Fix token_count for discriminated unions

[BrianPugh]

Addresses #785. @AntoninRousset can you please give this a try? thanks!

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 89.79%. Comparing base (d003ce3) to head (6e7b5b4).
⚠️ Report is 6 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #789      +/-   ##
==========================================
+ Coverage   89.75%   89.79%   +0.04%     
==========================================
  Files          71       71              
  Lines        8218     8251      +33     
  Branches     1861     1863       +2     
==========================================
+ Hits         7376     7409      +33     
  Misses        477      477              
  Partials      365      365              

Flag	Coverage Δ
unittests	89.72% <100.00%> (+0.04%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Copilot]

Pull request overview

Fixes incorrect token_count computation for Pydantic discriminated unions (notably within lists), preventing crashes like ValueError: Cannot Union types that consume different numbers of tokens (issue #785).

Changes:

• Adds a token_count() fast-path for Annotated[..., Field(discriminator=...)] so discriminated unions consume a single token.
• Adds a regression test covering list[DiscriminatedUnion] inside a Pydantic BaseModel.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
cyclopts/_convert.py	Updates token_count() to treat discriminated unions as single-token values.
tests/test_pydantic.py	Adds regression coverage for discriminated-union lists in Pydantic models.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

get_annotated_discriminator(resolve_optional(type_)) is called without first verifying the hint is Annotated[...]. Since get_annotated_discriminator() currently iterates over get_args(annotation)[1:] for any parameterized type, it can return a non-None value for non-Annotated hints whose type parameters happen to have a discriminator attribute, causing token_count() to incorrectly return (1, False). Consider guarding with is_annotated(...) / get_origin(...) is Annotated, or tightening get_annotated_discriminator() to return None unless the origin is Annotated.

Suggested change

	if get_annotated_discriminator(resolve_optional(type_)) is not None:
	resolved_optional_type = resolve_optional(type_)
	if (
	get_origin(resolved_optional_type) is typing.Annotated
	and get_annotated_discriminator(resolved_optional_type) is not None
	):

[Copilot]

This introduces a mutual dependency between cyclopts/_convert.py and cyclopts/argument/utils.py (argument.utils already imports convert_enum_flag from _convert). Even though the import here is inside token_count(), the circular module relationship is fragile and can become an import-time failure if future changes call token_count() during module initialization. Consider moving get_annotated_discriminator to a lower-level module (e.g., cyclopts.annotations) or duplicating a minimal discriminator check locally to avoid the cycle.

[AntoninRousset]

Yes it fixes it, thank you!

The reported missing argument is however wrong, when both Home and Cat or Dog have the same attribute name:

from dataclasses import dataclass
from typing import Annotated, Literal

import cyclopts
import pydantic

class AnimalBase(pydantic.BaseModel):
    name: str

class Cat(AnimalBase):
    dead_birds: int
    dead_rats: int
    type: Literal["cat"] = "cat"


class Dog(AnimalBase):
    type: Literal["dog"] = "dog"

Animal = Annotated[Cat | Dog, pydantic.Field(discriminator="type")]

class Home(pydantic.BaseModel):
    animals: list[Animal]
    name: str

app = cyclopts.App()


@app.default
def main(*, home: Home):
    pass


app(["--home.animals", '{"type": "dog"}', '--home.name', "ISS"])

Prints Parameter "--home.name" requires an argument., despite being given. It should instead print:

1 validation error for Home 
  animals.0.dog.name 
    Field required [type=missing, input_value={'type': 'dog'}, input_type=dict]
      For further information visit https://errors.pydantic.dev/2.12/v/missing 

Do you want me to open another issue for this?