Question 1

How to create a virtual environment?

Accepted Answer

1) python -m venv venv. 2) Activation on Windows: venv\Scripts\activate. 3) Linux/Mac: source venv/bin/activate. 4) Deactivation: deactivate. 5) pip install package.

Question 2

What are the differences between lists, tuples, sets?

Accepted Answer

1) List — mutable: [1,2,3]. 2) Tuple — immutable: (1,2,3). 3) Set — no duplicates: {1,2,3}. 4) Dictionary — key-value pairs. 5) Tuples are hashable.

Question 3

What are decorators?

Accepted Answer

1) A function that modifies another function. 2) @decorator above a definition. 3) def log(f): def wrapper(*a): return f(*a); return wrapper. 4) functools.wraps preserves name/docstring. 5) Examples: @property, @staticmethod.

Question 4

How to work with files?

Accepted Answer

1) with open('file.txt', 'r', encoding='utf-8') as f:. 2) f.read() or for line in f:. 3) Modes: r, w, a, rb. 4) with automatically closes the file. 5) pathlib.Path — modern approach.

Question 5

What are generators?

Accepted Answer

1) Function with yield. 2) Returns an iterator. 3) Lazy evaluation (saves memory). 4) next(gen) — get next element. 5) Generator expression: (x*2 for x in range(10)).

Question 6

How to install Python and verify it works?

Accepted Answer

1) Download from python.org or install via package manager. 2) Check: python --version (or python3 --version). 3) Check pip: python -m pip --version. 4) Update pip: python -m pip install -U pip. 5) Run REPL: python.

Question 7

What is the difference between python and python3 in Linux?

Accepted Answer

1) In some distributions, python = Python 2 (historically) or absent. 2) python3 always points to Python 3. 3) For scripts, better to use shebang: #!/usr/bin/env python3. 4) pip/pip3 similarly. 5) Virtual environments resolve version conflicts.

Question 8

How to properly set dependencies in a project?

Accepted Answer

1) Create venv. 2) Install packages via python -m pip install .... 3) Freeze dependencies: pip freeze > requirements.txt. 4) On another PC: pip install -r requirements.txt. 5) For larger projects, consider Poetry/uv.

Question 9

What is PEP 8 and why is it needed?

Accepted Answer

1) PEP 8 — Python code style. 2) Makes code readable and consistent in a team. 3) Indents: 4 spaces. 4) Names: snake_case for functions/variables, PascalCase for classes. 5) Autoformatter: black, linter: ruff.

Question 10

How does module import (import) work in Python?

Accepted Answer

1) import searches for a module in sys.path. 2) The local project folder is usually at the beginning of sys.path. 3) __init__.py file makes a folder a package (mandatory in older versions). 4) Relative imports: from .sub import x. 5) Avoid cyclic imports.

Question 11

What is the difference between shallow copy and deep copy?

Accepted Answer

1) Shallow copy copies only the top level of the container. 2) Nested objects remain shared. 3) Deep copy recursively copies everything. 4) Shallow: list.copy(), copy.copy(). 5) Deep: copy.deepcopy() (can be expensive).

Question 12

When to use list and when to use tuple?

Accepted Answer

1) list — if you need to modify contents. 2) tuple — if data should be immutable. 3) tuple can be used as a dict key (if elements are hashable). 4) tuple is usually slightly more compact. 5) For returning “packets” of values, tuple is convenient.

Question 13

How are dicts structured and which operations are fast?

Accepted Answer

1) dict — hash table. 2) Access/insertion/deletion by key is usually O(1). 3) Key must be hashable (immutable). 4) Insertion order is preserved (Python 3.7+ guaranteed). 5) Use collections.Counter for counting.

Question 14

How to nicely format strings (f-string)?

Accepted Answer

1) f'Hello, {name}'. 2) Format: f'{value:.2f}' for float. 3) Dates: f'{dt:%Y-%m-%d}'. 4) Width: f'{n:>10}'. 5) For debugging: f'{var=}' (3.8+).

Question 15

How does unpacking *args and **kwargs work?

Accepted Answer

1) *args — positional arguments as a tuple. 2) **kwargs — named arguments as a dict. 3) You can proxy a call: f(*args, **kwargs). 4) Collection unpacking: [*a, *b]. 5) Dict unpacking: {**d1, **d2}.

Question 16

What is list comprehension and when to use it?

Accepted Answer

1) Short syntax for creating a list. 2) Example: [x*x for x in range(10) if x % 2 == 0]. 3) Readable compared to manual loop (if expression is simple). 4) Use (...) for generator. 5) Be careful with nested loops, don't overcomplicate.

Question 17

How does generator expression differ from list comprehension?

Accepted Answer

1) List comprehension creates a list immediately. 2) Generator expression creates items lazily. 3) Generator saves memory on large data. 4) Example: sum(x*x for x in data). 5) Generator can be “exhausted” once.

Question 18

How does enumerate work and why is it convenient?

Accepted Answer

1) enumerate(iterable) gives (index, value). 2) Example: for i, item in enumerate(items, start=1): ... 3) Eliminates manual counter. 4) Works with any iterator. 5) Improves readability.

Question 19

How to properly sort lists (sorted vs sort)?

Accepted Answer

1) list.sort() sorts in place and returns None. 2) sorted(iterable) returns a new list. 3) key= — key function: sorted(users, key=lambda u: u.age). 4) reverse=True — reverse order. 5) Stable sorting (important for multi-level).

Question 20

How to combine and filter collections (map/filter) and when is comprehension better?

Accepted Answer

1) map/filter return iterators. 2) Example: map(str, nums). 3) filter(fn, items) keeps elements where fn=True. 4) Comprehension is often easier to read. 5) For complex logic, a regular loop is better.

Question 21

How is exception handling structured with try/except/else/finally?

Accepted Answer

1) try — code that might fail. 2) except catches specific exceptions (don't catch bare except). 3) else runs if no exception occurred. 4) finally always runs (cleanup). 5) raise to propagate/create an error.

Question 22

How to create a custom exception?

Accepted Answer

1) Inherit from Exception: class MyError(Exception): pass. 2) Use a meaningful name. 3) Pass a message: raise MyError('...'). 4) For parameters — store them in attributes. 5) Do not inherit from BaseException.

Question 23

What is a context manager and with?

Accepted Answer

1) with guarantees resource release. 2) Files, locks, connections. 3) Custom manager: implement __enter__ and __exit__. 4) Or use contextlib.contextmanager. 5) Simplifies safe code.

Question 24

How to work with date and time (datetime)?

Accepted Answer

1) datetime.datetime.now() — current local time. 2) Better to store UTC: datetime.datetime.now(datetime.timezone.utc). 3) Parsing: datetime.datetime.fromisoformat(...). 4) Formatting: dt.strftime('%Y-%m-%d'). 5) Difference: timedelta.

Question 25

How to safely work with paths (pathlib)?

Accepted Answer

1) from pathlib import Path. 2) p = Path('data') / 'file.txt'. 3) p.exists(), p.is_file(), p.read_text(encoding='utf-8'). 4) Create folder: p.parent.mkdir(parents=True, exist_ok=True). 5) Cross-platform and readable.

Question 26

How to read/write JSON?

Accepted Answer

1) import json. 2) json.loads(text) → dict/list. 3) json.dumps(obj, ensure_ascii=False) → string. 4) For files: json.load(fp), json.dump(obj, fp). 5) For datetime, convert (e.g., ISO string).

Question 27

What is a dataclass and why is it needed?

Accepted Answer

1) dataclass reduces boilerplate for data classes. 2) @dataclass automatically creates __init__/__repr__/__eq__. 3) You can set frozen=True (immutability). 4) default_factory for lists/dicts. 5) Useful for DTOs/models.

Question 28

How to create an immutable object in Python?

Accepted Answer

1) Use dataclass(frozen=True). 2) Or NamedTuple / typing.NamedTuple. 3) For collections: tuple instead of list, frozenset instead of set. 4) Do not mutate internal structures. 5) For configs, it reduces bugs.

Question 29

What is @property and when to use it?

Accepted Answer

1) Makes the method look like a field. 2) Allows computed properties. 3) You can add validation in setter. 4) Helps hide internal representation. 5) Don't abuse — complexity also increases.

Question 30

How do inheritance and super() work?

Accepted Answer

1) class Child(Parent): ... 2) super() calls methods of the base class according to MRO. 3) In multiple inheritance, MRO is critical. 4) Always use super() in cooperative classes. 5) Check MRO: Class.mro().

Question 31

What is MRO and why is it important?

Accepted Answer

1) MRO — the method search order in inheritance. 2) Uses the C3 linearization algorithm. 3) Affects super() in multiple inheritance. 4) MRO errors → TypeError when creating a class. 5) See Class.__mro__ / Class.mro().

Question 32

What are dunder methods (__str__, __repr__, __iter__)?

Accepted Answer

1) These are object behavior protocols. 2) __repr__ — for developers, __str__ — for users. 3) __len__ makes the object compatible with len(). 4) __iter__/__next__ — iterator. 5) __enter__/__exit__ — context manager.

Question 33

How to correctly implement __repr__?

Accepted Answer

1) Make an unambiguous representation. 2) Ideally: so that the object can be reconstructed (if really possible). 3) Use f'ClassName(x={self.x!r})'. 4) Do not perform heavy computations. 5) For dataclass, __repr__ is generated automatically.

Question 34

What is an iterator and an iterable object?

Accepted Answer

1) Iterable — something you can loop over with for. 2) Iterator — object with __iter__ and __next__. 3) iter(obj) returns an iterator. 4) next(it) gives the next element or raises StopIteration. 5) Generators are iterators.

Question 35

How to use itertools for efficient loops?

Accepted Answer

1) itertools.chain combines sequences. 2) itertools.islice creates a “slice” of an iterator. 3) itertools.groupby groups data (on sorted data). 4) itertools.product for Cartesian product. 5) Good for large data streams.

Question 36

What is the difference between * and / in function signatures (keyword-only and positional-only)?

Accepted Answer

1) f(a, /, b) — a is positional-only. 2) f(*, x, y) — x and y are keyword-only. 3) Reduces errors in calls. 4) Helps maintain API. 5) Often used in library code.

Question 37

What is typing and why are type hints needed?

Accepted Answer

1) Type hints — type annotations (not mandatory for the interpreter). 2) Improve IDE autocomplete. 3) Help static analyzers (mypy/pyright). 4) Document API. 5) Reduce bugs in large projects.

Question 38

How to use Optional, Union, and | (PEP 604)?

Accepted Answer

1) Optional[str] = Union[str, None]. 2) In 3.10+: str | None. 3) Union[int, str] = int | str. 4) Prefer | in modern code. 5) Don't forget about handling None.

Question 39

What is Protocol and when is it useful?

Accepted Answer

1) Protocol describes a "duck type" by methods. 2) Allows typing interfaces without inheritance. 3) Example: an object with a .read() method. 4) Useful for mocks and abstractions. 5) Requires typing.Protocol (3.8+) or typing_extensions.

Question 40

How to properly log (logging) instead of print?

Accepted Answer

1) import logging; log = logging.getLogger(__name__). 2) Use levels: debug/info/warning/error. 3) Configure format and handler. 4) Do not log secrets (tokens/passwords). 5) Do not configure root-logger in the library.

Question 41

How to read large files without loading into memory?

Accepted Answer

1) Iterate over lines: for line in f:. 2) For binary — read in chunks: f.read(8192). 3) Use generators for pipelines. 4) Do not do f.read() on gigabytes. 5) For CSV — csv.reader line by line.

Question 42

How to work with CSV?

Accepted Answer

1) import csv. 2) reader = csv.DictReader(f). 3) Use: csv.DictWriter(...).writeheader(); writer.writerow(...). 4) Specify newline='' when opening. 5) For encodings use encoding='utf-8'.

Question 43

How to quickly find minimum/maximum by key?

Accepted Answer

1) min(items, key=...). 2) max(items, key=...). 3) key can be lambda or attrgetter. 4) For dict: max(d, key=d.get). 5) More readable than manual loops.

Question 44

How to use collections.Counter and defaultdict?

Accepted Answer

1) Counter counts frequencies: Counter(words). 2) most_common(n) — top-N. 3) defaultdict(list) is convenient for grouping. 4) defaultdict(int) — counter without if. 5) Improves readability and speed of code.

Question 45

How to group list elements by key?

Accepted Answer

1) For small data: defaultdict(list). 2) group[key].append(item). 3) For groupby, sort first by key. 4) key function should be stable. 5) Result usually dict[key] -> list[items].

Question 46

How to make HTTP requests in Python?

Accepted Answer

1) Popular: requests (if available). 2) In standard library: urllib.request (less convenient). 3) For async: httpx or aiohttp. 4) Always set a timeout. 5) Check status and handle errors.

Question 47

How to safely parse user input?

Accepted Answer

1) Explicitly convert types: int(x), float(x) with try/except. 2) Check ranges and format. 3) Do not use eval/exec on input data. 4) For JSON use json.loads. 5) For paths — pathlib and whitelisting.

Question 48

What is an f-string and why is it better than format?

Accepted Answer

1) f-string is faster and more readable. 2) Supports expressions inside {}. 3) Convenient for formatting numbers/dates. 4) Easier to maintain than concatenation. 5) format is useful when template is stored separately.

Question 49

How to create a CLI script in Python?

Accepted Answer

1) Entry point: if __name__ == '__main__': main(). 2) Parsing arguments: argparse. 3) Return code: sys.exit(code). 4) Logs instead of print in complex utilities. 5) Specify shebang for Linux.

Question 50

How to use argparse for command-line arguments?

Accepted Answer

1) parser = argparse.ArgumentParser(). 2) parser.add_argument('--path', required=True). 3) args = parser.parse_args(). 4) Support subcommands: add_subparsers(). 5) Auto-generate help.

Question 51

What is asyncio and when is it needed?

Accepted Answer

1) asyncio — concurrency for I/O (network/files) without threads. 2) Use async def and await. 3) Does not speed up CPU-heavy tasks. 4) Suitable for web services/bots/parsers. 5) For CPU — multiprocessing.

Question 52

What is the difference between threading and multiprocessing?

Accepted Answer

1) threading — threads in one process, shared GIL. 2) Good for I/O tasks. 3) multiprocessing — separate processes, parallelize CPU. 4) IPC and serialization are more expensive. 5) Choice depends on workload profile.

Question 53

What is GIL and how does it affect performance?

Accepted Answer

1) GIL limits Python bytecode execution to one thread. 2) CPU-bound tasks do not scale with threads. 3) I/O tasks usually work fine with threads. 4) For CPU — multiprocessing or NumPy (C code). 5) Alternatives: PyPy, Cython, Rust/Go extensions.

Question 54

How to run multiple async tasks in parallel?

Accepted Answer

1) Create tasks: asyncio.create_task(coro()). 2) Gather: await asyncio.gather(*tasks). 3) Use Semaphore to limit concurrency. 4) Handle exceptions (return_exceptions). 5) Remember about timeouts.

Question 55

How to properly implement timeouts in asyncio?

Accepted Answer

1) asyncio.wait_for(coro, timeout=...). 2) In Python 3.11+: asyncio.timeout(). 3) Set timeouts at library level for network clients. 4) Catch asyncio.TimeoutError. 5) Retry with backoff if needed.

Question 56

How to write tests: unittest or pytest?

Accepted Answer

1) unittest — standard library, more “Java-style”. 2) pytest — simpler syntax, powerful fixtures. 3) Convenient to run in CI. 4) Structure tests by modules/features. 5) Isolate external dependencies with mocks.

Question 57

How to mock dependencies in tests?

Accepted Answer

1) Use unittest.mock (patch, MagicMock). 2) Mock at the point of use, not definition. 3) Prefer dependency injection where possible. 4) Check calls: assert_called_once_with. 5) Do not turn tests into copies of implementation.

Question 58

How to work with fixtures in pytest?

Accepted Answer

1) @pytest.fixture for setup data. 2) scope=function/module/session. 3) Fixture can return an object or yield for teardown. 4) Fixtures can be parameterized. 5) Reduces test duplication.

Question 59

How to measure test coverage?

Accepted Answer

1) coverage.py or pytest-cov. 2) Run: pytest --cov=package. 3) View the HTML report. 4) High coverage does not guarantee quality, but helps. 5) Goal: cover critical business logic.

Question 60

How to profile Python code?

Accepted Answer

1) cProfile for general profiling. 2) timeit for micro-benchmarks. 3) line_profiler for line-by-line analysis. 4) py-spy for production without injection. 5) Measure first, then optimize.

Question 61

What are context managers (contextlib.contextmanager)?

Accepted Answer

1) Allows writing with statements via generator. 2) @contextmanager decorator on a function. 3) yield separates enter/exit. 4) Do cleanup in finally. 5) Useful for temporary settings/resources.

Question 62

How to cache computations (functools.lru_cache)?

Accepted Answer

1) @lru_cache(maxsize=...). 2) Works for pure functions. 3) Arguments must be hashable. 4) cache_clear() clears cache. 5) For complex cases — external cache (Redis).

Question 63

How to use @dataclass with default_factory?

Accepted Answer

1) Mutable default values cannot be default=[]. 2) Use field(default_factory=list). 3) Similarly for dict/set. 4) Prevents shared list across instances. 5) Common beginner mistake.

Question 64

Why is it dangerous to write def f(x, arr=[])?

Accepted Answer

1) Default value is computed once at function definition. 2) List will be shared across calls. 3) Correct: def f(x, arr=None): arr = arr or []. 4) Same for dict/set. 5) Classic Python trap.

Python