在 C# 中，解析并按时间字符串排序的最佳方法是什么？

Question

I am reading and loading in files into Excel using C# VSTO and the filenames are something like this:

C:\myfiles\1000AM.csv
C:\myfiles\1100AM.csv
C:\myfiles\1200PM.csv
C:\myfiles\100PM.csv
C:\myfiles\200PM.csv

And then i am putting these in a list and need to sort these by "time".

How can i convert the string in the format above into a time object that i can use to sort on?

Answer 1

You need extract the time parts somehow and then compare them to each other.

You could for example do this using a Comparison<string> . Here is an example that uses the Span<T> type to do this without allocating any additional garbage:

List<string> list = new List<string>() { ... }

list.Sort((a, b) => 
{
    //compare AM/PM
    int compareAmAndPm = a.AsSpan().Slice(a.Length - 6, 2)
        .CompareTo(b.AsSpan().Slice(b.Length - 6, 2), StringComparison.Ordinal);
    if (compareAmAndPm != 0)
        return compareAmAndPm;

    //compare the times as integers
    int index = a.LastIndexOf('\\');
    var firstTime = int.Parse(a.AsSpan().Slice(index + 1, a.Length - index - 7));

    index = b.LastIndexOf('\\');
    var secondTime = int.Parse(b.AsSpan().Slice(index + 1, b.Length - index - 7));


    return firstTime.CompareTo(secondTime);
});

It should give you a result of this:

C:\myfiles\1000AM.csv
C:\myfiles\1100AM.csv
C:\myfiles\100PM.csv
C:\myfiles\200PM.csv
C:\myfiles\1200PM.csv

Answer 2

From practice we figured out, that a Time or Date on it's own does not work 99% of the cases. We need both plus the timezone to have any hope of processing them meaningfully.

That is why we only have things like DateTime , nowadays. Ideally those file names should consist of the full DateTime, in UTC and Invariant culture. If you got the option to change how those are created, use it.

However if you consistently only have one part that is not an issue: DateTime simply used default values for the other two. And as those two will be consistent, they will work. The only issue will be a finding a culture setting that eats that AM/PM format.