Question

I’m looking for a way to do this:

qs = MyModel.objects.filter(mystring__like="____10____")

#Which would create a sql clause
... LIKE '____10____'

instead of behave like this:

qs = MyModel.objects.filter(mystring__icontains="____10____")
#Which creates a sql clause
... LIKE %\_\\_\\_\\_10\\_\\_\\_\\_%

I know I can use a regex filter, but that’s substantially slower and more error prone than just using the built-in mysql wildcard feature (I’ve tested it directly in mysql, the query strings are long enough that the difference is substantial).

---

EDIT:
figured out how to do this with the .extra() method with madisvain’s help.

qs = MyModel.objects.extra(where=["`mystring` LIKE '____10____'"])

In terms of performance difference, 2000 random queries with the regex approach took 20.5 seconds, with this approach 2000 random queries take 6 seconds.

Answer 1

I didn’t even know django had this __like version. Well but according to the docs please read this!
https://docs.djangoproject.com/en/dev/topics/db/queries/#escaping-percent-signs-and-underscores-in-like-statements

You should be writing the query like this.

    qs = MyModel.objects.filter(mystring__contains="____10____")

Or this.

    qs = MyModel.objects.extra(where="mystring LIKE '____10____'")

Read more about the extra() method here:
https://docs.djangoproject.com/en/dev/ref/models/querysets/#extra