Exclude pbp participation for any years where the data file isn't ava…

…ilable (#107)

* Exclude pbp participation data if the file isn't available

* Fixed incorrect error import

nfl_data_py/__init__.py

@@ -10,6 +10,7 @@
 import numpy
 import pandas
 import appdirs
+from urllib.error import HTTPError
 
 # module level doc string
 __doc__ = """
@@ -143,13 +144,20 @@ def import_pbp_data(
                 raw = pandas.DataFrame(data)
                 raw['season'] = year
 
-                if all([include_participation, year >= 2016, not cache]):
+
+                if include_participation and not cache:
                     path = r'https://github.com/nflverse/nflverse-data/releases/download/pbp_participation/pbp_participation_{}.parquet'.format(year)
-                    partic = pandas.read_parquet(path)
-                    raw = raw.merge(partic,
-                                    how='left',
-                                    left_on=['play_id','game_id'],
-                                    right_on=['play_id','nflverse_game_id'])
+
+                    try:
+                        partic = pandas.read_parquet(path)
+                        raw = raw.merge(
+                            partic,
+                            how='left',
+                            left_on=['play_id','game_id'],
+                            right_on=['play_id','nflverse_game_id']
+                        )
+                    except HTTPError:
+                        pass
 
                 pbp_data.append(raw)
                 print(str(year) + ' done.')
@@ -158,8 +166,10 @@ def import_pbp_data(
                 print(e)
                 print('Data not available for ' + str(year))
 
-    if pbp_data:
-        plays = pandas.concat(pbp_data).reset_index(drop=True)
+    if not pbp_data:
+        return pandas.DataFrame()
+
+    plays = pandas.concat(pbp_data, ignore_index=True)
 
     # converts float64 to float32, saves ~30% memory
     if downcast:

nfl_data_py/tests/nfl_test.py

@@ -9,17 +9,17 @@
 
 
 class test_pbp(TestCase):
+    pbp = nfl.import_pbp_data([2020])
+
     def test_is_df_with_data(self):
-        s = nfl.import_pbp_data([2020])
-        self.assertIsInstance(s, pd.DataFrame)
-        self.assertTrue(len(s) > 0)
+        self.assertIsInstance(self.pbp, pd.DataFrame)
+        self.assertTrue(len(self.pbp) > 0)
 
     def test_is_df_with_data_thread_requests(self):
         s = nfl.import_pbp_data([2020, 2021], thread_requests=True)
         self.assertIsInstance(s, pd.DataFrame)
         self.assertTrue(len(s) > 0)
 
-
     def test_uses_cache_when_cache_is_true(self):
         cache = Path(__file__).parent/f"tmpcache-{random.randint(0, 10000)}"
         self.assertRaises(
@@ -33,6 +33,21 @@ def test_uses_cache_when_cache_is_true(self):
         self.assertIsInstance(data, pd.DataFrame)
 
         shutil.rmtree(cache)
+
+    def test_includes_participation_by_default(self):
+        self.assertIn("offense_players", self.pbp.columns)
+
+    def test_excludes_participation_when_requested(self):
+        data = nfl.import_pbp_data([2020], include_participation=False)
+        self.assertIsInstance(self.pbp, pd.DataFrame)
+        self.assertTrue(len(self.pbp) > 0)
+        self.assertNotIn("offense_players", data.columns)
+
+    def test_excludes_participation_if_not_available(self):
+        data = nfl.import_pbp_data([2024])
+        self.assertIsInstance(self.pbp, pd.DataFrame)
+        self.assertTrue(len(self.pbp) > 0)
+        self.assertNotIn("offense_players", data.columns)
 
 
 class test_weekly(TestCase):