Usage - mazharenko/aoc-agent GitHub Wiki
This page explains the universal conventions related to creating the agent. Some other essential things are explained in other pages related to different approaches preferred by the user.
1. Creating a project
The project must be a console application and reference the mazharenko.AoCAgent package
dotnet new console -n aoc
dotnet add aoc package mazharenko.AocAgent
2. Defining days
Define day classes named like DayXX. Don't forget the partial keyword.
partial class Day01
{
internal partial class Part1
{
}
internal partial class Part2
{
}
}
A lot of internal staff will be generated now to implement the agent functionality. Namely, an interface, a base class and a class for examples. At this moment the complier will require to implement the string Solve(string input) methods from the interface.
partial class Day01
{
internal partial class Part1
{
public string Solve(string input)
{
throw new NotImplementedException();
}
}
internal partial class Part2
{
public string Solve(string input)
{
throw new NotImplementedException();
}
}
}
3. Types
Strongly typed inputs and results are supported. The "implemented" Solve method can be changed according to the desired types.
partial class Day01
{
internal partial class Part1
{
public int Solve(int[] input)
{
throw new NotImplementedException();
}
}
internal partial class Part2 { ... }
}
A lot of internal staff will be regenerated now to reflect the desired types, so the compiler is going to stop complaining shortly. However, for non-string inputs it will now require to implement the T Parse(string input) method.
partial class Day01
{
internal partial class Part1
{
public int[] Parse(string input)
{
throw new NotImplementedException();
}
public int Solve(int[] input)
{
throw new NotImplementedException();
}
}
internal partial class Part2 { ... }
}
For the result's type, there still must be a way to convert it to a string. By default, its ToString method will be called. This can be overridden:
partial class Day01
{
internal partial class Part1
{
...
public override string Format(int res)
{
return base.Format(res);
}
...
}
...
}
4. Examples
Advent of Code problem descriptions usually include sample inputs and correct answers for them. These can be defined for a Part to simplify debugging. The agent calculates and submits answers only when all the provided examples pass. Until examples are given, no attempts are made.
internal partial class Day01
{
internal partial class Part1
{
private readonly Example example = new(
"""
Time: 7 15 30
Distance: 9 40 200
""", 288);
public (long, long)[] Parse(string input) => ... ;
public int Solve((long, long)[] input) => ... ;
}
}
Example is a generated type. Its Expectation member reflects the Solve output type.
5. Reusing stuff
The Part classes are nested into its Day, so its members are available within the Part classes. Additionally, Parts internally inherit from the Day, so the Parse method can be implemented within the Day, if it is expected to be the same for both parts.
partial class Day01
{
public int[] Parse(string input)
{
throw new NotImplementedException();
}
internal partial class Part1
{
public int Solve(int[] input)
{
throw new NotImplementedException();
}
}
internal partial class Part2 { ... }
}
Example inputs can be defined in the Day as well, but then a Part must specify the expectations for them:
internal partial class Day01
{
private readonly Example example = new(
"""
Time: 7 15 30
Distance: 9 40 200
""");
internal partial class Part1
{
public Part1()
{
Expect(example, 288);
}
public (long, long)[] Parse(string input) => ... ;
public int Solve((long, long)[] input) => ... ;
}
}
6. This is not enough
A few questions must be answered before moving on. Depending on one's preferences, further actions might differ.
-
Do you want the library to generate the Entry point for you?
An agent is a console application. Its entry point is supposed to call all the running logic. The entry point can be generated automatically or written manually.
-
Do you follow the repo-per-year scheme or have a single repository for all the Advent of Code events?
How days are discovered and associated with years needs a bit more attention for the latter.
Follow the link which answers these questions best.
Entry point generation, Single year in a project
Entry point generation, Multiple years in a project