From #13, I realized that the current state of the automation is slightly over-engineered and clunky.
After taking a look at the codebase again, getting some ideas about how to improve it with the help of Pydantic and Pytest, and finding some time to do this (as I'm very much interested in optimizing this even further, also general maintenance work), I implemented a change in the algorithm, data structure, and typing.
Algorithm
Before synchronizing the database, it will try to properly find the start and end dates by combining dict[str, DatabaseEntry] and list[Summary], where both of them will use the min and the max for the starting and ending dates, respectively. After this, the dates which are iterable will be generated by declaratively using functional programming (takehwhile, count). Once we have known all of the possible dates, the streak is generated by functional programming again (accumulate) to know whether we have broken our streak or not.
With this algorithm, it should now be possible to accurately find and synchronize dates effortlessly.
Data Structure
One thing that I find strange is that in a day, we can only have one information, but the data structure from the API is a list[Summary]. To leverage the type-system, I transformed it into dict[str, DatabaseEntry] so it could accurately represents the proper state. As this is implemented, impossible states should not be possible now.
Typing
A lot of typings are improved through the usage of Pydantic's BaseModel.
Web
Implemented better caching by calculating the data in advance, instead of deriving them in each calls, which can cause a computation cost.
Tests
Through tests have been created to ensure correctness, so now the coverage should be back to 100%.
Description
From #13, I realized that the current state of the automation is slightly over-engineered and clunky.
After taking a look at the codebase again, getting some ideas about how to improve it with the help of Pydantic and Pytest, and finding some time to do this (as I'm very much interested in optimizing this even further, also general maintenance work), I implemented a change in the algorithm, data structure, and typing.
Algorithm
Before synchronizing the database, it will try to properly find the start and end dates by combining
dict[str, DatabaseEntry]andlist[Summary], where both of them will use the min and the max for the starting and ending dates, respectively. After this, the dates which are iterable will be generated by declaratively using functional programming (takehwhile,count). Once we have known all of the possible dates, the streak is generated by functional programming again (accumulate) to know whether we have broken our streak or not.With this algorithm, it should now be possible to accurately find and synchronize dates effortlessly.
Data Structure
One thing that I find strange is that in a day, we can only have one information, but the data structure from the API is a
list[Summary]. To leverage the type-system, I transformed it intodict[str, DatabaseEntry]so it could accurately represents the proper state. As this is implemented, impossible states should not be possible now.Typing
A lot of typings are improved through the usage of Pydantic's
BaseModel.Web
Implemented better caching by calculating the
datain advance, instead of deriving them in each calls, which can cause a computation cost.Tests
Through tests have been created to ensure correctness, so now the coverage should be back to 100%.