SWDB/parser/parse.py

#!/usr/bin/python3

import urllib3
import sys
import os
import requests


urls = [
  'http://summonerswar.wikia.com/wiki/Fire_Monsters',
  'http://summonerswar.wikia.com/wiki/Water_Monsters',
  'http://summonerswar.wikia.com/wiki/Wind_Monsters',
  'http://summonerswar.wikia.com/wiki/Light_Monsters',
  'http://summonerswar.wikia.com/wiki/Dark_Monsters'
]
http = urllib3.PoolManager()
fsql = open('data.sql', 'w')
fget = open('img.sh', 'w')
id = 1;

for url in urls:
    page = http.request('GET', url).data.decode('utf-8')
    page = page[page.index('</th></tr>'):]
    element = url[url.rfind('/') + 1:url.rfind('_')]

    while '</td></tr>' in page:
        
        murl = 'http://summonerswar.wikia.com/' + page[page.index('a href="') + 8:page.index('"', page.index('a href="') + 8)]
        print(murl)
        page = page[page.index('</td></tr>') + 1:]

        '''page = page[page.index('<td style="text-align:center;">') + 1:]
        monster = page[:page.index('</tr>') + 5]
        name = monster[monster.index('title="') + 7:monster.index(' (', monster.index('title="') + 8)]
        monster = monster[monster.index('"', monster.index('title="') + 8):]
        img = monster[monster.index('data-src="') + 10:monster.index('"', monster.index('data-src="') + 11)]
        monster = monster[monster.index('<td style="text-align:center;">') + 1:]
        monster = monster[monster.index('<td style="text-align:center;">') + 1:]
        tpe = monster[monster.index('>') + 1:monster.index('\n')]
        monster = monster[monster.index('<td style="text-align:center;">') + 1:]
        monster = monster[monster.index('<td style="text-align:center;">') + 1:]
        hp = monster[monster.index('>') + 1:monster.index('\n')]
        monster = monster[monster.index('<td style="text-align:center;">') + 1:]
        atk = monster[monster.index('>') + 1:monster.index('\n')]
        monster = monster[monster.index('<td style="text-align:center;">') + 1:]
        df = monster[monster.index('>') + 1:monster.index('\n')]
        monster = monster[monster.index('<td style="text-align:center;">') + 1:]
        spd = monster[monster.index('>') + 1:monster.index('\n')]
        print("NAME:    " + name)
        print("ELEMENT: " + element)
        print("IMG:     " + img)
        print("TYPE:    " + tpe)
        print("HP:      " + hp)
        print("ATK:     " + atk)
        print("DEF:     " + df)
        print("SPD:     " + spd)
        sys.exit("DEBUG EXIT")

        

        element = monster[monster.index('"element">') + 10:monster.index('<', monster.index('"element">'))]
        print("ELEMENT: " + element)
        monster = monster[monster.index('<', monster.index('"element">')):]

        name = monster[monster.index('<h3>') + 4:monster.index('</h3>')]
        print("NAME: " + name)
        monster = monster[monster.index('</h3>'):]

        img_aw = monster[monster.index('data-src="', monster.index('icon')) + 10:monster.index('"', monster.index('data-src="', monster.index('icon')) + 11)]
        print("IMAGE (AW): " + img_aw)
        monster = monster[monster.index('"', monster.index('data-src="', monster.index('icon')) + 11):]

        name_aw = monster[monster.index('>', monster.index('"name"')) + 1:monster.index('<', monster.index('"name"'))]
        print("NAME (AW): " + name_aw)
        monster = monster[monster.index('<', monster.index('"name"')):]

        print("wget " + img + " -O monster/" + "{:04d}".format(id) + ".png", file=fget)
        print("wget " + img_aw + " -O monster/awaken/" + "{:04d}".format(id) + ".png", file=fget)
        print("INSERT INTO monster (id, name, name_aw, element, stars, base) VALUES (" + str(id) + ", '" + name + "', '" + name_aw + "', '" + element + "', " + str(star) + ", null);", file=fsql)

        id = id + 1
        #sys.exit("DEBUG EXIT")
        '''