remove a bunch of warnings

pyproject.toml

@@ -9,16 +9,16 @@ readme = "README.md"
 python = "^3.12"
 psycopg2 = "^2.6.2"
 luigi = "^3.3.0"
-pandas = "^2.0.3"
+pandas = "^2.2.2"
 chess = "^1.10.0"
 python-lichess = "^0.10"
 sendgrid = "6.0.5"
 stockfish = "3.5.0"
-sqlalchemy = "1.3.0"
-seaborn = "^0.10.0"
-matplotlib = "3.8.4"
+sqlalchemy = "^2.0.32"
+seaborn = "^0.13.0"
 beautifulsoup4 = "^4.8.0"
 scikit-learn = "^1.0.1"
+more-itertools = "^10.4.0"
 
 [tool.poetry.group.dev.dependencies]
 pytest = "^8.0.1"

src/pipeline_import/models.py

@@ -23,11 +23,11 @@ def create_wp_features(df: pd.DataFrame) -> pd.DataFrame:
     df = df[df['clock'] != -1]
 
     df['opponent_clock'] = df.groupby(['game_link'])['clock'].shift(-1)
-    df['opponent_clock'].fillna(df['opponent_clock'].shift(2), inplace=True)
+    df['opponent_clock'] = df['opponent_clock'].fillna(df['opponent_clock'].shift(2))  # noqa
 
     # in situations where there were only one or two moves,
     # fill with the clock time
-    df['opponent_clock'].fillna(df['clock'], inplace=True)
+    df['opponent_clock'] = df['opponent_clock'].fillna(df['clock'])
 
     # start with white
     df['player_to_move'] = df['half_move'] % 2

src/pipeline_import/plots.py

@@ -3,6 +3,7 @@
 import os
 
 from matplotlib import use
+from more_itertools import one
 from seaborn import set as sns_set
 
 
@@ -78,9 +79,9 @@ def make_elo_by_weekday_plot(elo, fig_loc, filename):
                  xticks=range(0, 7),
                  )
 
-        min_last_day = elo[-1:]['min'].values
-        max_last_day = elo[-1:]['max'].values
-        mean_last_day = elo[-1:]['mean'].values
+        min_last_day = one(elo[-1:]['min'].values)
+        max_last_day = one(elo[-1:]['max'].values)
+        mean_last_day = one(elo[-1:]['mean'].values)
 
         # annotate the lines individually
         ax.annotate('min',

src/pipeline_import/transforms.py

@@ -19,7 +19,6 @@
     to_numeric,
     to_timedelta,
 )
-from psycopg2 import connect
 from utils.types import Json, Visitor
 
 
@@ -107,7 +106,7 @@ def fix_provisional_columns(json_df: pd.DataFrame) -> pd.DataFrame:
     for side in ['black', 'white']:
         col = f'players_{side}_provisional'
         if col in json_df.columns:
-            json_df[col].fillna(False, inplace=True)
+            json_df[col] = json_df[col].fillna(False)
         else:
             json_df[col] = False
     return json_df
@@ -288,7 +287,7 @@ def get_elo_by_weekday(df, category='blitz'):
     df['weekday_played'] = df['datetime_played'].dt.weekday
 
     # change to sunday-first, not monday-first
-    df['weekday_played'].replace(6, -1, inplace=True)
+    df['weekday_played'] = df['weekday_played'].replace(6, -1)
     df['weekday_played'] += 1  # what a dumb way of fixing this
 
     elo = (df.groupby('weekday_played')
@@ -307,18 +306,16 @@ def get_elo_by_weekday(df, category='blitz'):
 
 
 def get_weekly_data(pg_cfg, player):
-    db_connection_string = 'postgresql://{}:{}@{}:{}/{}'
-
-    with connect(db_connection_string.format(pg_cfg.read_user,
-                                             pg_cfg.read_password,
-                                             pg_cfg.host,
-                                             pg_cfg.port,
-                                             pg_cfg.database)) as con:
-
-        sql = f"""SELECT * from chess_games
-                  WHERE player = '{player}'
-                  AND datetime_played >= now()::date - interval '7 days';
-               """
-
-        df = read_sql_query(sql, con)
+    db_conn_string = 'postgresql+psycopg2://{}:{}@{}:{}/{}'
+    db_conn_string = db_conn_string.format(pg_cfg.read_user,
+                                           pg_cfg.read_password,
+                                           pg_cfg.host,
+                                           pg_cfg.port,
+                                           pg_cfg.database)
+    sql = f"""SELECT * from chess_games
+              WHERE player = '{player}'
+              AND datetime_played >= now()::date - interval '7 days';
+           """
+
+    df = read_sql_query(sql, db_conn_string)
     return df

src/utils/db.py

@@ -1,25 +1,19 @@
 
 import pandas as pd
-import psycopg2
 from pipeline_import.configs import postgres_cfg
 
 
 def run_remote_sql_query(sql, **params) -> pd.DataFrame:
     pg_cfg = postgres_cfg()
-    user = pg_cfg.user
-    password = pg_cfg.password
-    host = pg_cfg.host
-    port = pg_cfg.port
-    database = pg_cfg.database
-
-    db = psycopg2.connect(host=host,
-                          database=database,
-                          user=user,
-                          password=password,
-                          port=port,
-                          )
-
-    df: pd.DataFrame = pd.read_sql_query(sql, db, params=params)
+
+    db_conn_string = 'postgresql+psycopg2://{}:{}@{}:{}/{}'
+    db_conn_string = db_conn_string.format(pg_cfg.user,
+                                           pg_cfg.password,
+                                           pg_cfg.host,
+                                           pg_cfg.port,
+                                           pg_cfg.database)
+
+    df: pd.DataFrame = pd.read_sql_query(sql, db_conn_string, params=params)
 
     return df
 

src/vendors/stockfish.py

@@ -84,6 +84,8 @@ def get_evals(df: pd.DataFrame,
                                      errors='coerce')
 
     df.dropna(inplace=True)
-    df = pd.concat([df, db_evaluations], axis=0, ignore_index=True)
+
+    if not db_evaluations.empty:
+        df = pd.concat([df, db_evaluations], axis=0, ignore_index=True)
 
     return df

tests/__snapshots__/test_plots.ambr

@@ -1,4 +1,7 @@
 # serializer version: 1
+# name: test_color_stats_plot
+  '4a9c086a905bde320cfeed437e131934'
+# ---
 # name: test_create_newsletter
   Mail(
     asm=None,
@@ -81,6 +84,12 @@
     tracking_settings=None,
   )
 # ---
+# name: test_elo_by_weekday_plot
+  '574de65d835cd72b0c470efd78483315'
+# ---
+# name: test_elo_by_weekday_without_games
+  '2c72ebcedd63e778d31e3ca25711e9d3'
+# ---
 # name: test_generate_elo_by_weekday_text_generic
   "This week, your highest elo in bullet was 300 and your lowest elo was 100. <br><img alt='Elo by weekday' src='cid:elo-by-weekday'><br>"
 # ---

tests/test_plots.py

@@ -17,7 +17,7 @@
 )
 
 
-def test_color_stats_plot():
+def test_color_stats_plot(snapshot):
 
     multiindex = pd.MultiIndex.from_arrays([['blitz', 'blitz', 'bullet'],
                                             ['white', 'black', 'black']],
@@ -41,14 +41,12 @@ def test_color_stats_plot():
     with open(file_loc, 'rb') as f:
         md5 = hashlib.md5(f.read()).hexdigest()
 
-    true_md5 = 'c5db29597bc7936db111444dd8bc35df'
-
-    assert md5 == true_md5
+    assert md5 == snapshot
 
     os.remove(file_loc)
 
 
-def test_elo_by_weekday_plot():
+def test_elo_by_weekday_plot(snapshot):
     elo = pd.DataFrame([[0, 1666, 0, 1666.0, 1666.0],
                         [1, 1685, 7.071, 1680.0, 1690.0],
                         [2, 1685, 7.071, 1680.0, 1690.0],
@@ -73,14 +71,12 @@ def test_elo_by_weekday_plot():
     with open(file_loc, 'rb') as f:
         md5 = hashlib.md5(f.read()).hexdigest()
 
-    true_md5 = 'dad710a3d32903926277ea59a4d1e2cc'
-
-    assert md5 == true_md5
+    assert md5 == snapshot
 
     os.remove(file_loc)
 
 
-def test_elo_by_weekday_without_games():
+def test_elo_by_weekday_without_games(snapshot):
 
     empty_elo = pd.DataFrame([],
                              columns=['weekday_played',
@@ -94,7 +90,7 @@ def test_elo_by_weekday_without_games():
                              )
 
     fig_loc = '.'
-    filename = 'elo_by_weekday_test.png'
+    filename = 'elo_by_weekday_without_games.png'
 
     plots.make_elo_by_weekday_plot(empty_elo, fig_loc, filename)
 
@@ -103,9 +99,7 @@ def test_elo_by_weekday_without_games():
     with open(file_loc, 'rb') as f:
         md5 = hashlib.md5(f.read()).hexdigest()
 
-    true_md5 = '4e93a5e395a23f57b6700aba8d630796'
-
-    assert md5 == true_md5
+    assert md5 == snapshot
 
     os.remove(file_loc)