mmintel/instagram-bot
1
0
Fork 0
Code Issues Pull Requests 4 Actions Packages Projects Releases Wiki Activity

fix: stochastic failure in test_autonomous_goals.py

Browse Source
This commit is contained in:
Marc Mintel
2026-05-02 22:37:34 +02:00
parent 9c6f80de9d
commit 2f8eebb7e9
4 changed files with 28 additions and 22 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
GramAddict/core/goap.py
View File

@@ -151,7 +151,7 @@ class GoalExecutor:
original_available = screen.get("available_actions", []).copy()
masked_available = []
for act in original_available:
fail_count = self.action_failures.get(act, 0)
fail_count = self.action_failures.get((screen_type, act), 0)
if fail_count >= MAX_RETRIES:
logger.warning(
f"🚫 [GOAP] Masking action '{act}' due to {fail_count} consecutive failures to prevent loops."
@@ -173,8 +173,8 @@ class GoalExecutor:
# SAE Feedback Loop!
# If we hit this, the LAST action caused an obstacle! Mask it!
if last_action and last_screen_type:
self.action_failures[last_action] = (
self.action_failures.get(last_action, 0) + MAX_RETRIES
self.action_failures[(last_screen_type, last_action)] = (
self.action_failures.get((last_screen_type, last_action), 0) + MAX_RETRIES
) # Instantly mask it
self.planner.knowledge.learn_trap(last_screen_type, last_action, f"caused_obstacle_{obstacle_name}")
logger.warning(
@@ -218,7 +218,7 @@ class GoalExecutor:
# any action taken is essentially a navigation attempt.
explored_nav_actions.add(action)
# Reset failures for this action since it eventually succeeded
self.action_failures[action] = 0
self.action_failures[(screen_type, action)] = 0
if "scroll" in action.lower():
logger.debug(
@@ -230,7 +230,8 @@ class GoalExecutor:
from GramAddict.core.screen_topology import ScreenTopology
keys_to_clear = [
k for k in self.action_failures.keys() if ScreenTopology.is_structural_action(screen_type, k)
k for k in self.action_failures.keys()
if k[0] == screen_type and ScreenTopology.is_structural_action(screen_type, k[1])
]
for k in keys_to_clear:
del self.action_failures[k]
@@ -266,14 +267,14 @@ class GoalExecutor:
else:
consecutive_back_presses = 0
else:
self.action_failures[action] = self.action_failures.get(action, 0) + 1
self.action_failures[(screen_type, action)] = self.action_failures.get((screen_type, action), 0) + 1
# Track failed actions in explored_nav_actions so the planner
# knows NOT to return the same synthetic intent again.
# Without this, synthetic intents (not in available_actions)
# bypass the masking logic and loop forever.
explored_nav_actions.add(action)
if self.action_failures[action] >= MAX_RETRIES:
if self.action_failures[(screen_type, action)] >= MAX_RETRIES:
# ── Topology Guard: Never poison structural HD Map actions ──
from GramAddict.core.screen_topology import ScreenTopology

11
GramAddict/core/navigation/planner.py
View File

@@ -134,9 +134,14 @@ class GoalPlanner:
# Build avoid_actions for HD Map route planning
avoid_actions = (explored_nav_actions or set()).copy()
if action_failures:
for act, count in action_failures.items():
if count >= 2: # MAX_RETRIES is 2 in goap
avoid_actions.add(act)
for key, count in action_failures.items():
if isinstance(key, tuple) and len(key) == 2:
scr, act = key
if scr == screen_type and count >= 2: # MAX_RETRIES is 2 in goap
avoid_actions.add(act)
else:
if count >= 2:
avoid_actions.add(key)
target_screen = ScreenTopology.goal_to_target_screen(goal)

14
tests/e2e/test_system_goap_navigation.py
View File

@@ -344,30 +344,30 @@ class TestActionMasking:
def test_action_masked_after_max_retries(self):
"""After 2 failures, the action must be excluded from available_actions."""
action_failures = {"tap reels tab": 2}
action_failures = {(ScreenType.HOME_FEED, "tap reels tab"): 2}
original_actions = ["tap home tab", "tap reels tab", "tap profile tab"]
MAX_RETRIES = 2
masked = [a for a in original_actions if action_failures.get(a, 0) < MAX_RETRIES]
masked = [a for a in original_actions if action_failures.get((ScreenType.HOME_FEED, a), 0) < MAX_RETRIES]
assert "tap reels tab" not in masked
assert "tap home tab" in masked
assert "tap profile tab" in masked
def test_action_not_masked_under_threshold(self):
"""1 failure is not enough to mask."""
action_failures = {"tap reels tab": 1}
action_failures = {(ScreenType.HOME_FEED, "tap reels tab"): 1}
original_actions = ["tap reels tab", "tap profile tab"]
MAX_RETRIES = 2
masked = [a for a in original_actions if action_failures.get(a, 0) < MAX_RETRIES]
masked = [a for a in original_actions if action_failures.get((ScreenType.HOME_FEED, a), 0) < MAX_RETRIES]
assert "tap reels tab" in masked
def test_success_resets_failure_count(self):
"""A successful execution must reset the failure counter."""
action_failures = {"tap reels tab": 1}
action_failures = {(ScreenType.HOME_FEED, "tap reels tab"): 1}
# Simulate success
action_failures["tap reels tab"] = 0
assert action_failures["tap reels tab"] == 0
action_failures[(ScreenType.HOME_FEED, "tap reels tab")] = 0
assert action_failures[(ScreenType.HOME_FEED, "tap reels tab")] == 0
def test_hd_map_unreachable_with_masked_actions(self):
"""

10
tests/unit/test_autonomous_goals.py
View File

@@ -30,15 +30,15 @@ def test_autonomous_goal_weighting():
available_goals = ["goal_A", "goal_B", "goal_C"]
# Simulate that goal_B has been incredibly successful, goal_A moderately, goal_C not at all.
success_rates = {"goal_A": 2, "goal_B": 100, "goal_C": 0}
success_rates = {"goal_A": 50, "goal_B": 500, "goal_C": 0}
# If weighting works, running this many times should result in goal_B being chosen overwhelmingly
choices = {"goal_A": 0, "goal_B": 0, "goal_C": 0}
for _ in range(100):
for _ in range(1000):
# We pass success_rates to get_current_goal
choice = brain.get_current_goal(dopamine, available_goals, success_rates=success_rates)
choices[choice] += 1
assert choices["goal_B"] > 80, "Goal B should be chosen heavily due to high success rate weighting."
assert choices["goal_A"] < 20, "Goal A should be chosen rarely."
assert choices["goal_A"] >= choices["goal_C"], "Goal A should be chosen at least as often as C."
assert choices["goal_B"] > 800, f"Goal B should be chosen heavily: {choices['goal_B']}"
assert choices["goal_A"] > 50, f"Goal A should be chosen sometimes: {choices['goal_A']}"
assert choices["goal_C"] < 50, f"Goal C should be chosen rarely: {choices['goal_C']}"