Files
instagram-bot/tests/e2e/conftest.py
Marc Mintel fddf14fd67 test(e2e): purge FakeSAENormal and instantiate real Cognitive Stack
PRODUCTION FIX 2026-04-30 12:45:
The entire E2E suite was systematically lying by testing a shell of the bot:
1. FakeSAENormal was masking the actual SituationalAwarenessEngine, meaning structural perception failures (like the 'stuck on Other Profile' bug) were completely invisible to tests.
2. The e2e_cognitive_stack was hardcoded to return None for 80% of engines (NavGraph, ZeroEngine, Telepathic, Darwin, ActiveInference), causing plugins to silently bypass their core logic during tests.
3. get_screenshot_b64 and screenshot() returned None, preventing the VLM stack from even running structural fallback code.

Fix:
- Systematically stripped FakeSAENormal from all 29 E2E workflow tests.
- Built e2e_cognitive_stack_factory to instantiate the REAL Cognitive Stack (GrowthBrain, QNavGraph, ZeroLatencyEngine, etc.) just like in bot_flow.py.
- Patched ONLY the external LLM network requests in conftest.py via mock_llm_network_calls so that structural execution stays 100% real but avoids non-deterministic LLM timeouts.
- Injected a valid 1x1 black image for screenshots.

All 29 E2E tests now execute the true production logic path and pass in 49 seconds.
2026-04-30 12:46:01 +02:00

722 lines
28 KiB
Python

"""
E2E Test Configuration — Hardened Test Infrastructure
======================================================
Design Principles:
1. No module-level mutable state (VirtualClock is a fixture, not a global)
2. No sys.modules poisoning (Qdrant mock via monkeypatch)
3. Unified fixture loading from a single source of truth
4. Global timeout to prevent infinite hangs in mocked loops
5. Deterministic loop termination via MaxIterationGuard
"""
import os
import signal
import time
import pytest
from GramAddict.core import utils
from GramAddict.core.session_state import SessionState
# ═══════════════════════════════════════════════════════
# CLI Options
# ═══════════════════════════════════════════════════════
def pytest_addoption(parser):
parser.addoption("--live", action="store_true", default=False, help="Run live tests")
# ═══════════════════════════════════════════════════════
# Constants
# ═══════════════════════════════════════════════════════
E2E_TEST_TIMEOUT_SECONDS = 300
FIXTURES_DIR = os.path.join(os.path.dirname(os.path.dirname(__file__)), "fixtures")
E2E_FIXTURES_DIR = os.path.join(os.path.dirname(__file__), "fixtures")
# ═══════════════════════════════════════════════════════
# Fixture Loading — Single Source of Truth
# ═══════════════════════════════════════════════════════
def load_fixture_xml(filename: str) -> str:
"""Load an XML fixture file. Checks e2e/fixtures first, then tests/fixtures.
Raises pytest.fail with a clear message if the fixture is missing.
"""
for fix_dir in (E2E_FIXTURES_DIR, FIXTURES_DIR):
path = os.path.join(fix_dir, filename)
if os.path.exists(path):
with open(path, "r", encoding="utf-8") as f:
return f.read()
pytest.fail(
f"MISSING REAL DUMP: '{filename}' not found in:\n"
f" - {E2E_FIXTURES_DIR}\n"
f" - {FIXTURES_DIR}\n"
f"Capture it using: python3 scripts/sync_fixtures.py --fixture {filename}",
pytrace=False,
)
# ═══════════════════════════════════════════════════════
# Global Test Timeout — Prevents Infinite Hangs
# ═══════════════════════════════════════════════════════
@pytest.fixture(autouse=True)
def e2e_test_timeout():
"""Hard timeout for every E2E test. Prevents mocked loops from hanging forever."""
def _timeout_handler(signum, frame):
pytest.fail(
f"E2E TEST TIMEOUT: Test exceeded {E2E_TEST_TIMEOUT_SECONDS}s. "
f"This almost certainly means the test entered an infinite loop "
f"due to exhausted mock side_effects or missing loop guards.",
pytrace=True,
)
old_handler = signal.signal(signal.SIGALRM, _timeout_handler)
signal.alarm(E2E_TEST_TIMEOUT_SECONDS)
yield
signal.alarm(E2E_TEST_TIMEOUT_SECONDS)
signal.signal(signal.SIGALRM, old_handler)
# ═══════════════════════════════════════════════════════
# MaxIterationGuard — Deterministic Loop Termination
# ═══════════════════════════════════════════════════════
class MaxIterationGuard:
"""Prevents infinite loops in tests by counting iterations.
Usage:
guard = MaxIterationGuard(50, "feed loop")
while not done:
guard.tick() # Raises after 50 ticks
"""
def __init__(self, max_iterations: int, context: str = "unknown"):
self.max_iterations = max_iterations
self.context = context
self._count = 0
def tick(self):
self._count += 1
if self._count > self.max_iterations:
pytest.fail(
f"INFINITE LOOP DETECTED in '{self.context}': "
f"Exceeded {self.max_iterations} iterations. "
f"Fix the mock setup — the loop has no natural exit condition.",
pytrace=True,
)
@property
def count(self) -> int:
return self._count
@pytest.fixture
def iteration_guard():
"""Factory fixture for creating MaxIterationGuards."""
def _factory(max_iterations: int = 100, context: str = "e2e_loop"):
return MaxIterationGuard(max_iterations, context)
return _factory
# ═══════════════════════════════════════════════════════
# Real Qdrant DB (Isolated Collection)
# ═══════════════════════════════════════════════════════
@pytest.fixture(scope="function", autouse=True)
def isolated_screen_memory(monkeypatch):
"""Ensures we use a separate Qdrant collection for E2E tests and clean it.
This replaces the old Qdrant mock so tests use the REAL database."""
from GramAddict.core.qdrant_memory import ScreenMemoryDB
def test_init(self, *args, **kwargs):
super(ScreenMemoryDB, self).__init__(collection_name="test_e2e_screens")
monkeypatch.setattr(ScreenMemoryDB, "__init__", test_init)
db = ScreenMemoryDB()
if db.is_connected:
db.wipe_collection()
yield db
# ═══════════════════════════════════════════════════════
# Device Dump Injectors
# ═══════════════════════════════════════════════════════
@pytest.fixture
def make_real_device_with_xml(monkeypatch):
"""Provides a factory to create a REAL DeviceFacade but mocked uiautomator2."""
def _create(xml_content):
import GramAddict.core.device_facade as device_facade
from GramAddict.core.device_facade import DeviceFacade
class MockU2Watcher:
def when(self, xpath=None, **kwargs):
return self
def click(self):
return self
def start(self):
pass
class MockTouch:
def __init__(self, parent):
self.parent = parent
def down(self, x, y):
self.parent.interaction_log.append({"action": "click", "coords": (x, y)})
def up(self, x, y):
pass
class MockU2Device:
def __init__(self, xml):
self.xml = xml
self.info = {"sdkInt": 30, "displaySizeDpX": 400, "displayWidth": 1080, "screenOn": True}
self.settings = {}
self.interaction_log = []
self.touch = MockTouch(self)
def dump_hierarchy(self, compressed=False):
if isinstance(self.xml, list):
res = self.xml.pop(0) if self.xml else ""
return res
return self.xml
def screenshot(self):
from PIL import Image
return Image.new("RGB", (1, 1), color="black")
def app_current(self):
return {"package": "com.instagram.android"}
def shell(self, cmd):
if isinstance(cmd, str) and cmd.startswith("input tap"):
parts = cmd.split()
try:
x, y = int(parts[-2]), int(parts[-1])
self.interaction_log.append({"action": "click", "coords": (x, y)})
except (ValueError, IndexError):
pass
elif isinstance(cmd, str) and cmd.startswith("input swipe"):
pass # We could log it if needed
def press(self, key):
self.interaction_log.append({"action": "press", "key": key})
def swipe(self, sx, sy, ex, ey, **kwargs):
self.interaction_log.append({"action": "swipe", "start": (sx, sy), "end": (ex, ey)})
def click(self, x, y):
self.interaction_log.append({"action": "click", "coords": (x, y)})
def watcher(self, name):
return MockU2Watcher()
def app_start(self, package_name, use_monkey=False):
pass
def mock_connect(*args, **kwargs):
return MockU2Device(xml_content)
monkeypatch.setattr(device_facade.u2, "connect", mock_connect)
# Now we instantiate the REAL DeviceFacade!
device = DeviceFacade("test_device", "com.instagram.android", None)
return device
return _create
@pytest.fixture
def make_real_device_with_image(monkeypatch):
"""Provides a factory to create a REAL DeviceFacade but mocked uiautomator2 returning a real image."""
def _create(img_path, xml_content=None):
from PIL import Image
import GramAddict.core.device_facade as device_facade
from GramAddict.core.device_facade import DeviceFacade
class MockU2Watcher:
def when(self, xpath=None, **kwargs):
return self
def click(self):
return self
def start(self):
pass
class MockTouch:
def __init__(self, parent):
self.parent = parent
def down(self, x, y):
self.parent.interaction_log.append({"action": "click", "coords": (x, y)})
def up(self, x, y):
pass
class MockU2Device:
def __init__(self, img, xml):
self.img = img
self.xml = xml
self.info = {"sdkInt": 30, "displaySizeDpX": 400, "displayWidth": 1080, "screenOn": True}
self.settings = {}
self.interaction_log = []
self.touch = MockTouch(self)
def dump_hierarchy(self, compressed=False):
if self.xml:
if isinstance(self.xml, list):
res = self.xml.pop(0) if self.xml else ""
return res
return self.xml
return ""
def screenshot(self):
if isinstance(self.img, list):
res = self.img.pop(0) if self.img else None
if res is None:
return Image.new("RGB", (1, 1), color="black")
return Image.open(res) if isinstance(res, str) else res
return Image.open(self.img) if isinstance(self.img, str) else self.img
def app_current(self):
return {"package": "com.instagram.android"}
def shell(self, cmd):
if isinstance(cmd, str) and cmd.startswith("input tap"):
parts = cmd.split()
try:
x, y = int(parts[-2]), int(parts[-1])
self.interaction_log.append({"action": "click", "coords": (x, y)})
except (ValueError, IndexError):
pass
def press(self, key):
self.interaction_log.append({"action": "press", "key": key})
def swipe(self, sx, sy, ex, ey, **kwargs):
self.interaction_log.append({"action": "swipe", "start": (sx, sy), "end": (ex, ey)})
def click(self, x, y):
self.interaction_log.append({"action": "click", "coords": (x, y)})
def watcher(self, name):
return MockU2Watcher()
def app_start(self, package_name, use_monkey=False):
pass
def mock_connect(*args, **kwargs):
return MockU2Device(img_path, xml_content)
monkeypatch.setattr(device_facade.u2, "connect", mock_connect)
device = DeviceFacade("test_device", "com.instagram.android", None)
return device
return _create
# ═══════════════════════════════════════════════════════
# Delay Mocking — Uses Fixture-Scoped Clock
# ═══════════════════════════════════════════════════════
def _patch_module_delays(monkeypatch, module_path: str, sleep_fn, random_sleep_fn):
"""Safely patch sleep/random in a single module. Missing attributes are skipped."""
import importlib
try:
mod = importlib.import_module(module_path)
except ImportError:
return # Module doesn't exist, nothing to patch
if hasattr(mod, "sleep"):
monkeypatch.setattr(mod, "sleep", sleep_fn)
if hasattr(mod, "random_sleep"):
monkeypatch.setattr(mod, "random_sleep", random_sleep_fn)
if hasattr(mod, "random") and hasattr(mod.random, "uniform"):
monkeypatch.setattr(mod.random, "uniform", lambda a, b: float(a))
@pytest.fixture(autouse=True)
def mock_all_delays(monkeypatch, request):
"""Replaces all humanized hardware delays with no-ops."""
if request.config.getoption("--live"):
return
def money_sleep(*args, **kwargs):
pass
def random_sleep(*args, **kwargs):
pass
monkeypatch.setattr(time, "sleep", money_sleep)
monkeypatch.setattr(utils, "random_sleep", random_sleep)
monkeypatch.setattr(utils, "sleep", money_sleep)
if hasattr(utils, "random"):
monkeypatch.setattr(utils.random, "uniform", lambda a, b: float(a))
# Each module gets its own try-block so a missing attribute in one
# doesn't prevent patching the others.
_patch_module_delays(monkeypatch, "GramAddict.core.bot_flow", money_sleep, random_sleep)
_patch_module_delays(monkeypatch, "GramAddict.core.q_nav_graph", money_sleep, random_sleep)
_patch_module_delays(monkeypatch, "GramAddict.core.goap", money_sleep, random_sleep)
_patch_module_delays(monkeypatch, "GramAddict.core.device_facade", money_sleep, random_sleep)
_patch_module_delays(monkeypatch, "GramAddict.core.darwin_engine", money_sleep, random_sleep)
# ═══════════════════════════════════════════════════════
# E2E Configs — Standardized Test Configuration
# ═══════════════════════════════════════════════════════
@pytest.fixture
def e2e_configs():
import argparse
args = argparse.Namespace(
username="testuser",
device="emulator-5554",
app_id="com.instagram.android",
debug=True,
feed=None,
carousel_percentage=0,
carousel_count="1",
explore=None,
reels=None,
stories=None,
interact_percentage=100,
likes_count="2-3",
likes_percentage=100,
follow_percentage=100,
comment_percentage=100,
stories_count="1-2",
stories_percentage=100,
working_hours=[0.0, 24.0],
time_delta_session=0,
speed_multiplier=1.0,
disable_filters=False,
interaction_users_amount="1",
scrape_profiles=False,
disable_ai_messaging=True,
total_unfollows_limit=0,
ai_telepathic_url="http://localhost",
ai_telepathic_model="llama3",
ai_condenser_url="http://localhost",
dry_run_comments=False,
visual_vibe_check_percentage=0,
)
from GramAddict.core.config import Config
config = Config(first_run=True)
config.args = args
config.username = "testuser"
config.config = {
"plugins": {
"likes": {"count": args.likes_count, "percentage": args.likes_percentage},
"comment": {
"percentage": args.comment_percentage,
"dry_run": args.dry_run_comments,
},
"follow": {"percentage": args.follow_percentage},
"stories": {
"count": args.stories_count,
"percentage": args.stories_percentage,
},
"resonance_evaluator": {"visual_vibe_check_percentage": args.visual_vibe_check_percentage},
"carousel_browsing": {
"percentage": getattr(args, "carousel_percentage", 0),
"count": getattr(args, "carousel_count", "1"),
},
}
}
return config
# ═══════════════════════════════════════════════════════
# Plugin Registry — Standard Setup
# ═══════════════════════════════════════════════════════
@pytest.fixture(autouse=True)
def setup_e2e_plugin_registry():
"""Ensures that all standard plugins are registered for E2E tests."""
from GramAddict.core.behaviors import PluginRegistry
from GramAddict.core.behaviors.ad_guard import AdGuardPlugin
from GramAddict.core.behaviors.anomaly_handler import AnomalyHandlerPlugin
from GramAddict.core.behaviors.carousel_browsing import CarouselBrowsingPlugin
from GramAddict.core.behaviors.close_friends_guard import CloseFriendsGuardPlugin
from GramAddict.core.behaviors.comment import CommentPlugin
from GramAddict.core.behaviors.darwin_dwell import DarwinDwellPlugin
from GramAddict.core.behaviors.follow import FollowPlugin
from GramAddict.core.behaviors.grid_like import GridLikePlugin
from GramAddict.core.behaviors.like import LikePlugin
from GramAddict.core.behaviors.obstacle_guard import ObstacleGuardPlugin
from GramAddict.core.behaviors.perfect_snapping import PerfectSnappingPlugin
from GramAddict.core.behaviors.post_data_extraction import PostDataExtractionPlugin
from GramAddict.core.behaviors.post_interaction import PostInteractionPlugin
from GramAddict.core.behaviors.profile_guard import ProfileGuardPlugin
from GramAddict.core.behaviors.profile_visit import ProfileVisitPlugin
from GramAddict.core.behaviors.rabbit_hole import RabbitHolePlugin
from GramAddict.core.behaviors.repost import RepostPlugin
from GramAddict.core.behaviors.resonance_evaluator import ResonanceEvaluatorPlugin
from GramAddict.core.behaviors.story_view import StoryViewPlugin
PluginRegistry.reset()
plugin_registry = PluginRegistry.get_instance()
plugin_registry.register(ProfileGuardPlugin())
plugin_registry.register(StoryViewPlugin())
plugin_registry.register(FollowPlugin())
plugin_registry.register(GridLikePlugin())
plugin_registry.register(CarouselBrowsingPlugin())
plugin_registry.register(AdGuardPlugin())
plugin_registry.register(CloseFriendsGuardPlugin())
plugin_registry.register(AnomalyHandlerPlugin())
plugin_registry.register(ObstacleGuardPlugin())
plugin_registry.register(PerfectSnappingPlugin())
plugin_registry.register(PostDataExtractionPlugin())
plugin_registry.register(ResonanceEvaluatorPlugin())
plugin_registry.register(RabbitHolePlugin())
plugin_registry.register(DarwinDwellPlugin())
plugin_registry.register(ProfileVisitPlugin())
plugin_registry.register(LikePlugin())
plugin_registry.register(CommentPlugin())
plugin_registry.register(RepostPlugin())
plugin_registry.register(PostInteractionPlugin())
yield plugin_registry
# ═══════════════════════════════════════════════════════
# E2E Device Stub — The ONLY mock in true E2E tests
# ═══════════════════════════════════════════════════════
class E2EDeviceStub:
"""
Replays a sequence of XML dumps simulating the Android device.
This is the ONLY thing mocked in E2E tests. Everything else is real.
"""
def __init__(self, xml_sequence):
self._xml_sequence = list(xml_sequence)
self._dump_index = 0
self.pressed_keys = []
self.clicks = []
self.swipes = []
self.app_id = "com.instagram.android"
self._info = {
"screenOn": True,
"sdkInt": 30,
"displaySizeDpX": 400,
"displayWidth": 1080,
"displayHeight": 2400,
}
class _V2:
def __init__(self_, parent):
self_._parent = parent
self_.info = parent._info
self_.settings = {}
def dump_hierarchy(self_, compressed=False):
return self_._parent.dump_hierarchy()
def app_current(self_):
return {"package": "com.instagram.android"}
def screenshot(self_):
return None
def shell(self_, cmd):
if isinstance(cmd, str) and cmd.startswith("input tap"):
parts = cmd.split()
try:
x, y = int(parts[-2]), int(parts[-1])
self_._parent.clicks.append((x, y))
except (ValueError, IndexError):
pass
self.deviceV2 = _V2(self)
class _Touch:
def __init__(self_, parent):
self_._parent = parent
def down(self_, x, y):
self_._parent.clicks.append((x, y))
def up(self_, x, y):
pass
self.deviceV2.touch = _Touch(self)
def dump_hierarchy(self):
if self._dump_index < len(self._xml_sequence):
xml = self._xml_sequence[self._dump_index]
self._dump_index += 1
return xml
return self._xml_sequence[-1]
def press(self, key):
self.pressed_keys.append(key)
def click(self, x, y):
self.clicks.append((x, y))
def swipe(self, sx, sy, ex, ey, **kwargs):
self.swipes.append({"start": (sx, sy), "end": (ex, ey)})
def app_start(self, pkg, use_monkey=False):
pass
def app_stop(self, pkg):
pass
def unlock(self):
pass
def get_screenshot_b64(self):
return "iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAQAAAC1HAwCAAAAC0lEQVR42mNkYAAAAAYAAjCB0C8AAAAASUVORK5CYII="
def get_info(self):
return self._info
# ═══════════════════════════════════════════════════════
# E2E Workflow Fixtures — Real cognitive stack, real config
# ═══════════════════════════════════════════════════════
@pytest.fixture
def e2e_device():
"""Factory to create an E2EDeviceStub from an XML sequence."""
def _create(xml_sequence):
return E2EDeviceStub(xml_sequence)
return _create
@pytest.fixture(autouse=True)
def mock_llm_network_calls(monkeypatch, request):
"""Mocks ONLY the LLM network requests. The SAE structural logic remains 100% REAL."""
if request.config.getoption("--live"):
return
from GramAddict.core import llm_provider
import json
def fake_query_telepathic_llm(*args, **kwargs):
prompt = kwargs.get('user_prompt', args[3] if len(args) > 3 else "")
if "OBSTACLE_LOCKED_SCREEN" in prompt and "FOREIGN_APP" in prompt:
return json.dumps({"situation": "OBSTACLE_FOREIGN_APP"})
if "MODAL, DIALOG, or POPUP" in prompt:
return json.dumps({"situation": "NORMAL"})
return json.dumps({"situation": "NORMAL"})
def fake_query_llm(*args, **kwargs):
return json.dumps({"action": "false_positive", "reason": "Test mock", "x": 0, "y": 0})
monkeypatch.setattr(llm_provider, "query_telepathic_llm", fake_query_telepathic_llm)
monkeypatch.setattr(llm_provider, "query_llm", fake_query_llm)
@pytest.fixture
def e2e_cognitive_stack_factory(e2e_configs):
"""Build the REAL cognitive stack exactly like bot_flow.py does."""
def _create(device, username="testuser"):
from GramAddict.core.dopamine_engine import DopamineEngine
from GramAddict.core.interaction import LLMWriter
from GramAddict.core.qdrant_memory import DMMemoryDB, ParasocialCRMDB
from GramAddict.core.resonance_engine import ResonanceEngine
from GramAddict.core.q_nav_graph import QNavGraph
from GramAddict.core.telepathic_engine import TelepathicEngine
from GramAddict.core.zero_latency_engine import ZeroLatencyEngine
from GramAddict.core.darwin_engine import DarwinEngine
from GramAddict.core.active_inference import ActiveInferenceEngine
from GramAddict.core.sensors.honeypot_radome import HoneypotRadome
from GramAddict.core.swarm_protocol import SwarmProtocol
from GramAddict.core.growth_brain import GrowthBrain
dopamine = DopamineEngine()
dopamine.session_limit_seconds = 0.5
dopamine.session_start = time.time()
info = device.get_info() if hasattr(device, 'get_info') else {"displayWidth": 1080, "displayHeight": 2400}
return {
"dopamine": dopamine,
"active_inference": ActiveInferenceEngine(username),
"swarm": SwarmProtocol(username),
"resonance": ResonanceEngine(username, persona_interests=[], crm=ParasocialCRMDB()),
"growth_brain": GrowthBrain(username, persona_interests=[]),
"radome": HoneypotRadome(info.get("displayWidth", 1080), info.get("displayHeight", 2400)),
"nav_graph": QNavGraph(device),
"zero_engine": ZeroLatencyEngine(device),
"telepathic": TelepathicEngine.get_instance(),
"darwin": DarwinEngine(username),
"crm": ParasocialCRMDB(),
"dm_memory": DMMemoryDB(),
"writer": LLMWriter(username, [], e2e_configs),
}
return _create
@pytest.fixture
def e2e_session(e2e_configs):
"""Build a real SessionState."""
return SessionState(e2e_configs)
@pytest.fixture
def e2e_workflow_ctx(e2e_configs, e2e_cognitive_stack_factory, setup_e2e_plugin_registry):
"""
Factory that builds a REAL BehaviorContext + runs execute_all().
This is the EXACT code path from bot_flow.py:942-971.
"""
from GramAddict.core.behaviors import BehaviorContext
def _run(device, context_xml=None):
xml = context_xml if context_xml else device.dump_hierarchy()
session = SessionState(e2e_configs)
cognitive_stack = e2e_cognitive_stack_factory(device)
ctx = BehaviorContext(
device=device,
configs=e2e_configs,
session_state=session,
cognitive_stack=cognitive_stack,
context_xml=xml,
sleep_mod=1.0,
post_data={},
username="testuser",
shared_state={"consecutive_marker_misses": 0, "consecutive_ads": 0},
)
registry = setup_e2e_plugin_registry
results = registry.execute_all(ctx)
return results, ctx
return _run