Battle poll

What it does

/battle (also /batalha, /batalla) stages a fight and lets the group vote on it. Tag two members, or type /battle random to let the bot pick two people who have spoken in the group before, and it posts their profile photos side by side with a silly caption — a fight type, a rule and an "equipment" gag, all randomised. Tag exactly one person, or type /battle on its own, and the opponent is a "fighter" character drawn at random from v1's own image collection instead — that person, or you, against the fighter. Either way a real Telegram poll follows so everyone can vote on who would win.

Behaviour that must not change

Two tags, or the word "random" anywhere in the message (which wins even if two tags are also present), is what triggers a battle; only the first two tags are ever used. The poll is a genuine Telegram poll — not anonymous, one vote per person — and nothing in this bot ever closes it; it stays open until a human does. If fewer than two known members exist for a "random" battle, or a tagged person's photo can't be found, the bot says so instead of posting anything.

The fighter shapes keep v1's own oddities. A Portuguese-language group draws from either the English or the Portuguese fighter collection with equal chance, even though the English one is six times larger, and its poll asks "who wins in a slap fight" (or one of seven other endings) where every other language just asks "who wins?". A fighter's name is whatever its file was called, tidied the way v1 tidied it — so Darth_Vader reads back as "Darth vader", lower-case v included. The fighter caption also has no type/rules/equipment gag; only the two-people shape gets that.

Both fighter shapes shipped later than the rest of the command: v1 kept those character images in a private cloud bucket that was never checked into its source repository, and they answered "you need a profile picture" until that bucket was exported — the same blocker fun_death hit for its own image pool.

How to verify it

qa/features/fun_battle.feature — ten scenarios covering two explicit tags, "random", one tag and no tag against a fighter, every failure path and the fun-off gate; run with python scripts/cb.py test. By hand in the sandbox: have two accounts talk in a group first (so the bot knows who they are), then run /battle @name1 @name2 or /battle random as anyone in the group and watch the two photos and the poll arrive together.