vmandic

ISet<HzzoMedsDownloadDto> ParseMedsLiElements(IEnumerable<IElement> elems) =>
  elems.Aggregate(new HashSet<HzzoMedsDownloadDto>(), (medsList, li) =>
  {
    var href = li.QuerySelector("a").GetAttribute("href");

    // NOTE: this domain is not available, links don't work :-(
    if (!href.Contains("cdn.hzzo.hr"))
    {
      var dtParts = li.TextContent.TrimEnd().Split(' ').LastOrDefault().Split('.');
      var downloadDto = new HzzoMedsDownloadDto(

vmandic

public class HzzoMedsDownloadDto
{
  private readonly string _rootLocation;

  public HzzoMedsDownloadDto(string href, string validFrom, string rootLocation)
  {
    this.Href = href;
    this.ValidFrom = DateTime.Parse(validFrom);
    this._rootLocation = rootLocation;
  }

vmandic

public async Task<ActionResult> Index(
  [FromServices] HzzoHtmlScraper scraper, 
  [FromServices] HzzoExcelDownloader downloader)
{
  var startTime = DateTime.Now;
  var meds = await downloader.Run(await scraper.Run());
  var totalTime = startTime - DateTime.Now;

  return Ok(
    $"Done! Handler duration: {totalTime.Duration()}" +

vmandic

using System.Collections.Generic;
using System.Threading.Tasks;
using MedsProcessor.Common.Models;

namespace MedsProcessor.Parser
{
	public class HzzoExcelParser
	{
		public Task<ISet<HzzoMedsDownloadDto>> Run(ISet<HzzoMedsDownloadDto> meds)
		{

vmandic

public async Task<ISet<HzzoMedsDownloadDto>> Run(ISet<HzzoMedsDownloadDto> meds)
{
  await Task.WhenAll(
    // NOTE: due to excel docs designed in different ways, we do this separation of work
    StartLongRunning(() => ParsePrimaryListsStartingWith2014_02(meds)),
    StartLongRunning(() => ParseSupplementaryListsStartingWith2014_02(meds)),
    StartLongRunning(() => ParsePrimaryListsUpTo2014_01(meds)),
    StartLongRunning(() => ParseSupplementaryListsUpTo2014_01(meds))
  );

vmandic

static readonly DateTime filterDtStartWith2014 = new DateTime(2014, 1, 3);

void ParseSupplementaryListsUpTo2014_01(ISet<HzzoMedsDownloadDto> meds) =>
  ParseHzzoExcelDocuments(meds.Where(x =>
    x.ValidFrom <= filterDtStartWith2014 &&
    (
      x.FileName.ToLowerInvariant().Contains("dopunska") ||
      x.FileName.ToLowerInvariant().Contains("dll")
    )), DrugListType.Supplementary, false);

vmandic

public class DrugImportDto
{
    public int RowId { get; set; }
    public DrugListType ListType { get; set; }
    public DateTime ValidFrom { get; set; }

    public string AtkCode { get; set; }
    public DrugApplicationTypeLimitation ApplicationTypeLimitation { get; set; }

    public string GenericName { get; set; }

vmandic

using System;

namespace MedsProcessor.Parser
{
	public enum DrugListType
	{
		Undefined = 0,
		Primary = 1,
		Supplementary = 2
	}

vmandic

static ISheet OpenWorkbookSheetWithNpoi(FileStream stream, HzzoMedsDownloadDto med, HzzoMedsDownloadDto latestMed)
{
  ISheet drugListSheet = null;

  try
  {
    if (med.FileName.ToLowerInvariant().EndsWith(".xls"))
    {
      var hssfWorkbook = new HSSFWorkbook(stream);
      drugListSheet = hssfWorkbook.GetSheetAt(0);

vmandic

void ParseHzzoExcelDocuments(IEnumerable<HzzoMedsDownloadDto> filteredMeds, DrugListType listType, bool isListStartingWith2014)
{
  HzzoMedsDownloadDto latestMed = null;
  int latestRow = 0;
  int latestCol = 0;

  try
  {
    Parallel.ForEach(filteredMeds, med =>
    {