Introducing django-bulk-sync
django-bulk-sync is a package to support the Django ORM that synthesizes the concepts of bulk_create, bulk_update, and delete into a single call: bulk_sync()
.
A Usage Scenario
Companies have zero or more Employees. You want to efficiently sync the names of all employees for a single Company
from an import from that company, but some are added, updated, or removed. The simple approach is inefficient – read the import line by line, and:
For each of N records:
- SELECT to check for the employee’s existence
- UPDATE if it exists, INSERT if it doesn’t
Then figure out some way to identify what was missing and delete it. As is so often the case, the speed of this process is controlled mostly by the number of queries run, and here it is about two queries for every record, and so O(N).
Instead, with bulk_sync
, we can avoid the O(N) number of queries, and simplify the logic we have to write as well.
Example Usage
from django.db.models import Q
from bulk_sync import bulk_sync
new_models = []
for line in company_import_file:
# The `.id` (or `.pk`) field should not be set. Instead, `key_fields`
# tells it how to match.
e = Employee(name=line['name'], phone_number=line['phone_number'], ...)
new_models.append(e)
# `filters` controls the subset of objects considered when deciding to
# update or delete.
filters = Q(company_id=501)
# `key_fields` matches an existing object if all `key_fields` are equal.
key_fields = ('name', )
ret = bulk_sync(
new_models=new_models,
filters=filters,
key_fields=key_fields)
print("Results of bulk_sync: "
"{created} created, {updated} updated, {deleted} deleted."
.format(**ret['stats']))
Under the hood, it will fetch the primary keys of all filtered objects, then atomically call bulk_create
, bulk_update
, and a single queryset delete()
call, to correctly and efficiently update all fields of all employees for the filtered Company, using name
to match properly.
Argument Reference
def bulk_sync(new_models, key_fields, filters, batch_size=None):
Combine bulk create, update, and delete. Make the DB match a set of in-memory objects.
new_models
: An iterable of Django ORMModel
objects that you want stored in the database. They may or may not haveid
set, but you should not have already calledsave()
on them.key_fields
: Identifying attribute name(s) to match upnew_models
items with database rows. If a foreign key is being used as a key field, be sure to pass thefieldname_id
rather than thefieldname
.filters
: Q() filters specifying the subset of the database to work in.batch_size
: passes through to Djangobulk_create.batch_size
andbulk_update.batch_size
, and controls how many objects are created/updated per SQL query.
def bulk_compare(old_models, new_models, key_fields, ignore_fields=None):
Compare two sets of models by key_fields
.
old_models
: Iterable of Django ORM objects to compare.new_models
: Iterable of Django ORM objects to compare.key_fields
: Identifying attribute name(s) to match upnew_models
items with database rows. If a foreign key is being used as a key field, be sure to pass thefieldname_id
rather than thefieldname
.ignore_fields
: (optional) If set, provide field names that should not be considered when comparing objects.- Returns dict of:
{ 'added': list of all added objects. 'unchanged': list of all unchanged objects. 'updated': list of all updated objects. 'updated_details': dict of {obj: {field_name: (old_value, new_value)}} for all changed fields in each updated object. 'removed': list of all removed objects. }
Other techniques
A simpler overall approach might be to simply do this:
with transaction.atomic():
MyObject.objects.filter(filters).delete()
MyObject.objects.bulk_create(new_models)
In some cases, this may be faster (as it skips the initial fetch). However, the bulk_sync
approach has a number of advantages:
- Primary key of existing objects will not change unnecessarily: This may be critical if you use it as a foreign key.
- You cannot find statistics on how many were created/updated/deleted.
auto_now_add
fields will be changed, incorrectly.
Installation and Quick Start
The package is available on pip as django-bulk-sync. Run:
pip install django-bulk-sync
then import via:
from bulk_sync import bulk_sync
And use as in the example above.