From f3ce6149b7bb9a449c8e219b21ad3065d2a8442a Mon Sep 17 00:00:00 2001 From: Rich Date: Wed, 26 Jul 2023 16:44:20 +0200 Subject: [PATCH 01/16] Move tutorials to separate file to reduce readme bloat --- README.md | 1033 +------------------------------------------------ TUTORIALS.md | 1034 ++++++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 1035 insertions(+), 1032 deletions(-) create mode 100644 TUTORIALS.md diff --git a/README.md b/README.md index a7dd456..64411a7 100644 --- a/README.md +++ b/README.md @@ -30,1036 +30,5 @@ Boot your microservices-enabled system using docker-compose. You can shut down using `docker-compose stop` and remove everything using `docker-compose rm`. ## Tutorials -If you aren't familiar with the semantic.works stack/microservices yet, you might want to check out [why semantic tech?](https://mu.semte.ch/2017/03/23/adding-ember-fastboot-to-your-mu-project/) -- [Creating a JSON API](#creating-a-json-api) -- [Adding authentication to your mu-project](#adding-authentication-to-your-mu-project) -- [Creating a mail service](#building-a-mail-handling-service) -- [Adding Ember Fastboot to your project](#adding-ember-fastboot-to-your-project) -- [Adding a machine learning microservice to your mu.semte.ch project](#adding-a-machine-learning-microservice-to-your-musemtech-project) - -### Creating a JSON API -Repetition is boring. Web applications oftentimes require the same functionality: to create, read, update and delete resources. Even if they operate in different domains. Or, in terms of a REST API, endpoints to GET, POST, PATCH and DELETE resources. Since productivity is one of the driving forces behind the mu.semte.ch architecture, the platform provides a microservice – [mu-cl-resources](https://github.com/mu-semtech/mu-cl-resources) – that generates a [JSONAPI](http://jsonapi.org/) compliant API for your resources based on a simple configuration describing the domain. In this tutorial we will explain how to setup such a configuration. - -#### Adding mu-cl-resources to your project -Like all microservices in the mu.semte.ch stack, mu-cl-resources is published as a Docker image. It just needs two configuration files in `./config/resources/`: - -- `domain.lisp`: describing the resources and relationships between them in your domain -- `repository.lisp`: defining prefixes for the vocabularies used in your domain - -To provide the configuration files, you can mount the files in the /config folder of the Docker image: -```yaml -services: - # ... - resource: - image: semtech/mu-cl-resources:1.20.0 - links: - - db:database - volumes: - - ./config:/config - # ... -``` -Alternatively you can build your own Docker image by extending the mu-cl-resources image and copying the configuration files in /config. See the [mu-cl-resources repo](https://github.com/mu-semtech/mu-cl-resources#mounting-the-config-files). - -The former option may be easier during development, while the latter is better suited in a production environment. With this last variant you can publish and release your service with its own version, independent of the mu-cl-resources version. - -When adding mu-cl-resources to our application, we also have to update the dispatcher configuration such that incoming requests get dispatched to our new service. We will update the dispatcher configuration at the end of this tutorial once we know on which endpoints our resources will be available. - -#### Describing your domain -Next step is to describe your domain in the configuration files. The configuration is written in Common Lisp. Don’t be intimidated, just follow the examples, make abstraction of all the parentheses and your’re good to go 🙂 As an example, we will describe the domain of the [ember-data-table demo](http://ember-data-table.semte.ch/) which [consists of books and their authors](https://github.com/erikap/books-service/tree/ember-data-table-example). - -##### repository.lisp -The `repository.lisp` file describes the prefixes for the vocabularies used in our domain model. - -To start, each configuration file starts with: - -```lisp -(in-package :mu-cl-resources) -``` - -Next, the prefixes are listed one per line as follows: - -```lisp -(in-package :mu-cl-resources) - -(add-prefix "dcterms" "http://purl.org/dc/terms/") -(add-prefix "schema" "http://schema.org/") -``` - -##### domain.lisp -The domain.lisp file describes your resources and the relationships between them. In this post we will describe the model of a book. Later on we will add an author model and specify the relationship between books and authors. - -Also start the `domain.lisp` file with the following line: -```lisp -(in-package :mu-cl-resources) -``` - -Next, add the basis of the book model: -```lisp -(in-package :mu-cl-resources) - -(define-resource book () - :class (s-prefix "schema:Book") - :resource-base (s-url "http://mu.semte.ch/services/github/madnificent/book-service/books/") -:on-path "books") -``` - -Although you may not have written a letter of Common Lisp before, you will probably be able to understand the lines of code above. - -- We define a book resource -- Each book will get schema:Book as RDF class -- Each book instance will be identified with a URI starting with “http://mu.semte.ch/services/github/madnificent/book-service/books/” – mu-cl-resources just appends a generated UUID to it -- The resources will be published on the /books API endpoint - -Finally, define the properties of a book: -```lisp -(in-package :mu-cl-resources) - -(define-resource book () - :class (s-prefix "schema:Book") - :properties `((:title :string ,(s-prefix "schema:headline")) - (:isbn :string ,(s-prefix "schema:isbn")) - (:publication-date :date ,(s-prefix "schema:datePublished")) - (:genre :string ,(s-prefix "schema:genre")) - (:language :string ,(s-prefix "schema:inLanguage")) - (:number-of-pages :integer ,(s-prefix "schema:numberOfPages"))) - :resource-base (s-url "http://mu.semte.ch/services/github/madnificent/book-service/books/") -:on-path "books") -``` - -Each property is described according to the format: - -```lisp -(:dasherized-property-name :type, (s-prefix "my-prefix:my-predicate")) -``` - -and will result in a triple: -``` - my-prefix:my-predicate "some-value" -``` - -#### Configuring the dispatcher - -Our book resources will be available on the /books paths. The mu-cl-resources service provides GET, POST, PATCH and DELETE operations on this path for free Assuming the books service is known as ‘resource’ in our dispatcher, we will add the following dispatch rule to our dispatcher configuration to forward the incoming requests to the books service: - -```lisp -match "/books/*path" do - Proxy.forward conn, path, "http://resource/books/" -end -``` - -Good job! The books can now be produced and consumed by the frontend through your JSONAPI compliant API. Now we will add relationships to the model. - - -#### The author model -Each book is written by (at least one) an author. An author isn’t a regular property of a book like - for example - the book’s title. It's a resource on its own. An author has its own properties like a name, a birth date etc. And it is related to a book. Before we can define the relationship between books and authors, we first need to specify the model of an author. - -The definition of the model is very similar to that of the book. Add the following lines to your `domain.lisp`: - -```lisp -(define-resource author () - :class (s-prefix "schema:Author") - :properties `((:name :string ,(s-prefix "schema:name"))) - :resource-base (s-url "http://mu.semte.ch/services/github/madnificent/book-service/authors/") -:on-path "authors") -``` - -Expose the author endpoints in the `dispatcher.ex` configuration: -``` -match "/authors/*path" do - Proxy.forward conn, path, "http://resource/authors/" -end -``` - -#### Defining relationships -Now that the author model is added, we can define the relationship between a book and an author. Let’s suppose a one-to-many relationship. A book has one author and an author may have written multiple books. - -First, extend the book’s model: -```lisp -(define-resource book () - :class (s-prefix "schema:Book") - :properties `((:title :string ,(s-prefix "schema:headline")) - (:isbn :string ,(s-prefix "schema:isbn")) - (:publication-date :date ,(s-prefix "schema:datePublished")) - (:genre :string ,(s-prefix "schema:genre")) - (:language :string ,(s-prefix "schema:inLanguage")) - (:number-of-pages :integer ,(s-prefix "schema:numberOfPages"))) - :has-one `((author :via ,(s-prefix "schema:author") - :as "author")) - :resource-base (s-url "http://mu.semte.ch/services/github/madnificent/book-service/books/") -:on-path "books") -``` - -Adding an author to a book will now result in the following triple in the store: -``` - - schema:author . -``` - -The :as “author” portion of the definition specifies the path on which the relationship will be exposed in the JSON representation of a book. - -```json -{ - "attributes": { - "title": "Rock & Roll with Ember" - }, - "id": "620f7a1c-9d31-4b8a-a627-2eb6904fe1f3", - "type": "books", - "relationships": { - "author": { - "links": { - "self": "/books/620f7a1c-9d31-4b8a-a627-2eb6904fe1f3/links/author", - "related": "/books/620f7a1c-9d31-4b8a-a627-2eb6904fe1f3/author" - } - } - } -} -``` - -Next, add the inverse relationship to the author’s model: - -```lisp -(define-resource author () - :class (s-prefix "schema:Author") - :properties `((:name :string ,(s-prefix "schema:name"))) - :has-many `((book :via (s-prefix "schema:author") - :inverse t - :as "books")) - :resource-base (s-url "http://mu.semte.ch/services/github/madnificent/book-service/authors/") -:on-path "authors") -``` - -The ‘:inverse t’ indicates that the relationship from author to books is the inverse relation. As you can see the the relationship’s path “books” is in plural since it’s a has-many relation in this case. - -If you want to define a many-to-many relationships between books and authors, just change the :has-one to :has-many and pluralize the “author” path to “authors” in the book’s model. Don’t forget to restart your microservice if you’ve updated the model. - -Now simply restart your microservice by running `docker-compose restart`, and you're done! - -#### Conclusion -That's it! Now you can [fetch](http://jsonapi.org/format/#fetching-relationships) and [update](http://jsonapi.org/format/#crud-updating-relationships) the relationships as specified by [jsonapi.org](http://jsonapi.org/). The generated API also supports the [include query parameter](http://jsonapi.org/format/#fetching-includes) to include related resources in the response when fetching one or more resource. That’s a lot you get for just a few lines of code, isn’t it? - -*This tutorial has been adapted from Erika Pauwels' mu.semte.ch articles. You can view them [here](https://mu.semte.ch/2017/07/27/generating-a-jsonapi-compliant-api-for-your-resources/) and [here](https://mu.semte.ch/2017/08/17/generating-a-jsonapi-compliant-api-for-your-resources-part-2/).* - - -### Adding authentication to your mu-project -![](http://mu.semte.ch/wp-content/uploads/2017/08/customumize_for_user-1024x768.png) - -Web applications oftentimes require a user to be authenticated to access (part of) their application. For example a webshop may require a user to be logged in before placing an order. In a previous blog post we already explained [the semantic model to represent logged in users](https://mu.semte.ch/2017/08/24/representing-logged-in-users/). In this post we will show how to enable authentication in your app. We assume you already have  a [mu-project](https://github.com/mu-semtech/mu-project) running. - -Adding authentication to your application consists of two tasks: - -- Adding registration so users can create a new account -- Adding a login service so users can authenticate themselves - -Both tasks require changes in the backend as well as in the frontend. Let’s start with the registration. - -#### Registration - -First, we will add registration to the project. The backend will be enriched with a microservice to manage accounts. The frontend will be augmented with an Ember addon providing components to register, unregister and  change a password. - -##### In the backend - -The [registration service](https://github.com/mu-semtech/registration-service) provides a service to create new accounts with a nickname and a password. To integrate the service in your project, add the following snippet to the `docker-compose.yml`. -```yaml -registration: - image: semtech/mu-registration-service:2.6.0 - links: - - database:database -``` - -(Re)start the project. -```bash -docker-compose up -``` -Next, configure the following routes in your dispatcher configuration in `config/dispatcher/dispatcher.ex`. -```ex -match "/accounts/\*path" do - Proxy.forward conn, path, "http://registration/accounts/" -end -``` - -Restart the dispatcher service . -```bash -docker-compose restart dispatcher -``` - -From now on all requests starting with ‘/accounts’ will be forwarded to the registration service. - -##### In the frontend - -We now have an endpoint for registration in the backend. We need a complementary component in the frontend that provides a GUI to communicate with this backend.  This component is offered by the [ember-mu-registration](https://www.npmjs.com/package/ember-mu-registration) addon. - -First, install the addon by executing the following command in your Ember project. -```bash -ember install ember-mu-registration -``` - -Next, just include the `{{mu-register}}`, `{{mu-unregister}}` and `{{mu-change-password}}` component in your template. - -```hbs -{{!-- app/templates/registration.hbs --}} -{{mu-register}} -``` - -The components will automatically send the correct requests to the backend. You can customize the component’s template and/or behavior as explained in the addon’s [README](https://github.com/mu-semtech/ember-mu-registration#advanced-usage). - -Finally create a new user account through the newly added mu-register component. We can use this user to validate the login in the next step. - -#### Login -Users can now create a new account, but how can they authenticate themselves in the app? In the next step we will enrich the backend with a login microservice and the frontend with a login form and a logout button. - -##### In the backend -The [login service](https://github.com/mu-semtech/login-service) provides a service to associate a session with a user’s account if the correct user credentials are provided. Have a look at [the semantic works docs](https://github.com/Denperidge-Redpencil/project/blob/master/docs/references/representing-logged-in-users.md) if you want to know the semantic model behind the users, sessions and accounts. To integrate the service in your project, add the following snippet to the `docker-compose.yml`. -```yaml -login: - image: semtech/mu-login-service:2.8.0 - links: - - database:database -``` - -(Re)start the project. -```bash -docker-compose up -``` - -Next, configure the following routes in your dispatcher configuration in `config/dispatcher/dispatcher.ex`. -```ex -match "/sessions/\*path" do - Proxy.forward conn, path, "http://login/sessions/" -end -``` - -Restart the dispatcher service . -```bash -docker-compose restart dispatcher -``` - -From now on all requests starting with ‘/sessions’ will be forwarded to the login service. - -#### In the frontend -Users can now be authenticated in the backend. Next, we need GUI components to login and logout and a mechanism to protect parts of the application so they are only accessible by authenticated users. These components are offered by the [ember-mu-login addon](https://github.com/mu-semtech/ember-mu-login) which requires [ember-simple-auth](https://github.com/simplabs/ember-simple-auth) to be installed, too. - -First, install the addons by executing the following commands in your Ember project. -```bash -ember install ember-simple-auth -ember install ember-mu-login -``` - -##### Login form -Next, we will generate a login route with a login form where the user can enter his credentials to authenticate. -```bash -ember generate route login -``` - -Add the `mu-login` component to the template. -```hbs -{{!-- app/templates/login.hbs --}} - -{{mu-login}} -``` - -##### Logout button -Once the user logged in, we will show a button so the user can logout. We will use [ember-simple-auth’s ‘isAuthenticated’ property](https://github.com/simplabs/ember-simple-auth#basic-usage) to check the current session’s state. The session service needs to be injected in the application controller. - -```js -// app/controllers/application.js -import Ember from 'ember'; - -export default Ember.Controller.extend({ - session: Ember.inject.service('session') - - // … -}); -``` - -Next, update the application’s template to show the logout button if the user is authenticated. -```hbs -{{!-- app/templates/application.hbs --}} -{{#if session.isAuthenticated}} - {{mu-logout}} -{{/if}} -``` - -Finally, mix the `ApplicationRouteMixin` in your application’s route. This mixin will automatically handle the [authenticationSucceeded](http://ember-simple-auth.com/api/classes/SessionService.html#event_authenticationSucceeded) and [invalidationSucceeded](http://ember-simple-auth.com/api/classes/SessionService.html#event_invalidationSucceeded) events. - -```js -// app/routes/application.js -import Ember from 'ember'; -import ApplicationRouteMixin from 'ember-simple-auth/mixins/application-route-mixin'; - -export default Ember.Route.extend(ApplicationRouteMixin); -``` - -##### Protecting routes - -Users can now login in the application, but they are still able to access all pages regardless whether they are authenticated or not. To make a route in the application accessible only when the session is authenticated, mix the [AuthenticatedRouteMixin](http://ember-simple-auth.com/api/classes/AuthenticatedRouteMixin.html) into the respective route: -```js -// app/routes/protected.js -import Ember from 'ember'; -import AuthenticatedRouteMixin from 'ember-simple-auth/mixins/authenticated-route-mixin'; - -export default Ember.Route.extend(AuthenticatedRouteMixin); -``` - -This will make the route (and all of its subroutes) transition to the ‘login’ route if the session is not authenticated. - -#### Finished -And that's it! Now you know how your mu-project can be easily augmented with authentication using a custom user registration service. - -*This tutorial has been adapted from Erika Pauwels' mu.semte.ch article. You can view it [here](https://mu.semte.ch/2017/11/23/adding-authentication-to-your-mu-project/)* - - - - -### Building a mail handling service -My goal for this short coding session is to have a mail handling service that will allow me to list and maninpulate mails through a JSON:API REST back-end. And have that service pick up when I write a mail to the database and send it automatically. You can see the result of this project at https://github.com/langens-jonathan/ReactiveMailServiceExample. - -#### Gain a head-start with mu-project -For this project I started with cloning the mu-project repository: -```bash -git clone https://github.com/mu-semtech/mu-project -``` - -This will give me the CRUD endpoint I need to manipulate my mail related resources. After cloning I rename the repository to MailBox and set the remote origin to a new one. For now I will leave the `README.md` file as it is. - -For the first block we will modify the `config/resources/domain.lisp`, `config/resourecs/repository.lisp` and the `config/dispatcher/dispatcher.ex` files. - -To add the necessary resource definitions, add them to the `domain.lisp` file as follows: - -```lisp -(define-resource mail () - :class (s-prefix "example:Mail") - :properties `((:sender :string ,(s-prefix "example:sender")) - (:subject :string ,(s-prefix "example:subject")) - (:content :string ,(s-prefix "example:content")) - (:ready :string ,(s-prefix "example:ready"))) - :resource-base (s-url "http://example.com/mails/") - :on-path "mails") -``` - -This will create a resource description that we can manipulate on route `/mails` with the properties sender, title, body and ready. - -Then add the prefix to the `repository.lisp` file: - -```lisp - (add-prefix "example" "http://example.com/") -``` - -We are almost there for a first test! The only thing left to do is to add the `/mails` route to the dispatcher (for more info check the documentation on http://mu.semte.ch). To do this add the following block of code to the `dispatcher.ex` file: - -``` -match "/mails/*path" do - Proxy.forward conn, path, "http://resource/mails/" -end -``` - -Now fire this up and lets see what we have by running the following command in the project root directory: - -```bash -docker-compose up -``` - -*Note: We don’t have a front-end, but with a tool like postman we can make GET, PATCH and POST calls to test the backend functionality.* - -A GET call to http://localhost/mails produces: -```json -{ - "data": [], - "links": { - "last": "/mails/", - "first": "/mails/" - } -} -``` - -Alright! Ok, no data yet, but we get back resource information. - -Lets do a post request to make a new mail resource: - -```conf -URL: http://localhost/mails -Headers: {"Content-Type":"application/vnd.api+json"} -Body: - { - "data":{ - "attributes":{ - "sender":"flowofcontrol@gmail.com", - "subject":"Mu Semtech Mail Server", - "content":"This is a test for the Mu Semtech Mail Server.", - "ready":"no" - }, - "type":"mails" - } - } -``` - -This gives us the following reponse: -```json -{ - "data": { - "attributes": { - "sender": "flowofcontrol@gmail.com", - "subject": "Mu Semtech Mail Server", - "content": "This is a test for the Mu Semtech Mail Server.", - "ready": "no" - }, - "id": "58978C2A6460170009000001", - "type": "mails", - "relationships": {} - } -} -``` - -That worked! In about 30 minutes we have a fully functional REST API endpoint for managing mail resources! - -To verify the original get request again, this now produces: -```json -{ - "data": { - "attributes": { - "sender": "flowofcontrol@gmail.com", - "subject": "Mu Semtech Mail Server", - "content": "This is a test for the Mu Semtech Mail Server.", - "ready": "no" - }, - "id": "58978C3A6460170009000002", - "type": "mails", - "relationships": {} - } -} -``` - -#### Enabling the reactive database -Before we can start writing our reactive mail managing micro-service, we will need to add a monitoring service to monitor the DB. This will be a lot easier than it sounds with mu.semte.ch. To start, open the `docker-compose.yml` file and add the following lines at the bottom of the file: - -```yaml -# ... -delta: - image: semtech/mu-delta-service:beta-0.7 - links: - - db:db - volumes: - - ./config/delta-service:/config - environment: - CONFIGFILE: "/config/config.properties" - SUBSCRIBERSFILE: "/config/subscribers.json" -``` - -This will add the monitoring service to our installation. The last thing to do for now is to change the link on the `resource` microservice by replacing -```yaml -links: - - db:database -``` -with -```yaml -links: - - delta:database -``` - -The final steps are to create the configuration and subscribers files. Create a file called `config.properties` at the location `config/delta-service/config.properties` and write the following lines in that file: - -```conf -# made by Langens Jonathan -queryURL=http://db:8890/sparql -updateURL=http://db:8890/sparql -sendUpdateInBody=true -calculateEffectives=true -``` - -and then create `config/delta-service/subscribers.json` and put this JSON inside: - -```json -{ - "potentials":[ - ], - "effectives":[ - ] -} -``` - -If we run `docker-compose rm` and then `docker-compose up` again, the delta service will be booting and already monitoring the changes that happen in the database! Of course we are not doing anything with them yet. So we will create a new micro-service just for this purpose. - -#### The mail-fetching microservice -The next step is to build our mail handling microservice. To do this we create a new directory called `mail-service` in our base directory. Then we create a file in that directory called `Dockerfile`. We will start from a mu.semte.ch template to make developing this microservice that much quicker. Mu.semte.ch has templates for a bunch of languages ruby, javascript, python, … For this microservice we will go for python 2.7. To do this we simply need to create a `web.py` file which will serve as the location for our code. Next add the following to the Dockerfile: - -```dockerfile -# mail-service/Dockerfile -FROM semtech/mu-python-template - -MAINTAINER Langens Jonathan -``` - -I know it doesn’t say much, but it doesn’t need to. The python template will handle the rest. - -Then we need to add some mail manipulating functionality. Since this is not really the objective of this post I create a `mail_helpers.py` file and paste the following code in there: -```python -# mail-service/mail_helpers.py -import sys -import imaplib -import getpass -import email -import datetime -import uuid -import helpers - -def save_mail(sender, date, subject, content): - str_uuid = str(uuid.uuid4()) - insert_query = "INSERT DATA\n{\nGRAPH \n{\n a ;\n" - insert_query += " \"" + sender + "\";\n" - insert_query += " \"" + date + "\";\n" - insert_query += " \"" + content + "\";\n" - insert_query += " \"" + subject + "\";\n" - insert_query += " \"" + str_uuid + "\".\n" - insert_query += "}\n}" - print "query:\n", insert_query - helpers.update(insert_query) - -def process_mailbox(mailbox): - rv, data = mailbox.search(None, "ALL") - if rv != 'OK': - print "No messages found!" - return - - for num in data[0].split(): - rv, data = mailbox.fetch(num, '(RFC822)') - if rv != 'OK': - print "ERROR getting message", num - return - - msg = email.message_from_string(data[0][1]) - content = str(msg.get_payload()) - content = content.replace('\n','') - - save_mail(msg['From'], msg['Date'], msg['Subject'], content) -``` - -As you can see the mail_helpers contain 2 functions, one to iterate over all emails in a mailbox and the other to save a single email to the triple store. Easy peasy! - -Next we create `web.py`. For more information on how the python template can be used you can visit: https://github.com/mu-semtech/mu-python-template. I created the following method to process all mails: -```python -# mail-service/web.py -@app.route("/fetchMails") -def fetchMailMethod(): - EMAIL_ADDRESS = "address" - EMAIL_PWD = "pwd" - - MAIL_SERVER = imaplib.IMAP4_SSL('imap.gmail.com') - - try: - MAIL_SERVER.login(EMAIL_ADDRESS, EMAIL_PWD) - except imaplib.IMAP4.error: - print "Logging into mailbox failed! " - - rv, data = MAIL_SERVER.select("INBOX") - if rv == 'OK': - mail_helpers.process_mailbox(MAIL_SERVER) - MAIL_SERVER.close() - - MAIL_SERVER.logout() - - return "ok" -``` - -This method is rather straightforward: it just opens a connection to an email address and opens the inbox mailbox. It then selects it for processing, thus inserting all mails into the triple store. - -At this point, we have: -- Defined a JSONAPI through which we can access our emails, using the standard mu.semte.ch stack -- Built a custom service which fetches the emails from our mail account and inserts them into the triplestore using the right model - -Now we will use these services in combination with the delta service, to discover which emails were inserted into the database, and to perform reactive computations on it. - -#### The delta service - -The delta service’s responsibilities are: - -- Acting as the SPARQL endpoint for the microservices -- Calculating the differences (deltas) that a query will introduce in the database -- Notifying interested parties of these differences - -For this hands on we use version beta-0.8 of the delta service. - -##### What do these delta reports look like? -There are 2 types of delta reports, you have potential inserts and effective inserts. A report for either will look like: -```json -{ - "delta": [ - { - "type": "effective", - "graph": "http://mu.semte.ch/application", - "inserts": [ - { - "s": { - "value": "http://example.com/mails/58B187FA6AA88E0009000001", - "type": "uri" - }, - "p": { - "value": "http://example.com/subject", - "type": "uri" - }, - "o": { - "value": "Mu Semtech Mail Server", - "type": "literal" - } - }, - ... -} -``` -*You can view the full version [here](https://gist.githubusercontent.com/langens-jonathan/cd5db8e9f68861662d888dad77f93662/raw/84adc69f9fd3143f45c05c0a5cefdf1ca9b95b55/gistfile1.txt).* - -A report states the query that was send, an array of inserted objects and an array of deleted objects: Inserted or deleted objects represent a single triple with s, p and o being subject, predicate and object. - -#### Expanding our mail handling microservice -We need to notify the delta service of the existence of our mail handling service. We do this using the `subscribers.json` file that was created before. Change it so it looks like: - -```json -{ - "potentials":[ - ], - "effectives":[ - "http://mailservice/process_delta" - ] -} -``` - -In the `docker-compose.yml` file we need to alter the delta-service definition to look like: - -```yaml - delta: - image: semtech/mu-delta-service:beta-0.8 - links: - - db:db - - mailservice:mailservice - volumes: - - ./config/delta-service:/config - environment: - CONFIGFILE: "/config/config.properties" - SUBSCRIBERSFILE: "/config/subscribers.json" -``` - -That way the delta service can talk to the mailservice. - -To handle delta reports in our mail handling microservice we will need 2 things: - -- Get access to the POST body of a request -- Process and manipulate JSON data - -To get access to this add the following imports to your `web.py` file: - -```python -import json -from flask import request -``` - -Then we define a new method that will: -- Handle the incoming delta reports -- Load the delta report into a variable -- Define some variables. - -Lastly we define an array that will hold the URI’s of all emails that need to be send. - - -```python -# mail-service/web.py -@app.route("/process_delta", methods=['POST']) -def processDelta(): - delta_report = json.loads(request.data) - mails_to_send = set() - predicate_mail_is_ready = "http://example.com/ready" - value_mail_is_ready = "yes" - # continued later... -``` - -We will loop over all inserted triples to check for mails that are ready to be send: -```python -# mail-service/web.py -def processDelta(): - # ... - # ...continuation - for delta in delta_report['delta']: - for triple in delta['inserts']: - if(triple['p']['value'] == predicate_mail_is_ready): - if(triple['o']['value'] == value_mail_is_ready): - mails_to_send.add(triple['s']['value']) - # continued later... -``` - -After this for loop has run, all the URI’s of mails that are ready to be send will be in the `mails_to_send` array. Now we loop over the array and query the database for each URI in the set. And then we will fetch a mail object for every URI that is in the set. - -Add the following code to `mail_helpers.py`: -```python -# mail-service/mail_helpers.py -def load_mail(uri): - # this query will find the mail (if it exists) - select_query = "SELECT DISTINCT ?uuid ?from ?ready ?subject ?content\n" - select_query += "WHERE \n{\n" - select_query += "<" + str(uri) + "> ?from;\n" - select_query += "a ;\n" - select_query += " ?content;\n" - select_query += " ?subject;\n" - select_query += " ?ready;\n" - select_query += " ?uuid.\n" - select_query += "}" - - # execute the query... - result = helpers.query(select_query) - - # if the length of the result array is 0 we return nil - if len(result['results']['bindings']) < 1: - return {} - - # I should probably check here but for a quick test application - # it doesn't matter that much. If there is more than 1 result - # that would indicate a data error - bindings = result['results']['bindings'][0] - - # we extract an object - mail = dict() - mail['uuid'] = bindings['uuid']['value'] - mail['sender'] = bindings['from']['value'] - mail['ready'] = bindings['ready']['value'] - mail['subject'] = bindings['subject']['value'] - mail['content'] = bindings['content']['value'] - - return mail -``` - -This function will load the mail object from the triple store. There is still the chance that the ready predicate was sent for some other object, for a mail that does not have all required fields, or for an object that is not a mail but happens to use the same predicate. - -We will use this function to try to load a mail object for each URI. Because the query was built without OPTIONAL statements, we are certain that an the dictionary returned by the load_mail function will either have all keys or none. - -To send the mail I have copied the entire `send_mail` function from http://naelshiab.com/tutorial-send-email-python/ and modified it slightly to take into account the dictionary object that now describes the mail. - -```python -# mail-service/mail_helpers.py -def send_mail(mail): - - fromaddr = "YOUR EMAIL" - toaddr = "EMAIL ADDRESS YOU SEND TO" - - msg = MIMEMultipart() - - msg['From'] = mail['from'] - msg['To'] = mail['to'] - msg['Subject'] = mail['subject'] - - body = mail['content'] - msg.attach(MIMEText(body, 'plain')) - - server = smtplib.SMTP('smtp.gmail.com', 587) - server.starttls() - server.login(fromaddr, "YOUR PASSWORD") - text = msg.as_string() - server.sendmail(fromaddr, toaddr, text) - server.quit() -``` - -The last thing that we need to do is to connect the list of URI’s to the send_mail function: -```python -# mail-service/web.py -def processDelta(): - # ...continuation - for uri in mails_to_send: - mail = mail_helpers.load_mail(uri) - if 'uuid' in mail.keys(): - mail_helpers.send_mail(mail, EMAIL_ADDRESS, EMAIL_PWD) -``` - -To test this you can send a POST request similar to this one to your local mu.semte.ch application on http://localhost/mails: - -```json -{"data":{ - "attributes":{ - "from":"flowofcontrol@gmail.com", - "subject":"A mail from the triple store", - "content":"This mail was sent by a micro service that listens to your triple store.", - "ready":"yes", - "to":"flowofcontrol@gmail.com" - }, - "type":"mails" - } -} -``` - -If all went well then the person whose email address you filled in in the to field will have gotten a mail from you. Good job! You've just created a mailing microservice. - -*This tutorial has been adapted from Jonathan Langens' mu.semte.ch articles. You can view them [here](https://mu.semte.ch/2017/02/16/reactive-microservice-hands-on-tutorial-part-1/) and [here](https://mu.semte.ch/2017/03/16/reactive-microservice-hands-on-tutorial-part-2/).* - -### Adding Ember FastBoot to your project - -![](http://mu.semte.ch/wp-content/uploads/2017/03/kuifje_op_de_maan-248x300.png) -In this post, we’re going elaborate a little on how to add Ember FastBoot to your mu-project. This should not be considered as a full blown tutorial, but rather as a set of notes to get you started. - -In a nutshell, Ember FastBoot introduces server side rendering on your ember app, which should not only improve user experience by serving static content first, but also make your website more SEO friendly. For more info, I would recommend you to check out [https://ember-fastboot.com/](https://ember-fastboot.com/). - -#### Setting the scene -All right, let’s get started. Assume you’re writing the new blogging app, called “mu-fastboot-example”. -It has a very simple data model with two entities. A blog post, which has a title, content, an author and many comments.  You can find the definition [here](https://github.com/cecemel/mu-fastboot-example-backend/blob/master/config/resources/domain.lisp). The backend needs a frontend of course and this has been published [here](https://github.com/cecemel/mu-fastboot-example-frontend). - -Assume for now, we only need an index page, which displays an overview of the current posts along with the number of comments to this post, and the authors of the comments.  A blog-post-summary component was created and  its template may be found [here](https://github.com/cecemel/mu-fastboot-example-frontend/blob/master/app/templates/components/blog-post-summary.hbs). - -Firing up both frontend and backend, your home page would look like this. - -![](http://mu.semte.ch/wp-content/uploads/2017/03/Screen-Shot-2017-03-23-at-13.48.46-226x300.png) - -Fetching your index page with a JavaScript disabled client, like e.g. curl, results in a totally SEO unfriendly, user unfriendly blank page, which waits till all resources are loaded before showing something. - -#### Adding FastBoot - -As [https://ember-fastboot.com/docs/user-guide#architecture](https://ember-fastboot.com/docs/user-guide#architecture) will tell you, two components are involved: the ember addon fastboot, and the application server itself, which will pre-prender your app. -Installing fast boot add on is as simple as typing: -```bash - ember install ember-cli-fastboot -``` - -The nice thing is, locally, you can immediately test the result of adding fastboot. Type -```bash - ember fastboot -``` -And to see the result (on port 3000): - -```bash - curl localhost:3000 -``` - -```hbs - -

Another even better post

- by cecemel. -

Celebrating the voidness!

-

comments: 0

-

comment authors:

- ` -``` - -#### caveats - -There is still an issue. As you might have noticed,  the second blog post doesn’t contain any comments or any comment authors. -This because, FastBoot decides returning the page to the client, once the _model()_ hook resolves (or _beforeModel(), afterModel()_). -If there is a component making an asynchronous call, e.g. counting the comments for each post, FastBook won’t consider this . -The trick is, to make sure these async calls are resolved before, the _model()_ hook is resolved. You could change _app/routes/index.js_  e.g. to the following: - -```js - import Ember from 'ember'; - - export default Ember.Route.extend({ - fastboot: Ember.inject.service(), - model() { - if (this.get('fastboot.isFastBoot')) { - return this.store.findAll('blog-post', {include: "comments"}); - } - return this.store.findAll('blog-post'); - } - }); -``` - -This result of this change, can be seen immediately: -```bash - curl localhost:3000 -``` - -```hbs - -

Another even better post

- by cecemel. -

Celebrating the voidness!

-

comments: 2

-

comment authors: An anonymous stranger A popular blogger

- -``` - -Unfortunately, this makes FastBoot a little less transparent then one would have initially hoped for. More information may be found at [FastBoot](https://ember-fastboot.com/docs/user-guide#use-model-hooks-to-defer-rendering). - -#### Deploying - -FastBoot has some nice deploy possibilities such as with AWS and Heroku, but in our case, we’ll go for the [custom server](https://ember-fastboot.com/docs/deploying#custom-server).  At the time of writing, the documentation is NOT up to date,  and as an app server you should use, [fastboot-app-server](https://github.com/ember-fastboot/fastboot-app-server) instead of [fastboot](https://github.com/ember-fastboot/fastboot). - -Just as with normal ember apps, in your app root, you build with -```bash - ember build -``` - -which should, among the normal files, also create a `dist/fastboot/` folder. -To host everything, you should follow the instructions described [here](https://github.com/ember-fastboot/fastboot-app-server). - -FORTUNATELY, to ease the deploy, a [docker image](https://github.com/cecemel/ember-fastboot-proxy-service) has been created, which can easily be added to your mu-project, like e.g. [here](https://github.com/cecemel/mu-fastboot-example-backend/blob/master/docker-compose.yml) . - -As usual, firing the project up with -```bash - docker-compose stop; docker-compose rm -f; docker-compose up -``` -and you should have a working app. - -So, that’s it. In case of questions, feel free to reach out. - -*This tutorial has been adapted from Felix Ruiz De Arcaute's mu.semte.ch article. You can view it [here](https://mu.semte.ch/2017/03/23/adding-ember-fastboot-to-your-mu-project/)* - - -### Adding a machine learning microservice to your mu.semte.ch project -In this post I want to explore how to add a machine learning microservice to any existing [mu.semte.ch](http://mu.semte.ch/) project. I want to be able to upload an image and add the labels for that image to the SPARQL store. - -![](http://mu.semte.ch/wp-content/uploads/2017/08/docter-semtec-farious.png) - -#### TensorFlow - -TensorFlow’s inception library is great for image classification, it is a deep neural network that is trained to recognize objects. We can remove and retrain the outer layer easily (you can find a tutorial by Google on it). The microservice we will use wraps Inception and offers 3 routes: - -- Add-Training-Example: through this route you tell the system that a certain image file is of a certain label -- Train: trains the model -- Classify: takes an image file and adds the classification to the triple store - -For an exisiting [mu.semte.ch](http://mu.semte.ch/) project you will probably want to add this microservice together with a trained graph and the use the classify route. While it is possible for a production system to learn, this may not be the best idea. Computers love to train, so they allocate all their computational resources to that, rather than keeping the rest smoothly running. - -#### Mu-image-classifier demo -I have prepared [a small example project](https://github.com/langens-jonathan/mu-image-classifier) where you can see and test the classifier microservice. After you clone this it has no trained graph so the classify route will not work. The architecture of this demo app is as in the image below: -![](http://mu.semte.ch/wp-content/uploads/2017/08/mu-image-classifier.png) - -#### Train the model - -So you have cloned the repository, cd in to the directory and type: - -```bash -docker-compose up -``` - -Open a browser and surf to [localhost](http://localhost). Click on the training route. To train the model you have to add classes (min 2) and then add images for those classes. Be sure not to use any other image format than JPG. After you have prepared the training set you can click the “Train” button. Your computer will be busy for a while now. If you need a more detailed tutorial on how to train there is one on the main page of the project you have just cloned, check the ‘/’ route! - -![](http://mu.semte.ch/wp-content/uploads/2017/08/Screenshot-from-2017-08-01-09-49-06s.png) - -#### Add the image classifier to a generic mu.semte.ch project -After you have a trained graph it is really as simple as adding the image classifier microservice to your [mu.semte.ch](http://mu.semte.ch/) project and all will work. The assumption is though that the vocabulary that is used to describe the files in your triple store is the one that is documented on the [mu-file-service](https://github.com/mu-semtech/file-service). - -Add this snippet to your docker-compose.yml: -```yml -classifier: - image: flowofcontrol/mu-tf-image-classifier - links: - - db:database - environment: - CLASSIFIER_TRESHHOLD: 0.7 - volumes: - - ./data/classifier/tf_files:/tf_files - - ./data/classifier/images:/images - - ./data/files:/files - ports: - - "6006:6006" -``` - - -As you can see we include 3 folders and expose a port. On that port you can also make use of TensorBoard, which is TensorFlow’s administration board and that gives you access to all kind of statistics about our image classifier. Add this to your dispatcher (if you want to be able to retrain, you also have to add the other routes): -```ex -match "/classify/*path" do - Proxy.forward conn, path, "http://classifier:5000/classify/" -end -``` - -The architecture of your app might then look somewhat like: -![](http://mu.semte.ch/wp-content/uploads/2017/08/integrating_mu-image-classifier.png) - -#### Classifying -If you then have an image you want to classify with it’s metadata correctly in the triple store then you can call the classify route and your image will be tagged. The response of the classify route will also tell you the probabilities for other labels. Below is what I get when I use the demo app to classify a random “Darth Vader” search result from images.google.com: - -![](http://mu.semte.ch/wp-content/uploads/2017/08/Screenshot-from-2017-08-01-10-04-13s.png) - -That’s all folks! - -*This tutorial has been adapted from Jonathan Langens' mu.semte.ch article. You can view it [here](https://mu.semte.ch/2017/08/03/adding-a-machine-learning-microservice-to-your-mu-semte-ch-project/)* +To help you find your feet with your first semantic works projects, we've collected [a few tutorials](TUTORIALS.md). diff --git a/TUTORIALS.md b/TUTORIALS.md new file mode 100644 index 0000000..06c94aa --- /dev/null +++ b/TUTORIALS.md @@ -0,0 +1,1034 @@ +## Tutorials +If you aren't familiar with the semantic.works stack/microservices yet, you might want to check out [why semantic tech?](https://mu.semte.ch/2017/03/23/adding-ember-fastboot-to-your-mu-project/) + +- [Creating a JSON API](#creating-a-json-api) +- [Adding authentication to your mu-project](#adding-authentication-to-your-mu-project) +- [Creating a mail service](#building-a-mail-handling-service) +- [Adding Ember Fastboot to your project](#adding-ember-fastboot-to-your-project) +- [Adding a machine learning microservice to your mu.semte.ch project](#adding-a-machine-learning-microservice-to-your-musemtech-project) + +### Creating a JSON API +Repetition is boring. Web applications oftentimes require the same functionality: to create, read, update and delete resources. Even if they operate in different domains. Or, in terms of a REST API, endpoints to GET, POST, PATCH and DELETE resources. Since productivity is one of the driving forces behind the mu.semte.ch architecture, the platform provides a microservice – [mu-cl-resources](https://github.com/mu-semtech/mu-cl-resources) – that generates a [JSONAPI](http://jsonapi.org/) compliant API for your resources based on a simple configuration describing the domain. In this tutorial we will explain how to setup such a configuration. + +#### Adding mu-cl-resources to your project +Like all microservices in the mu.semte.ch stack, mu-cl-resources is published as a Docker image. It just needs two configuration files in `./config/resources/`: + +- `domain.lisp`: describing the resources and relationships between them in your domain +- `repository.lisp`: defining prefixes for the vocabularies used in your domain + +To provide the configuration files, you can mount the files in the /config folder of the Docker image: +```yaml +services: + # ... + resource: + image: semtech/mu-cl-resources:1.20.0 + links: + - db:database + volumes: + - ./config:/config + # ... +``` +Alternatively you can build your own Docker image by extending the mu-cl-resources image and copying the configuration files in /config. See the [mu-cl-resources repo](https://github.com/mu-semtech/mu-cl-resources#mounting-the-config-files). + +The former option may be easier during development, while the latter is better suited in a production environment. With this last variant you can publish and release your service with its own version, independent of the mu-cl-resources version. + +When adding mu-cl-resources to our application, we also have to update the dispatcher configuration such that incoming requests get dispatched to our new service. We will update the dispatcher configuration at the end of this tutorial once we know on which endpoints our resources will be available. + +#### Describing your domain +Next step is to describe your domain in the configuration files. The configuration is written in Common Lisp. Don’t be intimidated, just follow the examples, make abstraction of all the parentheses and your’re good to go 🙂 As an example, we will describe the domain of the [ember-data-table demo](http://ember-data-table.semte.ch/) which [consists of books and their authors](https://github.com/erikap/books-service/tree/ember-data-table-example). + +##### repository.lisp +The `repository.lisp` file describes the prefixes for the vocabularies used in our domain model. + +To start, each configuration file starts with: + +```lisp +(in-package :mu-cl-resources) +``` + +Next, the prefixes are listed one per line as follows: + +```lisp +(in-package :mu-cl-resources) + +(add-prefix "dcterms" "http://purl.org/dc/terms/") +(add-prefix "schema" "http://schema.org/") +``` + +##### domain.lisp +The domain.lisp file describes your resources and the relationships between them. In this post we will describe the model of a book. Later on we will add an author model and specify the relationship between books and authors. + +Also start the `domain.lisp` file with the following line: +```lisp +(in-package :mu-cl-resources) +``` + +Next, add the basis of the book model: +```lisp +(in-package :mu-cl-resources) + +(define-resource book () + :class (s-prefix "schema:Book") + :resource-base (s-url "http://mu.semte.ch/services/github/madnificent/book-service/books/") +:on-path "books") +``` + +Although you may not have written a letter of Common Lisp before, you will probably be able to understand the lines of code above. + +- We define a book resource +- Each book will get schema:Book as RDF class +- Each book instance will be identified with a URI starting with “http://mu.semte.ch/services/github/madnificent/book-service/books/” – mu-cl-resources just appends a generated UUID to it +- The resources will be published on the /books API endpoint + +Finally, define the properties of a book: +```lisp +(in-package :mu-cl-resources) + +(define-resource book () + :class (s-prefix "schema:Book") + :properties `((:title :string ,(s-prefix "schema:headline")) + (:isbn :string ,(s-prefix "schema:isbn")) + (:publication-date :date ,(s-prefix "schema:datePublished")) + (:genre :string ,(s-prefix "schema:genre")) + (:language :string ,(s-prefix "schema:inLanguage")) + (:number-of-pages :integer ,(s-prefix "schema:numberOfPages"))) + :resource-base (s-url "http://mu.semte.ch/services/github/madnificent/book-service/books/") +:on-path "books") +``` + +Each property is described according to the format: + +```lisp +(:dasherized-property-name :type, (s-prefix "my-prefix:my-predicate")) +``` + +and will result in a triple: +``` + my-prefix:my-predicate "some-value" +``` + +#### Configuring the dispatcher + +Our book resources will be available on the /books paths. The mu-cl-resources service provides GET, POST, PATCH and DELETE operations on this path for free Assuming the books service is known as ‘resource’ in our dispatcher, we will add the following dispatch rule to our dispatcher configuration to forward the incoming requests to the books service: + +```lisp +match "/books/*path" do + Proxy.forward conn, path, "http://resource/books/" +end +``` + +Good job! The books can now be produced and consumed by the frontend through your JSONAPI compliant API. Now we will add relationships to the model. + + +#### The author model +Each book is written by (at least one) an author. An author isn’t a regular property of a book like - for example - the book’s title. It's a resource on its own. An author has its own properties like a name, a birth date etc. And it is related to a book. Before we can define the relationship between books and authors, we first need to specify the model of an author. + +The definition of the model is very similar to that of the book. Add the following lines to your `domain.lisp`: + +```lisp +(define-resource author () + :class (s-prefix "schema:Author") + :properties `((:name :string ,(s-prefix "schema:name"))) + :resource-base (s-url "http://mu.semte.ch/services/github/madnificent/book-service/authors/") +:on-path "authors") +``` + +Expose the author endpoints in the `dispatcher.ex` configuration: +``` +match "/authors/*path" do + Proxy.forward conn, path, "http://resource/authors/" +end +``` + +#### Defining relationships +Now that the author model is added, we can define the relationship between a book and an author. Let’s suppose a one-to-many relationship. A book has one author and an author may have written multiple books. + +First, extend the book’s model: +```lisp +(define-resource book () + :class (s-prefix "schema:Book") + :properties `((:title :string ,(s-prefix "schema:headline")) + (:isbn :string ,(s-prefix "schema:isbn")) + (:publication-date :date ,(s-prefix "schema:datePublished")) + (:genre :string ,(s-prefix "schema:genre")) + (:language :string ,(s-prefix "schema:inLanguage")) + (:number-of-pages :integer ,(s-prefix "schema:numberOfPages"))) + :has-one `((author :via ,(s-prefix "schema:author") + :as "author")) + :resource-base (s-url "http://mu.semte.ch/services/github/madnificent/book-service/books/") +:on-path "books") +``` + +Adding an author to a book will now result in the following triple in the store: +``` + + schema:author . +``` + +The :as “author” portion of the definition specifies the path on which the relationship will be exposed in the JSON representation of a book. + +```json +{ + "attributes": { + "title": "Rock & Roll with Ember" + }, + "id": "620f7a1c-9d31-4b8a-a627-2eb6904fe1f3", + "type": "books", + "relationships": { + "author": { + "links": { + "self": "/books/620f7a1c-9d31-4b8a-a627-2eb6904fe1f3/links/author", + "related": "/books/620f7a1c-9d31-4b8a-a627-2eb6904fe1f3/author" + } + } + } +} +``` + +Next, add the inverse relationship to the author’s model: + +```lisp +(define-resource author () + :class (s-prefix "schema:Author") + :properties `((:name :string ,(s-prefix "schema:name"))) + :has-many `((book :via (s-prefix "schema:author") + :inverse t + :as "books")) + :resource-base (s-url "http://mu.semte.ch/services/github/madnificent/book-service/authors/") +:on-path "authors") +``` + +The ‘:inverse t’ indicates that the relationship from author to books is the inverse relation. As you can see the the relationship’s path “books” is in plural since it’s a has-many relation in this case. + +If you want to define a many-to-many relationships between books and authors, just change the :has-one to :has-many and pluralize the “author” path to “authors” in the book’s model. Don’t forget to restart your microservice if you’ve updated the model. + +Now simply restart your microservice by running `docker-compose restart`, and you're done! + +#### Conclusion +That's it! Now you can [fetch](http://jsonapi.org/format/#fetching-relationships) and [update](http://jsonapi.org/format/#crud-updating-relationships) the relationships as specified by [jsonapi.org](http://jsonapi.org/). The generated API also supports the [include query parameter](http://jsonapi.org/format/#fetching-includes) to include related resources in the response when fetching one or more resource. That’s a lot you get for just a few lines of code, isn’t it? + +*This tutorial has been adapted from Erika Pauwels' mu.semte.ch articles. You can view them [here](https://mu.semte.ch/2017/07/27/generating-a-jsonapi-compliant-api-for-your-resources/) and [here](https://mu.semte.ch/2017/08/17/generating-a-jsonapi-compliant-api-for-your-resources-part-2/).* + + +### Adding authentication to your mu-project +![](http://mu.semte.ch/wp-content/uploads/2017/08/customumize_for_user-1024x768.png) + +Web applications oftentimes require a user to be authenticated to access (part of) their application. For example a webshop may require a user to be logged in before placing an order. In a previous blog post we already explained [the semantic model to represent logged in users](https://mu.semte.ch/2017/08/24/representing-logged-in-users/). In this post we will show how to enable authentication in your app. We assume you already have  a [mu-project](https://github.com/mu-semtech/mu-project) running. + +Adding authentication to your application consists of two tasks: + +- Adding registration so users can create a new account +- Adding a login service so users can authenticate themselves + +Both tasks require changes in the backend as well as in the frontend. Let’s start with the registration. + +#### Registration + +First, we will add registration to the project. The backend will be enriched with a microservice to manage accounts. The frontend will be augmented with an Ember addon providing components to register, unregister and  change a password. + +##### In the backend + +The [registration service](https://github.com/mu-semtech/registration-service) provides a service to create new accounts with a nickname and a password. To integrate the service in your project, add the following snippet to the `docker-compose.yml`. +```yaml +registration: + image: semtech/mu-registration-service:2.6.0 + links: + - database:database +``` + +(Re)start the project. +```bash +docker-compose up +``` +Next, configure the following routes in your dispatcher configuration in `config/dispatcher/dispatcher.ex`. +```ex +match "/accounts/\*path" do + Proxy.forward conn, path, "http://registration/accounts/" +end +``` + +Restart the dispatcher service . +```bash +docker-compose restart dispatcher +``` + +From now on all requests starting with ‘/accounts’ will be forwarded to the registration service. + +##### In the frontend + +We now have an endpoint for registration in the backend. We need a complementary component in the frontend that provides a GUI to communicate with this backend.  This component is offered by the [ember-mu-registration](https://www.npmjs.com/package/ember-mu-registration) addon. + +First, install the addon by executing the following command in your Ember project. +```bash +ember install ember-mu-registration +``` + +Next, just include the `{{mu-register}}`, `{{mu-unregister}}` and `{{mu-change-password}}` component in your template. + +```hbs +{{!-- app/templates/registration.hbs --}} +{{mu-register}} +``` + +The components will automatically send the correct requests to the backend. You can customize the component’s template and/or behavior as explained in the addon’s [README](https://github.com/mu-semtech/ember-mu-registration#advanced-usage). + +Finally create a new user account through the newly added mu-register component. We can use this user to validate the login in the next step. + +#### Login +Users can now create a new account, but how can they authenticate themselves in the app? In the next step we will enrich the backend with a login microservice and the frontend with a login form and a logout button. + +##### In the backend +The [login service](https://github.com/mu-semtech/login-service) provides a service to associate a session with a user’s account if the correct user credentials are provided. Have a look at [the semantic works docs](https://github.com/Denperidge-Redpencil/project/blob/master/docs/references/representing-logged-in-users.md) if you want to know the semantic model behind the users, sessions and accounts. To integrate the service in your project, add the following snippet to the `docker-compose.yml`. +```yaml +login: + image: semtech/mu-login-service:2.8.0 + links: + - database:database +``` + +(Re)start the project. +```bash +docker-compose up +``` + +Next, configure the following routes in your dispatcher configuration in `config/dispatcher/dispatcher.ex`. +```ex +match "/sessions/\*path" do + Proxy.forward conn, path, "http://login/sessions/" +end +``` + +Restart the dispatcher service . +```bash +docker-compose restart dispatcher +``` + +From now on all requests starting with ‘/sessions’ will be forwarded to the login service. + +#### In the frontend +Users can now be authenticated in the backend. Next, we need GUI components to login and logout and a mechanism to protect parts of the application so they are only accessible by authenticated users. These components are offered by the [ember-mu-login addon](https://github.com/mu-semtech/ember-mu-login) which requires [ember-simple-auth](https://github.com/simplabs/ember-simple-auth) to be installed, too. + +First, install the addons by executing the following commands in your Ember project. +```bash +ember install ember-simple-auth +ember install ember-mu-login +``` + +##### Login form +Next, we will generate a login route with a login form where the user can enter his credentials to authenticate. +```bash +ember generate route login +``` + +Add the `mu-login` component to the template. +```hbs +{{!-- app/templates/login.hbs --}} + +{{mu-login}} +``` + +##### Logout button +Once the user logged in, we will show a button so the user can logout. We will use [ember-simple-auth’s ‘isAuthenticated’ property](https://github.com/simplabs/ember-simple-auth#basic-usage) to check the current session’s state. The session service needs to be injected in the application controller. + +```js +// app/controllers/application.js +import Ember from 'ember'; + +export default Ember.Controller.extend({ + session: Ember.inject.service('session') + + // … +}); +``` + +Next, update the application’s template to show the logout button if the user is authenticated. +```hbs +{{!-- app/templates/application.hbs --}} +{{#if session.isAuthenticated}} + {{mu-logout}} +{{/if}} +``` + +Finally, mix the `ApplicationRouteMixin` in your application’s route. This mixin will automatically handle the [authenticationSucceeded](http://ember-simple-auth.com/api/classes/SessionService.html#event_authenticationSucceeded) and [invalidationSucceeded](http://ember-simple-auth.com/api/classes/SessionService.html#event_invalidationSucceeded) events. + +```js +// app/routes/application.js +import Ember from 'ember'; +import ApplicationRouteMixin from 'ember-simple-auth/mixins/application-route-mixin'; + +export default Ember.Route.extend(ApplicationRouteMixin); +``` + +##### Protecting routes + +Users can now login in the application, but they are still able to access all pages regardless whether they are authenticated or not. To make a route in the application accessible only when the session is authenticated, mix the [AuthenticatedRouteMixin](http://ember-simple-auth.com/api/classes/AuthenticatedRouteMixin.html) into the respective route: +```js +// app/routes/protected.js +import Ember from 'ember'; +import AuthenticatedRouteMixin from 'ember-simple-auth/mixins/authenticated-route-mixin'; + +export default Ember.Route.extend(AuthenticatedRouteMixin); +``` + +This will make the route (and all of its subroutes) transition to the ‘login’ route if the session is not authenticated. + +#### Finished +And that's it! Now you know how your mu-project can be easily augmented with authentication using a custom user registration service. + +*This tutorial has been adapted from Erika Pauwels' mu.semte.ch article. You can view it [here](https://mu.semte.ch/2017/11/23/adding-authentication-to-your-mu-project/)* + + + + +### Building a mail handling service +My goal for this short coding session is to have a mail handling service that will allow me to list and maninpulate mails through a JSON:API REST back-end. And have that service pick up when I write a mail to the database and send it automatically. You can see the result of this project at https://github.com/langens-jonathan/ReactiveMailServiceExample. + +#### Gain a head-start with mu-project +For this project I started with cloning the mu-project repository: +```bash +git clone https://github.com/mu-semtech/mu-project +``` + +This will give me the CRUD endpoint I need to manipulate my mail related resources. After cloning I rename the repository to MailBox and set the remote origin to a new one. For now I will leave the `README.md` file as it is. + +For the first block we will modify the `config/resources/domain.lisp`, `config/resourecs/repository.lisp` and the `config/dispatcher/dispatcher.ex` files. + +To add the necessary resource definitions, add them to the `domain.lisp` file as follows: + +```lisp +(define-resource mail () + :class (s-prefix "example:Mail") + :properties `((:sender :string ,(s-prefix "example:sender")) + (:subject :string ,(s-prefix "example:subject")) + (:content :string ,(s-prefix "example:content")) + (:ready :string ,(s-prefix "example:ready"))) + :resource-base (s-url "http://example.com/mails/") + :on-path "mails") +``` + +This will create a resource description that we can manipulate on route `/mails` with the properties sender, title, body and ready. + +Then add the prefix to the `repository.lisp` file: + +```lisp + (add-prefix "example" "http://example.com/") +``` + +We are almost there for a first test! The only thing left to do is to add the `/mails` route to the dispatcher (for more info check the documentation on http://mu.semte.ch). To do this add the following block of code to the `dispatcher.ex` file: + +``` +match "/mails/*path" do + Proxy.forward conn, path, "http://resource/mails/" +end +``` + +Now fire this up and lets see what we have by running the following command in the project root directory: + +```bash +docker-compose up +``` + +*Note: We don’t have a front-end, but with a tool like postman we can make GET, PATCH and POST calls to test the backend functionality.* + +A GET call to http://localhost/mails produces: +```json +{ + "data": [], + "links": { + "last": "/mails/", + "first": "/mails/" + } +} +``` + +Alright! Ok, no data yet, but we get back resource information. + +Lets do a post request to make a new mail resource: + +```conf +URL: http://localhost/mails +Headers: {"Content-Type":"application/vnd.api+json"} +Body: + { + "data":{ + "attributes":{ + "sender":"flowofcontrol@gmail.com", + "subject":"Mu Semtech Mail Server", + "content":"This is a test for the Mu Semtech Mail Server.", + "ready":"no" + }, + "type":"mails" + } + } +``` + +This gives us the following reponse: +```json +{ + "data": { + "attributes": { + "sender": "flowofcontrol@gmail.com", + "subject": "Mu Semtech Mail Server", + "content": "This is a test for the Mu Semtech Mail Server.", + "ready": "no" + }, + "id": "58978C2A6460170009000001", + "type": "mails", + "relationships": {} + } +} +``` + +That worked! In about 30 minutes we have a fully functional REST API endpoint for managing mail resources! + +To verify the original get request again, this now produces: +```json +{ + "data": { + "attributes": { + "sender": "flowofcontrol@gmail.com", + "subject": "Mu Semtech Mail Server", + "content": "This is a test for the Mu Semtech Mail Server.", + "ready": "no" + }, + "id": "58978C3A6460170009000002", + "type": "mails", + "relationships": {} + } +} +``` + +#### Enabling the reactive database +Before we can start writing our reactive mail managing micro-service, we will need to add a monitoring service to monitor the DB. This will be a lot easier than it sounds with mu.semte.ch. To start, open the `docker-compose.yml` file and add the following lines at the bottom of the file: + +```yaml +# ... +delta: + image: semtech/mu-delta-service:beta-0.7 + links: + - db:db + volumes: + - ./config/delta-service:/config + environment: + CONFIGFILE: "/config/config.properties" + SUBSCRIBERSFILE: "/config/subscribers.json" +``` + +This will add the monitoring service to our installation. The last thing to do for now is to change the link on the `resource` microservice by replacing +```yaml +links: + - db:database +``` +with +```yaml +links: + - delta:database +``` + +The final steps are to create the configuration and subscribers files. Create a file called `config.properties` at the location `config/delta-service/config.properties` and write the following lines in that file: + +```conf +# made by Langens Jonathan +queryURL=http://db:8890/sparql +updateURL=http://db:8890/sparql +sendUpdateInBody=true +calculateEffectives=true +``` + +and then create `config/delta-service/subscribers.json` and put this JSON inside: + +```json +{ + "potentials":[ + ], + "effectives":[ + ] +} +``` + +If we run `docker-compose rm` and then `docker-compose up` again, the delta service will be booting and already monitoring the changes that happen in the database! Of course we are not doing anything with them yet. So we will create a new micro-service just for this purpose. + +#### The mail-fetching microservice +The next step is to build our mail handling microservice. To do this we create a new directory called `mail-service` in our base directory. Then we create a file in that directory called `Dockerfile`. We will start from a mu.semte.ch template to make developing this microservice that much quicker. Mu.semte.ch has templates for a bunch of languages ruby, javascript, python, … For this microservice we will go for python 2.7. To do this we simply need to create a `web.py` file which will serve as the location for our code. Next add the following to the Dockerfile: + +```dockerfile +# mail-service/Dockerfile +FROM semtech/mu-python-template + +MAINTAINER Langens Jonathan +``` + +I know it doesn’t say much, but it doesn’t need to. The python template will handle the rest. + +Then we need to add some mail manipulating functionality. Since this is not really the objective of this post I create a `mail_helpers.py` file and paste the following code in there: +```python +# mail-service/mail_helpers.py +import sys +import imaplib +import getpass +import email +import datetime +import uuid +import helpers + +def save_mail(sender, date, subject, content): + str_uuid = str(uuid.uuid4()) + insert_query = "INSERT DATA\n{\nGRAPH \n{\n a ;\n" + insert_query += " \"" + sender + "\";\n" + insert_query += " \"" + date + "\";\n" + insert_query += " \"" + content + "\";\n" + insert_query += " \"" + subject + "\";\n" + insert_query += " \"" + str_uuid + "\".\n" + insert_query += "}\n}" + print "query:\n", insert_query + helpers.update(insert_query) + +def process_mailbox(mailbox): + rv, data = mailbox.search(None, "ALL") + if rv != 'OK': + print "No messages found!" + return + + for num in data[0].split(): + rv, data = mailbox.fetch(num, '(RFC822)') + if rv != 'OK': + print "ERROR getting message", num + return + + msg = email.message_from_string(data[0][1]) + content = str(msg.get_payload()) + content = content.replace('\n','') + + save_mail(msg['From'], msg['Date'], msg['Subject'], content) +``` + +As you can see the mail_helpers contain 2 functions, one to iterate over all emails in a mailbox and the other to save a single email to the triple store. Easy peasy! + +Next we create `web.py`. For more information on how the python template can be used you can visit: https://github.com/mu-semtech/mu-python-template. I created the following method to process all mails: +```python +# mail-service/web.py +@app.route("/fetchMails") +def fetchMailMethod(): + EMAIL_ADDRESS = "address" + EMAIL_PWD = "pwd" + + MAIL_SERVER = imaplib.IMAP4_SSL('imap.gmail.com') + + try: + MAIL_SERVER.login(EMAIL_ADDRESS, EMAIL_PWD) + except imaplib.IMAP4.error: + print "Logging into mailbox failed! " + + rv, data = MAIL_SERVER.select("INBOX") + if rv == 'OK': + mail_helpers.process_mailbox(MAIL_SERVER) + MAIL_SERVER.close() + + MAIL_SERVER.logout() + + return "ok" +``` + +This method is rather straightforward: it just opens a connection to an email address and opens the inbox mailbox. It then selects it for processing, thus inserting all mails into the triple store. + +At this point, we have: +- Defined a JSONAPI through which we can access our emails, using the standard mu.semte.ch stack +- Built a custom service which fetches the emails from our mail account and inserts them into the triplestore using the right model + +Now we will use these services in combination with the delta service, to discover which emails were inserted into the database, and to perform reactive computations on it. + +#### The delta service + +The delta service’s responsibilities are: + +- Acting as the SPARQL endpoint for the microservices +- Calculating the differences (deltas) that a query will introduce in the database +- Notifying interested parties of these differences + +For this hands on we use version beta-0.8 of the delta service. + +##### What do these delta reports look like? +There are 2 types of delta reports, you have potential inserts and effective inserts. A report for either will look like: +```json +{ + "delta": [ + { + "type": "effective", + "graph": "http://mu.semte.ch/application", + "inserts": [ + { + "s": { + "value": "http://example.com/mails/58B187FA6AA88E0009000001", + "type": "uri" + }, + "p": { + "value": "http://example.com/subject", + "type": "uri" + }, + "o": { + "value": "Mu Semtech Mail Server", + "type": "literal" + } + }, + ... +} +``` +*You can view the full version [here](https://gist.githubusercontent.com/langens-jonathan/cd5db8e9f68861662d888dad77f93662/raw/84adc69f9fd3143f45c05c0a5cefdf1ca9b95b55/gistfile1.txt).* + +A report states the query that was send, an array of inserted objects and an array of deleted objects: Inserted or deleted objects represent a single triple with s, p and o being subject, predicate and object. + +#### Expanding our mail handling microservice +We need to notify the delta service of the existence of our mail handling service. We do this using the `subscribers.json` file that was created before. Change it so it looks like: + +```json +{ + "potentials":[ + ], + "effectives":[ + "http://mailservice/process_delta" + ] +} +``` + +In the `docker-compose.yml` file we need to alter the delta-service definition to look like: + +```yaml + delta: + image: semtech/mu-delta-service:beta-0.8 + links: + - db:db + - mailservice:mailservice + volumes: + - ./config/delta-service:/config + environment: + CONFIGFILE: "/config/config.properties" + SUBSCRIBERSFILE: "/config/subscribers.json" +``` + +That way the delta service can talk to the mailservice. + +To handle delta reports in our mail handling microservice we will need 2 things: + +- Get access to the POST body of a request +- Process and manipulate JSON data + +To get access to this add the following imports to your `web.py` file: + +```python +import json +from flask import request +``` + +Then we define a new method that will: +- Handle the incoming delta reports +- Load the delta report into a variable +- Define some variables. + +Lastly we define an array that will hold the URI’s of all emails that need to be send. + + +```python +# mail-service/web.py +@app.route("/process_delta", methods=['POST']) +def processDelta(): + delta_report = json.loads(request.data) + mails_to_send = set() + predicate_mail_is_ready = "http://example.com/ready" + value_mail_is_ready = "yes" + # continued later... +``` + +We will loop over all inserted triples to check for mails that are ready to be send: +```python +# mail-service/web.py +def processDelta(): + # ... + # ...continuation + for delta in delta_report['delta']: + for triple in delta['inserts']: + if(triple['p']['value'] == predicate_mail_is_ready): + if(triple['o']['value'] == value_mail_is_ready): + mails_to_send.add(triple['s']['value']) + # continued later... +``` + +After this for loop has run, all the URI’s of mails that are ready to be send will be in the `mails_to_send` array. Now we loop over the array and query the database for each URI in the set. And then we will fetch a mail object for every URI that is in the set. + +Add the following code to `mail_helpers.py`: +```python +# mail-service/mail_helpers.py +def load_mail(uri): + # this query will find the mail (if it exists) + select_query = "SELECT DISTINCT ?uuid ?from ?ready ?subject ?content\n" + select_query += "WHERE \n{\n" + select_query += "<" + str(uri) + "> ?from;\n" + select_query += "a ;\n" + select_query += " ?content;\n" + select_query += " ?subject;\n" + select_query += " ?ready;\n" + select_query += " ?uuid.\n" + select_query += "}" + + # execute the query... + result = helpers.query(select_query) + + # if the length of the result array is 0 we return nil + if len(result['results']['bindings']) < 1: + return {} + + # I should probably check here but for a quick test application + # it doesn't matter that much. If there is more than 1 result + # that would indicate a data error + bindings = result['results']['bindings'][0] + + # we extract an object + mail = dict() + mail['uuid'] = bindings['uuid']['value'] + mail['sender'] = bindings['from']['value'] + mail['ready'] = bindings['ready']['value'] + mail['subject'] = bindings['subject']['value'] + mail['content'] = bindings['content']['value'] + + return mail +``` + +This function will load the mail object from the triple store. There is still the chance that the ready predicate was sent for some other object, for a mail that does not have all required fields, or for an object that is not a mail but happens to use the same predicate. + +We will use this function to try to load a mail object for each URI. Because the query was built without OPTIONAL statements, we are certain that an the dictionary returned by the load_mail function will either have all keys or none. + +To send the mail I have copied the entire `send_mail` function from http://naelshiab.com/tutorial-send-email-python/ and modified it slightly to take into account the dictionary object that now describes the mail. + +```python +# mail-service/mail_helpers.py +def send_mail(mail): + + fromaddr = "YOUR EMAIL" + toaddr = "EMAIL ADDRESS YOU SEND TO" + + msg = MIMEMultipart() + + msg['From'] = mail['from'] + msg['To'] = mail['to'] + msg['Subject'] = mail['subject'] + + body = mail['content'] + msg.attach(MIMEText(body, 'plain')) + + server = smtplib.SMTP('smtp.gmail.com', 587) + server.starttls() + server.login(fromaddr, "YOUR PASSWORD") + text = msg.as_string() + server.sendmail(fromaddr, toaddr, text) + server.quit() +``` + +The last thing that we need to do is to connect the list of URI’s to the send_mail function: +```python +# mail-service/web.py +def processDelta(): + # ...continuation + for uri in mails_to_send: + mail = mail_helpers.load_mail(uri) + if 'uuid' in mail.keys(): + mail_helpers.send_mail(mail, EMAIL_ADDRESS, EMAIL_PWD) +``` + +To test this you can send a POST request similar to this one to your local mu.semte.ch application on http://localhost/mails: + +```json +{"data":{ + "attributes":{ + "from":"flowofcontrol@gmail.com", + "subject":"A mail from the triple store", + "content":"This mail was sent by a micro service that listens to your triple store.", + "ready":"yes", + "to":"flowofcontrol@gmail.com" + }, + "type":"mails" + } +} +``` + +If all went well then the person whose email address you filled in in the to field will have gotten a mail from you. Good job! You've just created a mailing microservice. + +*This tutorial has been adapted from Jonathan Langens' mu.semte.ch articles. You can view them [here](https://mu.semte.ch/2017/02/16/reactive-microservice-hands-on-tutorial-part-1/) and [here](https://mu.semte.ch/2017/03/16/reactive-microservice-hands-on-tutorial-part-2/).* + +### Adding Ember FastBoot to your project + +![](http://mu.semte.ch/wp-content/uploads/2017/03/kuifje_op_de_maan-248x300.png) +In this post, we’re going elaborate a little on how to add Ember FastBoot to your mu-project. This should not be considered as a full blown tutorial, but rather as a set of notes to get you started. + +In a nutshell, Ember FastBoot introduces server side rendering on your ember app, which should not only improve user experience by serving static content first, but also make your website more SEO friendly. For more info, I would recommend you to check out [https://ember-fastboot.com/](https://ember-fastboot.com/). + +#### Setting the scene +All right, let’s get started. Assume you’re writing the new blogging app, called “mu-fastboot-example”. +It has a very simple data model with two entities. A blog post, which has a title, content, an author and many comments.  You can find the definition [here](https://github.com/cecemel/mu-fastboot-example-backend/blob/master/config/resources/domain.lisp). The backend needs a frontend of course and this has been published [here](https://github.com/cecemel/mu-fastboot-example-frontend). + +Assume for now, we only need an index page, which displays an overview of the current posts along with the number of comments to this post, and the authors of the comments.  A blog-post-summary component was created and  its template may be found [here](https://github.com/cecemel/mu-fastboot-example-frontend/blob/master/app/templates/components/blog-post-summary.hbs). + +Firing up both frontend and backend, your home page would look like this. + +![](http://mu.semte.ch/wp-content/uploads/2017/03/Screen-Shot-2017-03-23-at-13.48.46-226x300.png) + +Fetching your index page with a JavaScript disabled client, like e.g. curl, results in a totally SEO unfriendly, user unfriendly blank page, which waits till all resources are loaded before showing something. + +#### Adding FastBoot + +As [https://ember-fastboot.com/docs/user-guide#architecture](https://ember-fastboot.com/docs/user-guide#architecture) will tell you, two components are involved: the ember addon fastboot, and the application server itself, which will pre-prender your app. +Installing fast boot add on is as simple as typing: +```bash + ember install ember-cli-fastboot +``` + +The nice thing is, locally, you can immediately test the result of adding fastboot. Type +```bash + ember fastboot +``` +And to see the result (on port 3000): + +```bash + curl localhost:3000 +``` + +```hbs + +

Another even better post

+ by cecemel. +

Celebrating the voidness!

+

comments: 0

+

comment authors:

+ ` +``` + +#### caveats + +There is still an issue. As you might have noticed,  the second blog post doesn’t contain any comments or any comment authors. +This because, FastBoot decides returning the page to the client, once the _model()_ hook resolves (or _beforeModel(), afterModel()_). +If there is a component making an asynchronous call, e.g. counting the comments for each post, FastBook won’t consider this . +The trick is, to make sure these async calls are resolved before, the _model()_ hook is resolved. You could change _app/routes/index.js_  e.g. to the following: + +```js + import Ember from 'ember'; + + export default Ember.Route.extend({ + fastboot: Ember.inject.service(), + model() { + if (this.get('fastboot.isFastBoot')) { + return this.store.findAll('blog-post', {include: "comments"}); + } + return this.store.findAll('blog-post'); + } + }); +``` + +This result of this change, can be seen immediately: +```bash + curl localhost:3000 +``` + +```hbs + +

Another even better post

+ by cecemel. +

Celebrating the voidness!

+

comments: 2

+

comment authors: An anonymous stranger A popular blogger

+ +``` + +Unfortunately, this makes FastBoot a little less transparent then one would have initially hoped for. More information may be found at [FastBoot](https://ember-fastboot.com/docs/user-guide#use-model-hooks-to-defer-rendering). + +#### Deploying + +FastBoot has some nice deploy possibilities such as with AWS and Heroku, but in our case, we’ll go for the [custom server](https://ember-fastboot.com/docs/deploying#custom-server).  At the time of writing, the documentation is NOT up to date,  and as an app server you should use, [fastboot-app-server](https://github.com/ember-fastboot/fastboot-app-server) instead of [fastboot](https://github.com/ember-fastboot/fastboot). + +Just as with normal ember apps, in your app root, you build with +```bash + ember build +``` + +which should, among the normal files, also create a `dist/fastboot/` folder. +To host everything, you should follow the instructions described [here](https://github.com/ember-fastboot/fastboot-app-server). + +FORTUNATELY, to ease the deploy, a [docker image](https://github.com/cecemel/ember-fastboot-proxy-service) has been created, which can easily be added to your mu-project, like e.g. [here](https://github.com/cecemel/mu-fastboot-example-backend/blob/master/docker-compose.yml) . + +As usual, firing the project up with +```bash + docker-compose stop; docker-compose rm -f; docker-compose up +``` +and you should have a working app. + +So, that’s it. In case of questions, feel free to reach out. + +*This tutorial has been adapted from Felix Ruiz De Arcaute's mu.semte.ch article. You can view it [here](https://mu.semte.ch/2017/03/23/adding-ember-fastboot-to-your-mu-project/)* + + +### Adding a machine learning microservice to your mu.semte.ch project +In this post I want to explore how to add a machine learning microservice to any existing [mu.semte.ch](http://mu.semte.ch/) project. I want to be able to upload an image and add the labels for that image to the SPARQL store. + +![](http://mu.semte.ch/wp-content/uploads/2017/08/docter-semtec-farious.png) + +#### TensorFlow + +TensorFlow’s inception library is great for image classification, it is a deep neural network that is trained to recognize objects. We can remove and retrain the outer layer easily (you can find a tutorial by Google on it). The microservice we will use wraps Inception and offers 3 routes: + +- Add-Training-Example: through this route you tell the system that a certain image file is of a certain label +- Train: trains the model +- Classify: takes an image file and adds the classification to the triple store + +For an exisiting [mu.semte.ch](http://mu.semte.ch/) project you will probably want to add this microservice together with a trained graph and the use the classify route. While it is possible for a production system to learn, this may not be the best idea. Computers love to train, so they allocate all their computational resources to that, rather than keeping the rest smoothly running. + +#### Mu-image-classifier demo +I have prepared [a small example project](https://github.com/langens-jonathan/mu-image-classifier) where you can see and test the classifier microservice. After you clone this it has no trained graph so the classify route will not work. The architecture of this demo app is as in the image below: +![](http://mu.semte.ch/wp-content/uploads/2017/08/mu-image-classifier.png) + +#### Train the model + +So you have cloned the repository, cd in to the directory and type: + +```bash +docker-compose up +``` + +Open a browser and surf to [localhost](http://localhost). Click on the training route. To train the model you have to add classes (min 2) and then add images for those classes. Be sure not to use any other image format than JPG. After you have prepared the training set you can click the “Train” button. Your computer will be busy for a while now. If you need a more detailed tutorial on how to train there is one on the main page of the project you have just cloned, check the ‘/’ route! + +![](http://mu.semte.ch/wp-content/uploads/2017/08/Screenshot-from-2017-08-01-09-49-06s.png) + +#### Add the image classifier to a generic mu.semte.ch project +After you have a trained graph it is really as simple as adding the image classifier microservice to your [mu.semte.ch](http://mu.semte.ch/) project and all will work. The assumption is though that the vocabulary that is used to describe the files in your triple store is the one that is documented on the [mu-file-service](https://github.com/mu-semtech/file-service). + +Add this snippet to your docker-compose.yml: +```yml +classifier: + image: flowofcontrol/mu-tf-image-classifier + links: + - db:database + environment: + CLASSIFIER_TRESHHOLD: 0.7 + volumes: + - ./data/classifier/tf_files:/tf_files + - ./data/classifier/images:/images + - ./data/files:/files + ports: + - "6006:6006" +``` + + +As you can see we include 3 folders and expose a port. On that port you can also make use of TensorBoard, which is TensorFlow’s administration board and that gives you access to all kind of statistics about our image classifier. Add this to your dispatcher (if you want to be able to retrain, you also have to add the other routes): +```ex +match "/classify/*path" do + Proxy.forward conn, path, "http://classifier:5000/classify/" +end +``` + +The architecture of your app might then look somewhat like: +![](http://mu.semte.ch/wp-content/uploads/2017/08/integrating_mu-image-classifier.png) + +#### Classifying +If you then have an image you want to classify with it’s metadata correctly in the triple store then you can call the classify route and your image will be tagged. The response of the classify route will also tell you the probabilities for other labels. Below is what I get when I use the demo app to classify a random “Darth Vader” search result from images.google.com: + +![](http://mu.semte.ch/wp-content/uploads/2017/08/Screenshot-from-2017-08-01-10-04-13s.png) + +That’s all folks! + +*This tutorial has been adapted from Jonathan Langens' mu.semte.ch article. You can view it [here](https://mu.semte.ch/2017/08/03/adding-a-machine-learning-microservice-to-your-mu-semte-ch-project/)* From 089358bb76b734dd91761eebcd959e8f77eb4845 Mon Sep 17 00:00:00 2001 From: Rich Date: Thu, 27 Jul 2023 10:57:51 +0200 Subject: [PATCH 02/16] Add ember UI tutorial from https://mu.semte.ch/getting-started/ --- TUTORIALS.md | 159 +++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 159 insertions(+) diff --git a/TUTORIALS.md b/TUTORIALS.md index 06c94aa..64f2552 100644 --- a/TUTORIALS.md +++ b/TUTORIALS.md @@ -1,7 +1,10 @@ ## Tutorials If you aren't familiar with the semantic.works stack/microservices yet, you might want to check out [why semantic tech?](https://mu.semte.ch/2017/03/23/adding-ember-fastboot-to-your-mu-project/) +Each of these tutorials starts with a mu-project docker-compose set-up. + - [Creating a JSON API](#creating-a-json-api) +- [Adding an ember UI to your project](#adding-an-ember-ui-to-your-project) - [Adding authentication to your mu-project](#adding-authentication-to-your-mu-project) - [Creating a mail service](#building-a-mail-handling-service) - [Adding Ember Fastboot to your project](#adding-ember-fastboot-to-your-project) @@ -210,6 +213,162 @@ That's it! Now you can [fetch](http://jsonapi.org/format/#fetching-relationships *This tutorial has been adapted from Erika Pauwels' mu.semte.ch articles. You can view them [here](https://mu.semte.ch/2017/07/27/generating-a-jsonapi-compliant-api-for-your-resources/) and [here](https://mu.semte.ch/2017/08/17/generating-a-jsonapi-compliant-api-for-your-resources-part-2/).* +### Adding an ember UI to your project +This tutorial builds on the [previous one](#creating-a-json-api) to add a UI to manage books using [EmberJS](https://www.emberjs.com/). + +#### Ember in the frontend + +Our end-users access the services through EmberJS application. This provides us with an integrated, styled and flexible view of the enabled microservices. We’ll create a new ember application to allow end-users to list, create, and delete authors. The advised way to build and develop EmberJS applications is using ember-cli. + +You can install ember-cli from ember-cli.com, or you can use the ember-docker found at https://github.com/madnificent/docker-ember . Our examples assume you’ll use ember-docker. + +#### Build a new app + +First we create the new application. The command is short, but it may take a while to fetch all NPM dependencies. Grab a coffee while the computer works for you. + +```sh +edi ember new books +``` + +#### Live reloading changes + +Let’s see if our new application runs. Go into the books directory and run the ember serve command (available as eds). Once the files have compiled, you can visit the site in your browser at localhost:4200. + +```sh +cd books +eds --proxy http://host # alt: ember serve --proxy http://localhost:80/ +``` + +The proxy connects to our localhost on port 80 (yes, it’s called host in the ember-docker, rather than localhost). We’ll use this later to fetch content from the microservices. Let’s alter the title of our application, the browser’s view will update automatically. Open app/application.hbs and change the following: + +```diff +- {{!-- The following component displays Ember's default welcome message. --}} +- {{welcome-page}} +- {{!-- Feel free to remove this! --}} ++

My books

+``` + +Boom, automatic updates in the browser. + +#### Connecting + +EmberJS applications roughly follow the Web-MVC pattern. The applications have a rigid folder-structure, most content being in the app folder. Ember-cli uses generators to generate basic stubs of content. We create the books model, route and controller using ember-cli. Check the helpers for ember generate model, ember generate route and ember generate controller, or the following: + +```sh +edi ember generate model book title:string isbn:string +edi ember generate route book +edi ember generate controller book +``` + +The terminal output shows the created and updated files. (note: generating new files can make watched files fail in Docker, just kill and restart eds should that happen.) + +We will fetch all books and render them in our template. In routes/book.js: + +```diff += export default class BooksRoute extends Route { ++ model(){ ++ return this.store.findAll('book'); ++ } += } +``` + +We’ll display the found records in our template so we’re able to see the created records later on. Add the following to templates/book.hbs + +```diff ++
    ++ {{#each @model as |book|}} ++
  • {{book.title}} {{book.isbn}}
    • ++ {{/each}} ++
+``` + +#### Creating new books + +We’ll add a small input-form through which we can create new books at the bottom of our listing. Two input fields and a create button will suffice for our example. + +In the app/templates/book.hbs template, we’ll add our create fields and button: + +```diff ++
++
++
++
Book title
++
++ ++
++
ISBN
++
++ ++
++
++ ++
+``` + +We’ll add this action in the controller and make it create the new book. In app/controllers/book.hbs add the following: + +```diff += import Controller from '@ember/controller'; ++ import { action } from '@ember/object'; ++ import { tracked } from '@glimmer/tracking'; ++ import { inject as service } from '@ember/service'; += += export default class BooksController extends Controller { ++ @tracked newTitle = ''; ++ @tracked newIsbn = ''; ++ ++ @service store; ++ ++ @action ++ createBook(event) { ++ event.preventDefault(); ++ // create the new book ++ const book = this.store.createRecord('book', { ++ title: this.newTitle, ++ isbn: this.newIsbn ++ }); ++ book.save() ++ // clear the input fields ++ this.newTitle = ''; ++ this.newIsbn = ''; ++ } += }); +``` + +#### Removing books + +Removing books follows a similar path to creating new books. We add a delete button to the template, and a delete action to the controller. + +In app/templates/book.hbs we alter: + +```diff +=
    += {{#each @model as |book|}} ++
  • ++ {{book.title}}{{book.isbn}} ++ ++
    • += {{/each}} +=
+``` + +In app/controllers/book.hbs we alter: + +```diff += this.newTitle = ''; += this.newIsbn = ''; += } ++ ++ @action ++ removeBook( book, event ) { ++ event.preventDefault(); ++ book.destroyRecord(); += } += } +``` + ### Adding authentication to your mu-project ![](http://mu.semte.ch/wp-content/uploads/2017/08/customumize_for_user-1024x768.png) From 99df2deced5de5ea9324f13f851f7d669f5bf28e Mon Sep 17 00:00:00 2001 From: Rich Date: Thu, 27 Jul 2023 11:12:28 +0200 Subject: [PATCH 03/16] Pull some author properties from 'masterclass' slide --- TUTORIALS.md | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/TUTORIALS.md b/TUTORIALS.md index 64f2552..23d3c76 100644 --- a/TUTORIALS.md +++ b/TUTORIALS.md @@ -131,7 +131,9 @@ The definition of the model is very similar to that of the book. Add the followi ```lisp (define-resource author () :class (s-prefix "schema:Author") - :properties `((:name :string ,(s-prefix "schema:name"))) + :properties `((:name :string ,(s-prefix "schema:name")) + (:given-name :string ,(s-prefix "foaf:givenName")) + (:family-name :string ,(s-prefix "foaf:familyName"))) :resource-base (s-url "http://mu.semte.ch/services/github/madnificent/book-service/authors/") :on-path "authors") ``` From de135297d1fe2d92d85eb9dfbe15ad3e55206813 Mon Sep 17 00:00:00 2001 From: Rich Date: Thu, 27 Jul 2023 11:38:31 +0200 Subject: [PATCH 04/16] Improve note around use of docker-ember in ember tutorial --- TUTORIALS.md | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/TUTORIALS.md b/TUTORIALS.md index 23d3c76..716ef37 100644 --- a/TUTORIALS.md +++ b/TUTORIALS.md @@ -222,7 +222,7 @@ This tutorial builds on the [previous one](#creating-a-json-api) to add a UI to Our end-users access the services through EmberJS application. This provides us with an integrated, styled and flexible view of the enabled microservices. We’ll create a new ember application to allow end-users to list, create, and delete authors. The advised way to build and develop EmberJS applications is using ember-cli. -You can install ember-cli from ember-cli.com, or you can use the ember-docker found at https://github.com/madnificent/docker-ember . Our examples assume you’ll use ember-docker. +You can install ember-cli by following the instructions on [the emberjs website](https://cli.emberjs.com/), or you can keep everything in docker using the [ember-docker scripts](https://github.com/madnificent/docker-ember). Our examples assume you’ll use ember-docker, which provides the `edi` and `eds` commands. If you're using a globally installed ember-cli, simply remove `edi` from any commands and replace `eds` with `ember serve`. #### Build a new app @@ -230,6 +230,8 @@ First we create the new application. The command is short, but it may take a wh ```sh edi ember new books +# Or if using a globally installed ember-cli (see above) +# ember new books ``` #### Live reloading changes From 186af76d7e5704581d7c347a5dff39ca6586a3a4 Mon Sep 17 00:00:00 2001 From: Rich Date: Thu, 27 Jul 2023 12:44:04 +0200 Subject: [PATCH 05/16] Ember tut: Tweak diffs and plurals in commands to match reality and conventions --- TUTORIALS.md | 40 ++++++++++++++++++++++------------------ 1 file changed, 22 insertions(+), 18 deletions(-) diff --git a/TUTORIALS.md b/TUTORIALS.md index 716ef37..eb67a95 100644 --- a/TUTORIALS.md +++ b/TUTORIALS.md @@ -260,28 +260,35 @@ EmberJS applications roughly follow the Web-MVC pattern. The applications have ```sh edi ember generate model book title:string isbn:string -edi ember generate route book -edi ember generate controller book +edi ember generate route books +edi ember generate controller books ``` The terminal output shows the created and updated files. (note: generating new files can make watched files fail in Docker, just kill and restart eds should that happen.) -We will fetch all books and render them in our template. In routes/book.js: +We will fetch all books and render them in our template. In routes/books.js: ```diff += import Route from '@ember/routing/route'; ++ import { inject as service } from '@ember/service'; += = export default class BooksRoute extends Route { ++ @service store; ++ + model(){ + return this.store.findAll('book'); + } = } ``` -We’ll display the found records in our template so we’re able to see the created records later on. Add the following to templates/book.hbs +We’ll display the found records in our template so we’re able to see the created records later on. Add the following to templates/books.hbs ```diff +
    + {{#each @model as |book|}} -+
  • {{book.title}} {{book.isbn}}
    • ++
  • ++ {{book.title}} {{book.isbn}} ++
    • + {{/each}} +
``` @@ -311,7 +318,7 @@ In the app/templates/book.hbs template, we’ll add our create fields and button + ``` -We’ll add this action in the controller and make it create the new book. In app/controllers/book.hbs add the following: +We’ll add this action in the controller and make it create the new book. In app/controllers/books.js add the following: ```diff = import Controller from '@ember/controller'; @@ -319,7 +326,8 @@ We’ll add this action in the controller and make it create the new book. In a + import { tracked } from '@glimmer/tracking'; + import { inject as service } from '@ember/service'; = -= export default class BooksController extends Controller { +- export default class BooksController extends Controller {} ++ export default class BooksController extends Controller { + @tracked newTitle = ''; + @tracked newIsbn = ''; + @@ -338,7 +346,7 @@ We’ll add this action in the controller and make it create the new book. In a + this.newTitle = ''; + this.newIsbn = ''; + } -= }); ++ }); ``` #### Removing books @@ -348,17 +356,13 @@ Removing books follows a similar path to creating new books. We add a delete bu In app/templates/book.hbs we alter: ```diff -=
    -= {{#each @model as |book|}} -+
  • -+ {{book.title}}{{book.isbn}} +=
  • += {{book.title}}{{book.isbn}} + -+
    • -= {{/each}} -=
+= ``` -In app/controllers/book.hbs we alter: +In app/controllers/books.js we alter: ```diff = this.newTitle = ''; @@ -366,10 +370,10 @@ In app/controllers/book.hbs we alter: = } + + @action -+ removeBook( book, event ) { ++ removeBook(book, event) { + event.preventDefault(); + book.destroyRecord(); -= } ++ } = } ``` From 71f6e11d4b4c6e77b60d2eb6a01a4b378326d32e Mon Sep 17 00:00:00 2001 From: Rich Date: Thu, 27 Jul 2023 12:55:19 +0200 Subject: [PATCH 06/16] Add adapter and serializer generation to ember tutorial --- TUTORIALS.md | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) diff --git a/TUTORIALS.md b/TUTORIALS.md index eb67a95..f87e556 100644 --- a/TUTORIALS.md +++ b/TUTORIALS.md @@ -256,7 +256,14 @@ Boom, automatic updates in the browser. #### Connecting -EmberJS applications roughly follow the Web-MVC pattern. The applications have a rigid folder-structure, most content being in the app folder. Ember-cli uses generators to generate basic stubs of content. We create the books model, route and controller using ember-cli. Check the helpers for ember generate model, ember generate route and ember generate controller, or the following: +EmberJS applications roughly follow the Web-MVC pattern. The applications have a rigid folder-structure, most content being in the app folder. Ember-cli uses generators to generate basic stubs of content. Since the APIs we're using follow the json-api specification, we can avoid writing custom adapter and serialiser code by simply generating default ones for our application to use if a specific one is not specified: + +```sh +edi ember generate adapter application +edi ember generate serializer application +``` + +We create the books model, route and controller using ember-cli. Check the helpers for ember generate model, ember generate route and ember generate controller, or the following: ```sh edi ember generate model book title:string isbn:string From 8327482ea2dcd0582992a84bd292b10c75169b98 Mon Sep 17 00:00:00 2001 From: Rich Date: Thu, 27 Jul 2023 13:03:02 +0200 Subject: [PATCH 07/16] Linting tweaks to ember tutorial --- TUTORIALS.md | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/TUTORIALS.md b/TUTORIALS.md index f87e556..b698118 100644 --- a/TUTORIALS.md +++ b/TUTORIALS.md @@ -282,7 +282,7 @@ We will fetch all books and render them in our template. In routes/books.js: = export default class BooksRoute extends Route { + @service store; + -+ model(){ ++ model() { + return this.store.findAll('book'); + } = } @@ -313,12 +313,12 @@ In the app/templates/book.hbs template, we’ll add our create fields and button +
Book title
+
+ ++ placeholder="Thinking Fast and Slow" /> +
+
ISBN
+
+ ++ placeholder="978-0374533557" /> +
+ + @@ -346,9 +346,9 @@ We’ll add this action in the controller and make it create the new book. In a + // create the new book + const book = this.store.createRecord('book', { + title: this.newTitle, -+ isbn: this.newIsbn ++ isbn: this.newIsbn, + }); -+ book.save() ++ book.save(); + // clear the input fields + this.newTitle = ''; + this.newIsbn = ''; @@ -365,7 +365,7 @@ In app/templates/book.hbs we alter: ```diff =
  • = {{book.title}}{{book.isbn}} -+ ++ =
    • ``` From c25bede3878f652687c3b57c2352295ff215c137 Mon Sep 17 00:00:00 2001 From: Rich Date: Thu, 27 Jul 2023 15:34:47 +0200 Subject: [PATCH 08/16] Add warning about open file limits to readme --- README.md | 29 +++++++++++++++++++++++++++++ 1 file changed, 29 insertions(+) diff --git a/README.md b/README.md index 64411a7..cda8b00 100644 --- a/README.md +++ b/README.md @@ -12,6 +12,9 @@ Setting up your environment is done in three easy steps: 2. Then, configure how requests are dispatched in `config/dispatcher.ex` 3. Lastly, simply start the docker-compose. +> [!WARNING] +> Many of the containers used have issues with high limits on open file descriptors, so you might need [to work around this](#containers-stuck-while-starting-using-100-cpu) + #### Hooking things up with docker-compose Alter the `docker-compose.yml` file so it contains all microservices you need. The example content should be clear, but you can find more information in the [Docker Compose documentation](https://docs.docker.com/compose/). Don't remove the `identifier` and `db` container, they are respectively the entry-point and the database of your application. Don't forget to link the necessary microservices to the dispatcher and the database to the microservices. @@ -32,3 +35,29 @@ You can shut down using `docker-compose stop` and remove everything using `docke ## Tutorials To help you find your feet with your first semantic works projects, we've collected [a few tutorials](TUTORIALS.md). + +## Troubleshooting + +### Containers stuck while starting, using 100% CPU +Some docker images used in mu-project, notably those based on sbcl (lisp) and elixir images, are very slow and CPU intensive to start if the limits of open file descriptors are very high for the container. This leads to a process using 100% of a CPU for some time before that container becomes usable. This can be worked around by setting the defaults for new containers in the docker daemon config (/etc/docker/daemon.json (create it if it doesn't exist)): + +```json +{ + "default-ulimits": { + "nofile": { + "Hard": 104583, + "Name": "nofile", + "Soft": 104583 + } + } +} +``` + +Or, if you want these high defaults for some reason, you can set per-container limits in a docker-compose file for each of the mu-project services: + +```yml + ulimits: + nofile: + soft: 104583 + hard: 104583 +``` From 80555b6aa02f25195cb49f7187e7266f9b0edbc5 Mon Sep 17 00:00:00 2001 From: Rich Date: Thu, 14 Sep 2023 15:25:35 +0200 Subject: [PATCH 09/16] Move how-to on fixing container starting to project repo If you're reading this trying to figure out why this link doesn't work, maybe the [pull request](https://github.com/mu-semtech/project/pull/5) hasn't been merged yet. In which case, you want to look here: https://github.com/Denperidge-Redpencil/docs-update-project/blob/master/docs/how-tos/troubleshooting---slow-starting-containers.md --- README.md | 27 +-------------------------- 1 file changed, 1 insertion(+), 26 deletions(-) diff --git a/README.md b/README.md index cda8b00..6498bc5 100644 --- a/README.md +++ b/README.md @@ -13,7 +13,7 @@ Setting up your environment is done in three easy steps: 3. Lastly, simply start the docker-compose. > [!WARNING] -> Many of the containers used have issues with high limits on open file descriptors, so you might need [to work around this](#containers-stuck-while-starting-using-100-cpu) +> Many of the containers used have issues with high limits on open file descriptors, so you might need [to work around this](https://github.com/mu-semtech/project/blob/master/docs/how-tos/troubleshooting---slow-starting-containers.md) #### Hooking things up with docker-compose @@ -36,28 +36,3 @@ You can shut down using `docker-compose stop` and remove everything using `docke To help you find your feet with your first semantic works projects, we've collected [a few tutorials](TUTORIALS.md). -## Troubleshooting - -### Containers stuck while starting, using 100% CPU -Some docker images used in mu-project, notably those based on sbcl (lisp) and elixir images, are very slow and CPU intensive to start if the limits of open file descriptors are very high for the container. This leads to a process using 100% of a CPU for some time before that container becomes usable. This can be worked around by setting the defaults for new containers in the docker daemon config (/etc/docker/daemon.json (create it if it doesn't exist)): - -```json -{ - "default-ulimits": { - "nofile": { - "Hard": 104583, - "Name": "nofile", - "Soft": 104583 - } - } -} -``` - -Or, if you want these high defaults for some reason, you can set per-container limits in a docker-compose file for each of the mu-project services: - -```yml - ulimits: - nofile: - soft: 104583 - hard: 104583 -``` From c882ca40d1eabf9c52964491f766536fe66d1d48 Mon Sep 17 00:00:00 2001 From: Rich Date: Wed, 13 Sep 2023 16:48:15 +0200 Subject: [PATCH 10/16] Fix some typos, old names and add some precisions to tutorials --- README.md | 2 +- TUTORIALS.md | 27 ++++++++++++++------------- 2 files changed, 15 insertions(+), 14 deletions(-) diff --git a/README.md b/README.md index 6498bc5..3fe3125 100644 --- a/README.md +++ b/README.md @@ -17,7 +17,7 @@ Setting up your environment is done in three easy steps: #### Hooking things up with docker-compose -Alter the `docker-compose.yml` file so it contains all microservices you need. The example content should be clear, but you can find more information in the [Docker Compose documentation](https://docs.docker.com/compose/). Don't remove the `identifier` and `db` container, they are respectively the entry-point and the database of your application. Don't forget to link the necessary microservices to the dispatcher and the database to the microservices. +Alter the `docker-compose.yml` file so it contains all microservices you need. The example content should be clear, but you can find more information in the [Docker Compose documentation](https://docs.docker.com/compose/). Don't remove the `identifier` and `database` container, they are respectively the entry-point and the database of your application. Don't forget to link the necessary microservices to the dispatcher and the database to the microservices. #### Configure the dispatcher diff --git a/TUTORIALS.md b/TUTORIALS.md index b698118..0af261f 100644 --- a/TUTORIALS.md +++ b/TUTORIALS.md @@ -26,7 +26,7 @@ services: resource: image: semtech/mu-cl-resources:1.20.0 links: - - db:database + - database:database volumes: - ./config:/config # ... @@ -387,7 +387,7 @@ In app/controllers/books.js we alter: ### Adding authentication to your mu-project ![](http://mu.semte.ch/wp-content/uploads/2017/08/customumize_for_user-1024x768.png) -Web applications oftentimes require a user to be authenticated to access (part of) their application. For example a webshop may require a user to be logged in before placing an order. In a previous blog post we already explained [the semantic model to represent logged in users](https://mu.semte.ch/2017/08/24/representing-logged-in-users/). In this post we will show how to enable authentication in your app. We assume you already have  a [mu-project](https://github.com/mu-semtech/mu-project) running. +Web applications oftentimes require a user to be authenticated to access (part of) their application. For example a webshop may require a user to be logged in before placing an order. In a previous blog post we already explained [the semantic model to represent logged in users](https://mu.semte.ch/2017/08/24/representing-logged-in-users/). In this post we will show how to enable authentication in your app. We assume you already have  a [mu-project](https://github.com/mu-semtech/mu-project) running, with an ember front-end project. Adding authentication to your application consists of two tasks: @@ -555,7 +555,7 @@ And that's it! Now you know how your mu-project can be easily augmented with aut ### Building a mail handling service -My goal for this short coding session is to have a mail handling service that will allow me to list and maninpulate mails through a JSON:API REST back-end. And have that service pick up when I write a mail to the database and send it automatically. You can see the result of this project at https://github.com/langens-jonathan/ReactiveMailServiceExample. +My goal for this short coding session is to have a mail handling service that will allow me to list and manipulate mails through a JSON:API REST back-end. And have that service pick up when I write a mail to the database and send it automatically. You can see the result of this project at https://github.com/langens-jonathan/ReactiveMailServiceExample. #### Gain a head-start with mu-project For this project I started with cloning the mu-project repository: @@ -610,7 +610,8 @@ A GET call to http://localhost/mails produces: "data": [], "links": { "last": "/mails/", - "first": "/mails/" + "first": "/mails/", + "self": "mails" } } ``` @@ -680,7 +681,7 @@ Before we can start writing our reactive mail managing micro-service, we will ne delta: image: semtech/mu-delta-service:beta-0.7 links: - - db:db + - database:database volumes: - ./config/delta-service:/config environment: @@ -691,7 +692,7 @@ delta: This will add the monitoring service to our installation. The last thing to do for now is to change the link on the `resource` microservice by replacing ```yaml links: - - db:database + - database:database ``` with ```yaml @@ -703,8 +704,8 @@ The final steps are to create the configuration and subscribers files. Create a ```conf # made by Langens Jonathan -queryURL=http://db:8890/sparql -updateURL=http://db:8890/sparql +queryURL=http://database:8890/sparql +updateURL=http://database:8890/sparql sendUpdateInBody=true calculateEffectives=true ``` @@ -849,7 +850,7 @@ There are 2 types of delta reports, you have potential inserts and effective ins ``` *You can view the full version [here](https://gist.githubusercontent.com/langens-jonathan/cd5db8e9f68861662d888dad77f93662/raw/84adc69f9fd3143f45c05c0a5cefdf1ca9b95b55/gistfile1.txt).* -A report states the query that was send, an array of inserted objects and an array of deleted objects: Inserted or deleted objects represent a single triple with s, p and o being subject, predicate and object. +A report states the query that was sent, an array of inserted objects and an array of deleted objects: Inserted or deleted objects represent a single triple with s, p and o being subject, predicate and object. #### Expanding our mail handling microservice We need to notify the delta service of the existence of our mail handling service. We do this using the `subscribers.json` file that was created before. Change it so it looks like: @@ -870,7 +871,7 @@ In the `docker-compose.yml` file we need to alter the delta-service definition t delta: image: semtech/mu-delta-service:beta-0.8 links: - - db:db + - database:database - mailservice:mailservice volumes: - ./config/delta-service:/config @@ -898,7 +899,7 @@ Then we define a new method that will: - Load the delta report into a variable - Define some variables. -Lastly we define an array that will hold the URI’s of all emails that need to be send. +Lastly we define an array that will hold the URI’s of all emails that need to be sent. ```python @@ -912,7 +913,7 @@ def processDelta(): # continued later... ``` -We will loop over all inserted triples to check for mails that are ready to be send: +We will loop over all inserted triples to check for mails that are ready to be sent: ```python # mail-service/web.py def processDelta(): @@ -1176,7 +1177,7 @@ Add this snippet to your docker-compose.yml: classifier: image: flowofcontrol/mu-tf-image-classifier links: - - db:database + - database:database environment: CLASSIFIER_TRESHHOLD: 0.7 volumes: From 84e27ee54be295ea324d8d3f6dd3658e350c9431 Mon Sep 17 00:00:00 2001 From: Rich Date: Wed, 13 Sep 2023 17:22:10 +0200 Subject: [PATCH 11/16] Update usage of ember-mu-registration --- TUTORIALS.md | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/TUTORIALS.md b/TUTORIALS.md index 0af261f..737afeb 100644 --- a/TUTORIALS.md +++ b/TUTORIALS.md @@ -435,13 +435,14 @@ We now have an endpoint for registration in the backend. We need a complementary First, install the addon by executing the following command in your Ember project. ```bash ember install ember-mu-registration +ember install ember-resolver ``` Next, just include the `{{mu-register}}`, `{{mu-unregister}}` and `{{mu-change-password}}` component in your template. ```hbs {{!-- app/templates/registration.hbs --}} -{{mu-register}} + ``` The components will automatically send the correct requests to the backend. You can customize the component’s template and/or behavior as explained in the addon’s [README](https://github.com/mu-semtech/ember-mu-registration#advanced-usage). From aa2cdb2d8801d47e2afc27692a75b39530324654 Mon Sep 17 00:00:00 2001 From: Rich Date: Tue, 19 Sep 2023 14:40:53 +0200 Subject: [PATCH 12/16] Fix link to why-semantic-tech to (yet to be merged) project repo If the why-semantic-tech link does not work, please see https://github.com/Denperidge-Redpencil/du-project/blob/master/docs/discussions/why-semantic-tech.md --- TUTORIALS.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/TUTORIALS.md b/TUTORIALS.md index 737afeb..535aeeb 100644 --- a/TUTORIALS.md +++ b/TUTORIALS.md @@ -1,5 +1,5 @@ ## Tutorials -If you aren't familiar with the semantic.works stack/microservices yet, you might want to check out [why semantic tech?](https://mu.semte.ch/2017/03/23/adding-ember-fastboot-to-your-mu-project/) +If you aren't familiar with the semantic.works stack/microservices yet, you might want to check out [why semantic tech?](https://github.com/mu-semtech/project/blob/master/docs/discussions/why-semantic-tech.md) Each of these tutorials starts with a mu-project docker-compose set-up. From ea1c04072fc81cbec938402f21c97e637517c886 Mon Sep 17 00:00:00 2001 From: Rich Date: Mon, 18 Sep 2023 13:38:41 +0200 Subject: [PATCH 13/16] Update email tutorial to read a mailbox successfully --- TUTORIALS.md | 135 ++++++++++++++++++++++++++++++++------------------- 1 file changed, 84 insertions(+), 51 deletions(-) diff --git a/TUTORIALS.md b/TUTORIALS.md index 535aeeb..8500ebd 100644 --- a/TUTORIALS.md +++ b/TUTORIALS.md @@ -573,7 +573,8 @@ To add the necessary resource definitions, add them to the `domain.lisp` file as ```lisp (define-resource mail () :class (s-prefix "example:Mail") - :properties `((:sender :string ,(s-prefix "example:sender")) + :properties `((:from :string ,(s-prefix "example:from")) + (:to :string ,(s-prefix "example:to")) (:subject :string ,(s-prefix "example:subject")) (:content :string ,(s-prefix "example:content")) (:ready :string ,(s-prefix "example:ready"))) @@ -628,7 +629,8 @@ Body: { "data":{ "attributes":{ - "sender":"flowofcontrol@gmail.com", + "from":"flowofcontrol@gmail.com", + "to": "mail@example.com", "subject":"Mu Semtech Mail Server", "content":"This is a test for the Mu Semtech Mail Server.", "ready":"no" @@ -643,7 +645,8 @@ This gives us the following reponse: { "data": { "attributes": { - "sender": "flowofcontrol@gmail.com", + "from": "flowofcontrol@gmail.com", + "to": "mail@example.com", "subject": "Mu Semtech Mail Server", "content": "This is a test for the Mu Semtech Mail Server.", "ready": "no" @@ -662,7 +665,8 @@ To verify the original get request again, this now produces: { "data": { "attributes": { - "sender": "flowofcontrol@gmail.com", + "from": "flowofcontrol@gmail.com", + "to": "mail@example.com", "subject": "Mu Semtech Mail Server", "content": "This is a test for the Mu Semtech Mail Server.", "ready": "no" @@ -725,11 +729,11 @@ and then create `config/delta-service/subscribers.json` and put this JSON inside If we run `docker-compose rm` and then `docker-compose up` again, the delta service will be booting and already monitoring the changes that happen in the database! Of course we are not doing anything with them yet. So we will create a new micro-service just for this purpose. #### The mail-fetching microservice -The next step is to build our mail handling microservice. To do this we create a new directory called `mail-service` in our base directory. Then we create a file in that directory called `Dockerfile`. We will start from a mu.semte.ch template to make developing this microservice that much quicker. Mu.semte.ch has templates for a bunch of languages ruby, javascript, python, … For this microservice we will go for python 2.7. To do this we simply need to create a `web.py` file which will serve as the location for our code. Next add the following to the Dockerfile: +The next step is to build our mail handling microservice. To do this we create a new directory called `mail-service` in our base directory. Then we create a file in that directory called `Dockerfile`. We will start from a mu.semte.ch template to make developing this microservice that much quicker. Mu.semte.ch has templates for a bunch of languages ruby, javascript, python, … For this microservice we will go for python 3. To do this we simply need to create a dockerfile to build the container and a `web.py` file which will serve as the location for our code. First we create the file 'Dockerfile' in our mail-service directory: ```dockerfile # mail-service/Dockerfile -FROM semtech/mu-python-template +FROM semtech/mu-python-template:2.0.0-beta.2 MAINTAINER Langens Jonathan ``` @@ -739,74 +743,103 @@ I know it doesn’t say much, but it doesn’t need to. The python template will Then we need to add some mail manipulating functionality. Since this is not really the objective of this post I create a `mail_helpers.py` file and paste the following code in there: ```python # mail-service/mail_helpers.py -import sys -import imaplib -import getpass import email -import datetime import uuid import helpers - -def save_mail(sender, date, subject, content): - str_uuid = str(uuid.uuid4()) - insert_query = "INSERT DATA\n{\nGRAPH \n{\n a ;\n" - insert_query += " \"" + sender + "\";\n" - insert_query += " \"" + date + "\";\n" - insert_query += " \"" + content + "\";\n" - insert_query += " \"" + subject + "\";\n" - insert_query += " \"" + str_uuid + "\".\n" - insert_query += "}\n}" - print "query:\n", insert_query - helpers.update(insert_query) +from escape_helpers import sparql_escape_string + +def save_mail(sender, subject, content): + str_uuid = str(uuid.uuid4()) + insert_query = ( + f'INSERT DATA\n{{\nGRAPH \n{{' + f' a ;\n' + f' {sparql_escape_string(sender)};\n' + f' {sparql_escape_string(content)};\n' + f' {sparql_escape_string(subject)};\n' + f' "{str_uuid}".\n' + f'}}\n}}' + ) + helpers.log(f"query:\n{insert_query}") + helpers.update(insert_query) def process_mailbox(mailbox): - rv, data = mailbox.search(None, "ALL") - if rv != 'OK': - print "No messages found!" - return + rv, data = mailbox.search(None, "ALL") + if rv != 'OK' or not data[0]: + helpers.log("No messages found!") + return + else: + helpers.log("You've got mail!") + + for num in data[0].split(): + rv, data = mailbox.fetch(num, '(RFC822)') + if rv != 'OK': + helpers.log("ERROR getting message", num) + return - for num in data[0].split(): - rv, data = mailbox.fetch(num, '(RFC822)') - if rv != 'OK': - print "ERROR getting message", num - return - - msg = email.message_from_string(data[0][1]) - content = str(msg.get_payload()) - content = content.replace('\n','') + msg = email.message_from_string(data[0][1].decode()) + content = str(msg.get_payload()) - save_mail(msg['From'], msg['Date'], msg['Subject'], content) + save_mail(msg['From'], msg['Date'], msg['Subject'], content) ``` As you can see the mail_helpers contain 2 functions, one to iterate over all emails in a mailbox and the other to save a single email to the triple store. Easy peasy! -Next we create `web.py`. For more information on how the python template can be used you can visit: https://github.com/mu-semtech/mu-python-template. I created the following method to process all mails: +Next we create `web.py`. For more information on how the python template can be used you can visit: https://github.com/mu-semtech/mu-python-template. We create the following method to add a GET route to process all mails: ```python # mail-service/web.py +from os import environ +from imaplib import IMAP4, IMAP4_SSL + +import mail_helpers + +EMAIL_ADDRESS = environ['EMAIL_ADDRESS'] +EMAIL_PWD = environ['EMAIL_PWD'] + @app.route("/fetchMails") def fetchMailMethod(): - EMAIL_ADDRESS = "address" - EMAIL_PWD = "pwd" - - MAIL_SERVER = imaplib.IMAP4_SSL('imap.gmail.com') + MAIL_SERVER = IMAP4_SSL(environ['IMAP_SERVER']) - try: - MAIL_SERVER.login(EMAIL_ADDRESS, EMAIL_PWD) - except imaplib.IMAP4.error: - print "Logging into mailbox failed! " + try: + MAIL_SERVER.login(EMAIL_ADDRESS, EMAIL_PWD) + except IMAP4.error: + return "Unable to log in to IMAP server", 503 - rv, data = MAIL_SERVER.select("INBOX") - if rv == 'OK': - mail_helpers.process_mailbox(MAIL_SERVER) - MAIL_SERVER.close() + rv, data = MAIL_SERVER.select("INBOX") + if rv == 'OK': + mail_helpers.process_mailbox(MAIL_SERVER) + MAIL_SERVER.close() - MAIL_SERVER.logout() + MAIL_SERVER.logout() - return "ok" + return "ok" ``` This method is rather straightforward: it just opens a connection to an email address and opens the inbox mailbox. It then selects it for processing, thus inserting all mails into the triple store. +The last step to create this service is to add it to our docker-compose.yml file: + +```yaml + mailservice: + build: ./mail-service + links: + - database:database + ports: + # Forward a port for testing purposes + - "8888:80" + environment: + # Set the python template to development mode to enable auto reloading + MODE: "development" + LOG_LEVEL: "debug" + # This email should work but won't send any real mails, check ethereal.email for details + # You can replace this with a real email SMTP server but beware, this will send real mails! + EMAIL_ADDRESS: "kurtis.stamm@ethereal.email" + EMAIL_PWD: "zwaFZ3P5RDnDcWwRhA" + IMAP_SERVER: "imap.ethereal.email" + volumes: + # Bind our repo to the container, so we can edit files without having to rebuild the container + - ./mail-service:/app +``` + At this point, we have: - Defined a JSONAPI through which we can access our emails, using the standard mu.semte.ch stack - Built a custom service which fetches the emails from our mail account and inserts them into the triplestore using the right model From f0964cd32497e959663c63341d014984a8dbc15e Mon Sep 17 00:00:00 2001 From: Rich Date: Tue, 19 Sep 2023 10:27:06 +0200 Subject: [PATCH 14/16] Update email tutorial to send email correctly --- TUTORIALS.md | 91 +++++++++++++++++++++++++--------------------------- 1 file changed, 44 insertions(+), 47 deletions(-) diff --git a/TUTORIALS.md b/TUTORIALS.md index 8500ebd..9693a60 100644 --- a/TUTORIALS.md +++ b/TUTORIALS.md @@ -921,43 +921,35 @@ To handle delta reports in our mail handling microservice we will need 2 things: - Get access to the POST body of a request - Process and manipulate JSON data -To get access to this add the following imports to your `web.py` file: - -```python -import json -from flask import request -``` - -Then we define a new method that will: +To get access to this we edit `web.py` to define a new method that will: - Handle the incoming delta reports - Load the delta report into a variable - Define some variables. -Lastly we define an array that will hold the URI’s of all emails that need to be sent. +Lastly we define an array that will hold the URIs of all emails that need to be sent. ```python # mail-service/web.py -@app.route("/process_delta", methods=['POST']) -def processDelta(): - delta_report = json.loads(request.data) - mails_to_send = set() - predicate_mail_is_ready = "http://example.com/ready" - value_mail_is_ready = "yes" - # continued later... -``` +import json +from flask import request -We will loop over all inserted triples to check for mails that are ready to be sent: -```python -# mail-service/web.py +# ... + +@app.route("/process_delta", methods=['POST']) def processDelta(): - # ... - # ...continuation - for delta in delta_report['delta']: - for triple in delta['inserts']: - if(triple['p']['value'] == predicate_mail_is_ready): - if(triple['o']['value'] == value_mail_is_ready): - mails_to_send.add(triple['s']['value']) + delta_report = json.loads(request.data) + mails_to_send = set() + predicate_mail_is_ready = "http://example.com/ready" + value_mail_is_ready = "yes" + + # Loop over all inserted triples to check for mails that are ready to be sent: + + for delta in delta_report['delta']: + for triple in delta['inserts']: + if(triple['p']['value'] == predicate_mail_is_ready): + if(triple['o']['value'] == value_mail_is_ready): + mails_to_send.add(triple['s']['value']) # continued later... ``` @@ -968,15 +960,18 @@ Add the following code to `mail_helpers.py`: # mail-service/mail_helpers.py def load_mail(uri): # this query will find the mail (if it exists) - select_query = "SELECT DISTINCT ?uuid ?from ?ready ?subject ?content\n" - select_query += "WHERE \n{\n" - select_query += "<" + str(uri) + "> ?from;\n" - select_query += "a ;\n" - select_query += " ?content;\n" - select_query += " ?subject;\n" - select_query += " ?ready;\n" - select_query += " ?uuid.\n" - select_query += "}" + select_query = ( + f'SELECT DISTINCT ?uuid ?from ?to ?ready ?subject ?content\n' + f'WHERE {{\n' + f' <{str(uri)}> a ;\n' + f' ?from;\n' + f' ?to;\n' + f' ?content;\n' + f' ?subject;\n' + f' ?ready;\n' + f' ?uuid.\n' + f'}}' + ) # execute the query... result = helpers.query(select_query) @@ -993,7 +988,8 @@ def load_mail(uri): # we extract an object mail = dict() mail['uuid'] = bindings['uuid']['value'] - mail['sender'] = bindings['from']['value'] + mail['from'] = bindings['from']['value'] + mail['to'] = bindings['to']['value'] mail['ready'] = bindings['ready']['value'] mail['subject'] = bindings['subject']['value'] mail['content'] = bindings['content']['value'] @@ -1009,12 +1005,9 @@ To send the mail I have copied the entire `send_mail` function from http://naels ```python # mail-service/mail_helpers.py -def send_mail(mail): - - fromaddr = "YOUR EMAIL" - toaddr = "EMAIL ADDRESS YOU SEND TO" - +def send_mail(mail, from_addr, password): msg = MIMEMultipart() + helpers.log(f"sending... {mail}") msg['From'] = mail['from'] msg['To'] = mail['to'] @@ -1023,15 +1016,15 @@ def send_mail(mail): body = mail['content'] msg.attach(MIMEText(body, 'plain')) - server = smtplib.SMTP('smtp.gmail.com', 587) + server = SMTP(environ['SMTP_SERVER'], 587) server.starttls() - server.login(fromaddr, "YOUR PASSWORD") + server.login(from_addr, password) text = msg.as_string() - server.sendmail(fromaddr, toaddr, text) + server.sendmail(from_addr, mail['to'], text) server.quit() ``` -The last thing that we need to do is to connect the list of URI’s to the send_mail function: +The last thing that we need to do is to connect the list of URIs to the send_mail function: ```python # mail-service/web.py def processDelta(): @@ -1040,6 +1033,10 @@ def processDelta(): mail = mail_helpers.load_mail(uri) if 'uuid' in mail.keys(): mail_helpers.send_mail(mail, EMAIL_ADDRESS, EMAIL_PWD) + else: + helpers.log(f"Either no mail found or not ready: {mail}") + + return "ok" ``` To test this you can send a POST request similar to this one to your local mu.semte.ch application on http://localhost/mails: @@ -1058,7 +1055,7 @@ To test this you can send a POST request similar to this one to your local mu.se } ``` -If all went well then the person whose email address you filled in in the to field will have gotten a mail from you. Good job! You've just created a mailing microservice. +If all went well then the person whose email address you filled in in the to field will have gotten a mail from you (or there's a 'sent' mail in the [Ethereal Mail](https://ethereal.email/messages/) mailbox). Good job! You've just created a mailing microservice. *This tutorial has been adapted from Jonathan Langens' mu.semte.ch articles. You can view them [here](https://mu.semte.ch/2017/02/16/reactive-microservice-hands-on-tutorial-part-1/) and [here](https://mu.semte.ch/2017/03/16/reactive-microservice-hands-on-tutorial-part-2/).* From fb94640cfcd44283d411cf8baf03e1f6b2fbf595 Mon Sep 17 00:00:00 2001 From: Rich Date: Tue, 19 Sep 2023 12:10:37 +0200 Subject: [PATCH 15/16] Modify email tutorial to use mu-authorization and delta-notifier --- TUTORIALS.md | 178 ++++++++++++++++----------------------------------- 1 file changed, 56 insertions(+), 122 deletions(-) diff --git a/TUTORIALS.md b/TUTORIALS.md index 9693a60..5a04e9a 100644 --- a/TUTORIALS.md +++ b/TUTORIALS.md @@ -590,7 +590,9 @@ Then add the prefix to the `repository.lisp` file: (add-prefix "example" "http://example.com/") ``` -We are almost there for a first test! The only thing left to do is to add the `/mails` route to the dispatcher (for more info check the documentation on http://mu.semte.ch). To do this add the following block of code to the `dispatcher.ex` file: +We're almost there for a first test! Two small pieces of wiring left to do: + +First add the `/mails` route to the dispatcher (for more info check [the documentation](https://github.com/mu-semtech/mu-dispatcher/)). To do this add the following block of code to the `dispatcher.ex` file: ``` match "/mails/*path" do @@ -598,6 +600,19 @@ match "/mails/*path" do end ``` +Second, configure `mu-authorization` to allow anyone to edit Mail resources. We won't add proper access rules as that's a topic for a different tutorial and more info can be found [in the repo for the project](https://github.com/mu-semtech/mu-authorization). We edit the file `config/authorization/config.ex` to add to the public `GroupSpec`: + +```diff + graphs: [ %GraphSpec{ + graph: "http://mu.semte.ch/graphs/public", + constraint: %ResourceConstraint{ + resource_types: [ +- "http://xmlns.com/foaf/0.1/Person" ++ "http://xmlns.com/foaf/0.1/Person", ++ "http://example.com/Mail" + ], +``` + Now fire this up and lets see what we have by running the following command in the project root directory: ```bash @@ -678,56 +693,6 @@ To verify the original get request again, this now produces: } ``` -#### Enabling the reactive database -Before we can start writing our reactive mail managing micro-service, we will need to add a monitoring service to monitor the DB. This will be a lot easier than it sounds with mu.semte.ch. To start, open the `docker-compose.yml` file and add the following lines at the bottom of the file: - -```yaml -# ... -delta: - image: semtech/mu-delta-service:beta-0.7 - links: - - database:database - volumes: - - ./config/delta-service:/config - environment: - CONFIGFILE: "/config/config.properties" - SUBSCRIBERSFILE: "/config/subscribers.json" -``` - -This will add the monitoring service to our installation. The last thing to do for now is to change the link on the `resource` microservice by replacing -```yaml -links: - - database:database -``` -with -```yaml -links: - - delta:database -``` - -The final steps are to create the configuration and subscribers files. Create a file called `config.properties` at the location `config/delta-service/config.properties` and write the following lines in that file: - -```conf -# made by Langens Jonathan -queryURL=http://database:8890/sparql -updateURL=http://database:8890/sparql -sendUpdateInBody=true -calculateEffectives=true -``` - -and then create `config/delta-service/subscribers.json` and put this JSON inside: - -```json -{ - "potentials":[ - ], - "effectives":[ - ] -} -``` - -If we run `docker-compose rm` and then `docker-compose up` again, the delta service will be booting and already monitoring the changes that happen in the database! Of course we are not doing anything with them yet. So we will create a new micro-service just for this purpose. - #### The mail-fetching microservice The next step is to build our mail handling microservice. To do this we create a new directory called `mail-service` in our base directory. Then we create a file in that directory called `Dockerfile`. We will start from a mu.semte.ch template to make developing this microservice that much quicker. Mu.semte.ch has templates for a bunch of languages ruby, javascript, python, … For this microservice we will go for python 3. To do this we simply need to create a dockerfile to build the container and a `web.py` file which will serve as the location for our code. First we create the file 'Dockerfile' in our mail-service directory: @@ -840,81 +805,54 @@ The last step to create this service is to add it to our docker-compose.yml file - ./mail-service:/app ``` +Running `docker-compose up` again will start our new service, which should auto reload with changes to the python code, though changing the configuration of our other services will require to restart them for the changes to be picked up. + At this point, we have: - Defined a JSONAPI through which we can access our emails, using the standard mu.semte.ch stack - Built a custom service which fetches the emails from our mail account and inserts them into the triplestore using the right model -Now we will use these services in combination with the delta service, to discover which emails were inserted into the database, and to perform reactive computations on it. - -#### The delta service - -The delta service’s responsibilities are: +Now we will use these services in combination with the delta notifier, to discover which emails were inserted into the database, and to perform reactive computations on it. -- Acting as the SPARQL endpoint for the microservices -- Calculating the differences (deltas) that a query will introduce in the database -- Notifying interested parties of these differences +#### Mu-authorization and the delta-notifier -For this hands on we use version beta-0.8 of the delta service. +You may have noticed that the default mu-project docker-compose.yml contains a service called `database` but the image is `semtech/mu-authorization`, this has two responsibilities: -##### What do these delta reports look like? -There are 2 types of delta reports, you have potential inserts and effective inserts. A report for either will look like: -```json -{ - "delta": [ - { - "type": "effective", - "graph": "http://mu.semte.ch/application", - "inserts": [ - { - "s": { - "value": "http://example.com/mails/58B187FA6AA88E0009000001", - "type": "uri" - }, - "p": { - "value": "http://example.com/subject", - "type": "uri" - }, - "o": { - "value": "Mu Semtech Mail Server", - "type": "literal" - } - }, - ... -} -``` -*You can view the full version [here](https://gist.githubusercontent.com/langens-jonathan/cd5db8e9f68861662d888dad77f93662/raw/84adc69f9fd3143f45c05c0a5cefdf1ca9b95b55/gistfile1.txt).* +- Act as the SPARQL endpoint, handling authorization logic before forwarding approved queries or updates to the triplestore +- Producing 'delta's describing the changes and forwarding them to clients defined in `config/authorization/delta.ex` -A report states the query that was sent, an array of inserted objects and an array of deleted objects: Inserted or deleted objects represent a single triple with s, p and o being subject, predicate and object. +This is already configured for us to send these deltas to another service, the delta-notifier. This service is configured in `config/delta/rules.js`, which defines which triples different microservices are interested in. It sends matching deltas to these services using REST. The format for these messages is detailed [in the delta-notifier repository](https://github.com/mu-semtech/delta-notifier#delta-formats). #### Expanding our mail handling microservice -We need to notify the delta service of the existence of our mail handling service. We do this using the `subscribers.json` file that was created before. Change it so it looks like: -```json -{ - "potentials":[ - ], - "effectives":[ - "http://mailservice/process_delta" - ] -} -``` +We need to notify the delta-notifier of the existence of our mail handling service. To do this we replace the `config/delta/rules.js` file to send any deltas for subjects of `rdf:type` `example:Mail` to the mail service: -In the `docker-compose.yml` file we need to alter the delta-service definition to look like: - -```yaml - delta: - image: semtech/mu-delta-service:beta-0.8 - links: - - database:database - - mailservice:mailservice - volumes: - - ./config/delta-service:/config - environment: - CONFIGFILE: "/config/config.properties" - SUBSCRIBERSFILE: "/config/subscribers.json" +```js +export default [ + { + match: { + predicate: { + type: "uri", + value: "http://www.w3.org/1999/02/22-rdf-syntax-ns#type", + }, + object: { + type: "uri", + value: "http://example.com/Mail", + }, + }, + callback: { + url: "http://mailservice/process_delta", + method: "POST", + }, + options: { + resourceFormat: "v0.0.1", + gracePeriod: 250, + ignoreFromSelf: true, + }, + }, +]; ``` -That way the delta service can talk to the mailservice. +Don't forget to restart your delta notifier with `docker compose restart delta-notifier`. To handle delta reports in our mail handling microservice we will need 2 things: @@ -926,9 +864,6 @@ To get access to this we edit `web.py` to define a new method that will: - Load the delta report into a variable - Define some variables. -Lastly we define an array that will hold the URIs of all emails that need to be sent. - - ```python # mail-service/web.py import json @@ -944,16 +879,15 @@ def processDelta(): value_mail_is_ready = "yes" # Loop over all inserted triples to check for mails that are ready to be sent: - - for delta in delta_report['delta']: - for triple in delta['inserts']: - if(triple['p']['value'] == predicate_mail_is_ready): - if(triple['o']['value'] == value_mail_is_ready): - mails_to_send.add(triple['s']['value']) + for delta in delta_report: + for insert in delta['inserts']: + if (insert['predicate']['value'] == predicate_mail_is_ready + and insert['object']['value'] == value_mail_is_ready): + mails_to_send.add(insert['subject']['value']) # continued later... ``` -After this for loop has run, all the URI’s of mails that are ready to be send will be in the `mails_to_send` array. Now we loop over the array and query the database for each URI in the set. And then we will fetch a mail object for every URI that is in the set. +After this for loop has run, all the URIs of mails that are ready to be send will be in the `mails_to_send` array. Now we loop over the array and query the database for each URI in the set. And then we will fetch a mail object for every URI that is in the set. Add the following code to `mail_helpers.py`: ```python From cab9d38a1e43e16fa6e6c51d2624f3d37c00c667 Mon Sep 17 00:00:00 2001 From: Rich Date: Fri, 22 Sep 2023 16:14:46 +0200 Subject: [PATCH 16/16] Modify email tutorial to use delta-notifier While doing this took the time to make it a little more coherent as a project and to suggest ways to test as you go. --- TUTORIALS.md | 320 ++++++++++++++++++++++++++++++--------------------- 1 file changed, 187 insertions(+), 133 deletions(-) diff --git a/TUTORIALS.md b/TUTORIALS.md index 5a04e9a..f3b1a8d 100644 --- a/TUTORIALS.md +++ b/TUTORIALS.md @@ -6,7 +6,7 @@ Each of these tutorials starts with a mu-project docker-compose set-up. - [Creating a JSON API](#creating-a-json-api) - [Adding an ember UI to your project](#adding-an-ember-ui-to-your-project) - [Adding authentication to your mu-project](#adding-authentication-to-your-mu-project) -- [Creating a mail service](#building-a-mail-handling-service) +- [Reacting to data changes: Creating a mail management service](#reacting-to-data-changes-building-a-mail-management-service) - [Adding Ember Fastboot to your project](#adding-ember-fastboot-to-your-project) - [Adding a machine learning microservice to your mu.semte.ch project](#adding-a-machine-learning-microservice-to-your-musemtech-project) @@ -555,18 +555,21 @@ And that's it! Now you know how your mu-project can be easily augmented with aut -### Building a mail handling service -My goal for this short coding session is to have a mail handling service that will allow me to list and manipulate mails through a JSON:API REST back-end. And have that service pick up when I write a mail to the database and send it automatically. You can see the result of this project at https://github.com/langens-jonathan/ReactiveMailServiceExample. +### Reacting to data changes: Building a mail management service + +The aim of this tutorial is to demonstrate how to react to data changes by creating a mail handling service that will allow us to read and send mails through a JSON:API REST back-end. To do this we'll read emails from an inbox, store them as semantically linked data, and importantly, pick up any mails that are created and send them. Since we're looking at reacting to data, we won't build a front-end, but that could easily be done. #### Gain a head-start with mu-project -For this project I started with cloning the mu-project repository: + +To start clone and rename the mu-project repository: + ```bash -git clone https://github.com/mu-semtech/mu-project +git clone https://github.com/mu-semtech/mu-project mail-box ``` -This will give me the CRUD endpoint I need to manipulate my mail related resources. After cloning I rename the repository to MailBox and set the remote origin to a new one. For now I will leave the `README.md` file as it is. +For now, since we're not publishing this, leave the `README.md` file and the remote the repo is using as they are. -For the first block we will modify the `config/resources/domain.lisp`, `config/resourecs/repository.lisp` and the `config/dispatcher/dispatcher.ex` files. +To get us started with a CRUD JSON:API for mail objects, modify the `config/resources/domain.lisp`, `config/resourecs/repository.lisp` and the `config/dispatcher/dispatcher.ex` files. To add the necessary resource definitions, add them to the `domain.lisp` file as follows: @@ -577,12 +580,12 @@ To add the necessary resource definitions, add them to the `domain.lisp` file as (:to :string ,(s-prefix "example:to")) (:subject :string ,(s-prefix "example:subject")) (:content :string ,(s-prefix "example:content")) - (:ready :string ,(s-prefix "example:ready"))) + (:ready :boolean ,(s-prefix "example:ready"))) :resource-base (s-url "http://example.com/mails/") :on-path "mails") ``` -This will create a resource description that we can manipulate on route `/mails` with the properties sender, title, body and ready. +This will create a resource description that we can manipulate on route `/mails` with the self-explanatory properties from, to, title and content as well as a boolean flag to show if a new mail is ready to send. Then add the prefix to the `repository.lisp` file: @@ -642,15 +645,15 @@ URL: http://localhost/mails Headers: {"Content-Type":"application/vnd.api+json"} Body: { - "data":{ - "attributes":{ - "from":"flowofcontrol@gmail.com", + "data": { + "attributes": { + "from": "flowofcontrol@gmail.com", "to": "mail@example.com", - "subject":"Mu Semtech Mail Server", - "content":"This is a test for the Mu Semtech Mail Server.", - "ready":"no" + "subject": "Mu Semtech Mail Server", + "content": "This is a test for the Mu Semtech Mail Server.", + "ready": false }, - "type":"mails" + "type": "mails" } } ``` @@ -664,7 +667,7 @@ This gives us the following reponse: "to": "mail@example.com", "subject": "Mu Semtech Mail Server", "content": "This is a test for the Mu Semtech Mail Server.", - "ready": "no" + "ready": false }, "id": "58978C2A6460170009000001", "type": "mails", @@ -675,26 +678,31 @@ This gives us the following reponse: That worked! In about 30 minutes we have a fully functional REST API endpoint for managing mail resources! -To verify the original get request again, this now produces: +To verify the original GET request again, this now produces: ```json { - "data": { - "attributes": { - "from": "flowofcontrol@gmail.com", - "to": "mail@example.com", - "subject": "Mu Semtech Mail Server", - "content": "This is a test for the Mu Semtech Mail Server.", - "ready": "no" - }, - "id": "58978C3A6460170009000002", - "type": "mails", - "relationships": {} - } + "data": [ + { + "attributes": { + "from": "flowofcontrol@gmail.com", + "to": "mail@example.com", + "subject": "Mu Semtech Mail Server", + "content": "This is a test for the Mu Semtech Mail Server.", + "ready": false + }, + "id": "58978C3A6460170009000002", + "type": "mails", + "relationships": {} + } + ], + "links": { + ... + } } ``` #### The mail-fetching microservice -The next step is to build our mail handling microservice. To do this we create a new directory called `mail-service` in our base directory. Then we create a file in that directory called `Dockerfile`. We will start from a mu.semte.ch template to make developing this microservice that much quicker. Mu.semte.ch has templates for a bunch of languages ruby, javascript, python, … For this microservice we will go for python 3. To do this we simply need to create a dockerfile to build the container and a `web.py` file which will serve as the location for our code. First we create the file 'Dockerfile' in our mail-service directory: +The next step is to build our mail handling microservice. To do this we create a new directory called `mail-service` in our base directory. We will start from a mu.semte.ch template to make developing this microservice that much quicker. Mu.semte.ch has templates for a bunch of languages ruby, javascript, python, … For this microservice we will go for python 3 (using Flask). To do this we simply need to create a Dockerfile to build the container and a `web.py` file which will serve as the location for our code. First we create the file 'Dockerfile' in our mail-service directory: ```dockerfile # mail-service/Dockerfile @@ -705,81 +713,124 @@ MAINTAINER Langens Jonathan I know it doesn’t say much, but it doesn’t need to. The python template will handle the rest. -Then we need to add some mail manipulating functionality. Since this is not really the objective of this post I create a `mail_helpers.py` file and paste the following code in there: +Then we need to add some mail manipulating functionality. Since manipulating email APIs is not really the objective of this post we create a `mail_helpers.py` file and paste the following code in there: + ```python # mail-service/mail_helpers.py import email import uuid import helpers from escape_helpers import sparql_escape_string +from imaplib import IMAP4, IMAP4_SSL +from os import environ -def save_mail(sender, subject, content): - str_uuid = str(uuid.uuid4()) - insert_query = ( - f'INSERT DATA\n{{\nGRAPH \n{{' - f' a ;\n' - f' {sparql_escape_string(sender)};\n' - f' {sparql_escape_string(content)};\n' - f' {sparql_escape_string(subject)};\n' - f' "{str_uuid}".\n' - f'}}\n}}' +SMTP_SERVER = environ['SMTP_SERVER'] +IMAP_SERVER = environ['IMAP_SERVER'] +MAILBOX_ADDRESS = environ['MAILBOX_ADDRESS'] +MAILBOX_PWD = environ['MAILBOX_PWD'] + +def sanitise_mailbox_id(id): + return id.replace('<', '').replace('>', '') + +def save_mails(mails): + for mail in mails: + content = str(mail.get_payload()) + if mail.is_multipart(): + # This doesn't handle multipart messages well, but that's outside this example's scope + content = "" + for part in mail.get_payload(): + content += str(part) + + str_uuid = str(uuid.uuid4()) + uri = f'' + escaped_id = sparql_escape_string(sanitise_mailbox_id(mail['Message-ID'])) + insert_query = ( + f'INSERT DATA\n{{\nGRAPH \n{{' + f' {uri} a ;\n' + f' {sparql_escape_string(mail["From"])};\n' + f' {sparql_escape_string(mail["To"])};\n' + f' {sparql_escape_string(content)};\n' + f' {sparql_escape_string(mail["Subject"])};\n' + f' {escaped_id};\n' + f' "{str_uuid}".\n' + f'}}\n}}' + ) + helpers.log(f"query:\n{insert_query}") + helpers.update(insert_query) + +def get_stored_mailbox_ids(): + query = ( + f'SELECT ?mailboxid\n' + f'WHERE {{\n' + f' ?mail a ;\n' + f' ?mailboxid .\n' + f'}}' ) - helpers.log(f"query:\n{insert_query}") - helpers.update(insert_query) + result = helpers.query(query) + bindings = result['results']['bindings'] + ids = set() + for bound in bindings: + ids.add(bound['mailboxid']['value']) + return ids + +def filter_saved_mails(mails): + existing_ids = get_stored_mailbox_ids() + unsaved = [] + for mail in mails: + id = sanitise_mailbox_id(mail['Message-ID']) + if id not in existing_ids: + unsaved.append(mail) + return unsaved + +def fetch_all_mails(): + mailbox = IMAP4_SSL(IMAP_SERVER) -def process_mailbox(mailbox): - rv, data = mailbox.search(None, "ALL") - if rv != 'OK' or not data[0]: - helpers.log("No messages found!") - return - else: - helpers.log("You've got mail!") + try: + mailbox.login(MAILBOX_ADDRESS, MAILBOX_PWD) + except IMAP4.error: + raise Exception("Unable to log in to IMAP server") + + rv, _ = mailbox.select("INBOX") + mails = [] + if rv == 'OK': + rv, search_data = mailbox.search(None, "ALL") + if rv != 'OK' or not search_data[0]: + helpers.log("No messages found!") + return mails + else: + helpers.log(f"You've got mail! {search_data[0]}") - for num in data[0].split(): - rv, data = mailbox.fetch(num, '(RFC822)') - if rv != 'OK': - helpers.log("ERROR getting message", num) - return + for num in search_data[0].split(): + rv, data = mailbox.fetch(num, '(RFC822)') + helpers.log(f'is this data {data}') + if rv != 'OK': + helpers.log("ERROR getting message", num) + return mails + mails.append(email.message_from_string(data[0][1].decode())) - msg = email.message_from_string(data[0][1].decode()) - content = str(msg.get_payload()) + mailbox.close() + mailbox.logout() - save_mail(msg['From'], msg['Date'], msg['Subject'], content) + return mails ``` -As you can see the mail_helpers contain 2 functions, one to iterate over all emails in a mailbox and the other to save a single email to the triple store. Easy peasy! +As you can see the mail_helpers contains 5 functions, together these fetch all the mails from an IMAP mailbox, filter them by those which have already been saved (by the id used on the server) and save those in the triplestore. Next we create `web.py`. For more information on how the python template can be used you can visit: https://github.com/mu-semtech/mu-python-template. We create the following method to add a GET route to process all mails: + ```python # mail-service/web.py -from os import environ -from imaplib import IMAP4, IMAP4_SSL - import mail_helpers -EMAIL_ADDRESS = environ['EMAIL_ADDRESS'] -EMAIL_PWD = environ['EMAIL_PWD'] - @app.route("/fetchMails") def fetchMailMethod(): - MAIL_SERVER = IMAP4_SSL(environ['IMAP_SERVER']) - - try: - MAIL_SERVER.login(EMAIL_ADDRESS, EMAIL_PWD) - except IMAP4.error: - return "Unable to log in to IMAP server", 503 - - rv, data = MAIL_SERVER.select("INBOX") - if rv == 'OK': - mail_helpers.process_mailbox(MAIL_SERVER) - MAIL_SERVER.close() - - MAIL_SERVER.logout() - + all_mails = mail_helpers.fetch_all_mails() + unsaved = mail_helpers.filter_saved_mails(all_mails) + mail_helpers.save_mails(unsaved) return "ok" ``` -This method is rather straightforward: it just opens a connection to an email address and opens the inbox mailbox. It then selects it for processing, thus inserting all mails into the triple store. +This method is rather straightforward as it just composes our `mail_helper` functions to fetch, filter then save the mails. The last step to create this service is to add it to our docker-compose.yml file: @@ -795,8 +846,9 @@ The last step to create this service is to add it to our docker-compose.yml file # Set the python template to development mode to enable auto reloading MODE: "development" LOG_LEVEL: "debug" - # This email should work but won't send any real mails, check ethereal.email for details - # You can replace this with a real email SMTP server but beware, this will send real mails! + # To make this example easy to jump into, we use a mailbox from ethereal.email. + # This should just work, but you may need to set up a new one. You can also use a real + # email but our code will be able to send and read any emails in the account. EMAIL_ADDRESS: "kurtis.stamm@ethereal.email" EMAIL_PWD: "zwaFZ3P5RDnDcWwRhA" IMAP_SERVER: "imap.ethereal.email" @@ -811,6 +863,8 @@ At this point, we have: - Defined a JSONAPI through which we can access our emails, using the standard mu.semte.ch stack - Built a custom service which fetches the emails from our mail account and inserts them into the triplestore using the right model +You can test this set-up by sending mails to the email address your code is looking at, then sending a GET request to our service at `localhost:8888/fetchMails`. You should see logging of the mails found and you can look at the data in the triplestore at `localhost:8890/sparql`. + Now we will use these services in combination with the delta notifier, to discover which emails were inserted into the database, and to perform reactive computations on it. #### Mu-authorization and the delta-notifier @@ -824,7 +878,7 @@ This is already configured for us to send these deltas to another service, the d #### Expanding our mail handling microservice -We need to notify the delta-notifier of the existence of our mail handling service. To do this we replace the `config/delta/rules.js` file to send any deltas for subjects of `rdf:type` `example:Mail` to the mail service: +We need to notify the delta-notifier of the existence of our mail handling service. To do this we replace the `config/delta/rules.js` file to send any deltas for subjects of `rdf:type` `example:Mail` to the mail service. Importantly, with `ignoreFromSelf`, we ask to not receive deltas from our own changes, which means the mail-service will only be notified of changes made by other services, such as `mu-cl-resources`, so we don't try to send emails that have been pulled from the mailbox: ```js export default [ @@ -854,44 +908,25 @@ export default [ Don't forget to restart your delta notifier with `docker compose restart delta-notifier`. -To handle delta reports in our mail handling microservice we will need 2 things: +To handle delta reports in our mail handling microservice we edit `web.py` to define a new method that will: -- Get access to the POST body of a request -- Process and manipulate JSON data +- Examine the POST body of the request +- Collect the URIs for mails that need to be sent +- Load those mails from the triplestore +- Send them -To get access to this we edit `web.py` to define a new method that will: -- Handle the incoming delta reports -- Load the delta report into a variable -- Define some variables. +For this we'll need some more helper functions, so we add the following code to `mail_helpers.py`: ```python -# mail-service/web.py -import json -from flask import request - +# mail-service/mail_helpers.py # ... +from email.mime.text import MIMEText +from email.mime.multipart import MIMEMultipart +from smtplib import SMTP -@app.route("/process_delta", methods=['POST']) -def processDelta(): - delta_report = json.loads(request.data) - mails_to_send = set() - predicate_mail_is_ready = "http://example.com/ready" - value_mail_is_ready = "yes" - - # Loop over all inserted triples to check for mails that are ready to be sent: - for delta in delta_report: - for insert in delta['inserts']: - if (insert['predicate']['value'] == predicate_mail_is_ready - and insert['object']['value'] == value_mail_is_ready): - mails_to_send.add(insert['subject']['value']) - # continued later... -``` - -After this for loop has run, all the URIs of mails that are ready to be send will be in the `mails_to_send` array. Now we loop over the array and query the database for each URI in the set. And then we will fetch a mail object for every URI that is in the set. +SMTP_SERVER = environ['SMTP_SERVER'] +# ... -Add the following code to `mail_helpers.py`: -```python -# mail-service/mail_helpers.py def load_mail(uri): # this query will find the mail (if it exists) select_query = ( @@ -929,17 +964,8 @@ def load_mail(uri): mail['content'] = bindings['content']['value'] return mail -``` - -This function will load the mail object from the triple store. There is still the chance that the ready predicate was sent for some other object, for a mail that does not have all required fields, or for an object that is not a mail but happens to use the same predicate. -We will use this function to try to load a mail object for each URI. Because the query was built without OPTIONAL statements, we are certain that an the dictionary returned by the load_mail function will either have all keys or none. - -To send the mail I have copied the entire `send_mail` function from http://naelshiab.com/tutorial-send-email-python/ and modified it slightly to take into account the dictionary object that now describes the mail. - -```python -# mail-service/mail_helpers.py -def send_mail(mail, from_addr, password): +def send_mail(mail): msg = MIMEMultipart() helpers.log(f"sending... {mail}") @@ -950,29 +976,55 @@ def send_mail(mail, from_addr, password): body = mail['content'] msg.attach(MIMEText(body, 'plain')) - server = SMTP(environ['SMTP_SERVER'], 587) + server = SMTP(SMTP_SERVER, 587) server.starttls() - server.login(from_addr, password) - text = msg.as_string() - server.sendmail(from_addr, mail['to'], text) + server.login(MAILBOX_ADDRESS, MAILBOX_PWD) + server.sendmail(mail['from'], mail['to'], msg.as_string()) server.quit() ``` -The last thing that we need to do is to connect the list of URIs to the send_mail function: +The first function will load the mail object from the triple store. There is still the chance that the ready predicate was sent for some other object, for a mail that does not have all required fields, or for an object that is not a mail but happens to use the same predicate. Because the query was built without OPTIONAL statements, we are certain that an the dictionary returned by the load_mail function will either have all keys or none. + +To send the mail I have copied the entire `send_mail` function from http://naelshiab.com/tutorial-send-email-python/ and modified it slightly to take into account the dictionary object that now describes the mail. + +Now we can actually add our endpoint: + ```python # mail-service/web.py +import json +from flask import request +import helpers + +# ... + +@app.route("/process_delta", methods=['POST']) def processDelta(): - # ...continuation + delta_report = json.loads(request.data) + mails_to_send = set() + predicate_mail_ready = "http://example.com/ready" + value_mail_is_ready = "true" + + helpers.log(f"got delta {delta_report}") + # Loop over all inserted triples to check for mails that are ready to be sent: + for delta in delta_report: + for insert in delta['inserts']: + helpers.log(f"examining {insert}") + if (insert['predicate']['value'] == predicate_mail_ready + and insert['object']['value'] == value_mail_is_ready): + mails_to_send.add(insert['subject']['value']) + for uri in mails_to_send: mail = mail_helpers.load_mail(uri) if 'uuid' in mail.keys(): - mail_helpers.send_mail(mail, EMAIL_ADDRESS, EMAIL_PWD) + mail_helpers.send_mail(mail) else: helpers.log(f"Either no mail found or not ready: {mail}") return "ok" ``` +This first goes through all the newly inserted triples to find any that corespond to a ready to send mail. Then we loop over these and query the database for each URI and then send the mail. + To test this you can send a POST request similar to this one to your local mu.semte.ch application on http://localhost/mails: ```json @@ -989,7 +1041,9 @@ To test this you can send a POST request similar to this one to your local mu.se } ``` -If all went well then the person whose email address you filled in in the to field will have gotten a mail from you (or there's a 'sent' mail in the [Ethereal Mail](https://ethereal.email/messages/) mailbox). Good job! You've just created a mailing microservice. +If all went well then the person whose email address you filled in in the to field will have gotten a mail from you (or there's a 'sent' mail in the [Ethereal Mail](https://ethereal.email/messages/) mailbox). + +Good job! You've just created a microservice to manage an email inbox through REST. *This tutorial has been adapted from Jonathan Langens' mu.semte.ch articles. You can view them [here](https://mu.semte.ch/2017/02/16/reactive-microservice-hands-on-tutorial-part-1/) and [here](https://mu.semte.ch/2017/03/16/reactive-microservice-hands-on-tutorial-part-2/).*