Pruny is a small example app with a Sinatra component that implements an HTTP API that wraps an hypothetical FaultyTreeSevice adding two main features:

• Filtering of trees by indicators nodes;
• Service resilience.

Pruny was the product of an exercise I made for a previous job interview and I'm making it available here hoping that it might be useful to someone else as a learning resource.

It mainly to illustrate:

• How to create a microservce that exposes an HTTP API with a Sinatra component;
• A homebrew tree structure that can be created from a json-like structure (array and or hash in Ruby), searched, filtered (pruned) and converted back to json and served as application/json through HTTP.

The original private repo had a URL pointing to another service (what I called FaultyTreeService above) that returned a big tree but that was also, as its name imples, flaky - it failed several times during requests. In this version I'm removing the URL to this service, but you can see an example of that tree in /spec/fixtures/input-tree-ficture.json.

To filter the trees, the json structure returned by FaultyTreeService is:

1. Converted to a tree-like data structure;
2. One or more nodes are searched using a BFS(ish) algorithm;
3. These nodes are grouped by common parents and finally the direct path torwards the root node is calculated, excluding any non-relevant siblings;
4. A copy of the tree for each of these nodes starting from the root is then returned.

The external FaultyTreeService fails a lot. Pruny will retry up to 4 times before losing faith on the Micro-service gods and returning to the Monolith church.

Pruny requires:

• Ruby 2.5.1;
• Bundler.

To get it running, follow the steps below:

1. Install rbenv https://github.com/rbenv/rbenv;
2. Install Ruby 2.5.1 (see instructions at the rbenv page above);
3. Install the Bundler gem: $ gem install bundler;
4. Install the Dotenv gem: $ gem install dotenv;
5. Clone this repo somewhere, cd into it;
6. Run $ bundle install and wait for all gems to be downloaded and installed;
7. Open an account at rollbar.com if you don't have one already, get your API token;
8. Run $ cp .env.example .env, edit it using your favorite editor (emacs!);
9. Set the value for the ROLLBAR_ACCESS_TOKEN to the token you copied in step #5;
10. Run the app with $ bundle exec dotenv rackup, it should be accessible @ localhost:9292.

At the moment, Pruny has ony one endpoint called /tree. It accepts two parameters:

1. The tree name (as in/tree/<name>). Right now only input is officially supported and tested;
2. The id of indicators to filter by (i.e /tree/input?indicator_ids[]=1&indicator_ids[]=2...).

Use your favorite HTTP client or API testing tool (Postman is a good one) or if you're in a hurry just open the following URL in your browser or use curl:

$ curl -XGET "localhost:9292/tree/input?indicator_ids[]=32&indicator_ids[]=31&indicator_ids[]=1" --globoff

This request will filter the upstream input tree by the indicator nodes with id 32, 31 and 1. You should get the following pruned tree structure json:

[
    {
        "id":2,
        "name":"Demographics",
        "sub_themes": [
            {
                "id":4,
                "name":"Births and Deaths",
                "categories":[
                    {
                        "id": 11,
                        "name": "Crude death rate",
                        "unit": "(deaths per 1000 people)",
                        "indicators": [
                            {"id":1,"name":"total"}
                        ]
                    }
                ]
            }
        ]
    },
    {
        "id":3,
        "name":"Jobs",
        "sub_themes": [
            {
                "id":8,
                "name": "Unemployment", 
                "categories": [
                    {
                        "id":23,
                        "name": "Unemployment rate, 15–24 years, usual",
                        "unit": "(percent of labor force)",
                        "indicators": [
                            {"id":31,"name":"Total"},
                            {"id":32,"name":"Female"}
                        ]
                    }
                ]
            }
        ]
    }
]

If the upstream FaultyTreeService fails after too many retries (4 at the time of this writing) or if there's any other general error then Pruny will return an error in the form of a json, like:

{"error":"Woopsie! The server got confused. We have been notified."}

The end-user will not see the specific error, but it will be sent alongisde a backtrace will to Rollbar, where the awesome Pruny developers can deal with them in a timely manner (hopefully!).

The suite of specs are located in the spec directory. To run the whole suite just run rake at the project root directory.

If you'd like to run a specific spec, use the m cli helper:

$ bundle exec m spec/path/to/the/spec.rb

To run a specific example within a spec, pass in the line for it after the spec path after a colon:

$ bundle exec m spec/path/to/the/spec.rb:55

If m can't find the example in the line you specified, it will let you know and will list all examples in the spec alongside their lines, like so:

No tests found on line 22. Valid tests to run:

test_0001_anonymous: m spec/integration/app_spec.rb:82
test_0001_anonymous: m spec/integration/app_spec.rb:93
test_0001_anonymous: m spec/integration/app_spec.rb:104
test_0001_anonymous: m spec/integration/app_spec.rb:118

Just pick the right line number and run it again ;)