Brain overengineering

I got a simple entertaining task: collect data on water and air temperatures from a couple of HTML pages and return the result in JSON from the API. The task is trivial, it is solved by a code of lines of 40 (or so) with comments. Of course, if you write guided by the Quick & Dirty principle. Then the written code will be smelly and will not correspond to modern programming standards.

---

Let's take a simple execution as a basis and see what happens if we refactor it ( code with commits )

public async Task<ActionResult<MeasurementSet>>
        Get([FromQuery]DateTime? from = null, 
            [FromQuery]DateTime? to = null)
{
	// Defaulting values
  from ??= DateTime.Today.AddDays(-1);
  to ??= DateTime.Today;
  // Defining URLs
  var query = $"beginn={from:dd.MM.yyyy}&ende={to:dd.MM.yyyy}";
  var baseUrl = new Uri("https://BaseURL/");
  using (var client = new HttpClient { BaseAddress = baseUrl })
  {
    // Collecting data
    return Ok(new MeasurementSet 
    { 
      Temperature = await GetMeasures(query, client, "wassertemperatur"), 
      Level = await GetMeasures(query, client, "wasserstand"), 
    });
  }
}

private static async Task<IEnumerable<Measurement>>
        GetMeasures(string query, HttpClient client, string apiName)
{
  // Retrieving the data
  var response = await client.GetAsync($"{apiName}/some/other/url/part/?{query}");
  var html = await response.Content.ReadAsStringAsync();
  // Parsing HTML response
  var bodyMatch = Regex.Match(html, "<tbody>(.*)<\\/tbody>");
  var rowsHtml = bodyMatch.Groups.Values.Last();
  return Regex.Matches(rowsHtml.Value, "<tr  class=\"row2?\"><td >([^<]*)<\\/td><td  class=\"whocares\">([^<]*)<\\/td>")
    // Building the results
    .Select(match => new Measurement
    {
      Date = DateTime.Parse(match.Groups[1].Value),
      Value = decimal.Parse(match.Groups[2].Value)
    });
}
      
      

        
        
        
      

    
        
        
        
      
      

        
        
        
      

    
    

1. Error handling

- , .

if (response.IsSuccessStatusCode)
{
  throw new Exception($"{apiName} gathering failed with. [{response.StatusCode}] {html}");
}
// Parsing HTML response
var bodyMatch = Regex.Match(html, "<tbody>(.*)<\\/tbody>");
if (!bodyMatch.Success)
{
  throw new Exception($"Failed to define data table body. Content: {html}");
}
      
      

        
        
        
      

    
        
        
        
      
      

        
        
        
      

    
    

2.

, , , . , MeasureParser RawMeasuresCollector . , , .

3.

enum, , :

var apiName = measure switch
{
  MeasureType.Temperature => "wassertemperatur",
  MeasureType.Level => "wasserstand",
  _ => throw new NotImplementedException($"Measure type {measure} not implemented")
};
      
      

        
        
        
      

    
        
        
        
      
      

        
        
        
      

    
    

, . :

public class UrlQueryBuilder
{
  public DateTime From { get; set; } = DateTime.Today.AddDays(-1);
  public DateTime To { get; set; } = DateTime.Today;

  public string Build(MeasureType measure)
  {
    var query = $"beginn={From:dd.MM.yyyy}&ende={To:dd.MM.yyyy}";
    var apiName = measure switch
    {
      MeasureType.Temperature => "wassertemperatur",
      MeasureType.Level => "wasserstand",
      _ => throw new NotImplementedException($"Measure type {measure} not implemented")
    };
    return $"{apiName}/some/other/url/part/?{query}";
  }
}
      
      

        
        
        
      

    
        
        
        
      
      

        
        
        
      

    
    

4. (coupling)

. , , . URL :

var settings = Configuration.GetSection("AppSettings").Get<AppSettings>();
services.AddHttpClient(Constants.ClientName, client =>
{
  client.BaseAddress = new Uri(settings.BaseUrl);
});
services.AddTransient<IRawMeasuresCollector, RawMeasuresCollector>();
      
      

        
        
        
      

    
        
        
        
      
      

        
        
        
      

    
    

5.

. , :

    [TestMethod]
public async Task TestHtmlTemperatureParsing()
{
  var collector = new Mock<IRawMeasuresCollector>();
  collector
    .Setup(c => c.CollectRawMeasurement(MeasureType.Temperature))
    .Returns(Task.FromResult(_temperatureDataA));

  var actual = (await new MeasureParser(collector.Object)
    .GetMeasures(MeasureType.Temperature)
    ).ToArray();

  Assert.AreEqual(165, actual.Length);
  Assert.AreEqual(7.1M, actual
      .First(l => l.Date == DateTime.Parse("24.11.2020 10:15")).Value);
}
      
      

        
        
        
      

    
        
        
        
      
      

        
        
        
      

    
    

6.

, , . DOM , . .

:

public async Task<ActionResult<MeasurementSet>> Get
    ([FromQuery] DateTime? from = null, [FromQuery] DateTime? to = null)
{
  var parser = new MeasureParser(_collectorFactory.CreateCollector(from, to));
  return Ok(new MeasurementSet
  {
    Temperature = await parser.GetTemperature(),
    Level = await parser.GetLevel(),
  });
}
      
      

        
        
        
      

    
        
        
        
      
      

        
        
        
      

    
    

The previous fixes led to the uselessness of the enum with the dimension type, and we had to get rid of the code specified in point 3, which is rather positive, since it reduces the degree of code branching and helps to avoid errors.

Outcome

As a result, the one page method has grown into a decent project

Of course, the project is flexible, supported, etc., etc. But there is a feeling that it is too big. According to VS analytics, out of 277 lines of code, only 67 are executable.

Perhaps the example is not correct, since the functionality is not so broad, or the refactoring was done incorrectly, not completely.

Share your experience.