Capture stdout and stderr from the code under test using the pytest capsys fixture.

Provides access to the captured output so that tests can assert on what the program printed to stdout or stderr.

Used in pytest test functions when you need to verify printed output, error messages, CLI command output, or any side‑effect that writes to sys.stdout or sys.stderr.

Inside a test function that receives the capsys fixture, after executing the code under test but before any further output that could interfere with the capture.

out, err = capsys.readouterr()

Tuple unpacking assignment where the right‑hand side is the tuple returned by capsys.readouterr(), containing (stdout, stderr) as strings.

out: variable name for captured stdout (str); err: variable name for captured stderr (str); capsys: the pytest fixture providing the capture mechanism (usually named capsys).

• assert out == expected_output
• assert err == expected_error
• capsys
• capsysbinary
• capfd
• monkeypatch
• print
• logging
• caplog

• Use different variable names (e.g.
• stdout
• stderr)
• ignore one output with an underscore (out
• _ = capsys.readouterr() or _
• err = capsys.readouterr())
• use capsysbinary for binary data
• use capfd for low‑level file‑descriptor capture.

1. Calling capsys.readouterr() before running the code under test → captures empty strings, causing assertions to fail because output is missed. 2. Swapping the order of unpacked variables (err, out = capsys.readouterr()) → assertions check the wrong streams, leading to false test failures or passes. 3. Forgetting to call capsys.readouterr() and asserting on raw print output → tests miss captured output and may produce flaky results due to actual console interference. 4. Calling capsys.readouterr() after a fixture or teardown that has already restored sys.stdout/stderr → capture returns empty strings, missing the output. 5. Assuming captured output includes exactly the same newlines as printed strings without accounting for print’s automatic newline → off‑by‑one errors in string comparisons.

1. capfd – captures at the file‑descriptor level, works even when code bypasses sys.stdout (e.g., os.write). Difference: capfd works with low‑level descriptors, capsys works with sys.stdout/stderr redirection. 2. capsysbinary – same as capsys but returns bytes instead of str. Difference: use when testing binary output or encoding specifics. 3. monkeypatch.setattr('sys.stdout', io.StringIO()) – manual mocking approach. Difference: more verbose, requires manual cleanup, less integrated with pytest fixtures. 4. caplog – captures logging output rather than print output. Difference: targets the logging module, not stdout/stderr.

Coming from unittest.mock: may try to patch sys.stdout with unittest.mock.patch instead of using pytest's capsys fixture → use capsys for simpler, automatic cleanup. Coming from bash scripting: may expect to capture output by redirecting to a file and then reading the file → in pytest use capsys.readouterr() to capture in‑memory without filesystem I/O. Coming from C/C++: may think stdout/stderr are global and cannot be redirected per test → pytest's capsys temporarily redirects sys.stdout/sys.stderr for each test, providing isolation.

• pytest fixtures
• output capture
• testing utilities

1. When NOT to use: if you need to capture logging output, use caplog; if you need to capture subprocess output, use subprocess.PIPE or asyncio subprocess pipes. 2. Performance: capturing adds only minimal overhead (a few microseconds per test) as it merely redirects sys.stdout/sys.stderr to StringIO objects; negligible for most test suites. 3. Boundary conditions: capture only works while the capsys fixture is active (i.e., inside the test function or any fixture that uses it); outside that scope sys.stdout/stderr are restored and further output will not be captured.

Enables reliable assertions on program output, ensuring CLI tools and scripts produce the expected user‑facing messages and preventing regressions in visible output.

Think of capsys as a secretary who quietly takes notes of everything your program says out loud or whispers as errors, so you can later check the notes for correctness.

capsys is a built‑in pytest fixture that temporarily replaces sys.stdout and sys.stderr with io.StringIO objects to capture output for the duration of a test.

Move to capsysbinary for binary data or capfd for low‑level file‑descriptor capture when you need more control over output capture.