Skip to content
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
32 changes: 31 additions & 1 deletion smartscore/scripts/backtrack/run_daily_picks.py
Original file line number Diff line number Diff line change
Expand Up @@ -1254,6 +1254,35 @@ def write_historic_picks_if_missing(picks, on_date, env):
return

columns = _historic_pick_columns(env)

# Keep the table to the most recent DAYS_TO_KEEP_HISTORIC_DATA days (mirrors
# the site's own publish pruning) so it stays a ~week visual instead of
# growing forever. Prune before inserting today's rows.
DAYS_TO_KEEP_HISTORIC_DATA = 8
try:
all_rows = SUPABASE_ADMIN_CLIENT.table(table).select("date").execute().data or []
distinct_dates = sorted({r.get("date") for r in all_rows if r.get("date")})
if len(distinct_dates) >= DAYS_TO_KEEP_HISTORIC_DATA:
to_remove = distinct_dates[: len(distinct_dates) - (DAYS_TO_KEEP_HISTORIC_DATA - 1)]
for d in to_remove:
SUPABASE_ADMIN_CLIENT.table(table).delete().eq("date", d).execute()
except Exception as exc: # noqa: BLE001
print(f"WARNING: Historic-Picks pruning failed: {exc}", flush=True)

# `id` is the table's (non-null, no-default) primary key. The site's own
# publish renumbers every row sequentially on each write, but this ML pass
# INSERTs brand-new rows, so a fixed 1..N would collide with the rows
# already stored from earlier dates. Derive the next free id from the
# current max instead (and fall back to 1..N only on an empty/odd table).
next_id = 1
if "id" in columns:
try:
existing_ids = SUPABASE_ADMIN_CLIENT.table(table).select("id").execute().data or []
parsed = [int(r.get("id")) for r in existing_ids if r.get("id") is not None]
if parsed:
next_id = max(parsed) + 1
except (TypeError, ValueError):
next_id = 1
records = []
for i, (_, row) in enumerate(picks.iterrows()):
record = {"date": on_date}
Expand All @@ -1262,14 +1291,15 @@ def write_historic_picks_if_missing(picks, on_date, env):
pid = pid.item()
record["player_id"] = int(pid) if pid is not None and not (isinstance(pid, float) and np.isnan(pid)) else None
if "id" in columns:
record["id"] = i + 1
record["id"] = next_id + i
if "Scored" in columns:
record["Scored"] = None
for col in columns:
if col in ("id", "date", "player_id", "Scored"):
continue
if col == "stat":
value = stat_value(row.get("ensemble", np.nan))
record[col] = value
elif col in getattr(row, "index", []):
value = row.get(col)
if isinstance(value, (np.generic,)):
Expand Down
Loading