2018-10-15~2018-11-02 와룡전설 이벤트 글 추적기 (python 3.5)

wa-event-parser.py

from bs4 import BeautifulSoup
from tkinter import Tk, Label, IntVar, Entry, Button
import requests
import ctypes
import win32process
import os
import sys

# variables
base_url = "http://wa.ilovegame.co.kr/event/read/"
first_page = 1


hwnd = ctypes.windll.kernel32.GetConsoleWindow()
if hwnd != 0:
    ctypes.windll.user32.ShowWindow(hwnd, 0)
    ctypes.windll.kernel32.CloseHandle(hwnd)
    _, pid = win32process.GetWindowThreadProcessId(hwnd)
    os.system('taskkill /PID ' + str(pid) + ' /f')


# GUI
def dialog():
    root = Tk()
    root.title('와룡전설 이벤트글 추적기')
    root.geometry('320x120')

    def validate(str, action_type):
        if action_type == '1':  # insert
            if not str.isdigit():
                return False
        return True

    label_f = Label(root, text="여기부터")
    label_f.pack()
    var_f = IntVar()
    var_f.set(first_page)
    entry_f = Entry(root, textvariable=var_f, validate='key')
    entry_f['validatecommand'] = (entry_f.register(validate), '%P', '%d')
    entry_f.pack()

    label_e = Label(root, text="여기까지")
    label_e.pack()
    var_e = IntVar()
    var_e.set(first_page)
    entry_e = Entry(root, textvariable=var_e, validate='key')
    entry_e['validatecommand'] = (entry_e.register(validate), '%P', '%d')
    entry_e.pack()
    entry_e.focus_set()

    def callback():
        global first_page
        first_page = int(entry_f.get())
        parser(int(entry_e.get()), True, True)
        root.destroy()

    btn = Button(root, text="확인", width=10, command=callback)
    btn.pack()

    root.mainloop()


# wa-event parser
def parser(max_pages=1, multiple=False, log=False):
    # initialize first page no.
    if type(max_pages) != int:
        exit()

    if multiple:
        page = first_page - 1
    else:
        page = max_pages - 1

    # save output
    if log:
        sys.stdout = open('output.txt', 'w')

    while page < max_pages:
        page += 1
        url = base_url + str(page)
        source_code = requests.get(url)
        plain_text = source_code.text
        soup = BeautifulSoup(plain_text, 'lxml')

        # skip if event was expired
        if soup.string:
            if soup.string.find('종료된 이벤트'):
                continue

        # Page No. & Title parsing
        for title in soup.select('li.tit > strong'):
            print('Page No. ' + str(page) + '<br>')
            print('Title: ' + title.string + '<br>')
            print('URL: ' + url + '<br>')

        # Image parsing (<img> tag)
        for link in soup.select('li.evt_ctt img'):
            print('<img src="' + link['src'] + '">')

        # Image parsing (<style> tag)
        if soup.style:
            string = soup.style.string
            print('<img src="' + string[string.find('background:url(')+len('background:url('):string.find('.jpg)')+len('.jpg')] + '">')

        print('<br>')


if __name__ == "__main__":
    dialog()