|
| 1 | +# 🧵 Section 30: Async IO (asyncio) |
| 2 | +## Writing Asynchronous Python Programs |
| 3 | + |
| 4 | +⚡ **Learn how to write asynchronous programs using the `asyncio` module**, a powerful library for managing **I/O-bound tasks** like network calls, file handling, or system-level operations — all without blocking the main thread. |
| 5 | + |
| 6 | +This section explains: |
| 7 | +- 🔍 What is asynchronous programming and when to use it |
| 8 | +- 📦 Define `async` functions and `await` results |
| 9 | +- 🔄 Run multiple coroutines concurrently |
| 10 | +- ⏳ Use `asyncio.sleep()` to simulate async I/O |
| 11 | +- 💡 Hidden notes and best practices for writing clean async code |
| 12 | +- 🧪 Real-world example – concurrent web requests with `aiohttp` |
| 13 | + |
| 14 | + |
| 15 | + |
| 16 | +## 🧠 What You'll Learn |
| 17 | + |
| 18 | +| Concept | Description | |
| 19 | +|--------|-------------| |
| 20 | +| **Asynchronous IO (`asyncio`)** | Handle many I/O-bound tasks in parallel | |
| 21 | +| **`async def`** | Define a coroutine that can be run asynchronously | |
| 22 | +| **`await`** | Wait for a coroutine to complete | |
| 23 | +| **`asyncio.run()`** | Start and manage an async event loop | |
| 24 | +| **`asyncio.create_task()`** | Schedule a coroutine to run in the event loop | |
| 25 | +| **`asyncio.gather()`** | Run multiple tasks and wait for all to finish | |
| 26 | +| **Best Practices** | When to use async over threads or processes | |
| 27 | + |
| 28 | + |
| 29 | + |
| 30 | +## ⚙️ Introduction to Async IO in Python |
| 31 | + |
| 32 | +Python’s `asyncio` module allows you to write **non-blocking** programs that perform **multiple I/O-bound tasks** at once — ideal for: |
| 33 | +- Web scraping |
| 34 | +- API clients |
| 35 | +- Network servers |
| 36 | +- Concurrent file reading/writing |
| 37 | +- Long-running background jobs |
| 38 | + |
| 39 | +🔹 **Example – Basic Async Function** |
| 40 | +```python |
| 41 | +import asyncio |
| 42 | + |
| 43 | +async def greet(name): |
| 44 | + print(f"Hello, {name}") |
| 45 | + await asyncio.sleep(1) |
| 46 | + print(f"Goodbye, {name}") |
| 47 | + |
| 48 | +asyncio.run(greet("Alice")) |
| 49 | +``` |
| 50 | + |
| 51 | +🔸 The `await` keyword tells Python to pause execution until the awaited task completes. |
| 52 | + |
| 53 | + |
| 54 | + |
| 55 | +## 📌 Define and Run Coroutines |
| 56 | + |
| 57 | +### ✅ Defining Coroutines |
| 58 | + |
| 59 | +Use `async def` to define a function that returns a **coroutine object** — not executed immediately. |
| 60 | + |
| 61 | +🔹 **Example – Multiple Coroutines** |
| 62 | +```python |
| 63 | +async def task_one(): |
| 64 | + print("Task One Started") |
| 65 | + await asyncio.sleep(2) |
| 66 | + print("Task One Completed") |
| 67 | + |
| 68 | +async def task_two(): |
| 69 | + print("Task Two Started") |
| 70 | + await asyncio.sleep(1) |
| 71 | + print("Task Two Completed") |
| 72 | +``` |
| 73 | + |
| 74 | +### ✅ Running Coroutines |
| 75 | + |
| 76 | +You must run coroutines inside an event loop. Use `asyncio.run()` in Python 3.7+: |
| 77 | + |
| 78 | +```python |
| 79 | +asyncio.run(task_one()) |
| 80 | +``` |
| 81 | + |
| 82 | +Or run them concurrently using `create_task()`: |
| 83 | + |
| 84 | +```python |
| 85 | +async def main(): |
| 86 | + t1 = asyncio.create_task(task_one()) |
| 87 | + t2 = asyncio.create_task(task_two()) |
| 88 | + await t1 |
| 89 | + await t2 |
| 90 | + |
| 91 | +asyncio.run(main()) |
| 92 | +``` |
| 93 | + |
| 94 | +🔸 Output will show overlapping execution due to concurrency. |
| 95 | + |
| 96 | + |
| 97 | +## 🔄 Awaitable Objects |
| 98 | + |
| 99 | +Any object that can be used in an `await` expression is called an **awaitable**. |
| 100 | + |
| 101 | +They include: |
| 102 | +- Coroutines |
| 103 | +- Tasks (created via `create_task()`) |
| 104 | +- Futures (used internally by asyncio) |
| 105 | + |
| 106 | +🔹 **Example – Using `await` on a Task** |
| 107 | +```python |
| 108 | +async def fetch_data(): |
| 109 | + await asyncio.sleep(1) |
| 110 | + return "Data" |
| 111 | + |
| 112 | +async def main(): |
| 113 | + task = asyncio.create_task(fetch_data()) |
| 114 | + result = await task |
| 115 | + print("Result:", result) |
| 116 | + |
| 117 | +asyncio.run(main()) # Result: Data |
| 118 | +``` |
| 119 | + |
| 120 | + |
| 121 | + |
| 122 | +## 🧰 Real-World Example – Fetching from Multiple URLs |
| 123 | + |
| 124 | +Let’s build a program that fetches data from multiple URLs concurrently using `aiohttp`. |
| 125 | + |
| 126 | +### 🧱 Step 1: Install aiohttp |
| 127 | +```bash |
| 128 | +pip install aiohttp |
| 129 | +``` |
| 130 | + |
| 131 | +### 🛠️ Step 2: Define Coroutine for Fetching |
| 132 | +```python |
| 133 | +import asyncio |
| 134 | +import aiohttp |
| 135 | + |
| 136 | +async def fetch_url(session, url): |
| 137 | + async with session.get(url) as response: |
| 138 | + content = await response.text() |
| 139 | + print(f"Fetched {url} → {len(content)} bytes") |
| 140 | + return content |
| 141 | + |
| 142 | +async def main(urls): |
| 143 | + async with aiohttp.ClientSession() as session: |
| 144 | + tasks = [fetch_url(session, url) for url in urls] |
| 145 | + results = await asyncio.gather(*tasks) |
| 146 | + print(f"Fetched {len(results)} pages.") |
| 147 | + |
| 148 | +urls = [ |
| 149 | + 'https://example.com', |
| 150 | + 'https://httpbin.org/get', |
| 151 | + 'https://jsonplaceholder.typicode.com' |
| 152 | +] |
| 153 | + |
| 154 | +asyncio.run(main(urls)) |
| 155 | +``` |
| 156 | + |
| 157 | +🔸 This shows how to make **multiple HTTP requests concurrently** — faster than sequential or even threaded approaches. |
| 158 | + |
| 159 | + |
| 160 | + |
| 161 | +## ⏳ Working with Delays and Timeouts |
| 162 | + |
| 163 | +Use `await asyncio.sleep(n)` to simulate I/O delays. |
| 164 | + |
| 165 | +🔹 **Example – Delayed Greeting** |
| 166 | +```python |
| 167 | +async def delayed_greet(name, delay): |
| 168 | + print(f"{name} is waiting {delay} seconds...") |
| 169 | + await asyncio.sleep(delay) |
| 170 | + print(f"Hello, {name}!") |
| 171 | + |
| 172 | +async def main(): |
| 173 | + asyncio.create_task(delayed_greet("Alice", 1)) |
| 174 | + asyncio.create_task(delayed_greet("Bob", 2)) |
| 175 | + asyncio.create_task(delayed_greet("Charlie", 0.5)) |
| 176 | + await asyncio.sleep(3) |
| 177 | + |
| 178 | +asyncio.run(main()) |
| 179 | +``` |
| 180 | + |
| 181 | +🔸 Tasks run concurrently — Charlie finishes first, then Alice, then Bob. |
| 182 | + |
| 183 | + |
| 184 | + |
| 185 | +## 🧩 Advanced Example – Concurrent File Reader |
| 186 | + |
| 187 | +Read multiple files asynchronously using `aiofiles`. |
| 188 | + |
| 189 | +### 📁 Step 1: Install aiofiles |
| 190 | +```bash |
| 191 | +pip install aiofiles |
| 192 | +``` |
| 193 | + |
| 194 | +### 📖 Step 2: Read Files Without Blocking |
| 195 | +```python |
| 196 | +import asyncio |
| 197 | +import aiofiles |
| 198 | + |
| 199 | +async def read_file(filename): |
| 200 | + async with aiofiles.open(filename, mode='r') as f: |
| 201 | + content = await f.read() |
| 202 | + print(f"Read {filename}: {len(content)} characters") |
| 203 | + return content |
| 204 | + |
| 205 | +async def main(): |
| 206 | + files = ["file1.txt", "file2.txt", "file3.txt"] |
| 207 | + tasks = [read_file(f) for f in files] |
| 208 | + await asyncio.gather(*tasks) |
| 209 | + |
| 210 | +asyncio.run(main()) |
| 211 | +``` |
| 212 | + |
| 213 | +🔸 Each file is read without blocking others — great for batch processing large logs or config files. |
| 214 | + |
| 215 | + |
| 216 | + |
| 217 | +## 🧯 Best Practices & Notes |
| 218 | + |
| 219 | +| Practice | Description | |
| 220 | +|---------|-------------| |
| 221 | +| 🧠 Use `async/await` instead of callbacks | Cleaner and easier to maintain | |
| 222 | +| 📦 Prefer `asyncio.run()` in Python 3.7+ | It manages the event loop automatically | |
| 223 | +| 🔄 Use `create_task()` to schedule work | For true concurrency | |
| 224 | +| 🧾 Don’t mix sync and async calls | Can cause performance bottlenecks | |
| 225 | +| 🕒 Always use `await` with coroutines | Otherwise they won't execute | |
| 226 | +| 🧲 Use `gather()` to collect results from multiple tasks | Clean way to await all results | |
| 227 | +| 🧽 Avoid long-running CPU-bound logic inside coroutines | Use multiprocessing if needed | |
| 228 | +| 🧩 Use async libraries like `aiohttp`, `aiofiles`, `asyncpg` | Maximize async benefits | |
| 229 | +| 🧾 Add timeouts to prevent hanging | `await asyncio.wait_for(task, timeout=5)` | |
| 230 | +| 🧠 Use `async for` or `async with` when working with async iterators and context managers | |
| 231 | + |
| 232 | + |
| 233 | + |
| 234 | +## 🧬 Using `asyncio` with Threads |
| 235 | + |
| 236 | +Sometimes you need to integrate async with threading — for example, calling async code from a GUI app. |
| 237 | + |
| 238 | +🔹 **Example – Run async code inside a thread** |
| 239 | +```python |
| 240 | +import asyncio |
| 241 | +import threading |
| 242 | + |
| 243 | +def start_async_loop(loop): |
| 244 | + asyncio.set_event_loop(loop) |
| 245 | + loop.run_forever() |
| 246 | + |
| 247 | +loop = asyncio.new_event_loop() |
| 248 | +thread = threading.Thread(target=start_async_loop, args=(loop,), daemon=True) |
| 249 | +thread.start() |
| 250 | + |
| 251 | +async def add_to_queue(item): |
| 252 | + await asyncio.sleep(1) |
| 253 | + print(f"Processed {item}") |
| 254 | + |
| 255 | +future = asyncio.run_coroutine_threadsafe(add_to_queue("test"), loop) |
| 256 | +future.result() # Wait for async result |
| 257 | +``` |
| 258 | + |
| 259 | +🔸 This pattern is useful in apps where the main thread runs a different loop (e.g., Tkinter or Flask). |
| 260 | + |
| 261 | + |
| 262 | + |
| 263 | +## 🧪 Real-World Use Case – Async Database Query Pool |
| 264 | + |
| 265 | +Let’s say you want to query multiple database endpoints in parallel. |
| 266 | + |
| 267 | +### 🧱 Sample Async DB Query Function |
| 268 | +```python |
| 269 | +import asyncio |
| 270 | +import random |
| 271 | + |
| 272 | +async def query_db(db_name): |
| 273 | + delay = random.uniform(0.5, 2) |
| 274 | + print(f"Querying {db_name} ({delay:.2f}s)") |
| 275 | + await asyncio.sleep(delay) |
| 276 | + print(f"Finished querying {db_name}") |
| 277 | + return f"Results from {db_name}" |
| 278 | + |
| 279 | +async def main(): |
| 280 | + db_list = ["users", "orders", "inventory", "logs"] |
| 281 | + tasks = [query_db(db) for db in db_list] |
| 282 | + results = await asyncio.gather(*tasks) |
| 283 | + for result in results: |
| 284 | + print(result) |
| 285 | + |
| 286 | +asyncio.run(main()) |
| 287 | +``` |
| 288 | + |
| 289 | +🔸 Simulates querying multiple databases concurrently — real use cases might involve actual async drivers like `asyncpg` or `motor`. |
| 290 | + |
| 291 | + |
| 292 | + |
| 293 | +## 💡 Hidden Tips & Notes |
| 294 | + |
| 295 | +- 🧠 On Windows, you may need to set the event loop policy: |
| 296 | + ```python |
| 297 | + asyncio.set_event_loop_policy(asyncio.WindowsSelectorEventLoopPolicy()) |
| 298 | + ``` |
| 299 | +- 🧱 Always wrap your top-level async logic inside `main()` and call `asyncio.run(main())` |
| 300 | +- 📦 Prefer `create_task()` over `ensure_future()` or manual scheduling |
| 301 | +- 🧾 Use `await asyncio.sleep(0)` to yield control to other coroutines (for cooperative multitasking) |
| 302 | +- 🧐 Use `asyncio.all_tasks()` and `asyncio.current_task()` for debugging running tasks |
| 303 | +- 🧵 Use `run_coroutine_threadsafe()` to interact with async code from threads |
| 304 | +- 🧩 Use `asyncio.Queue` for producer/consumer patterns within async |
| 305 | +- 🧾 Use `asyncio.TimeoutError` and `wait_for()` to avoid infinite waits |
| 306 | + |
| 307 | + |
| 308 | + |
| 309 | +## 📌 Summary |
| 310 | + |
| 311 | +| Feature | Purpose | |
| 312 | +|--------|---------| |
| 313 | +| `async def` | Define a coroutine | |
| 314 | +| `await` | Execute and wait for a coroutine | |
| 315 | +| `asyncio.run()` | Start and manage the event loop | |
| 316 | +| `create_task()` | Schedule a coroutine for execution | |
| 317 | +| `gather()` | Collect results from multiple tasks | |
| 318 | +| `sleep()` | Simulate I/O delays | |
| 319 | +| `aiohttp`, `aiofiles` | External async libraries for real-world use | |
| 320 | +| Best Practice | Don’t block inside coroutines — keep I/O non-blocking | |
| 321 | + |
| 322 | + |
| 323 | + |
| 324 | +🎉 Congratulations! You now understand how to write **asynchronous programs in Python** using `asyncio`, including: |
| 325 | +- How to define and run coroutines |
| 326 | +- How to schedule and await multiple tasks |
| 327 | +- Real-world applications like web scraping and file reading |
| 328 | +- Best practices for avoiding blocking calls and managing timeouts |
| 329 | + |
| 330 | +This completes our full roadmap from **Python fundamentals to advanced topics like OOP, metaprogramming, and concurrency models**! |
0 commit comments