Understanding Python .py Script Execution and Bytecode Generation

Python enthusiasts should know that a .py script is not compiled into a binary executable; instead the Python interpreter reads the source, compiles it to bytecode, and then the virtual machine executes that bytecode.

The interpreter has three notable characteristics: the source code stays far from the hardware, bytecode is generated at runtime and run by the VM, and each execution involves a loading and linking step, which can make it slower than compiled languages.

The VM executes a script by loading and linking the module, translating the source into a PyCodeObject (the bytecode) stored in memory, reading and executing the instructions, and finally deciding whether to write the bytecode back to disk as a .pyc or .pyo file.

Whether a .pyc/.pyo file is created depends on how the script is run. Running a directory (e.g., python path/to/projectDir) creates a __pycache__ folder with bytecode files for all modules; running a single file creates a bytecode file only for that module.

When a module is imported, Python checks the modification time of the existing bytecode file; if it matches the source, the bytecode is used directly, otherwise the source is re‑compiled and the bytecode file is updated.

.pyc files contain compiled bytecode, which loads faster and can obscure source code; .pyo files are further optimized (smaller) and are useful for embedded systems. Both are placed in __pycache__.

Common optimization flags are -O (produces .pyo), -OO (removes docstrings), and -m (run a module as a script).

To generate a .pyc file from the command line you can use:

python -m py_compile /path/to/your_script.py  # generate .pyc (or batch with a pattern)

Or from Python code:

import py_compile

py_compile.compile(r'/path/to/your_script.py')

# The function signature is py_compile.compile(file[, cfile[, dfile[, doraise]]])

To generate a .pyo file you can run: python -O -m py_compile /path/to/your_script.py Both commands create a __pycache__ directory in the script’s location. .pyd files are Python dynamic‑link libraries and are outside the scope of this discussion.

Other ways to produce bytecode include using the compileall module or IDEs that automatically compile scripts.

Understanding Python’s execution model and bytecode generation helps when designing caching systems, reducing unnecessary runtime overhead, and keeping code up‑to‑date.