importer: rgl: Support combining teams

Add support for combining teams that have been linked after already having been imported. This occasionally happens by request. Signed-off-by: Sean Anderson <[email protected]>
Forty-Bot · Apr 13, 2024 · 9acd45e · 9acd45e
1 parent 0f8a44b
commit 9acd45e
Show file tree

Hide file tree

Showing 4 changed files with 74 additions and 13 deletions.
diff --git a/test/create.py b/test/create.py
@@ -13,6 +13,22 @@
 from trends.importer.fetch import DemoFileFetcher, ETF2LFileFetcher, FileFetcher, RGLFileFetcher
 from trends.sql import db_connect, db_init, db_schema
 
+# Pretend these two teams aren't linked so we can test combining them
+class RGLFirstFetcher(RGLFileFetcher):
+    def get_team(self, teamid):
+        ret = super().get_team(teamid)
+        if teamid == 5574:
+            ret['linkedTeams'] = []
+            ret['fetched'] = 1
+        if teamid == 4882:
+            ret['linkedTeams'] = []
+            ret['fetched'] = 1
+        return ret
+
+class RGLSecondFetcher(RGLFileFetcher):
+    def get_matchids(self):
+        yield 4664
+
 def create_test_db(url):
     # We use separate connections for importing because we use temporary tables which will alias
     # other queries.
@@ -82,9 +98,10 @@ def create_test_db(url):
         trends.importer.etf2l.import_etf2l(c, fetcher)
 
     with db_connect(url) as c:
-        fetcher = RGLFileFetcher(dir=f"{os.path.dirname(__file__)}/rgl")
-        trends.importer.rgl.import_rgl(c, fetcher)
-
+        dir = f"{os.path.dirname(__file__)}/rgl"
+        trends.importer.rgl.import_rgl(c, RGLFirstFetcher(dir=dir))
+        trends.importer.rgl.import_rgl(c, RGLSecondFetcher(dir=dir),
+                                       filter=trends.importer.rgl.no_filter_matchids)
     with db_connect(url) as c:
         cur = c.cursor()
         cur.execute("ANALYZE;")

diff --git a/trends/importer/league.py b/trends/importer/league.py
@@ -37,6 +37,38 @@ def import_compdiv(c, cd):
                (SELECT div_nameid FROM div_name WHERE division = %(division)s), %(tier)s
            ) ON CONFLICT DO NOTHING;""", cd)
 
+def combine_teams(c, league, teamids):
+    """Combine teams
+    Combine the teams specified into one team. This may occasionally be necessary when RGL teams are
+    retroactively linked.
+    """
+
+    new_teamid = min(teamids)
+    teamids = tuple(teamid for teamid in teamids if teamid != new_teamid)
+    def set_constraints(behavior):
+        c.execute(
+            f"""SET CONSTRAINTS
+                    team_player_league_teamid_fkey,
+                    team_player_league_teamid_compid_fkey,
+                    match_league_compid_teamid1_fkey,
+                    match_league_compid_teamid2_fkey
+                {behavior};""");
+
+    set_constraints("DEFERRED")
+    for table, col in (
+        ('team_comp_backing', 'teamid'),
+        ('team_player', 'teamid'),
+        ('match', 'teamid1'),
+        ('match', 'teamid2'),
+    ):
+        c.execute(
+            f"""UPDATE {table}
+                SET {col} = %s
+                WHERE league = %s AND {col} IN %s;""", (new_teamid, league, teamids))
+    set_constraints("IMMEDIATE")
+
+    return new_teamid
+
 def import_team(c, t):
     """Import a team
     The competition and division should already be imported. t should be a dict with the following
@@ -69,13 +101,14 @@ def import_team(c, t):
     if teamids := t.get('rgl_teamids'):
         c.execute("SELECT teamid FROM team_comp WHERE rgl_teamid IN %s GROUP BY teamid",
                   (teamids,))
-        if c.rowcount not in (0, 1):
-            logging.warning("Too many matching teams for linked RGL teamids %s", teamids)
-
-        if row := c.fetchone():
-            t['teamid'] = row[0]
+        if c.rowcount > 1:
+            logging.info("Combining RGL teamids %s", teamids)
+            t['teamid'] = combine_teams(c, t['league'], tuple(row[0] for row in c.fetchall()))
         else:
-            teamid_col = min(teamids)
+            if row := c.fetchone():
+                t['teamid'] = row[0]
+            else:
+                teamid_col = min(teamids)
 
     c.execute("INSERT INTO team_name (team) VALUES (%(name)s) ON CONFLICT DO NOTHING", t)
     c.execute(

diff --git a/trends/migrations/deferrable_teamid.sql b/trends/migrations/deferrable_teamid.sql
@@ -0,0 +1,8 @@
+BEGIN;
+ALTER TABLE team_player
+	ALTER CONSTRAINT team_player_league_teamid_fkey DEFERRABLE,
+	ALTER CONSTRAINT team_player_league_teamid_compid_fkey DEFERRABLE;
+ALTER TABLE match
+	ALTER CONSTRAINT match_league_compid_teamid1_fkey DEFERRABLE,
+	ALTER CONSTRAINT match_league_compid_teamid2_fkey DEFERRABLE;
+COMMIT;
diff --git a/trends/schema.sql b/trends/schema.sql
@@ -238,8 +238,9 @@ CREATE TABLE IF NOT EXISTS team_player (
 		rostered WITH &&,
 		playerid WITH =
 	) WHERE (compid ISNULL),
-	FOREIGN KEY (league, teamid) REFERENCES league_team (league, teamid),
-	FOREIGN KEY (league, teamid, compid) REFERENCES team_comp_backing (league, teamid, compid),
+	FOREIGN KEY (league, teamid) REFERENCES league_team (league, teamid) DEFERRABLE,
+	FOREIGN KEY (league, teamid, compid)
+		REFERENCES team_comp_backing (league, teamid, compid) DEFERRABLE,
 	CHECK (league_team_per_comp(league) = (compid NOTNULL))
 );
 
@@ -295,8 +296,10 @@ CREATE TABLE IF NOT EXISTS match (
 	FOREIGN KEY (league, round_compid, round_seq)
 		REFERENCES comp_round (league, compid, round_seq),
 	FOREIGN KEY (league, divid, round_seq) REFERENCES div_round (league, divid, round_seq),
-	FOREIGN KEY (league, compid, teamid1) REFERENCES team_comp_backing (league, compid, teamid),
-	FOREIGN KEY (league, compid, teamid2) REFERENCES team_comp_backing (league, compid, teamid),
+	FOREIGN KEY (league, compid, teamid1)
+		REFERENCES team_comp_backing (league, compid, teamid) DEFERRABLE,
+	FOREIGN KEY (league, compid, teamid2)
+		REFERENCES team_comp_backing (league, compid, teamid) DEFERRABLE,
 	CHECK (teamid1 < teamid2),
 	CHECK (league_div_optional(league) OR (divid NOTNULL)),
 	CHECK (league_time_optional(league) OR forfeit OR (scheduled NOTNULL)),