Cprofile

cProfile is a built-in Python module that provides a deterministic profiler for Python programs. It records how many times each function is called and how much total time is spent executing in that function, including time spent in subsequent called subfunctions. The results are designed to help identify bottlenecks in Python code and optimize performance. It focuses on Python-level execution and is generally preferred for CPU profiling of Python programs due to relatively low overhead compared to pure-Python profilers.

How it works: cProfile instruments Python code to collect per-function statistics during a run. It reports information such as the number of calls, the time spent in a function excluding subcalls (tottime), and the total time spent including subcalls (cumtime). The collected data can be analyzed to determine which functions contribute most to runtime and how call flows contribute to overall performance.

Usage: The profiler can be run from the command line or within a script. From the command line, you can profile a script with: python -m cProfile script.py. In code, you can use a Profile object: import cProfile, pstats; pr = cProfile.Profile(); pr.enable(); main(); pr.disable(); ps = pstats.Stats(pr); ps.sort_stats('cumtime'); ps.print_stats(20). This prints the top 20 functions by cumulative time. The Stats object supports various sort options and can be directed to a file or stream for later analysis.

Limitations and extensions: cProfile profiles Python code and may not reflect time spent purely in C extensions. For finer-grained insights, options such as line-by-line profilers can be used in addition to cProfile, depending on the debugging needs.