update versions (#74)

* unpins versions

* test py3.8+ only

* formatting

* make clean

* docker tests

* fixes statcast daily and retrosheet tables

* approx test for trig

.github/workflows/pythonpackage.yml

@@ -9,7 +9,7 @@ jobs:
     strategy:
       max-parallel: 4
       matrix:
-        python-version: [3.6, 3.7, 3.8]
+        python-version: ["3.8", "3.9", "3.10", "3.11"]
 
     steps:
     - uses: actions/checkout@v1

Dockerfile

@@ -0,0 +1,11 @@
+FROM python:3.10
+
+WORKDIR /workdir
+
+COPY requirements.txt /workdir/requirements.txt
+COPY requirements-dev.txt /workdir/requirements-dev.txt
+
+RUN python3.10 -m pip install -r requirements-dev.txt
+RUN python3.10 -m pip install  -r requirements.txt
+
+RUN make clean-data

Makefile

@@ -1,6 +1,6 @@
 .PHONY: lint \
 test-analysis test-data test-markov test \
-clean \
+clean clean-pyc \
 install-dev install \
 dist
 
@@ -39,12 +39,18 @@ clean-docs:
 clean-data:
 	rm -rf pybbda/data/assets/*
 
-clean:
+clean-pyc:
+	find . -name '*.pyc' -exec rm -f {} +
+	find . -name '*.pyo' -exec rm -f {} +
+	find . -name '*__pycache__' -exec rm -fr {} +
+
+clean: clean-pyc
 	rm -fr pybbda.egg-info
 	rm -fr build
 	rm -fr dist
 	rm -fr .pytest_cache
 
+
 dist: clean
 	python setup.py bdist_wheel
 	python setup.py sdist

docker-compose.yaml

@@ -0,0 +1,9 @@
+version: "3.9"
+services:
+  initdata:
+    build: .
+    volumes:
+      - .:/workdir
+    command:
+      - python3.10 
+      - -V

pybbda/analysis/simulations/components/state.py

@@ -34,7 +34,6 @@ def evolve(
         second_base_running_event=SecondBaseRunningEvent.DEFAULT,
         third_base_running_event=ThirdBaseRunningEvent.DEFAULT,
     ):
-
         if batting_event == BattingEvent.OUT:
             base_state = attr.evolve(self)
 
@@ -48,7 +47,6 @@ def evolve(
             )
 
         elif batting_event == BattingEvent.SINGLE:
-
             running_events = get_running_events_cached(
                 batting_event,
                 first_base_running_event,
@@ -354,7 +352,6 @@ def evolve(
         second_base_running_event=SecondBaseRunningEvent.DEFAULT,
         third_base_running_event=ThirdBaseRunningEvent.DEFAULT,
     ):
-
         outs = self.outs
 
         if outs == self.max_outs:

pybbda/data/sources/baseball_reference/_update.py

@@ -20,7 +20,6 @@ def _download_csv(url):
 
 
 def _save(lines, file_name, output_path):
-
     output_file_path = os.path.join(output_path, file_name)
     output_payload = "\n".join(str(line, "utf-8") for line in lines)
     logger.info("saving file to {}".format(output_file_path))

pybbda/data/sources/data_source/base.py

@@ -30,8 +30,15 @@ def _locate_file(self, name):
             raise FileNotFoundError(f"Cannot find file {full_path}")
 
     def _load(self, name):
-        file_full_path = self._locate_file(name)
-        return pd.read_csv(file_full_path)
+        if isinstance(name, str):
+            file_full_path = self._locate_file(name)
+            return pd.read_csv(file_full_path)
+        elif isinstance(name, list):
+            file_full_paths = [self._locate_file(n) for n in name]
+            dfs = [pd.read_csv(file_full_path) for file_full_path in file_full_paths]
+            return pd.concat(dfs, axis=0)
+        else:
+            raise TypeError
 
     def __getattr__(self, name):
         if name not in self.tables.keys():

pybbda/data/sources/retrosheet/_update.py

@@ -21,7 +21,7 @@ def _download_csv(url):
     logger.info("downloading file from {}".format(url))
     response = requests.get(url, stream=True)
     if response.status_code != 200:
-        logger.info("there was a download error code={}", response.status_code)
+        logger.info("there was a download error code=", response.status_code)
         raise FileNotFoundError
     it = response.iter_lines()
     return list(it)

pybbda/data/sources/retrosheet/data.py

@@ -9,14 +9,19 @@
 from pybbda.data.sources.data_source.base import DataSource
 
 RETROSHEET_DATA_PATH = PYBBDA_DATA_ROOT / "retrosheet"
-RETROSHEET_TABLES = {"people": "people.csv"}
-RETROSHEET_URLS = {
-    "people": "https://raw.githubusercontent.com/"
+RETROSHEET_TABLES = {"people": ["people{c}.csv" for c in "0123456789abcdef"]}
+
+RETROSHEET_URL_FMT = (
+    "https://raw.githubusercontent.com/"
     "chadwickbureau/"
     "register/"
     "master/"
     "data/"
-    "people.csv"
+    "people-{}.csv"
+)
+
+RETROSHEET_URLS = {
+    f"people{c}": RETROSHEET_URL_FMT.format(c) for c in "0123456789abcdef"
 }
 
 logger = logging.getLogger(__name__)

pybbda/data/sources/statcast/_update.py

@@ -88,7 +88,6 @@ def _pool_do_update(overwrite=False, season_stats=None):
 def _update(
     output_root=None, min_date=None, max_date=None, num_threads=2, overwrite=False
 ):
-
     today = datetime.date.today()
     min_date = min_date or (today - datetime.timedelta(1)).strftime("%Y-%m-%d")
     max_date = max_date or today.strftime("%Y-%m-%d")

pybbda/data/sources/statcast/data.py

@@ -1,5 +1,7 @@
 import logging
 import datetime
+import io
+import requests
 
 import pandas as pd
 
@@ -67,7 +69,8 @@ def get_statcast_daily(self, player_type, start_date, end_date, player_id=""):
             }
         )
 
-        daily_df = pd.read_csv(url)
+        file_handle = io.BytesIO(requests.get(url).content)
+        daily_df = pd.read_csv(file_handle)
         if len(daily_df) == STATCAST_QUERY_DATA_SIZE_LIMIT:
             logger.warning(
                 "Statcast query returned %d rows which probably "

pybbda/graphics/graphical_standings.py

@@ -32,7 +32,7 @@
 
 # defaults: (1.05, 1.20)
 k = 0.2
-EXPAND_TEXT = ExpandMultiplier(1.05*k, 1.2*k)
+EXPAND_TEXT = ExpandMultiplier(1.05 * k, 1.2 * k)
 EXPAND_POINTS = ExpandMultiplier(1.05, 1.2)
 EXPAND_OBJECTS = ExpandMultiplier(1.05, 1.2)
 

requirements-dev.txt

@@ -1,12 +1,12 @@
 pytest>=6.0.0
-tox==3.14.3
+tox
 setuptools
-Sphinx==3.0.3
-recommonmark==0.6.0
+Sphinx
+recommonmark
 Pygments>=2.5.1
 black>=22.3.0
-flake8==3.7.8
-sphinx-rtd-theme==0.4.3
-sphinx-gallery==0.7.0
-pytest-cov~=2.10.1
-pytest-xdist~=2.1.0
+flake8
+sphinx-rtd-theme
+sphinx-gallery
+pytest-cov
+pytest-xdist

requirements.txt

@@ -1,15 +1,14 @@
-attrs~=19.3.0
-beautifulsoup4~=4.8.0
-lxml~=4.6.2
-numpy~=1.18.1
-pandas~=1.1.0
-requests~=2.22.0
-scipy~=1.4.1
-sqlalchemy~=1.3.13
-tqdm~=4.46.1
-pychadwick~=0.5.0
-matplotlib~=3.1.3
-seaborn~=0.10.1
-plotnine~=0.7.1
-adjustText==0.7.3
+attrs
+beautifulsoup4
+lxml
+pandas
+requests
+scipy
+sqlalchemy
+tqdm
+pychadwick>=0.6.1
+matplotlib
+seaborn
+plotnine
+adjustText
 namedframes>=0.1.0

scripts/compute_default_running_probabilities.py

@@ -47,7 +47,6 @@ def process_df(df):
 
 
 def get_first_to_home_on_double(ana_df):
-
     first_to_third = ana_df.query(
         "event_cd==21 and start_first_base and end_third_base"
     ).event_ct.sum()

scripts/graphical_standings.py

@@ -3,6 +3,7 @@
 from pybbda.graphics.graphical_standings import plot_graphical_standings
 
 import os
+
 root = os.environ.get("PYBBDA_DATA_ROOT")
 print(root)
 data_root = "/home/bdilday/.pybbda/data/Lahman"
@@ -11,6 +12,7 @@
 teams = ld.teams
 
 import sys
+
 if len(sys.argv) == 2:
     yr = sys.argv[1]
 else:
@@ -19,4 +21,4 @@
 
 p = plot_graphical_standings(standings)
 
-print(p)
+print(p)

scripts/mlb_standings.py

@@ -1,4 +1,3 @@
-
 import requests
 from pybbda.graphics.graphical_standings import plot_graphical_standings
 import pandas as pd
@@ -16,18 +15,22 @@
 )
 
 
-
 def process_team_record(team_record):
-    return {"Team": team_record["team"]["teamName"],
-            "W": team_record["wins"],
-            "L": team_record["losses"],
-            "RS_G": team_record["runsScored"]/team_record["gamesPlayed"],
-            "RA_G": team_record["runsAllowed"] / team_record["gamesPlayed"]
-            }
+    return {
+        "Team": team_record["team"]["teamName"],
+        "W": team_record["wins"],
+        "L": team_record["losses"],
+        "RS_G": team_record["runsScored"] / team_record["gamesPlayed"],
+        "RA_G": team_record["runsAllowed"] / team_record["gamesPlayed"],
+    }
+
 
 def process_record(record):
     lg_div = record["division"]["abbreviation"]
-    return [{"lg_div": lg_div, **process_team_record(team_record)} for team_record in record["teamRecords"]]
+    return [
+        {"lg_div": lg_div, **process_team_record(team_record)}
+        for team_record in record["teamRecords"]
+    ]
 
 
 payload = requests.get(url).json()
@@ -38,4 +41,4 @@ def process_record(record):
 standings = pd.DataFrame(data)
 print(standings)
 p = plot_graphical_standings(standings)
-print(p)
+print(p)

tests/analysis/projections/marcels/test_batting.py

@@ -12,7 +12,6 @@ def test_batting_projections():
     "season, expected", [(2020, 36), (2019, 38), (2018, 41), (2017, 34), (2004, 42)]
 )
 def test_batting_metric_projections(season, expected):
-
     md = MarcelProjectionsBatting()
     proj = md.metric_projection("HR", season)
     assert round(proj.HR.max()) == expected

tests/analysis/projections/marcels/test_pitching.py

@@ -13,7 +13,6 @@ def test_pitching_projections():
     [(2020, 242), (2019, 235), (2018, 229), (2017, 224), (2004, 207)],
 )
 def test_pitching_metric_projections(season, expected):
-
     md = MarcelProjectionsPitching()
     proj = md.metric_projection("SO", season)
     assert round(proj.SO.max()) == expected

tests/analysis/trajectories/batted_balls/test_batted_balls_utils.py

@@ -5,7 +5,7 @@
 
 def test_trig_utils():
     deg = 1
-    assert cos_in_degrees(deg) ** 2 + sin_in_degrees(deg) ** 2 == 1
+    assert cos_in_degrees(deg) ** 2 + sin_in_degrees(deg) ** 2 == pytest.approx(1)
 
     deg = 30
     assert cos_in_degrees(deg) == pytest.approx(sqrt(3) / 2)

tests/data/test_statcast/test_statcast.py

@@ -13,12 +13,11 @@ def test_statcast_init():
 
 def test_statcast_get_daily(statcast_data):
     statcast_data.get_statcast_daily(
-        player_type="batter", start_date="2018-01-01", end_date="2018-01-02"
+        player_type="batter", start_date="2018-06-01", end_date="2018-06-02"
     )
 
 
 def test_statcast_validate_dates(statcast_data):
-
     with pytest.raises(ValueError):
         statcast_data.get_statcast_daily(
             player_type="batter", start_date="2018-01-01", end_date="2017-12-31"