RATSQL
Machines and Complexity
2nd bachelor Computer Science - University of Antwerp
Made by Cédric Leclercq, Maarten Peirsman, Pablo Deputter and Robbe Nooyens
RATSQL was created for the course Machines and Complexity at the University of Antwerp. Queries in relational algebra form can be entered inside the editor with the help of the special symbol buttons.The query can be translated to SQL using the convert button. Some options can be aplied like the choice to optimize the query or choosing the amount of deviations can take place when requesting a replacement for a selected word.
This year, next to machines and computability, we have another computer-related course that is completely new to us, namely the course "introduction to databases". During these classes so far we learned how to set up entity-relationship models, how to write queries in relational algebra and how to then translate them into a usable SQL query. Of course, with the naked eye and a little background knowledge, converting a simple expression to a SQL query is not that difficult. But what if the relational algebra expression becomes so complicated that the structure is hard to recognize? What if there was a program that could do this?
That is what we too, were wondering. In combination with the parsers that we saw in the subject of machines and computability, this seemed to us the perfect combination to make a project around. So we wanted to create a program that takes as input an expression in relational algebra and gives as output a SQL query. We used an earley parser to do this. The basis of our project is the conversion of an expression to a query. We then included an error detector that can see if a user wrote something wrong, and replace it is the user wants it. We also used the shunting yard algorithm to make a relational algebra expression as optimal as possible. For error detection we used the "Levenshtein distance". This method is used in computer science to calculate how many operations are needed to convert a string into another string b by adding, removing or replacing a character. Thus, it allows us to detect and replace simple spelling errors to avoid errors in our system. Like we said before, to implement nad make the query as optimal as possible, we are going to implement the Shunting-yard algorithm. This algorithm allows us to treat expressions that are contained in another expression first.
Notes
It needs to be noted that no enters are allowed in the direct input for parsing a relational algebra expression. When writing a new enter, a new relational algebra expression is expected.
Some tests of the query tests fail, this is because the tests are written to the optimal output, and our generated view creates some extra views (but is still fully correct).
Usage
Make sure that Qt 6 is installed (there is some support for Qt 5 for certain configurations, but limited). Secondly, make sure to set the working directory to the project root.