Update SQL

SQL

@@ -1,50 +1,243 @@
-We now have the actual failure:
-
-422 Client Error: Unprocessable Entity for:
-GET /v4/sports/baseball_mlb/odds?markets=batter_home_runs&bookmakers=draftkings,fanduel,betmgm...
-
-This confirms the current endpoint/request combination is invalid.
-
-Important documented behavior from The Odds API:
-- The aggregate MLB odds endpoint `/v4/sports/baseball_mlb/odds` is for featured markets like h2h, spreads, totals.
-- For additional markets including player props such as `batter_home_runs`, the docs say to use the event-odds endpoint:
-  `/v4/sports/baseball_mlb/events/{eventId}/odds`
-- Therefore the root cause is that our upcoming HR props path is hitting the wrong endpoint for player props.
-
-Your task:
-Implement the smallest safe fix with strong downstream-breakage awareness.
-
-Scope:
-ONLY change the upcoming HR props path.
-Do NOT modify live props behavior unless absolutely necessary.
-Do NOT do a large refactor.
-
-Implementation plan:
-1. In the Odds API provider, replace the current sport-level upcoming HR props request flow with:
-   a. GET `/v4/sports/baseball_mlb/events`
-   b. for each relevant event id, GET `/v4/sports/baseball_mlb/events/{eventId}/odds`
-      with:
-      - markets=batter_home_runs
-      - bookmakers=draftkings,fanduel,betmgm
-      - regions=us if appropriate for this endpoint
-      - oddsFormat=american
-      - dateFormat=iso if supported
-2. Normalize the event-level responses into the exact same dataframe schema currently expected by the downstream props page and mapper.
-3. Keep current logging/diagnostics in place.
-4. If one event call fails, log it and continue the rest rather than hard-failing the whole page.
-5. Add a conservative max-events guard so we do not accidentally burn too many credits.
-6. Before editing, show me:
-   - exact files/functions to change
-   - exact expected downstream schema
-   - any risk to live props path
-   - whether this changes only upcoming HR props
-7. After editing, explain:
-   - exact old path
-   - exact new path
-   - why it fixes the 422
-   - any quota/credit tradeoff
-
-Critical:
-- Do not change unrelated odds paths.
-- Do not assume FanDuel is missing until after the event-level calls are tested.
-- Preserve the existing downstream dataframe contract.
+Here is Claude's plan:                                                                                                                                            ╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌ Fix: HR Props 422 — Replace Upcoming Props Endpoint                                                                                                               
+  
+ Root Cause (Confirmed)
+
+ GET /v4/sports/baseball_mlb/odds?markets=batter_home_runs → HTTP 422.
+ The aggregate /odds endpoint only serves h2h, spreads, totals.
+ Player props require /v4/sports/baseball_mlb/events/{eventId}/odds.
+
+ ---
+ Exact File/Function to Change
+
+ One function, one file.
+
+ - File: data/provider_theoddsapi.py
+ - Function: TheOddsAPIProvider.fetch_all_upcoming_hr_props (lines 193–305)
+ - What changes: The entire body after the two guard checks — everything from the
+ sport_key = "baseball_mlb" line onward is replaced.
+ - Signature: unchanged (self, sportsbooks: list[str] | None = None) → no call sites break.
+
+ Nothing else is modified. fetch_live_prop_odds is untouched.
+
+ ---
+ Downstream Schema Contract
+
+ normalize_prop_odds in live_prop_odds.py:11-34 requires these columns
+ (missing ones are filled with None — order does not matter):
+
+ provider, event_id, commence_time, away_team, home_team,
+ sportsbook, market, player_name_raw, player_name, odds_american, line
+
+ The new code produces the identical row dict as the old code — same keys, same
+ derivation logic. The per-event /events/{id}/odds response uses the same
+ bookmakers → markets → outcomes JSON structure as the aggregate endpoint.
+ market_key and sportsbook_key are also produced (extra columns — safe, just ignored by
+ normalize_prop_odds, used downstream by the execution layer if present).
+
+ ---
+ Live Props Path: Zero Risk
+
+ fetch_live_prop_odds (lines 101–191) is a separate method. It is not called, not
+ imported, and not refactored. The Dashboard debug panel that calls it will continue to
+ behave identically (including its own 422 bug, which is out of scope).
+
+ ---
+ New Request Flow
+
+ Old (broken):
+ 1 call → GET /v4/sports/baseball_mlb/odds?markets=batter_home_runs → 422
+
+ New:
+ Step 1 → GET /v4/sports/baseball_mlb/events
+            params: apiKey, dateFormat=iso,
+                    commenceTimeFrom=<now UTC>,
+                    commenceTimeTo=<now+7d UTC>
+            → list of {id, away_team, home_team, commence_time}
+            → 1 API credit
+
+ Step 2 → for each event_id (capped at _MAX_EVENTS=20):
+            GET /v4/sports/baseball_mlb/events/{event_id}/odds
+            params: apiKey, markets=batter_home_runs,
+                    bookmakers=draftkings,fanduel,betmgm,
+                    oddsFormat=american, dateFormat=iso
+            → {id, bookmakers:[{key, markets:[{key, outcomes:[{description,price,point}]}]}]}
+            → 1 API credit per event; failures logged and skipped
+
+ Step 2 response is parsed with the identical bookmaker→market→outcome loop
+ already in the function — only the HTTP call changes.
+
+ Why commenceTimeFrom/To: limits events to upcoming games only; avoids fetching
+ a full past-season list; keeps credit cost predictable.
+
+ Note on regions: The per-event /events/{id}/odds endpoint accepts bookmakers
+ to filter by specific sportsbook keys. regions is not needed and not sent (it would
+ be ignored anyway when bookmakers is specified, per Odds API docs).
+
+ ---
+ Credit Cost
+
+ ┌─────────────────────────────────────────────────────┬──────────────────────────────────┐
+ │                      Scenario                       │             Credits              │
+ ├─────────────────────────────────────────────────────┼──────────────────────────────────┤
+ │ Old code (every call)                               │ 1 credit then 422 — wasted       │
+ ├─────────────────────────────────────────────────────┼──────────────────────────────────┤
+ │ New code, Opening Day (~15 games)                   │ 1 (events) + 15 (per-event) = 16 │
+ ├─────────────────────────────────────────────────────┼──────────────────────────────────┤
+ │ New code, mid-season full slate (~30 games, capped) │ 1 + 20 = 21 (MAX_EVENTS cap)     │
+ ├─────────────────────────────────────────────────────┼──────────────────────────────────┤
+ │ Each additional Props page load (no cache)          │ same 16–21 credits               │
+ └─────────────────────────────────────────────────────┴──────────────────────────────────┘
+
+ No caching is added in this fix. The credit cost per load is real. If credit burn becomes
+ a concern, add a @st.cache_data(ttl=300) wrapper in app.py as a follow-up.
+
+ ---
+ Exact Code Change
+
+ Add one constant after BOOK_KEY_MAP (line ~50):
+ _MAX_EVENTS = 20
+
+ Replace fetch_all_upcoming_hr_props body — keep the first two guard blocks
+ (if not ODDS_API_KEY and if not books), then replace the rest:
+
+ from datetime import datetime, timezone, timedelta
+
+ now = datetime.now(timezone.utc)
+ events_url = f"{ODDS_API_BASE}/baseball_mlb/events"
+ events_params = {
+     "apiKey": ODDS_API_KEY,
+     "dateFormat": "iso",
+     "commenceTimeFrom": now.strftime("%Y-%m-%dT%H:%M:%SZ"),
+     "commenceTimeTo": (now + timedelta(days=7)).strftime("%Y-%m-%dT%H:%M:%SZ"),
+ }
+
+ _diag_log.info("[upcoming_hr_props] Step1 events window=now+7d")
+ try:
+     r1 = requests.get(events_url, params=events_params, timeout=30)
+     _diag_log.info(
+         "[upcoming_hr_props] events HTTP %s | remaining=%s",
+         r1.status_code, r1.headers.get("x-requests-remaining", "?"),
+     )
+     r1.raise_for_status()
+ except requests.HTTPError as exc:
+     body = (exc.response.text[:300] if exc.response is not None else "")
+     raise RuntimeError(
+         f"Odds API events list HTTP {exc.response.status_code}: {body}"
+     ) from exc
+ except requests.RequestException as exc:
+     raise RuntimeError(f"Odds API events network error: {exc}") from exc
+
+ events = r1.json()
+ _diag_log.info(
+     "[upcoming_hr_props] events found=%d (cap=%d)", len(events), _MAX_EVENTS
+ )
+ events = events[:_MAX_EVENTS]
+
+ rows: list[dict[str, Any]] = []
+ for event in events:
+     event_id = str(event.get("id", "") or "")
+     if not event_id:
+         continue
+     away_team = str(event.get("away_team", "") or "")
+     home_team = str(event.get("home_team", "") or "")
+     commence_time = str(event.get("commence_time", "") or "")
+
+     odds_url = f"{ODDS_API_BASE}/baseball_mlb/events/{event_id}/odds"
+     odds_params = {
+         "apiKey": ODDS_API_KEY,
+         "markets": "batter_home_runs",
+         "bookmakers": ",".join(books),
+         "oddsFormat": "american",
+         "dateFormat": "iso",
+     }
+     try:
+         r2 = requests.get(odds_url, params=odds_params, timeout=30)
+         r2.raise_for_status()
+     except (requests.HTTPError, requests.RequestException) as exc:
+         _diag_log.warning(
+             "[upcoming_hr_props] event %s@%s odds failed: %s",
+             away_team, home_team, exc,
+         )
+         continue
+
+     event_data = r2.json()
+     bookmakers = (
+         event_data.get("bookmakers", []) if isinstance(event_data, dict) else []
+     )
+     _diag_log.info(
+         "[upcoming_hr_props] %s@%s books=%s",
+         away_team, home_team, [b.get("key") for b in bookmakers],
+     )
+
+     for bookmaker in bookmakers:
+         book_key = str(bookmaker.get("key", "") or "")
+         book_name = BOOK_KEY_MAP.get(book_key, book_key)
+
+         for market in bookmaker.get("markets", []) or []:
+             market_key = str(market.get("key", "") or "")
+             if market_key != "batter_home_runs":
+                 continue
+             market_name = MARKET_NAME_MAP.get(market_key, market_key)
+
+             for outcome in market.get("outcomes", []) or []:
+                 player_name_raw = str(
+                     outcome.get("description", "") or outcome.get("name", "") or ""
+                 ).strip()
+                 if not player_name_raw:
+                     continue
+
+                 price = outcome.get("price")
+                 if price is None:
+                     continue
+
+                 rows.append(
+                     {
+                         "provider": self.provider_name,
+                         "event_id": event_id,
+                         "commence_time": commence_time,
+                         "away_team": away_team,
+                         "home_team": home_team,
+                         "sportsbook": book_name,
+                         "sportsbook_key": book_key,
+                         "market_key": market_key,
+                         "market": market_name,
+                         "player_name_raw": player_name_raw,
+                         "player_name": map_odds_name_to_model_name(player_name_raw),
+                         "odds_american": int(price),
+                         "line": _safe_float(outcome.get("point")),
+                     }
+                 )
+
+ _diag_log.info("[upcoming_hr_props] total rows=%d", len(rows))
+ return pd.DataFrame(rows)
+
+ The inner bookmaker/market/outcome loop is identical to the existing code — only
+ the outer HTTP layer changed. Downstream schema is preserved exactly.
+
+ ---
+ Summary
+
+ ┌───────────────────────────┬───────────────────────────────────────────────────────────────────┐
+ │           What            │                              Answer                               │
+ ├───────────────────────────┼───────────────────────────────────────────────────────────────────┤
+ │ Files changed             │ 1: data/provider_theoddsapi.py                                    │
+ ├───────────────────────────┼───────────────────────────────────────────────────────────────────┤
+ │ Functions changed         │ 1: fetch_all_upcoming_hr_props                                    │
+ ├───────────────────────────┼───────────────────────────────────────────────────────────────────┤
+ │ Signature changed         │ No                                                                │
+ ├───────────────────────────┼───────────────────────────────────────────────────────────────────┤
+ │ Live props affected       │ No — fetch_live_prop_odds not touched                             │
+ ├───────────────────────────┼───────────────────────────────────────────────────────────────────┤
+ │ Downstream schema changed │ No — identical row dicts                                          │
+ ├───────────────────────────┼───────────────────────────────────────────────────────────────────┤
+ │ Why it fixes 422          │ Per-event endpoint accepts batter_home_runs; aggregate does not   │
+ ├───────────────────────────┼───────────────────────────────────────────────────────────────────┤
+ │ Credit tradeoff           │ ~16 credits/call on Opening Day vs 1 (which 422'd and was wasted) │
+ └───────────────────────────┴───────────────────────────────────────────────────────────────────┘
+╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌
+
+ Claude has written up a plan and is ready to execute. Would you like to proceed?
+
+ ❯ 1. Yes, auto-accept edits
+   2. Yes, manually approve edits
+   3. Type here to tell Claude what to change