Parser: field classes defined in the project itself are not recognised
Nobody has claimed this yet.
Assessment
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Newbie friendliness
- 35/100
Research direction
Start in cli/orm_lens/parser.py: BARE_FIELD_TYPES and BARE_FIELD_RE (around lines 82 and 176) decide which bare calls count as fields, and _pull_imported_bases() (around line 935) already follows imports to the defining file. Agree the approach on the issue first. Done means the repro in the issue reports tax_rate, key_filter and groups correctly, a new test covers the two-hop subclass case, and the golden snapshots are updated.
Written by the indexing model from the issue text.
Description
Source: pretix — src/pretix/helpers/models.py:79 (NormalizedDecimalField), src/pretix/base/models/fields.py:31 (MultiStringField), src/pretix/helpers/countries.py:68 (FastCountryField); weblate — weblate/trans/fields.py:10 (RegexField), weblate/utils/fields.py:34 (UsernameField), weblate/auth/models.py:916 (GroupManyToManyField, used at line 1004)
What happens
Big projects wrap Django fields in their own subclasses:
# helpers/models.py
class NormalizedDecimalField(DecimalField): ...
# models.py
tax_rate = NormalizedDecimalField(max_digits=7, decimal_places=4)
A bare call is only a field when its name is in BARE_FIELD_TYPES, so all of these are dropped: from describe, er and cascade (a dropped relation hides its whole edge), and drift lists them as "migrated but no longer declared".
On pretix most of its ~45 drift entries are this (tax_rate, country, phone, restrict_to_status, ...). On weblate it's User.username, User.groups (a M2M subclass, so the relation is lost too), Component.language_regex/variant_regex/key_filter, Screenshot.image.
Repro
core/fields.py
from django.db import models
from django.db.models import DecimalField
class NormalizedDecimalField(DecimalField):
pass
class RegexField(models.TextField):
pass
class GroupManyToManyField(models.ManyToManyField):
pass
shop/models.py
from django.contrib.auth.models import Group
from django.db import models
from core.fields import GroupManyToManyField, NormalizedDecimalField, RegexField
class Product(models.Model):
name = models.CharField(max_length=100)
tax_rate = NormalizedDecimalField(max_digits=7, decimal_places=4)
key_filter = RegexField(blank=True)
groups = GroupManyToManyField(Group, blank=True)
$ orm-lens describe shop.Product
| `name` | CharField | `max_length=100` |
Expected: tax_rate (DecimalField), key_filter (TextField) and groups as a ManyToManyField relation to Group.
Where
cli/orm_lens/parser.py
- line 82 / 176,
BARE_FIELD_TYPESandBARE_FIELD_RE; - line 935,
_pull_imported_bases(): already follows a module's imports to the file that defines a name. That's the piece to reuse here.
Suggested fix
One possible shape:
- In the body loop, when a line looks like
name = SomeName(butSomeNameisn't whitelisted, remember it as a candidate instead of dropping it. - After the scan, resolve each candidate through the module's imports (same machinery as
_pull_imported_bases). If its class inherits, directly or a few hops up, from a known Django field, record the field with that base's type. If it's aForeignKey/OneToOneField/ManyToManyFieldsubclass, fill the relation data the usual way. - Unresolvable names stay dropped, as today.
Third-party fields (PhoneNumberField, CountryField, I18nCharField, AutoOneToOneField from django-annoying) are a separate question: a small map of well-known ones is fine as a follow-up, not needed here.
This is the bigger one in the batch, hence help wanted. Comment on the issue with your plan first, happy to talk it through.
How to test
- A new test file with the repro (two modules, resolved through the import).
- A subclass two hops away (
class A(models.CharField),class B(A)). - A name that isn't a field class (
x = make_thing(...)) is still ignored. - Run the golden snapshots; saleor/zulip may gain fields, update them and list the changes in the PR.
cd cli && python -m pytest -q,ruffandmypypass.
Difficulty: help wanted, most of a day.
- Dominant language
- Python
- Stars
- 84
- Forks
- 13
- PR merge metrics
- No merged PRs in 30d
Getting set up
- Ships a Dockerfile or Docker Compose file
- Has a pull request template
- Read the contributing guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
More from FROWNINGdev/django-orm-lens
-
suggest-indexes proposes composite indexes that include the primary key or a unique fieldPossibly taken @bfs2021 claimed this today. Openbug good first issue python
Difficulty 2/5 1-3 hours Newbie friendliness 82/100
FROWNINGdev/django-orm-lens#269 ·
-
`migration-risk` text output prints `app.None` for graph-level findingsPossibly taken @phoenixbycaselka claimed this today. Openbug good first issue python
Difficulty 1/5 1-3 hours Newbie friendliness 88/100
FROWNINGdev/django-orm-lens#268 ·
-
Finish the Chinese rule reference — DOL001–DOL007May be free again @harbinresearcher claimed this 64 days ago, and no pull request is open. Opendocumentation good first issue help wanted
Difficulty 2/5 1-3 hours Newbie friendliness 68/100
FROWNINGdev/django-orm-lens#81 · 2 comments ·
-
Fields inherited from Django's `AbstractBaseUser` / `PermissionsMixin` / `AbstractUser` are missingOpenbug good first issue python
Difficulty 3/5 1-2 days Newbie friendliness 62/100
FROWNINGdev/django-orm-lens#267 ·
-
bug good first issue python
Difficulty 3/5 Half a day Newbie friendliness 78/100
FROWNINGdev/django-orm-lens#266 ·
All issues in FROWNINGdev/django-orm-lens
Similar issues
-
Difficulty 2/5 1-3 hours Newbie friendliness 62/100
roman-corgi/corgisim#327 ·
Maintainers usually reply within 2 days
-
[Bug]: opencode_go chat route drops reasoning_effort (even 'none'), while the same endpoint honors it via custom_llm_provider=openaiPossibly taken @arvindkumar-ship-it claimed this today. Openllm translation
Difficulty 2/5 1-3 hours Newbie friendliness 70/100
BerriAI/litellm#51871 · 1 comment ·
Maintainers usually reply within 1 day
-
Difficulty 2/5 1-3 hours Newbie friendliness 72/100
Maintainers usually reply within 1 day
-
PolicyEngine backdates South Carolina’s EITC into 2015–2017Possibly taken @MaxGhenis claimed this today. Open
Difficulty 2/5 1-3 hours Newbie friendliness 72/100
PolicyEngine/policyengine-us#10944 ·
Maintainers usually reply within 2 days
-
fix(frontend): resolve SSE parser crash on unhandled Ref methods and undefined actions identifierOpenbug
Difficulty 2/5 1-3 hours Newbie friendliness 75/100
kodokbakar/MiroFish-Local#94 ·
Maintainers usually reply within 1 day