Grabs snapshots of rubattle.net from «wayback machine» (thanks for API ;) and gets the online of Diablo II LOD server. Split results by UTC time of snapshot (don't forget UTC+4 cause moscow) by three groups. Build some nice graph to svg via pygal.
| import datetime | |
| import shelve | |
| import json | |
| import operator | |
| import requests | |
| from bs4 import BeautifulSoup as bs | |
| import pygal | |
| class SnapshotNotAvailable(Exception): | |
| pass | |
| def request_wayback(t): | |
| r = requests.get( | |
| "http://archive.org/wayback/available", | |
| params={ | |
| "url": "http://rubattle.net", | |
| "timestamp": t.strftime("%Y%m%d"), | |
| } | |
| ) | |
| info = json.loads(r.text) | |
| snap = info["archived_snapshots"]["closest"] | |
| if snap["available"]: | |
| return ( | |
| datetime.datetime.strptime(snap["timestamp"], "%Y%m%d%H%M%S"), | |
| snap["url"] | |
| ) | |
| else: | |
| raise SnapshotNotAvailable | |
| def request_users(url): | |
| r = requests.get(url) | |
| phtml = bs(r.text) | |
| status = phtml.find(id="serverStatusList") | |
| for count, name in zip(status("span"), status("a")): | |
| yield name.text, int(count.text) | |
| def request_and_store(t, db): | |
| t, url = request_wayback(t) | |
| key = t.strftime(datetime_format) | |
| if key not in db: | |
| online = tuple(request_users(url)) | |
| db[key] = online | |
| db.sync() | |
| print(len(db), t, online) | |
| return t | |
| def find_all_times(left, right, db): | |
| print("->", left, right) | |
| dt = (right - left) / 2 | |
| if dt >= delta: | |
| mid = left + dt | |
| t = request_and_store(mid, db) | |
| if left <= t <= right: | |
| find_all_times(left, min(t, mid), db) | |
| find_all_times(max(t, mid), right, db) | |
| with shelve.open("store.shelve") as db: | |
| datetime_format = "%Y-%m-%d-%H" | |
| delta = datetime.timedelta(seconds=60 * 60) | |
| default_start = datetime.datetime(year=2010, month=1, day=1).strftime(datetime_format) | |
| left = request_and_store(datetime.datetime.strptime(max(db, default=default_start), datetime_format), db) | |
| right = request_and_store(datetime.datetime.now(), db) | |
| find_all_times(left, right, db) | |
| print(str.format("Found {} pages", len(db))) | |
| plot = pygal.DateY( | |
| show_minor_x_labels=False, | |
| x_label_rotation=30, | |
| x_labels_major_count=7, | |
| x_label_format="%Y-%m-%d", | |
| legend_at_bottom=True, | |
| include_x_axis=True, | |
| ) | |
| plot.title = "PG players online per hours interval" | |
| d2_tag = "Diablo II LOD" | |
| get_d2_point = lambda key: (datetime.datetime.strptime(key, datetime_format), dict(db[key])[d2_tag]) | |
| night, day, prime = [], [], [] | |
| for t, c in sorted(map(get_d2_point, db)): | |
| t += datetime.timedelta(hours=4) # utc+4 moscow | |
| point = (t, c) | |
| if 1 <= t.hour < 10: | |
| night.append(point) | |
| elif 10 <= t.hour < 18: | |
| day.append(point) | |
| else: | |
| prime.append(point) | |
| for data, name in zip((night, day, prime), ("night [1; 10)", "day [10; 18)", "prime [18; 1)")): | |
| plot.add("Diablo II " + name, data) | |
| plot.render_to_file("PG players online.svg") |
