gistfile1.txt

# -*- coding: utf-8 -*-

import lxml.html as jq
import uuid
import hashlib
import oursql

db = oursql.connect(host="127.0.0.1", user="user", passwd="pwd", db="mydata", charset='utf8')
query = db.cursor(plain_query=True)

file = open('/path/to/bookmarks.html', 'r')
content = file.read()

mydata = jq.document_fromstring(content)

query.execute("SHOW VARIABLES LIKE 'character_set%';")
print query.fetchone()

links = {}

for value in mydata.xpath("//dl"):
	for link in value.xpath("//a"):
		try:
			uid = str(uuid.uuid1())
			sha = hashlib.sha1()
			sha.update(uid)
			myUid = sha.hexdigest()
			query.execute("set names utf8")
			query.execute("INSERT INTO bookmarks (uid, url, title) VALUES(?, ?, ?)", (myUid, link.attrib['href'].decode('utf8'), link.text_content().decode('utf8')))
			
		except:
			pass
			print "Error"
			print link.attrib['href']+" "+(link.text_content().decode('utf-8'))