AacpsBusAlert/BinaryDad.AacpsBusAlert.Services/BusRouteService/HtmlScrapeBusRouteService.cs

using BinaryDad.Extensions;
using Newtonsoft.Json;
using System.Text.RegularExpressions;

namespace BinaryDad.AacpsBusAlert.Services
{
    public class HtmlScrapeBusRouteService : IBusRouteService
    {
        private readonly HttpClient httpClient;

        public HtmlScrapeBusRouteService(IHttpClientFactory httpClientFactory)
        {
            httpClient = httpClientFactory.CreateClient();
        }

        public async Task<ICollection<BusRoute>> GetBusRoutesAsync()
        {
            var url = "https://busstops.aacps.org/public/BusRouteIssues.aspx";
            var content = await httpClient.GetStringAsync(url);

            var dataArrayMatch = Regex.Match(content, "var dataArray \\= (.*)\\;");

            if (dataArrayMatch.Success && dataArrayMatch.Groups.Count == 2)
            {
                var dataArray = dataArrayMatch.Groups[1].Value;

                var parsedArray = JsonConvert.DeserializeObject<string[][]>(dataArray);

                return parsedArray
                    .Select(r => new BusRoute
                    {
                        BusNumber = r[0].To<int>(),
                        SubBusNumber = r[1].To<int>(),
                        Schools = r[2],
                        Schedules = r[3],
                        Impact = r[4]
                    })
                    .ToList();
            }

            throw new ApplicationException("Unable to parse bus route page");
        }
    }
}