Opened 5 years ago

Closed 5 years ago

#30899 closed Cleanup/optimization (fixed)

Lazily compile regular expressions.

Reported by: Adam Johnson Owned by: Hasan Ramezani
Component: Core (Other) Version: dev
Severity: Normal Keywords:
Cc: me@… Triage Stage: Ready for checkin
Has patch: yes Needs documentation: no
Needs tests: no Patch needs improvement: no
Easy pickings: no UI/UX: no

Description

Inspired by [this article from Instagram](​https://instagram-engineering.com/python-at-scale-strict-modules-c0bb9245c834)

Currently Django lazily compiles some regular expressions such as those in validators: ​https://github.com/django/django/blob/54ea290e5bbd19d87bd8dba807738eeeaf01a362/django/core/validators.py#L17

This is to save on import time.

There are other import-time re.compile calls throughout the codebase, these could be migrated to lazy regex compiles to save on their import time:

​https://github.com/django/django/search?q=re.compile&unscoped_q=re.compile

Change History (8)

comment:1 by Mariusz Felisiak, 5 years ago

Summary: Lazily compile large regular expressions → Lazily compile regular expressions.
Triage Stage: Unreviewed → Accepted

comment:2 by Hasan Ramezani, 5 years ago

Owner: changed from nobody to Hasan Ramezani
Status: new → assigned

I can work on this an prepare a patch.
Should I move the ​_lazy_re_compile function to ​django.utils.text?

comment:3 by Adam Johnson, 5 years ago

Yeah django.utils.text sounds like a sensible location to me. Keeping it private (starting with underscore, undocumented) for now seems wise.

comment:4 by Hasan Ramezani, 5 years ago

Has patch: set

Description provided on the ​PR

Last edited 5 years ago by Hasan Ramezani (previous) (diff)

comment:5 by Mariusz Felisiak, 5 years ago

Triage Stage: Accepted → Ready for checkin

comment:6 by Mariusz Felisiak <felisiak.mariusz@…>, 5 years ago

In c4cba148:

Refs #30899 -- Moved _lazy_re_compile() to the django.utils.regex_helper.

comment:7 by Mariusz Felisiak <felisiak.mariusz@…>, 5 years ago

In 39a34d4:

Refs #30899 -- Made _lazy_re_compile() support bytes.

comment:8 by Mariusz Felisiak <felisiak.mariusz@…>, 5 years ago

Resolution: → fixed
Status: assigned → closed

In e3d0b4d5:

Fixed #30899 -- Lazily compiled import time regular expressions.

Note: See TracTickets for help on using tickets.
Back to Top