I am trying to work with basic HTTP using FastCGI and Nginx in c++. I have found the link for fastcgi here: http://chriswu.me/blog/getting-request-uri-and-content-in-c-plus-plus-fcgi/
But there is no clear distinction for HTTP methods like GET and POST. Also, I am unable to figure out how to perform redirection of url using fastcgi. P
I do not have experience with fastCGI and nginx, but since I have used CGI/Apache and took a look of fastCGI samples, I can suggest the following things (and risk to fail with the answer):
GET request is part of URL, so I would parse const char* uri = FCGX_GetParam("REQUEST_URI", request.envp); to check if parameters are given (i.e. if there are key/value pairs after the question mark).
If the previous condition is false, then check if media type in the header is application/x-www-form-urlencoded (meaning it's a POST) and parse HTTP request body to obtain key/value pairs. More info on that can be found at Wikipedia
To perform redirection, use the example but modify response to return HTTP redirection response as described at Wikipedia.
Perhaps fastCGI offers more advanced functions, so all of these can be achieved in a fancy way.
Related
I have been into web development from sometime. But recently came across an old technology, REST. I read various places about REST calls, what I have understood about REST service is,
REST service responds back with JSON or XML data, which can be used on client side for rendering the DOM elements.
It enhances the use of HTTP protocol.
The URL difference between a REST call and normal URL is:
REST CALL: wwww.xyz.com/getCart/12
URL: wwww.xyz.com/getCart.php?cartId=12
I got the basic difference, hitting the URL would render a page at the server end and would return the response, whereas making an AJAX Call to the REST service would simply return a JSON or a XML output which can be parsed at the client end.
My question is:
If I make my .php page to render a JSON string, and the application makes a AJAX call to the php page to get the JSON response back and use it on client side to render the DOM, then what is the difference between REST call and a normal URL call.?
How REST calls are configured differently from normal URLs?
There's a lot of misinformation and confusion about REST. I'm not surprised that these three points are what you understood from the information available, but they are wrong.
REST isn't coupled to any particular data format or media type. The most important constraint in REST is the emphasis on an uniform interface, which means in this case that the server should be able to respond with whatever data format or media type the clients accept. Under HTTP, the client will tell what formats it can understand through the Accept header, and the server should comply or fail with a 406 Not Acceptable error.
In the same way, REST isn't coupled to any particular protocol, although it's often convoluted with HTTP. Again, following the uniform interface, the clients should be able to follow any links provided by the server, for any protocol with a valid URI scheme.
The semantics of URLs are completely irrelevant to REST. All that matters to REST is that an URL identifies one and only one resource. The URL is an atomic identifier and the client shouldn't rely on any semantics embedded in it for any operations. The two examples you give are both valid in REST. There's nothing more or less RESTful about any of them.
To answer your question, under a REST application the difference you imagine doesn't exist. Hitting an URL will return a response. If the client is requesting with an Accept: text/html header, it may return the human-friendly html page to be rendered by a browser. If the client requests with an Accept: application/json or Accept: application/xml, it may return a machine-friendly format to be read by another application.
REST is just an architectural style, there is no technical difference.
One of the things that REST defines is that your URL needs to be atomic identifiers that refer to only one resource.
GET /users/:id (return the user with the given :id)
PUT /users/:id (update the user with the given :id)
Here is an answer about using a framework to make a REST API in php.
Rest puts more emphasis on the verbs, like GET, PUT, POST... You can call one method like
/api/Customers
and depending on the verb you use it will do a get, post, put or delete. You can also make more easy URL's like
/api/Customers/{id}/Orders/{id}
instead of making a method that would be
api/GetCustomersOrders?id=x&id=y.
All Web Services are APIs, but not all APIs are Web services.
APIs are application interfaces, meaning that one application is able to interact with another application in a standardized way.
Web services are a type of API, which must be accessed through a network connection.
REST APIs are a standardized architecture for building web APIs using HTTP methods.
I am wanting to expose a restful web service for posting and retrieving data, this may be consumed by mobile devices or a web site.
Now the actual creation of the service isn't a problem, what does seem to be a problem is communicating from a different domain.
I have made a simple example service deployed on the ASP.NET development server, which just exposes a simple POST action to send a request with JSON content. Then I have created a simple web page using jquery ajax to send some dummy data over, yet I believe I am getting stung with the same origin policy.
Is this a common thing, and how do you get around it? Some places have mentioned having a proxy on the domain that you always request a get to, but then you cannot use it in a restful manner...
So is this a common issue with a simple fix? As there seem to be plenty of restful services out there that allow 3rd parties to use their service...
How exactly are you "getting stung with the same origin policy"? From your description, I don't see how it could be relevant. If yourdomain.com/some-path/defined-request.json returns a certain JSON response, then it will return that response regardless of what is requesting the file, unless you have specifically defined required credentials that are not satisfied.
Here is an example of such a web service. It will return the same JSON object regardless of from where the request is made: http://maps.googleapis.com/maps/api/geocode/json?address=1600+Amphitheatre+Parkway,+Mountain+View,+CA&sensor=true
Unless I am misunderstanding you (in which case you should clarify your actual problem), the same origin policy doesn't really seem to apply here.
Update Re: Comment
"I make a simple HTML page and load it as file://myhtmlfilelocation/myhtmlfile.html and try to make an ajax request"
The cause of your problem is that you are using the file:// URL scheme, instead of the http:// protocol scheme. You can find information about this scheme in Section 3.10 of RFC 1738. Here is an excerpt:
The file URL scheme is used to designate files accessible on a particular host computer. This scheme, unlike most other URL schemes, does not designate a resource that is universally accessible over the Internet.
You should be able to resolve your issue by using the http:// scheme instead of the file:// scheme when you make your asynchronous HTTP request.
I'm making an HTTP server in c++, I notice that the way apache works is if you request a directory without adding a forward slash at the end, firefox still somehow knows that it's a directory you are requesting (which seems impossible for firefox to do, which is why I'm assuming apache is doing a redirect).
Is that assumption right? Does apache check to see that you are requesting a directory and then does an http redirect to a request with the forward slash? If that is how apache works, how do I implement that in c++? Thanks to anyone who replies.
Determine if the resource represents a directory, if so reply with a:
HTTP/1.X 301 Moved Permanently
Location: URI-including-trailing-slash
Using 301 allows user agents to cache the redirect.
If you wanted to do this, you would:
call stat on the pathname
determine that it is a directory
send the necesssary HTTP response for a redirect
I'm not at all sure that you need to do this. Install the Firefox 'web developer' add-on to see exactly what goes back and forth.
Seriously, this should not be a problem. Suggestions for how to proceed:
Get the source code for Apache and look at what it does
Build a debug build of Apache and step through the code in a debugger in such a case; examine which pieces of code get run.
Install Wireshark (network analysis tool), Live HTTP Headers (Firefox extension) etc, and look at what's happening on the network
Read the relevant RFCs for HTTP - which presumably you should be keeping under your pillow anyway if you're writing a server.
Once you've done those things, it should be obvious how to do it. If you can't do those things, you should not be trying to develop a web server in C++.
The assumption is correct and make sure your response includes a Location header to the URL that allows directory listing and a legal 301/302 first line. It is not a C++ question, it is more of a HTTP protocol question, since you are trying to write a HTTP server, as one of the other posts suggests, read the RFC.
You should install Fiddler and observe the HTTP headers sent by other web servers.
Your question is impossible to answer precisely without more details, but you want to send an HTTP 3xx status code with a Location header.
I'm writing a small content server as a web service. There are 2 units - one authenticates the application requesting content and when authentication succeeds, the request is forwarded to the other unit that serves the content.
[1] If I want to do this using CGI
scripts, is there any equivalent of
jsp:forward in CGI?
[2] Suppose if
forwarding is not possible, the
client application shouldn't be able
to request the second unit directly.
What is the proper way to do this?
Another attempt, since you are not after HTTP redirect...
The short answer is: Yes, it is possible.
However, it is highly dependent on the tools you are using. What web server and CGI scripting language you are using?
CGI scripts can do practically anything they want to do, for example they could execute code from other CGI scripts. Thus, they can provide the behavior you are looking for.
CGI (Common Gateway Interface) just describes how a web server starts a CGI script and gives the script input data via environment variables. CGI also describes how the script returns data to web server. That's all.
So if your authorization script wants to delegate some operation to other some script, it is up to that authorization script to implement it somehow. The CGI protocol does not help here.
The concept you might be looking for is called HTTP redirect, where the server sends a response to browser's request, telling the browser to fetch a new page from another URL.
CGI can do HTTP redirects just fine just like jsp:forward. You need just to output the right HTTP headers.
You need to return a 302 response code in HTTP headers, and provide location URL where browser should go next. Have your CGI script output these kind of headers:
HTTP/1.1 302 Redirect
Location: http://www.example.org/
These headers tell browser to fetch a page from URL http://www.example.org/ .
I'm trying to send a file and other POST variables to a PHP script on my server. There are no good resources on Google and the code samples I've found don't work. Preferably without using cURL.
If you're going to roll your own you'd need the relevant RFC for HTTP file uploading (googling on "rfc http file upload" will yield the same result). This RFC also shows how to handle a mix of files and other FORM-data (or POST variables). The problem is of course that you'll probably want to read the MIME RFC as well...
Just a couple of resources make it pretty easy to roll your own
Here is an example of a GET request via ASIO (the C++ networking library in Boost)
Here is the HTTP protocol made really easy
The GET request is how you can view any page on your site. With that code you can download any page and get it as raw text. As you can see it sends a GET header to the server. As explained in that HTTP protocol page, the POST request looks like this
POST /path/script.cgi HTTP/1.0 From:
frog#jmarshall.com User-Agent:
HTTPTool/1.0 Content-Type:
application/x-www-form-urlencoded
Content-Length: 32
home=Cosby&favorite+flavor=flies
To send a file:
You put your URL after post
change the content type to the type of file you are trying to upload.
Set Content-Length to the number of bytes in that file
Append the file after a carrage return (replace "home=Cosby&favorite+flavor=flies")
Another (more quick-n-dirty) solution is to use a utility, via a system() or similar call.
For example the wget utility has a --post-file option.
I'd say roll your own. Its not too complicated.
Capture an HTTP post sent from a browser in Wireshark and reverse engineer as necessary using the spec as your guide. (See Andreas Magnusson's answer below for perhaps more relevant specs.)
I would recommend this approach personally for learning the protocol rather than just going pure spec. Its pretty difficult to learn things just from the spec. I would rather explore the different behaviors by known http clients and try to figure out how things are working by using the spec as my guide.
Format and send the data accordingly over a socket once you're comfortable with HTTP.
Also, If you are not familiar with socket programming, check out Beej's guide to socket programming.
this worked great for me on debian (http get, http post):
http://cpp-netlib.github.com
I use v 0.9.3 that requires boost 1.49