RoboHydra: advanced techniques
Introduction
This is the third article we've published about RoboHydra. The first two introduced RoboHydra and explored common, simple uses of RoboHydra; this one however is going to focus on advanced techniques you can use for more complicated and custom purposes. In particular, this article will explain some of the more interesting things you can do with the low-level RoboHydra heads (RoboHydraHead objects) that we explored briefly in the second article. This kind of head doesn't do anything concrete, instead receiving a JavaScript function that will do whatever you choose. That function receives the request and response objects (and, optionally, a third argument, as we'll see later) and can do pretty much anything you can implement in JavaScript.
Preparing the environment
To prepare your computer to execute these examples, simply create a directory (we'll call it roboexamples throughout this article), make sure you have Node.js installed, open a terminal, go to your new directory and type npm install robohydra. As always, that'll create a directory called node_modules, but you don't need to do anything in there. Simply stay in roboexamples.
Next download the demo files for this article (ZIP) and uncompress them in your roboexamples folder. After that, you should have a directory called roboexamples/files.
The simplest RoboHydraHead head
The handler function you pass to a RoboHydraHead is supposed to decide what to send as a response (maybe based on the contents of the request object), then write that response in the response object. Let's look at a trivial example that always returns a given response text without setting any headers or checking anything in the incoming request. This would be roughly equivalent to a RoboHydraHeadStatic head:
var heads = require('robohydra').heads,
RoboHydraHead = heads.RoboHydraHead;
exports.getBodyParts = function(conf) {
return {
heads: [
new RoboHydraHead({
path: '/.*',
handler: function(req, res) {
res.send("Always the same response text");
}
})
]
};
};Save the code above to a file named roboexamples/robohydra/plugins/simple/index.js, and create a file roboexamples/simple.conf with the following contents:
{"plugins": ["simple"]}Now, type the following command in your terminal (remember, you have to be in the roboexamples directory) and you'll have a simple RoboHydra server replying "Always the same response text" for any request:
./node_modules/.bin/robohydra simple.confSpicing it up
However, that's not terribly exciting, is it? Now we'll try something a bit more involved: we'll make a head that listens in /guess-number and will expect a GET parameter guess. According to whether that number is correct or not, we'll show a message. Plus, we'll print the messages in HTML and set the correct headers so the browser knows how to interpret the output:
var heads = require('robohydra').heads,
RoboHydraHead = heads.RoboHydraHead;
exports.getBodyParts = function(conf) {
return {
heads: [
new RoboHydraHead({
path: "/guess-number",
handler: function(req, res) {
var message = "Sorry, try again";
// Chosen by fair dice roll, guaranteed to be random
var secretNumber = 4;
if (parseInt(req.queryParams.guess, 10) === secretNumber) {
message = "You guessed right!!!";
}
// By convention all headers are in lowercase
res.headers["content-type"] = "text/html";
res.headers["x-comic-number"] = "221";
res.send("<!doctype>\n<html><body>" +
message +
"</body></html>");
}
})
]
};
};Replace the contents of roboexamples/robohydra/plugins/simple/index.js with the above code,
and restart RoboHydra (i.e. kill with Ctrl-C, start again by typing ./node_modules/.bin/robohydra simple.conf). If you now open URLs like http://localhost:3000/guess-number?guess=4 or http://localhost:3000/guess-number?guess=1 in your browser, you'll see the two different results.
You can write your own handler functions to do whatever you please. The response object has the properties headers and statusCode for the headers and status code respectively, and if you want to write the response body, you have two possibilities:
- First, you can send the response in chunks, using the
writemethod in the response object (and then signal when you're done with your response by calling theendmethod), in case you need streaming or you want to simulate slow networks - Second, you can send the response all at once by calling the
sendmethod in the response object (which is equivalent to a singlewritecall plus anendcall).
If you need to inspect the incoming request you can check the properties headers, queryParams and bodyParams on the request object. See the RoboHydra documentation for full details.
Storing values for later
Another interesting possibility is storing values in variables outside of the function, so we can check them later. For example, we might want to have one URL that saves whatever comes to it, and another URL that shows what we have saved so far. Consider the following example:
var heads = require('robohydra').heads,
RoboHydraHead = heads.RoboHydraHead;
exports.getBodyParts = function(conf) {
var visits = {};
return {
heads: [
new RoboHydraHead({
path: "/articles/:articleId",
handler: function(req, res) {
var id = req.params.articleId;
visits[id] = visits[id] || 0;
visits[id]++;
res.headers["content-type"] = "text/html";
res.send("<!doctype>\n<html><body>" +
"<h1>" + id + "</h1>" +
"<p>Text for article '" +
id + "'</p></body></html>");
}
}),
new RoboHydraHead({
path: "/admin/visits",
handler: function(req, res) {
res.headers["content-type"] = "text/html";
res.write("<!doctype>\n<html><body>" +
"<h1>Visits per article</h1>" +
"<ul>");
for (var articleId in visits) {
res.write("<li>" + articleId + " (" +
visits[articleId] + " visits)</li>");
}
res.write("</ul></body></html>");
res.end();
}
})
]
};
};
In this example, we have URLs like /articles/foo, /articles/bar, etc. and the URL /admin/visits that shows how many times articles have been visited. In the first case, we show some stub text for the article and count the visit (note how we can use the params object in the request object to access interesting parts of the URL). In the second case, we show all the information we have gathered so far through the visits to the other URLs.
To try it out, replace the contents of roboexamples/robohydra/plugins/simple/index.js with the new code, restart RoboHydra, and load several URLs with structures like localhost:3000/articles/robohydra-intro, localhost:3000/articles/robohydra-advanced, etc. Then go to localhost:3000/admin/visits to see the results.
Talking heads: the next function
One very powerful feature we haven't talked about yet is the next function. Remember how when a request comes, RoboHydra tries to match the request with the first head, then the second, etc. until it finds a matching head. That means that even if there is more than one head matching the request, only the first one gets to serve it. However, the next function, which is passed as an optional third parameter to the handler function, allows a head to call the next head that matches the request. Let's see this with an example. Imagine you have a RoboHydra server with the following heads:
| Name | Path |
|---|---|
| onlyFoo | /foo |
| onlyBar | /bar |
| atLeastFoo | /foo.* |
| catchAll | /.* |
When a request for /foo comes, RoboHydra will process the request with the head onlyFoo, even though also atLeastFoo matches. However, if the onlyFoo calls the next function, the head atLeastFoo will process the request. Once it's finished, the onlyFoo will be able to inspect the response atLeastFoo produced, and decide if it wants to return that response, or tweak it and then return it, or ignore it completely.
The next function can be called as many times as we want to (but typically only once, if at all), and receives two parameters: a request object and a response object. RoboHydra will then try to find any heads below the current one that can process the request object we passed. The head will execute normally, and the response will be written to the response object we passed as a parameter. We could of course pass the same request and response objects we received ourselves, or pass other objects. This is useful in a variety of situations, but most commonly to tweak the URLs of a given request before processing, or to tweak the response before it is sent to the client.
We'll illustrate this first of all with a trivial example. Let's say we have a head A that depends on a GET parameter called type, but we don't always want to pass it: when we don't pass it, we want it to default to basic. We can write a small head B that sets the type GET parameter only if it's not already set, then pass the request to head A for actual processing. Head A will be very simple in this example, but it could be a proxy or anything else. Replace the contents of roboexamples/robohydra/plugins/simple/index.js with the below code block, and restart RoboHydra:
var heads = require('robohydra').heads,
RoboHydraHead = heads.RoboHydraHead;
exports.getBodyParts = function(conf) {
return {
heads: [
new RoboHydraHead({
name: "defaultTypeSetter",
path: "/.*",
handler: function(req, res, next) {
if (req.queryParams.type === undefined) {
req.queryParams.type = "basic";
}
next(req, res);
}
}),
new RoboHydraHead({
name: "content",
path: "/.*",
handler: function(req, res) {
res.send("The type in the request was: '" +
req.queryParams.type +
"'");
}
})
]
};
};
Now load URLs into your browser like http://localhost:3000/foo?type=bar and http://localhost:3000/foo and see the different results.
Let's look at another example. Let's say we are superstitious and we don't like the number 3. So, when we're browsing Dev Opera we don't want to see the contents of page 3, and instead want the contents of page 4. As we can see if we check on this site, the pagination URLs look like ../../../?page=number. What we will do, then, is create a head that will replace any page=3 we see in the URL with page=4 before passing it on to the proxy head that serves Dev Opera. Replace the contents of roboexamples/robohydra/plugins/simple/index.js again — this time with the following code — and restart RoboHydra:
var heads = require("robohydra").heads,
RoboHydraHead = heads.RoboHydraHead,
RoboHydraHeadProxy = heads.RoboHydraHeadProxy;
exports.getBodyParts = function(conf) {
return {
heads: [
new RoboHydraHead({
name: "superstitiousPagination",
path: "/.*",
handler: function(req, res, next) {
req.url = req.url.replace(/page=3/, "page=4");
next(req, res);
}
}),
new RoboHydraHeadProxy({
name: "devOpera",
mountPath: "/",
proxyTo: "http://dev.opera.com",
setHostHeader: true
})
]
};
};Now go to http://localhost:3000/ and start clicking on the next pages. Notice how when you click on page 3, page 4 is shown instead (you can tell because the pagination links mark page 4, not 3, as the current one), even though in the address bar you can still see ?page=3.
In this example we only modified the GET parameters, but we could have changed the URL of the request, or any other property. If we had changed the URL, the request would have been processed by the first head matching the new URL, regardless of what the original URL was. In the following sections we'll see how to inspect and tweak the response of the second head.
Inspecting responses
Apart from modifying the request before sending over to the next head, we can also inspect the response we received from it. To do that, we pass a fake response object to the next head, inspect that response, and decide if we want to use it. We could then make our own response, or modify the existing response before sending it back to the client.
Let's look at another simple example: we'll create a simple file server with a custom 404 error message. To do that, we'll need two heads: a regular RoboHydraHeadFilesystem head and a second head that will call the first and check the status code. If it's 404, it will return our own error message. If not, it will simply serve whatever the filesystem head returned. Replace the contents of roboexamples/robohydra/plugins/simple/index.js with the following code and restart RoboHydra:
var robohydra = require('robohydra'),
heads = robohydra.heads,
Response = robohydra.Response,
RoboHydraHead = heads.RoboHydraHead,
RoboHydraHeadFilesystem = heads.RoboHydraHeadFilesystem;
exports.getBodyParts = function(conf) {
return {
heads: [
new RoboHydraHead({
name: "special404",
path: "/.*",
handler: function(req, res, next) {
// Create a fake response object to pass to the
// filesystem head so we can capture its
// response. Once we have it, decide if we should
// respond with our error message, or with
// whatever the filesystem head returned
var fakeRes = new Response().on('end', function(evt) {
if (evt.response.statusCode === 404) {
res.send("OH NOES, YOU FOUND A DEAD LINK. " +
"Have you been playing Zelda again?");
} else {
res.copyFrom(evt.response);
res.end();
}
});
next(req, fakeRes);
}
}),
new RoboHydraHeadFilesystem({
name: "realFileServer",
mountPath: "/",
documentRoot: "files"
})
]
};
};As you can see, we're using the Response class to create the fake response object we'll pass to the filesystem head. In that fake response object we hook a function to the end event, and decide what to do there (pass the same response the filesystem head returned, or make our own response).
If you make requests to http://localhost:3000/ or http://localhost:3000/foo.html you will get a normal page (it's included in the ZIP file you downloaded before!). Otherwise, it will return our custom error message.
Modifying responses
As we can inspect the response given by the "next" head, we can also tweak its contents before sending them back to the client. Imagine we want to modify the code in this very site, Dev.Opera, to emphasize that it's all about developers. We could, for example, replace the word "developers" on the frontpage with "DEVELOPERS, DEVELOPERS, DEVELOPERS, DEVELOPERS". This can be done with the RoboHydraHeadFilter head; however, we'll implement this with the RoboHydraHead as it's easy enough and will demonstrate how to tweak responses with the next function.
Replace the contents of roboexamples/robohydra/plugins/simple/index.js with the following
code, restart RoboHydra and go to http://localhost:3000 in your browser:
var robohydra = require('robohydra'),
heads = robohydra.heads,
RoboHydraHead = heads.RoboHydraHead,
RoboHydraHeadFilesystem = heads.RoboHydraHeadFilesystem,
RoboHydraHeadProxy = heads.RoboHydraHeadProxy,
Response = robohydra.Response;
exports.getBodyParts = function(conf) {
var developersDevelopersDevelopersDevelopers =
'<span onmouseover="var a = document.createElement(\'audio\'); a.src = \'/ogg/developers.ogg\'; a.autoplay = true; document.body.appendChild(a);">DEVELOPERS, DEVELOPERS, DEVELOPERS, DEVELOPERS</span>';
return {
heads: [
new RoboHydraHeadFilesystem({
name: "static",
mountPath: "/ogg",
documentRoot: "files"
}),
new RoboHydraHead({
name: "emphasizer",
path: "/.*",
handler: function(req, res, next) {
// Create a fake response object to pass to the
// proxy head, so we can capture the proxy
// response. Once we have the proxy response,
// substitute the emphasized string for
// "developers" and return the modified response
var fakeRes = new Response().on('end', function(evt) {
res.copyFrom(evt.response);
res.body =
res.body.toString().replace(
/developers</,
developersDevelopersDevelopersDevelopers + "<"
);
res.end();
});
// Remove the Accept-Encoding header from the
// original request to make sure the server won't
// compress the response, and pass the fake
// response object created above
delete req.headers["accept-encoding"];
next(req, fakeRes);
}
}),
new RoboHydraHeadProxy({
name: "realDevOpera",
mountPath: "/",
proxyTo: "http://dev.opera.com",
setHostHeader: true
})
]
};
};How does this work? First of all, in the emphasizer head, we remove the Accept-Encoding header from the original request to make sure the server doesn't reply with a compressed response. Then we call the next function passing the modified request object and a fake response object that we'll use only to save the response from the realDevOpera head. This fake response object has an event listener for the end event, when the "next" head (in this case, the proxy) has finished writing its response. At that moment we copy the contents of the proxy response to our own response, modify the body to emphasize the text we want, and then signal that we're done by calling the end method in the original response.
Conclusion
In this article, we have seen some of the most advanced uses of RoboHydra. Now you know how to use the low-level functionality to create the most flexible and powerful RoboHydra-based servers possible. If you want to know more, and follow updates and new features, check out robohydra.org and RoboHydra on GitHub, Twitter and YouTube.
Esteban Manchado Velázquez
Esteban is a quality assurance engineer, project manager, developer, and other things at Opera. He has worked on projects like Opera Link, Opera Unite, My Opera, Dev Opera and others. Outside work, he likes music, playing drums, reading... and hacking, of course.
Comments
No new comments accepted.