Creating a RESTful API and website with PHP - php

I've got a PHP application I wrote earlier that I'd like to add a RESTful API to. I'd also like to expand the site to behave more like a Rails application in terms of the URLs you call to get the items in the system.
Is there any way to call items in PHP in a Railsy way without creating all kinds of folders and index pages? How can I call information in PHP without using a GET query tag?

If you have some form of mod_rewrite going you can do this quite easily with a .htaccess file.
If you have something like this:
RewriteEngine On
RewriteCond %{REQUEST_FILENAME} !-d
RewriteCond %{REQUEST_FILENAME} !-f
RewriteRule ^(.*)$ index.php?url=$1 [QSA,L]
It will check that the file you are trying to access doesn't exist already. (Say you had a file hello.html that you still needed people to access via http://yoursite.com/hello.html)
Then if the file doesn't already exist it will load your index.php file with the rest of the URL stored in the url variable.
This means you can call something like this http://yoursite.com/pages/edit/24 and it will load index.php with /pages/edit/24 inside the url variable.
That should get you started and you won't need all kinds of folders and index pages, just mod_rewrite, .htaccess file and an index.php that will be used to load whatever you need.

You also might consider to use one of the PHP frameworks with built-in REST support, for example CakePHP.

Quick note in respopnse to Pascal MARTIN: Zend_Rest_Server has absolutely nothing to do with REST. They just do RPC with slightly nicer URLs and call it REST so that it's more trendy.
If you want to do REST, you'll need to do a bit more work yourself as I have not found a good REST library for PHP yet. So inspect $_SERVER['REQUEST_METHOD'] to decide what to do with the called resource, etcetera.

Easiest way would probably using a framework that provides you with REST-oriented functionnalities. I know Zend Framework does that, with the class Zend_Rest_Server, that allows easy creation of a REST server.
I suppose many other frameworks do just the same.
But, if you already have an application that doesn't use a framework (or that is based on a Framework that doesn't embed that kind of class), a couple of URLrEwriting rules would do just fine ; you'd just have a bit more work to map URLS/parameters to classes/methods :-(

The design pattern you are looking for is called a front controller.
In its simplest form you use mod_rewrite to pass the incoming requests and pass it to a single php script. The url is then parsed with regular expressions and mapped to different response actions. However mapping an existing application may require extensive rewriting.
If you want to play around with this concept I recommend the Silex microframework.

Related

REST-like URLs with PHP

According to what I've read about RESTful services, one is supposed to be able to get a "resource" (like a database record) using a request to a URL with the record id at the end, like this: localhost/my_records/1. Accessing such a URL would display a web page with the record information for id 1.
I need to implement something like this in PHP/Apache/MySQL but I am still relatively new to the language. The most I can manage to create is something like
localhost/my_records/index.php?id=1.
Creating a ["/id"] url seems to require creating that folder. Is there any way to achieve this?
You can have a RESTful service with any kind of URL, you don't need to do a URL with the id at the end, but this type of URL is better for convenience and good design principles.
Otherwise this type of URLs is also know as "friendly URL", most PHP frameworks provide this type of URL, for example, Symfony, Code Igniter, Laravel... provide this functionality.
If you want this type of URL without use any PHP framework, you can check this answer and responses: How to create friendly URL in php?
In my opinion, a good framework to implement a good REST Api is Code igniter + Phil Sturgeon Rest Client. More info:
http://ellislab.com/codeigniter
https://github.com/philsturgeon/codeigniter-restclient
Configure your webserver to route all requests through a single PHP script. This will allow you to parse the retrieved URL. Since you are using Apache mod_rewite is the way to go.
RewriteEngine On
RewriteCond %{REQUEST_FILENAME} !-d
RewriteCond %{REQUEST_FILENAME} !-f
RewriteRule ^(.*)$ index.php [L]
This will route all requests for paths/files that do not exist (remove lines 2 and 3 if you want anything routed to that script) through index.php. In the script you can then access the accessed URL via $_SERVER['REQUEST_URL']
In essence what you are looking for is something commonly referred to as request routing. Typically this is implemented by a combination of configurations at both the server level and app level.
Some sort of server redirect (using RewriteRule or FallBackResource directives) which directs all requests (or requests of a certain format) to single application resource which is to handle it.
Application logic to parse the URI and HTTPS action (GET, POST, PUT, DELETE, HEAD, etc.) and determine how to route the request (i.e. which controller should handle it)
Oftentimes there is some sort of logic to format the response according to the format requested in the request headers (i.e. JSON/XML/HTML)
There are a number of PHP frameworks available which do this, that at least you could look to for guidance (if not to just use them altogether).

Url representation in php

I am used to representing embedded url information like this:
http://test.com/reports/statement.php?company=ABC&q=1
how would I do it like this instead?
http://test.com/reports/ABC/Q1
You need to use Apache mod_rewrite to achieve this.
If your server has it enabled, you could do something like this in .htaccess:
RewriteEngine on
RewriteRule ^([^/\.]+)/([^/\.]+)/?$ /statement.php?company=$1&q=$2 [L]
You can use $_SERVER['PATH_INFO'] to access anything in the URL docpath after the address of your script.
e.g.
http://test.com/reports/statement.php/ABC/Q1
...then in statement.php you would have the string "/ABC/Q1" in $_SERVER['PATH_INFO']
Of course, you'll need to setup your webserver to match the URL and target the correct script based on the HTTP request.
As stated by others, you have to use url rewriting.
Usually a php application that make use of it, it applies the pattern called Front Controller.
This means that almost every url is rewritten to point to a single file, where the $_SERVER['PATH_INFO'] is used to decide what to do, usually by matching with patterns you define for your actions, or return a 404 error if the url doesn't match any of the specified patterns.
This is called routing and every modern php framework has a component that helps doing this work.
A smart move would also be providing a tool to generate urls for your resources instead of handwriting them, so that if you change an url pattern you do not have to rewrite it everywhere.
If you need a complex routing system, check out the routing component of some major frameworks out there, e.g. Symfony2, Zend Framework 2, Auraphp.
For the simplest php router out there, check instead GluePHP. The codebase is tiny so you can make yourself an idea on how the stuff works, and even implement a small router that fits your needs if you want to.

Routing REST requests without framework?

I have been reading the article to learn how to build a rest API:
http://www.gen-x-design.com/archives/create-a-rest-api-with-php/
At one point it says "Assuming you’ve routed your request to the correct controller for users"
How can I do this without a framework?
I am writing a REST API that I can interact with from a different application. I ready the tutorial above, and it makes sense mostly, but I don't exactly understand what it means to route my request to the correct controller for users.
Assuming you are using Apache, you can accomplish this easily using a combination of mod_rewrite and some PHP-based logic. For example, in your .htaccess or vhost definition, you could route all requests through a single handler, possibly index.php:
# Don't rewrite requests for e.g. assets
RewriteCond %{REQUEST_FILENAME} !-f
RewriteRule ^(.*+)$ index.php?q=$1 [L]
...then in your index.php file do something like:
$target = $_REQUEST['q'];
/* parse the request and include the appropriate controller PHP */
For example, a request for /products/1234 might result in a controllers/products.php handler being included. That handler could then act on product 1234. Because you're using REST, you shouldn't need to be concerned with the original request having a query string parameter.
There are multiple ways to accomplish what it sounds like you're trying to do, this is just one of them. Ultimately what you go with will depend on what your specific requirements dictate. The above pattern is fairly common however, many frameworks use it or something like it.
I think this is a matter of terminology. Every code with some level of generalization can be called "framework". And since you're asking about "routing", which provides a starting level of generalization, every implementation becomes a framework.
If you don't want to use existing fully-fledged frameworks, you can elaborate your own light-weight-implementation. Here is some articles to start:
Write your own PHP MVC framework
PHP MVC framework in one hour
(the author has decided to remove this post because he thinks that using a modern fully-fledged framework is more appropriate way of programming such things, yet some people find such simple and stripped-down approache more suitable in many aspects: learning, efficiency, no huge dependencies, etc); the post is available on some other sites, for example, in the wayback machine or copies
The Model View Controller in PHP
All these intros include explanations of the routing mechanizm and demonstrate its implementation.
Basically, a router is a kind of internal "DNS" (in a figurative sense) inside your application. When a request arrives to your site, you need to dispatch it to appropriate worker class, according to that request properties. This is the router's task.

Alternatives for accessing PHP functions

Currently, I'm developing an application using standard PHP. As such, there are assorted functions that are called from forms. Some examples include registration, login, logout, and others.
Rather than creating a separate file for each function, I'm using a phpscripts.php file in the includes folder. To execute a function from html input, I pad the URL with ?action=functionName. Inside the script file, I use a switch/case tree to execute the proper function based on the S_GET['action'] variable.
Is there a better way to implement HTML calls to PHP functions?
No, that is the correct way to interface a HTML form with a PHP script. Other alternatives would include employing URL rewriting (to convert something like /actions/functionname/param to ?action=functionname&param=x), but the basic principle remains the same.
For your sanity... consider having one script that handles all requests like your above. Something like /process_action.php. You do handle unified actions such as: checking to make sure the function exists, checking for a logged in user, connecting to a database, creating shared resources, etc...
If you're using Apache, you can make the URLs pretty by using mod rewrite. You can turn
/actions/function/?restofquery
into
/process_action.php?action=function&restofquery
by using something like the following placed in a .htaccess file in your website.
RewriteEngine On
RewriteCond %{QUERY_STRING} ^(.*)$
RewriteRule ^/actions/([-_a-zA-Z0-9]+)/?$ /process_action.php?action=$1&%1 [L]
As a side note, your question sounds like you're on the verge of discovering the model/view/controller (MVC). There are several great frame-works, but don't let that discourage you from rolling your own. Learn by doing :)

passing GET parameters that look like URI directories

I've seen a lot of URIs that look something like this:
www.fakesite.net/stories/1234/man_invents_fire
and I was wondering if the /1234/man_invents_fire part of the URI are actually directories or if they are GET parameters (or something else). I've noticed that a lot of times the /man_invents_fire segment is unnecessary (can be removed with no consequences), which led me to believe that the /1234/ is the id number for the story in a database table (or something along those lines).
If those segments of the URI are GET parameters, is there an easy way of achieving this?
If they aren't, what is being done?
(also, I am aware that CodeIgnitor gives this kind of functionality, but I was curious to find out if it could be easily achieved without CodeIgnitor. I am, however, generally PHP, if that is relevant to an answer)
Thanks
Easiest thing to do is route everything into a main index.php file and figure out your routing from there by running $pieces = explode("/", $_SERVER['REQUEST_URI']);
After installing/enabling mod_rewrite, make sure allow override is not set to false in your apache config (to allow .htaccess to be read), then throw this in your docroot's .htaccess file.
<ifModule mod_rewrite.c>
RewriteCond %{REQUEST_FILENAME} !-s #Make sure the file doesn't actually exist (needed for not re-routing things like /images/header.jpg)
RewriteRule . /index.php [L,QSA] #re-route everything into index.php
</IfModule>
That is called url rewriting, google for it, you will find a lot of information about that.
Implementing this in PHP is typically done via an .htaccess file and using apache's mod_rewrite module.
They make the url like that so that people can easily bookmark it, and it can return safely in the search.
Depends on what language you're using to decode it. In this case, it appears "stories" is the main script, and "1234" is the id, and "man_invent_fires" is the title.
If you're using php, you can use the $_SERVER['PHP_SELF'] or $_SERVER['REQUEST_URI'] variable to decode it.
If you're planning to make a website like that, certain safety must be kept in mind. Look them up in google, but key one to look out for is sql injectors.
Just like permalinks in WordPress, this is done typically done via Apache's mod_rewrite (or an equivalent thereof if not using Apache); however, you can also use a 404 error page handler to achieve the same result (but this is not usually recommended).
Typically, all page requests are redirected to a gateway that parses the requested URI to determine if it fits the specified pattern (in your case likely to be /{category}/{article_id}/{article_title}). From there, the gateway can typically use just the article_id to retrieve the appropriate content.
Depending on the system, category and article_title can usually be thrown away/ignored and are typically for SEO value; however, in some cases category might be used to augment article_id in some way (e.g.: to determine what DB table to query, etc).
MVC's, like Zend, also use a similar technique to determine which controller and method therein to execute. An example format for this type of use is /{module}/{controller}/{method}; however, this is highly customizable.
Well, you are kind of right in assuming that the 1234 and main_invents_fire are parameters. They are not truly GET parameters in the sense that the HTTP protocol describes them but they accomplish the same task, while keeping the URL "friendly". The technique is called URL rewriting and the web is full of info on this these days..
Here's an article about friendly URLs in PHP but I'm sure googling for the topic will render more useful results.
As some background information in addition to the answers before me, a URL is just that - a 'Uniform Resource Locator'. Although in the old days, it often used to map 1:1 to a file/directory structure, all that is required in the HTTP spec is to identify a certain resource. Basically that means that, given a certain string, it should indicate a certain 'resource' on the server. In practice, in a HTTP environment this is usually implemented with a rewriting mechanism such as mod_rewrite. RFC's such as this one: http://www.ietf.org/rfc/rfc1738.txt give a nice, albeit abstract, overview. The concepts only come to life after designing and implementing some non-obvious uses, though.
If you are using Symfony, then you can use the routing feature to do this.

Categories