Create create_mlbplayers.py

2024-05-05 14:36:46 -05:00 · 2024-05-05 14:36:46 -05:00 · 43c9fd0e32
commit 43c9fd0e32
parent 4a70faafbc
1 changed files with 124 additions and 0 deletions
--- a/create_mlbplayers.py
+++ b/create_mlbplayers.py
@ -0,0 +1,124 @@
+import asyncio
+import datetime
+import logging
+import random
+import sys
+
+from creation_helpers import get_all_pybaseball_ids, get_args, sanitize_name
+from db_calls import db_get, db_patch, player_desc, db_post
+
+date = f'{datetime.datetime.now().year}-{datetime.datetime.now().month}-{datetime.datetime.now().day}'
+log_level = logging.INFO
+logging.basicConfig(
+    filename=f'logs/{date}.log',
+    format='%(asctime)s - card-creation - %(levelname)s - %(message)s',
+    level=log_level
+)
+
+
+async def main(args):
+    arg_data = get_args(args)
+    post_players = True if 'post_players' not in arg_data or arg_data['post_players'].lower() == 'true' else False
+    max_run = 9999999 if 'max_run' not in arg_data else int(arg_data['max_run'])
+
+    p_query = await db_get('players')
+    if p_query['count'] == 0:
+        print(f'Well fuck. No players were found.')
+        return
+
+    player_cache = {
+        # bbref_id: mlbplayer_id
+    }
+
+    second_touch_players = []
+    second_touch_bbref_ids = []
+    new_mlbplayers = []
+
+    async def set_mlbplayer(pd_player: dict):
+        if post_players:
+            await db_patch(
+                'players',
+                object_id=pd_player['player_id'],
+                params=[('mlbplayer_id', player_cache[pd_player['bbref_id']])]
+            )
+            logging.info(f'posting update to API')
+        else:
+            logging.info(f'not posting update to API')
+
+    async def search_and_match(pd_player: dict):
+        m_query = await db_get('mlbplayers', params=[('key_bbref', pd_player['bbref_id'])])
+        if m_query['count'] > 0:
+            this_mlb = m_query['players'][0]
+            player_cache[pd_player['bbref_id']] = this_mlb['id']
+            pd_player['mlbplayer_id'] = this_mlb['id']
+            logging.info(
+                f'Set {player_desc(pd_player)} / player_id {pd_player["player_id"]} mlbplayer_id to cached '
+                f'{player_cache[pd_player["bbref_id"]]}')
+            await set_mlbplayer(pd_player)
+
+        else:
+            all_ids = get_all_pybaseball_ids([pd_player['bbref_id']], 'bbref')
+            if all_ids is not None:
+                if all_ids['key_bbref'] not in second_touch_bbref_ids:
+                    new_mlbplayers.append({
+                        'first_name': sanitize_name(all_ids['name_first']).title(),
+                        'last_name': sanitize_name(all_ids['name_last']).title(),
+                        'key_mlbam': int(all_ids['key_mlbam']),
+                        'key_fangraphs': int(all_ids['key_fangraphs']),
+                        'key_bbref': all_ids['key_bbref'],
+                        'key_retro': all_ids['key_retro'],
+                        'offense_col': random.randint(1, 3)
+                    })
+                    second_touch_bbref_ids.append(all_ids['key_bbref'])
+                else:
+                    print(f'Duplicate bbref ID: {all_ids['key_bbref']}; not adding to new_mlbplayers')
+                second_touch_players.append(pd_player)
+                logging.info(f'Added {player_desc(pd_player)} to new_mlbplayers')
+
+            else:
+                logging.info(f'No return from pybaseball for {player_desc(x)} / player_id {x["player_id"]}')
+                print(f'No data for {player_desc(x)}')
+
+    start_time = datetime.datetime.now()
+    count = 0
+    for x in p_query['players']:
+        count += 1
+        if count > max_run:
+            print(f'Count ({count}) exceeded max_run ({max_run})')
+            break
+        if count % 10 == 0:
+            print(f'Processing player #{count}...')
+
+        if x['mlbplayer'] is None:
+            if x['bbref_id'] in player_cache:
+                await set_mlbplayer(x)
+            else:
+                await search_and_match(x)
+
+        else:
+            logging.info(f'{player_desc(x)} / player_id {x["player_id"]} / bbref_id {x["bbref_id"]} not found')
+
+    start_time_two = datetime.datetime.now()
+    run_time = start_time_two - start_time
+    print(f'Done first pass ({run_time}s), posting {len(new_mlbplayers)} now...')
+    count = 0
+
+    resp = await db_post(endpoint='mlbplayers', payload={'players': new_mlbplayers})
+    if resp is not None:
+        print(f'Starting second pass now...')
+        for x in second_touch_players:
+            count += 1
+            if count % 10 == 0:
+                print(f'Processing player #{count}...')
+            if x['bbref_id'] in player_cache:
+                await set_mlbplayer(x)
+            else:
+                await search_and_match(x)
+
+    start_time_three = datetime.datetime.now()
+    p_run_time = datetime.datetime.now() - start_time_two
+    print(f'Done second pass ({p_run_time}s), total run time: {datetime.datetime.now() - start_time}s')
+
+
+if __name__ == '__main__':
+    asyncio.run(main(sys.argv[1:]))