UnicodeEncodeError when running `tortoise makemigrations` due to ASCII encoding in migration writer

[chuwei-dev]

Describe the bug
When running tortoise makemigrations, a UnicodeEncodeError is raised if any model metadata, field description, choices, or comments contain non-ASCII characters (e.g., Chinese characters).

The issue appears to be caused by the migration writer explicitly using encoding="ascii" when writing migration files.

This prevents projects using non-English metadata from generating migrations successfully.

Environment

• tortoise-orm: 1.0.0
• Python: 3.12.3
• OS: Windows 10
• Database: SQLite (also reproducible with others)

To Reproduce

1. Create a model containing non-ASCII characters:

from tortoise import fields, models

class User(models.Model):
    name = fields.CharField(max_length=50, description="用户名")

    class Meta:
        table = "user"

2. Run:

tortoise makemigrations

3. Observe the error:

UnicodeEncodeError: 'ascii' codec can't encode characters in position ...

Stack trace shows the failure occurs at:

tortoise/migrations/writer.py  line 266
path.write_text(self.as_string(), encoding="ascii")

Expected behavior
Migration files should be written using UTF-8 encoding (or default system encoding), allowing non-ASCII characters in:

• field descriptions
• choices labels
• Meta attributes
• comments
• docstrings

Additional context
Migration generation fails if any non-ASCII characters are present, because migration files are written using ASCII encoding.

Analysis
In tortoise/migrations/writer.py, migration files are written with:

path.write_text(self.as_string(), encoding="ascii")

ASCII encoding cannot represent Unicode characters outside the 0–127 range.

Since Python 3 defaults to UTF-8 in most modern environments, explicitly forcing ASCII seems unnecessary and restrictive.

Suggested Fix
Replace:

path.write_text(self.as_string(), encoding="ascii")

With either:

path.write_text(self.as_string(), encoding="utf-8")

or simply:

path.write_text(self.as_string())

Using UTF-8 would align with modern Python defaults and support internationalized projects.