Is Python an Interpreted language?

When I decided to study it, I felt the same as Naruto learning the Rasengan: It was difficult to learn: Naruto took 3 years to learn; and I took 10 😂, since I firstly heard the words “interpreted”/“compiled” language when I was learning R.

Python, itself is an interpreted language, and there are many nuances that I want to discuss.

Let’s start by defining what is a programming language.

1 Programming language

A programming language can be defined by two main pillars:

1. Specification (Rules):
   1. Syntax: how the code is written; some languages may be more verbose than others. For example, in C you need to define the type of each variable and finish the line with ;. Whereas Python does not require it, but indentation is obligatory for your code to run.
   2. Semantics: It defines basic operations, such as, the operator +, does what we think it does: add two values together (in case of floats/ints) or concatenate two lists in Python.
   3. Data Types & Core concepts: In Python, we have dict, but we do not have dataframes; R is the opposite.
2. Implementation:
   1. Machines do not understand source code, only binary machine code (0s and 1s). So, for every programming language, a tool is necessary that translates the source code into binary code that can be:
      1. Compiler: It will translate upfront the entire source code into a standalone binary executable. It is a characteristic of C, Rust and Go programming languages;
      2. Interpreter: It will translate the source code into binary code at the runtime line by line. This is the primary implementation strategy for Python, JavaScript and R.

2 Python’s implementation

What makes a programming language interpreted or compiled is its implementation. Python has different implementations, such as:

1. Nuitka and Cython make the Python a compiled language, so, all machine code is known ahead of time.
2. PyPy uses JIT (Just-In-Time) compilation: it takes the Python source file, converts it to bytecode¹, and directly compiles it into machine code during the runtime (not ahead of time). It is a just-in-time compiler, rather than an ahead-of-time compiler.
3. CPython, the official implementation, first compiles the Python source code to bytecode². Then the Python Virtual Machine (PVM) - itself a program in C - reads that bytecode and hands each instruction off to an already-existing C function inside the interpreter that carries it out. That’s why Python is known as an “interpreted language”.

• Bytecode plays a vital role here: Python caches the first time a module is imported, saving it as a .pyc file so it does not need to be recompiled on every subsequent import.
• It never translates Python code into machine code directly; instead, the PVM executes the bytecode by invoking the corresponding pre-compiled C function for each instruction.

Finally, what makes a language interpreted or compiled is the implementation, not the language itself.

This picture describes the whole process of executing Python source code:

There is more! Keep reading Part 2!

Happy Coding!

Footnotes

1. bytecode is an intermediate low-level code that can be read and interpreted by a machine, being originated by a source code.

2. you can inspect the bytecode of any Python function by calling dis.dis(<your_function>), see is python a dynamic language.