I have an API that produces JSON like this:
)]}',
{
//JSON DATA
}
The //JSON DATA is valid JSON, but the )]}', up top is not.
When I try to GET this data via a Logic App, I get:
BadRequest. Http request failed: the content was not a valid JSON.
So, a few related questions:
1) Can I tell the logic app to return the invalid JSON anyway?
2) How can debug the issue better? I happen to know that the response is invalid, but what if I didn't? Can I see the raw data somewhere?
3) This is all done via the Azure web portal. Are there better tools? Visual Studio?
I should also mention that if I call a route on the same API that returns XML instead of JSON, then the Logic App works fine. So it definitely doesn't like the JSON response in particular.
Thanks!
First of all, please do not post three questions as a single question.
Question 1). The best thing you can do is make the API return a valid JSON object. This is good for million reasons. Here're a few:
it's pretty much a standard (either valid JSON or XML -- yeah, old school way);
therefore, no users of this API (including you) will need to struggle and guess what's going on and why;
your Logic App's step will just work without adding extra complexity;
you will make this world and your karma better.
If API-side changes are not within your reach, I don't think you can do much. If you're lucky and the HTTP action is successful (Status Code 2xx), you can try to use a Query Action with a function that truncates the first characters. It will look something like this (I don't know the exact syntax): #Substring(body('myHttpGet'), 4, length(body('myHttpGet')) - 4) where myHttpGet is the id of the Http Get action.
However, once again, if possible, I strongly recommend fixing up the API which is the root cause of the problem, instead of dealing with garbage response after that.
UPDATE Another thing you can do is wrap the dirty API. For example, you could create a trivial Azure Function that invokes the API you don't directly control, and sanitizes the response for you consumption requirements. This Azure Function function should be easy to call from the Logic App. It costs almost nothing (unless we're talking millions of requests/month). The only drawback here is the increasing latency, which may be not an issue at all -- test it and see whether it adds less than 100ms or so... Oh, and don't forget to file a ticket with the API owner, they make our world a bad place!
Question 2) In Azure Logic App web UI you can Look into the execution details and the error will definitely be there.
Question 3) You're asking for a tool recommendation which is by definition a highly subjective thing and is off-topic on StackOverflow.
TL/DR: The other app is not producing valid JSON.
Meaning, this is not a problem for you to solve. The other app has to return valid JSON if the owner claims it should.
If they cannot or will not produce valid JSON, then the first thing you need to do is inform your management that you will have to spend a lot of extra time accommodating their non-standard format.
Related
I'm trying to parse the last received message from Pushbullet. I'm currently doing it using Clicks, which means that I do every single clicks, sendkeys and all the rest needed actions automatically. In other words it is just a simulation of the user. One of the biggest con of the method is that whenever xpath or a class or id of the element changes which I'm aiming with Selenium, whole cycle fails and the test case can not proceed operating.
I want to change the way I'm doing. Particularly, I want to send a json request to Pushbullet API and then get the response in return.
I just couldn't decide from where and how to start doing.
Could you guys please tell me the way from where should I start and what are the steps that needs to be covered in order to finally reach the finish line?
Thank you in advance.
I noticed that this question is tagged under Selenium, but in your question, you express interest in switching to an API approach. I will try to provide some advice to you on this.
Selenium Approach
You mention that you tests are brittle, and if anything changes then they fail. This is usually the case with UI tests. If you would like to stick with the Selenium approach, I can try to help you write more robust locators for your WebElements that will not break constantly.
API Approach
You will need to start with the Pushbullet API documentation -- https://docs.pushbullet.com/
To get messages, it looks like you will want to use the chat endpoint -- a sample request looks like this: https://api.pushbullet.com/v2/chats, plus authentication.
Once you fetch the chat objects, you will need to write your own logic to parse each chat object and fetch the most recent message from there.
Depending on what language you are using, you will need to install a REST client package onto your project. I use C#, so RestSharp is the client I like to use.
I recommend installing a REST client interface, such as Postman, to start practicing your API calls. Once you get your API calls working in Postman, you can start writing code to make these API calls.
What other questions do you have about this?
I’m in the middle of making an Express app. It’s just a learning project.
I’m getting some info from an Anime api called jikan.me, it provides info about different Anime series like a picture url and synopsis.
For example one is at https://api.jikan.me/anime/16 .
Now, the jikan api might have a json endpoint at anime/1 but there's nothing at anime/2.
I want to find a list of all the numbers (https://api.jikan.me/anime/[numbers]) that actually contain endpoints.
I've tried simply going to https://api.jikan.me/anime but it returns error: No ID/Path Given.
I'm expecting there is likely no absolute answer to this problem but that I might learn something about server-side code along the way.
Where would I begin to look to find this info?
This is a bit late but, Jikan is an unofficial REST API for MyAnimeList. The IDs are respective to the IDs on MAL. For example; https://myanimelist.net/anime/1 can be parsed through https://api.jikan.moe/anime/1 but the ID 2 does not exist on MAL. It's a 404, hence that error.
To initially get some IDs, you can try the search endpoint.
Furthermore, I'll be releasing REST 2.2 quite soon (this month) which will give you the ability to parse from pages like these and thus you'll get another endpoint that provides a handful of IDs to get their data from.
Source: I'm the developer of Jikan
If it's not in the documentation it's probably information not available to you... a REST api needs to be specifically configured to offer certain endpoints, that number at the end might just be an ID that's searched for in an internal database and there's no way for the application to know if there's gonna be something there; all they can do is return an error message for you to handle as is the case here.
I am writing a web service in node, and testing it with Postman. I spent a long timing looking for an error. When I finally found it, it turned out to be a simple error formatting the response body, which is json.
If I leave off the final brace in the response body, Postman waits for two minutes, and then reports that it received everything, just fine.
If I leave off the closing quote in the last value in the json, Postman says the server didn't respond, perhaps I should check my security certificates.
I would much rather Postman said "Hey, Buddy, you left off a quote!"
If there some validation service I can talk to? Or a plugin in Postman?
Here there are some validation javascript libraries, you can use:
Validator provides a declarative way of validating javascript objects.
Express-validator acts as an express.js middleware for node-validator.
Meanwhile, Postman got API testing and Collection Runner that can help you through this; which you can write some pre-request script as well as test script for each request.
Also, they got Newman which is a command-line collection runner. It allows you to effortlessly run and test a Postman collection directly from the command-line. It is built with extensibility in mind so that you can easily integrate it with your continuous integration servers and build systems.
I found that Paw worked (https://paw.cloud/). And so far I haven't paid for it.
Where Postman said "check your security certificates," Paw said "we were expecting 376 bytes but you only sent us 312."
Cuts down my time solving the problem a lot!
I use Fiddler for this. It is very good at identifying (with an error message that pops up) problems and bad implementations of the HTTP protocol. Browse the web with it running, and within a few minutes you'll undoubtedly hit a poorly implemented server.
Postman won't be able to handle these cases since it's insulated from poor behavior by the browser's framework.
That's not your problem though.
When I finally found it, it turned out to be a simple error formatting the response body, which is json.
That has absolutely nothing to do with HTTP. HTTP doesn't know or care what your request/response bodies are.
The problem you face is that your API endpoint could be returning whatever it wants. You need a custom solution to your problem, as there is no standard API server in this case.
Most folks will run unit tests that hit common endpoints of your service to ensure they're alive and well.
I should also point out that it should be all but impossible for you to break the JSON response if you're doing it correctly. Sounds like you're serializing JSON manually... never do that, we have JSON serializers for this purpose. Send in an object and let it worry about building the JSON output for you. Otherwise, you'll waste a lot of time on problems like these.
I have a simple RESTful web service and I wish to test the PUT method on a certain resource. I would like to do it in the most simple way using as few additional tools as possible.
For instance, testing the GET method of a resource is the peak of simplicity - just going to the resource URL in the browser. I understand that it is impossible to reach the same level of simplicity when testing a PUT method.
The following two assumptions should ease the task:
The request body is a json string prepared beforehand. Meaning, whatever is the solution to my problem it does not have to compose a json string from the user input - the user input is the final json string.
The REST engine I use (OpenRasta) understands certain URL decorators, which tell it what is the desired HTTP method. Hence I can issue a POST request, which would be treated as a PUT request inside the REST engine. This means, regular html form can be used to test the PUT action.
However, I wish the user to be able to enter the URL of the resource to be PUT to, which makes the task more complicated, but eases the testing.
Thanks to all the good samaritans out there in advance.
P.S.
I have neither PHP nor PERL installed, but I do have python. However, staying within the realm of javascript seems to be the simplest approach, if possible. My OS is Windows, if that matters.
I'd suggest using the Poster add-on for Firefox. You can find it over here.
As well as providing a means to inspect HTTP requests coming from desktop and web applications, Fiddler allows you to create arbitrary HTTP requests (as well as resend ones that were previously sent by an application).
It is browser-agnostic.
I use the RESTClient firefox plugin (you can not use an URL for the message body but at least you can save your request) but also would recommend curl on the command line.
Maybe you should also have a look at this SO question.
If you view the source of a Google+ profile page, it appears rather complex. It seems most of the data is kept in a huge JSON-like objects. However, they don't seem to be really JSON, since they don't get recognized when I try to decode them. I am hoping the format is more clear to other people here. How would you go about parsing it? It seems it would fairly trivial, if you know where to start.
Here is a sample profile, for example: http://plus.google.com/104560124403688998123
Here's a PHP API I'm working on. It can download and parse the data for a profile page and people's public relationships.
https://github.com/jmstriegel/php.googleplusapi
The JSON piece is a bit mangled. To generate valid JSON, you basically have to remove the first 5 characters that prevent XSRF attacks and then add in all the nulls that have been removed. Here's the code specific to handling parsing the weird Google Plus JSON responses:
https://github.com/jmstriegel/php.googleplusapi/blob/master/lib/GooglePlus/GoogleUtil.php
Call GoogleUtil::FetchGoogleJSON( $url ) and you'll get back a giant array that you can then pull data from. Using this, it should be trivial to make a proxy service to translate stuff into valid json(p) for you to use in your own apps.
I don't have access to Google+ yet, so I'll just answer the general question - that is, how to parse JSON.
JSON is just JavaScript, so parsing it is as simple as evaluating the script. To do this, use the eval() JavaScript function.
var obj = eval('{"JSON":"goes here"}');
Another option is to leverage a console tool. Popular modern browsers pretty much all have them. I recommend Firebug for Firefox in particular.
Using Firefox, log into Google+, then open the Firebug console. You can use the console's dir() command to create a browseable representation of the data. Ex:
console.dir(eval('{"JSON":"goes here"}'));
Sorry I can't be more specific about how to get a handle on Google+'s JSON in particular; without access to the service, this is about the best I can do blind. Good luck!
Thanks to Jason for the excellent php class which reads a profile page into an array.
I've used this class as a base and then parsed it, based upon Russell Beattie's python code from the original appspot rss feed application.
Code here
A few notes:
I use this to merge G+ and WP feeds, hence writing posts into an intermediate array ($items).
I have a convention of creating a pseudo title in Google Plus posts, by emboldening a line and adding two newlines before writing the post. The function getTitle strips this out as a better formatted title in my website and getSummary produces the rest of the post with duplicating the title.
It's made up of a number of parts, an object describing your picasa images, one describing the fields on your profile, one describing your friends.
Most of the long numbers are the internal IDs of people, posts and photos. For instance, my ID is 105249724614922381234. Other than that, it could be parsed if you needed to.