NOTES.txt

BUILT IN COMMANDS
- reserved words - can't be used for any other elements
	- make list in hashmap/hashset so can check?
	- does this mean can't be used within commands, or can't be made
		into actions, or both?
	-

- graphs:
	location ALWAYS first in entities file
		paths ALWAYS after locations
	NB: test files will ALWAYS be in a valid format

- command flexibility:
	all commands are case insensitive
		(incl entity names/locos/built in cmds/action triggers)
	MUST CONTAIN TRIGGER && AT LEAST 1 SUBJECT
	MUST cope w/ 'decorative' extra words
		eg cmd 'chop tree with axe' === 'please chop the tree using the axe'
	MUST cope w/ partial cmds
		eg 'unlock door' 'unlock with key' === 'unlock door with key'
			both give enough detail to try match action
	CANNOT use extraneous entities (ie that don't belong)
	action must match ALL subjects given in the command
		ie can't 'open potion with hammer'
		inappropriate subject always === NO match
	ambiguous commands:
		if >1 valid + performable action: do nothing
		send warning
			(eg 'there is more than one 'open' action available - which one would you like to perform?')
	composite commands:
		NOT supported (eg 'get axe and coin' 'get key and open door')
		these are doing more than one thing - not allowed
		but: can use two triggers for same exact action

- error msgs:
	won't be tested for specific messages, just return sth sensible

- they will test by checking game state using cmds (look, inv etc)
	testing will focus on 'fair and reasonable' inputs
	so not so much pvp edges - more flexibility for valid, likely inputs
		user must give enough info to uniquely ID the action
		but still have flexibility about exact expression of cmd
		ie focus on normal inputs, test cases



- CAN CREATE OTHER CONFIG FILES FOR TESTING

- location unique ID MUST follow 'goto'

- location descriptions:
	- ONLY when in that location + do a 'look'
	- don't need to display when in other locos
		and do a look - just the name

- actions and entities:
	only contain alphanumerics
	action triggers can have spaces
	no tests involving numbers

- actions
	- CAN accept two triggers for same action
		eg 'chop with axe to cut down tree'
		both == same action so this is fine
	- BUT NOT FOR BUILT IN CMDS - IF REPEATED
		DON'T ACCEPT? BUT THIS ISNT TESTED
- Built in cmds are:
	predefined, fixed, well understood
	don't need any flexible trigger matching
		=> 'look look' / 'goto cabin goto cabin'
		aren't valid - 2 cmds in a sentence

- entities
	locations / artefacts / furniture/ chars/ player (special kind of char)
	entity names (defined in config) are unique
	only single instance of an entity within game


- actions
	have these elements:
		1+ poss trigger phrases
			any can be used to initiate action
		1+ subject entities to act upon
			ALL must be available b4 doing action
		narration: human-readable explanation of what happened
		(optional) set consumed entities
		(optional) set of produced entities

	being available means: must be in inv / in that location
		== shortcut so can use entities in ur loco w/o picking up
	subjects can be locos/chars/furniture which obv can't be picked up

	trigger phrases NOT unique - can be multiple open actions
		which act on diff entities
	trigger phrases cannot contain entity names
	(BUT overall cmds obv can)

	when received:
	- try match trigger phrase
		remember trigger phrase can be same for diff actions
		eg 'open trapdoor' 'open cellar' 'open treasure chest' etc
		so each individ instance of an action is a GameAction
		store in HashMap <String, HashSet<GameAction>>
		ie < (trigger phrase) <((set of actions))>>
			eg <'open', <'open door', 'open cellar'>>

		so: 'cut' and 'chop' trigger maps can
		both contain set containing 'chop down tree' action


DESIGN:
- get user input, tokenise(just split by whitespace?), match trigger phrase
	- do i need to check for valid inputs?
	- how does this actually work
	- maybe just make it work for valid ones first ?

- action class:
	vars for triggers/subjects/consumed/produced/narration (/location?)
	method to check if valid based on current game state
	method to execute action
	use Factory pattern?




- locations:
	paths to other locations / chars / artefacts / furniture


- storeroom
	not in game world - no paths to/from it
	contains all entities with no starting loco
	these won't enter game until an action puts them in a loco



- multiple players:
	ALL before ':' == name
	ALL after ':' == command
	valid names: upper/lower chars, spaces, apostrophes, hyphens
	NO registration process for new players
		ie if new user suddenly enters:
		create them + place in start location
	command MUST be applied to right player
	==> some game state must be stored separately for players
		eg current location, inventory - what else?
	'look' command: SHOULD show other players in the description of a loco
	(but not yourself)



- 'cut tree chop down tree' - this is allowed as same action