Advertisement
Guest User

Untitled

a guest
Jun 1st, 2017
115
0
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 1.39 KB | None | 0 0
  1. #!/usr/bin/env python3
  2.  
  3. import psycopg2
  4. import sys
  5.  
  6. if __name__ =="__main__":
  7. conn = psycopg2.connect('dbname=grf2 user=postgres password=asdwsx host=127.0.0.1')
  8. #conn = psycopg2.connect('dbname=giraff user=giraff password=g1r@ff host=cobalt.da')
  9.  
  10. cur = conn.cursor()
  11. cur2 = conn.cursor()
  12.  
  13. cur.execute("""SELECT id, ftrack_id FROM sites""")
  14. sites = {}
  15. for row in cur:
  16. sites[row[1]] = int(row[0])
  17.  
  18.  
  19. urls = {}
  20. counter = 0
  21. cur.execute("""SELECT date, site, url FROM stats.day_page WHERE url_id IS NULL LIMIT 1000000""")
  22. for row in cur:
  23. counter +=1
  24. if (counter % 1000 == 0):
  25. print(counter)
  26.  
  27. if row[1] in sites:
  28. site_id = sites[row[1]]
  29. else:
  30. print(row[1])
  31. continue
  32.  
  33. if not (site_id, row[2]) in urls:
  34. cur2.execute("""SELECT id FROM urls WHERE site_id = %s AND url = %s""", (site_id, row[2]))
  35. url_id = cur2.fetchone()
  36. if not url_id:
  37. cur2.execute("""INSERT INTO urls(site_id, url) VALUES(%s,%s) RETURNING id""", (site_id, row[2]))
  38. url_id = cur2.fetchone()
  39.  
  40. urls[(site_id, row[2])] = url_id[0]
  41.  
  42. cur2.execute("""UPDATE stats.day_page SET url_id = %s WHERE date = %s AND site = %s AND url = %s""", (urls[(site_id, row[2])], row[0], row[1], row[2]))
  43.  
  44. conn.commit()
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement