Best practice for updating a structured resource via REST? - rest

I have a client-side interface that allows the user to perform multiple edits against a tree-like outline. I consider the aggregate of the records making up that outline, in totality, a single resource (/outlines/39) even though its parts could be accessed as separate resources via different URLs.
The problem is the user can edit existing nodes in the outline as well as add new nodes to the outline. Normally, when you edit something you PUT its changes and when you add something new you POST it; however, in some cases you'll want to wrap all the changes--including both adds and edits--in a single transaction. What are some practical ways people have handled this?
Even though the outline already exists and a PUT seems appropriate, the embedded adds violate the idempotence of the PUT. I'm not sure that POST seems appropriate either. For design purposes, I have decided not to save each discrete update the user makes though I guess this offers one solution. Still, there must be others who have dealt with my issue or have ideas about it.

Is there any way you could make the add idempotent? E.g. if nodes had a natural key, then when the client tried to add a node a second time you could do nothing.

How about: make a new resource: /outlines/39/transactions, and POST your transaction to that resource, e.g.
POST "addNode=node1, addNode=node2, editNode=node3,newName=foobar" to /outlines/39/transactions

Related

Rest API design: Managing access to sub-entities

Note: I realize that this is close to being off-topic for being opinion-based, but I am hoping that there is some accepted best practice to handle this that I just don't know about.
My problem is the following: I need to design a Rest API for a program where users can create their own projects, and each project contains files that can only be seen by users that have access. I am stuck with how to design the "List all files of a project" query.
Standard Rest API practice would suggest two endpoints, like:
`GET /projects` # List all projects
`POST /projects` # Create new project
`GET /projects/id` # Get specific project
etc.
and the same for the files of a project.
However, there should never be a reason to list all files - only the files of a single project. To make it more complicated, access management needs to be a thing, users should never see files that are in projects they don't have access to.
I can see multiple ways to handle that:
So the obvious way is to implement the GET function, optionally with a filter. However, this isn't optimal, since if the user doesn't set a filter, it would have to crawl through all projects, check for each project whether the user has access, and then list all files the user has access to:
GET /files?project=test1
I could also make the files command a subcommand of the projects command - e.g.
GET /projects/#id/files
However, I have the feeling this isn't too restful, since it doesn't expose entities directly?
Is there any concencus on which should usually be implemented? Is it okay to "force" users to set a parameter in the first one? Or is there a third alternative that solves what I am looking for? Happy about any literature recommendations on how to design this as well.
Standard Rest API practice would suggest two endpoints
No, it wouldn't. REST practice would suggest figuring out the resources in your resource model.
Think "documents": I should be able to retrieve (GET) a document that describes all of the files in the project. Great! This document should only be accessible when the request authorization matches some access control list. Also good.
Maybe there should also be a document for each user, so they can see a list of all of the projects they have access to, where that document includes links to the "all of the files in the project" documents. And of course that document should also be subject to access control.
Note that "documents" here might be text, or media files, or scripts, or CSS, or pretty much any kind of information that you can transmit over the network. We can gloss the details, because "uniform interface" means that we manage them all the same way.
In other words, we're just designing a "web site" filled with interlinked documents, with access control.
Each document is going to need a unique identifier. That identifier can be anything we want: /5393d5b0-0517-4c13-a821-c6578cb97668 is fine. Because it can be anything we want, we have extra degrees of freedom.
For example, we might design our identifiers such that the document whose identifiers begin with /users/12345 are only accessible by requests with authorization headers that match user 12345, and that all documents whose identifiers begin with /projects/12345 are only accessible by requests with authorization headers that match any of the users that have access to that specific project, and so on.
In other words, it is completely acceptable to choose resource identifier spellings that make your implementation easier.
(Note: in an ideal world, you would have "cool" identifiers that are implementation agnostic, so that they still work even if you change the underlying implementation details of your server.)
I have the feeling this isn't too restful, since it doesn't expose entities directly?
It's fine. Resource models and entity models are different things; we shouldn't expect them to always match one to one.
After looking further, I came across this document from Microsoft. Some quotes:
Also consider the relationships between different types of resources and how you might expose these associations. For example, the /customers/5/orders might represent all of the orders for customer 5. You could also go in the other direction, and represent the association from an order back to a customer with a URI such as /orders/99/customer. However, extending this model too far can become cumbersome to implement. A better solution is to provide navigable links to associated resources in the body of the HTTP response message. This mechanism is described in more detail in the section Use HATEOAS to enable navigation to related resources.
In more complex systems, it can be tempting to provide URIs that enable a client to navigate through several levels of relationships, such as /customers/1/orders/99/products. However, this level of complexity can be difficult to maintain and is inflexible if the relationships between resources change in the future. Instead, try to keep URIs relatively simple. Once an application has a reference to a resource, it should be possible to use this reference to find items related to that resource. The preceding query can be replaced with the URI /customers/1/orders to find all the orders for customer 1, and then /orders/99/products to find the products in this order.
This makes me think that using solution 2 is probably the best case for me, since each file will be associated with only a single project, and should be deleted when a project is deleted. Files cannot exist on their own, outside of projects.

In general, would it be redundant to have two GET routes for users (one for ID and one for username)?

I'm building a CRUD for users in my rest API, and currently my GET route looks like this:
get("/api/users/:id")
But this just occured to me: what if a users tries to search for other users via their username?
So I thought about implementing another route, like so:
get("api/users/username/:id")
But this just looks a bit reduntant to me. Even more so if ever my app should allow searching for actual names as well. Would I then need 3 routes?
So in this wonderful community, are there any experienced web developers that could tell me how they would handle having to search for a user via their username?
Obs: if you need more details, just comment about it and I'll promptly update my question 🙃
how they would handle having to search for a user via their username?
How would you support this on a web site?
You would probably have a form; that form would have an input control that would allow the user to provide a user name. When the user submit the form, the browser would copy the form input controls into an application/x-www-form-urlencoded document (as described by the HTTP standard), then substitute that document as the query_part of the form action, and submit the query.
So the resulting request would perhaps look like
GET /api/users?username=GuiMendel HTTP/x.y
You could, of course, have as many different forms as you like, with different combinations of input controls. Some of those forms might share actions, but not necessarily.
so I could just have my controller for GET "/api/users" redirect to an action based on the inputs?
REST doesn't care about "controllers" -- that's an implementation detail; the whole point is that the client doesn't need to know how the server produces a representation of the resource, we just need to know how to ask for it (via the "uniform interface").
Your routing framework might care a great deal, but again that's just another implementation detail hiding behind the facade.
for example, there were no inputs, it would return all users (index), but with the input you suggested, it would filter out only users whose usernames matched the input? Did I get it right?
Yup, that's fine.
From the point of view of a REST client
/api/users
/api/users?username=GuiMendel
These identify different resources; the two resources don't have to have any meaningful relationship with each other at all. The machines don't care (human beings do care, so we normally design our identifiers in such a way that at least some human beings have an easy time of it -- for example, we might optimize our identifiers to make things easy when operators are reading the access logs).

Proper usage of OpenAPI 3.0.2

I started using OpenAPI a few months ago, but I'm finding multiple caveats while generating documentation. This of course leads me to believe that I must be doing something wrong.
So, I usually begin always by creating the schemes for the multiple resources of my API, I then place a reference to the schemes on the request bodies for each respective endpoint. However I find that this approach makes generated documentation confusing, adding for example an id parameter on the request body of a user create operation when it's clearly not supposed to be there (Of course the user is being created at this point, so there isn't even an ID in the first place, and why would an ID be part of the request body?).
I ended up just making multiple schemes for a single data model, I would choose which one I used based on the operation (E.g. Creating vs Displaying), this way I could generate documentation more accurately. This helps me avoid situations like displaying relations that are generated post-creation on a create operation. (I know I'm probably not phrasing all this correctly but just bear with me).
In a nutshell, I'll end up with approximately 3 or 4 schemes per data model. For the User model I'll have for example User, UserNew, UserSubscriptions, UserAll. All schemes represent the same resource, they just differ in the data that they display.
Again, I feel like this is not the way that OpenAPI is meant to be used. So I hoped that maybe someone here that has more experience with this could guide me on the right path! Thank you.

RESTful - GET or POST - what to do?

Im working on a web service that i want to be RESTful. I know about the CRUD way of doing things, but I have a few things that im not completly clear with. So this is the case:
I have a tracking service that collects some data in the browser (client) and then sends it off to the tracking server. There are 2 cases, one where the profile exists and one where it does not. Finally the service returns some elements that has to be injected to the DOM.
So basically i need 2 web services:
http://mydomain.tld/profiles/
http://mydomain.tld/elements/
Question 1:
Right now im only using GET, but im rewriting the server to support CRUD. So in that case i have to use POST if the profile does not exist. Something like http://mydomain.tld/profiles/ and then POST payload have the information to save. If the profile is existing i use PUT and http://mydomain.tld/profiles// and payload of PUT has data to save. All good, but problem is that as far as i understand, xmlhttp does not support PUT. Now is it ok to use POST even though its an update?
Question 2:
As said my service returns some elements to be injected into the DOM, when a track is made. Logically, to keep it RESTful, i guess that i would have to use POST/PUT to update the profile and then GET to get the elements to inject. But to save bandwidth and resources on the serverside, it makes more sense to return the elements with the POST/PUT to profiles, even though its a different resource. What are your take on this?
BR/Sune
EDIT:
Question 3:
In some cases i only want to update the profile and NOT receive back elements. Could i still use same resource and then using a payload parameter to specify if i want elements, e.g. "dont_receive_elements:true"
On question #1, are you sure that xmlhttp does not support "put"? I just ran http://www.mnot.net/javascript/xmlhttprequest/ on three browsers (Chrome, Firefox, IE) and according to the output, "put" was successful on all browsers. Following the information on http://www.slideshare.net/apigee/rest-design-webinar (and I highly recommend checking out the many Apigee videos and slideshows on restful API), "put" is recommended for the use case you mention.
But you may be able to avoid this issue entirely by thinking a little differently about your data. Is it possible to consider that you have a profile and that for each profile you have 0 or more sets of payload information? In this model the two cases are:
1. No profile exists, create profile with a POST on .../profiles/ Then add elements/tracking data with posts to .../profile/123/tracks/ (or .../profile/123/elements/)
2. Profile exists, just add the elements/tracking data
(Sorry without understanding your model in detail, it is hard to be very precise).
As for question #2 - going with a data model where a profile has 0 or more elements, you could update the profile (adding the necessary elements) and then return the updated profile (and its full graph of elements), saving you any additional gets.
More generally on question #2, as the developer of the API you have a fair amount of freedom in the REST world - if you are focused on making it easy and obvious for the consumers of your API then you are probably fine.
Bottom line: Check out www.apigee.com - they know much more than I.
#Richard - thanks alot for your links and feedback. The solution i came down to is to make the API simple and clean as you suggest in your comment, having seperate calls to each resouce.
Then to be able to save bandwidth and keep performance up, I made a "non-official" function in the API that works like a proxy internally and are called with a single GET, that updates a profile and returns an element. This, i know, is not very restful etc, but it handles my situation and is not part of the official API. The reason i need it to support GET for this i need to call it from javascript and cross domain.
I guess i could have solved the cross domain by using JSONP, but i would still have to make the API "unclean" :)

Adding pages "on the fly" with a CMS system

I am in the process of building a website content management system for one of my clients. It's a highly customized system, so I cannot use any "of the shelve" solution.
I need to allow my client to add pages to the website on the fly. I have two options here:
(1) Create a database driven page in the format of www.mycompany.com/page.aspx?catID=5&pageID=3 (query the database with the category and page ID's, grab the data and show it on the page) - or -
(2) Allow the management system to create static pages, something like www.mycompany.com/company/aboutus.aspx and www.mycompany.com/company/company_history.aspx , etc.
I believe that, while the former is much easier to implement, the latter is a better both for the user AND for Google.
My questions are (finally): (1) Would you agree that the latter is a better solution, and (2) What is the best way to implement such a solution? Should I create and update each file using the FileSystem (i.e. - the site's management system requires the user to supply a page/file name, page title and content, and creates the page on the fly based on these parameters)? Is there a better way?
Thank you!
It's entirely possible to have database driven pages with nice URLs. StackOverflow itself is a great example - this question's URL is http://stackoverflow.com/questions/1119274/adding-pages-on-the-fly-with-a-cms-system, but the page is built from the database, not static HTML.
I would use the first solution, but mask the addresses using a custom request handler. Basically, give each of your pages a unique string ID (such as about-us) and then, with your request handler that takes all requests, find this particular page in the database and render it.
See this article for some additional info (found it when googling for custom http handlers in ASP.NET.) In that article, it has the following handler added:
<add verb="*" path="*.piechart" type="PieChartHandler"/>
You would probably want to catch all paths (*), excluding certain media paths used for CSS, images and JavaScript.
More resources:
Custom HTTP Handler
HttpHandler in ASP.Net
I'd stay clear of static pages if I where you. Dynamic Data, MVC and some good planning should take you a long way!
What you need to do is to create some or many templates that each view/controller in mvc can use. Let whoever is responsible for the content handle it through dynamic data entities.
I would use the first idea, but work out a better URL scheme. If the system doesn't provide nice URLs (without ?), you'll have trouble getting the search engines to parse the whole site. Also using numbers instead of words make it hard on users to pass around URLs.
If you start to have performance problems you could add caching that would generate static pages from time to time. I would avoid doing that until you have to; caching can cause many headaches along the way to getting it right.
Although the existing advice is more-or-less sound, the commentators have failed to consider one factor which, admittedly, you haven't given much detail on. Are these pages that they'll edit once they're built, or a they one-shot creations? If the latter, your plan of generating static pages isn't quite so bad as they suggest. Why bother even having to think about database schemas and caching, when you can just serve flat content.
It will probably make for pretty lifeless, end-of-the-road pages, but if that's what you want ...