rpgnlp 0.1.2

A natural language parser for RPG text commands. Extracts actions, subjects, directions, instruments, modifiers, and topics from player input.

rpgnlp

A natural language parser for RPG text commands. Extracts structured data — actions, subjects, directions, instruments, modifiers, and topics — from free-form player input.

Installation

pip install rpgnlp

On first use, the required spaCy language model (en_core_web_md) and NLTK data (averaged_perceptron_tagger_eng, punkt_tab) will be downloaded automatically if not already present.

Quick Start

from rpgnlp import NLPEngine

engine = NLPEngine()

result = engine.run("attack the goblin with a sword")
print(result)
# {
#     "action": "attack",
#     "subject": "goblin",
#     "direction": "",
#     "instrument": [{"name": "sword", "quantity": 1}],
#     "modifiers": [],
#     "topic": ""
# }

Output Fields

Field	Type	Description
action	str	Canonical action (e.g., "attack", "travel", "speak")
subject	str	Target noun phrase (e.g., "goblin", "red door")
direction	str	Cardinal/compound direction (e.g., "north", "south east")
instrument	list[dict]	Instruments with name and quantity
modifiers	list[str]	Adverbs or manner words (e.g., "quickly", "haste")
topic	str	Conversation topic after "about" (e.g., "plan")

Some Supported Actions

Action	Examples
attack	attack, hit, strike, slash, stab, shoot, kick
light_attack	poke, prod, tap, flick, slap, nip
heavy_attack	smash, crush, demolish, obliterate, shatter
travel	go, run, walk, sprint, climb, swim, fly
speak	talk, say, ask, whisper, shout, demand
inspect	look, examine, observe, study, check
search	search, hunt, investigate, scout, seek
defend	block, parry, dodge, guard, protect
use	use, activate, equip, drink, eat, open
take	take, grab, loot, steal, collect
drop	drop, place, put, discard, give
cast	cast, conjure, summon, enchant, heal
sneak	sneak, hide, lurk, prowl, shadow
rest	rest, sleep, camp, meditate, wait
buy	buy, purchase, acquire, obtain, shop
sell	sell, vend, trade, auction, market

Examples

engine = NLPEngine()

# Movement with direction
engine.run("run north east")
# action: "travel", direction: "north east"

# Speech with topic
engine.run("tell the guard about the plan")
# action: "speak", subject: "guard", topic: "plan"

# Multiple instruments with quantities
engine.run("attack the goblin with two daggers and a shield")
# action: "attack", subject: "goblin",
# instrument: [{"name": "daggers", "quantity": 2}, {"name": "shield", "quantity": 1}]

# Modifiers
engine.run("run south with haste")
# action: "travel", direction: "south", modifiers: ["haste"]

# Compound actions (split by "and" or "then")
results = engine.run("go north then attack the troll")
# Returns the last action; compound actions are processed sequentially

Requirements

• Python >= 3.9
• spaCy with en_core_web_md model
• NLTK

License

MIT