The goal of this API is to make the scraped articles accessible to our frontend as well as external requests. It is written with the help of Akka HTTP, which provides a toolkit to create simple HTTP-based services.
In addition we use elastic4s, a Scala client for Elasticsearch. The client is type safe, enabling us to write Elastic-queries programmatically and be notified about errors at compile time.
Configurations for the Webserver and the Elasticsearch connection are stored in src/main/resources/application.conf. Information on how the app should be run on production server are located in inewshttpapi.service.
The data classes are stored in src/main/scala/de/htwBerlin/ai/inews/core and are used to store Elasticsearch query results and define GET responses. To be able to serialize our objects into JSON we use a custom JsonFormat which utilizes functions of spray-json.
The routes are implemented in src/main/scala/de/htwBerlin/ai/inews/http/routes.
The article routes provides all functionality that is needed to display the scraped data in the frontend.
| Method | Route | Parameters | Description |
|---|---|---|---|
GET |
/api/articles |
offset : Int count : Int query : String department : List[String] newspaper : List[String] author : String |
Get, search and filter multiple articles |
GET |
/api/articles/{articleId} |
- | Get one single article by id |
GET |
/api/articles/authors |
query : String |
Get and filter authors |
GET |
/api/articles/departments |
- | Get all departments |
GET |
/api/articles/newspapers |
- | Get all newspapers |
The analytics routes aggregate additional information on the articles, needed to provide analysis functions.
| Method | Route | Parameters | Description |
|---|---|---|---|
GET |
/api/analytics/lemmas |
- | Get the 10 most relevant lemmas for the past 7 days |
GET |
/api/analytics/terms |
query : String timeFrom : Long timeTo : Long |
Get the occurrences for a term in a specified time frame |
All queries and logic needed to return Elasticsearch data is implemented in src/main/scala/de/htwBerlin/ai/inews/data/ArticleService.scala, utilizing elastic4s. Articles are parsed into Scala classes with the help of the ArticleHitReader.
To run the API on your local machine you first have to connect to the Elasticsearch interface on the INews-Server via port forwarding. Therefore run ssh -L 9200:localhost:9200 [email protected]. After that the API should be able to access Elasticsearch.
To simply create an executable file bundled with all dependencies we use sbt-assembly plugin, which is added in project/plugins.sbt.
To build a .jar file, run sbt clean assembly in the project directory. The generated file will be located in target/scala<SCALA_VERSION>/inews-backend-assembly-<APP_VERSION>.jar.
To deploy the app, the .jar file as well as the inewshttpapi.service have to be copied to the INews-Server.
Commands to update the .service file:
- Copy file to INews-server:
scp inewshttpapi.service [email protected]:/home/local/inewshttpapi.service - Login at INews-server:
ssh [email protected] - Enable sudo access:
su - Move file to correct location:
mv inewshttpapi.service /lib/systemd/system - Restart service:
systemctl restart inewshttpapi.service
Commands to update the .jar file:
- Copy file to INews-server:
scp target/scala-2.13/inews-http-api-assembly-0.1.jar [email protected]:/home/local/inews-http-api-assembly-0.1.jar - Login at INews-server:
ssh [email protected] - Enable sudo access:
su - Move file to correct location:
mv inews-http-api-assembly-0.1.jar /home/httpapi/ - Restart service:
systemctl restart inewshttpapi.service
At the time of writing, the API is available within the HTW network at http://news.f4.htw-berlin.de:8081.