Bump version to 0.8.0 and update changelog

* chore: upgrade deps, update changelog, drop test db before create

* tests: clear operators for sqlite ddl after NotSupportError raised

.github/workflows/ci.yml

@@ -2,10 +2,10 @@ name: ci
 on:
   push:
     branches-ignore:
-      - master
+      - main
   pull_request:
     branches-ignore:
-      - master
+      - main
 jobs:
   ci:
     runs-on: ubuntu-latest
@@ -18,13 +18,22 @@ jobs:
           POSTGRES_PASSWORD: 123456
           POSTGRES_USER: postgres
         options: --health-cmd=pg_isready --health-interval 10s --health-timeout 5s --health-retries 5
+    strategy:
+      matrix:
+        python-version: ["3.8", "3.9", "3.10", "3.11", "3.12"]
     steps:
       - name: Start MySQL
         run: sudo systemctl start mysql.service
-      - uses: actions/checkout@v2
+      - uses: actions/cache@v4
-      - uses: actions/setup-python@v2
         with:
-          python-version: '3.x'
+          path: ~/.cache/pip
+          key: ${{ runner.os }}-pip-${{ hashFiles('**/poetry.lock') }}
+          restore-keys: |
+            ${{ runner.os }}-pip-
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: ${{ matrix.python-version }}
       - name: Install and configure Poetry
         run: |
           pip install -U pip poetry

.github/workflows/pypi.yml

@@ -7,8 +7,8 @@ jobs:
   publish:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v4
-      - uses: actions/setup-python@v2
+      - uses: actions/setup-python@v5
         with:
           python-version: '3.x'
       - name: Install and configure Poetry

CHANGELOG.md

@@ -1,7 +1,59 @@
 # ChangeLog
 
+## 0.8
+
+### [0.8.0](../../releases/tag/v0.8.0) - 2024-12-04
+
+- Fix the issue of parameter concatenation when generating ORM with inspectdb (#331)
+- Fix KeyError when deleting a field with unqiue=True. (#364)
+- Correct the click import. (#360)
+- Improve CLI help text and output. (#355)
+- Fix mysql drop unique index raises OperationalError. (#346)
+  **Upgrade note:**
+    1. Use column name as unique key name for mysql
+    2. Drop support for Python3.7
+
+## 0.7
+
+### [0.7.2](../../releases/tag/v0.7.2) - 2023-07-20
+
+- Support virtual fields.
+- Fix modify multiple times. (#279)
+- Added `-i` and `--in-transaction` options to `aerich migrate` command. (#296)
+- Fix generates two semicolons in a row. (#301)
+
+### 0.7.1
+
+- Fix syntax error with python3.8.10. (#265)
+- Fix sql generate error. (#263)
+- Fix initialize an empty database. (#267)
+
+### 0.7.1rc1
+
+- Fix postgres sql error (#263)
+
+### 0.7.0
+
+**Now aerich use `.py` file to record versions.**
+
+Upgrade Note:
+
+1. Drop `aerich` table
+2. Delete `migrations/models` folder
+3. Run `aerich init-db`
+
+- Improve `inspectdb` adding support to `postgresql::numeric` data type
+- Add support for dynamically load DDL classes easing to add support to
+  new databases without changing `Migrate` class logic
+- Fix decimal field change. (#246)
+- Support add/remove field with index.
+
 ## 0.6
 
+### 0.6.3
+
+- Improve `inspectdb` and support `postgres` & `sqlite`.
+
 ### 0.6.2
 
 - Support migration for specified index. (#203)

Makefile

@@ -14,14 +14,20 @@ up:
 deps:
 	@poetry install -E asyncpg -E asyncmy
 
-style: deps
+_style:
-	isort -src $(checkfiles)
+	@isort -src $(checkfiles)
-	black $(black_opts) $(checkfiles)
+	@black $(black_opts) $(checkfiles)
+style: deps _style
 
-check: deps
+_check:
-	black --check $(black_opts) $(checkfiles) || (echo "Please run 'make style' to auto-fix style issues" && false)
+	@black --check $(black_opts) $(checkfiles) || (echo "Please run 'make style' to auto-fix style issues" && false)
-	flake8 $(checkfiles)
+	@ruff check $(checkfiles)
-	bandit -x tests -r $(checkfiles)
+	@mypy $(checkfiles)
+ifneq ($(shell python -c 'import sys;is_py38=sys.version_info<(3,9);rc=int(is_py38);sys.exit(rc)'),)
+	# Run bandit with Python3.9+, as the `usedforsecurity=...` parameter of `hashlib.new` is only added from Python 3.9 onwards.
+	@bandit -r aerich
+endif
+check: deps _check
 
 test: deps
 	$(py_warn) TEST_DB=sqlite://:memory: py.test
@@ -35,9 +41,10 @@ test_mysql:
 test_postgres:
 	$(py_warn) TEST_DB="postgres://postgres:$(POSTGRES_PASS)@$(POSTGRES_HOST):$(POSTGRES_PORT)/test_\{\}" pytest -vv -s
 
-testall: deps test_sqlite test_postgres test_mysql
+_testall: test_sqlite test_postgres test_mysql
+testall: deps _testall
 
 build: deps
 	@poetry build
 
-ci: check testall
+ci: check _testall

README.md

@@ -5,6 +5,8 @@
 [![image](https://github.com/tortoise/aerich/workflows/pypi/badge.svg)](https://github.com/tortoise/aerich/actions?query=workflow:pypi)
 [![image](https://github.com/tortoise/aerich/workflows/ci/badge.svg)](https://github.com/tortoise/aerich/actions?query=workflow:ci)
 
+English | [Русский](./README_RU.md)
+
 ## Introduction
 
 Aerich is a database migrations tool for TortoiseORM, which is like alembic for SQLAlchemy, or like Django ORM with
@@ -44,7 +46,7 @@ Commands:
 
 ## Usage
 
-You need add `aerich.models` to your `Tortoise-ORM` config first. Example:
+You need to add `aerich.models` to your `Tortoise-ORM` config first. Example:
 
 ```python
 TORTOISE_ORM = {
@@ -101,22 +103,30 @@ e.g. `aerich --app other_models init-db`.
 ```shell
 > aerich migrate --name drop_column
 
-Success migrate 1_202029051520102929_drop_column.sql
+Success migrate 1_202029051520102929_drop_column.py
 ```
 
 Format of migrate filename is
-`{version_num}_{datetime}_{name|update}.sql`.
+`{version_num}_{datetime}_{name|update}.py`.
 
 If `aerich` guesses you are renaming a column, it will ask `Rename {old_column} to {new_column} [True]`. You can choose
 `True` to rename column without column drop, or choose `False` to drop the column then create. Note that the latter may
 lose data.
 
+If you need to manually write migration, you could generate empty file:
+
+```shell
+> aerich migrate --name add_index --empty
+
+Success migrate 1_202326122220101229_add_index.py
+```
+
 ### Upgrade to latest version
 
 ```shell
 > aerich upgrade
 
-Success upgrade 1_202029051520102929_drop_column.sql
+Success upgrade 1_202029051520102929_drop_column.py
 ```
 
 Now your db is migrated to latest.
@@ -142,7 +152,7 @@ Options:
 ```shell
 > aerich downgrade
 
-Success downgrade 1_202029051520102929_drop_column.sql
+Success downgrade 1_202029051520102929_drop_column.py
 ```
 
 Now your db is rolled back to the specified version.
@@ -152,7 +162,7 @@ Now your db is rolled back to the specified version.
 ```shell
 > aerich history
 
-1_202029051520102929_drop_column.sql
+1_202029051520102929_drop_column.py
 ```
 
 ### Show heads to be migrated
@@ -160,12 +170,12 @@ Now your db is rolled back to the specified version.
 ```shell
 > aerich heads
 
-1_202029051520102929_drop_column.sql
+1_202029051520102929_drop_column.py
 ```
 
 ### Inspect db tables to TortoiseORM model
 
-Currently `inspectdb` only supports MySQL.
+Currently `inspectdb` support MySQL & Postgres & SQLite.
 
 ```shell
 Usage: aerich inspectdb [OPTIONS]
@@ -189,7 +199,44 @@ Inspect a specified table in the default app and redirect to `models.py`:
 aerich inspectdb -t user > models.py
 ```
 
-Note that this command is limited and cannot infer some fields, such as `IntEnumField`, `ForeignKeyField`, and others.
+For example, you table is:
+
+```sql
+CREATE TABLE `test`
+(
+    `id`       int            NOT NULL AUTO_INCREMENT,
+    `decimal`  decimal(10, 2) NOT NULL,
+    `date`     date                                    DEFAULT NULL,
+    `datetime` datetime       NOT NULL                 DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP,
+    `time`     time                                    DEFAULT NULL,
+    `float`    float                                   DEFAULT NULL,
+    `string`   varchar(200) COLLATE utf8mb4_general_ci DEFAULT NULL,
+    `tinyint`  tinyint                                 DEFAULT NULL,
+    PRIMARY KEY (`id`),
+    KEY `asyncmy_string_index` (`string`)
+) ENGINE = InnoDB
+  DEFAULT CHARSET = utf8mb4
+  COLLATE = utf8mb4_general_ci
+```
+
+Now run `aerich inspectdb -t test` to see the generated model:
+
+```python
+from tortoise import Model, fields
+
+
+class Test(Model):
+    date = fields.DateField(null=True, )
+    datetime = fields.DatetimeField(auto_now=True, )
+    decimal = fields.DecimalField(max_digits=10, decimal_places=2, )
+    float = fields.FloatField(null=True, )
+    id = fields.IntField(pk=True, )
+    string = fields.CharField(max_length=200, null=True, )
+    time = fields.TimeField(null=True, )
+    tinyint = fields.BooleanField(null=True, )
+```
+
+Note that this command is limited and can't infer some fields, such as `IntEnumField`, `ForeignKeyField`, and others.
 
 ### Multiple databases
 

README_RU.md

@@ -0,0 +1,274 @@
+# Aerich
+ 
+[![image](https://img.shields.io/pypi/v/aerich.svg?style=flat)](https://pypi.python.org/pypi/aerich)
+[![image](https://img.shields.io/github/license/tortoise/aerich)](https://github.com/tortoise/aerich)
+[![image](https://github.com/tortoise/aerich/workflows/pypi/badge.svg)](https://github.com/tortoise/aerich/actions?query=workflow:pypi)
+[![image](https://github.com/tortoise/aerich/workflows/ci/badge.svg)](https://github.com/tortoise/aerich/actions?query=workflow:ci)
+ 
+[English](./README.md) | Русский
+
+## Введение
+ 
+Aerich - это инструмент для миграции базы данных для TortoiseORM, который аналогичен Alembic для SQLAlchemy или встроенному решению миграций в Django ORM.
+ 
+## Установка
+ 
+Просто установите из pypi:
+ 
+```shell
+pip install aerich
+```
+ 
+## Быстрый старт
+ 
+```shell
+> aerich -h
+ 
+Usage: aerich [OPTIONS] COMMAND [ARGS]...
+ 
+Options:
+  -V, --version      Show the version and exit.
+  -c, --config TEXT  Config file.  [default: pyproject.toml]
+  --app TEXT         Tortoise-ORM app name.
+  -h, --help         Show this message and exit.
+ 
+Commands:
+  downgrade  Downgrade to specified version.
+  heads      Show current available heads in migrate location.
+  history    List all migrate items.
+  init       Init config file and generate root migrate location.
+  init-db    Generate schema and generate app migrate location.
+  inspectdb  Introspects the database tables to standard output as...
+  migrate    Generate migrate changes file.
+  upgrade    Upgrade to specified version.
+```
+ 
+## Использование
+ 
+Сначала вам нужно добавить aerich.models в конфигурацию вашего Tortoise-ORM. Пример:
+ 
+```python
+TORTOISE_ORM = {
+    "connections": {"default": "mysql://root:123456@127.0.0.1:3306/test"},
+    "apps": {
+        "models": {
+            "models": ["tests.models", "aerich.models"],
+            "default_connection": "default",
+        },
+    },
+}
+```
+ 
+### Инициализация
+ 
+```shell
+> aerich init -h
+ 
+Usage: aerich init [OPTIONS]
+ 
+  Init config file and generate root migrate location.
+ 
+Options:
+  -t, --tortoise-orm TEXT  Tortoise-ORM config module dict variable, like
+                           settings.TORTOISE_ORM.  [required]
+  --location TEXT          Migrate store location.  [default: ./migrations]
+  -s, --src_folder TEXT    Folder of the source, relative to the project root.
+  -h, --help               Show this message and exit.
+```
+ 
+Инициализируйте файл конфигурации и задайте местоположение миграций:
+ 
+```shell
+> aerich init -t tests.backends.mysql.TORTOISE_ORM
+ 
+Success create migrate location ./migrations
+Success write config to pyproject.toml
+```
+ 
+### Инициализация базы данных
+ 
+```shell
+> aerich init-db
+ 
+Success create app migrate location ./migrations/models
+Success generate schema for app "models"
+```
+ 
+Если ваше приложение Tortoise-ORM не является приложением по умолчанию с именем models, вы должны указать правильное имя приложения с помощью параметра --app, например: aerich --app other_models init-db.
+ 
+### Обновление моделей и создание миграции
+ 
+```shell
+> aerich migrate --name drop_column
+ 
+Success migrate 1_202029051520102929_drop_column.py
+```
+ 
+Формат имени файла миграции следующий: `{версия}_{дата_и_время}_{имя|обновление}.py`.
+ 
+Если aerich предполагает, что вы переименовываете столбец, он спросит: 
+Переименовать `{старый_столбец} в {новый_столбец} [True]`. Вы можете выбрать `True`,
+чтобы переименовать столбец без удаления столбца, или выбрать `False`, чтобы удалить столбец,
+а затем создать новый. Обратите внимание, что последний вариант может привести к потере данных.
+ 
+ 
+### Обновление до последней версии
+ 
+```shell
+> aerich upgrade
+ 
+Success upgrade 1_202029051520102929_drop_column.py
+```
+ 
+Теперь ваша база данных обновлена до последней версии.
+ 
+### Откат до указанной версии
+ 
+```shell
+> aerich downgrade -h
+ 
+Usage: aerich downgrade [OPTIONS]
+ 
+  Downgrade to specified version.
+ 
+Options:
+  -v, --version INTEGER  Specified version, default to last.  [default: -1]
+  -d, --delete           Delete version files at the same time.  [default:
+                         False]
+ 
+  --yes                  Confirm the action without prompting.
+  -h, --help             Show this message and exit.
+```
+ 
+```shell
+> aerich downgrade
+ 
+Success downgrade 1_202029051520102929_drop_column.py
+```
+ 
+Теперь ваша база данных откатилась до указанной версии.
+ 
+### Показать историю
+ 
+```shell
+> aerich history
+ 
+1_202029051520102929_drop_column.py
+```
+ 
+### Чтобы узнать, какие миграции должны быть применены, можно использовать команду:
+ 
+```shell
+> aerich heads
+ 
+1_202029051520102929_drop_column.py
+```
+ 
+### Осмотр таблиц базы данных для модели TortoiseORM
+ 
+В настоящее время inspectdb поддерживает MySQL, Postgres и SQLite.
+ 
+```shell
+Usage: aerich inspectdb [OPTIONS]
+ 
+  Introspects the database tables to standard output as TortoiseORM model.
+ 
+Options:
+  -t, --table TEXT  Which tables to inspect.
+  -h, --help        Show this message and exit.
+```
+ 
+Посмотреть все таблицы и вывести их на консоль:
+ 
+```shell
+aerich --app models inspectdb
+```
+ 
+Осмотреть указанную таблицу в приложении по умолчанию и перенаправить в models.py:
+ 
+```shell
+aerich inspectdb -t user > models.py
+```
+ 
+Например, ваша таблица выглядит следующим образом:
+ 
+```sql
+CREATE TABLE `test`
+(
+    `id`       int            NOT NULL AUTO_INCREMENT,
+    `decimal`  decimal(10, 2) NOT NULL,
+    `date`     date                                    DEFAULT NULL,
+    `datetime` datetime       NOT NULL                 DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP,
+    `time`     time                                    DEFAULT NULL,
+    `float`    float                                   DEFAULT NULL,
+    `string`   varchar(200) COLLATE utf8mb4_general_ci DEFAULT NULL,
+    `tinyint`  tinyint                                 DEFAULT NULL,
+    PRIMARY KEY (`id`),
+    KEY `asyncmy_string_index` (`string`)
+) ENGINE = InnoDB
+  DEFAULT CHARSET = utf8mb4
+  COLLATE = utf8mb4_general_ci
+```
+ 
+Теперь выполните команду aerich inspectdb -t test, чтобы увидеть сгенерированную модель:
+ 
+```python
+from tortoise import Model, fields
+ 
+ 
+class Test(Model):
+    date = fields.DateField(null=True, )
+    datetime = fields.DatetimeField(auto_now=True, )
+    decimal = fields.DecimalField(max_digits=10, decimal_places=2, )
+    float = fields.FloatField(null=True, )
+    id = fields.IntField(pk=True, )
+    string = fields.CharField(max_length=200, null=True, )
+    time = fields.TimeField(null=True, )
+    tinyint = fields.BooleanField(null=True, )
+```
+ 
+Обратите внимание, что эта команда имеет ограничения и не может автоматически определить некоторые поля, такие как `IntEnumField`, `ForeignKeyField` и другие.
+ 
+### Несколько баз данных
+ 
+```python
+tortoise_orm = {
+    "connections": {
+        "default": expand_db_url(db_url, True),
+        "second": expand_db_url(db_url_second, True),
+    },
+    "apps": {
+        "models": {"models": ["tests.models", "aerich.models"], "default_connection": "default"},
+        "models_second": {"models": ["tests.models_second"], "default_connection": "second", },
+    },
+}
+```
+ 
+Вам нужно указать `aerich.models` только в одном приложении и должны указывать `--app` при запуске команды `aerich migrate` и т.д.
+ 
+## Восстановление рабочего процесса aerich
+ 
+В некоторых случаях, например, при возникновении проблем после обновления `aerich`, вы не можете запустить `aerich migrate` или `aerich upgrade`. В таком случае вы можете выполнить следующие шаги:
+ 
+1. удалите таблицы `aerich`.
+2. удалите директорию `migrations/{app}`.
+3. rerun `aerich init-db`.
+ 
+Обратите внимание, что эти действия безопасны, и вы можете использовать их для сброса миграций, если у вас слишком много файлов миграции.
+ 
+## Использование aerich в приложении
+ 
+Вы можете использовать `aerich` вне командной строки, используя класс `Command`.
+ 
+```python
+from aerich import Command
+ 
+command = Command(tortoise_config=config, app='models')
+await command.init()
+await command.migrate('test')
+```
+ 
+## Лицензия
+ 
+Этот проект лицензирован в соответствии с лицензией
+[Apache-2.0](https://github.com/long2ice/aerich/blob/master/LICENSE) Лицензия.

aerich/__init__.py

@@ -1,6 +1,6 @@
 import os
 from pathlib import Path
-from typing import List
+from typing import TYPE_CHECKING, List, Optional, Type
 
 from tortoise import Tortoise, generate_schema_for_client
 from tortoise.exceptions import OperationalError
@@ -8,17 +8,21 @@ from tortoise.transactions import in_transaction
 from tortoise.utils import get_schema_sql
 
 from aerich.exceptions import DowngradeError
-from aerich.inspectdb import InspectDb
+from aerich.inspectdb.mysql import InspectMySQL
-from aerich.migrate import Migrate
+from aerich.inspectdb.postgres import InspectPostgres
+from aerich.inspectdb.sqlite import InspectSQLite
+from aerich.migrate import MIGRATE_TEMPLATE, Migrate
 from aerich.models import Aerich
 from aerich.utils import (
     get_app_connection,
     get_app_connection_name,
     get_models_describe,
-    get_version_content_from_file,
+    import_py_file,
-    write_version_file,
 )
 
+if TYPE_CHECKING:
+    from aerich.inspectdb import Inspect  # noqa:F401
+
 
 class Command:
     def __init__(
@@ -26,16 +30,27 @@ class Command:
         tortoise_config: dict,
         app: str = "models",
         location: str = "./migrations",
-    ):
+    ) -> None:
         self.tortoise_config = tortoise_config
         self.app = app
         self.location = location
         Migrate.app = app
 
-    async def init(self):
+    async def init(self) -> None:
         await Migrate.init(self.tortoise_config, self.app, self.location)
 
-    async def upgrade(self):
+    async def _upgrade(self, conn, version_file) -> None:
+        file_path = Path(Migrate.migrate_location, version_file)
+        m = import_py_file(file_path)
+        upgrade = getattr(m, "upgrade")
+        await conn.execute_script(await upgrade(conn))
+        await Aerich.create(
+            version=version_file,
+            app=self.app,
+            content=get_models_describe(self.app),
+        )
+
+    async def upgrade(self, run_in_transaction: bool = True) -> List[str]:
         migrated = []
         for version_file in Migrate.get_all_version_files():
             try:
@@ -43,24 +58,18 @@ class Command:
             except OperationalError:
                 exists = False
             if not exists:
-                async with in_transaction(
+                app_conn_name = get_app_connection_name(self.tortoise_config, self.app)
-                    get_app_connection_name(self.tortoise_config, self.app)
+                if run_in_transaction:
-                ) as conn:
+                    async with in_transaction(app_conn_name) as conn:
-                    file_path = Path(Migrate.migrate_location, version_file)
+                        await self._upgrade(conn, version_file)
-                    content = get_version_content_from_file(file_path)
+                else:
-                    upgrade_query_list = content.get("upgrade")
+                    app_conn = get_app_connection(self.tortoise_config, self.app)
-                    for upgrade_query in upgrade_query_list:
+                    await self._upgrade(app_conn, version_file)
-                        await conn.execute_script(upgrade_query)
-                    await Aerich.create(
-                        version=version_file,
-                        app=self.app,
-                        content=get_models_describe(self.app),
-                    )
                 migrated.append(version_file)
         return migrated
 
-    async def downgrade(self, version: int, delete: bool):
+    async def downgrade(self, version: int, delete: bool) -> List[str]:
-        ret = []
+        ret: List[str] = []
         if version == -1:
             specified_version = await Migrate.get_last_version()
         else:
@@ -73,25 +82,25 @@ class Command:
             versions = [specified_version]
         else:
             versions = await Aerich.filter(app=self.app, pk__gte=specified_version.pk)
-        for version in versions:
+        for version_obj in versions:
-            file = version.version
+            file = version_obj.version
             async with in_transaction(
                 get_app_connection_name(self.tortoise_config, self.app)
             ) as conn:
                 file_path = Path(Migrate.migrate_location, file)
-                content = get_version_content_from_file(file_path)
+                m = import_py_file(file_path)
-                downgrade_query_list = content.get("downgrade")
+                downgrade = getattr(m, "downgrade")
-                if not downgrade_query_list:
+                downgrade_sql = await downgrade(conn)
+                if not downgrade_sql.strip():
                     raise DowngradeError("No downgrade items found")
-                for downgrade_query in downgrade_query_list:
+                await conn.execute_script(downgrade_sql)
-                    await conn.execute_query(downgrade_query)
+                await version_obj.delete()
-                await version.delete()
                 if delete:
                     os.unlink(file_path)
                 ret.append(file)
         return ret
 
-    async def heads(self):
+    async def heads(self) -> List[str]:
         ret = []
         versions = Migrate.get_all_version_files()
         for version in versions:
@@ -99,22 +108,28 @@ class Command:
                 ret.append(version)
         return ret
 
-    async def history(self):
+    async def history(self) -> List[str]:
-        ret = []
         versions = Migrate.get_all_version_files()
-        for version in versions:
+        return [version for version in versions]
-            ret.append(version)
-        return ret
 
-    async def inspectdb(self, tables: List[str]):
+    async def inspectdb(self, tables: Optional[List[str]] = None) -> str:
         connection = get_app_connection(self.tortoise_config, self.app)
-        inspect = InspectDb(connection, tables)
+        dialect = connection.schema_generator.DIALECT
-        await inspect.inspect()
+        if dialect == "mysql":
+            cls: Type["Inspect"] = InspectMySQL
+        elif dialect == "postgres":
+            cls = InspectPostgres
+        elif dialect == "sqlite":
+            cls = InspectSQLite
+        else:
+            raise NotImplementedError(f"{dialect} is not supported")
+        inspect = cls(connection, tables)
+        return await inspect.inspect()
 
-    async def migrate(self, name: str = "update"):
+    async def migrate(self, name: str = "update", empty: bool = False) -> str:
-        return await Migrate.migrate(name)
+        return await Migrate.migrate(name, empty)
 
-    async def init_db(self, safe: bool):
+    async def init_db(self, safe: bool) -> None:
         location = self.location
         app = self.app
         dirname = Path(location, app)
@@ -132,7 +147,7 @@ class Command:
             app=app,
             content=get_models_describe(app),
         )
-        content = {
+        version_file = Path(dirname, version)
-            "upgrade": [schema],
+        content = MIGRATE_TEMPLATE.format(upgrade_sql=schema, downgrade_sql="")
-        }
+        with open(version_file, "w", encoding="utf-8") as f:
-        write_version_file(Path(dirname, version), content)
+            f.write(content)

aerich/cli.py

@@ -1,14 +1,11 @@
-import asyncio
 import os
-from functools import wraps
 from pathlib import Path
-from typing import List
+from typing import Dict, List, cast
 
-import click
+import asyncclick as click
 import tomlkit
-from click import Context, UsageError
+from asyncclick import Context, UsageError
 from tomlkit.exceptions import NonExistentKey
-from tortoise import Tortoise
 
 from aerich import Command
 from aerich.enums import Color
@@ -21,21 +18,6 @@ CONFIG_DEFAULT_VALUES = {
 }
 
 
-def coro(f):
-    @wraps(f)
-    def wrapper(*args, **kwargs):
-        loop = asyncio.get_event_loop()
-
-        # Close db connections at the end of all all but the cli group function
-        try:
-            loop.run_until_complete(f(*args, **kwargs))
-        finally:
-            if f.__name__ != "cli":
-                loop.run_until_complete(Tortoise.close_connections())
-
-    return wrapper
-
-
 @click.group(context_settings={"help_option_names": ["-h", "--help"]})
 @click.version_option(__version__, "-V", "--version")
 @click.option(
@@ -47,59 +29,70 @@ def coro(f):
 )
 @click.option("--app", required=False, help="Tortoise-ORM app name.")
 @click.pass_context
-@coro
+async def cli(ctx: Context, config, app) -> None:
-async def cli(ctx: Context, config, app):
     ctx.ensure_object(dict)
     ctx.obj["config_file"] = config
 
     invoked_subcommand = ctx.invoked_subcommand
     if invoked_subcommand != "init":
-        if not Path(config).exists():
+        config_path = Path(config)
-            raise UsageError("You must exec init first", ctx=ctx)
+        if not config_path.exists():
-        with open(config, "r") as f:
+            raise UsageError(
-            content = f.read()
+                "You need to run `aerich init` first to create the config file.", ctx=ctx
-        doc = tomlkit.parse(content)
+            )
+        content = config_path.read_text()
+        doc: dict = tomlkit.parse(content)
         try:
-            tool = doc["tool"]["aerich"]
+            tool = cast(Dict[str, str], doc["tool"]["aerich"])
             location = tool["location"]
             tortoise_orm = tool["tortoise_orm"]
             src_folder = tool.get("src_folder", CONFIG_DEFAULT_VALUES["src_folder"])
         except NonExistentKey:
-            raise UsageError("You need run aerich init again when upgrade to 0.6.0+")
+            raise UsageError("You need run `aerich init` again when upgrading to aerich 0.6.0+.")
         add_src_path(src_folder)
         tortoise_config = get_tortoise_config(ctx, tortoise_orm)
-        app = app or list(tortoise_config.get("apps").keys())[0]
+        if not app:
+            apps_config = cast(dict, tortoise_config.get("apps"))
+            app = list(apps_config.keys())[0]
         command = Command(tortoise_config=tortoise_config, app=app, location=location)
         ctx.obj["command"] = command
         if invoked_subcommand != "init-db":
             if not Path(location, app).exists():
-                raise UsageError("You must exec init-db first", ctx=ctx)
+                raise UsageError(
+                    "You need to run `aerich init-db` first to initialize the database.", ctx=ctx
+                )
             await command.init()
 
 
-@cli.command(help="Generate migrate changes file.")
+@cli.command(help="Generate a migration file for the current state of the models.")
-@click.option("--name", default="update", show_default=True, help="Migrate name.")
+@click.option("--name", default="update", show_default=True, help="Migration name.")
+@click.option("--empty", default=False, is_flag=True, help="Generate an empty migration file.")
 @click.pass_context
-@coro
+async def migrate(ctx: Context, name, empty) -> None:
-async def migrate(ctx: Context, name):
     command = ctx.obj["command"]
-    ret = await command.migrate(name)
+    ret = await command.migrate(name, empty)
     if not ret:
         return click.secho("No changes detected", fg=Color.yellow)
-    click.secho(f"Success migrate {ret}", fg=Color.green)
+    click.secho(f"Success creating migration file {ret}", fg=Color.green)
 
 
-@cli.command(help="Upgrade to specified version.")
+@cli.command(help="Upgrade to specified migration version.")
+@click.option(
+    "--in-transaction",
+    "-i",
+    default=True,
+    type=bool,
+    help="Make migrations in a single transaction or not. Can be helpful for large migrations or creating concurrent indexes.",
+)
 @click.pass_context
-@coro
+async def upgrade(ctx: Context, in_transaction: bool) -> None:
-async def upgrade(ctx: Context):
     command = ctx.obj["command"]
-    migrated = await command.upgrade()
+    migrated = await command.upgrade(run_in_transaction=in_transaction)
     if not migrated:
         click.secho("No upgrade items found", fg=Color.yellow)
     else:
         for version_file in migrated:
-            click.secho(f"Success upgrade {version_file}", fg=Color.green)
+            click.secho(f"Success upgrading to {version_file}", fg=Color.green)
 
 
 @cli.command(help="Downgrade to specified version.")
@@ -108,8 +101,8 @@ async def upgrade(ctx: Context):
     "--version",
     default=-1,
     type=int,
-    show_default=True,
+    show_default=False,
-    help="Specified version, default to last.",
+    help="Specified version, default to last migration.",
 )
 @click.option(
     "-d",
@@ -117,59 +110,56 @@ async def upgrade(ctx: Context):
     is_flag=True,
     default=False,
     show_default=True,
-    help="Delete version files at the same time.",
+    help="Also delete the migration files.",
 )
 @click.pass_context
 @click.confirmation_option(
-    prompt="Downgrade is dangerous, which maybe lose your data, are you sure?",
+    prompt="Downgrade is dangerous: you might lose your data! Are you sure?",
 )
-@coro
+async def downgrade(ctx: Context, version: int, delete: bool) -> None:
-async def downgrade(ctx: Context, version: int, delete: bool):
     command = ctx.obj["command"]
     try:
         files = await command.downgrade(version, delete)
     except DowngradeError as e:
         return click.secho(str(e), fg=Color.yellow)
     for file in files:
-        click.secho(f"Success downgrade {file}", fg=Color.green)
+        click.secho(f"Success downgrading to {file}", fg=Color.green)
 
 
-@cli.command(help="Show current available heads in migrate location.")
+@cli.command(help="Show currently available heads (unapplied migrations).")
 @click.pass_context
-@coro
+async def heads(ctx: Context) -> None:
-async def heads(ctx: Context):
     command = ctx.obj["command"]
     head_list = await command.heads()
     if not head_list:
-        return click.secho("No available heads, try migrate first", fg=Color.green)
+        return click.secho("No available heads.", fg=Color.green)
     for version in head_list:
         click.secho(version, fg=Color.green)
 
 
-@cli.command(help="List all migrate items.")
+@cli.command(help="List all migrations.")
 @click.pass_context
-@coro
+async def history(ctx: Context) -> None:
-async def history(ctx: Context):
     command = ctx.obj["command"]
     versions = await command.history()
     if not versions:
-        return click.secho("No history, try migrate", fg=Color.green)
+        return click.secho("No migrations created yet.", fg=Color.green)
     for version in versions:
         click.secho(version, fg=Color.green)
 
 
-@cli.command(help="Init config file and generate root migrate location.")
+@cli.command(help="Initialize aerich config and create migrations folder.")
 @click.option(
     "-t",
     "--tortoise-orm",
     required=True,
-    help="Tortoise-ORM config module dict variable, like settings.TORTOISE_ORM.",
+    help="Tortoise-ORM config dict location, like `settings.TORTOISE_ORM`.",
 )
 @click.option(
     "--location",
     default="./migrations",
     show_default=True,
-    help="Migrate store location.",
+    help="Migrations folder.",
 )
 @click.option(
     "-s",
@@ -179,8 +169,7 @@ async def history(ctx: Context):
     help="Folder of the source, relative to the project root.",
 )
 @click.pass_context
-@coro
+async def init(ctx: Context, tortoise_orm, location, src_folder) -> None:
-async def init(ctx: Context, tortoise_orm, location, src_folder):
     config_file = ctx.obj["config_file"]
 
     if os.path.isabs(src_folder):
@@ -192,52 +181,52 @@ async def init(ctx: Context, tortoise_orm, location, src_folder):
     # check that we can find the configuration, if not we can fail before the config file gets created
     add_src_path(src_folder)
     get_tortoise_config(ctx, tortoise_orm)
-    if Path(config_file).exists():
+    config_path = Path(config_file)
-        with open(config_file, "r") as f:
+    if config_path.exists():
-            content = f.read()
+        content = config_path.read_text()
-        doc = tomlkit.parse(content)
     else:
-        doc = tomlkit.parse("[tool.aerich]")
+        content = "[tool.aerich]"
+    doc: dict = tomlkit.parse(content)
     table = tomlkit.table()
     table["tortoise_orm"] = tortoise_orm
     table["location"] = location
     table["src_folder"] = src_folder
     doc["tool"]["aerich"] = table
 
-    with open(config_file, "w") as f:
+    config_path.write_text(tomlkit.dumps(doc))
-        f.write(tomlkit.dumps(doc))
 
     Path(location).mkdir(parents=True, exist_ok=True)
 
-    click.secho(f"Success create migrate location {location}", fg=Color.green)
+    click.secho(f"Success creating migrations folder {location}", fg=Color.green)
-    click.secho(f"Success write config to {config_file}", fg=Color.green)
+    click.secho(f"Success writing aerich config to {config_file}", fg=Color.green)
 
 
-@cli.command(help="Generate schema and generate app migrate location.")
+@cli.command(help="Generate schema and generate app migration folder.")
 @click.option(
+    "-s",
     "--safe",
     type=bool,
+    is_flag=True,
     default=True,
-    help="When set to true, creates the table only when it does not already exist.",
+    help="Create tables only when they do not already exist.",
     show_default=True,
 )
 @click.pass_context
-@coro
+async def init_db(ctx: Context, safe: bool) -> None:
-async def init_db(ctx: Context, safe):
     command = ctx.obj["command"]
     app = command.app
     dirname = Path(command.location, app)
     try:
         await command.init_db(safe)
-        click.secho(f"Success create app migrate location {dirname}", fg=Color.green)
+        click.secho(f"Success creating app migration folder {dirname}", fg=Color.green)
-        click.secho(f'Success generate schema for app "{app}"', fg=Color.green)
+        click.secho(f'Success generating initial migration file for app "{app}"', fg=Color.green)
     except FileExistsError:
         return click.secho(
-            f"Inited {app} already, or delete {dirname} and try again.", fg=Color.yellow
+            f"App {app} is already initialized. Delete {dirname} and try again.", fg=Color.yellow
         )
 
 
-@cli.command(help="Introspects the database tables to standard output as TortoiseORM model.")
+@cli.command(help="Prints the current database tables to stdout as Tortoise-ORM models.")
 @click.option(
     "-t",
     "--table",
@@ -246,13 +235,13 @@ async def init_db(ctx: Context, safe):
     required=False,
 )
 @click.pass_context
-@coro
+async def inspectdb(ctx: Context, table: List[str]) -> None:
-async def inspectdb(ctx: Context, table: List[str]):
     command = ctx.obj["command"]
-    await command.inspectdb(table)
+    ret = await command.inspectdb(table)
+    click.secho(ret)
 
 
-def main():
+def main() -> None:
     cli()
 
 

aerich/coder.py

@@ -1,31 +1,32 @@
 import base64
 import json
-import pickle  # nosec: B301
+import pickle  # nosec: B301,B403
+from typing import Any, Union
 
 from tortoise.indexes import Index
 
 
 class JsonEncoder(json.JSONEncoder):
-    def default(self, obj):
+    def default(self, obj) -> Any:
         if isinstance(obj, Index):
             return {
                 "type": "index",
-                "val": base64.b64encode(pickle.dumps(obj)).decode(),
+                "val": base64.b64encode(pickle.dumps(obj)).decode(),  # nosec: B301
-            }  # nosec: B301
+            }
         else:
             return super().default(obj)
 
 
-def object_hook(obj):
+def object_hook(obj) -> Any:
     _type = obj.get("type")
     if not _type:
         return obj
     return pickle.loads(base64.b64decode(obj["val"]))  # nosec: B301
 
 
-def encoder(obj: dict):
+def encoder(obj: dict) -> str:
     return json.dumps(obj, cls=JsonEncoder)
 
 
-def decoder(obj: str):
+def decoder(obj: Union[str, bytes]) -> Any:
     return json.loads(obj, object_hook=object_hook)

aerich/ddl/__init__.py

@@ -1,5 +1,5 @@
 from enum import Enum
-from typing import List, Type
+from typing import Any, List, Type, cast
 
 from tortoise import BaseDBAsyncClient, Model
 from tortoise.backends.base.schema_generator import BaseSchemaGenerator
@@ -23,30 +23,38 @@ class BaseDDL:
     _DROP_INDEX_TEMPLATE = 'ALTER TABLE "{table_name}" DROP INDEX "{index_name}"'
     _ADD_FK_TEMPLATE = 'ALTER TABLE "{table_name}" ADD CONSTRAINT "{fk_name}" FOREIGN KEY ("{db_column}") REFERENCES "{table}" ("{field}") ON DELETE {on_delete}'
     _DROP_FK_TEMPLATE = 'ALTER TABLE "{table_name}" DROP FOREIGN KEY "{fk_name}"'
-    _M2M_TABLE_TEMPLATE = 'CREATE TABLE "{table_name}" ("{backward_key}" {backward_type} NOT NULL REFERENCES "{backward_table}" ("{backward_field}") ON DELETE CASCADE,"{forward_key}" {forward_type} NOT NULL REFERENCES "{forward_table}" ("{forward_field}") ON DELETE {on_delete}){extra}{comment}'
+    _M2M_TABLE_TEMPLATE = (
+        'CREATE TABLE "{table_name}" (\n'
+        '    "{backward_key}" {backward_type} NOT NULL REFERENCES "{backward_table}" ("{backward_field}") ON DELETE CASCADE,\n'
+        '    "{forward_key}" {forward_type} NOT NULL REFERENCES "{forward_table}" ("{forward_field}") ON DELETE {on_delete}\n'
+        "){extra}{comment}"
+    )
     _MODIFY_COLUMN_TEMPLATE = 'ALTER TABLE "{table_name}" MODIFY COLUMN {column}'
     _CHANGE_COLUMN_TEMPLATE = (
         'ALTER TABLE "{table_name}" CHANGE {old_column_name} {new_column_name} {new_column_type}'
     )
     _RENAME_TABLE_TEMPLATE = 'ALTER TABLE "{old_table_name}" RENAME TO "{new_table_name}"'
 
-    def __init__(self, client: "BaseDBAsyncClient"):
+    def __init__(self, client: "BaseDBAsyncClient") -> None:
         self.client = client
         self.schema_generator = self.schema_generator_cls(client)
 
-    def create_table(self, model: "Type[Model]"):
+    def create_table(self, model: "Type[Model]") -> str:
-        return self.schema_generator._get_table_sql(model, True)["table_creation_string"]
+        return self.schema_generator._get_table_sql(model, True)["table_creation_string"].rstrip(
+            ";"
+        )
 
-    def drop_table(self, table_name: str):
+    def drop_table(self, table_name: str) -> str:
         return self._DROP_TABLE_TEMPLATE.format(table_name=table_name)
 
     def create_m2m(
         self, model: "Type[Model]", field_describe: dict, reference_table_describe: dict
-    ):
+    ) -> str:
-        through = field_describe.get("through")
+        through = cast(str, field_describe.get("through"))
         description = field_describe.get("description")
-        reference_id = reference_table_describe.get("pk_field").get("db_column")
+        pk_field = cast(dict, reference_table_describe.get("pk_field"))
-        db_field_types = reference_table_describe.get("pk_field").get("db_field_types")
+        reference_id = pk_field.get("db_column")
+        db_field_types = cast(dict, pk_field.get("db_field_types"))
         return self._M2M_TABLE_TEMPLATE.format(
             table_name=through,
             backward_table=model._meta.db_table,
@@ -59,34 +67,30 @@ class BaseDDL:
             forward_type=db_field_types.get(self.DIALECT) or db_field_types.get(""),
             on_delete=field_describe.get("on_delete"),
             extra=self.schema_generator._table_generate_extra(table=through),
-            comment=self.schema_generator._table_comment_generator(
+            comment=(
-                table=through, comment=description
+                self.schema_generator._table_comment_generator(table=through, comment=description)
-            )
                 if description
-            else "",
+                else ""
+            ),
         )
 
-    def drop_m2m(self, table_name: str):
+    def drop_m2m(self, table_name: str) -> str:
         return self._DROP_TABLE_TEMPLATE.format(table_name=table_name)
 
-    def _get_default(self, model: "Type[Model]", field_describe: dict):
+    def _get_default(self, model: "Type[Model]", field_describe: dict) -> Any:
         db_table = model._meta.db_table
         default = field_describe.get("default")
         if isinstance(default, Enum):
             default = default.value
-        db_column = field_describe.get("db_column")
+        db_column = cast(str, field_describe.get("db_column"))
         auto_now_add = field_describe.get("auto_now_add", False)
         auto_now = field_describe.get("auto_now", False)
         if default is not None or auto_now_add:
-            if (
+            if field_describe.get("field_type") in [
-                field_describe.get("field_type")
-                in [
                 "UUIDField",
                 "TextField",
                 "JSONField",
-                ]
+            ] or is_default_function(default):
-                or is_default_function(default)
-            ):
                 default = ""
             else:
                 try:
@@ -103,64 +107,55 @@ class BaseDDL:
             default = None
         return default
 
-    def add_column(self, model: "Type[Model]", field_describe: dict, is_pk: bool = False):
+    def add_column(self, model: "Type[Model]", field_describe: dict, is_pk: bool = False) -> str:
+        return self._add_or_modify_column(model, field_describe, is_pk)
+
+    def _add_or_modify_column(self, model, field_describe: dict, is_pk: bool, modify=False) -> str:
         db_table = model._meta.db_table
         description = field_describe.get("description")
-        db_column = field_describe.get("db_column")
+        db_column = cast(str, field_describe.get("db_column"))
-        db_field_types = field_describe.get("db_field_types")
+        db_field_types = cast(dict, field_describe.get("db_field_types"))
         default = self._get_default(model, field_describe)
         if default is None:
             default = ""
-        return self._ADD_COLUMN_TEMPLATE.format(
+        if modify:
+            unique = ""
+            template = self._MODIFY_COLUMN_TEMPLATE
+        else:
+            unique = "UNIQUE" if field_describe.get("unique") else ""
+            template = self._ADD_COLUMN_TEMPLATE
+        return template.format(
             table_name=db_table,
             column=self.schema_generator._create_string(
                 db_column=db_column,
                 field_type=db_field_types.get(self.DIALECT, db_field_types.get("")),
                 nullable="NOT NULL" if not field_describe.get("nullable") else "",
-                unique="UNIQUE" if field_describe.get("unique") else "",
+                unique=unique,
-                comment=self.schema_generator._column_comment_generator(
+                comment=(
+                    self.schema_generator._column_comment_generator(
                         table=db_table,
                         column=db_column,
-                    comment=field_describe.get("description"),
+                        comment=description,
                     )
                     if description
-                else "",
+                    else ""
+                ),
                 is_primary_key=is_pk,
                 default=default,
             ),
         )
 
-    def drop_column(self, model: "Type[Model]", column_name: str):
+    def drop_column(self, model: "Type[Model]", column_name: str) -> str:
         return self._DROP_COLUMN_TEMPLATE.format(
             table_name=model._meta.db_table, column_name=column_name
         )
 
-    def modify_column(self, model: "Type[Model]", field_describe: dict, is_pk: bool = False):
+    def modify_column(self, model: "Type[Model]", field_describe: dict, is_pk: bool = False) -> str:
-        db_table = model._meta.db_table
+        return self._add_or_modify_column(model, field_describe, is_pk, modify=True)
-        db_field_types = field_describe.get("db_field_types")
-        default = self._get_default(model, field_describe)
-        if default is None:
-            default = ""
-        return self._MODIFY_COLUMN_TEMPLATE.format(
-            table_name=db_table,
-            column=self.schema_generator._create_string(
-                db_column=field_describe.get("db_column"),
-                field_type=db_field_types.get(self.DIALECT) or db_field_types.get(""),
-                nullable="NOT NULL" if not field_describe.get("nullable") else "",
-                unique="",
-                comment=self.schema_generator._column_comment_generator(
-                    table=db_table,
-                    column=field_describe.get("db_column"),
-                    comment=field_describe.get("description"),
-                )
-                if field_describe.get("description")
-                else "",
-                is_primary_key=is_pk,
-                default=default,
-            ),
-        )
 
-    def rename_column(self, model: "Type[Model]", old_column_name: str, new_column_name: str):
+    def rename_column(
+        self, model: "Type[Model]", old_column_name: str, new_column_name: str
+    ) -> str:
         return self._RENAME_COLUMN_TEMPLATE.format(
             table_name=model._meta.db_table,
             old_column_name=old_column_name,
@@ -169,7 +164,7 @@ class BaseDDL:
 
     def change_column(
         self, model: "Type[Model]", old_column_name: str, new_column_name: str, new_column_type: str
-    ):
+    ) -> str:
         return self._CHANGE_COLUMN_TEMPLATE.format(
             table_name=model._meta.db_table,
             old_column_name=old_column_name,
@@ -177,17 +172,17 @@ class BaseDDL:
             new_column_type=new_column_type,
         )
 
-    def add_index(self, model: "Type[Model]", field_names: List[str], unique=False):
+    def add_index(self, model: "Type[Model]", field_names: List[str], unique=False) -> str:
         return self._ADD_INDEX_TEMPLATE.format(
             unique="UNIQUE " if unique else "",
             index_name=self.schema_generator._generate_index_name(
                 "idx" if not unique else "uid", model, field_names
             ),
             table_name=model._meta.db_table,
-            column_names=", ".join([self.schema_generator.quote(f) for f in field_names]),
+            column_names=", ".join(self.schema_generator.quote(f) for f in field_names),
         )
 
-    def drop_index(self, model: "Type[Model]", field_names: List[str], unique=False):
+    def drop_index(self, model: "Type[Model]", field_names: List[str], unique=False) -> str:
         return self._DROP_INDEX_TEMPLATE.format(
             index_name=self.schema_generator._generate_index_name(
                 "idx" if not unique else "uid", model, field_names
@@ -195,45 +190,52 @@ class BaseDDL:
             table_name=model._meta.db_table,
         )
 
-    def drop_index_by_name(self, model: "Type[Model]", index_name: str):
+    def drop_index_by_name(self, model: "Type[Model]", index_name: str) -> str:
         return self._DROP_INDEX_TEMPLATE.format(
             index_name=index_name,
             table_name=model._meta.db_table,
         )
 
-    def add_fk(self, model: "Type[Model]", field_describe: dict, reference_table_describe: dict):
+    def _generate_fk_name(
+        self, db_table, field_describe: dict, reference_table_describe: dict
+    ) -> str:
+        """Generate fk name"""
+        db_column = cast(str, field_describe.get("raw_field"))
+        pk_field = cast(dict, reference_table_describe.get("pk_field"))
+        to_field = cast(str, pk_field.get("db_column"))
+        to_table = cast(str, reference_table_describe.get("table"))
+        return self.schema_generator._generate_fk_name(
+            from_table=db_table,
+            from_field=db_column,
+            to_table=to_table,
+            to_field=to_field,
+        )
+
+    def add_fk(
+        self, model: "Type[Model]", field_describe: dict, reference_table_describe: dict
+    ) -> str:
         db_table = model._meta.db_table
 
         db_column = field_describe.get("raw_field")
-        reference_id = reference_table_describe.get("pk_field").get("db_column")
+        pk_field = cast(dict, reference_table_describe.get("pk_field"))
-        fk_name = self.schema_generator._generate_fk_name(
+        reference_id = pk_field.get("db_column")
-            from_table=db_table,
-            from_field=db_column,
-            to_table=reference_table_describe.get("table"),
-            to_field=reference_table_describe.get("pk_field").get("db_column"),
-        )
         return self._ADD_FK_TEMPLATE.format(
             table_name=db_table,
-            fk_name=fk_name,
+            fk_name=self._generate_fk_name(db_table, field_describe, reference_table_describe),
             db_column=db_column,
             table=reference_table_describe.get("table"),
             field=reference_id,
             on_delete=field_describe.get("on_delete"),
         )
 
-    def drop_fk(self, model: "Type[Model]", field_describe: dict, reference_table_describe: dict):
+    def drop_fk(
+        self, model: "Type[Model]", field_describe: dict, reference_table_describe: dict
+    ) -> str:
         db_table = model._meta.db_table
-        return self._DROP_FK_TEMPLATE.format(
+        fk_name = self._generate_fk_name(db_table, field_describe, reference_table_describe)
-            table_name=db_table,
+        return self._DROP_FK_TEMPLATE.format(table_name=db_table, fk_name=fk_name)
-            fk_name=self.schema_generator._generate_fk_name(
-                from_table=db_table,
-                from_field=field_describe.get("raw_field"),
-                to_table=reference_table_describe.get("table"),
-                to_field=reference_table_describe.get("pk_field").get("db_column"),
-            ),
-        )
 
-    def alter_column_default(self, model: "Type[Model]", field_describe: dict):
+    def alter_column_default(self, model: "Type[Model]", field_describe: dict) -> str:
         db_table = model._meta.db_table
         default = self._get_default(model, field_describe)
         return self._ALTER_DEFAULT_TEMPLATE.format(
@@ -242,13 +244,13 @@ class BaseDDL:
             default="SET" + default if default is not None else "DROP DEFAULT",
         )
 
-    def alter_column_null(self, model: "Type[Model]", field_describe: dict):
+    def alter_column_null(self, model: "Type[Model]", field_describe: dict) -> str:
         return self.modify_column(model, field_describe)
 
-    def set_comment(self, model: "Type[Model]", field_describe: dict):
+    def set_comment(self, model: "Type[Model]", field_describe: dict) -> str:
         return self.modify_column(model, field_describe)
 
-    def rename_table(self, model: "Type[Model]", old_table_name: str, new_table_name: str):
+    def rename_table(self, model: "Type[Model]", old_table_name: str, new_table_name: str) -> str:
         db_table = model._meta.db_table
         return self._RENAME_TABLE_TEMPLATE.format(
             table_name=db_table, old_table_name=old_table_name, new_table_name=new_table_name

aerich/ddl/mysql/__init__.py

@@ -1,7 +1,12 @@
+from typing import TYPE_CHECKING, List, Type
+
 from tortoise.backends.mysql.schema_generator import MySQLSchemaGenerator
 
 from aerich.ddl import BaseDDL
 
+if TYPE_CHECKING:
+    from tortoise import Model  # noqa:F401
+
 
 class MysqlDDL(BaseDDL):
     schema_generator_cls = MySQLSchemaGenerator
@@ -22,6 +27,37 @@ class MysqlDDL(BaseDDL):
     _DROP_INDEX_TEMPLATE = "ALTER TABLE `{table_name}` DROP INDEX `{index_name}`"
     _ADD_FK_TEMPLATE = "ALTER TABLE `{table_name}` ADD CONSTRAINT `{fk_name}` FOREIGN KEY (`{db_column}`) REFERENCES `{table}` (`{field}`) ON DELETE {on_delete}"
     _DROP_FK_TEMPLATE = "ALTER TABLE `{table_name}` DROP FOREIGN KEY `{fk_name}`"
-    _M2M_TABLE_TEMPLATE = "CREATE TABLE `{table_name}` (`{backward_key}` {backward_type} NOT NULL REFERENCES `{backward_table}` (`{backward_field}`) ON DELETE CASCADE,`{forward_key}` {forward_type} NOT NULL REFERENCES `{forward_table}` (`{forward_field}`) ON DELETE CASCADE){extra}{comment}"
+    _M2M_TABLE_TEMPLATE = (
+        "CREATE TABLE `{table_name}` (\n"
+        "    `{backward_key}` {backward_type} NOT NULL REFERENCES `{backward_table}` (`{backward_field}`) ON DELETE CASCADE,\n"
+        "    `{forward_key}` {forward_type} NOT NULL REFERENCES `{forward_table}` (`{forward_field}`) ON DELETE CASCADE\n"
+        "){extra}{comment}"
+    )
     _MODIFY_COLUMN_TEMPLATE = "ALTER TABLE `{table_name}` MODIFY COLUMN {column}"
     _RENAME_TABLE_TEMPLATE = "ALTER TABLE `{old_table_name}` RENAME TO `{new_table_name}`"
+
+    def _index_name(self, unique: bool, model: "Type[Model]", field_names: List[str]) -> str:
+        if unique:
+            if len(field_names) == 1:
+                # Example: `email = CharField(max_length=50, unique=True)`
+                # Generate schema: `"email" VARCHAR(10) NOT NULL UNIQUE`
+                # Unique index key is the same as field name: `email`
+                return field_names[0]
+            index_prefix = "uid"
+        else:
+            index_prefix = "idx"
+        return self.schema_generator._generate_index_name(index_prefix, model, field_names)
+
+    def add_index(self, model: "Type[Model]", field_names: List[str], unique=False) -> str:
+        return self._ADD_INDEX_TEMPLATE.format(
+            unique="UNIQUE " if unique else "",
+            index_name=self._index_name(unique, model, field_names),
+            table_name=model._meta.db_table,
+            column_names=", ".join(self.schema_generator.quote(f) for f in field_names),
+        )
+
+    def drop_index(self, model: "Type[Model]", field_names: List[str], unique=False) -> str:
+        return self._DROP_INDEX_TEMPLATE.format(
+            index_name=self._index_name(unique, model, field_names),
+            table_name=model._meta.db_table,
+        )

aerich/ddl/postgres/__init__.py

@@ -1,4 +1,4 @@
-from typing import Type
+from typing import Type, cast
 
 from tortoise import Model
 from tortoise.backends.asyncpg.schema_generator import AsyncpgSchemaGenerator
@@ -18,7 +18,7 @@ class PostgresDDL(BaseDDL):
     _SET_COMMENT_TEMPLATE = 'COMMENT ON COLUMN "{table_name}"."{column}" IS {comment}'
     _DROP_FK_TEMPLATE = 'ALTER TABLE "{table_name}" DROP CONSTRAINT "{fk_name}"'
 
-    def alter_column_null(self, model: "Type[Model]", field_describe: dict):
+    def alter_column_null(self, model: "Type[Model]", field_describe: dict) -> str:
         db_table = model._meta.db_table
         return self._ALTER_NULL_TEMPLATE.format(
             table_name=db_table,
@@ -26,9 +26,9 @@ class PostgresDDL(BaseDDL):
             set_drop="DROP" if field_describe.get("nullable") else "SET",
         )
 
-    def modify_column(self, model: "Type[Model]", field_describe: dict, is_pk: bool = False):
+    def modify_column(self, model: "Type[Model]", field_describe: dict, is_pk: bool = False) -> str:
         db_table = model._meta.db_table
-        db_field_types = field_describe.get("db_field_types")
+        db_field_types = cast(dict, field_describe.get("db_field_types"))
         db_column = field_describe.get("db_column")
         datatype = db_field_types.get(self.DIALECT) or db_field_types.get("")
         return self._MODIFY_COLUMN_TEMPLATE.format(
@@ -38,12 +38,14 @@ class PostgresDDL(BaseDDL):
             using=f' USING "{db_column}"::{datatype}',
         )
 
-    def set_comment(self, model: "Type[Model]", field_describe: dict):
+    def set_comment(self, model: "Type[Model]", field_describe: dict) -> str:
         db_table = model._meta.db_table
         return self._SET_COMMENT_TEMPLATE.format(
             table_name=db_table,
             column=field_describe.get("db_column") or field_describe.get("raw_field"),
-            comment="'{}'".format(field_describe.get("description"))
+            comment=(
+                "'{}'".format(field_describe.get("description"))
                 if field_describe.get("description")
-            else "NULL",
+                else "NULL"
+            ),
         )

aerich/inspectdb.py

@@ -1,87 +0,0 @@
-import sys
-from typing import List, Optional
-
-from ddlparse import DdlParse
-from tortoise import BaseDBAsyncClient
-
-
-class InspectDb:
-    _table_template = "class {table}(Model):\n"
-    _field_template_mapping = {
-        "INT": "    {field} = fields.IntField({pk}{unique}{comment})",
-        "SMALLINT": "    {field} = fields.IntField({pk}{unique}{comment})",
-        "TINYINT": "    {field} = fields.BooleanField({null}{default}{comment})",
-        "VARCHAR": "    {field} = fields.CharField({pk}{unique}{length}{null}{default}{comment})",
-        "LONGTEXT": "    {field} = fields.TextField({null}{default}{comment})",
-        "TEXT": "    {field} = fields.TextField({null}{default}{comment})",
-        "DATETIME": "    {field} = fields.DatetimeField({null}{default}{comment})",
-        "FLOAT": "    {field} = fields.FloatField({null}{default}{comment})",
-        "DATE": "    {field} = fields.DateField({null}{default}{comment})",
-    }
-
-    def __init__(self, conn: BaseDBAsyncClient, tables: Optional[List[str]] = None):
-        self.conn = conn
-        self.tables = tables
-        self.DIALECT = conn.schema_generator.DIALECT
-
-    async def show_create_tables(self):
-        if self.DIALECT == "mysql":
-            if not self.tables:
-                sql_tables = f"SELECT table_name FROM information_schema.tables WHERE table_schema = '{self.conn.database}';"  # nosec: B608
-                ret = await self.conn.execute_query(sql_tables)
-                self.tables = map(lambda x: x["TABLE_NAME"], ret[1])
-            for table in self.tables:
-                sql_show_create_table = f"SHOW CREATE TABLE {table}"
-                ret = await self.conn.execute_query(sql_show_create_table)
-                yield ret[1][0]["Create Table"]
-        else:
-            raise NotImplementedError("Currently only support MySQL")
-
-    async def inspect(self):
-        ddl_list = self.show_create_tables()
-        result = "from tortoise import Model, fields\n\n\n"
-        tables = []
-        async for ddl in ddl_list:
-            parser = DdlParse(ddl, DdlParse.DATABASE.mysql)
-            table = parser.parse()
-            name = table.name.title()
-            columns = table.columns
-            fields = []
-            model = self._table_template.format(table=name)
-            for column_name, column in columns.items():
-                comment = default = length = unique = null = pk = ""
-                if column.primary_key:
-                    pk = "pk=True, "
-                if column.unique:
-                    unique = "unique=True, "
-                if column.data_type == "VARCHAR":
-                    length = f"max_length={column.length}, "
-                if not column.not_null:
-                    null = "null=True, "
-                if column.default is not None:
-                    if column.data_type == "TINYINT":
-                        default = f"default={'True' if column.default == '1' else 'False'}, "
-                    elif column.data_type == "DATETIME":
-                        if "CURRENT_TIMESTAMP" in column.default:
-                            if "ON UPDATE CURRENT_TIMESTAMP" in ddl:
-                                default = "auto_now_add=True, "
-                            else:
-                                default = "auto_now=True, "
-                    else:
-                        default = f"default={column.default}, "
-
-                if column.comment:
-                    comment = f"description='{column.comment}', "
-
-                field = self._field_template_mapping[column.data_type].format(
-                    field=column_name,
-                    pk=pk,
-                    unique=unique,
-                    length=length,
-                    null=null,
-                    default=default,
-                    comment=comment,
-                )
-                fields.append(field)
-            tables.append(model + "\n".join(fields))
-        sys.stdout.write(result + "\n\n\n".join(tables))

aerich/inspectdb/__init__.py

@@ -0,0 +1,186 @@
+from __future__ import annotations
+
+from typing import Any, Callable, Dict, Optional, TypedDict
+
+from pydantic import BaseModel
+from tortoise import BaseDBAsyncClient
+
+
+class ColumnInfoDict(TypedDict):
+    name: str
+    pk: str
+    index: str
+    null: str
+    default: str
+    length: str
+    comment: str
+
+
+FieldMapDict = Dict[str, Callable[..., str]]
+
+
+class Column(BaseModel):
+    name: str
+    data_type: str
+    null: bool
+    default: Any
+    comment: Optional[str] = None
+    pk: bool
+    unique: bool
+    index: bool
+    length: Optional[int] = None
+    extra: Optional[str] = None
+    decimal_places: Optional[int] = None
+    max_digits: Optional[int] = None
+
+    def translate(self) -> ColumnInfoDict:
+        comment = default = length = index = null = pk = ""
+        if self.pk:
+            pk = "pk=True, "
+        else:
+            if self.unique:
+                index = "unique=True, "
+            else:
+                if self.index:
+                    index = "index=True, "
+        if self.data_type in ("varchar", "VARCHAR"):
+            length = f"max_length={self.length}, "
+        elif self.data_type in ("decimal", "numeric"):
+            length_parts = []
+            if self.max_digits:
+                length_parts.append(f"max_digits={self.max_digits}")
+            if self.decimal_places:
+                length_parts.append(f"decimal_places={self.decimal_places}")
+            if length_parts:
+                length = ", ".join(length_parts) + ", "
+        if self.null:
+            null = "null=True, "
+        if self.default is not None:
+            if self.data_type in ("tinyint", "INT"):
+                default = f"default={'True' if self.default == '1' else 'False'}, "
+            elif self.data_type == "bool":
+                default = f"default={'True' if self.default == 'true' else 'False'}, "
+            elif self.data_type in ("datetime", "timestamptz", "TIMESTAMP"):
+                if "CURRENT_TIMESTAMP" == self.default:
+                    if "DEFAULT_GENERATED on update CURRENT_TIMESTAMP" == self.extra:
+                        default = "auto_now=True, "
+                    else:
+                        default = "auto_now_add=True, "
+            else:
+                if "::" in self.default:
+                    default = f"default={self.default.split('::')[0]}, "
+                elif self.default.endswith("()"):
+                    default = ""
+                elif self.default == "":
+                    default = 'default=""'
+                else:
+                    default = f"default={self.default}, "
+
+        if self.comment:
+            comment = f"description='{self.comment}', "
+        return {
+            "name": self.name,
+            "pk": pk,
+            "index": index,
+            "null": null,
+            "default": default,
+            "length": length,
+            "comment": comment,
+        }
+
+
+class Inspect:
+    _table_template = "class {table}(Model):\n"
+
+    def __init__(self, conn: BaseDBAsyncClient, tables: list[str] | None = None) -> None:
+        self.conn = conn
+        try:
+            self.database = conn.database  # type:ignore[attr-defined]
+        except AttributeError:
+            pass
+        self.tables = tables
+
+    @property
+    def field_map(self) -> FieldMapDict:
+        raise NotImplementedError
+
+    async def inspect(self) -> str:
+        if not self.tables:
+            self.tables = await self.get_all_tables()
+        result = "from tortoise import Model, fields\n\n\n"
+        tables = []
+        for table in self.tables:
+            columns = await self.get_columns(table)
+            fields = []
+            model = self._table_template.format(table=table.title().replace("_", ""))
+            for column in columns:
+                field = self.field_map[column.data_type](**column.translate())
+                fields.append("    " + field)
+            tables.append(model + "\n".join(fields))
+        return result + "\n\n\n".join(tables)
+
+    async def get_columns(self, table: str) -> list[Column]:
+        raise NotImplementedError
+
+    async def get_all_tables(self) -> list[str]:
+        raise NotImplementedError
+
+    @classmethod
+    def decimal_field(cls, **kwargs) -> str:
+        return "{name} = fields.DecimalField({pk}{index}{length}{null}{default}{comment})".format(
+            **kwargs
+        )
+
+    @classmethod
+    def time_field(cls, **kwargs) -> str:
+        return "{name} = fields.TimeField({null}{default}{comment})".format(**kwargs)
+
+    @classmethod
+    def date_field(cls, **kwargs) -> str:
+        return "{name} = fields.DateField({null}{default}{comment})".format(**kwargs)
+
+    @classmethod
+    def float_field(cls, **kwargs) -> str:
+        return "{name} = fields.FloatField({null}{default}{comment})".format(**kwargs)
+
+    @classmethod
+    def datetime_field(cls, **kwargs) -> str:
+        return "{name} = fields.DatetimeField({null}{default}{comment})".format(**kwargs)
+
+    @classmethod
+    def text_field(cls, **kwargs) -> str:
+        return "{name} = fields.TextField({null}{default}{comment})".format(**kwargs)
+
+    @classmethod
+    def char_field(cls, **kwargs) -> str:
+        return "{name} = fields.CharField({pk}{index}{length}{null}{default}{comment})".format(
+            **kwargs
+        )
+
+    @classmethod
+    def int_field(cls, **kwargs) -> str:
+        return "{name} = fields.IntField({pk}{index}{comment})".format(**kwargs)
+
+    @classmethod
+    def smallint_field(cls, **kwargs) -> str:
+        return "{name} = fields.SmallIntField({pk}{index}{comment})".format(**kwargs)
+
+    @classmethod
+    def bigint_field(cls, **kwargs) -> str:
+        return "{name} = fields.BigIntField({pk}{index}{default}{comment})".format(**kwargs)
+
+    @classmethod
+    def bool_field(cls, **kwargs) -> str:
+        return "{name} = fields.BooleanField({null}{default}{comment})".format(**kwargs)
+
+    @classmethod
+    def uuid_field(cls, **kwargs) -> str:
+        return "{name} = fields.UUIDField({pk}{index}{default}{comment})".format(**kwargs)
+
+    @classmethod
+    def json_field(cls, **kwargs) -> str:
+        return "{name} = fields.JSONField({null}{default}{comment})".format(**kwargs)
+
+    @classmethod
+    def binary_field(cls, **kwargs) -> str:
+        return "{name} = fields.BinaryField({null}{default}{comment})".format(**kwargs)

aerich/inspectdb/mysql.py

@@ -0,0 +1,71 @@
+from __future__ import annotations
+
+from aerich.inspectdb import Column, FieldMapDict, Inspect
+
+
+class InspectMySQL(Inspect):
+    @property
+    def field_map(self) -> FieldMapDict:
+        return {
+            "int": self.int_field,
+            "smallint": self.smallint_field,
+            "tinyint": self.bool_field,
+            "bigint": self.bigint_field,
+            "varchar": self.char_field,
+            "char": self.char_field,
+            "longtext": self.text_field,
+            "text": self.text_field,
+            "datetime": self.datetime_field,
+            "float": self.float_field,
+            "date": self.date_field,
+            "time": self.time_field,
+            "decimal": self.decimal_field,
+            "json": self.json_field,
+            "longblob": self.binary_field,
+        }
+
+    async def get_all_tables(self) -> list[str]:
+        sql = "select TABLE_NAME from information_schema.TABLES where TABLE_SCHEMA=%s"
+        ret = await self.conn.execute_query_dict(sql, [self.database])
+        return list(map(lambda x: x["TABLE_NAME"], ret))
+
+    async def get_columns(self, table: str) -> list[Column]:
+        columns = []
+        sql = """select c.*, s.NON_UNIQUE, s.INDEX_NAME
+from information_schema.COLUMNS c
+         left join information_schema.STATISTICS s on c.TABLE_NAME = s.TABLE_NAME
+    and c.TABLE_SCHEMA = s.TABLE_SCHEMA
+    and c.COLUMN_NAME = s.COLUMN_NAME
+where c.TABLE_SCHEMA = %s
+  and c.TABLE_NAME = %s"""
+        ret = await self.conn.execute_query_dict(sql, [self.database, table])
+        for row in ret:
+            non_unique = row["NON_UNIQUE"]
+            if non_unique is None:
+                unique = False
+            else:
+                unique = not non_unique
+            index_name = row["INDEX_NAME"]
+            if index_name is None:
+                index = False
+            else:
+                index = row["INDEX_NAME"] != "PRIMARY"
+            columns.append(
+                Column(
+                    name=row["COLUMN_NAME"],
+                    data_type=row["DATA_TYPE"],
+                    null=row["IS_NULLABLE"] == "YES",
+                    default=row["COLUMN_DEFAULT"],
+                    pk=row["COLUMN_KEY"] == "PRI",
+                    comment=row["COLUMN_COMMENT"],
+                    unique=row["COLUMN_KEY"] == "UNI",
+                    extra=row["EXTRA"],
+                    # TODO: why `unque`?
+                    unque=unique,  # type:ignore
+                    index=index,
+                    length=row["CHARACTER_MAXIMUM_LENGTH"],
+                    max_digits=row["NUMERIC_PRECISION"],
+                    decimal_places=row["NUMERIC_SCALE"],
+                )
+            )
+        return columns

aerich/inspectdb/postgres.py

@@ -0,0 +1,79 @@
+from __future__ import annotations
+
+from typing import TYPE_CHECKING
+
+from aerich.inspectdb import Column, FieldMapDict, Inspect
+
+if TYPE_CHECKING:
+    from tortoise.backends.base_postgres.client import BasePostgresClient
+
+
+class InspectPostgres(Inspect):
+    def __init__(self, conn: "BasePostgresClient", tables: list[str] | None = None) -> None:
+        super().__init__(conn, tables)
+        self.schema = conn.server_settings.get("schema") or "public"
+
+    @property
+    def field_map(self) -> FieldMapDict:
+        return {
+            "int4": self.int_field,
+            "int8": self.int_field,
+            "smallint": self.smallint_field,
+            "varchar": self.char_field,
+            "text": self.text_field,
+            "bigint": self.bigint_field,
+            "timestamptz": self.datetime_field,
+            "float4": self.float_field,
+            "float8": self.float_field,
+            "date": self.date_field,
+            "time": self.time_field,
+            "decimal": self.decimal_field,
+            "numeric": self.decimal_field,
+            "uuid": self.uuid_field,
+            "jsonb": self.json_field,
+            "bytea": self.binary_field,
+            "bool": self.bool_field,
+            "timestamp": self.datetime_field,
+        }
+
+    async def get_all_tables(self) -> list[str]:
+        sql = "select TABLE_NAME from information_schema.TABLES where table_catalog=$1 and table_schema=$2"
+        ret = await self.conn.execute_query_dict(sql, [self.database, self.schema])
+        return list(map(lambda x: x["table_name"], ret))
+
+    async def get_columns(self, table: str) -> list[Column]:
+        columns = []
+        sql = f"""select c.column_name,
+       col_description('public.{table}'::regclass, ordinal_position) as column_comment,
+       t.constraint_type as column_key,
+       udt_name as data_type,
+       is_nullable,
+       column_default,
+       character_maximum_length,
+       numeric_precision,
+       numeric_scale
+from information_schema.constraint_column_usage const
+         join information_schema.table_constraints t
+              using (table_catalog, table_schema, table_name, constraint_catalog, constraint_schema, constraint_name)
+         right join information_schema.columns c using (column_name, table_catalog, table_schema, table_name)
+where c.table_catalog = $1
+  and c.table_name = $2
+  and c.table_schema = $3"""  # nosec:B608
+        ret = await self.conn.execute_query_dict(sql, [self.database, table, self.schema])
+        for row in ret:
+            columns.append(
+                Column(
+                    name=row["column_name"],
+                    data_type=row["data_type"],
+                    null=row["is_nullable"] == "YES",
+                    default=row["column_default"],
+                    length=row["character_maximum_length"],
+                    max_digits=row["numeric_precision"],
+                    decimal_places=row["numeric_scale"],
+                    comment=row["column_comment"],
+                    pk=row["column_key"] == "PRIMARY KEY",
+                    unique=False,  # can't get this simply
+                    index=False,  # can't get this simply
+                )
+            )
+        return columns

aerich/inspectdb/sqlite.py

@@ -0,0 +1,61 @@
+from __future__ import annotations
+
+from aerich.inspectdb import Column, FieldMapDict, Inspect
+
+
+class InspectSQLite(Inspect):
+    @property
+    def field_map(self) -> FieldMapDict:
+        return {
+            "INTEGER": self.int_field,
+            "INT": self.bool_field,
+            "SMALLINT": self.smallint_field,
+            "VARCHAR": self.char_field,
+            "TEXT": self.text_field,
+            "TIMESTAMP": self.datetime_field,
+            "REAL": self.float_field,
+            "BIGINT": self.bigint_field,
+            "DATE": self.date_field,
+            "TIME": self.time_field,
+            "JSON": self.json_field,
+            "BLOB": self.binary_field,
+        }
+
+    async def get_columns(self, table: str) -> list[Column]:
+        columns = []
+        sql = f"PRAGMA table_info({table})"
+        ret = await self.conn.execute_query_dict(sql)
+        columns_index = await self._get_columns_index(table)
+        for row in ret:
+            try:
+                length = row["type"].split("(")[1].split(")")[0]
+            except IndexError:
+                length = None
+            columns.append(
+                Column(
+                    name=row["name"],
+                    data_type=row["type"].split("(")[0],
+                    null=row["notnull"] == 0,
+                    default=row["dflt_value"],
+                    length=length,
+                    pk=row["pk"] == 1,
+                    unique=columns_index.get(row["name"]) == "unique",
+                    index=columns_index.get(row["name"]) == "index",
+                )
+            )
+        return columns
+
+    async def _get_columns_index(self, table: str) -> dict[str, str]:
+        sql = f"PRAGMA index_list ({table})"
+        indexes = await self.conn.execute_query_dict(sql)
+        ret = {}
+        for index in indexes:
+            sql = f"PRAGMA index_info({index['name']})"
+            index_info = (await self.conn.execute_query_dict(sql))[0]
+            ret[index_info["name"]] = "unique" if index["unique"] else "index"
+        return ret
+
+    async def get_all_tables(self) -> list[str]:
+        sql = "select tbl_name from sqlite_master where type='table' and name!='sqlite_sequence'"
+        ret = await self.conn.execute_query_dict(sql)
+        return list(map(lambda x: x["tbl_name"], ret))

aerich/migrate.py

@@ -1,10 +1,11 @@
+import hashlib
+import importlib
 import os
 from datetime import datetime
-from hashlib import md5
 from pathlib import Path
-from typing import Dict, List, Optional, Tuple, Type, Union
+from typing import Dict, Iterable, List, Optional, Set, Tuple, Type, Union, cast
 
-import click
+import asyncclick as click
 from dictdiffer import diff
 from tortoise import BaseDBAsyncClient, Model, Tortoise
 from tortoise.exceptions import OperationalError
@@ -12,12 +13,20 @@ from tortoise.indexes import Index
 
 from aerich.ddl import BaseDDL
 from aerich.models import MAX_VERSION_LENGTH, Aerich
-from aerich.utils import (
+from aerich.utils import get_app_connection, get_models_describe, is_default_function
-    get_app_connection,
+
-    get_models_describe,
+MIGRATE_TEMPLATE = """from tortoise import BaseDBAsyncClient
-    is_default_function,
+
-    write_version_file,
+
-)
+async def upgrade(db: BaseDBAsyncClient) -> str:
+    return \"\"\"
+        {upgrade_sql}\"\"\"
+
+
+async def downgrade(db: BaseDBAsyncClient) -> str:
+    return \"\"\"
+        {downgrade_sql}\"\"\"
+"""
 
 
 class Migrate:
@@ -28,68 +37,68 @@ class Migrate:
     _upgrade_m2m: List[str] = []
     _downgrade_m2m: List[str] = []
     _aerich = Aerich.__name__
-    _rename_old = []
+    _rename_old: List[str] = []
-    _rename_new = []
+    _rename_new: List[str] = []
 
     ddl: BaseDDL
+    ddl_class: Type[BaseDDL]
     _last_version_content: Optional[dict] = None
     app: str
     migrate_location: Path
     dialect: str
     _db_version: Optional[str] = None
 
+    @staticmethod
+    def get_field_by_name(name: str, fields: List[dict]) -> dict:
+        return next(filter(lambda x: x.get("name") == name, fields))
+
     @classmethod
     def get_all_version_files(cls) -> List[str]:
         return sorted(
-            filter(lambda x: x.endswith("sql"), os.listdir(cls.migrate_location)),
+            filter(lambda x: x.endswith("py"), os.listdir(cls.migrate_location)),
             key=lambda x: int(x.split("_")[0]),
         )
 
     @classmethod
     def _get_model(cls, model: str) -> Type[Model]:
-        return Tortoise.apps.get(cls.app).get(model)
+        return Tortoise.apps[cls.app][model]
 
     @classmethod
     async def get_last_version(cls) -> Optional[Aerich]:
         try:
             return await Aerich.filter(app=cls.app).first()
         except OperationalError:
-            pass
+            return None
 
     @classmethod
-    async def _get_db_version(cls, connection: BaseDBAsyncClient):
+    async def _get_db_version(cls, connection: BaseDBAsyncClient) -> None:
         if cls.dialect == "mysql":
             sql = "select version() as version"
             ret = await connection.execute_query(sql)
             cls._db_version = ret[1][0].get("version")
 
     @classmethod
-    async def init(cls, config: dict, app: str, location: str):
+    async def load_ddl_class(cls) -> Type[BaseDDL]:
+        ddl_dialect_module = importlib.import_module(f"aerich.ddl.{cls.dialect}")
+        return getattr(ddl_dialect_module, f"{cls.dialect.capitalize()}DDL")
+
+    @classmethod
+    async def init(cls, config: dict, app: str, location: str) -> None:
         await Tortoise.init(config=config)
         last_version = await cls.get_last_version()
         cls.app = app
         cls.migrate_location = Path(location, app)
         if last_version:
-            cls._last_version_content = last_version.content
+            cls._last_version_content = cast(dict, last_version.content)
 
         connection = get_app_connection(config, app)
         cls.dialect = connection.schema_generator.DIALECT
-        if cls.dialect == "mysql":
+        cls.ddl_class = await cls.load_ddl_class()
-            from aerich.ddl.mysql import MysqlDDL
+        cls.ddl = cls.ddl_class(connection)
-
-            cls.ddl = MysqlDDL(connection)
-        elif cls.dialect == "sqlite":
-            from aerich.ddl.sqlite import SqliteDDL
-
-            cls.ddl = SqliteDDL(connection)
-        elif cls.dialect == "postgres":
-            from aerich.ddl.postgres import PostgresDDL
-
-            cls.ddl = PostgresDDL(connection)
         await cls._get_db_version(connection)
 
     @classmethod
-    async def _get_last_version_num(cls):
+    async def _get_last_version_num(cls) -> Optional[int]:
         last_version = await cls.get_last_version()
         if not last_version:
             return None
@@ -97,50 +106,68 @@ class Migrate:
         return int(version.split("_", 1)[0])
 
     @classmethod
-    async def generate_version(cls, name=None):
+    async def generate_version(cls, name=None) -> str:
         now = datetime.now().strftime("%Y%m%d%H%M%S").replace("/", "")
         last_version_num = await cls._get_last_version_num()
         if last_version_num is None:
-            return f"0_{now}_init.sql"
+            return f"0_{now}_init.py"
-        version = f"{last_version_num + 1}_{now}_{name}.sql"
+        version = f"{last_version_num + 1}_{now}_{name}.py"
         if len(version) > MAX_VERSION_LENGTH:
             raise ValueError(f"Version name exceeds maximum length ({MAX_VERSION_LENGTH})")
         return version
 
     @classmethod
-    async def _generate_diff_sql(cls, name):
+    async def _generate_diff_py(cls, name) -> str:
         version = await cls.generate_version(name)
         # delete if same version exists
         for version_file in cls.get_all_version_files():
             if version_file.startswith(version.split("_")[0]):
                 os.unlink(Path(cls.migrate_location, version_file))
-        content = {
+
-            "upgrade": list(dict.fromkeys(cls.upgrade_operators)),
+        content = cls._get_diff_file_content()
-            "downgrade": list(dict.fromkeys(cls.downgrade_operators)),
+        Path(cls.migrate_location, version).write_text(content, encoding="utf-8")
-        }
-        write_version_file(Path(cls.migrate_location, version), content)
         return version
 
     @classmethod
-    async def migrate(cls, name) -> str:
+    async def migrate(cls, name: str, empty: bool) -> str:
         """
         diff old models and new models to generate diff content
-        :param name:
+        :param name: str name for migration
+        :param empty: bool if True generates empty migration
         :return:
         """
+        if empty:
+            return await cls._generate_diff_py(name)
         new_version_content = get_models_describe(cls.app)
-        cls.diff_models(cls._last_version_content, new_version_content)
+        last_version = cast(dict, cls._last_version_content)
-        cls.diff_models(new_version_content, cls._last_version_content, False)
+        cls.diff_models(last_version, new_version_content)
+        cls.diff_models(new_version_content, last_version, False)
 
         cls._merge_operators()
 
         if not cls.upgrade_operators:
             return ""
 
-        return await cls._generate_diff_sql(name)
+        return await cls._generate_diff_py(name)
 
     @classmethod
-    def _add_operator(cls, operator: str, upgrade=True, fk_m2m_index=False):
+    def _get_diff_file_content(cls) -> str:
+        """
+        builds content for diff file from template
+        """
+
+        def join_lines(lines: List[str]) -> str:
+            if not lines:
+                return ""
+            return ";\n        ".join(lines) + ";"
+
+        return MIGRATE_TEMPLATE.format(
+            upgrade_sql=join_lines(cls.upgrade_operators),
+            downgrade_sql=join_lines(cls.downgrade_operators),
+        )
+
+    @classmethod
+    def _add_operator(cls, operator: str, upgrade=True, fk_m2m_index=False) -> None:
         """
         add operator,differentiate fk because fk is order limit
         :param operator:
@@ -148,6 +175,7 @@ class Migrate:
         :param fk_m2m_index:
         :return:
         """
+        operator = operator.rstrip(";")
         if upgrade:
             if fk_m2m_index:
                 cls._upgrade_fk_m2m_index_operators.append(operator)
@@ -160,19 +188,37 @@ class Migrate:
                 cls.downgrade_operators.append(operator)
 
     @classmethod
-    def _handle_indexes(cls, model: Type[Model], indexes: List[Union[Tuple[str], Index]]):
+    def _handle_indexes(cls, model: Type[Model], indexes: List[Union[Tuple[str], Index]]) -> list:
-        ret = []
+        ret: list = []
-        for index in indexes:
+
-            if isinstance(index, Index):
+        def index_hash(self) -> str:
-                index.__hash__ = lambda self: md5(  # nosec: B303
+            h = hashlib.new("MD5", usedforsecurity=False)  # type:ignore[call-arg]
+            h.update(
                 self.index_name(cls.ddl.schema_generator, model).encode()
                 + self.__class__.__name__.encode()
-                ).hexdigest()
+            )
+            return h.hexdigest()
+
+        for index in indexes:
+            if isinstance(index, Index):
+                index.__hash__ = index_hash  # type:ignore[method-assign,assignment]
             ret.append(index)
         return ret
 
     @classmethod
-    def diff_models(cls, old_models: Dict[str, dict], new_models: Dict[str, dict], upgrade=True):
+    def _get_indexes(cls, model, model_describe: dict) -> Set[Union[Index, Tuple[str, ...]]]:
+        indexes: Set[Union[Index, Tuple[str, ...]]] = set()
+        for x in cls._handle_indexes(model, model_describe.get("indexes", [])):
+            if isinstance(x, Index):
+                indexes.add(x)
+            else:
+                indexes.add(cast(Tuple[str, ...], tuple(x)))
+        return indexes
+
+    @classmethod
+    def diff_models(
+        cls, old_models: Dict[str, dict], new_models: Dict[str, dict], upgrade=True
+    ) -> None:
         """
         diff models and add operators
         :param old_models:
@@ -185,39 +231,35 @@ class Migrate:
         new_models.pop(_aerich, None)
 
         for new_model_str, new_model_describe in new_models.items():
-            model = cls._get_model(new_model_describe.get("name").split(".")[1])
+            model = cls._get_model(new_model_describe["name"].split(".")[1])
 
-            if new_model_str not in old_models.keys():
+            if new_model_str not in old_models:
                 if upgrade:
                     cls._add_operator(cls.add_model(model), upgrade)
                 else:
                     # we can't find origin model when downgrade, so skip
                     pass
             else:
-                old_model_describe = old_models.get(new_model_str)
+                old_model_describe = cast(dict, old_models.get(new_model_str))
                 # rename table
-                new_table = new_model_describe.get("table")
+                new_table = cast(str, new_model_describe.get("table"))
-                old_table = old_model_describe.get("table")
+                old_table = cast(str, old_model_describe.get("table"))
                 if new_table != old_table:
                     cls._add_operator(cls.rename_table(model, old_table, new_table), upgrade)
                 old_unique_together = set(
-                    map(lambda x: tuple(x), old_model_describe.get("unique_together"))
+                    map(
+                        lambda x: tuple(x),
+                        cast(List[Iterable[str]], old_model_describe.get("unique_together")),
+                    )
                 )
                 new_unique_together = set(
-                    map(lambda x: tuple(x), new_model_describe.get("unique_together"))
-                )
-                old_indexes = set(
                     map(
-                        lambda x: x if isinstance(x, Index) else tuple(x),
+                        lambda x: tuple(x),
-                        cls._handle_indexes(model, old_model_describe.get("indexes", [])),
+                        cast(List[Iterable[str]], new_model_describe.get("unique_together")),
-                    )
-                )
-                new_indexes = set(
-                    map(
-                        lambda x: x if isinstance(x, Index) else tuple(x),
-                        cls._handle_indexes(model, new_model_describe.get("indexes", [])),
                     )
                 )
+                old_indexes = cls._get_indexes(model, old_model_describe)
+                new_indexes = cls._get_indexes(model, new_model_describe)
                 old_pk_field = old_model_describe.get("pk_field")
                 new_pk_field = new_model_describe.get("pk_field")
                 # pk field
@@ -227,12 +269,19 @@ class Migrate:
                     if action == "change" and option == "name":
                         cls._add_operator(cls._rename_field(model, *change), upgrade)
                 # m2m fields
-                old_m2m_fields = old_model_describe.get("m2m_fields")
+                old_m2m_fields = cast(List[dict], old_model_describe.get("m2m_fields"))
-                new_m2m_fields = new_model_describe.get("m2m_fields")
+                new_m2m_fields = cast(List[dict], new_model_describe.get("m2m_fields"))
                 for action, option, change in diff(old_m2m_fields, new_m2m_fields):
                     if change[0][0] == "db_constraint":
                         continue
-                    table = change[0][1].get("through")
+                    new_value = change[0][1]
+                    if isinstance(new_value, str):
+                        for new_m2m_field in new_m2m_fields:
+                            if new_m2m_field["name"] == new_value:
+                                table = cast(str, new_m2m_field.get("through"))
+                                break
+                    else:
+                        table = new_value.get("through")
                     if action == "add":
                         add = False
                         if upgrade and table not in cls._upgrade_m2m:
@@ -242,12 +291,9 @@ class Migrate:
                             cls._downgrade_m2m.append(table)
                             add = True
                         if add:
+                            ref_desc = cast(dict, new_models.get(new_value.get("model_name")))
                             cls._add_operator(
-                                cls.create_m2m(
+                                cls.create_m2m(model, new_value, ref_desc),
-                                    model,
-                                    change[0][1],
-                                    new_models.get(change[0][1].get("model_name")),
-                                ),
                                 upgrade,
                                 fk_m2m_index=True,
                             )
@@ -268,28 +314,36 @@ class Migrate:
                 for index in old_unique_together.difference(new_unique_together):
                     cls._add_operator(cls._drop_index(model, index, True), upgrade, True)
                 # add indexes
-                for index in new_indexes.difference(old_indexes):
+                for idx in new_indexes.difference(old_indexes):
-                    cls._add_operator(cls._add_index(model, index, False), upgrade, True)
+                    cls._add_operator(cls._add_index(model, idx, False), upgrade, True)
                 # remove indexes
-                for index in old_indexes.difference(new_indexes):
+                for idx in old_indexes.difference(new_indexes):
-                    cls._add_operator(cls._drop_index(model, index, False), upgrade, True)
+                    cls._add_operator(cls._drop_index(model, idx, False), upgrade, True)
-                old_data_fields = old_model_describe.get("data_fields")
+                old_data_fields = list(
-                new_data_fields = new_model_describe.get("data_fields")
+                    filter(
+                        lambda x: x.get("db_field_types") is not None,
+                        cast(List[dict], old_model_describe.get("data_fields")),
+                    )
+                )
+                new_data_fields = list(
+                    filter(
+                        lambda x: x.get("db_field_types") is not None,
+                        cast(List[dict], new_model_describe.get("data_fields")),
+                    )
+                )
 
-                old_data_fields_name = list(map(lambda x: x.get("name"), old_data_fields))
+                old_data_fields_name = cast(List[str], [i.get("name") for i in old_data_fields])
-                new_data_fields_name = list(map(lambda x: x.get("name"), new_data_fields))
+                new_data_fields_name = cast(List[str], [i.get("name") for i in new_data_fields])
 
                 # add fields or rename fields
                 for new_data_field_name in set(new_data_fields_name).difference(
                     set(old_data_fields_name)
                 ):
-                    new_data_field = next(
+                    new_data_field = cls.get_field_by_name(new_data_field_name, new_data_fields)
-                        filter(lambda x: x.get("name") == new_data_field_name, new_data_fields)
-                    )
                     is_rename = False
                     for old_data_field in old_data_fields:
                         changes = list(diff(old_data_field, new_data_field))
-                        old_data_field_name = old_data_field.get("name")
+                        old_data_field_name = cast(str, old_data_field.get("name"))
                         if len(changes) == 2:
                             # rename field
                             if (
@@ -347,44 +401,52 @@ class Migrate:
                             ),
                             upgrade,
                         )
+                        if new_data_field["indexed"]:
+                            cls._add_operator(
+                                cls._add_index(
+                                    model, (new_data_field["db_column"],), new_data_field["unique"]
+                                ),
+                                upgrade,
+                                True,
+                            )
                 # remove fields
                 for old_data_field_name in set(old_data_fields_name).difference(
                     set(new_data_fields_name)
                 ):
-                    # don't remove field if is rename
+                    # don't remove field if is renamed
                     if (upgrade and old_data_field_name in cls._rename_old) or (
                         not upgrade and old_data_field_name in cls._rename_new
                     ):
                         continue
+                    old_data_field = cls.get_field_by_name(old_data_field_name, old_data_fields)
+                    db_column = cast(str, old_data_field["db_column"])
                     cls._add_operator(
-                        cls._remove_field(
+                        cls._remove_field(model, db_column),
-                            model,
-                            next(
-                                filter(
-                                    lambda x: x.get("name") == old_data_field_name, old_data_fields
-                                )
-                            ).get("db_column"),
-                        ),
                         upgrade,
                     )
-                old_fk_fields = old_model_describe.get("fk_fields")
+                    if old_data_field["indexed"]:
-                new_fk_fields = new_model_describe.get("fk_fields")
+                        is_unique_field = old_data_field.get("unique")
+                        cls._add_operator(
+                            cls._drop_index(model, {db_column}, is_unique_field),
+                            upgrade,
+                            True,
+                        )
 
-                old_fk_fields_name = list(map(lambda x: x.get("name"), old_fk_fields))
+                old_fk_fields = cast(List[dict], old_model_describe.get("fk_fields"))
-                new_fk_fields_name = list(map(lambda x: x.get("name"), new_fk_fields))
+                new_fk_fields = cast(List[dict], new_model_describe.get("fk_fields"))
+
+                old_fk_fields_name: List[str] = [i.get("name", "") for i in old_fk_fields]
+                new_fk_fields_name: List[str] = [i.get("name", "") for i in new_fk_fields]
 
                 # add fk
                 for new_fk_field_name in set(new_fk_fields_name).difference(
                     set(old_fk_fields_name)
                 ):
-                    fk_field = next(
+                    fk_field = cls.get_field_by_name(new_fk_field_name, new_fk_fields)
-                        filter(lambda x: x.get("name") == new_fk_field_name, new_fk_fields)
-                    )
                     if fk_field.get("db_constraint"):
+                        ref_describe = cast(dict, new_models[fk_field["python_type"]])
                         cls._add_operator(
-                            cls._add_fk(
+                            cls._add_fk(model, fk_field, ref_describe),
-                                model, fk_field, new_models.get(fk_field.get("python_type"))
-                            ),
                             upgrade,
                             fk_m2m_index=True,
                         )
@@ -392,41 +454,44 @@ class Migrate:
                 for old_fk_field_name in set(old_fk_fields_name).difference(
                     set(new_fk_fields_name)
                 ):
-                    old_fk_field = next(
+                    old_fk_field = cls.get_field_by_name(
-                        filter(lambda x: x.get("name") == old_fk_field_name, old_fk_fields)
+                        old_fk_field_name, cast(List[dict], old_fk_fields)
                     )
                     if old_fk_field.get("db_constraint"):
+                        ref_describe = cast(dict, old_models[old_fk_field["python_type"]])
                         cls._add_operator(
-                            cls._drop_fk(
+                            cls._drop_fk(model, old_fk_field, ref_describe),
-                                model, old_fk_field, old_models.get(old_fk_field.get("python_type"))
-                            ),
                             upgrade,
                             fk_m2m_index=True,
                         )
                 # change fields
                 for field_name in set(new_data_fields_name).intersection(set(old_data_fields_name)):
-                    old_data_field = next(
+                    old_data_field = cls.get_field_by_name(field_name, old_data_fields)
-                        filter(lambda x: x.get("name") == field_name, old_data_fields)
+                    new_data_field = cls.get_field_by_name(field_name, new_data_fields)
-                    )
-                    new_data_field = next(
-                        filter(lambda x: x.get("name") == field_name, new_data_fields)
-                    )
                     changes = diff(old_data_field, new_data_field)
+                    modified = False
                     for change in changes:
                         _, option, old_new = change
                         if option == "indexed":
                             # change index
-                            unique = new_data_field.get("unique")
                             if old_new[0] is False and old_new[1] is True:
+                                unique = new_data_field.get("unique")
                                 cls._add_operator(
                                     cls._add_index(model, (field_name,), unique), upgrade, True
                                 )
                             else:
+                                unique = old_data_field.get("unique")
                                 cls._add_operator(
                                     cls._drop_index(model, (field_name,), unique), upgrade, True
                                 )
                         elif option == "db_field_types.":
-                            # continue since repeated with others
+                            if new_data_field.get("field_type") == "DecimalField":
+                                # modify column
+                                cls._add_operator(
+                                    cls._modify_field(model, new_data_field),
+                                    upgrade,
+                                )
+                            else:
                                 continue
                         elif option == "default":
                             if not (
@@ -442,110 +507,131 @@ class Migrate:
                         elif option == "nullable":
                             # change nullable
                             cls._add_operator(cls._alter_null(model, new_data_field), upgrade)
+                        elif option == "description":
+                            # change comment
+                            cls._add_operator(cls._set_comment(model, new_data_field), upgrade)
                         else:
+                            if modified:
+                                continue
                             # modify column
                             cls._add_operator(
                                 cls._modify_field(model, new_data_field),
                                 upgrade,
                             )
+                            modified = True
 
-        for old_model in old_models:
+        for old_model in old_models.keys() - new_models.keys():
-            if old_model not in new_models.keys():
+            cls._add_operator(cls.drop_model(old_models[old_model]["table"]), upgrade)
-                cls._add_operator(cls.drop_model(old_models.get(old_model).get("table")), upgrade)
 
     @classmethod
-    def rename_table(cls, model: Type[Model], old_table_name: str, new_table_name: str):
+    def rename_table(cls, model: Type[Model], old_table_name: str, new_table_name: str) -> str:
         return cls.ddl.rename_table(model, old_table_name, new_table_name)
 
     @classmethod
-    def add_model(cls, model: Type[Model]):
+    def add_model(cls, model: Type[Model]) -> str:
         return cls.ddl.create_table(model)
 
     @classmethod
-    def drop_model(cls, table_name: str):
+    def drop_model(cls, table_name: str) -> str:
         return cls.ddl.drop_table(table_name)
 
     @classmethod
-    def create_m2m(cls, model: Type[Model], field_describe: dict, reference_table_describe: dict):
+    def create_m2m(
+        cls, model: Type[Model], field_describe: dict, reference_table_describe: dict
+    ) -> str:
         return cls.ddl.create_m2m(model, field_describe, reference_table_describe)
 
     @classmethod
-    def drop_m2m(cls, table_name: str):
+    def drop_m2m(cls, table_name: str) -> str:
         return cls.ddl.drop_m2m(table_name)
 
     @classmethod
-    def _resolve_fk_fields_name(cls, model: Type[Model], fields_name: Tuple[str]):
+    def _resolve_fk_fields_name(cls, model: Type[Model], fields_name: Iterable[str]) -> List[str]:
         ret = []
         for field_name in fields_name:
+            try:
                 field = model._meta.fields_map[field_name]
-            if field.source_field:
+            except KeyError:
-                ret.append(field.source_field)
+                # field dropped or to be add
-            elif field_name in model._meta.fk_fields:
+                pass
-                ret.append(field_name + "_id")
             else:
+                if field.source_field:
+                    field_name = field.source_field
+                elif field_name in model._meta.fk_fields:
+                    field_name += "_id"
             ret.append(field_name)
         return ret
 
     @classmethod
-    def _drop_index(cls, model: Type[Model], fields_name: Union[Tuple[str], Index], unique=False):
+    def _drop_index(
+        cls, model: Type[Model], fields_name: Union[Iterable[str], Index], unique=False
+    ) -> str:
         if isinstance(fields_name, Index):
             return cls.ddl.drop_index_by_name(
                 model, fields_name.index_name(cls.ddl.schema_generator, model)
             )
-        fields_name = cls._resolve_fk_fields_name(model, fields_name)
+        field_names = cls._resolve_fk_fields_name(model, fields_name)
-        return cls.ddl.drop_index(model, fields_name, unique)
+        return cls.ddl.drop_index(model, field_names, unique)
 
     @classmethod
-    def _add_index(cls, model: Type[Model], fields_name: Union[Tuple[str], Index], unique=False):
+    def _add_index(
+        cls, model: Type[Model], fields_name: Union[Iterable[str], Index], unique=False
+    ) -> str:
         if isinstance(fields_name, Index):
             return fields_name.get_sql(cls.ddl.schema_generator, model, False)
-        fields_name = cls._resolve_fk_fields_name(model, fields_name)
+        field_names = cls._resolve_fk_fields_name(model, fields_name)
-        return cls.ddl.add_index(model, fields_name, unique)
+        return cls.ddl.add_index(model, field_names, unique)
 
     @classmethod
-    def _add_field(cls, model: Type[Model], field_describe: dict, is_pk: bool = False):
+    def _add_field(cls, model: Type[Model], field_describe: dict, is_pk: bool = False) -> str:
         return cls.ddl.add_column(model, field_describe, is_pk)
 
     @classmethod
-    def _alter_default(cls, model: Type[Model], field_describe: dict):
+    def _alter_default(cls, model: Type[Model], field_describe: dict) -> str:
         return cls.ddl.alter_column_default(model, field_describe)
 
     @classmethod
-    def _alter_null(cls, model: Type[Model], field_describe: dict):
+    def _alter_null(cls, model: Type[Model], field_describe: dict) -> str:
         return cls.ddl.alter_column_null(model, field_describe)
 
     @classmethod
-    def _set_comment(cls, model: Type[Model], field_describe: dict):
+    def _set_comment(cls, model: Type[Model], field_describe: dict) -> str:
         return cls.ddl.set_comment(model, field_describe)
 
     @classmethod
-    def _modify_field(cls, model: Type[Model], field_describe: dict):
+    def _modify_field(cls, model: Type[Model], field_describe: dict) -> str:
         return cls.ddl.modify_column(model, field_describe)
 
     @classmethod
-    def _drop_fk(cls, model: Type[Model], field_describe: dict, reference_table_describe: dict):
+    def _drop_fk(
+        cls, model: Type[Model], field_describe: dict, reference_table_describe: dict
+    ) -> str:
         return cls.ddl.drop_fk(model, field_describe, reference_table_describe)
 
     @classmethod
-    def _remove_field(cls, model: Type[Model], column_name: str):
+    def _remove_field(cls, model: Type[Model], column_name: str) -> str:
         return cls.ddl.drop_column(model, column_name)
 
     @classmethod
-    def _rename_field(cls, model: Type[Model], old_field_name: str, new_field_name: str):
+    def _rename_field(cls, model: Type[Model], old_field_name: str, new_field_name: str) -> str:
         return cls.ddl.rename_column(model, old_field_name, new_field_name)
 
     @classmethod
-    def _change_field(cls, model: Type[Model], old_field_describe: dict, new_field_describe: dict):
+    def _change_field(
-        db_field_types = new_field_describe.get("db_field_types")
+        cls, model: Type[Model], old_field_describe: dict, new_field_describe: dict
+    ) -> str:
+        db_field_types = cast(dict, new_field_describe.get("db_field_types"))
         return cls.ddl.change_column(
             model,
-            old_field_describe.get("db_column"),
+            cast(str, old_field_describe.get("db_column")),
-            new_field_describe.get("db_column"),
+            cast(str, new_field_describe.get("db_column")),
-            db_field_types.get(cls.dialect) or db_field_types.get(""),
+            cast(str, db_field_types.get(cls.dialect) or db_field_types.get("")),
         )
 
     @classmethod
-    def _add_fk(cls, model: Type[Model], field_describe: dict, reference_table_describe: dict):
+    def _add_fk(
+        cls, model: Type[Model], field_describe: dict, reference_table_describe: dict
+    ) -> str:
         """
         add fk
         :param model:
@@ -556,7 +642,7 @@ class Migrate:
         return cls.ddl.add_fk(model, field_describe, reference_table_describe)
 
     @classmethod
-    def _merge_operators(cls):
+    def _merge_operators(cls) -> None:
         """
         fk/m2m/index must be last when add,first when drop
         :return:

aerich/models.py

@@ -3,12 +3,13 @@ from tortoise import Model, fields
 from aerich.coder import decoder, encoder
 
 MAX_VERSION_LENGTH = 255
+MAX_APP_LENGTH = 100
 
 
 class Aerich(Model):
     version = fields.CharField(max_length=MAX_VERSION_LENGTH)
-    app = fields.CharField(max_length=20)
+    app = fields.CharField(max_length=MAX_APP_LENGTH)
-    content = fields.JSONField(encoder=encoder, decoder=decoder)
+    content: dict = fields.JSONField(encoder=encoder, decoder=decoder)
 
     class Meta:
         ordering = ["-id"]

aerich/utils.py

@@ -1,17 +1,18 @@
-import importlib
+import importlib.util
 import os
 import re
 import sys
 from pathlib import Path
-from typing import Dict, Union
+from types import ModuleType
+from typing import Dict, Optional, Union
 
-from click import BadOptionUsage, ClickException, Context
+from asyncclick import BadOptionUsage, ClickException, Context
 from tortoise import BaseDBAsyncClient, Tortoise
 
 
 def add_src_path(path: str) -> str:
     """
-    add a folder to the paths so we can import from there
+    add a folder to the paths, so we can import from there
     :param path: path to add
     :return: absolute path
     """
@@ -77,60 +78,6 @@ def get_tortoise_config(ctx: Context, tortoise_orm: str) -> dict:
     return config
 
 
-_UPGRADE = "-- upgrade --\n"
-_DOWNGRADE = "-- downgrade --\n"
-
-
-def get_version_content_from_file(version_file: Union[str, Path]) -> Dict:
-    """
-    get version content
-    :param version_file:
-    :return:
-    """
-    with open(version_file, "r", encoding="utf-8") as f:
-        content = f.read()
-        first = content.index(_UPGRADE)
-        try:
-            second = content.index(_DOWNGRADE)
-        except ValueError:
-            second = len(content) - 1
-        upgrade_content = content[first + len(_UPGRADE) : second].strip()  # noqa:E203
-        downgrade_content = content[second + len(_DOWNGRADE) :].strip()  # noqa:E203
-        ret = {
-            "upgrade": list(filter(lambda x: x or False, upgrade_content.split(";\n"))),
-            "downgrade": list(filter(lambda x: x or False, downgrade_content.split(";\n"))),
-        }
-        return ret
-
-
-def write_version_file(version_file: Path, content: Dict):
-    """
-    write version file
-    :param version_file:
-    :param content:
-    :return:
-    """
-    with open(version_file, "w", encoding="utf-8") as f:
-        f.write(_UPGRADE)
-        upgrade = content.get("upgrade")
-        if len(upgrade) > 1:
-            f.write(";\n".join(upgrade))
-            if not upgrade[-1].endswith(";"):
-                f.write(";\n")
-        else:
-            f.write(f"{upgrade[0]}")
-            if not upgrade[0].endswith(";"):
-                f.write(";")
-            f.write("\n")
-        downgrade = content.get("downgrade")
-        if downgrade:
-            f.write(_DOWNGRADE)
-            if len(downgrade) > 1:
-                f.write(";\n".join(downgrade) + ";\n")
-            else:
-                f.write(f"{downgrade[0]};\n")
-
-
 def get_models_describe(app: str) -> Dict:
     """
     get app models describe
@@ -138,11 +85,19 @@ def get_models_describe(app: str) -> Dict:
     :return:
     """
     ret = {}
-    for model in Tortoise.apps.get(app).values():
+    for model in Tortoise.apps[app].values():
         describe = model.describe()
         ret[describe.get("name")] = describe
     return ret
 
 
-def is_default_function(string: str):
+def is_default_function(string: str) -> Optional[re.Match]:
     return re.match(r"^<function.+>$", str(string or ""))
+
+
+def import_py_file(file: Union[str, Path]) -> ModuleType:
+    module_name, file_ext = os.path.splitext(os.path.split(file)[-1])
+    spec = importlib.util.spec_from_file_location(module_name, file)
+    module = importlib.util.module_from_spec(spec)  # type:ignore[arg-type]
+    spec.loader.exec_module(module)  # type:ignore[union-attr]
+    return module

aerich/version.py

@@ -1 +1 @@
-__version__ = "0.6.2"
+__version__ = "0.8.0"

conftest.py

@@ -1,19 +1,22 @@
 import asyncio
 import os
+from typing import Generator
 
 import pytest
 from tortoise import Tortoise, expand_db_url, generate_schema_for_client
 from tortoise.backends.asyncpg.schema_generator import AsyncpgSchemaGenerator
 from tortoise.backends.mysql.schema_generator import MySQLSchemaGenerator
 from tortoise.backends.sqlite.schema_generator import SqliteSchemaGenerator
+from tortoise.exceptions import DBConnectionError, OperationalError
 
 from aerich.ddl.mysql import MysqlDDL
 from aerich.ddl.postgres import PostgresDDL
 from aerich.ddl.sqlite import SqliteDDL
 from aerich.migrate import Migrate
 
-db_url = os.getenv("TEST_DB", "sqlite://:memory:")
+MEMORY_SQLITE = "sqlite://:memory:"
-db_url_second = os.getenv("TEST_DB_SECOND", "sqlite://:memory:")
+db_url = os.getenv("TEST_DB", MEMORY_SQLITE)
+db_url_second = os.getenv("TEST_DB_SECOND", MEMORY_SQLITE)
 tortoise_orm = {
     "connections": {
         "default": expand_db_url(db_url, True),
@@ -27,7 +30,7 @@ tortoise_orm = {
 
 
 @pytest.fixture(scope="function", autouse=True)
-def reset_migrate():
+def reset_migrate() -> None:
     Migrate.upgrade_operators = []
     Migrate.downgrade_operators = []
     Migrate._upgrade_fk_m2m_index_operators = []
@@ -37,20 +40,27 @@ def reset_migrate():
 
 
 @pytest.fixture(scope="session")
-def event_loop():
+def event_loop() -> Generator:
     policy = asyncio.get_event_loop_policy()
     res = policy.new_event_loop()
     asyncio.set_event_loop(res)
-    res._close = res.close
+    res._close = res.close  # type:ignore[attr-defined]
-    res.close = lambda: None
+    res.close = lambda: None  # type:ignore[method-assign]
 
     yield res
 
-    res._close()
+    res._close()  # type:ignore[attr-defined]
 
 
 @pytest.fixture(scope="session", autouse=True)
-async def initialize_tests(event_loop, request):
+async def initialize_tests(event_loop, request) -> None:
+    # Placing init outside the try block since it doesn't
+    # establish connections to the DB eagerly.
+    await Tortoise.init(config=tortoise_orm)
+    try:
+        await Tortoise._drop_databases()
+    except (DBConnectionError, OperationalError):
+        pass
     await Tortoise.init(config=tortoise_orm, _create_db=True)
     await generate_schema_for_client(Tortoise.get_connection("default"), safe=True)
 

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "aerich"
-version = "0.6.2"
+version = "0.8.0"
 description = "A database migrations tool for Tortoise ORM."
 authors = ["long2ice <long2ice@gmail.com>"]
 license = "Apache-2.0"
@@ -15,33 +15,57 @@ packages = [
 include = ["CHANGELOG.md", "LICENSE", "README.md"]
 
 [tool.poetry.dependencies]
-python = "^3.7"
+python = "^3.8"
 tortoise-orm = "*"
-click = "*"
 asyncpg = { version = "*", optional = true }
-asyncmy = { version = "*", optional = true }
+asyncmy = { version = "^0.2.9", optional = true, allow-prereleases = true }
-ddlparse = "*"
+pydantic = "^2.0"
 dictdiffer = "*"
 tomlkit = "*"
+asyncclick = "^8.1.7.2"
 
-[tool.poetry.dev-dependencies]
+[tool.poetry.group.dev.dependencies]
-flake8 = "*"
+ruff = "*"
 isort = "*"
 black = "*"
 pytest = "*"
 pytest-xdist = "*"
-pytest-asyncio = "*"
+# Breaking change in 0.23.*
+# https://github.com/pytest-dev/pytest-asyncio/issues/706
+pytest-asyncio = "^0.21.2"
 bandit = "*"
 pytest-mock = "*"
 cryptography = "*"
+mypy = "^1.10.0"
 
 [tool.poetry.extras]
 asyncmy = ["asyncmy"]
 asyncpg = ["asyncpg"]
 
+
+[tool.aerich]
+tortoise_orm = "conftest.tortoise_orm"
+location = "./migrations"
+src_folder = "./."
+
 [build-system]
-requires = ["poetry>=0.12"]
+requires = ["poetry-core>=1.0.0"]
-build-backend = "poetry.masonry.api"
+build-backend = "poetry.core.masonry.api"
 
 [tool.poetry.scripts]
 aerich = "aerich.cli:main"
+
+[tool.black]
+line-length = 100
+target-version = ['py38', 'py39', 'py310', 'py311', 'py312']
+
+[tool.pytest.ini_options]
+asyncio_mode = 'auto'
+
+[tool.mypy]
+pretty = true
+python_version = "3.8"
+ignore_missing_imports = true
+
+[tool.ruff.lint]
+ignore = ['E501']

setup.cfg

@@ -1,2 +0,0 @@
-[flake8]
-ignore = E501,W503

tests/models.py

@@ -29,14 +29,15 @@ class User(Model):
     is_active = fields.BooleanField(default=True, description="Is Active")
     is_superuser = fields.BooleanField(default=False, description="Is SuperUser")
     intro = fields.TextField(default="")
+    longitude = fields.DecimalField(max_digits=10, decimal_places=8)
 
 
 class Email(Model):
-    email_id = fields.IntField(pk=True)
+    email_id = fields.IntField(primary_key=True)
-    email = fields.CharField(max_length=200, index=True)
+    email = fields.CharField(max_length=200, db_index=True)
     is_primary = fields.BooleanField(default=False)
     address = fields.CharField(max_length=200)
-    users = fields.ManyToManyField("models.User")
+    users: fields.ManyToManyRelation[User] = fields.ManyToManyField("models.User")
 
 
 def default_name():
@@ -46,12 +47,15 @@ def default_name():
 class Category(Model):
     slug = fields.CharField(max_length=100)
     name = fields.CharField(max_length=200, null=True, default=default_name)
-    user = fields.ForeignKeyField("models.User", description="User")
+    user: fields.ForeignKeyRelation[User] = fields.ForeignKeyField(
+        "models.User", description="User"
+    )
+    title = fields.CharField(max_length=20, unique=False)
     created_at = fields.DatetimeField(auto_now_add=True)
 
 
 class Product(Model):
-    categories = fields.ManyToManyField("models.Category")
+    categories: fields.ManyToManyRelation[Category] = fields.ManyToManyField("models.Category")
     name = fields.CharField(max_length=50)
     view_num = fields.IntField(description="View Num", default=0)
     sort = fields.IntField()
@@ -71,9 +75,11 @@ class Product(Model):
 class Config(Model):
     label = fields.CharField(max_length=200)
     key = fields.CharField(max_length=20)
-    value = fields.JSONField()
+    value: dict = fields.JSONField()
     status: Status = fields.IntEnumField(Status)
-    user = fields.ForeignKeyField("models.User", description="User")
+    user: fields.ForeignKeyRelation[User] = fields.ForeignKeyField(
+        "models.User", description="User"
+    )
 
 
 class NewModel(Model):

tests/models_second.py

@@ -34,18 +34,24 @@ class User(Model):
 class Email(Model):
     email = fields.CharField(max_length=200)
     is_primary = fields.BooleanField(default=False)
-    user = fields.ForeignKeyField("models_second.User", db_constraint=False)
+    user: fields.ForeignKeyRelation[User] = fields.ForeignKeyField(
+        "models_second.User", db_constraint=False
+    )
 
 
 class Category(Model):
     slug = fields.CharField(max_length=200)
     name = fields.CharField(max_length=200)
-    user = fields.ForeignKeyField("models_second.User", description="User")
+    user: fields.ForeignKeyRelation[User] = fields.ForeignKeyField(
+        "models_second.User", description="User"
+    )
     created_at = fields.DatetimeField(auto_now_add=True)
 
 
 class Product(Model):
-    categories = fields.ManyToManyField("models_second.Category")
+    categories: fields.ManyToManyRelation[Category] = fields.ManyToManyField(
+        "models_second.Category"
+    )
     name = fields.CharField(max_length=50)
     view_num = fields.IntField(description="View Num")
     sort = fields.IntField()
@@ -61,5 +67,5 @@ class Product(Model):
 class Config(Model):
     label = fields.CharField(max_length=200)
     key = fields.CharField(max_length=20)
-    value = fields.JSONField()
+    value: dict = fields.JSONField()
     status: Status = fields.IntEnumField(Status, default=Status.on)

tests/old_models.py

@@ -29,23 +29,29 @@ class User(Model):
     is_superuser = fields.BooleanField(default=False, description="Is SuperUser")
     avatar = fields.CharField(max_length=200, default="")
     intro = fields.TextField(default="")
+    longitude = fields.DecimalField(max_digits=12, decimal_places=9)
 
 
 class Email(Model):
     email = fields.CharField(max_length=200)
     is_primary = fields.BooleanField(default=False)
-    user = fields.ForeignKeyField("models.User", db_constraint=False)
+    user: fields.ForeignKeyRelation[User] = fields.ForeignKeyField(
+        "models.User", db_constraint=False
+    )
 
 
 class Category(Model):
     slug = fields.CharField(max_length=200)
     name = fields.CharField(max_length=200)
-    user = fields.ForeignKeyField("models.User", description="User")
+    user: fields.ForeignKeyRelation[User] = fields.ForeignKeyField(
+        "models.User", description="User"
+    )
     created_at = fields.DatetimeField(auto_now_add=True)
 
 
 class Product(Model):
-    categories = fields.ManyToManyField("models.Category")
+    categories: fields.ManyToManyRelation[Category] = fields.ManyToManyField("models.Category")
+    uid = fields.IntField(source_field="uuid", unique=True)
     name = fields.CharField(max_length=50)
     view_num = fields.IntField(description="View Num")
     sort = fields.IntField()
@@ -59,9 +65,10 @@ class Product(Model):
 
 
 class Config(Model):
+    name = fields.CharField(max_length=100, unique=True)
     label = fields.CharField(max_length=200)
     key = fields.CharField(max_length=20)
-    value = fields.JSONField()
+    value: dict = fields.JSONField()
     status: Status = fields.IntEnumField(Status, default=Status.on)
 
     class Meta:

tests/test_ddl.py

@@ -14,10 +14,11 @@ def test_create_table():
     `id` INT NOT NULL PRIMARY KEY AUTO_INCREMENT,
     `slug` VARCHAR(100) NOT NULL,
     `name` VARCHAR(200),
+    `title` VARCHAR(20) NOT NULL,
     `created_at` DATETIME(6) NOT NULL  DEFAULT CURRENT_TIMESTAMP(6),
     `user_id` INT NOT NULL COMMENT 'User',
     CONSTRAINT `fk_category_user_e2e3874c` FOREIGN KEY (`user_id`) REFERENCES `user` (`id`) ON DELETE CASCADE
-) CHARACTER SET utf8mb4;"""
+) CHARACTER SET utf8mb4"""
         )
 
     elif isinstance(Migrate.ddl, SqliteDDL):
@@ -27,9 +28,10 @@ def test_create_table():
     "id" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL,
     "slug" VARCHAR(100) NOT NULL,
     "name" VARCHAR(200),
+    "title" VARCHAR(20) NOT NULL,
     "created_at" TIMESTAMP NOT NULL  DEFAULT CURRENT_TIMESTAMP,
     "user_id" INT NOT NULL REFERENCES "user" ("id") ON DELETE CASCADE /* User */
-);"""
+)"""
         )
 
     elif isinstance(Migrate.ddl, PostgresDDL):
@@ -39,10 +41,11 @@ def test_create_table():
     "id" SERIAL NOT NULL PRIMARY KEY,
     "slug" VARCHAR(100) NOT NULL,
     "name" VARCHAR(200),
+    "title" VARCHAR(20) NOT NULL,
     "created_at" TIMESTAMPTZ NOT NULL  DEFAULT CURRENT_TIMESTAMP,
     "user_id" INT NOT NULL REFERENCES "user" ("id") ON DELETE CASCADE
 );
-COMMENT ON COLUMN "category"."user_id" IS 'User';"""
+COMMENT ON COLUMN "category"."user_id" IS 'User'"""
         )
 
 
@@ -72,18 +75,16 @@ def test_modify_column():
     ret1 = Migrate.ddl.modify_column(User, User._meta.fields_map.get("is_active").describe(False))
     if isinstance(Migrate.ddl, MysqlDDL):
         assert ret0 == "ALTER TABLE `category` MODIFY COLUMN `name` VARCHAR(200)"
+        assert (
+            ret1
+            == "ALTER TABLE `user` MODIFY COLUMN `is_active` BOOL NOT NULL  COMMENT 'Is Active' DEFAULT 1"
+        )
     elif isinstance(Migrate.ddl, PostgresDDL):
         assert (
             ret0
             == 'ALTER TABLE "category" ALTER COLUMN "name" TYPE VARCHAR(200) USING "name"::VARCHAR(200)'
         )
 
-    if isinstance(Migrate.ddl, MysqlDDL):
-        assert (
-            ret1
-            == "ALTER TABLE `user` MODIFY COLUMN `is_active` BOOL NOT NULL  COMMENT 'Is Active' DEFAULT 1"
-        )
-    elif isinstance(Migrate.ddl, PostgresDDL):
         assert (
             ret1 == 'ALTER TABLE "user" ALTER COLUMN "is_active" TYPE BOOL USING "is_active"::BOOL'
         )
@@ -153,9 +154,7 @@ def test_add_index():
     index_u = Migrate.ddl.add_index(Category, ["name"], True)
     if isinstance(Migrate.ddl, MysqlDDL):
         assert index == "ALTER TABLE `category` ADD INDEX `idx_category_name_8b0cb9` (`name`)"
-        assert (
+        assert index_u == "ALTER TABLE `category` ADD UNIQUE INDEX `name` (`name`)"
-            index_u == "ALTER TABLE `category` ADD UNIQUE INDEX `uid_category_name_8b0cb9` (`name`)"
-        )
     elif isinstance(Migrate.ddl, PostgresDDL):
         assert index == 'CREATE INDEX "idx_category_name_8b0cb9" ON "category" ("name")'
         assert index_u == 'CREATE UNIQUE INDEX "uid_category_name_8b0cb9" ON "category" ("name")'
@@ -171,7 +170,7 @@ def test_drop_index():
     ret_u = Migrate.ddl.drop_index(Category, ["name"], True)
     if isinstance(Migrate.ddl, MysqlDDL):
         assert ret == "ALTER TABLE `category` DROP INDEX `idx_category_name_8b0cb9`"
-        assert ret_u == "ALTER TABLE `category` DROP INDEX `uid_category_name_8b0cb9`"
+        assert ret_u == "ALTER TABLE `category` DROP INDEX `name`"
     elif isinstance(Migrate.ddl, PostgresDDL):
         assert ret == 'DROP INDEX "idx_category_name_8b0cb9"'
         assert ret_u == 'DROP INDEX "uid_category_name_8b0cb9"'

tests/test_migrate.py

@@ -1,11 +1,15 @@
+from pathlib import Path
+from typing import List, cast
+
 import pytest
+import tortoise
 from pytest_mock import MockerFixture
 
 from aerich.ddl.mysql import MysqlDDL
 from aerich.ddl.postgres import PostgresDDL
 from aerich.ddl.sqlite import SqliteDDL
 from aerich.exceptions import NotSupportError
-from aerich.migrate import Migrate
+from aerich.migrate import MIGRATE_TEMPLATE, Migrate
 from aerich.utils import get_models_describe
 
 old_models_describe = {
@@ -100,6 +104,21 @@ old_models_describe = {
                 "constraints": {"ge": 1, "le": 2147483647},
                 "db_field_types": {"": "INT"},
             },
+            {
+                "name": "title",
+                "field_type": "CharField",
+                "db_column": "title",
+                "python_type": "str",
+                "generated": False,
+                "nullable": False,
+                "unique": True,
+                "indexed": True,
+                "default": None,
+                "description": None,
+                "docstring": None,
+                "constraints": {"max_length": 20},
+                "db_field_types": {"": "VARCHAR(20)"},
+            },
         ],
         "fk_fields": [
             {
@@ -169,6 +188,21 @@ old_models_describe = {
             "db_field_types": {"": "INT"},
         },
         "data_fields": [
+            {
+                "name": "name",
+                "field_type": "CharField",
+                "db_column": "name",
+                "python_type": "str",
+                "generated": False,
+                "nullable": False,
+                "unique": True,
+                "indexed": True,
+                "default": None,
+                "description": None,
+                "docstring": None,
+                "constraints": {"max_length": 100},
+                "db_field_types": {"": "VARCHAR(100)"},
+            },
             {
                 "name": "label",
                 "field_type": "CharField",
@@ -369,6 +403,21 @@ old_models_describe = {
                 "constraints": {"max_length": 50},
                 "db_field_types": {"": "VARCHAR(50)"},
             },
+            {
+                "name": "uid",
+                "field_type": "IntField",
+                "db_column": "uuid",
+                "python_type": "int",
+                "generated": False,
+                "nullable": False,
+                "unique": True,
+                "indexed": True,
+                "default": None,
+                "description": None,
+                "docstring": None,
+                "constraints": {"ge": -2147483648, "le": 2147483647},
+                "db_field_types": {"": "INT"},
+            },
             {
                 "name": "view_num",
                 "field_type": "IntField",
@@ -644,6 +693,21 @@ old_models_describe = {
                 "constraints": {},
                 "db_field_types": {"": "TEXT", "mysql": "LONGTEXT"},
             },
+            {
+                "name": "longitude",
+                "unique": False,
+                "default": None,
+                "indexed": False,
+                "nullable": False,
+                "db_column": "longitude",
+                "docstring": None,
+                "generated": False,
+                "field_type": "DecimalField",
+                "constraints": {},
+                "description": None,
+                "python_type": "decimal.Decimal",
+                "db_field_types": {"": "DECIMAL(12,9)", "sqlite": "VARCHAR(40)"},
+            },
         ],
         "fk_fields": [],
         "backward_fk_fields": [
@@ -758,6 +822,16 @@ old_models_describe = {
 }
 
 
+def should_add_user_id_column_type_alter_sql() -> bool:
+    if tortoise.__version__ < "0.21":
+        return False
+    # tortoise-orm>=0.21 changes IntField constraints
+    # from {"ge": 1, "le": 2147483647} to {"ge": -2147483648,"le": 2147483647}
+    data_fields = cast(List[dict], old_models_describe["models.Category"]["data_fields"])
+    user_id_constraints = data_fields[-1]["constraints"]
+    return tortoise.fields.data.IntField.constraints != user_id_constraints
+
+
 def test_migrate(mocker: MockerFixture):
     """
     models.py diff with old_models.py
@@ -768,53 +842,73 @@ def test_migrate(mocker: MockerFixture):
     - drop field: User.avatar
     - add index: Email.email
     - add many to many: Email.users
-    - remove unique: User.username
+    - remove unique: Category.title
+    - add unique: User.username
     - change column: length User.password
     - add unique_together: (name,type) of Product
+    - drop unique field: Config.name
     - alter default: Config.status
     - rename column: Product.image -> Product.pic
     """
-    mocker.patch("click.prompt", side_effect=(True,))
+    mocker.patch("asyncclick.prompt", side_effect=(True,))
 
     models_describe = get_models_describe("models")
     Migrate.app = "models"
     if isinstance(Migrate.ddl, SqliteDDL):
         with pytest.raises(NotSupportError):
             Migrate.diff_models(old_models_describe, models_describe)
+        Migrate.upgrade_operators.clear()
+        with pytest.raises(NotSupportError):
             Migrate.diff_models(models_describe, old_models_describe, False)
+        Migrate.downgrade_operators.clear()
     else:
         Migrate.diff_models(old_models_describe, models_describe)
         Migrate.diff_models(models_describe, old_models_describe, False)
         Migrate._merge_operators()
     if isinstance(Migrate.ddl, MysqlDDL):
-        assert sorted(Migrate.upgrade_operators) == sorted(
+        expected_upgrade_operators = {
-            [
             "ALTER TABLE `category` MODIFY COLUMN `name` VARCHAR(200)",
             "ALTER TABLE `category` MODIFY COLUMN `slug` VARCHAR(100) NOT NULL",
+            "ALTER TABLE `category` DROP INDEX `title`",
+            "ALTER TABLE `config` DROP COLUMN `name`",
+            "ALTER TABLE `config` DROP INDEX `name`",
             "ALTER TABLE `config` ADD `user_id` INT NOT NULL  COMMENT 'User'",
             "ALTER TABLE `config` ADD CONSTRAINT `fk_config_user_17daa970` FOREIGN KEY (`user_id`) REFERENCES `user` (`id`) ON DELETE CASCADE",
             "ALTER TABLE `config` ALTER COLUMN `status` DROP DEFAULT",
+            "ALTER TABLE `config` MODIFY COLUMN `value` JSON NOT NULL",
             "ALTER TABLE `email` ADD `address` VARCHAR(200) NOT NULL",
             "ALTER TABLE `email` DROP COLUMN `user_id`",
             "ALTER TABLE `configs` RENAME TO `config`",
+            "ALTER TABLE `product` DROP COLUMN `uuid`",
+            "ALTER TABLE `product` DROP INDEX `uuid`",
             "ALTER TABLE `product` RENAME COLUMN `image` TO `pic`",
             "ALTER TABLE `email` RENAME COLUMN `id` TO `email_id`",
             "ALTER TABLE `product` ADD INDEX `idx_product_name_869427` (`name`, `type_db_alias`)",
             "ALTER TABLE `email` ADD INDEX `idx_email_email_4a1a33` (`email`)",
             "ALTER TABLE `product` ADD UNIQUE INDEX `uid_product_name_869427` (`name`, `type_db_alias`)",
             "ALTER TABLE `product` ALTER COLUMN `view_num` SET DEFAULT 0",
+            "ALTER TABLE `product` MODIFY COLUMN `created_at` DATETIME(6) NOT NULL  DEFAULT CURRENT_TIMESTAMP(6)",
+            "ALTER TABLE `product` MODIFY COLUMN `is_reviewed` BOOL NOT NULL  COMMENT 'Is Reviewed'",
             "ALTER TABLE `user` DROP COLUMN `avatar`",
             "ALTER TABLE `user` MODIFY COLUMN `password` VARCHAR(100) NOT NULL",
-                "CREATE TABLE IF NOT EXISTS `newmodel` (\n    `id` INT NOT NULL PRIMARY KEY AUTO_INCREMENT,\n    `name` VARCHAR(50) NOT NULL\n) CHARACTER SET utf8mb4;",
+            "ALTER TABLE `user` MODIFY COLUMN `intro` LONGTEXT NOT NULL",
-                "ALTER TABLE `user` ADD UNIQUE INDEX `uid_user_usernam_9987ab` (`username`)",
+            "ALTER TABLE `user` MODIFY COLUMN `last_login` DATETIME(6) NOT NULL  COMMENT 'Last Login'",
-                "CREATE TABLE `email_user` (`email_id` INT NOT NULL REFERENCES `email` (`email_id`) ON DELETE CASCADE,`user_id` INT NOT NULL REFERENCES `user` (`id`) ON DELETE CASCADE) CHARACTER SET utf8mb4",
+            "ALTER TABLE `user` MODIFY COLUMN `is_active` BOOL NOT NULL  COMMENT 'Is Active' DEFAULT 1",
-            ]
+            "ALTER TABLE `user` MODIFY COLUMN `is_superuser` BOOL NOT NULL  COMMENT 'Is SuperUser' DEFAULT 0",
-        )
+            "ALTER TABLE `user` MODIFY COLUMN `longitude` DECIMAL(10,8) NOT NULL",
-
+            "ALTER TABLE `user` ADD UNIQUE INDEX `username` (`username`)",
-        assert sorted(Migrate.downgrade_operators) == sorted(
+            "CREATE TABLE `email_user` (\n    `email_id` INT NOT NULL REFERENCES `email` (`email_id`) ON DELETE CASCADE,\n    `user_id` INT NOT NULL REFERENCES `user` (`id`) ON DELETE CASCADE\n) CHARACTER SET utf8mb4",
-            [
+            "CREATE TABLE IF NOT EXISTS `newmodel` (\n    `id` INT NOT NULL PRIMARY KEY AUTO_INCREMENT,\n    `name` VARCHAR(50) NOT NULL\n) CHARACTER SET utf8mb4",
+            "ALTER TABLE `category` MODIFY COLUMN `created_at` DATETIME(6) NOT NULL  DEFAULT CURRENT_TIMESTAMP(6)",
+            "ALTER TABLE `product` MODIFY COLUMN `body` LONGTEXT NOT NULL",
+            "ALTER TABLE `email` MODIFY COLUMN `is_primary` BOOL NOT NULL  DEFAULT 0",
+        }
+        expected_downgrade_operators = {
             "ALTER TABLE `category` MODIFY COLUMN `name` VARCHAR(200) NOT NULL",
             "ALTER TABLE `category` MODIFY COLUMN `slug` VARCHAR(200) NOT NULL",
+            "ALTER TABLE `category` ADD UNIQUE INDEX `title` (`title`)",
+            "ALTER TABLE `config` ADD `name` VARCHAR(100) NOT NULL UNIQUE",
+            "ALTER TABLE `config` ADD UNIQUE INDEX `name` (`name`)",
             "ALTER TABLE `config` DROP COLUMN `user_id`",
             "ALTER TABLE `config` DROP FOREIGN KEY `fk_config_user_17daa970`",
             "ALTER TABLE `config` ALTER COLUMN `status` SET DEFAULT 1",
@@ -823,68 +917,125 @@ def test_migrate(mocker: MockerFixture):
             "ALTER TABLE `config` RENAME TO `configs`",
             "ALTER TABLE `product` RENAME COLUMN `pic` TO `image`",
             "ALTER TABLE `email` RENAME COLUMN `email_id` TO `id`",
+            "ALTER TABLE `product` ADD `uuid` INT NOT NULL UNIQUE",
+            "ALTER TABLE `product` ADD UNIQUE INDEX `uuid` (`uuid`)",
             "ALTER TABLE `product` DROP INDEX `idx_product_name_869427`",
             "ALTER TABLE `email` DROP INDEX `idx_email_email_4a1a33`",
             "ALTER TABLE `product` DROP INDEX `uid_product_name_869427`",
             "ALTER TABLE `product` ALTER COLUMN `view_num` DROP DEFAULT",
             "ALTER TABLE `user` ADD `avatar` VARCHAR(200) NOT NULL  DEFAULT ''",
-                "ALTER TABLE `user` DROP INDEX `idx_user_usernam_9987ab`",
+            "ALTER TABLE `user` DROP INDEX `username`",
             "ALTER TABLE `user` MODIFY COLUMN `password` VARCHAR(200) NOT NULL",
             "DROP TABLE IF EXISTS `email_user`",
             "DROP TABLE IF EXISTS `newmodel`",
-            ]
+            "ALTER TABLE `user` MODIFY COLUMN `intro` LONGTEXT NOT NULL",
+            "ALTER TABLE `config` MODIFY COLUMN `value` TEXT NOT NULL",
+            "ALTER TABLE `category` MODIFY COLUMN `created_at` DATETIME(6) NOT NULL  DEFAULT CURRENT_TIMESTAMP(6)",
+            "ALTER TABLE `product` MODIFY COLUMN `created_at` DATETIME(6) NOT NULL  DEFAULT CURRENT_TIMESTAMP(6)",
+            "ALTER TABLE `product` MODIFY COLUMN `is_reviewed` BOOL NOT NULL  COMMENT 'Is Reviewed'",
+            "ALTER TABLE `user` MODIFY COLUMN `last_login` DATETIME(6) NOT NULL  COMMENT 'Last Login'",
+            "ALTER TABLE `user` MODIFY COLUMN `is_active` BOOL NOT NULL  COMMENT 'Is Active' DEFAULT 1",
+            "ALTER TABLE `user` MODIFY COLUMN `is_superuser` BOOL NOT NULL  COMMENT 'Is SuperUser' DEFAULT 0",
+            "ALTER TABLE `user` MODIFY COLUMN `longitude` DECIMAL(12,9) NOT NULL",
+            "ALTER TABLE `product` MODIFY COLUMN `body` LONGTEXT NOT NULL",
+            "ALTER TABLE `email` MODIFY COLUMN `is_primary` BOOL NOT NULL  DEFAULT 0",
+        }
+        if should_add_user_id_column_type_alter_sql():
+            sql = "ALTER TABLE `category` MODIFY COLUMN `user_id` INT NOT NULL  COMMENT 'User'"
+            expected_upgrade_operators.add(sql)
+            expected_downgrade_operators.add(sql)
+        assert not set(Migrate.upgrade_operators).symmetric_difference(expected_upgrade_operators)
+
+        assert not set(Migrate.downgrade_operators).symmetric_difference(
+            expected_downgrade_operators
         )
 
     elif isinstance(Migrate.ddl, PostgresDDL):
-        assert sorted(Migrate.upgrade_operators) == sorted(
+        expected_upgrade_operators = {
-            [
+            'DROP INDEX "uid_category_title_f7fc03"',
             'ALTER TABLE "category" ALTER COLUMN "name" DROP NOT NULL',
             'ALTER TABLE "category" ALTER COLUMN "slug" TYPE VARCHAR(100) USING "slug"::VARCHAR(100)',
+            'ALTER TABLE "category" ALTER COLUMN "created_at" TYPE TIMESTAMPTZ USING "created_at"::TIMESTAMPTZ',
+            'ALTER TABLE "config" DROP COLUMN "name"',
+            'DROP INDEX "uid_config_name_2c83c8"',
             'ALTER TABLE "config" ADD "user_id" INT NOT NULL',
             'ALTER TABLE "config" ADD CONSTRAINT "fk_config_user_17daa970" FOREIGN KEY ("user_id") REFERENCES "user" ("id") ON DELETE CASCADE',
             'ALTER TABLE "config" ALTER COLUMN "status" DROP DEFAULT',
+            'ALTER TABLE "config" ALTER COLUMN "value" TYPE JSONB USING "value"::JSONB',
             'ALTER TABLE "configs" RENAME TO "config"',
             'ALTER TABLE "email" ADD "address" VARCHAR(200) NOT NULL',
             'ALTER TABLE "email" DROP COLUMN "user_id"',
             'ALTER TABLE "email" RENAME COLUMN "id" TO "email_id"',
+            'ALTER TABLE "email" ALTER COLUMN "is_primary" TYPE BOOL USING "is_primary"::BOOL',
+            'DROP INDEX "uid_product_uuid_d33c18"',
+            'ALTER TABLE "product" DROP COLUMN "uuid"',
             'ALTER TABLE "product" ALTER COLUMN "view_num" SET DEFAULT 0',
             'ALTER TABLE "product" RENAME COLUMN "image" TO "pic"',
+            'ALTER TABLE "product" ALTER COLUMN "is_reviewed" TYPE BOOL USING "is_reviewed"::BOOL',
+            'ALTER TABLE "product" ALTER COLUMN "body" TYPE TEXT USING "body"::TEXT',
+            'ALTER TABLE "product" ALTER COLUMN "created_at" TYPE TIMESTAMPTZ USING "created_at"::TIMESTAMPTZ',
             'ALTER TABLE "user" ALTER COLUMN "password" TYPE VARCHAR(100) USING "password"::VARCHAR(100)',
             'ALTER TABLE "user" DROP COLUMN "avatar"',
+            'ALTER TABLE "user" ALTER COLUMN "is_superuser" TYPE BOOL USING "is_superuser"::BOOL',
+            'ALTER TABLE "user" ALTER COLUMN "last_login" TYPE TIMESTAMPTZ USING "last_login"::TIMESTAMPTZ',
+            'ALTER TABLE "user" ALTER COLUMN "intro" TYPE TEXT USING "intro"::TEXT',
+            'ALTER TABLE "user" ALTER COLUMN "is_active" TYPE BOOL USING "is_active"::BOOL',
+            'ALTER TABLE "user" ALTER COLUMN "longitude" TYPE DECIMAL(10,8) USING "longitude"::DECIMAL(10,8)',
             'CREATE INDEX "idx_product_name_869427" ON "product" ("name", "type_db_alias")',
             'CREATE INDEX "idx_email_email_4a1a33" ON "email" ("email")',
-                'CREATE TABLE "email_user" ("email_id" INT NOT NULL REFERENCES "email" ("email_id") ON DELETE CASCADE,"user_id" INT NOT NULL REFERENCES "user" ("id") ON DELETE CASCADE)',
+            'CREATE TABLE "email_user" (\n    "email_id" INT NOT NULL REFERENCES "email" ("email_id") ON DELETE CASCADE,\n    "user_id" INT NOT NULL REFERENCES "user" ("id") ON DELETE CASCADE\n)',
-                'CREATE TABLE IF NOT EXISTS "newmodel" (\n    "id" SERIAL NOT NULL PRIMARY KEY,\n    "name" VARCHAR(50) NOT NULL\n);\nCOMMENT ON COLUMN "config"."user_id" IS \'User\';',
+            'CREATE TABLE IF NOT EXISTS "newmodel" (\n    "id" SERIAL NOT NULL PRIMARY KEY,\n    "name" VARCHAR(50) NOT NULL\n);\nCOMMENT ON COLUMN "config"."user_id" IS \'User\'',
             'CREATE UNIQUE INDEX "uid_product_name_869427" ON "product" ("name", "type_db_alias")',
             'CREATE UNIQUE INDEX "uid_user_usernam_9987ab" ON "user" ("username")',
-            ]
+        }
-        )
+        expected_downgrade_operators = {
-        assert sorted(Migrate.downgrade_operators) == sorted(
+            'CREATE UNIQUE INDEX "uid_category_title_f7fc03" ON "category" ("title")',
-            [
             'ALTER TABLE "category" ALTER COLUMN "name" SET NOT NULL',
             'ALTER TABLE "category" ALTER COLUMN "slug" TYPE VARCHAR(200) USING "slug"::VARCHAR(200)',
+            'ALTER TABLE "category" ALTER COLUMN "created_at" TYPE TIMESTAMPTZ USING "created_at"::TIMESTAMPTZ',
+            'ALTER TABLE "config" ADD "name" VARCHAR(100) NOT NULL UNIQUE',
+            'CREATE UNIQUE INDEX "uid_config_name_2c83c8" ON "config" ("name")',
             'ALTER TABLE "config" ALTER COLUMN "status" SET DEFAULT 1',
             'ALTER TABLE "config" DROP COLUMN "user_id"',
             'ALTER TABLE "config" DROP CONSTRAINT "fk_config_user_17daa970"',
             'ALTER TABLE "config" RENAME TO "configs"',
+            'ALTER TABLE "config" ALTER COLUMN "value" TYPE JSONB USING "value"::JSONB',
             'ALTER TABLE "email" ADD "user_id" INT NOT NULL',
             'ALTER TABLE "email" DROP COLUMN "address"',
             'ALTER TABLE "email" RENAME COLUMN "email_id" TO "id"',
+            'ALTER TABLE "email" ALTER COLUMN "is_primary" TYPE BOOL USING "is_primary"::BOOL',
+            'ALTER TABLE "product" ADD "uuid" INT NOT NULL UNIQUE',
+            'CREATE UNIQUE INDEX "uid_product_uuid_d33c18" ON "product" ("uuid")',
             'ALTER TABLE "product" ALTER COLUMN "view_num" DROP DEFAULT',
             'ALTER TABLE "product" RENAME COLUMN "pic" TO "image"',
             'ALTER TABLE "user" ADD "avatar" VARCHAR(200) NOT NULL  DEFAULT \'\'',
             'ALTER TABLE "user" ALTER COLUMN "password" TYPE VARCHAR(200) USING "password"::VARCHAR(200)',
+            'ALTER TABLE "user" ALTER COLUMN "last_login" TYPE TIMESTAMPTZ USING "last_login"::TIMESTAMPTZ',
+            'ALTER TABLE "user" ALTER COLUMN "is_superuser" TYPE BOOL USING "is_superuser"::BOOL',
+            'ALTER TABLE "user" ALTER COLUMN "is_active" TYPE BOOL USING "is_active"::BOOL',
+            'ALTER TABLE "user" ALTER COLUMN "intro" TYPE TEXT USING "intro"::TEXT',
+            'ALTER TABLE "user" ALTER COLUMN "longitude" TYPE DECIMAL(12,9) USING "longitude"::DECIMAL(12,9)',
+            'ALTER TABLE "product" ALTER COLUMN "created_at" TYPE TIMESTAMPTZ USING "created_at"::TIMESTAMPTZ',
+            'ALTER TABLE "product" ALTER COLUMN "is_reviewed" TYPE BOOL USING "is_reviewed"::BOOL',
+            'ALTER TABLE "product" ALTER COLUMN "body" TYPE TEXT USING "body"::TEXT',
             'DROP INDEX "idx_product_name_869427"',
             'DROP INDEX "idx_email_email_4a1a33"',
-                'DROP INDEX "idx_user_usernam_9987ab"',
+            'DROP INDEX "uid_user_usernam_9987ab"',
             'DROP INDEX "uid_product_name_869427"',
             'DROP TABLE IF EXISTS "email_user"',
             'DROP TABLE IF EXISTS "newmodel"',
-            ]
+        }
+        if should_add_user_id_column_type_alter_sql():
+            sql = 'ALTER TABLE "category" ALTER COLUMN "user_id" TYPE INT USING "user_id"::INT'
+            expected_upgrade_operators.add(sql)
+            expected_downgrade_operators.add(sql)
+        assert not set(Migrate.upgrade_operators).symmetric_difference(expected_upgrade_operators)
+        assert not set(Migrate.downgrade_operators).symmetric_difference(
+            expected_downgrade_operators
         )
+
     elif isinstance(Migrate.ddl, SqliteDDL):
         assert Migrate.upgrade_operators == []
-
         assert Migrate.downgrade_operators == []
 
 
@@ -892,18 +1043,30 @@ def test_sort_all_version_files(mocker):
     mocker.patch(
         "os.listdir",
         return_value=[
-            "1_datetime_update.sql",
+            "1_datetime_update.py",
-            "11_datetime_update.sql",
+            "11_datetime_update.py",
-            "10_datetime_update.sql",
+            "10_datetime_update.py",
-            "2_datetime_update.sql",
+            "2_datetime_update.py",
         ],
     )
 
     Migrate.migrate_location = "."
 
     assert Migrate.get_all_version_files() == [
-        "1_datetime_update.sql",
+        "1_datetime_update.py",
-        "2_datetime_update.sql",
+        "2_datetime_update.py",
-        "10_datetime_update.sql",
+        "10_datetime_update.py",
-        "11_datetime_update.sql",
+        "11_datetime_update.py",
     ]
+
+
+async def test_empty_migration(mocker, tmp_path: Path) -> None:
+    mocker.patch("os.listdir", return_value=[])
+    Migrate.app = "foo"
+    expected_content = MIGRATE_TEMPLATE.format(upgrade_sql="", downgrade_sql="")
+    Migrate.migrate_location = tmp_path
+
+    migration_file = await Migrate.migrate("update", True)
+
+    f = tmp_path / migration_file
+    assert f.read_text() == expected_content

tests/test_utils.py

@@ -0,0 +1,6 @@
+from aerich.utils import import_py_file
+
+
+def test_import_py_file() -> None:
+    m = import_py_file("aerich/utils.py")
+    assert getattr(m, "import_py_file")