feat(scripts): add sportstime-parser data pipeline

Complete Python package for scraping, normalizing, and uploading sports schedule data to CloudKit. Includes: - Multi-source scrapers for NBA, MLB, NFL, NHL, MLS, WNBA, NWSL - Canonical ID system for teams, stadiums, and games - Fuzzy matching with manual alias support - CloudKit uploader with batch operations and deduplication - Comprehensive test suite with fixtures - WNBA abbreviation aliases for improved team resolution - Alias validation script to detect orphan references All 5 phases of data remediation plan completed: - Phase 1: Alias fixes (team/stadium alias additions) - Phase 2: NHL stadium coordinate fixes - Phase 3: Re-scrape validation - Phase 4: iOS bundle update - Phase 5: Code quality improvements (WNBA aliases) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-20 18:56:25 -06:00
parent ac78042a7e
commit 52d445bca4
76 changed files with 25065 additions and 0 deletions
--- a/sportstime_parser/tests/test_canonical_id.py
+++ b/sportstime_parser/tests/test_canonical_id.py
@@ -0,0 +1,187 @@
+"""Tests for canonical ID generation."""
+
+import pytest
+from datetime import datetime, date
+
+from sportstime_parser.normalizers.canonical_id import (
+    generate_game_id,
+    generate_team_id,
+    generate_team_id_from_abbrev,
+    generate_stadium_id,
+    parse_game_id,
+    normalize_string,
+)
+
+
+class TestNormalizeString:
+    """Tests for normalize_string function."""
+
+    def test_basic_normalization(self):
+        """Test basic string normalization."""
+        assert normalize_string("New York") == "new_york"
+        assert normalize_string("Los Angeles") == "los_angeles"
+
+    def test_removes_special_characters(self):
+        """Test that special characters are removed."""
+        assert normalize_string("AT&T Stadium") == "att_stadium"
+        assert normalize_string("St. Louis") == "st_louis"
+        assert normalize_string("O'Brien Field") == "obrien_field"
+
+    def test_collapses_whitespace(self):
+        """Test that multiple spaces are collapsed."""
+        assert normalize_string("New   York") == "new_york"
+        assert normalize_string("  Los Angeles  ") == "los_angeles"
+
+    def test_empty_string(self):
+        """Test empty string handling."""
+        assert normalize_string("") == ""
+        assert normalize_string("   ") == ""
+
+    def test_unicode_normalization(self):
+        """Test unicode characters are handled."""
+        assert normalize_string("Café") == "cafe"
+        assert normalize_string("José") == "jose"
+
+
+class TestGenerateGameId:
+    """Tests for generate_game_id function."""
+
+    def test_basic_game_id(self):
+        """Test basic game ID generation."""
+        game_id = generate_game_id(
+            sport="nba",
+            season=2025,
+            away_abbrev="bos",
+            home_abbrev="lal",
+            game_date=date(2025, 12, 25),
+        )
+        assert game_id == "game_nba_2025_20251225_bos_lal"
+
+    def test_game_id_with_datetime(self):
+        """Test game ID generation with datetime object."""
+        game_id = generate_game_id(
+            sport="mlb",
+            season=2026,
+            away_abbrev="nyy",
+            home_abbrev="bos",
+            game_date=datetime(2026, 4, 1, 19, 0),
+        )
+        assert game_id == "game_mlb_2026_20260401_nyy_bos"
+
+    def test_game_id_with_game_number(self):
+        """Test game ID for doubleheader."""
+        game_id_1 = generate_game_id(
+            sport="mlb",
+            season=2026,
+            away_abbrev="nyy",
+            home_abbrev="bos",
+            game_date=date(2026, 7, 4),
+            game_number=1,
+        )
+        game_id_2 = generate_game_id(
+            sport="mlb",
+            season=2026,
+            away_abbrev="nyy",
+            home_abbrev="bos",
+            game_date=date(2026, 7, 4),
+            game_number=2,
+        )
+        assert game_id_1 == "game_mlb_2026_20260704_nyy_bos_1"
+        assert game_id_2 == "game_mlb_2026_20260704_nyy_bos_2"
+
+    def test_sport_lowercased(self):
+        """Test that sport is lowercased."""
+        game_id = generate_game_id(
+            sport="NBA",
+            season=2025,
+            away_abbrev="BOS",
+            home_abbrev="LAL",
+            game_date=date(2025, 12, 25),
+        )
+        assert game_id == "game_nba_2025_20251225_bos_lal"
+
+
+class TestParseGameId:
+    """Tests for parse_game_id function."""
+
+    def test_parse_basic_game_id(self):
+        """Test parsing a basic game ID."""
+        parsed = parse_game_id("game_nba_2025_20251225_bos_lal")
+        assert parsed["sport"] == "nba"
+        assert parsed["season"] == 2025
+        assert parsed["away_abbrev"] == "bos"
+        assert parsed["home_abbrev"] == "lal"
+        assert parsed["year"] == 2025
+        assert parsed["month"] == 12
+        assert parsed["day"] == 25
+        assert parsed["game_number"] is None
+
+    def test_parse_game_id_with_game_number(self):
+        """Test parsing game ID with game number."""
+        parsed = parse_game_id("game_mlb_2026_20260704_nyy_bos_2")
+        assert parsed["sport"] == "mlb"
+        assert parsed["season"] == 2026
+        assert parsed["away_abbrev"] == "nyy"
+        assert parsed["home_abbrev"] == "bos"
+        assert parsed["year"] == 2026
+        assert parsed["month"] == 7
+        assert parsed["day"] == 4
+        assert parsed["game_number"] == 2
+
+    def test_parse_invalid_game_id(self):
+        """Test parsing invalid game ID raises error."""
+        with pytest.raises(ValueError):
+            parse_game_id("invalid")
+        with pytest.raises(ValueError):
+            parse_game_id("nba_2025_bos")  # Missing game_ prefix
+        with pytest.raises(ValueError):
+            parse_game_id("")
+        with pytest.raises(ValueError):
+            parse_game_id("game_nba_2025_bos_lal")  # Missing date
+
+
+class TestGenerateTeamId:
+    """Tests for generate_team_id function."""
+
+    def test_basic_team_id(self):
+        """Test basic team ID generation from city and name."""
+        team_id = generate_team_id(sport="nba", city="Los Angeles", name="Lakers")
+        assert team_id == "team_nba_los_angeles_lakers"
+
+    def test_team_id_normalizes_input(self):
+        """Test that inputs are normalized."""
+        team_id = generate_team_id(sport="NBA", city="New York", name="Yankees")
+        assert team_id == "team_nba_new_york_yankees"
+
+
+class TestGenerateTeamIdFromAbbrev:
+    """Tests for generate_team_id_from_abbrev function."""
+
+    def test_basic_team_id_from_abbrev(self):
+        """Test team ID from abbreviation."""
+        team_id = generate_team_id_from_abbrev(sport="nba", abbreviation="LAL")
+        assert team_id == "team_nba_lal"
+
+    def test_lowercases_abbreviation(self):
+        """Test abbreviation is lowercased."""
+        team_id = generate_team_id_from_abbrev(sport="MLB", abbreviation="NYY")
+        assert team_id == "team_mlb_nyy"
+
+
+class TestGenerateStadiumId:
+    """Tests for generate_stadium_id function."""
+
+    def test_basic_stadium_id(self):
+        """Test basic stadium ID generation."""
+        stadium_id = generate_stadium_id(sport="mlb", name="Fenway Park")
+        assert stadium_id == "stadium_mlb_fenway_park"
+
+    def test_stadium_id_special_characters(self):
+        """Test stadium ID with special characters."""
+        stadium_id = generate_stadium_id(sport="nfl", name="AT&T Stadium")
+        assert stadium_id == "stadium_nfl_att_stadium"
+
+    def test_stadium_id_with_sponsor(self):
+        """Test stadium ID with sponsor name."""
+        stadium_id = generate_stadium_id(sport="nba", name="Crypto.com Arena")
+        assert stadium_id == "stadium_nba_cryptocom_arena"