sentiment-engine/MALKHUT/malkhut/tests/test_prod_tooling.py

"""
Tests for cognition launcher, news sources, and monitor.
"""
import json
import os
import tempfile
import time
import pytest
from malkhut.cognition_launcher import CognitionLauncher, CognitionConfig
from malkhut.training.news_sources import NewsSourceRepository, SourceMetadata
from malkhut.training.monitor import CognitionMonitor, CognitionMetrics


# ══════════════════════════════════════════════════════════════════════════════
# COGNITION LAUNCHER (8 tests)
# ══════════════════════════════════════════════════════════════════════════════

class TestCognitionLauncher:
    def test_launcher_creates(self):
        launcher = CognitionLauncher()
        assert launcher._running is False
        assert launcher._total_fetched == 0

    def test_launcher_config(self):
        cfg = CognitionConfig(rate_limit_rpm=10, fetch_interval_s=5)
        launcher = CognitionLauncher(config=cfg)
        assert launcher.config.rate_limit_rpm == 10

    def test_launcher_loads_regimes(self):
        path = '/tmp/test_regimes_e2e.json'
        with open(path, 'w') as f:
            json.dump({"flash_crash": {"source": "test"}}, f)
        try:
            launcher = CognitionLauncher(CognitionConfig(regime_db_path=path))
            assert launcher._total_regimes == 1
        finally:
            os.unlink(path)

    def test_launcher_saves_regimes(self):
        with tempfile.NamedTemporaryFile(suffix=".json", delete=False) as f:
            path = f.name
        try:
            launcher = CognitionLauncher(CognitionConfig(regime_db_path=path))
            launcher._discovered_regimes["test"] = {"source": "test"}
            launcher._save_regimes()
            with open(path) as f:
                data = json.load(f)
            assert "test" in data
        finally:
            os.unlink(path)


# ══════════════════════════════════════════════════════════════════════════════
# NEWS SOURCE REPOSITORY (15 tests)
# ══════════════════════════════════════════════════════════════════════════════

class TestNewsSourceRepository:
    def test_register_source(self):
        with tempfile.NamedTemporaryFile(suffix=".json", delete=False) as f:
            path = f.name
        try:
            repo = NewsSourceRepository(path)
            repo.register("s1", "Test", "https://example.com", "news", 0.8, 0.9, 1.0)
            assert repo.source_count == 1
        finally:
            os.unlink(path)

    def test_rank_by_relevance(self):
        with tempfile.NamedTemporaryFile(suffix=".json", delete=False) as f:
            path = f.name
        try:
            repo = NewsSourceRepository(path)
            repo.register("s1", "Low", "https://a.com", "news", 0.3, 0.8, 1.0)
            repo.register("s2", "High", "https://b.com", "news", 0.9, 0.9, 1.0)
            ranked = repo.rank_by_relevance()
            assert ranked[0].source_id == "s2"
        finally:
            os.unlink(path)

    def test_rank_by_health(self):
        with tempfile.NamedTemporaryFile(suffix=".json", delete=False) as f:
            path = f.name
        try:
            repo = NewsSourceRepository(path)
            repo.register("s1", "Good", "https://a.com", reliability=0.95)
            repo.register("s2", "Bad", "https://b.com", reliability=0.5)
            ranked = repo.rank_by_health()
            assert ranked[0].source_id == "s1"
        finally:
            os.unlink(path)

    def test_get_by_type(self):
        with tempfile.NamedTemporaryFile(suffix=".json", delete=False) as f:
            path = f.name
        try:
            repo = NewsSourceRepository(path)
            repo.register("s1", "News", "https://news.com", "news")
            repo.register("s2", "Data", "https://data.com", "data")
            assert len(repo.get_by_type("news")) == 1
            assert len(repo.get_by_type("data")) == 1
        finally:
            os.unlink(path)

    def test_get_by_tag(self):
        with tempfile.NamedTemporaryFile(suffix=".json", delete=False) as f:
            path = f.name
        try:
            repo = NewsSourceRepository(path)
            repo.register("s1", "Test", "https://example.com", tags=("crypto", "news"))
            assert len(repo.get_by_tag("crypto")) == 1
        finally:
            os.unlink(path)

    def test_disable_enable(self):
        with tempfile.NamedTemporaryFile(suffix=".json", delete=False) as f:
            path = f.name
        try:
            repo = NewsSourceRepository(path)
            repo.register("s1", "Test", "https://example.com")
            assert repo.enabled_count == 1
            repo.disable("s1")
            assert repo.enabled_count == 0
            repo.enable("s1")
            assert repo.enabled_count == 1
        finally:
            os.unlink(path)

    def test_persistence(self):
        with tempfile.NamedTemporaryFile(suffix=".json", delete=False) as f:
            path = f.name
        try:
            repo1 = NewsSourceRepository(path)
            repo1.register("s1", "Test", "https://example.com")
            repo1.record_fetch("s1", success=True)

            repo2 = NewsSourceRepository(path)
            assert repo2.source_count == 1
            assert repo2.get_enabled()[0].fetch_count == 1
        finally:
            os.unlink(path)

    def test_seed_defaults(self):
        with tempfile.NamedTemporaryFile(suffix=".json", delete=False) as f:
            path = f.name
        try:
            repo = NewsSourceRepository(path)
            repo.seed_defaults()
            assert repo.source_count >= 10
        finally:
            os.unlink(path)

    def test_health_score(self):
        s = SourceMetadata(source_id="s1", name="Test", url="https://x.com",
                          source_type="news", regime_relevance=0.5, reliability=0.9,
                          freshness_hours=1.0, fetch_count=10, error_count=1)
        assert s.health_score > 0
        assert s.health_score < 1.0


# ══════════════════════════════════════════════════════════════════════════════
# COGNITION MONITOR (10 tests)
# ══════════════════════════════════════════════════════════════════════════════

class TestCognitionMonitor:
    def test_record_fetch(self):
        with tempfile.NamedTemporaryFile(suffix=".log", delete=False) as f:
            path = f.name
        try:
            mon = CognitionMonitor(path)
            mon.record_fetch("s1", success=True)
            assert mon.total_fetched == 1
        finally:
            os.unlink(path)

    def test_record_error(self):
        with tempfile.NamedTemporaryFile(suffix=".log", delete=False) as f:
            path = f.name
        try:
            mon = CognitionMonitor(path)
            mon.record_fetch("s1", success=False)
            assert mon.total_errors == 1
        finally:
            os.unlink(path)

    def test_record_regime(self):
        with tempfile.NamedTemporaryFile(suffix=".log", delete=False) as f:
            path = f.name
        try:
            mon = CognitionMonitor(path)
            mon.record_regime(3)
            assert mon.discovered_regimes == 3
        finally:
            os.unlink(path)

    def test_snapshot(self):
        with tempfile.NamedTemporaryFile(suffix=".log", delete=False) as f:
            path = f.name
        try:
            mon = CognitionMonitor(path)
            mon.record_fetch("s1", success=True)
            metrics = mon.snapshot(total_sources=5, enabled_sources=4)
            assert isinstance(metrics, CognitionMetrics)
            assert metrics.total_fetched == 1
        finally:
            os.unlink(path)

    def test_metrics_logged(self):
        with tempfile.NamedTemporaryFile(suffix=".log", delete=False) as f:
            path = f.name
        try:
            mon = CognitionMonitor(path)
            mon.snapshot(total_sources=5, enabled_sources=4)
            with open(path) as f:
                lines = f.readlines()
            assert len(lines) == 1
            record = json.loads(lines[0])
            assert "sources" in record
        finally:
            os.unlink(path)

    def test_check_alerts_high_error(self):
        with tempfile.NamedTemporaryFile(suffix=".log", delete=False) as f:
            path = f.name
        try:
            mon = CognitionMonitor(path)
            for _ in range(20):
                mon.record_fetch("s1", success=False)
            mon.record_fetch("s1", success=True)
            mon.snapshot(total_sources=1, enabled_sources=1)
            alerts = mon.check_alerts()
            assert len(alerts) > 0
        finally:
            os.unlink(path)

    def test_check_alerts_low_fetch_rate(self):
        with tempfile.NamedTemporaryFile(suffix=".log", delete=False) as f:
            path = f.name
        try:
            mon = CognitionMonitor(path)
            mon._start_time = time.time() - 600  # 10 min ago
            mon.record_fetch("s1", success=True)  # only 1 fetch in 10 min
            mon.snapshot(total_sources=1, enabled_sources=1)
            alerts = mon.check_alerts()
            assert any("LOW_FETCH_RATE" in a for a in alerts)
        finally:
            os.unlink(path)

    def test_check_alerts_clean(self):
        with tempfile.NamedTemporaryFile(suffix=".log", delete=False) as f:
            path = f.name
        try:
            mon = CognitionMonitor(path)
            for _ in range(100):
                mon.record_fetch("s1", success=True)
            mon.snapshot(total_sources=1, enabled_sources=1)
            alerts = mon.check_alerts()
            assert len(alerts) == 0
        finally:
            os.unlink(path)