import cPickle
import MySQLdb
from db import DataBase
from pageRank import pageRank

db=DataBase();dbConn=db.db;dbCursor=db.cursor
dbCursor.execute("SELECT pl_from, page_id FROM pagelinks INNER JOIN page ON pagelinks.pl_title = page.page_title")
wikipedia_links = dbCursor.fetchall()
dbCursor.close();del dbCursor

sparse_matrix, these_links, last_link_from = [], [1], 1
for link_from, link_to in wikipedia_links:
    if link_from != last_link_from:
        sparse_matrix.append(these_links)
        while link_from != last_link_from+1:
            sparse_matrix.append([])
            last_link_from += 1
        these_links = [int(link_from)]
    if link_to <= wikipedia_links[-1][0]-1:
            these_links.append(int(link_to))
    last_link_from = link_from

page_rank = pageRank(sparse_matrix, alpha=.85, convergence=.9)

f=open("pagerank.pkl", "w")
cPickle.dump(page_rank, f, 1)
f.close()