Java Embedded Python (JEP)

Java Embedded Python (JEP) provides access to a locally-installed Python interpreter and makes it very easy to pass values from Java to Python, execute Python operations, functions, and objects, and retrieve Python data suitably translated into Java (e.g. a Python dictionary becomes a Java Map, and a NumPy array is treated as a Java array).

The biggest drawback of using JEP is that its installation on MS Windows requires Visual Studio Build Tools (a 9 GB download) to compile the necessary C++ glue code which links JEP to Python. However, the JEP developers have gone to some trouble to document the required steps with specific installation guides for Linux, OSX, and MS Windows.

I've also included a few additional tips that I picked up when installing JEP on MS Windows 10. They're listed in the "Other Downloads and Links" section at the bottom of this page.

My Examples

Most of these examples come with a corresponding Python script which can be executed in Python directly. That's a good strategy for testing that you have Python, and all the necessary third-party modules, correctly installed. It's also useful to compare the JEP version of code with similar Python.

Other Downloads and Links

• Batch files to compile and run my code: compile.bat and run.bat
   
  
• All the examples in one zipped file (20 KB)
   
  
• The excellent JEP documentation
   
  
• The JEP API documentation.
   
  
• My zipped copy of the JEP libraries, jep-4.1.1.jar and jep.dll, which will PROBABLY NOT WORK (224 KB). These libraries are intimately connected to the version of Python you have installed, and the OS. I've included them only as placeholders; you'll need to replace them with versions that you have compiled on your machine (see below). However, still place the unzipped libs folder in the same directory as the examples and batch files.
   
  
• The JEP installation details for MS Windows can be found here. I'd like to add a few extra tips:
   
  
  1. Before you install JEP, make sure that Python and NumPy are working correctly standalone. In particular, you should be able to execute a script by typing python script in a terminal window.
      
     
  2. There must be a JAVA_HOME environment variable set to point to your JDK.
      
     
  3. Install "Build Tools for Visual Studio 2022", which can be accessed at the bottom of the Visual Studio page. Open the tab marked "Tools for Visual Studio" and click on the relevant "Download" button. After downloading the installer, start it up and it'll open a window offering you numerous installation combinations – go for the "Desktop development with C++" option. The 10 GB download can be shrunk by deselecting the last 3 sub-options (C++ CMake, Testing tools, and C++ AddressSanitizer).
      
     
  4. At last (!), you can install JEP with pip install jep. This will build and install the necessary JEP JARs and DLLs.
      
     
  5. Now its time to replace the two library files, jep-4.1.1.jar and jep.dll in my libs/ folder by copies of your versions. On my test machines they're in Python 3.8's site-packages/ folder, which has a path something like C:\Users\HP\AppData\Local\Programs\Python\Python38\Lib\site-packages\jep-4.1.1-py3.8-win-amd64.egg\jep. Copy them over to the libs/ folder in jep-examples/, deleting my copies in the process.
  
   
  
• My examples use a number of third-party Python libraries, which you'll have to install in Python (using pip). These are NumPy, Matplotlib, Sympy, and Art.