Python Type Hints: A Practical Guide to Type Annotations

# Without type hints - what does this function expect?
def process_data(data, threshold, output):
    ...
 
# With type hints - instantly clear
def process_data(data: list[float], threshold: float, output: str) -> dict[str, float]:
    ...

# Basic variable annotations
name: str = "Alice"
age: int = 30
height: float = 5.9
is_active: bool = True
raw_data: bytes = b"hello"
 
# Annotations without assignment (declaration only)
username: str
count: int

def greet(name: str) -> str:
    return f"Hello, {name}!"
 
def calculate_average(numbers: list[float]) -> float:
    return sum(numbers) / len(numbers)
 
def save_record(record: dict[str, str], overwrite: bool = False) -> None:
    """Functions that return nothing use -> None."""
    ...

def parse_config(path: str, encoding: str = "utf-8") -> dict[str, str]:
    config: dict[str, str] = {}
    with open(path, encoding=encoding) as f:
        for line in f:
            key, _, value = line.partition("=")
            config[key.strip()] = value.strip()
    return config

# Lists
scores: list[int] = [95, 87, 92]
names: list[str] = ["Alice", "Bob"]
 
# Dictionaries
user_ages: dict[str, int] = {"Alice": 30, "Bob": 25}
config: dict[str, list[str]] = {"servers": ["a.com", "b.com"]}
 
# Tuples - fixed length with specific types
point: tuple[float, float] = (3.14, 2.72)
record: tuple[str, int, bool] = ("Alice", 30, True)
 
# Variable-length tuples (all same type)
values: tuple[int, ...] = (1, 2, 3, 4, 5)
 
# Sets and frozensets
tags: set[str] = {"python", "typing"}
constants: frozenset[int] = frozenset({1, 2, 3})

from typing import List, Dict, Tuple, Set, FrozenSet
 
scores: List[int] = [95, 87, 92]
user_ages: Dict[str, int] = {"Alice": 30}
point: Tuple[float, float] = (3.14, 2.72)
tags: Set[str] = {"python", "typing"}

# Matrix: list of lists of floats
matrix: list[list[float]] = [[1.0, 2.0], [3.0, 4.0]]
 
# Mapping users to their list of scores
gradebook: dict[str, list[int]] = {
    "Alice": [95, 87, 92],
    "Bob": [78, 85, 90],
}
 
# Configuration: nested dictionaries
app_config: dict[str, dict[str, str | int]] = {
    "database": {"host": "localhost", "port": 5432},
    "cache": {"host": "redis.local", "port": 6379},
}

from typing import Optional
 
# Pre-3.10 syntax
def find_user(user_id: int) -> Optional[str]:
    """Returns username or None if not found."""
    users = {1: "Alice", 2: "Bob"}
    return users.get(user_id)
 
# Python 3.10+ syntax (preferred)
def find_user(user_id: int) -> str | None:
    """Returns username or None if not found."""
    users = {1: "Alice", 2: "Bob"}
    return users.get(user_id)

from typing import Union
 
# Pre-3.10 syntax
def format_value(value: Union[int, float, str]) -> str:
    return str(value)
 
# Python 3.10+ syntax (preferred)
def format_value(value: int | float | str) -> str:
    return str(value)
 
# Common pattern: accepting multiple input formats
def load_data(source: str | Path) -> list[dict[str, str]]:
    path = Path(source) if isinstance(source, str) else source
    with open(path) as f:
        return json.load(f)

from typing import Any
 
def log_event(event: str, payload: Any) -> None:
    """Accepts any payload type - useful for generic logging."""
    print(f"[{event}] {payload}")
 
# Any is compatible with every type
log_event("click", {"x": 100, "y": 200})
log_event("error", 404)
log_event("message", "hello")

from typing import TypeVar
 
T = TypeVar("T")
 
def first_element(items: list[T]) -> T:
    """Returns the first element, preserving its type."""
    return items[0]
 
# Type checkers infer the correct return type
name = first_element(["Alice", "Bob"])     # type: str
score = first_element([95, 87, 92])        # type: int
 
# Bounded TypeVar - restrict to specific types
Numeric = TypeVar("Numeric", int, float)
 
def add(a: Numeric, b: Numeric) -> Numeric:
    return a + b

from typing import TypeVar, Generic
 
T = TypeVar("T")
 
class Stack(Generic[T]):
    def __init__(self) -> None:
        self._items: list[T] = []
 
    def push(self, item: T) -> None:
        self._items.append(item)
 
    def pop(self) -> T:
        return self._items.pop()
 
    def peek(self) -> T:
        return self._items[-1]
 
    def is_empty(self) -> bool:
        return len(self._items) == 0
 
# Usage with specific types
int_stack: Stack[int] = Stack()
int_stack.push(42)
value: int = int_stack.pop()
 
str_stack: Stack[str] = Stack()
str_stack.push("hello")

from typing import Callable
 
# Function that takes a callback
def apply_operation(
    values: list[float],
    operation: Callable[[float], float]
) -> list[float]:
    return [operation(v) for v in values]
 
# Usage
import math
results = apply_operation([1.0, 4.0, 9.0], math.sqrt)
# results: [1.0, 2.0, 3.0]
 
# More complex callable signatures
Comparator = Callable[[str, str], int]
EventHandler = Callable[[str, dict[str, str]], None]
 
def sort_with_comparator(
    items: list[str],
    compare: Comparator
) -> list[str]:
    import functools
    return sorted(items, key=functools.cmp_to_key(compare))

from typing import Literal
 
def set_log_level(level: Literal["DEBUG", "INFO", "WARNING", "ERROR"]) -> None:
    print(f"Log level set to {level}")
 
set_log_level("INFO")      # OK
set_log_level("VERBOSE")   # Type error: not a valid literal
 
# Useful for mode parameters
def read_file(
    path: str,
    mode: Literal["text", "binary"] = "text"
) -> str | bytes:
    if mode == "text":
        return open(path).read()
    return open(path, "rb").read()

from typing import TypedDict, NotRequired
 
class UserProfile(TypedDict):
    name: str
    email: str
    age: int
    bio: NotRequired[str]  # Optional key (Python 3.11+)
 
def display_user(user: UserProfile) -> str:
    return f"{user['name']} ({user['email']}), age {user['age']}"
 
# Type checker validates the structure
user: UserProfile = {
    "name": "Alice",
    "email": "alice@example.com",
    "age": 30,
}
 
display_user(user)  # OK

from typing import Protocol, runtime_checkable
 
@runtime_checkable
class Drawable(Protocol):
    def draw(self, x: int, y: int) -> None: ...
 
class Circle:
    def draw(self, x: int, y: int) -> None:
        print(f"Drawing circle at ({x}, {y})")
 
class Square:
    def draw(self, x: int, y: int) -> None:
        print(f"Drawing square at ({x}, {y})")
 
def render(shape: Drawable, x: int, y: int) -> None:
    shape.draw(x, y)
 
# Both work without inheriting from Drawable
render(Circle(), 10, 20)
render(Square(), 30, 40)
 
# runtime_checkable enables isinstance checks
print(isinstance(Circle(), Drawable))  # True

from typing import TypeAlias
 
# Simple aliases
UserId: TypeAlias = int
JsonDict: TypeAlias = dict[str, "JsonValue"]
JsonValue: TypeAlias = str | int | float | bool | None | list["JsonValue"] | JsonDict
 
# Complex aliases simplify signatures
Matrix: TypeAlias = list[list[float]]
Callback: TypeAlias = Callable[[str, int], bool]
Config: TypeAlias = dict[str, str | int | list[str]]
 
def transform_matrix(m: Matrix, factor: float) -> Matrix:
    return [[cell * factor for cell in row] for row in m]
 
# Python 3.12+ uses the type statement
# type Vector = list[float]

# Install mypy
# pip install mypy
 
# Check a single file
# mypy script.py
 
# Check an entire project
# mypy src/
 
# Check with specific Python version
# mypy --python-version 3.10 src/

# pyproject.toml configuration
# [tool.mypy]
# python_version = "3.10"
# warn_return_any = true
# warn_unused_configs = true
# disallow_untyped_defs = true
# check_untyped_defs = true
# no_implicit_optional = true
# strict_equality = true
 
# Per-module overrides
# [[tool.mypy.overrides]]
# module = "third_party_lib.*"
# ignore_missing_imports = true

# Error: Incompatible return value type (got "Optional[str]", expected "str")
# Fix: Handle the None case
def get_name(user_id: int) -> str:
    result = lookup(user_id)  # returns str | None
    if result is None:
        raise ValueError(f"User {user_id} not found")
    return result  # mypy knows result is str here
 
# Error: Item "None" of "Optional[str]" has no attribute "upper"
# Fix: Narrow the type first
def format_name(name: str | None) -> str:
    if name is not None:
        return name.upper()
    return "UNKNOWN"
 
# Error: Need type annotation for variable
# Fix: Add explicit annotation
items: list[str] = []  # Not just: items = []
 
# Error: Incompatible types in assignment
# Fix: Use Union or correct the type
value: int | str = 42
value = "hello"  # OK with union type
 
# Suppress specific errors when necessary
x = some_untyped_function()  # type: ignore[no-untyped-call]

from fastapi import FastAPI
from pydantic import BaseModel
 
app = FastAPI()
 
class UserCreate(BaseModel):
    name: str
    email: str
    age: int
    tags: list[str] = []
 
class UserResponse(BaseModel):
    id: int
    name: str
    email: str
 
@app.post("/users", response_model=UserResponse)
async def create_user(user: UserCreate) -> UserResponse:
    # FastAPI validates the request body against UserCreate
    # and serializes the response to match UserResponse
    return UserResponse(id=1, name=user.name, email=user.email)

import pandas as pd
from typing import Any
 
# Basic DataFrame typing
def clean_dataframe(df: pd.DataFrame) -> pd.DataFrame:
    return df.dropna().reset_index(drop=True)
 
# Using pandera for schema validation
# pip install pandera
import pandera as pa
from pandera.typing import DataFrame, Series
 
class SalesSchema(pa.DataFrameModel):
    product: Series[str]
    quantity: Series[int] = pa.Field(ge=0)
    price: Series[float] = pa.Field(gt=0)
    date: Series[pd.Timestamp]
 
@pa.check_types
def process_sales(df: DataFrame[SalesSchema]) -> DataFrame[SalesSchema]:
    """Type-checked DataFrame processing."""
    return df[df["quantity"] > 0]

# Mistake: mutable default with type hint
def bad_append(item: str, items: list[str] = []) -> list[str]:
    items.append(item)  # Shared mutable default!
    return items
 
# Fix: use None as default
def good_append(item: str, items: list[str] | None = None) -> list[str]:
    if items is None:
        items = []
    items.append(item)
    return items

from typing import TypedDict, Callable, TypeAlias
from pathlib import Path
import csv
 
# Define data shapes with TypedDict
class RawRecord(TypedDict):
    name: str
    value: str
    category: str
 
class ProcessedRecord(TypedDict):
    name: str
    value: float
    category: str
    normalized: float
 
# Type alias for transform functions
Transform: TypeAlias = Callable[[list[ProcessedRecord]], list[ProcessedRecord]]
 
def load_csv(path: Path) -> list[RawRecord]:
    """Load CSV data with typed output."""
    records: list[RawRecord] = []
    with open(path) as f:
        reader = csv.DictReader(f)
        for row in reader:
            records.append(RawRecord(
                name=row["name"],
                value=row["value"],
                category=row["category"],
            ))
    return records
 
def parse_records(raw: list[RawRecord]) -> list[ProcessedRecord]:
    """Convert raw string records to typed records."""
    max_value = max(float(r["value"]) for r in raw) or 1.0
    return [
        ProcessedRecord(
            name=r["name"],
            value=float(r["value"]),
            category=r["category"],
            normalized=float(r["value"]) / max_value,
        )
        for r in raw
    ]
 
def filter_by_category(
    records: list[ProcessedRecord],
    category: str
) -> list[ProcessedRecord]:
    """Filter records, fully typed input and output."""
    return [r for r in records if r["category"] == category]
 
def apply_transforms(
    records: list[ProcessedRecord],
    transforms: list[Transform]
) -> list[ProcessedRecord]:
    """Apply a chain of typed transform functions."""
    result = records
    for transform in transforms:
        result = transform(result)
    return result
 
# Usage
raw = load_csv(Path("data.csv"))
processed = parse_records(raw)
filtered = filter_by_category(processed, "electronics")

import pandas as pd
import pygwalker as pyg
 
# Your typed pipeline produces clean, structured data
data: list[ProcessedRecord] = parse_records(raw)
df = pd.DataFrame(data)
 
# PyGWalker renders it as an interactive visualization
walker = pyg.walk(df)