seniority.py

from __future__ import division

import requests
import BeautifulSoup

team_mapping = {
    "NJN": "BRK",
    "CHA": "CHO",
    "NOH": "NOP"
}


def get_page_body(uri):
    d = requests.get("http://www.basketball-reference.com%s" % uri)
    content = d.text.encode("utf-8")
    return BeautifulSoup.BeautifulSoup(content)


def get_seniority(team):
    team_page = get_page_body("/teams/%s/2015.html" % team)
    wins = get_winning_percentage(team_page)

    dd = team_page.find(id="totals")

    data = []
    for row in dd.findAll("tr"):
        cols = row.findAll("td")
        if not cols:
            continue
        link = cols[1].findAll("a")[0]["href"]
        mins = int(cols[5].text)
        if mins:
            total_mins = get_player_seniority(link, team)
        data.append((mins, total_mins))
    total = sum(mins for (mins, total_mins) in data)
    coef = sum(mins/total_mins*mins/total for (mins, total_mins) in data)
    print team, wins, coef


def get_player_seniority(link, team):
    teams = [team]
    teams.extend(key for (key, value) in team_mapping.items() if value == team)
    total_mins = 0
    dd = get_page_body(link).find(id="totals")
    for row in dd.findAll("tr"):
        css_class = row["class"]
        if not css_class or css_class.split()[-1] not in ("full_table", "partial_table"):
            continue
        cols = row.findAll("td")
        if cols[2].text in teams:
            total_mins += int(cols[7].text)
    return total_mins


def get_winning_percentage(team_page):
    info_box = team_page.find(id="info_box")
    record_box = info_box.find(text="Record:").parent.parent
    data = record_box.contents[1].split(",")[0].strip().split("-")
    wins, losses = int(data[0]), int(data[1])
    return wins / (wins + losses)


def list_teams():
    codes = []
    data = get_page_body("/teams").find(id="active")
    for row in data.findAll("tr"):
        cols = row.findAll("td")
        if not cols:
            continue
        links = cols[0].findAll("a")
        if not links:
            continue
        code = links[0]["href"].split("/")[2]
        codes.append(code)
    return codes


def main():
    teams = list_teams()
    for team in teams:
        if team in team_mapping:
            team = team_mapping[team]
        get_seniority(team)


if __name__ == "__main__":
    main()