Update

src/data/apinhle/data_pull_plays.py

@@ -29,62 +29,62 @@
 # ---------------------------------------------------
 
 for iter_year in [iter_year]: # or iter_years
-
-    # Load the previous game stats, if exist
-    try:
-
-        # Pull all game lists
+    try: # Pull all game lists, if exists
         gamecode = pd.read_csv(f"./latest/box/{iter_year}_box.csv")
+
+        # Load the previous game stats, if exist
+        try:
+            # If previously pulled data exist
+            df_playbyplay_exist = pd.read_csv(f"./latest/play/{iter_year}_playbyplay.csv")
+            df_playbyplay_player_exist = pd.read_csv(f"./latest/play/{iter_year}_playbyplay_player.csv")
+            # Unique of all existing game records
+            gameids_exist = df_playbyplay_exist["gameid"].unique()
+            # Remove the existing game records
+            # Pick up games with newest data points
+            gamecode = gamecode.loc[~gamecode["gameid"].isin(gameids_exist), :]
+            # NOTE: This process does not account for any record revisions
+            idx_exist = True
+        except:
+            # New data needed, no need to append the old one
+            idx_exist = False
 
-        # If previously pulled data exist
-        df_playbyplay_exist = pd.read_csv(f"./latest/play/{iter_year}_playbyplay.csv")
-        df_playbyplay_player_exist = pd.read_csv(f"./latest/play/{iter_year}_playbyplay_player.csv")
-        # Unique of all existing game records
-        gameids_exist = df_playbyplay_exist["gameid"].unique()
-        # Remove the existing game records
-        # Pick up games with newest data points
-        gamecode = gamecode.loc[~gamecode["gameid"].isin(gameids_exist), :]
-        # NOTE: This process does not account for any record revisions
-        idx_exist = True
-    except:
-        # New data needed, no need to append the old one
-        idx_exist = False
-
-    # ---------------------------------------------------
-    # Pull team/game lists of the games for the season
-    '''
-    Pulling full season game would take a long time. If some game records were already pulled, 
-        I recommend to skip those game records
-    '''
-    if len(gamecode["gameid"]) != 0:
-        # At least one records need to be pulled
-        df_playbyplay = []
-        df_playerinfo = []
-        for _, row in gamecode.iterrows():
-            # Pull game's play-by-play stat
-            r = requests.get(url=f'https://api-web.nhle.com/v1/gamecenter/{row.gameid}/play-by-play')
-
-            iter_playbyplay, iter_player = proc_playbyplay_clean(r, row)
-            # Append to save
-            df_playbyplay.append(iter_playbyplay)
-            df_playerinfo.append(iter_player)
-            # Pause to play safe with the API
-            time.sleep(1)
+        # ---------------------------------------------------
+        # Pull team/game lists of the games for the season
+        '''
+        Pulling full season game would take a long time. If some game records were already pulled, 
+            I recommend to skip those game records
+        '''
+        if len(gamecode["gameid"]) != 0:
+            # At least one records need to be pulled
+            df_playbyplay = []
+            df_playerinfo = []
+            for _, row in gamecode.iterrows():
+                # Pull game's play-by-play stat
+                r = requests.get(url=f'https://api-web.nhle.com/v1/gamecenter/{row.gameid}/play-by-play')
+
+                iter_playbyplay, iter_player = proc_playbyplay_clean(r, row)
+                # Append to save
+                df_playbyplay.append(iter_playbyplay)
+                df_playerinfo.append(iter_player)
+                # Pause to play safe with the API
+                time.sleep(1)
 
-        # Save, full data
-        playbyplay = pd.concat(df_playbyplay)
-        df_playerinfo = pd.concat(df_playerinfo)
+            # Save, full data
+            playbyplay = pd.concat(df_playbyplay)
+            df_playerinfo = pd.concat(df_playerinfo)
 
-        if idx_exist:
-            # If the old record exists, append the old record
-            playbyplay = pd.concat([df_playbyplay_exist, df_playbyplay], axis=0)
-            df_playerinfo = pd.concat([df_playbyplay_player_exist, df_playerinfo], axis=0)
-
-        # Save data
-        playbyplay.to_csv(f"./latest/play/{iter_year}_playbyplay.csv", index=False)
-        df_playerinfo.to_csv(f"./latest/play/{iter_year}_playbyplay_player.csv", index=False)
-    else:
-        # No records need to be pulled
-        print("All records currently existing, no need to pull records")
+            if idx_exist:
+                # If the old record exists, append the old record
+                playbyplay = pd.concat([df_playbyplay_exist, df_playbyplay], axis=0)
+                df_playerinfo = pd.concat([df_playbyplay_player_exist, df_playerinfo], axis=0)
+
+            # Save data
+            playbyplay.to_csv(f"./latest/play/{iter_year}_playbyplay.csv", index=False)
+            df_playerinfo.to_csv(f"./latest/play/{iter_year}_playbyplay_player.csv", index=False)
+        else:
+            # No records need to be pulled
+            print("All records currently existing, no need to pull records")
+    except: # Season data not yet exist
+        print("The season data not yet exist. Exit the process safely.")
 
-print("Good bye.")
+print("au revoir.")