change a db from a certain point in time, when the change doesn't fit the already existing data - database

Question

I have a model that looks like this:
class Report(models.Model):
updater = models.CharField(max_length=15)
pub_date = models.DateTimeField(auto_add_now=True)
identifier = models.CharField(max_length=100)
... and so on...
There are some more fields but they are irrelevant to the question. Now the site has very simple functions - the users can see older reports and their data, and can edit them or add new ones.
However, the identifier field is actually an integer that symbolizes a log file that is being reported. Most of the times, each report has one log. But sometimes it has more than one. I did it as a CharField because I built the site to replace an older sharepoint 2003 website, where that field was treated as simple text. So I want that in my next version, it would be like it should be, i.e. like this:
class Report(models.Model):
updater = models.CharField(max_length=15)
pub_date = models.DateTimeField(auto_add_now=True)
... and so on...
class Log(models.Model):
report = models.ForeignKey(Report)
identifier = models.IntegerField()
The problem is, since in the old site that field was a CharField, people used this as they liked. Meaning, even if they updated various logs in the same report they just did it like this <logid1>, <logid2>. Sometimes they added some text <logid1> which is related to <logid2>.
So I want to change this, but I don't want to lose all the old data, and I can't fix all those edge cases (the DB contains around 22 thousand reports). I thought about adding this to report:
def disp_id(self):
if self.pub_date < ... #the day I'll do the update
return self.identifier
else:
return ', '.join([log.identifier for log in self.log_set.all()])
But then I'm not really getting rid of the old field now am I? I'm just adding a new one and keeping the original null from a certain date.
As far as I know, what I want to do is impossible. I'm only asking because I know that maybe I'm not the first one to deal with this sort of thing and maybe there is a solution that I'm not aware of.
Hope my explanation is clear enough, thanks in advance!

Answer 1

class Report(models.Model):
updater = models.CharField(max_length=15)
pub_date = models.DateTimeField(auto_add_now=True)
identifier = models.CharField(null=True)
... and so on...
logs = models.ManyToManyField(Log,null=True)
class Log(models.Model):
identifier = models.IntegerField()
Make the above model , and then make a script as follow:
ident_list = []
for reports in Report.objects.all():
identifiers = reports.identifiers.split(',')
for idents in identifiers:
if not idents in ident_list:
log = Log.create(**{'identifier' : int(idents)})
ident_list.append(int(idents))
else:
log = Log.objects.get(identifier = int(idents))
report.log.add(log)
Check the data before removing the column identifiers from the table Report.
Does it solves your purpose now ?