Replace Existing Jobs Filter

feature
1

Hey, I saw the filter someone created which stopped new tasks from being created if one already exists. I needed something different where I would delete an existing job and replace it with a new one.

I filter on the job type and arguments passed. For this use-case, we only care about scheduled jobs. If one is already running, I ignore it… Here you go:

namespace Keg.Web.Filters
{
    using System;
    using System.Linq;

    using Hangfire;
    using Hangfire.Client;
    using Hangfire.Common;
    using Hangfire.States;
    using Hangfire.Storage;

    public class ReplaceScheduledJobFilter: JobFilterAttribute, IClientFilter
    {
        public void OnCreating(CreatingContext filterContext)
        {
            var connection = filterContext.Connection;
            var storageConnection = connection as JobStorageConnection;
            if (storageConnection == null)
            {
                return;
            }

            var currentArgs = filterContext.Job.Args.Select(Convert.ToString);
            var jobIdsToRemove = storageConnection
                .GetAllItemsFromSet("schedule")
                .Select(jobId => new
                {
                    Id = jobId,
                    JobData = storageConnection.GetJobData(jobId),
                })
                .Where(dataset => dataset.JobData.Job.Type == filterContext.Job.Type)
                .Where(dataset => 
                    currentArgs.Intersect(dataset.JobData.Job.Args.Select(Convert.ToString)).Any())
                .Select(dataset => dataset.Id);

            var client = new BackgroundJobClient(JobStorage.Current);
            foreach (var jobId in jobIdsToRemove)
            {
                client.ChangeState(jobId, new DeletedState { Reason = "[ReplaceScheduledJobFilter]" }, ScheduledState.StateName);
            }
        }

        public void OnCreated(CreatedContext filterContext)
        {
        }
    }
}

Let me know if there is a way to improve this. I need an API to select all the JobData in a single query instead of potentially doing hundreds/thousands of queries.