The project name is a misnomer, there is no Apache Wicket in it (yet?).
I'm trying to learn how to develop a web application in Java using some standard modern framework, and also using new Java APIs.
Originally, I wanted to create a web app using Apache Wicket. Then I discovered that Wicket only does the views but I needed controllers and a persistence layer, too. I spent a couple of hours on this phase.
So I turned to Spring Boot, which gave me something using JPA and Hibernate and Spring MVC and Thymeleaf. I liked Spring Boot because it told me that it's an opinionated collection of various frameworks. This is good because I don't know the Spring ecosystem (yet) and therefore I'm happy to go with whatever someone suggests. Once I have a bit of experience, I can swap different parts out.
After trying Spring Boot, I hit another brick wall: Spring Boot suggests to use the H2 database as an in-memory database. But I don't want the documents I've entered to disappear when I have to restart the server. So I tried MariaDB, but then I ran into the issue that the database would hang after running the application once or twice. So then I decided to investigate other databases, and found out that H2 allows me to store the data in files. That seemed to be good, so I used that. I guess it took me two or three hours to overcome this road block.
From then on, it was fairly smooth sailing.
I switched back and forth between different tutorials and reading various API descriptions, and the Thymeleaf documentation. Then, I added a bit of UI polish using Bootstrap and learned about "webjars".
And finally, I looked at the code and tried to see which code might be better suited to be expressed in a functional style (stream processing). Quite possible I've gone a bit overboard with this. The longest stream is less than ten lines of code, though, so maybe it's still okay.
The whole thing took maybe 15 or 20 hours.
Considering my Information Retrieval background, the idea is this: You are able to CRUD documents, where a document comprises a title and some content. The app will maintain an inverted index over these documents: Both title and content will be split into words, each word will be converted to its stem form (using the Porter stemmer, this only works for English). Then we count how often each stem occurs in each document.
You can also enter a search query (also a sequence of words). The search query will be split into words, stemmed, and counted just like a document would. Then we use the vector space model of IR (using tf*idf -- term frequency times inverted document frequency) to rank all the documents against the query.