databank/tests/test_analytics_team_elo.py

"""Basic unit tests for TeamExtractor and Elo analyzers.

This module uses a simple in-memory FakeDB to validate:
- Team aliases union behavior
- Elo snapshot and history persistence contracts
"""

import unittest
from typing import Any, Dict, List

from databank.analytics.teams import TeamExtractorAnalyzer
from databank.analytics.elo import EloAnalyzer
from databank.core.models import Document


class FakeDB:
    """A minimal in-memory DB stub implementing insert_many/find for tests."""

    def __init__(self) -> None:
        self.storage: Dict[str, List[Dict[str, Any]]] = {}

    def connect(self) -> None:  # for parity
        """No-op connect to match real DB interface."""
        return None

    def insert_many(self, docs: List[Document]) -> int:
        """Append documents by kind; return inserted count."""
        count = 0
        for d in docs:
            kind = d.kind
            rec = {"_id": d.id, "kind": kind, "data": d.data}
            self.storage.setdefault(kind, []).append(rec)
            count += 1
        return count

    def find(
        self,
        kind: str,
        query: Dict[str, Any] | None = None,
        _projection=None,
        _limit=None,
    ):  # noqa: A002 - filter naming for parity
        """Retrieve documents by kind; supports _id $in filter."""
        data = self.storage.get(kind, [])
        if not query:
            return list(data)
        # Support _id $in
        ids = None
        if "_id" in query and isinstance(query["_id"], dict) and "$in" in query["_id"]:
            ids = set(map(str, query["_id"]["$in"]))
        if ids is not None:
            return [d for d in data if str(d.get("_id")) in ids]
        return list(data)


class TestAnalyticsTeamElo(unittest.TestCase):
    """Tests for TeamExtractorAnalyzer and EloAnalyzer integrations."""

    def test_team_extractor_alias_union(self) -> None:
        """Teams analyzer should union new aliases with existing ones."""
        db = FakeDB()
        # Existing team with alias
        db.insert_many(
            [
                Document(
                    id="100",
                    kind="teams",
                    data={
                        "team_id": "100",
                        "name_canonical": "manchester city",
                        "aliases": ["曼城"],
                    },
                )
            ]
        )
        # Two matches with different alias spellings
        matches = [
            {
                "match": {
                    "homeTeamId": 100,
                    "homeTeamName": "Manchester City",
                    "awayTeamId": 200,
                    "awayTeamName": "Arsenal",
                }
            },
            {
                "match": {
                    "homeTeamId": 300,
                    "homeTeamName": "Chelsea",
                    "awayTeamId": 100,
                    "awayTeamName": "Man City",
                }
            },
        ]
        res = TeamExtractorAnalyzer().compute(matches, db=db)
        self.assertIn("upserted", res)
        # Check merged aliases contain existing + new
        teams = db.find("teams")
        # Find all records for _id 100 and pick the latest appended (upsert effect)
        mlist = [x for x in teams if str(x.get("_id")) == "100"]
        self.assertGreaterEqual(len(mlist), 1)
        mrec = mlist[-1]
        aliases = set(mrec["data"]["aliases"])  # type: ignore[index]
        self.assertTrue({"曼城", "Manchester City", "Man City"}.issubset(aliases))

    def test_elo_persist_snapshot_and_history(self) -> None:
        """Elo analyzer should persist snapshot and two history entries per match."""
        db = FakeDB()
        # Two matches: A beats B 1-0, then draw 0-0
        matches = [
            {
                "match": {
                    "matchId": 1,
                    "homeTeamId": "A",
                    "awayTeamId": "B",
                    "homeScore": 1,
                    "awayScore": 0,
                    "elapsedTime": "已完场",
                    "matchTime": 1_700_000_000,
                }
            },
            {
                "match": {
                    "matchId": 2,
                    "homeTeamId": "A",
                    "awayTeamId": "B",
                    "homeScore": 0,
                    "awayScore": 0,
                    "elapsedTime": "已完场",
                    "matchTime": 1_700_000_100,
                }
            },
        ]
        res = EloAnalyzer().compute(matches, db=db, persist=True)
        self.assertEqual(res["processed"], 2)
        # Snapshot should exist
        elo = db.find("elo_ratings")
        self.assertTrue(len(elo) >= 2)
        # History should have 4 entries (2 matches x 2 teams)
        hist = db.find("ratings_history")
        self.assertEqual(len(hist), 4)
        # Ensure history contains match_id and ts
        for h in hist:
            self.assertIn("match_id", h["data"])  # type: ignore[index]
            self.assertIn("ts", h["data"])  # type: ignore[index]


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