wip

Scripts/sportstime_parser/scrapers/nba.py

@@ -95,9 +95,11 @@ class NBAScraper(BaseScraper):
         BR organizes games by month with separate pages.
         Format: https://www.basketball-reference.com/leagues/NBA_YYYY_games-month.html
         where YYYY is the ending year of the season.
+        Bails early if first few months have no data (season doesn't exist).
         """
         all_games: list[RawGameData] = []
         end_year = self.season + 1
+        consecutive_empty_months = 0
 
         for month in BR_MONTHS:
             url = self._get_source_url("basketball_reference", month=month, year=end_year)
@@ -105,13 +107,23 @@ class NBAScraper(BaseScraper):
             try:
                 html = self.session.get_html(url)
                 games = self._parse_basketball_reference(html, url)
-                all_games.extend(games)
-                self._logger.debug(f"Found {len(games)} games in {month}")
+
+                if games:
+                    all_games.extend(games)
+                    consecutive_empty_months = 0
+                    self._logger.debug(f"Found {len(games)} games in {month}")
+                else:
+                    consecutive_empty_months += 1
 
             except Exception as e:
                 # Some months may not exist (e.g., no games in August)
                 self._logger.debug(f"No data for {month}: {e}")
-                continue
+                consecutive_empty_months += 1
+
+            # If first 3 months (Oct, Nov, Dec) all have no data, season doesn't exist
+            if consecutive_empty_months >= 3 and not all_games:
+                self._logger.info(f"No games found in first {consecutive_empty_months} months, season likely doesn't exist")
+                break
 
         return all_games
 
@@ -247,8 +259,11 @@ class NBAScraper(BaseScraper):
 
         ESPN API returns games for a specific date range.
         We iterate through each day of the season.
+        Bails out early if no games found after checking first month.
         """
         all_games: list[RawGameData] = []
+        consecutive_empty_days = 0
+        max_empty_days = 45  # Bail after ~1.5 months of no games
 
         for year, month in self._get_season_months():
             # Get number of days in month
@@ -267,10 +282,25 @@ class NBAScraper(BaseScraper):
 
                     data = self.session.get_json(url)
                     games = self._parse_espn_response(data, url)
-                    all_games.extend(games)
+
+                    if games:
+                        all_games.extend(games)
+                        consecutive_empty_days = 0
+                    else:
+                        consecutive_empty_days += 1
+
+                    # Bail early if no games found for a long stretch
+                    if consecutive_empty_days >= max_empty_days:
+                        self._logger.info(f"No games found for {max_empty_days} consecutive days, stopping ESPN scrape")
+                        return all_games
 
                 except Exception as e:
                     self._logger.debug(f"ESPN error for {year}-{month}-{day}: {e}")
+                    consecutive_empty_days += 1
+
+                    if consecutive_empty_days >= max_empty_days:
+                        self._logger.info(f"Too many consecutive failures, stopping ESPN scrape")
+                        return all_games
                     continue
 
         return all_games