Вопросы на собеседовании Python-разработчика на английском

Пример ответа · EN

I use a list when the sequence needs to change, and a tuple when its positions describe a fixed record. For example, coordinates can be a tuple. Tuple immutability prevents replacing its elements, but a list stored inside a tuple can still change. I choose by meaning before making performance claims.

Пример ответа · EN

The __init__ method sets up an instance after it has been created. For example, it can store a validated name on self. Object creation is handled by __new__, so I distinguish creating an object from initializing its state. An initializer must return None, rather than return the newly created instance.

Пример ответа · EN

A comprehension works well for a short transformation or filter, such as [n * 2 for n in values if n > 0]. A list comprehension builds a list immediately. Parentheses around the expression create a generator expression instead. For several branches or side effects, I prefer an explicit loop.

Пример ответа · EN

In a GIL-enabled CPython build, only one thread executes Python bytecode at a time. Threads can still help with I/O waits. For CPU-heavy Python work, I would consider processes and measure the overhead. Free-threaded CPython builds also exist, so I would check the interpreter build and library compatibility before generalizing.

Пример ответа · EN

I first decide whether absence is expected. For an optional setting, settings.get('timeout', 10) provides a default. For required data, direct indexing and a narrow KeyError handler can make the failure explicit. If None is a valid stored value, I avoid using it alone to distinguish a missing key.

Пример ответа · EN

A shallow copy creates a new outer container but keeps references to its contents. If I copy a list of lists and append to an inner list, both outer lists observe that change. A deep copy recursively copies contents, subject to the objects' copying behavior. I choose based on the isolation I actually need.

Пример ответа · EN

A decorator takes a function and returns the object that replaces its name. A timing decorator can return a wrapper that calls the original function and reports its duration. I use functools.wraps to preserve useful metadata. I also make sure the wrapper forwards arguments and returns the original result.

Пример ответа · EN

The with statement uses a context manager to enter a block and run its exit logic afterwards, including when the block raises an exception. For a file, this normally means closing it. A custom manager can implement __enter__ and __exit__, or use contextlib.contextmanager with cleanup in a finally block.

Пример ответа · EN

The else block runs when the try block finishes normally, without an exception or an early return, break, or continue. I keep the operation that may fail inside try and put subsequent success-only work in else. Exceptions from else are not caught by the preceding except handlers. Cleanup belongs in finally when it is required.

Пример ответа · EN

Dictionary keys must be hashable: their hash must remain stable, and equal objects must have equal hashes. Strings work, while lists do not. A tuple is usable only if all its elements are hashable. I would use a tuple of identifiers as a composite key rather than a mutable list.

Пример ответа · EN

Equality asks whether objects compare as equal; identity asks whether they are the same object. Two separately created empty lists compare equal but are not identical. I use == for value comparisons and is for a singleton such as None. I do not rely on integer or string caching to compare values.

Пример ответа · EN

A default argument expression is evaluated when the function is defined. Calls that omit the argument therefore reuse the same list. If each call needs fresh state, I use None as the default and create a list inside the function. I make that contract explicit when None could otherwise be meaningful input.

Пример ответа · EN

In a function definition, *args collects extra positional arguments into a tuple, and **kwargs collects extra keyword arguments into a dictionary. At a call site, the same symbols unpack values. A wrapper can forward both to another function. For a stable public interface, I prefer explicit parameters where they make the contract clearer.

Пример ответа · EN

For ordinary function scopes, I describe lookup as local, enclosing function scopes, module globals, then built-ins. A nested function can read an enclosing variable. To rebind that variable, it needs nonlocal; global targets the module scope. I avoid calling this a complete rule for every Python construct, since class and annotation scopes have additional details.

Пример ответа · EN

A generator produces values on demand and preserves its execution state between yields. That lets a consumer process records without building a complete result list first. I would use it for a large input stream. It can be exhausted, so repeated traversal needs a fresh generator or deliberately stored results.

Пример ответа · EN

No. Declaring async def creates a coroutine function, but blocking work inside it can still block the event-loop thread. I use await with asynchronous I/O and schedule tasks when operations should overlap. For a blocking I/O library, asyncio.to_thread may help. CPU-heavy work requires a separate execution strategy, rather than merely adding async.

Пример ответа · EN

Type annotations describe intended types for tools and readers, but Python does not generally enforce them at runtime. A checker can flag an incompatible call before execution. External input still needs runtime validation. I use annotations to make interfaces clearer and checks more useful, without treating them as proof that incoming data is valid.

Пример ответа · EN

A virtual environment gives a project its own installed Python packages, normally isolated from the base environment's packages. It helps avoid dependency conflicts. I can create one with python -m venv .venv. It is not an operating-system sandbox, and recreating it reproducibly also requires a record of the project's dependencies.

Пример ответа · EN

An instance method receives the instance as self. A class method receives the class as cls, so it can implement an alternative constructor that respects subclasses. A static method receives neither automatically. I use the method type that matches the information required, and consider a module-level function when class grouping adds little value.

Пример ответа · EN

I would clarify which numeric field to compare and how to handle invalid or empty input. Then I would read records with csv.DictReader, convert that field explicitly, and track the best record during iteration. I would test a single record, ties, malformed values and empty input before deciding whether extra abstractions are useful.