CI / test (pull_request) Canceled after 0s
- analysis/tree/build_tree.py -> data/public/tree.json, analysis/reports/TREE.md: every screen the bot has seen, every option, where each led (fixed or dice, with ranges), best known picks per colour chosen cautiously, where any picks end, and the untested options. solver.js screenPick(): least-tried and Thompson picks from the tree. - The bot logs a pick it then overrode as a step: 37 runs were coded with choices they never made. Importer and graph.js now use the played steps only. - runs_clean.csv: the Unofficial Dragon Register's 288 bonded dragons are in the master list as outcome-only rows, flagged (no analysis uses them). - Cycle, refresh and the bot kit rebuild and carry the tree; tests check every generated report is current. Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_014jqCC2LZ4x1WHQ1V57nTsv
88 lines
4.3 KiB
Python
88 lines
4.3 KiB
Python
"""The screen tree (analysis/tree/build_tree.py) on a small game whose answers are known."""
|
|
import sys
|
|
import unittest
|
|
from pathlib import Path
|
|
|
|
ROOT = Path(__file__).resolve().parents[2]
|
|
sys.path.insert(0, str(ROOT / "analysis/tree"))
|
|
import build_tree as B # noqa: E402
|
|
|
|
SCENES = {"start": {"label": "Start", "stage": "S", "options": [{"id": "a", "text": "Left"}, {"id": "b", "text": "Right"}, {"id": "c", "text": "Never tried"}]},
|
|
"meadow": {"label": "Meadow", "stage": "M", "options": [{"id": "x", "text": "Approach"}]},
|
|
"blue": {"label": "Blue", "stage": "B", "options": [{"id": "g", "text": "Hold its gaze"}, {"id": "l", "text": "Lower eyes"}]}}
|
|
|
|
|
|
def step(scene, option, to=None, stage="S"):
|
|
return {"scene": scene, "option": option, "optionText": option, "to": to, "stage": stage}
|
|
|
|
|
|
def run(i, steps, kind, colour=None):
|
|
return {"id": f"r{i}", "steps": steps, "outcome": {"type": kind, "colour": colour}}
|
|
|
|
|
|
def world():
|
|
T = []
|
|
for i in range(100): # Left: the meadow; a Blue waits there 1 time in 10; holding its gaze bonds 8 times in 10
|
|
s = [step("start", "a", "meadow")]
|
|
if i % 10 == 0:
|
|
s += [step("meadow", "x", "blue"), step("blue", "g", "BURNED" if i in (0, 50) else "BONDED:blue")]
|
|
T.append(run(i, s, "BURNED" if i in (0, 50) else "BONDED", None if i in (0, 50) else "blue"))
|
|
else:
|
|
s += [step("meadow", "x", "BURNED", stage="INTENTIONAL_BURN")]
|
|
T.append(run(i, s, "BURNED"))
|
|
T.append(run(200, [step("start", "b", "BURNED")], "BURNED")) # Right: died, once
|
|
T.append(run(201, [step("start", "b", "BURNED"), step("start", "a", "meadow")], "ABORTED")) # overridden pick, then aborted
|
|
T.append(run(202, [step("blue", "l", "BONDED:blue")], "BONDED", "blue")) # one lucky run with lowered eyes
|
|
return T
|
|
|
|
|
|
class Tree(unittest.TestCase):
|
|
@classmethod
|
|
def setUpClass(cls):
|
|
cls.t = B.build(world(), SCENES)
|
|
|
|
def test_counts_kinds_and_overridden_picks(self):
|
|
S = self.t["screens"]
|
|
a = S["start"]["options"]["a"]
|
|
self.assertEqual((a["n"], a["kind"], list(a["to"])), (101, "fixed", ["meadow"])) # 100 runs + the aborted one
|
|
self.assertLess(a["missed"], 0.01)
|
|
self.assertEqual(S["start"]["options"]["b"]["n"], 1) # the overridden pick is not counted
|
|
self.assertEqual(self.t["builtFrom"]["overridden"], 1)
|
|
self.assertEqual(S["start"]["options"]["c"]["kind"], "untested")
|
|
x = S["meadow"]["options"]["x"]
|
|
self.assertEqual(x["kind"], "dice")
|
|
self.assertAlmostEqual(x["to"]["blue"]["p"], 0.1)
|
|
self.assertEqual(x["deliberate"], 90) # deliberate burns are real results
|
|
self.assertTrue(x["to"]["blue"]["lo"] < 0.1 < x["to"]["blue"]["hi"])
|
|
g = S["blue"]["options"]["g"]
|
|
self.assertEqual((g["to"]["BONDED:Blue"]["n"], g["to"]["DIED"]["n"]), (8, 2))
|
|
|
|
def test_best_known_chance_is_chosen_cautiously(self):
|
|
best = self.t["best"]["Blue"]
|
|
# one lucky run with lowered eyes is 100% as observed, but holding its gaze (8 of 10) is the cautious pick
|
|
self.assertEqual([k["option"] for k in best["picks"]], ["a", "x", "g"])
|
|
self.assertAlmostEqual(best["p"], 0.1 * 0.8)
|
|
self.assertLess(best["pCautious"], best["p"])
|
|
self.assertAlmostEqual(best["ends"]["BONDED:Blue"], 0.08)
|
|
self.assertAlmostEqual(best["ends"]["DIED"], 0.92)
|
|
self.assertEqual(self.t["best"]["Black"]["p"], 0.0)
|
|
|
|
def test_outcomes_of_any_picks(self):
|
|
self.assertEqual(B.outcomes(self.t, {"start": "b"}), {"DIED": 1.0})
|
|
self.assertEqual(B.outcomes(self.t, {"start": "c"}), {"not mapped": 1.0})
|
|
ends = B.outcomes(self.t, {"start": "a", "blue": "l"})
|
|
self.assertAlmostEqual(ends["BONDED:Blue"], 0.1)
|
|
|
|
def test_coverage_lists_the_gaps(self):
|
|
c = self.t["coverage"]
|
|
self.assertEqual((c["screens"], c["options"], c["tested"]), (3, 6, 5))
|
|
self.assertEqual([u["option"] for u in c["untested"]], ["c"])
|
|
self.assertEqual(c["endings"], ["BONDED:Blue", "DIED"])
|
|
|
|
def test_published_tree_is_up_to_date(self):
|
|
self.assertEqual(B.main(["--check"]), 0)
|
|
|
|
|
|
if __name__ == "__main__":
|
|
unittest.main()
|