parse.py 3.7 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283
  1. #!/usr/bin/python3
  2. import urllib3
  3. import sys
  4. import os
  5. import requests
  6. urls = [
  7. 'http://summonerswar.wikia.com/wiki/Fire_Monsters',
  8. 'http://summonerswar.wikia.com/wiki/Water_Monsters',
  9. 'http://summonerswar.wikia.com/wiki/Wind_Monsters',
  10. 'http://summonerswar.wikia.com/wiki/Light_Monsters',
  11. 'http://summonerswar.wikia.com/wiki/Dark_Monsters'
  12. ]
  13. http = urllib3.PoolManager()
  14. fsql = open('data.sql', 'w')
  15. fget = open('img.sh', 'w')
  16. id = 1;
  17. for url in urls:
  18. page = http.request('GET', url).data.decode('utf-8')
  19. page = page[page.index('</th></tr>'):]
  20. element = url[url.rfind('/') + 1:url.rfind('_')]
  21. while '</td></tr>' in page:
  22. murl = 'http://summonerswar.wikia.com/' + page[page.index('a href="') + 8:page.index('"', page.index('a href="') + 8)]
  23. print(murl)
  24. page = page[page.index('</td></tr>') + 1:]
  25. '''page = page[page.index('<td style="text-align:center;">') + 1:]
  26. monster = page[:page.index('</tr>') + 5]
  27. name = monster[monster.index('title="') + 7:monster.index(' (', monster.index('title="') + 8)]
  28. monster = monster[monster.index('"', monster.index('title="') + 8):]
  29. img = monster[monster.index('data-src="') + 10:monster.index('"', monster.index('data-src="') + 11)]
  30. monster = monster[monster.index('<td style="text-align:center;">') + 1:]
  31. monster = monster[monster.index('<td style="text-align:center;">') + 1:]
  32. tpe = monster[monster.index('>') + 1:monster.index('\n')]
  33. monster = monster[monster.index('<td style="text-align:center;">') + 1:]
  34. monster = monster[monster.index('<td style="text-align:center;">') + 1:]
  35. hp = monster[monster.index('>') + 1:monster.index('\n')]
  36. monster = monster[monster.index('<td style="text-align:center;">') + 1:]
  37. atk = monster[monster.index('>') + 1:monster.index('\n')]
  38. monster = monster[monster.index('<td style="text-align:center;">') + 1:]
  39. df = monster[monster.index('>') + 1:monster.index('\n')]
  40. monster = monster[monster.index('<td style="text-align:center;">') + 1:]
  41. spd = monster[monster.index('>') + 1:monster.index('\n')]
  42. print("NAME: " + name)
  43. print("ELEMENT: " + element)
  44. print("IMG: " + img)
  45. print("TYPE: " + tpe)
  46. print("HP: " + hp)
  47. print("ATK: " + atk)
  48. print("DEF: " + df)
  49. print("SPD: " + spd)
  50. sys.exit("DEBUG EXIT")
  51. element = monster[monster.index('"element">') + 10:monster.index('<', monster.index('"element">'))]
  52. print("ELEMENT: " + element)
  53. monster = monster[monster.index('<', monster.index('"element">')):]
  54. name = monster[monster.index('<h3>') + 4:monster.index('</h3>')]
  55. print("NAME: " + name)
  56. monster = monster[monster.index('</h3>'):]
  57. img_aw = monster[monster.index('data-src="', monster.index('icon')) + 10:monster.index('"', monster.index('data-src="', monster.index('icon')) + 11)]
  58. print("IMAGE (AW): " + img_aw)
  59. monster = monster[monster.index('"', monster.index('data-src="', monster.index('icon')) + 11):]
  60. name_aw = monster[monster.index('>', monster.index('"name"')) + 1:monster.index('<', monster.index('"name"'))]
  61. print("NAME (AW): " + name_aw)
  62. monster = monster[monster.index('<', monster.index('"name"')):]
  63. print("wget " + img + " -O monster/" + "{:04d}".format(id) + ".png", file=fget)
  64. print("wget " + img_aw + " -O monster/awaken/" + "{:04d}".format(id) + ".png", file=fget)
  65. print("INSERT INTO monster (id, name, name_aw, element, stars, base) VALUES (" + str(id) + ", '" + name + "', '" + name_aw + "', '" + element + "', " + str(star) + ", null);", file=fsql)
  66. id = id + 1
  67. #sys.exit("DEBUG EXIT")
  68. '''