51Degrees PHP Device Detection On-Premise

PHP Pipeline API

Developer Documentation

Introduction

This project contains the on-premise version of 51Degrees Device Detection for the Pipeline API.

When using on-premise device detection engines in the PHP pipeline, the appropriate extensions will need to be installed.

Dependencies

For runtime dependencies, see our dependencies page. The tested versions page shows the PHP versions that we currently test against. The software may run fine against other versions, but additional caution should be applied.

Data file

In order to perform device detection on-premise, you will need to use a 51Degrees data file. This repository includes a free, 'lite' file in the 'device-detection-data' sub-module, which has a significantly reduced set of properties. To obtain a file with a more complete set of device properties see the 51Degrees website. If you want to use the lite file, you will need to install GitLFS:

sudo apt-get install git-lfs
git lfs install

Then, navigate to 'on-premise/device-detection-cxx/device-detection-data' and execute:

git lfs pull

Install

Device detection on-premise uses a native binary. (i.e. compiled from C code to target a specific platform/architecture). This section explains how to build the binary and how to build the PHP extension that uses it.

Pre-requisites

• php & php-dev
  • sudo apt-get install php php-dev
• Install C build tools:
  • Windows:
  • You will need either Visual Studio 2019 or the C++ Build Tools installed.
    • Minimum platform toolset version is v142
    • Minimum Windows SDK version is 10.0.18362.0
  • Linux/MacOS:
  • sudo apt-get install g++ make libatomic1
• Pull git submodules:
  • git submodule update --init --recursive

Build Steps

Now, we can create the extension. Navigate to the on-premise directory and install it with:

phpize
./configure
make
sudo make install

The Hash engine extension will then be installed into the PHP extensions directory and can then be added to the active php.ini file.

Regenerating swig wrapper files.

Note - this step should not normally need to be performed outside 51Degrees. We include the instructions here for the rare case where it is necessary.

If the SWIG wrapper files need to be regenerated due to new code in the 'device-detection-cxx' submodule, add SWIG=1 to the ./configure step. Note that for backwards compatibility with PHP 5, SWIG 3.0.12 is used for the pregenerated files in this repository. Newer versions of SWIG can be used, provided the extension is being built only for PHP 7.

Currently, The generation of SWIG wrapper files is using a specific revision of SWIG and is done using the following steps:

Checkout and install SWIG:

git clone https://github.com/swig/swig.git
cd swig
git checkout fd96627b2fc65353c03b160efd60fdce864d386c
./autogen.sh
./configure
make
sudo make install

Generate wrapper files:

$:~/device-detection-php-onpremise
cd on-premise
phpize
swig -v -c++ -php7 -module FiftyOneDegreesHashEngine -outdir src/php8 -o src/php8/hash_wrap.cxx hash-php.i

Windows

Although building extensions on Windows is possible, we recommend using the Windows Subsystem for Linux and following the instructions for Linux above.

Configuration

The minimum configuration needed for the extensions is to add it to the active php.ini file, and set the data file. For example:

extension=/usr/lib/php/20170718/FiftyOneDegreesHashEngine.so
FiftyOneDegreesHashEngine.data_file=/path to your file/51Degrees-LiteV4.1.hash

is enough to set up the Hash extension with default configuration options.

NOTE: Make sure to check the highlighted parts. The first is dependent on the PHP version. This location will be printed when installing. The second is the path to where you have stored your 51Degrees data file.

More Options

Running Under Process Manager

When PHP program is run under a process manager such as Apache MPM, php-fpm or any other process manager, MaxPerformance profile must be specified for performance_profile option. HighPerformance can also be used but the performance cannot be compared to MaxPerformance profile.

The reason is that, in other profiles such as Balanced, BalancedTemp, LowMemory or Default, only a part of the data file is loaded into memory so from time to time, calls to read data from disk are required. This call use file handles from a file handle pool which was designed to optimise the performance. However, this pool was created when Device Detection engine module is loaded in the main process, so when process manager spawn child processes in response to incoming requests by forking the main process, this pool is copied to the child processes, causing the file handles to be shared. When multiple processes call to load data from file using shared file handles, problems could occur such as file position being changed unwantedly by other child processes. Thus, we recommend the MaxPerformance to be used as no call to data file is required.

While this is an issue in multi-processing environment, it is not a problem in multi-threading environment as the file handles are not shared and are managed by the file handle pool.

Apache MPM

When running under Apache MPM, only prefork mode is supported, since dl() call is required to load Device Detection engine module and it is not supported under multi-threading environment which worker and event mode does.

Refresh Internal Data

To reload the data file, refreshData() needs to be called as described in this example . Thus, unless a reload is performed in the main process and all child processes, the data will not be updated. There is not a proven solution to do so yet, so we recommend a full server restart to be performed instead.

Examples

Before running the examples, the configuration for the native module MUST be added to the php.ini file. See Configuration.

To run the examples, you will need PHP and composer installed. Once these are available, install the dependencies required by the examples. Navigate to the repository root and execute:

composer install

This will create the vendor directory containing autoload.php. Now navigate to the examples directory and start a PHP server with the relevant file. For example:

php -S localhost:3000 gettingStartedWeb.php

This will start a local web server listening on port 3000. Open your web browser and browse to http://localhost:3000/ to see the example in action.

The table below describes the examples that are available.

Tests

PHPUnit

This repo has tests for the examples. To run the tests, make sure PHPUnit is installed then, in the root of this repo, call:

phpunit --fail-on-warning --display-warnings --log-junit test-results.xml

Performance

Performance tests for the engine can be found in the performance-tests directory. To build this, the dependencies listed on ApacheBench must be installed.

To build the tests, enter the performance-tests directory and run the following.

mkdir build
cd build
cmake ..
cmake --build .

Once the build is completed, ensure that the php.ini file is correctly configured, and run the tests from the build directory with:

./runPerf.sh

This will give two output files: calibrate.out and process.out. These detail the performance for a calibration page where there is no pipeline processing, and for a page which processes the request through a pipeline containing the device detection on-premise engine, and fetches properties.

Development

When making changes to this repository, it may be necessary to link to a local development version of pipeline dependencies. For information on this, see Composer local path.

For exmaple, if a development version of 51degrees/fiftyone.pipeline.core was stored locally, the location would be added with:

"repositories": [
    {
        "type": "path",
        "url": "../../path/to/packages/pipeline-php-core"
    }
]

then the dependency changed to:

"51degrees/fiftyone.pipeline.core": "*"