def migrate_contacts(source, batch_size=1000):
# read in the content
f = open(source)
xml = etree.parse(f)
f.close()
contacts = xml.getroot()
print "migrating", str(len(contacts)), "contact records from", source
# first thing to do is locate all the duplicates in the logins
record = []
duplicates = []
for element in contacts:
login = element.find("login")
if login is not None and login.text is not None and login.text != "":
if login.text in record:
duplicates.append(login.text)
record.append(login.text)
# now go through and load all the user accounts
batch = []
for element in contacts:
login = element.find("login")
password = element.find("password")
name = element.find("name")
email = element.find("email")
issns = element.findall("issn")
if login is None or login.text is None or login.text == "":
print "ERROR: contact without login - providing login"
if len(issns) == 0:
# make a random 8 character login name
login = ''.join(random.choice(string.ascii_uppercase + string.digits) for x in range(8))
else:
# select the first issn
login = issns[0].text
else:
login = login.text
if password is None or password.text is None or password.text == "":
print "ERROR: contact without password", login, "- providing one"
# make a random 8 character password
password = ''.join(random.choice(string.ascii_uppercase + string.digits) for x in range(8))
else:
password = password.text
# check to see if this is a duplicate
if login in duplicates:
if len(issns) == 0:
print "INFO: duplicate detected, has no ISSNs, so skipping", login
continue
else:
print "INFO: duplicate detected, with ISSNs, so keeping", login
a = Account()
a.set_id(login)
a.set_password(password)
if name is not None and name.text is not None and name.text != "":
a.set_name(name.text)
if email is not None and email.text is not None and email.text != "":
a.set_email(email.text)
for issn in issns:
if issn is not None and issn.text is not None and issn.text != "":
jid = _get_journal_id_from_issn(issn.text)
a.add_journal(jid)
a.prep() # prep for saving, since we're not actually going to call save()
batch.append(a.data)
if len(batch) >= batch_size:
Account.bulk(batch, refresh=True)
del batch[:]
if len(batch) > 0:
Account.bulk(batch)
