new file mode 100644

@@ -0,0 +1,15 @@

+# NHL stats

+

+This repo will contain scripts for dumping full NHL game feeds.

+

+## API URLs

+

+This page contains docs on the API: https://gitlab.com/dword4/nhlapi/-/blob/master/stats-api.md

+

+The main API endpoint is https://statsapi.web.nhl.com/api/v1

+

+The important endpoints are /game/ID/feed/live which contains the full game info, including all plays,

+and /game/ID/linescore which contains aggregate statistics for each player during the match.

+

+Game discovery is via /schedule query strings are used to get schedule data between certain dates

+or for a whole season with: ?season=20172018

new file mode 100755

@@ -0,0 +1,77 @@

+#!/usr/bin/env python

+"""Download full game feeds."""

+

+import argparse

+import asyncio

+from functools import partial

+from pathlib import Path

+import sys

+

+import backoff

+import httpx

+import msgpack

+import tqdm

+

+api_url = "https://statsapi.web.nhl.com/api/v1{}".format

+

+

+def get_game_guids(season: int) -> list[int]:

+    path = Path(f"seasons/{season}{season+1}.msgpack")

+    if not path.exists():

+        raise ValueError(f"No game data for {season} season")

+

+    with path.open("rb") as f:

+        season_data = msgpack.load(f)

+

+    # TODO: replace with JSONPath query

+    return [

+        g["gamePk"]

+        for d in season_data["dates"]

+        for g in d["games"]

+        if g["status"]["abstractGameState"] == "Final"

+    ]

+

+

+@backoff.on_exception(

+    backoff.expo,

+    (httpx.ConnectTimeout, httpx.ConnectError),

+    max_time=120,

+)

+async def get_game_feed(client: httpx.Client, guid: int) -> dict:

+    resp = await client.get(api_url(f"/game/{guid}/feed/live"))

+    return resp.json()

+

+

+def feed_downloaded(guid: int) -> bool:

+    return Path(f"games/{guid}.msgpack").exists()

+

+

+def save_game_feed(feed):

+    guid = feed["gamePk"]

+    with open(f"games/{guid}.msgpack", "wb") as f:

+        msgpack.dump(feed, f)

+

+

+async def main(season: int, force: bool = False):

+    guids = [

+        guid for guid in get_game_guids(season=season) if not feed_downloaded(guid)

+    ]

+    if not guids and not force:

+        print(f"Games for {season} season already downloaded", file=sys.stderr)

+        return

+

+    c = httpx.AsyncClient(

+        limits=httpx.Limits(max_keepalive_connections=5, max_connections=10)

+    )

+    game_feeds = map(partial(get_game_feed, c), guids)

+    monitor = partial(tqdm.tqdm, desc=f"{season} season games", total=len(guids))

+    async with c:

+        for feed_coro in monitor(asyncio.as_completed(game_feeds)):

+            save_game_feed(await feed_coro)

+

+

+if __name__ == "__main__":

+    parser = argparse.ArgumentParser(description="Download NHL game feeds")

+    parser.add_argument("season", type=int, help="Which season to download")

+    args = parser.parse_args()

+    asyncio.run(main(args.season))

new file mode 100755

@@ -0,0 +1,19 @@

+#!/bin/bash

+

+let season_start="$1"

+

+if [[ -z $season_start ]]; then

+    echo "Expected Season (e.g. 2008), but none provided" >&2

+    exit 1

+fi

+

+# 2008 -> 20082009 as required by NHL API

+let season=$season_start$season_start+1

+url="https://statsapi.web.nhl.com/api/v1/schedule?season=$season"

+outfile="$season.json"

+

+echo

+echo "Hitting $url and saving to $season.msgpack"

+echo

+

+curl $url | ./json2msgpack > seasons/$season.msgpack

new file mode 100644

@@ -0,0 +1,13 @@

+name: nhl-stats

+channels:

+    - default

+    - conda-forge

+dependencies:

+    - python

+    - ipython

+    - ipykernel

+    # Actual deps

+    - backoff

+    - httpx

+    - msgpack-python

+    - tqdm

new file mode 100755

@@ -0,0 +1,7 @@

+#!/usr/bin/env -S python -u

+

+import json

+import sys

+import msgpack

+

+msgpack.dump(json.load(sys.stdin), sys.stdout.buffer)