Shulou(Shulou.com)06/03 Report--

The main content of this article is "introduction to the principle and architecture of PPython". Interested friends may wish to take a look. The method introduced in this paper is simple, fast and practical. Next let the editor to take you to learn the "introduction to the principle and architecture of PPython"!

Introduction

Python and PHP are both widely used languages, each with its own strong points, so it is expected that the combination of the two can achieve richer results.

Calling Python in PHP to implement some processing is a niche but practical requirement. At present, a lot of information can be found on the Internet that exec () (including system (), shell_exec (), passthru (), etc.) executes external Python files, but this is only a general way, the call cost is relatively high, in each call, you need to load the entire Python interpretation environment.

Developers with such requirements are well suited to take a look at PPython, a technology that fundamentally combines PHP with Python.

PPython first appeared in https://code.google.com/p/ppython, and the author ported lajp (a technology of PHP combined with Java) to Python.

The project was originally founded in 2012 and seems to have been out of maintenance for many years, but its ideas and results are commendable so far, so the project was migrated from the discontinued Google Code to GitHub and reissued and maintained under the original Apache license.

A few days ago, the author made an attempt on this and realized the convenience and ease of use of PPython.

Principle and architecture

There are two different socket mechanisms for PHP and Python communication: TCP socket and UNIX socket. UNIX sockets are Unix/Linux native sockets and have the following characteristics compared to TCP sockets:

Can only communicate on the same host (IPC), not across hosts

The transmission speed is higher than the TCP socket

The server only provides services to the local machine (there is no external listening port), which is relatively secure and easy to manage.

PHP and Python each have their own data types defined within the language. Usually, when the PHP process communicates with the Python process, it needs to transcode. If this kind of conversion is implemented by the application itself, it will add a lot of additional burden from development efficiency to running performance.

PPython's processing of communication between PHP and Python supports both TCP socket and UNIX socket mechanisms, taking into account communication efficiency and distribution, transcoding is handled uniformly by the service, and Python provides format compatibility for PHP data types, so that PHP developers do not have to worry about underlying communication.

Because of the language GIL feature of Python, the efficiency of multithread in the same process is not high. PPython can deploy services according to the needs of the project, and run Python in multiple processes to improve the overall performance of the application.

Usage

The code for PPython can be downloaded from the above project repository.

Among the downloaded files, the following three are the core codes of PPython, and their functions are as follows:

Php_python.py,Python process master file, complete the Python side to listen to the request and run the return

Process.py,Python core class to implement key processes such as Python internal process calls and data structure conversion between PHP and Python

The php_python.php,PPython client, which is referenced by the PHP side, can be called directly using the PPython function.

Place the above files in any directory. First modify the port on which you are going to run PPython, and the listening port is unlimited, as long as the modifications on both ends of php_python.py and php_python.php are the same. The author uniformly changed to 10240.

Running php_python.py in the current directory will run a PPython service as long as the Python environment is normal.

-PPython Service- Time: 2019-05-13 22:24:09---Listen port: 10240charset: utf-8Server startup...

When php_python.php is introduced into the PHP side, you can use the ppython function to communicate with the previously started PPython service. The incoming request is processed by the PPython service calling Python and the result is returned. For example, $res = ppython ('test::go') is the call to the go function in test.py, or more parameters can be added. The second parameter will pass more parameters to the tuned function.

Php_python.py is the global code that runs directly after PPython starts. Global configuration or general processing after process startup is written here, such as establishing a database connection in the native code, which should be optimized according to the situation in the project.

However, the main interesting aspect of Python is not only to describe business functions like PHP, but also to provide more complex data structures for computing tasks needed in artificial intelligence and other fields, so the combination of the two can bring more application scenarios to PHP.

Improve

In addition, the native php_python.py has some shortcomings. The author encountered three problems in calling custom code with ppython, and solved them accordingly:

Hongmeng official Strategic Cooperation to build HarmonyOS Technology Community

Complex (plural class) is not supported. Plural is a data type in mathematics, which mainly includes real (real part) and imag (imaginary part) data. Although it is rarely encountered in daily life, AI and various professional research fields may not be rare. There is a complex class in Python, which can calculate complex numbers directly, but PPython serialization and deserialization do not deal with complex. In order for the data included in complex to be returned normally, you only need to add serialization processing that meets the PHP requirements in the z_encode () method of process.py, as follows:

Elif isinstance (p, numpy.complex128):

T1 = str (p.real)

T2 = str (p.imag)

Return 'Orange7: "complex": 2: {SLAR4: "real"; dappl4% splaces4: "imag"; dpurs% s;}'% (t1Magol T2)

Ndarray (multidimensional array) is not supported. Compared with complex,ndarray is much more common, I believe that where the use of various computing functions of Python, ndarray is unavoidable, and even ndarray has achieved Python to a certain extent. But the original php_python.py does not recognize ndarray. However, it is not difficult to solve, find the z_encode () method in process.py, plus the following paragraph, you can directly convert ndarray into an array (numerical index) that meets the requirements of PHP.

Elif isinstance (p, numpy.ndarray):

S =''

I = 0

For d in p:

S + = 'iGV% dbot% s'% (iGrainzyme encode (d))

I + = 1

Return "av% d: {% s}"% (len (p), s)

The original code is not very robust. For example, if the data is of ndarray type, if p = = None: ValueError: The truth value of an array with more than one element is ambiguous. Use a.any () or a.all (), because the result of p = = None is also ndarray, does not return false, change the judgment method to if p is None: to avoid errors.

Accordingly, the PHP side should also pay attention to serialization and deserialization.

When dealing with object data such as complex in the reply, if the corresponding class is not defined in the system, PHP can be deserialized, but it will be displayed as "incomplete object", vardump can see real and imag data, but can not be operated directly. After defining the complex class, it is parsed according to the specified class, which is the same as the general object in PHP and can easily carry out all operations.

At this point, there is no problem with the function of PHP and Python.

Add: register as a service

Starting php_python.py under the command line is mainly convenient for debugging, and you can see the observation and feedback information. After all, it is not convenient to start PPython manually in the production environment. You can configure PPython as a service, and modify the port to configure different PPython for different applications.

It is easy to register a process as a service under Linux, as long as / usr/lib/systemd/system/ppython.service is established, as follows:

[Unit] Description=PHP-Python ServiceAfter=network.target remote-fs.target nss-lookup.target [Service] ExecStart= {PPYTHON_PATH} / php_python.py [Install] WantedBy=multi-user.target

Where {PPYTHON_PATH} should be changed to the actual path.

Summary

With PPython, you can discard the shell call exec () and return the development to the logic itself.

At this point, I believe you have a deeper understanding of the "introduction to the principle and architecture of PPython". You might as well do it in practice. Here is the website, more related content can enter the relevant channels to inquire, follow us, continue to learn!

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.