import _bootstrap  # noqa: F401

import unittest
from datetime import datetime, timezone

from fault_log_analyzer.cluster import (
    ClusterEngine,
    TfidfVectorizer,
    cosine_distance,
    cosine_similarity,
    dbscan,
)
from fault_log_analyzer.models import FaultLog


def make_log(fid, message, host="h-1"):
    return FaultLog(
        fault_log_id=fid,
        host_id=host,
        fingerprint="fp",
        level="ERROR",
        message=message,
        occurred_at=datetime(2025, 1, 1, tzinfo=timezone.utc),
    )


class TestTfidfVectorizer(unittest.TestCase):
    def test_fit_transform(self):
        vec = TfidfVectorizer().fit([["a", "b"], ["b", "c"]])
        out = vec.transform([["a", "b"], ["b", "c"]])
        self.assertEqual(len(out), 2)
        self.assertTrue(all(isinstance(v, dict) for v in out))

    def test_empty_documents(self):
        vec = TfidfVectorizer().fit([])
        self.assertEqual(vec.transform([]), [])


class TestCosineSimilarity(unittest.TestCase):
    def test_identical(self):
        self.assertEqual(cosine_similarity({"a": 1.0}, {"a": 1.0}), 1.0)

    def test_disjoint(self):
        self.assertEqual(cosine_similarity({"a": 1.0}, {"b": 1.0}), 0.0)

    def test_distance(self):
        self.assertAlmostEqual(cosine_distance({"a": 1.0}, {"a": 1.0}), 0.0)


class TestDbscan(unittest.TestCase):
    def test_empty(self):
        result = dbscan([])
        self.assertEqual(result.labels, [])
        self.assertEqual(result.cluster_ids, {})

    def test_single_noise(self):
        result = dbscan([{"a": 1.0}], eps=0.75, min_samples=2)
        self.assertEqual(result.labels, [-1])

    def test_two_similar_cluster(self):
        result = dbscan([{"a": 1.0}, {"a": 1.0}], eps=0.75, min_samples=1)
        self.assertEqual(result.labels[0], result.labels[1])
        self.assertNotEqual(result.labels[0], -1)

    def test_two_dissimilar_noise(self):
        result = dbscan([{"a": 1.0}, {"b": 1.0}], eps=0.75, min_samples=2)
        self.assertEqual(result.labels, [-1, -1])


class TestClusterEngine(unittest.TestCase):
    def test_cluster_batch_empty(self):
        ce = ClusterEngine()
        self.assertEqual(ce.cluster_batch([]), {})

    def test_cluster_batch_groups_identical(self):
        ce = ClusterEngine(eps=0.75, min_samples=1)
        logs = [make_log("fl1", "No space left on device"), make_log("fl2", "No space left on device")]
        assignment = ce.cluster_batch(logs)
        self.assertEqual(assignment["fl1"], assignment["fl2"])

    def test_incremental_match_existing_cluster(self):
        ce = ClusterEngine(eps=0.75, min_samples=1)
        first = ce.cluster_batch([make_log("fl1", "No space left on device")])
        second = ce.cluster_batch([make_log("fl2", "No space left on device")])
        self.assertEqual(second["fl2"], first["fl1"])


if __name__ == "__main__":
    unittest.main()