pyright

Pyright - 快速 Python 类型检查器

---

progressive_disclosure: entry_point: summary: "Microsoft 推出的快速 Python 类型检查器，集成 VS Code 并支持严格模式" when_to_use: - "需要比 mypy 更快的类型检查时" - "使用 VS Code（Pylance）时" - "需要更严格的类型检查时" - "从 mypy 迁移时" quick_start: - "npm install -g pyright" - "创建 pyrightconfig.json" - "pyright ." token_estimate: entry: 65-80 full: 3500-4500

安装

Node.js（推荐）

# Global installation
npm install -g pyright

# Per-project installation
npm install --save-dev pyright

# Verify installation
pyright --version

VS Code（Pylance）

Pyright 为 Pylance 扩展提供支持：

# Install Pylance extension (includes pyright)
code --install-extension ms-python.vscode-pylance

pip 安装

# Community wrapper
pip install pyright

# Note: Still requires Node.js runtime

配置

pyrightconfig.json

{
  "include": [
    "src"
  ],
  "exclude": [
    "**/node_modules",
    "**/__pycache__",
    ".venv"
  ],
  "typeCheckingMode": "basic",
  "pythonVersion": "3.11",
  "pythonPlatform": "Linux",
  "reportMissingImports": true,
  "reportMissingTypeStubs": false,
  "strictListInference": true,
  "strictDictionaryInference": true,
  "strictSetInference": true
}

类型检查模式

基础模式（默认）：

{
  "typeCheckingMode": "basic",
  "reportUnusedImport": "warning",
  "reportUnusedVariable": "warning"
}

标准模式：

{
  "typeCheckingMode": "standard",
  "reportUnknownParameterType": "error",
  "reportUnknownArgumentType": "error",
  "reportUnknownVariableType": "error"
}

严格模式（最高类型安全性）：

{
  "typeCheckingMode": "strict",
  "reportPrivateUsage": "error",
  "reportConstantRedefinition": "error",
  "reportIncompatibleMethodOverride": "error",
  "reportIncompatibleVariableOverride": "error",
  "reportUnnecessaryIsInstance": "warning",
  "reportUnnecessaryCast": "warning"
}

单文件配置

# pyright: strict
"""Strict type checking for this file."""

# pyright: basic
"""Basic type checking."""

# pyright: reportGeneralTypeIssues=false
"""Disable specific diagnostics."""

VS Code 集成

settings.json

{
  "python.languageServer": "Pylance",
  "python.analysis.typeCheckingMode": "basic",
  "python.analysis.diagnosticMode": "workspace",
  "python.analysis.autoImportCompletions": true,
  "python.analysis.inlayHints.functionReturnTypes": true,
  "python.analysis.inlayHints.variableTypes": true,
  "python.analysis.completeFunctionParens": true
}

工作区配置

{
  "python.analysis.extraPaths": [
    "./src",
    "./lib"
  ],
  "python.analysis.stubPath": "./typings",
  "python.analysis.diagnosticSeverityOverrides": {
    "reportUnusedImport": "warning",
    "reportUnusedVariable": "warning",
    "reportGeneralTypeIssues": "error"
  }
}

类型检查功能

类型收窄

from typing import Union

def process(value: Union[str, int]) -> str:
    # Pyright narrows type based on isinstance
    if isinstance(value, str):
        return value.upper()  # value is str here
    else:
        return str(value)  # value is int here

# Type guards
from typing import TypeGuard

def is_str_list(val: list[object]) -> TypeGuard[list[str]]:
    return all(isinstance(x, str) for x in val)

def process_list(items: list[object]) -> None:
    if is_str_list(items):
        # items is list[str] here
        print(", ".join(items))

协议支持

from typing import Protocol

class Drawable(Protocol):
    def draw(self) -> None: ...

class Circle:
    def draw(self) -> None:
        print("Drawing circle")

def render(obj: Drawable) -> None:
    obj.draw()

# Works with structural typing
render(Circle())  # ✓ No explicit inheritance needed

TypedDict

from typing import TypedDict, NotRequired

class User(TypedDict):
    name: str
    age: int
    email: NotRequired[str]  # Optional in Python 3.11+

def create_user(data: User) -> None:
    print(data["name"])  # ✓ Type-safe
    # print(data["missing"])  # ✗ Error

user: User = {
    "name": "Alice",
    "age": 30
}  # ✓ email is optional

字面量类型

from typing import Literal

Mode = Literal["read", "write", "append"]

def open_file(mode: Mode) -> None:
    ...

open_file("read")    # ✓
open_file("delete")  # ✗ Error

高级功能

变型与泛型

from typing import TypeVar, Generic, Sequence

T_co = TypeVar("T_co", covariant=True)
T_contra = TypeVar("T_contra", contravariant=True)

class Reader(Generic[T_co]):
    def read(self) -> T_co: ...

class Writer(Generic[T_contra]):
    def write(self, item: T_contra) -> None: ...

# Covariance: Reader[Dog] is subtype of Reader[Animal]
# Contravariance: Writer[Animal] is subtype of Writer[Dog]

ParamSpec

from typing import ParamSpec, TypeVar, Callable

P = ParamSpec("P")
R = TypeVar("R")

def add_logging(f: Callable[P, R]) -> Callable[P, R]:
    def wrapper(*args: P.args, **kwargs: P.kwargs) -> R:
        print(f"Calling {f.__name__}")
        return f(*args, **kwargs)
    return wrapper

@add_logging
def greet(name: str, age: int) -> str:
    return f"Hello {name}, {age}"

# Type-safe: greet("Alice", 30)

类型别名

from typing import TypeAlias

# Simple alias
UserId: TypeAlias = int
Username: TypeAlias = str

# Generic alias
from collections.abc import Sequence
Vector: TypeAlias = Sequence[float]

# Complex alias
JSON: TypeAlias = dict[str, "JSON"] | list["JSON"] | str | int | float | bool | None

CI/CD 集成

GitHub Actions

name: Type Check

on: [push, pull_request]

jobs:
  typecheck:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4

      - name: Set up Node.js
        uses: actions/setup-node@v4
        with:
          node-version: '20'

      - name: Install pyright
        run: npm install -g pyright

      - name: Set up Python
        uses: actions/setup-python@v5
        with:
          python-version: '3.11'

      - name: Install dependencies
        run: |
          python -m pip install --upgrade pip
          pip install -r requirements.txt

      - name: Run pyright
        run: pyright

Pre-commit 钩子

# .pre-commit-config.yaml
repos:
  - repo: local
    hooks:
      - id: pyright
        name: pyright
        entry: pyright
        language: node
        types: [python]
        pass_filenames: false
        additional_dependencies: ['[email protected]']

Makefile

.PHONY: typecheck typecheck-strict

typecheck:
	pyright

typecheck-strict:
	pyright --level strict

typecheck-watch:
	pyright --watch

typecheck-stats:
	pyright --stats

从 mypy 迁移

配置映射

{
  "// mypy: disallow_untyped_defs": "reportUntypedFunctionDecorator",
  "// mypy: disallow_any_generics": "reportMissingTypeArgument",
  "// mypy: warn_return_any": "reportUnknownArgumentType",
  "// mypy: strict_equality": "reportUnnecessaryComparison",
  "// mypy: warn_unused_ignores": "reportUnnecessaryTypeIgnoreComment"
}

注释语法

# mypy: ignore-errors
# ↓ pyright equivalent
# pyright: reportGeneralTypeIssues=false

# type: ignore
# ↓ pyright equivalent
# pyright: ignore

# type: ignore[error-code]
# ↓ pyright equivalent
# pyright: ignore[reportGeneralTypeIssues]

渐进式迁移

{
  "typeCheckingMode": "basic",
  "include": ["src/new_module"],
  "exclude": ["src/legacy"],
  "reportMissingImports": true,
  "reportMissingTypeStubs": false
}

性能优化

基准性能

# Create performance baseline
pyright --stats --createstub

# Compare after changes
pyright --stats

监视模式

# Fast incremental checking
pyright --watch

# With specific path
pyright --watch src/

并行检查

{
  "executionEnvironments": [
    {
      "root": "src",
      "pythonVersion": "3.11"
    },
    {
      "root": "tests",
      "pythonVersion": "3.11",
      "extraPaths": ["src"]
    }
  ]
}

常见模式

Optional 处理

from typing import Optional

def get_user(user_id: int) -> Optional[str]:
    return "Alice" if user_id == 1 else None

# Before pyright 1.1.200
user = get_user(1)
if user is not None:
    print(user.upper())

# With pyright type narrowing
user = get_user(1)
if user:  # Narrows to str
    print(user.upper())

联合类型收窄

from typing import Union

def process(value: Union[str, list[str]]) -> str:
    if isinstance(value, list):
        return ", ".join(value)  # value is list[str]
    return value  # value is str

重载

from typing import overload, Literal

@overload
def open_file(path: str, mode: Literal["r"]) -> str: ...

@overload
def open_file(path: str, mode: Literal["rb"]) -> bytes: ...

def open_file(path: str, mode: str) -> str | bytes:
    if mode == "rb":
        return b"binary data"
    return "text data"

# Type-safe usage
text: str = open_file("file.txt", "r")
data: bytes = open_file("file.bin", "rb")

断言函数

from typing import Never

def assert_never(value: Never) -> Never:
    raise AssertionError(f"Unexpected value: {value}")

def handle_status(status: Literal["success", "error"]) -> None:
    if status == "success":
        print("OK")
    elif status == "error":
        print("Failed")
    else:
        assert_never(status)  # Exhaustiveness check

Pyright 与 mypy 对比

性能比较

# Typical project (10K lines)
# mypy: ~5-10 seconds
# pyright: ~1-2 seconds

# Large project (100K lines)
# mypy: ~60-120 seconds
# pyright: ~10-20 seconds

功能差异

Pyright 的优势：

• 类型检查速度快 5–10 倍
• 类型推断更出色
• 集成 VS Code（Pylance）
• 由 Microsoft 积极开发
• 协议支持更出色
• 类型收窄能力更强

mypy 的优势：

• 生态更成熟
• 插件系统
• 控制更精细
• 文档更完善
• 配置选项更多

适用 Pyright 的场景

• ✅ VS Code 用户
• ✅ 需要快速反馈
• ✅ 希望进行严格的类型检查
• ✅ 使用现代 Python（3.10+）
• ✅ 开始新项目

适用 mypy 的场景

• ✅ 已有 mypy 配置
• ✅ 需要 mypy 插件
• ✅ 使用非 VS Code 编辑器
• ✅ 使用旧版 Python（<3.8）
• ✅ 团队更偏好 mypy

最佳实践

从基础模式开始

{
  "typeCheckingMode": "basic",
  "reportMissingImports": true,
  "reportUndefinedVariable": true
}

逐步提高严格程度

{
  "typeCheckingMode": "standard",
  "reportUnknownParameterType": "warning",
  "reportUnknownArgumentType": "warning"
}

使用类型存根

# Generate stubs for third-party packages
pyright --createstub package_name

# Custom stubs directory
mkdir -p typings

善用内嵌提示

{
  "python.analysis.inlayHints.variableTypes": true,
  "python.analysis.inlayHints.functionReturnTypes": true,
  "python.analysis.inlayHints.callArgumentNames": true
}

类型覆盖率

# Check type completeness
pyright --stats

# Output:
# Files analyzed: 42
# Lines of code: 3,421
# Type completeness: 87.3%

有针对性地忽略

# Avoid broad ignores
# pyright: ignore  # ✗ Too broad

# Prefer specific ignores
# pyright: ignore[reportGeneralTypeIssues]  # ✓ Specific

# Or fix the issue
value: str = cast(str, unknown_value)  # ✓ Best

故障排除

导入解析

{
  "extraPaths": ["src", "lib"],
  "stubPath": "typings",
  "venvPath": ".",
  "venv": ".venv"
}

存根生成

# Generate stubs for package
pyright --createstub requests

# Custom stub location
pyright --createstub requests --outputdir typings

性能问题

{
  "exclude": [
    "**/node_modules",
    "**/__pycache__",
    ".venv",
    "build",
    "dist"
  ]
}

VS Code 未使用 Pyright

{
  "python.languageServer": "Pylance",
  "python.analysis.typeCheckingMode": "basic",
  "python.analysis.diagnosticMode": "workspace"
}

资源

官方资源：

• Pyright 文档
• Pylance 扩展
• 类型检查模式指南

社区资源：

• awesome-pyright
• Python 类型提示
• PEP 484 - 类型提示

---

相关技能：mypy、FastAPI、Django、pytest Token 数量：约 3,850 个 token