Amalan Terbaik untuk Menggunakan Pydantic dalam Python

Pydantic ialah perpustakaan Python yang memudahkan pengesahan data menggunakan pembayang jenis. Ia memastikan integriti data dan menawarkan cara mudah untuk mencipta model data dengan semakan dan pengesahan jenis automatik.

Dalam aplikasi perisian, pengesahan data yang boleh dipercayai adalah penting untuk mengelakkan ralat, isu keselamatan dan tingkah laku yang tidak dapat diramalkan.

Panduan ini menyediakan amalan terbaik untuk menggunakan Pydantic dalam projek Python, meliputi definisi model, pengesahan data, pengendalian ralat dan pengoptimuman prestasi.

---

Memasang Pydantic

Untuk memasang Pydantic, gunakan pip, pemasang pakej Python, dengan arahan:

pip install pydantic

Arahan ini memasang Pydantic dan kebergantungannya.

Penggunaan Asas

Buat model Pydantic dengan membuat kelas yang diwarisi daripada BaseModel. Gunakan anotasi jenis Python untuk menentukan jenis setiap medan:

from pydantic import BaseModel

class User(BaseModel):
    id: int
    name: str
    email: str

Pydantic menyokong pelbagai jenis medan, termasuk int, str, float, bool, list dan dict. Anda juga boleh menentukan model bersarang dan jenis tersuai:

from typing import List, Optional
from pydantic import BaseModel

class Address(BaseModel):
    street: str
    city: str
    zip_code: Optional[str] = None

class User(BaseModel):
    id: int
    name: str
    email: str
    age: Optional[int] = None
    addresses: List[Address]

Setelah anda menentukan model Pydantic, buat kejadian dengan menyediakan data yang diperlukan. Pydantic akan mengesahkan data dan menimbulkan ralat jika mana-mana medan tidak memenuhi keperluan yang ditentukan:

user = User(
    id=1,
    name="John Doe",
    email="john.doe@example.com",
    addresses=[{"street": "123 Main St", "city": "Anytown", "zip_code": "12345"}]
)

print(user)

# Output:
# id=1 name='John Doe' email='john.doe@example.com' age=None addresses=[Address(street='123 Main St', city='Anytown', zip_code='12345')]

---

Mentakrifkan Model Pydantic

Model Pydantic menggunakan anotasi jenis Python untuk menentukan jenis medan data.

Mereka menyokong pelbagai jenis terbina dalam, termasuk:

• Jenis primitif: str, int, float, bool
• Jenis koleksi: senarai, tuple, set, dict
• Jenis pilihan: Pilihan daripada modul menaip untuk medan yang boleh Tiada
• Jenis kesatuan: Kesatuan daripada modul menaip untuk menentukan medan boleh menjadi salah satu daripada beberapa jenis

Contoh:

from typing import List, Dict, Optional, Union
from pydantic import BaseModel

class Item(BaseModel):
    name: str
    price: float
    tags: List[str]
    metadata: Dict[str, Union[str, int, float]]

class Order(BaseModel):
    order_id: int
    items: List[Item]
    discount: Optional[float] = None

Jenis Tersuai

Selain jenis terbina dalam, anda boleh mentakrifkan jenis tersuai menggunakan conint Pydantic, constr dan fungsi kekangan lain.

Ini membolehkan anda menambah peraturan pengesahan tambahan, seperti kekangan panjang pada rentetan atau julat nilai untuk integer.

Contoh:

from pydantic import BaseModel, conint, constr

class Product(BaseModel):
    name: constr(min_length=2, max_length=50)
    quantity: conint(gt=0, le=1000)
    price: float

product = Product(name="Laptop", quantity=5, price=999.99)

Medan Diperlukan lwn. Pilihan

Secara lalai, medan dalam model Pydantic diperlukan melainkan ditandakan secara eksplisit sebagai pilihan.

Jika medan yang diperlukan tiada semasa instansiasi model, Pydantic akan menimbulkan Ralat Pengesahan.

Contoh:

from pydantic import BaseModel

class User(BaseModel):
    id: int
    name: str
    email: str

user = User(id=1, name="John Doe")


# Output
#  Field required [type=missing, input_value={'id': 1, 'name': 'John Doe'}, input_type=dict]

Medan Pilihan dengan Nilai Lalai

Medan boleh dijadikan pilihan dengan menggunakan Pilihan daripada modul menaip dan memberikan nilai lalai.

Contoh:

from pydantic import BaseModel
from typing import Optional

class User(BaseModel):
    id: int
    name: str
    email: Optional[str] = None

user = User(id=1, name="John Doe")

Dalam contoh ini, e-mel adalah pilihan dan lalai kepada Tiada jika tidak disediakan.

Model Bersarang

Pydantic membenarkan model bersarang antara satu sama lain, membolehkan struktur data yang kompleks.

Model bersarang ditakrifkan sebagai medan model lain, memastikan integriti dan pengesahan data pada berbilang peringkat.

Contoh:

from pydantic import BaseModel
from typing import Optional, List


class Address(BaseModel):
    street: str
    city: str
    zip_code: Optional[str] = None

class User(BaseModel):
    id: int
    name: str
    email: str
    addresses: List[Address]

user = User(
    id=1,
    name="John Doe",
    email="john.doe@example.com",
    addresses=[{"street": "123 Main St", "city": "Anytown"}]
)

Amalan Terbaik untuk Mengurus Data Bersarang

Apabila bekerja dengan model bersarang, adalah penting untuk:

• Sahkan data pada setiap peringkat: Pastikan setiap model bersarang mempunyai peraturan dan kekangan pengesahan sendiri.
• Gunakan konvensyen penamaan yang jelas dan konsisten: Ini menjadikan struktur data anda lebih mudah dibaca dan diselenggara.
• Pastikan model mudah: Elakkan struktur bersarang yang terlalu kompleks. Jika model menjadi terlalu kompleks, pertimbangkan untuk memecahkannya kepada komponen yang lebih kecil dan lebih mudah diurus.

---

Pengesahan Data

Pydantic termasuk satu set pengesah terbina dalam yang mengendalikan tugas pengesahan data biasa secara automatik.

Pengesah ini termasuk:

• Pengesahan jenis: Memastikan medan sepadan dengan anotasi jenis yang ditentukan (cth., int, str, senarai).
• Pengesahan julat: Menguatkuasakan julat nilai dan panjang menggunakan kekangan seperti conint, constr, confloat.
• Pengesahan format: Menyemak format tertentu, seperti EmailStr untuk mengesahkan alamat e-mel.
• Pengesahan koleksi: Memastikan elemen dalam koleksi (cth., senarai, dict) mematuhi jenis dan kekangan yang ditentukan.

Pengesah ini memudahkan proses memastikan integriti dan pematuhan data dalam model anda.

Berikut ialah beberapa contoh yang menunjukkan pengesah terbina dalam:

daripada import pydantic BaseModel, EmailStr, conint, constr

class User(BaseModel):
    id: conint(gt=0)  # id must be greater than 0
    name: constr(min_length=2, max_length=50)  # name must be between 2 and 50 characters
    email: EmailStr  # email must be a valid email address
    age: conint(ge=18)  # age must be 18 or older

user = User(id=1, name="John Doe", email="john.doe@example.com", age=25)

Dalam contoh ini, model Pengguna menggunakan pengesah terbina dalam untuk memastikan id lebih besar daripada 0, nama antara 2 dan 50 aksara, e-mel ialah alamat e-mel yang sah dan umur 18 tahun ke atas.
Untuk dapat menggunakan pengesah e-mel, anda perlu memasang sambungan kepada pydantic:

pip install pydantic[email]

Custom Validators

Pydantic allows you to define custom validators for more complex validation logic.

Custom validators are defined using the @field_validator decorator within your model class.

Example of a custom validator:

from pydantic import BaseModel, field_validator


class Product(BaseModel):
    name: str
    price: float

    @field_validator('price')
    def price_must_be_positive(cls, value):
        if value <= 0:
            raise ValueError('Price must be positive')
        return value

product = Product(name="Laptop", price=999.99)

Here, the price_must_be_positive validator ensures that the price field is a positive number.

Custom validators are registered automatically when you define them within a model using the @field_validator decorator. Validators can be applied to individual fields or across multiple fields.

Example of registering a validator for multiple fields:

from pydantic import BaseModel, field_validator


class Person(BaseModel):
    first_name: str
    last_name: str

    @field_validator('first_name', 'last_name')
    def names_cannot_be_empty(cls, value):
        if not value:
            raise ValueError('Name fields cannot be empty')
        return value

person = Person(first_name="John", last_name="Doe")

In this example, the names_cannot_be_empty validator ensures that both the first_name and last_name fields are not empty.

Using Config Classes

Pydantic models can be customized using an inner Config class.

This class allows you to set various configuration options that affect the model's behavior, such as validation rules, JSON serialization, and more.

Example of a Config class:

from pydantic import BaseModel

class User(BaseModel):
    id: int
    name: str
    email: str

    class Config:
        str_strip_whitespace = True  # Strip whitespace from strings
        str_min_length = 1  # Minimum length for any string field

user = User(id=1, name="  John Doe  ", email="john.doe@example.com")

print(user)

# Output:
# id=1 name='John Doe' email='john.doe@example.com'

In this example, the Config class is used to strip whitespace from string fields and enforce a minimum length of 1 for any string field.

Some common configuration options in Pydantic's Config class include:

• str_strip_whitespace: Automatically strip leading and trailing whitespace from string fields.
• str_min_length: Set a minimum length for any string field.
• validate_default: Validate all fields, even those with default values.
• validate_assignment: Enable validation on assignment to model attributes.
• use_enum_values: Use the values of enums directly instead of the enum instances.
• json_encoders: Define custom JSON encoders for specific types.

---

Error Handling

When Pydantic finds data that doesn't conform to the model's schema, it raises a ValidationError.

This error provides detailed information about the issue, including the field name, the incorrect value, and a description of the problem.

Here's an example of how default error messages are structured:

from pydantic import BaseModel, ValidationError, EmailStr

class User(BaseModel):
    id: int
    name: str
    email: EmailStr

try:
    user = User(id='one', name='John Doe', email='invalid-email')
except ValidationError as e:
    print(e.json())

# Output:
# [{"type":"int_parsing","loc":["id"],"msg":"Input should be a valid integer, unable to parse string as an integer","input":"one","url":"https://errors.pydantic.dev/2.8/v/int_parsing"},{"type":"value_error","loc":["email"],"msg":"value is not a valid email address: An email address must have an @-sign.","input":"invalid-email","ctx":{"reason":"An email address must have an @-sign."},"url":"https://errors.pydantic.dev/2.8/v/value_error"}]

In this example, the error message will indicate that id must be an integer and email must be a valid email address.

Customizing Error Messages

Pydantic allows you to customize error messages for specific fields by raising exceptions with custom messages in validators or by setting custom configurations.

Here’s an example of customizing error messages:

from pydantic import BaseModel, ValidationError, field_validator

class Product(BaseModel):
    name: str
    price: float

    @field_validator('price')
    def price_must_be_positive(cls, value):
        if value <= 0:
            raise ValueError('Price must be a positive number')
        return value

try:
    product = Product(name='Laptop', price=-1000)
except ValidationError as e:
    print(e.json())

# Output:
# [{"type":"value_error","loc":["price"],"msg":"Value error, Price must be a positive number","input":-1000,"ctx":{"error":"Price must be a positive number"},"url":"https://errors.pydantic.dev/2.8/v/value_error"}]

In this example, the error message for price is customized to indicate that it must be a positive number.

Best Practices for Error Reporting

Effective error reporting involves providing clear, concise, and actionable feedback to users or developers.

Here are some best practices:

• Log errors: Use logging mechanisms to record validation errors for debugging and monitoring purposes.
• Return user-friendly messages: When exposing errors to end-users, avoid technical jargon. Instead, provide clear instructions on how to correct the data.
• Aggregate errors: When multiple fields are invalid, aggregate the errors into a single response to help users correct all issues at once.
• Use consistent formats: Ensure that error messages follow a consistent format across the application for easier processing and understanding.

Examples of best practices in error reporting:

from pydantic import BaseModel, ValidationError, EmailStr
import logging

logging.basicConfig(level=logging.INFO)

class User(BaseModel):
    id: int
    name: str
    email: EmailStr

def create_user(data):
    try:
        user = User(**data)
        return user
    except ValidationError as e:
        logging.error("Validation error: %s", e.json())
        return {"error": "Invalid data provided", "details": e.errors()}

user_data = {'id': 'one', 'name': 'John Doe', 'email': 'invalid-email'}
response = create_user(user_data)
print(response)

# Output:
# ERROR:root:Validation error: [{"type":"int_parsing","loc":["id"],"msg":"Input should be a valid integer, unable to parse string as an integer","input":"one","url":"https://errors.pydantic.dev/2.8/v/int_parsing"},{"type":"value_error","loc":["email"],"msg":"value is not a valid email address: An email address must have an @-sign.","input":"invalid-email","ctx":{"reason":"An email address must have an @-sign."},"url":"https://errors.pydantic.dev/2.8/v/value_error"}]
# {'error': 'Invalid data provided', 'details': [{'type': 'int_parsing', 'loc': ('id',), 'msg': 'Input should be a valid integer, unable to parse string as an integer', 'input': 'one', 'url': 'https://errors.pydantic.dev/2.8/v/int_parsing'}, {'type': 'value_error', 'loc': ('email',), 'msg': 'value is not a valid email address: An email address must have an @-sign.', 'input': 'invalid-email', 'ctx': {'reason': 'An email address must have an @-sign.'}}]}

In this example, validation errors are logged, and a user-friendly error message is returned, helping maintain application stability and providing useful feedback to the user.

---

Performance Considerations

Lazy initialization is a technique that postpones the creation of an object until it is needed.

In Pydantic, this can be useful for models with fields that are costly to compute or fetch. By delaying the initialization of these fields, you can reduce the initial load time and improve performance.

Example of lazy initialization:

from pydantic import BaseModel
from functools import lru_cache

class DataModel(BaseModel):
    name: str
    expensive_computation: str = None

    @property
    @lru_cache(maxsize=1)
    def expensive_computation(self):
        # Simulate an expensive computation
        result = "Computed Value"
        return result

data_model = DataModel(name="Test")
print(data_model.expensive_computation)

In this example, the expensive_computation field is computed only when accessed for the first time, reducing unnecessary computations during model initialization.

Redundant Validation

Pydantic models automatically validate data during initialization.

However, if you know that certain data has already been validated or if validation is not necessary in some contexts, you can disable validation to improve performance.

This can be done using the model_construct method, which bypasses validation:

Example of avoiding redundant validation:

from pydantic import BaseModel

class User(BaseModel):
    id: int
    name: str
    email: str

# Constructing a User instance without validation
data = {'id': 1, 'name': 'John Doe', 'email': 'john.doe@example.com'}
user = User.model_construct(**data)

In this example, User.model_construct is used to create a User instance without triggering validation, which can be useful in performance-critical sections of your code.

Efficient Data Parsing

When dealing with large datasets or high-throughput systems, efficiently parsing raw data becomes critical.

Pydantic provides the model_validate_json method, which can be used to parse JSON or other serialized data formats directly into Pydantic models.

Example of efficient data parsing:

from pydantic import BaseModel

class User(BaseModel):
    id: int
    name: str
    email: str

json_data = '{"id": 1, "name": "John Doe", "email": "john.doe@example.com"}'
user = User.model_validate_json(json_data)
print(user)

In this example, model_validate_json is used to parse JSON data into a User model directly, providing a more efficient way to handle serialized data.

Controlling Validation

Pydantic models can be configured to validate data only when necessary.

The validate_default and validate_assignment options in the Config class control when validation occurs, which can help improve performance:

• validate_default: When set to False, only fields that are set during initialization are validated.
• validate_assignment: When set to True, validation is performed on field assignment after the model is created.

Example configuration:

from pydantic import BaseModel

class User(BaseModel):
    id: int
    name: str
    email: str

    class Config:
        validate_default = False  # Only validate fields set during initialization
        validate_assignment = True  # Validate fields on assignment

user = User(id=1, name="John Doe", email="john.doe@example.com")
user.email = "new.email@example.com"  # This assignment will trigger validation

In this example, validate_default is set to False to avoid unnecessary validation during initialization, and validate_assignment is set to True to ensure that fields are validated when they are updated.

---

Settings Management

Pydantic's BaseSettings class is designed for managing application settings, supporting environment variable loading and type validation.

This helps in configuring applications for different environments (e.g., development, testing, production).

Consider this .env file:

database_url=db
secret_key=sk
debug=False

Example of using BaseSettings:

from pydantic_settings import BaseSettings

class Settings(BaseSettings):
    database_url: str
    secret_key: str
    debug: bool = False

    class Config:
        env_file = ".env"

settings = Settings()
print(settings.model_dump())

# Output:
# {'database_url': 'db', 'secret_key': 'sk', 'debug': False}

In this example, settings are loaded from environment variables, and the Config class specifies that variables can be loaded from a .env file.

For using BaseSettings you will need to install an additional package:

pip install pydantic-settings

Managing settings effectively involves a few best practices:

• Use environment variables: Store configuration values in environment variables to keep sensitive data out of your codebase.
• Provide defaults: Define sensible default values for configuration settings to ensure the application runs with minimal configuration.
• Separate environments: Use different configuration files or environment variables for different environments (e.g., .env.development, .env.production).
• Validate settings: Use Pydantic's validation features to ensure all settings are correctly typed and within acceptable ranges.

---

Common Pitfalls and How to Avoid Them

One common mistake when using Pydantic is misapplying type annotations, which can lead to validation errors or unexpected behavior.

Here are a few typical mistakes and their solutions:

• Misusing Union Types: Using Union incorrectly can complicate type validation and handling.
• Optional Fields without Default Values: Forgetting to provide a default value for optional fields can lead to None values causing errors in your application.
• Incorrect Type Annotations: Assigning incorrect types to fields can cause validation to fail. For example, using str for a field that should be an int.

Ignoring Performance Implications

Ignoring performance implications when using Pydantic can lead to slow applications, especially when dealing with large datasets or frequent model instantiations.

Here are some strategies to avoid performance bottlenecks:

• Leverage Configuration Options: Use Pydantic's configuration options like validate_default and validate_assignment to control when validation occurs.
• Optimize Nested Models: When working with nested models, ensure that you are not over-validating or duplicating validation logic.
• Use Efficient Parsing Methods: Utilize model_validate_json and model_validate for efficient data parsing.
• Avoid Unnecessary Validation: Use the model_construct method to create models without validation when the data is already known to be valid.

Overcomplicating Models

Overcomplicating Pydantic models can make them difficult to maintain and understand.

Here are some tips to keep models simple and maintainable:

• Dokumenkan Model Anda: Gunakan rentetan dokumen dan ulasan untuk menerangkan peraturan pengesahan kompleks atau logik perniagaan yang dibenamkan dalam model.
• Enkapsulasi Logik Dengan Sesuai: Simpan pengesahan dan logik perniagaan dalam kaedah model yang sesuai atau utiliti luaran untuk mengelakkan definisi model yang berselerak.
• Gunakan Warisan dengan Berhemat: Walaupun warisan boleh menggalakkan penggunaan semula kod, penggunaan yang berlebihan boleh menjadikan hierarki model kompleks dan lebih sukar untuk diikuti.
• Elakkan Bersarang Berlebihan: Model bersarang dalam mungkin sukar untuk diurus. Matlamat untuk tahap bersarang yang seimbang.

---

Kesimpulan

Dalam panduan ini, kami telah merangkumi pelbagai amalan terbaik untuk menggunakan Pydantic dengan berkesan dalam projek Python anda.

Kami bermula dengan asas untuk bermula dengan Pydantic, termasuk pemasangan, penggunaan asas dan menentukan model. Kami kemudiannya menyelidiki ciri lanjutan seperti jenis tersuai, penyiaran dan penyahsirilan serta pengurusan tetapan.

Pertimbangan prestasi utama, seperti mengoptimumkan permulaan model dan penghuraian data yang cekap, telah diserlahkan untuk memastikan aplikasi anda berjalan lancar.

Kami juga membincangkan masalah biasa, seperti menyalahgunakan anotasi jenis, mengabaikan implikasi prestasi dan model yang terlalu rumit serta menyediakan strategi untuk mengelakkannya.

Menggunakan amalan terbaik ini dalam projek dunia sebenar anda akan membantu anda memanfaatkan kuasa penuh Pydantic, menjadikan kod anda lebih teguh, boleh diselenggara dan berprestasi.

Atas ialah kandungan terperinci Amalan Terbaik untuk Menggunakan Pydantic dalam Python. Untuk maklumat lanjut, sila ikut artikel berkaitan lain di laman web China PHP!