minor optimizations (#13)

* makes roster warning debug. adds dataframe methods

- removes relative imports

* bumps version

* fix

pychadwick/__init__.py

@@ -1,7 +1,8 @@
 from pandas import Int32Dtype
 from pychadwick.libutils import ChadwickLibrary
 
-_version = "0.2.1"
+
+_version = "0.2.2"
 
 __all__ = ["ChadwickLibrary"]
 

pychadwick/chadwick.py

@@ -13,12 +13,12 @@
 
 import pandas as pd
 
-from .league import CWLeague
-from .game import CWGame
-from .gameiter import CWGameIterator
-from .roster import CWRoster
-from .utils import CWEventFieldStruct
-from . import EVENT_DATA_TYPES, ChadwickLibrary
+from pychadwick.league import CWLeague
+from pychadwick.game import CWGame
+from pychadwick.gameiter import CWGameIterator
+from pychadwick.roster import CWRoster
+from pychadwick.utils import CWEventFieldStruct
+from pychadwick import EVENT_DATA_TYPES, ChadwickLibrary
 
 
 class Chadwick:
@@ -193,11 +193,11 @@ def process_game(self, game_ptr, roster_visitor=None, roster_home=None):
         gameiter = self.cw_gameiter_create(game_ptr)
 
         if not roster_visitor:
-            logging.warning("roster for %s is undefined.", "visitor")
+            logging.debug("roster for %s is undefined.", "visitor")
             roster_visitor = pointer(CWRoster())
 
         if not roster_home:
-            logging.warning("roster for %s is undefined.", "home")
+            logging.debug("roster for %s is undefined.", "home")
             roster_home = pointer(CWRoster())
 
         event_str = create_string_buffer(b" ", 4096)
@@ -229,6 +229,17 @@ def convert_data_frame_types(df, data_type_mapping):
                     raise TypeError
         return df
 
+    def games_to_dataframe(self, games, data_type_mapping=None):
+        if data_type_mapping is None:
+            data_type_mapping = EVENT_DATA_TYPES
+        dfs = [
+            pd.DataFrame(list(self.process_game(game_ptr)), dtype="f8")
+            for game_ptr in games
+        ]
+        return self.convert_data_frame_types(
+            pd.concat(dfs, axis=0, ignore_index=True), data_type_mapping
+        )
+
     def game_to_dataframe(self, game_ptr, data_type_mapping=None):
         if data_type_mapping is None:
             data_type_mapping = EVENT_DATA_TYPES
@@ -237,6 +248,25 @@ def game_to_dataframe(self, game_ptr, data_type_mapping=None):
             data_type_mapping,
         )
 
+    def event_file_to_dataframe(self, event_file, data_type_mapping=None):
+        if data_type_mapping is None:
+            data_type_mapping = EVENT_DATA_TYPES
+        return self.games_to_dataframe(self.games(event_file), data_type_mapping)
+
+    def event_files_to_dataframe(self, event_files, data_type_mapping=None):
+        if data_type_mapping is None:
+            data_type_mapping = EVENT_DATA_TYPES
+        data = []
+        for event_file in event_files:
+            games = self.games(event_file)
+            for game in games:
+                data += list(self.process_game(game))
+        return self.convert_data_frame_types(
+            pd.DataFrame(data, dtype="f8"),
+            data_type_mapping,
+        )
+
+
     def register_function(self, func_name, func_arg_types, func_res_type):
         func = self.libchadwick.__getattr__(func_name)
         func.argtypes = func_arg_types

setup.cfg

@@ -1,5 +1,5 @@
 [bumpversion]
-current_version = 0.2.1
+current_version = 0.2.2
 
 [flake8]
 max-line-length = 90

setup.py

@@ -5,7 +5,7 @@
     long_description = fh.read()
 setup(
     name="pychadwick",
-    version="0.2.1",
+    version="0.2.2",
     author="Ben Dilday",
     author_email="ben.dilday.phd@gmail.com",
     description="Python bindings to the Chadwick library",

tests/pychadwick/test_pychadwick.py → tests/pychadwick/chadwick/test_pychadwick.py

@@ -47,7 +47,10 @@ def test_load_games_to_df(chadwick, team_events):
             f"https://raw.githubusercontent.com/chadwickbureau/retrosheet/master/event/regular/{team_event}"
         )
         games = chadwick.games(event_path)
-        dfs = [chadwick.game_to_dataframe(game) for game in games]
+        df = chadwick.games_to_dataframe(games)
+
+    games = chadwick.games(event_path)
+    df = chadwick.game_to_dataframe(next(games))
 
 
 def test_game_to_csv(chadwick, team_events):