connection = MySQLdb.connect(host = "127.0.0.1", user = "******", passwd = "paper1ens", db = "paperlens")
cursor = connection.cursor()
connection.commit()
try:
paper_hash = dict()
cursor.execute("select id,title from paper")
n = 0
while 1:
row = cursor.fetchone()
if row == None:
break
paper_id = int(row[0])
title = row[1]
paper_hash[paper_id] = paperlens_import.intHash(title.lower())
n = n + 1
if n % 10000 == 0:
print str(n)
n = 0
for (paper_id, hash_value) in paper_hash.items():
cursor.execute("update paper set hashvalue=%s where id=%s",(hash_value,paper_id))
n = n + 1
if n % 10000 == 0:
print str(n)
connection.commit()
cursor.close()
connection.close()
except MySQLdb.Error, e:
print e.args[0], e.args[1]
connection.commit()
try:
data = open("../../../data/citeseer.txt")
cursor.execute("truncate table paper_citeseer")
cursor.execute("truncate table cite_citeseer")
citeseer_id_map = dict()
title = ''
citeseer_id = ''
n = 0
for line in data:
(key, value) = Extract(line)
if line.find("<record>") >= 0:
if len(title) > 20:
hashvalue = paperlens_import.intHash(title.lower())
cursor.execute("select count(*),id from paper where hashvalue=%s",(hashvalue))
row = cursor.fetchone()
if int(row[0]) == 1:
paper_id = int(row[1])
if citeseer_id not in citeseer_id_map:
citeseer_id_map[citeseer_id] = paper_id
#cursor.execute("replace into paper_citeseer (paper_id, citeseer_key) values (%s, %s)",(paper_id, citeseer_id))
if n % 10000 == 0:
print n, title, citeseer_id
n = n + 1
title = ''
citeseer_id = ''
if key == "<dc:title>":
passwd="paper1ens",
db="paperlens")
cursor = connection.cursor()
connection.commit()
try:
data = open("../../../data/citeseer.txt")
cursor.execute("truncate table paper_citeseer")
title = ''
citeseer_id = ''
n = 0
for line in data:
(key, value) = Extract(line)
if line.find("<record>") >= 0:
if len(title) > 20:
hashvalue = paperlens_import.intHash(title.lower())
cursor.execute(
"select count(*),id from paper where hashvalue=%s",
(hashvalue))
row = cursor.fetchone()
if int(row[0]) == 1:
paper_id = int(row[1])
cursor.execute(
"replace into paper_citeseer (paper_id, citeseer_key) values (%s, %s)",
(paper_id, citeseer_id))
if n % 10000 == 0:
print n, title, citeseer_id
n = n + 1
title = ''
