scripts/fetch_bbref_draft_hs.py

#!/usr/bin/env python3
"""Fetch NBA draft picks + finishing high school from Basketball-Reference.

Polite scraper: ~1 request / 3.5s. Caches player pages so re-runs are cheap.

Usage:
  python fetch_bbref_draft_hs.py --years 2016-2025
  python fetch_bbref_draft_hs.py --years 2022,2023,2024
"""

from __future__ import annotations

import argparse
import csv
import json
import re
import time
from pathlib import Path

import requests
from bs4 import BeautifulSoup

ROOT = Path(__file__).resolve().parents[1]
RAW = ROOT / "sources" / "bbref_raw"
CACHE = RAW / "player_hs_cache.json"

UA = (
    "athletic_scrapper-nba-draft-hs-panel/0.1 "
    "(research dataset; polite scrape; github.com/Dharit13)"
)
SLEEP_S = 3.5
BASE = "https://www.basketball-reference.com"


def parse_years(spec: str) -> list[int]:
    years: list[int] = []
    for part in spec.split(","):
        part = part.strip()
        if not part:
            continue
        if "-" in part:
            a, b = part.split("-", 1)
            years.extend(range(int(a), int(b) + 1))
        else:
            years.append(int(part))
    return sorted(set(years))


def load_cache() -> dict:
    if CACHE.exists():
        return json.loads(CACHE.read_text())
    return {}


def save_cache(cache: dict) -> None:
    RAW.mkdir(parents=True, exist_ok=True)
    CACHE.write_text(json.dumps(cache, indent=2, sort_keys=True) + "\n")


def get_html(session: requests.Session, url: str) -> str:
    r = session.get(url, timeout=45)
    r.raise_for_status()
    # Basketball-Reference is UTF-8; avoid mojibake from mis-detected encoding.
    r.encoding = "utf-8"
    return r.text


def parse_draft_table(html: str, year: int) -> list[dict]:
    soup = BeautifulSoup(html, "html.parser")
    table = soup.find("table", id="stats")
    if table is None:
        raise RuntimeError(f"No #stats table for {year}")

    rows_out: list[dict] = []
    for tr in table.find("tbody").find_all("tr"):
        if tr.get("class") and "thead" in tr.get("class"):
            continue
        pick_cell = tr.find("td", {"data-stat": "pick_overall"})
        if pick_cell is None:
            continue
        pick_txt = pick_cell.get_text(strip=True)
        if not pick_txt.isdigit():
            continue
        pick = int(pick_txt)
        team_cell = tr.find("td", {"data-stat": "team_id"})
        player_cell = tr.find("td", {"data-stat": "player"})
        college_cell = tr.find("td", {"data-stat": "college_name"})
        a = player_cell.find("a") if player_cell else None
        player_name = player_cell.get_text(strip=True) if player_cell else ""
        player_url = BASE + a["href"] if a and a.get("href") else ""
        player_slug = a["href"].rstrip("/").split("/")[-1].replace(".html", "") if a else ""
        team = team_cell.get_text(strip=True) if team_cell else ""
        college = college_cell.get_text(strip=True) if college_cell else ""
        rnd = 1 if pick <= 30 else 2
        rows_out.append(
            {
                "draft_year": year,
                "pick": pick,
                "round": rnd,
                "player_name": player_name,
                "nba_team": team,
                "college_or_prior": college,
                "bbref_player_url": player_url,
                "bbref_player_slug": player_slug,
            }
        )
    return rows_out


def _parse_one_hs_entry(entry: str) -> tuple[str, str, str]:
    """Parse 'School in City, State' → (school, city, state)."""
    entry = entry.strip(" ,")
    if not entry:
        return "", "", ""
    if " in " in entry:
        school, loc = entry.rsplit(" in ", 1)
        school = school.strip()
        loc = loc.strip()
        if "," in loc:
            city, state = loc.rsplit(",", 1)
            return school, city.strip(), state.strip()
        return school, loc, ""
    return entry, "", ""


_HS_ENTRY_RE = re.compile(
    r"(.+?)\s+in\s+(.+?)(?=,\s*[^,]+?\s+in\s+|$)",
    re.IGNORECASE,
)


def split_hs_entries(body: str) -> list[str]:
    """Split BBRef multi-school bio lines into individual 'School in City, Region' entries."""
    body = body.strip()
    if not body:
        return []
    matches = list(_HS_ENTRY_RE.finditer(body))
    if not matches:
        return [body]
    return [
        f"{m.group(1).strip().lstrip(',').strip()} in {m.group(2).strip()}"
        for m in matches
    ]


def apply_hs_body(out: dict, body: str) -> None:
    parts = split_hs_entries(body)
    out["hs_all"] = " | ".join(parts)
    school, city, state = _parse_one_hs_entry(parts[-1])
    out["high_school_finish"] = school
    out["hs_city"] = city
    out["hs_state"] = state


def parse_player_hs(html: str) -> dict:
    """Return high_school_finish, hs_city, hs_state, birth_country_hint.

    Basketball-Reference uses:
      - 'High School: X in City, State' for a single school
      - 'High Schools: A in ..., B in ...' for multiple — we take the **last**
        entry as the finishing school (same convention as the 2026 case study).
    """
    soup = BeautifulSoup(html, "html.parser")
    out = {
        "high_school_finish": "",
        "hs_city": "",
        "hs_state": "",
        "hs_raw": "",
        "birth_place": "",
        "hs_all": "",
    }
    meta = soup.find("div", id="meta")
    if meta is None:
        return out

    for p in meta.find_all("p"):
        text = " ".join(p.get_text(" ", strip=True).split())
        if text.startswith("Born:"):
            out["birth_place"] = text.replace("Born:", "").strip()
        if text.startswith("High School:") or text.startswith("High Schools:"):
            out["hs_raw"] = text
            if text.startswith("High Schools:"):
                body = text[len("High Schools:") :].strip()
            else:
                body = text[len("High School:") :].strip()
            apply_hs_body(out, body)
    return out


def parse_hs_raw_line(hs_raw: str) -> dict:
    """Re-parse a cached hs_raw line without network I/O."""
    out = {
        "high_school_finish": "",
        "hs_city": "",
        "hs_state": "",
        "hs_raw": hs_raw or "",
        "hs_all": "",
    }
    if not hs_raw:
        return out
    text = hs_raw.strip()
    if text.startswith("High Schools:"):
        body = text[len("High Schools:") :].strip()
    elif text.startswith("High School:"):
        body = text[len("High School:") :].strip()
    else:
        body = text
    apply_hs_body(out, body)
    return out


def fetch_year(session: requests.Session, year: int, cache: dict, force: bool) -> list[dict]:
    draft_path = RAW / f"draft_{year}.csv"
    url = f"{BASE}/draft/NBA_{year}.html"
    print(f"[{year}] draft page {url}")
    time.sleep(SLEEP_S)
    html = get_html(session, url)
    picks = parse_draft_table(html, year)
    print(f"[{year}] {len(picks)} picks")

    for i, row in enumerate(picks, 1):
        slug = row["bbref_player_slug"]
        if not slug:
            row.update(
                {
                    "high_school_finish": "",
                    "hs_city": "",
                    "hs_state": "",
                    "hs_raw": "",
                    "birth_place": "",
                    "hs_source": "missing_player_page",
                }
            )
            continue
        if slug in cache and not force:
            hs = cache[slug]
        else:
            print(f"  [{i}/{len(picks)}] player {row['player_name']} ({slug})")
            time.sleep(SLEEP_S)
            try:
                ph = get_html(session, row["bbref_player_url"])
                hs = parse_player_hs(ph)
                hs["fetched"] = True
            except requests.HTTPError as e:
                hs = {
                    "high_school_finish": "",
                    "hs_city": "",
                    "hs_state": "",
                    "hs_raw": "",
                    "birth_place": "",
                    "error": str(e),
                }
            cache[slug] = hs
            if i % 5 == 0:
                save_cache(cache)
        row["high_school_finish"] = hs.get("high_school_finish", "")
        row["hs_city"] = hs.get("hs_city", "")
        row["hs_state"] = hs.get("hs_state", "")
        row["hs_raw"] = hs.get("hs_raw", "")
        row["birth_place"] = hs.get("birth_place", "")
        row["hs_source"] = "basketball_reference_player_page"

    fields = [
        "draft_year",
        "pick",
        "round",
        "player_name",
        "nba_team",
        "college_or_prior",
        "high_school_finish",
        "hs_city",
        "hs_state",
        "hs_raw",
        "birth_place",
        "bbref_player_url",
        "bbref_player_slug",
        "hs_source",
    ]
    with draft_path.open("w", newline="", encoding="utf-8") as f:
        w = csv.DictWriter(f, fieldnames=fields, extrasaction="ignore")
        w.writeheader()
        w.writerows(picks)
    print(f"[{year}] wrote {draft_path}")
    return picks


def main() -> None:
    ap = argparse.ArgumentParser()
    ap.add_argument("--years", default="2016-2025", help="e.g. 2016-2025 or 2022,2023")
    ap.add_argument("--force", action="store_true", help="refetch player pages even if cached")
    args = ap.parse_args()
    years = parse_years(args.years)
    RAW.mkdir(parents=True, exist_ok=True)
    cache = load_cache()
    session = requests.Session()
    session.headers.update({"User-Agent": UA, "Accept-Language": "en-US,en;q=0.9"})

    for year in years:
        fetch_year(session, year, cache, force=args.force)
        save_cache(cache)
    print("done")


if __name__ == "__main__":
    main()