casebench

Test cases go straight to the board. The runtime never sees them. Reaches the board and nothing else.

commands

casebench put FOLDER --player HANDLE
casebench note CASE "text" [--item ITEM] --player HANDLE
casebench note CASE -f note.md [--item ITEM] --player HANDLE
casebench lock CASE [--answer A ...] [--rest 3/60] [--text locked.] --player HANDLE
casebench unlock CASE --player HANDLE
casebench list [--player HANDLE]
casebench pull CASE --player HANDLE

every command that writes takes --player (more than once is fine) or --all, for every board. list without --player lists every board.

• put puts the case on the board, or replaces its title and items if it's there already. notes and a lock stay. it won't put a case over one the runtime put there.
• note writes under the case's title, or on an item with --item. -f - reads stdin.
• lock takes one --answer per line of the lock, or asks for them if there's none. a|b takes either. lines are checked together, so a wrong try never says which line was right. only salted hashes are stored.
• --rest 3/60: after 3 wrong tries in a row the lock takes nothing for 60 minutes.
• pull only pulls what the bench put there.

configuration

• CASEBENCH_DSN: the hub's postgres.
• CASEBENCH_REST: the default for --rest. 3/60 if unset.

schema.sql has the board tables as the bench expects them, for a local database.

a case folder

case.yaml
items/ticket.md
items/umbrella.md
items/umbrella.svg

id: test-lost-umbrella
title: The lost umbrella
items:
  - id: ticket
    title: Cloakroom ticket
    text: items/ticket.md
  - id: umbrella
    title: The umbrella
    text: items/umbrella.md
    images: [items/umbrella.svg]

ids are lowercase, digits and dashes. images: png, jpg, webp, svg. examples/lost-umbrella is a whole one.

tests

python -m unittest

against the memstore. no database needed.