suggest-indexes proposes composite indexes that include the primary key or a unique field

Open Beginner friendly
  • Is something wrong?
#269 0 comments 0 reactions 0 assignees View on GitHub

@bfs2021 is already working on this.

Since Nov 14, 2026.

  • #272 by @bfs2021 — open

Assessment

Difficulty
2/5
Estimated time
1-3 hours
Newbie friendliness
82/100
Issue type
Bug
Clarity
Clearly specified
Activity status
Active
Tech stack
django, python
Domain
cli, database

Research direction

Start in cli/orm_lens/query_analyzer.py at suggest_indexes(), where pk_name and target_model are available, then read the composite loop at the top of _propose_indexes(). Build a set of unique single columns (the PK, pk, primary_key=True and unique=True fields, and OneToOneFields) and skip any composite that contains one of them, leaving the single-field logic alone. Done when the Story repro yields only ['status'] and ['title'], pk= behaves like id=, and cli/tests/test_query_analyzer.py plus pytest, ruff and mypy pass.

Written by the indexing model from the issue text.

Description

bug good first issue python

Source: weblate — trans.Unit: models.Index(fields=['id', 'translation'])

What happens

filter(id=pk, title=title) seen twice gives models.Index(fields=['id', 'title']). The primary key is unique: an equality lookup on it returns at most one row, so the PK index already answers the query and a composite led by id only costs writes and disk. The same is true for any unique=True field.

_propose_indexes() uses already_indexed for single-field proposals only. The composite loop checks the M2M/reverse filter from #252/#263 and existing Meta groups, nothing else. pk= is folded into id before that, so both spellings hit it.

On weblate trans.Unit gets models.Index(fields=['id', 'translation']).

Repro

board/models.py

from django.db import models


class Story(models.Model):
    title = models.CharField(max_length=200)
    slug = models.SlugField(unique=True)
    status = models.CharField(max_length=20)

board/stats.py

from .models import Story


def by_id_and_title(pk, title):
    return Story.objects.filter(id=pk, title=title).first()


def by_id_and_title_again(pk, title):
    return Story.objects.get(id=pk, title=title)


def by_pk_and_status(pk, status):
    return Story.objects.filter(pk=pk, status=status).exists()


def by_pk_and_status_again(pk, status):
    return Story.objects.filter(pk=pk, status=status).first()


def by_slug_and_status(slug, status):
    return Story.objects.filter(slug=slug, status=status).first()


def by_slug_and_status_again(slug, status):
    return Story.objects.filter(slug=slug, status=status).exists()

$ orm-lens suggest-indexes board.Story
target: board.Story
proposed Meta.indexes (5):
  models.Index(fields=['id', 'status']),  # 2 composite filter sites
  models.Index(fields=['id', 'title']),  # 2 composite filter sites
  models.Index(fields=['slug', 'status']),  # 2 composite filter sites
  models.Index(fields=['status']),  # 4 filter sites
  models.Index(fields=['title']),  # 2 filter sites

Expected: only ['status'] and ['title']. The usage table can keep listing id + title etc.

Where

cli/orm_lens/query_analyzer.py

  • line 497, _propose_indexes(); the composite loop is right at the top of the function body;
  • line 309, _indexed_single_fields(): builds already_indexed, but it also contains db_index=True fields and FKs, which are not unique, so it can't be reused as is;
  • line 598, suggest_indexes(): pk_name (line 675) and target_model are both there.
Suggested fix

In suggest_indexes(), build the set of unique single columns: pk_name, "pk", every field with primary_key=True or unique=True (the _PK_TRUE_RE / _UNIQUE_TRUE_RE regexes are already in the module), and OneToOneFields (unique by definition). Pass it to _propose_indexes() as a new keyword argument and skip a composite combo that contains any of these names.

Don't touch the single-field and order_by logic.

How to test
  • cli/tests/test_query_analyzer.py: the repro gives exactly ['status'] and ['title'].
  • filter(pk=..., ...) is dropped the same as filter(id=..., ...).
  • A model with code = models.CharField(primary_key=True) drops a ['code', 'title'] composite.
  • A composite of two non-unique fields (filter(title=..., status=...)) is still proposed.
  • cd cli && python -m pytest -q, ruff and mypy pass.

Difficulty: good first issue, one set and one if.

Dominant language
Python
Stars
84
Forks
13
No merged PRs in 30d

Getting set up

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from FROWNINGdev/django-orm-lens

All issues in FROWNINGdev/django-orm-lens

Similar issues

More Python issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.