Fixes and formatting

.github/workflows/python-package.yml

@@ -16,7 +16,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        python-version: ["3.9", "3.10", "3.11"]
+        python-version: ["3.10", "3.11", "3.12"]
 
     steps:
     - uses: actions/checkout@v4

.gitignore

@@ -113,4 +113,5 @@ venv.bak/
 
 # Custom
 NOTES
-test.*
+test.*
+*.lock

README.md

@@ -15,15 +15,18 @@ git clone https://github.com/gdifiore/pyball.git
 
 cd pyball
 
-pip install . OR poetry install
+python -m venv .venv
+
+(activate .venv)
+
+poetry install
+
+poetry run post-install
 ```
 
 ## Docs
-
 Read the [docs](https://gdifiore.github.io/pyball/docs/pyball/index.html) for function descriptions.
 
-For examples, look at my [MLBResearch Repo](https://github.com/gdifiore/MLBResearch/blob/main/Parse_BBRef_Table/bbref_table.ipynb) where I test most of the functions.
-
 ## Comments and Suggestions
 Leave any comments or suggestions in [an issue](https://github.com/SummitCode/pyball/issues/new) or directly make make [a pull request](https://github.com/SummitCode/pyball/compare) adding code.
 
@@ -32,4 +35,5 @@ Leave any comments or suggestions in [an issue](https://github.com/SummitCode/py
 `pyball` is licensed under the [MIT license](https://github.com/SummitCode/pyball/blob/master/LICENSE)
 
 ## TODO
-- update documentation
+- update documentation
+- add checks to URLs passd into functions (e.g. team into team parsing)

pyball/__init__.py

@@ -5,4 +5,9 @@
 from .pitching_stats import *
 from .team_batting_stats import *
 from .team_pitching_stats import *
-from .savant import *
+from .savant import *
+
+import subprocess
+
+def post_install():
+    subprocess.run(["playwright", "install"], check=True)

pyball/batting_stats.py

@@ -8,10 +8,10 @@
 
 import pandas as pd
 
-from pyball.utils import readURL
+from pyball.utils import read_url, is_player_url
 
 
-def findBattingTable(soup):
+def _find_batting_table(soup):
     """
     Function to find the batting stats table (Baseball-Reference) in the soup
 
@@ -44,8 +44,11 @@ def batting_stats(url):
     pandas dataframe
         Contains the batting stats for the player
     """
-    soup = readURL(url)
-    table = findBattingTable(soup)
+    if not is_player_url(url):
+        return None
+
+    soup = read_url(url)
+    table = _find_batting_table(soup)
 
     df = pd.read_html(str(table))[0]
 

pyball/pitching_stats.py

@@ -8,10 +8,10 @@
 
 import pandas as pd
 
-from pyball.utils import readURL
+from pyball.utils import read_url, is_player_url
 
 
-def findPitchingTable(soup):
+def _find_pitching_table(soup):
     """
     Function to find the pitching stats table (Baseball-Reference) in the soup
 
@@ -44,8 +44,11 @@ def pitching_stats(url):
     pandas dataframe
         Contains the pitching stats for the player
     """
-    soup = readURL(url)
-    table = findPitchingTable(soup)
+    if not is_player_url(url):
+        return None
+
+    soup = read_url(url)
+    table = _find_pitching_table(soup)
 
     df = pd.read_html(str(table))[0]
 

pyball/playerid_lookup.py

@@ -47,7 +47,7 @@ def download_file(file_number):
         io.StringIO(response.content.decode("utf-8")),
         dtype={"key_sr_nfl": object, "key_sr_nba": object, "key_sr_nhl": object},
     )
-    
+
 @lru_cache(maxsize=1)
 def get_lookup_table():
     """
@@ -100,4 +100,4 @@ def playerid_lookup(last, first=None):
     query_string += f" and name_first == '{first}'" if first else ""
     results = table.query(query_string)
 
-    return results.reset_index(drop=True)
+    return results.reset_index(drop=True)

pyball/savant.py

@@ -8,10 +8,10 @@
 
 import pandas as pd
 
-from pyball.utils import readURL
+from pyball.utils import read_url
 
 
-def findPercentilesTable(soup):
+def _find_percentiles_table(soup):
     """
     Function to find the stat percentiles table (Baseball Savant) in the soup
 
@@ -30,7 +30,7 @@ def findPercentilesTable(soup):
     return table
 
 
-def savantPercentileStats(url):
+def savant_percentile_stats(url):
     """
     Function to return the (Baseball Savant) percentile stats for a player as a pandas dataframe
 
@@ -44,15 +44,15 @@ def savantPercentileStats(url):
     pandas dataframe
         Contains the percentile stats for the player
     """
-    soup = readURL(url)
-    table = findPercentilesTable(soup)
+    soup = read_url(url)
+    table = _find_percentiles_table(soup)
 
     df = pd.read_html(str(table))[0]
 
     return df.dropna(how="all")
 
 
-def findStatcastPitchingStatsTable(soup):
+def _find_statcast_pitching_stats_table(soup):
     """
     Function to find the statcast pitching stats table (Baseball Savant) in the soup
 
@@ -67,13 +67,16 @@ def findStatcastPitchingStatsTable(soup):
         Contains the html of the statcast pitching stats table
     """
     div = soup.find("div", id="statcast_stats_pitching")
+    if div is None:
+        print("Not a pitcher page")
+        return None
     # get table inside div
     table = div.find("table")
 
     return table
 
 
-def savantPitchingStatcastStats(url):
+def savant_pitching_statcast_stats(url):
     """
     Function to return the (Baseball Savant) pitching stats for a player as a pandas dataframe
 
@@ -87,16 +90,19 @@ def savantPitchingStatcastStats(url):
     pandas dataframe
         Contains the savant pitching stats for the player
     """
-    soup = readURL(url)
-    table = findStatcastPitchingStatsTable(soup)
+    soup = read_url(url)
+    table = _find_statcast_pitching_stats_table(soup)
+
+    if table is None:
+        return None
 
     df = pd.read_html(str(table))[0]
 
     # drop a row of all NA and drop last row of MLB average
     return df.dropna(how="all").drop(df.index[-1])
 
 
-def findStatcastBattingStatsTable(soup):
+def _find_statcast_batting_stats_table(soup):
     """
     Function to find the statcast batting stats table (Baseball Savant) in the soup
 
@@ -112,12 +118,16 @@ def findStatcastBattingStatsTable(soup):
     """
     # Find div with id 'statcast_glance_batter' and get table inside
     div = soup.find("div", id="statcast_glance_batter")
+    if div is None:
+        print("Not a pitcher page")
+        return None
+
     table = div.find("table")
 
     return table
 
 
-def savantBattingStatcastStats(url):
+def savant_batting_statcast_stats(url):
     """
     Function to return the (Baseball Savant) batting stats for a player as a pandas dataframe
 
@@ -131,16 +141,19 @@ def savantBattingStatcastStats(url):
     pandas dataframe
         Contains the savant batting stats for the player
     """
-    soup = readURL(url)
-    table = findStatcastBattingStatsTable(soup)
+    soup = read_url(url)
+    table = _find_statcast_batting_stats_table(soup)
+
+    if table is None:
+        return None
 
     df = pd.read_html(str(table))[0]
 
     # drop a row of all NA and drop last row of MLB average
     return df.dropna(how="all").drop(df.index[-1])
 
 
-def findBattedBallProfileTable(soup):
+def _find_batted_ball_profile_table(soup):
     """
     Function to find the batted ball profile table (Baseball Savant) in the soup
 
@@ -159,7 +172,7 @@ def findBattedBallProfileTable(soup):
     return table
 
 
-def savantBattedBallProfile(url):
+def savant_batted_ball_profile(url):
     """
     Function to return the (Baseball Savant) batted ball profile for a player as a pandas dataframe
 
@@ -173,15 +186,15 @@ def savantBattedBallProfile(url):
     pandas dataframe
         Contains the batted ball profile for the player
     """
-    soup = readURL(url)
-    table = findBattedBallProfileTable(soup)
+    soup = read_url(url)
+    table = _find_batted_ball_profile_table(soup)
 
     df = pd.read_html(str(table))[0]
 
     return df.dropna(how="all")
 
 
-def findPitchTrackingTable(soup):
+def _find_pitch_tracking_table(soup):
     """
     Function to find the pitch tracking table (Baseball Savant) in the soup
 
@@ -200,7 +213,7 @@ def findPitchTrackingTable(soup):
     return table
 
 
-def savantPitchTracking(url):
+def savant_pitch_tracking(url):
     """
     Function returns the (Baseball Savant) pitch-specific results for a player as a pandas dataframe
 
@@ -214,8 +227,8 @@ def savantPitchTracking(url):
     pandas dataframe
         Contains the pitch-specific results for the player
     """
-    soup = readURL(url)
-    table = findPitchTrackingTable(soup)
+    soup = read_url(url)
+    table = _find_pitch_tracking_table(soup)
 
     df = pd.read_html(str(table))[0]
 

pyball/team_batting_stats.py

@@ -8,10 +8,10 @@
 
 import pandas as pd
 
-from pyball.utils import readURL
+from pyball.utils import read_url, is_team_url
 
 
-def findTeamBattingTable(soup):
+def _find_team_batting_table(soup):
     """
     Function to find the team batting stats table (Baseball-Reference) in the soup
 
@@ -44,8 +44,11 @@ def team_batting_stats(url):
     pandas dataframe
         containing the team batting stats for the team
     """
-    soup = readURL(url)
-    table = findTeamBattingTable(soup)
+    if not is_team_url(url):
+        return None
+
+    soup = read_url(url)
+    table = _find_team_batting_table(soup)
 
     df = pd.read_html(str(table))[0]
 

pyball/team_pitching_stats.py

@@ -9,10 +9,10 @@
 
 import pandas as pd
 
-from pyball.utils import readURL
+from pyball.utils import read_url, is_team_url
 
 
-def findTeamPitchingTable(soup):
+def _find_team_pitching_table(soup):
     """
     Function to find the team pitching stats table (Baseball-Reference) in the soup
 
@@ -45,8 +45,11 @@ def team_pitching_stats(url):
     pandas dataframe
         containing the team batting stats for the team
     """
-    soup = readURL(url)
-    table = findTeamPitchingTable(soup)
+    if not is_team_url(url):
+        return None
+
+    soup = read_url(url)
+    table = _find_team_pitching_table(soup)
 
     df = pd.read_html(str(table))[0]