching message.

Match groups can be referenced in the output using $1, $2, and so on.

Example:

3. Reactions

The bot can react to certain words in a message with configured emojis.

It can also react to mismatched brackets if bracket reaction emojis are configured.

Example:

CLI commands

The CLI is the local command interface for the bot process. When it is enabled, the default prefix is ..

What it is for:

• quick local checks without Discord
• testing tag execution with a fake message
• testing script execution with fake author, guild, channel, or tag data

Available commands:

• .help
• .version
• .uptime
• .clear
• .eval (expression)
• .vm_eval [--debug/-d] (script)
• .execute_tag (name) [args]
• .reload_commands
• .restart
• .stop

General behavior:

• .eval runs a plain JavaScript expression in the local REPL context
• .vm_eval runs a full script through the main script environment
• .execute_tag runs a stored tag, or an emulated tag if you provide emulation values
• .clear clears the terminal
• .stop stops the process

File path input:

• .eval and .vm_eval can take a local file path instead of inline code.
• .execute_tag can also take a file path for an emulated tag body.
• Absolute paths and file:// URLs must exist.
• Relative paths are only treated as files when the target exists. Otherwise the text is treated as normal input.

Common emulation options:

• Message emulation lets you supply author, guild, channel, attachment, and embed data.
• Tag emulation lets you supply tag body, type, language, owner, and tag arguments.
• If you provide tag emulation fields, .execute_tag uses the emulated tag instead of loading one from storage.

Examples:

.eval 1 + 2
.eval ./scripts/test.js
.vm_eval --authorId 123 --authorUsername Alex "return msg.author.username;"
.execute_tag my_tag some args
.execute_tag --tagBody ./tags/example.js --tagType script --tagLanguage js

WebSocket API

The WebSocket API is a command interface for external tools.

Request format:

{
    "id": "optional-request-id",
    "op": "command_name",
    "data": {}
}

Response format:

{
    "id": "same-request-id",
    "op": "command_name",
    "status": "success",
    "data": {}
}

status is either success or error.

Available commands:

• help lists available websocket commands
• version returns the bot version
• uptime returns uptime information
• eval runs JavaScript from data.code and returns the output
• vm_eval runs a script from data.code, with optional debug and emulated message data
• execute_tag runs a stored or emulated tag
• reload_commands, restart, and stop return a success flag and a short message

Validation and rules:

• eval.code and vm_eval.code must be non-empty strings.
• execute_tag.name is required unless you provide a tag object for emulation.
• Group objects such as msg and tag reject unknown properties.
• Wrong types are rejected before the command runs.
• Raw websocket commands do not guess file paths. If you want to run code from a file, read the file yourself and send its contents.

Emulated objects:

• msg can provide fake message, author, guild, and channel data
• tag can provide name, aliasName, body, owner, args, type, and language

Inspector events:

• vm_eval with "debug": true sends an inspector_ready event before the final response.
• The event payload contains the connection details for the debugger.

Helper scripts:

• node ./scripts/websocket-client.js starts an interactive client for the websocket API
• node ./scripts/websocket-client.js help sends one command and prints the JSON reply
• node ./scripts/vscode-debug.js ./path/to/script.js reads a local file and sends it to vm_eval with debugging enabled

The interactive websocket client accepts either full JSON or a lightweight command syntax. Examples:

help
version
eval 1 + 2
vm_eval debug return "hello";
execute_tag my_tag some args
execute_tag my_tag arg1 arg2 authorId="123" guildName="Test Server"

Evaluation API

1. eval / pure JS API

Mirrors the API of Leveret; see Neeve's API documentation.

Main differences:

• util.dumpTags(true) returns a full dump instead of only names
• msg.reply exits the script right away
• util.fetchMessage(ch_id | null, msg_id) allows fetching a single message
• util.fetchMessages accepts message fetch options
• util.findUserById allows fetching a user that is not necessarily in the same server as the bot
• Script HTTP requests (request / fetch) block localhost/private IP addresses and file:// URLs by default unless allowLocalhostRequests or allowFilesystemRequests is enabled in config/config.json

Output rules:

• Strings are sent as-is.
• Objects and arrays are formatted automatically.
• Empty output becomes Cannot send an empty message.

Example:

Inspector behavior:

• When inspector support is enabled without user-facing sessions, the bot exposes one console-driven inspector session at a time.
• When user-facing sessions are enabled, users can start isolated sessions with %eval debug (script) and %t debug (tag_name) [tag_args].
• In user-inspector mode, the bot reply includes the DevTools and VS Code connection details.

2. vm2 / NodeJS API

This backend allows more advanced scripts than the pure JS API, including async flows and importing from the allowed module list.

Tags can use VM2 scripts with:

%t add (name) vm2 (script)

Result rules:

• If you use reply, the script can send directly.
• If you do not use reply, return the final value.

Internal module whitelist:

• assert
• buffer
• crypto
• events
• path
• querystring
• url
• util
• zlib

Global scope:

• tag is available when the script is running from a tag
• msg exposes the emulated or real message
• reply, request, fetchTag, dumpTags, and findUsers are async helpers

Example:

%eval vm2 ```js
return "Amogus";
```

Startup

First-time setup:

1. Install a supported Node.js version.
2. Run npm install.
3. Provide bot credentials with either config/auth.json, config/auth.env, or the LEVERET_TOKEN and LEVERET_OWNER environment variables.
4. Adjust config/config.json and config/reactions.json if needed.

Starting the bot:

npm start

What to expect:

• logs are printed to the console
• logs are also written to the configured log files in logs/
• the CLI starts only if it is enabled in configuration
• the WebSocket server starts only if it is enabled in configuration

Importing

To export the current tag database from the bot, use:

%tag dump full [spaces]

This sends a tags.json file that can be fed to the importer.

Importer usage:

npm run importer -- --help
npm run importer -- --path "path-to-tags.json"
npm run importer -- --path "path-to-tags.db" --owners owner1 owner2
npm run importer -- --path "path-to-tags.json" --amend
npm run importer -- --fix
npm run importer -- --purge-old

What each mode does:

• -p, --path imports a tags file (.json or .db / SQLite database)
• -j, --json-path legacy path alias for importing a JSON tag dump
• -o, --owners filters imported tags to only specified owner IDs (--owners owner1 owner2)
• --amend updates existing tags instead of treating them as conflicts
• --fix checks and repairs database issues
• --purge-old removes old imported tags

Amogus

⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣠⣤⣤⣤⣤⣤⣤⣤⣤⣄⡀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⢀⣴⣿⡿⠛⠉⠙⠛⠛⠛⠛⠻⢿⣿⣷⣤⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⣼⣿⠋⠀⠀⠀⠀⠀⠀⠀⢀⣀⣀⠈⢻⣿⣿⡄⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⣸⣿⡏⠀⠀⠀⣠⣶⣾⣿⣿⣿⠿⠿⠿⢿⣿⣿⣿⣄⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⣿⣿⠁⠀⠀⢰⣿⣿⣯⠁⠀⠀⠀⠀⠀⠀⠀⠈⠙⢿⣷⡄⠀
⠀⠀⣀⣤⣴⣶⣶⣿⡟⠀⠀⠀⢸⣿⣿⣿⣆⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣿⣷⠀
⠀⢰⣿⡟⠋⠉⣹⣿⡇⠀⠀⠀⠘⣿⣿⣿⣿⣷⣦⣤⣤⣤⣶⣶⣶⣶⣿⣿⠀
⠀⢸⣿⡇⠀⠀⣿⣿⡇⠀⠀⠀⠀⠹⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⡿
⠀⣸⣿⡇⠀⠀⣿⣿⡇⠀⠀⠀⠀⠀⠉⠻⠿⣿⣿⣿⣿⡿⠿⠿⠛⢻⣿⡇⠀⠀
⠀⣿⣿⠁⠀⠀⣿⣿⡇⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸⣿⣧⠀⠀
⠀⣿⣿⠀⠀⠀⣿⣿⡇⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸⣿⣿⠀⠀
⠀⣿⣿⠀⠀⠀⣿⣿⡇⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸⣿⣿⠀⠀
⠀⢿⣿⡆⠀⠀⣿⣿⡇⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸⣿⡇⠀⠀
⠀⠸⣿⣧⡀⠀⣿⣿⡇⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣿⣿⠃⠀⠀
⠀⠀⠛⢿⣿⣿⣿⣿⣇⠀⠀⠀⠀⣰⣿⣿⣷⣶⣶⣶⣶⠶⠀⠀⢠⣿⣿⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⣿⣿⠀⠀⠀⠀⠀⣿⣿⡇⠀⣽⣿⡏⠁⠀⠀⠀⢸⣿⡇⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⣿⣿⠀⠀⠀⠀⠀⣿⣿⡇⠀⢹⣿⡆⠀⠀⠀⠀⣸⣿⠇⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⢿⣿⣦⣄⣀⣠⣴⣿⣿⠁⠀⠈⠻⣿⣿⣿⣿⡿⠏⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠈⠛⠻⠿⠿⠿⠿⠋⠁⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀