waqwaq2.py

# coding: utf-8

import datetime
import re
import sys
import urllib

from bs4 import BeautifulSoup
import xlsxwriter
import pandas as pd


# コンソール出力をリダイレクトするときのおまじない
reload(sys)
sys.setdefaultencoding('utf-8')

# 設定
wikipedia_base = "http://ja.wikipedia.org"
get_limit = 50

# 期間設定
start_date = datetime.datetime(2012, 1, 1)
end_date = datetime.datetime(2013, 12, 31)

# スコア設定
scores = {"New": 5, "Edit": 3, "MinorEdit": 1, "Delete": 2}

# 参加者設定（仮に過去のWAQWAQプロジェクト参加者を入れてます）
names = [u"Next49", u"Mishika", u"Theta K", u"蒋龍"]
# names = [u"Next49"]

# 出力CSVのエンコード設定(utf-8-sig: UTF-8+BOM)
csv_encoding = "utf-8-sig"


def get_wikipedia_history(username, offset=None):
    # 結果格納場所
    result = []
    # URLの組み立て
    urlparts = u"特別:投稿記録"
    urlparts_quote = urllib.quote(urlparts.encode("utf-8"))
    username_quote = urllib.quote_plus(username.encode("utf-8"))
    url = u"http://ja.wikipedia.org/w/index.php?limit=%d&tagfilter=&title=%s&contribs=user&target=%s&namespace=0&tagfilter=&year=2014&month=-1" % (
        get_limit, urlparts_quote, username_quote)
    if offset is not None:
        url += u"&offset=%s" % offset

    # 取得＆パース
    req = urllib.urlopen(url)
    soup = BeautifulSoup(req)

    found_old = False

    print("****************************************")
    for li in soup.ul.find_all("li"):
        item = {u"執筆者": username}

        print(li)

        # 日時の抽出
        a_date = li.find(class_=["mw-changeslist-date", "history-deleted"])
        a_date_str = a_date.string
        # m = re.match(u"(\d+)年(\d+)月(\d+)日（.*）(\d+):(\d+)",a_date_str)
        m = re.match(u"(\d+)年(\d+)月(\d+)日.*(\d+):(\d+)", a_date_str)
        a_y = int(m.group(1))
        a_m = int(m.group(2))
        a_d = int(m.group(3))
        a_h = int(m.group(4))
        a_min = int(m.group(5))
        item[u"変更日時"] = datetime.datetime(a_y, a_m, a_d, a_h, a_min)

        if item[u"変更日時"] > end_date:
            continue

        if item[u"変更日時"] < start_date:
            found_old = True
            break

        # 該当ページ・該当版のURL取得
        if a_date.get("href") is not None:
            item[u"当該ページの当該版のURL"] = wikipedia_base + a_date["href"]
        else:
            item[u"当該ページの当該版のURL"] = ""

        a_page = li.find(class_=["exitw", "mw-contributions-title"])
        item[u"当該ページのURL"] = wikipedia_base + a_page["href"]

        # 変更種別の取得
        abbr = li.abbr
        if abbr is None:
            if li.find(class_="history-deleted") is not None:
                item[u"変更種別"] = "Delete"
            else:
                item[u"変更種別"] = "Edit"
        else:
            if abbr["class"][0] == "newpage":
                item[u"変更種別"] = "New"
            elif abbr["class"][0] == "minoredit":
                item[u"変更種別"] = "MinorEdit"

        # 変更したページ名の取得
        a_title = li.find(class_="mw-contributions-title")
        item[u"ページ名"] = a_title.string

        # 変更バイト数の取得
        plusminus = li.find(
            class_=["mw-plusminus-pos", "mw-plusminus-null", "mw-plusminus-neg"])
        m = re.match(r"\([+-]?([0-9,]+)\)", plusminus.string)
        item[u"変更バイト数"] = int(m.group(1).replace(",", ""))

        # スコアの計算
        item[u"スコア"] = scores[item[u"変更種別"]]

        # 結果の先頭に挿入
        print item
        print("--------------------------------------")
        result.insert(0, item)
    a_next = soup.find(class_="mw-nextlink")
    if a_next is not None and not found_old:
        print a_next
        print ":::::::::::::::::::::::::::::::::::::::"
        a_next_href = a_next["href"]
        m = re.search(r"offset=(\d+)", a_next_href)
        offset = m.group(1)
        result_old = get_wikipedia_history(username, offset)
        result = result_old + result

    return result


def write_worksheet(wb, username, changes, date_format):
    # 結果、チャートのワークシート作成
    ws = wb.add_worksheet(username)
    ws_c = wb.add_chartsheet(u"%s_c" % username)

    # カラムヘッダ作成
    ws.write("A1", u"変更日時")
    ws.write("B1", u"変更種別")
    ws.write("C1", u"変更バイト数")
    ws.write("D1", u"ページ名")
    ws.write("E1", u"当該ページのURL")
    ws.write("F1", u"当該ページの当該版のURL")
    ws.write("G1", u"スコア")
    ws.write("H1", u"スコア累計")

    # 各行データの埋め込み
    row = 0
    score_sum = 0
    for change in changes:
        row += 1
        score_sum += change[u"スコア"]
        ws.write(row, 0, change[u"変更日時"], date_format)
        ws.write(row, 1, change[u"変更種別"])
        ws.write(row, 2, change[u"変更バイト数"])
        ws.write(row, 3, change[u"ページ名"])
        ws.write(row, 4, change[u"当該ページのURL"])
        ws.write(row, 5, change[u"当該ページの当該版のURL"])
        ws.write(row, 6, change[u"スコア"])
        ws.write(row, 7, score_sum)

    row += 1
    ws.write(row, 0, datetime.datetime.now(), date_format)
    print "row = %d" % row

    # チャートの作成
    chart = wb.add_chart({"type": "line"})
    chart.add_series({
        "name": username,
        "categories": [ws.get_name(), 1, 0, row, 0],
        "values": [ws.get_name(), 1, 7, row, 7],
        # "data_labels": {"value": True},
        })
    chart.set_legend({"none": True})
    ws_c.set_chart(chart)
    return

if __name__ == "__main__":
    wb = xlsxwriter.Workbook("waqwaq.xlsx")
    date_format = wb.add_format({"num_format": "yyyy/mm/dd hh:mm:ss"})

    # df(DataFrame): CSVにするためにpandasのデータ構造を使う
    df = None
    for username in names:
        result = get_wikipedia_history(username)
        write_worksheet(wb, username, result, date_format)
        if df is None:
            df = pd.DataFrame.from_dict(result)
        else:
            df_tmp = pd.DataFrame.from_dict(result)
            df = pd.concat([df, df_tmp])
    df.sort(u"変更日時", inplace=True)
    df.to_csv("waqwaq.csv", encoding=csv_encoding, index=False)
    wb.close()